article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
one of the major problems in modern astrophysics is to determine physical parameters of stars , masses and ages , and their kinematics .
whereas various methods exist , most of them are not suitable for large stellar samples .
thus the most common approach is fitting to stellar evolution models , i.e. , comparison of an observed location of a star in the @xmath9 plane with isochrones or evolutionary tracks for a given metallicity and @xmath10-enhancement .
this method demands stellar atmospheric parameters as an input , which can be derived from photometry ( e.g. , * ? ? ?
* ; * ? ? ?
* ) or spectroscopy @xcite , or some combination thereof ( e.g. application to sdss / segue data in * ? ? ?
the drawback of photometry is that it depends on reddening , which is difficult to estimate , and demands calibration of metallicity and alpha - enhancement on spectroscopic data .
additionally , information on gravity is not obtained directly , a quantity needed to accurately determine distances .
the alternative and preferable method is to use spectroscopy directly , that allows a consistent determination of all atmospheric quantitites , effective temperature , surface gravity , and element abundances , simultaneously .
thus , it minimizes the the total error caused by various sources of uncertainties inherent to different methods of parameter determination . with the advent of large - scale stellar surveys such as rave @xcite , gcs @xcite , sdss / segue @xcite , gaia - eso @xcite
much attention has been devoted to developing sophisticated statistical methods to best exploit the available spectroscopic and photometric information .
the basic technique of isochrone fitting ( see e.g. @xcite ) , or a combination of complementary techniques @xcite , have evolved into more systematic methods that can be applied to very large amounts of data .
different alternatives based on a frequentist approach ( e.g. @xcite ) or bayesian analysis of ever increasing sophistication @xcite have been presented and their advantages and disadvantages discussed @xcite . as a result ,
spectrophotometric determination of masses , ages , and distances for late - type stars is nowadays routinely applied in large - scale stellar surveys and , although with different implementations , lies at the core of almost all studies aimed at characterizing stellar populations in the milky way , both in the intrinsic stellar properties such as age - metallicity relations @xcite as well as in their kinematical properties @xcite .
however , in these studies little or no attention has been given to the physical accuracy of stellar atmospheric parameters , which , in one way or another , always rely on models describing radiative transport in stellar atmospheres . in spectroscopy ,
all theoretical grids of synthetic stellar spectra for late - type stars have been computed based on the assumptions of local thermodynamic and 1d hydrostatic equilibrium ( hereafter 1d lte ) .
these are , for example , the widely - used synthetic spectra grids by @xcite , which rely on the kurucz 1d lte model atmosphere and radiative transfer codes .
however , recent observational and theoretial studies revealed substantial systematic biases in basic stellar parameters at low metallicity and/or gravity caused by the breakdown of lte and 1d hydrostatic equilibrium approximations ( e.g. , * ? ? ? * ; * ? ? ? * ) .
it must be pointed out that these biases are ubiquitous , and affect the results obtained from the analysis of high- and low - resolution spectra .
the reason is that departures from 1d he and lte often change equivalent widths ( or , equivalently , line indices ) , and consequently abundances inferred from a diagnostic spectral feature .
this effect does not vanish with a decreasing spectral resolution , which merely alters the line profile shape without affecting the total energy absorbed in a line .
although no detailed quantitative estimates have been performed so far , one may also expect a non - negligible impact on certain photometric indices . as a consequence ,
it is completely unknown whether , and how , systematic errors in basic stellar parameters caused by the assumptions of 1d lte impact the inferred stellar properties , that is , in our context , ages , masses , and distances . in ( * ? ?
* hereafter paper 1 ) , we presented a comprehensive study of systematic errors in the determination of spectroscopic stellar parameters arising from using the classical method of excitation - ionization balance of fe@xmath11i and fe@xmath11ii lines based on 1d lte model atmospheres .
based on the results for the large sample of metal - poor stars ( @xmath1 [ fe / h ] @xmath12 ) selected from the rave catalogue , we concluded that the impact of using inaccurate effective temperature scale and neglecting the effects of non - local thermodynamic equilibrium ( nlte ) on the determination of gravities and metallicities leads to systematic biases of up to @xmath13 k in effective temperature , @xmath14 dex in surface gravity and @xmath15 dex in metallicity can be obtained by independent means , i.e. , from high - quality photometry with a careful correction for reddening .
a more reasonable way is to resort to standard methods relying on global @xmath16 minimization , as e.g. in @xcite , using , however , the synthetic grids computed completely in nlte . such a code is currently being developed by one of the co - authors and will be applied to a large dataset of low - resolution spectra . ] . in this work ,
we use results from paper 1 to analyze the impact of using physically accurate spectroscopic stellar parameters on the determination of their masses , ages , and distances .
our paper is structured as follows : section [ sec : obs ] gives details on the observational data and the different techniques used in the determination of basic spectroscopic parameters .
section [ sec : models ] is a brief description of the grid of stellar evolution models computed for this work . in section [ sec : bayes ] we describe the novel bayesian method we apply to obtain masses , ages and , distances , and discuss the limitations , particularly imposed by the degeneracy of stellar models on the observational plane .
we present the results in section [ sec : results ] and compare them with those published by the rave collaboration .
finally , in section [ sec : summary ] we summarize the impact of our findings in the broader context of galactic structure studies and emphasize the absolute need of accurate spectroscopic stellar parameters in order to fully exploit the enormous amount and quality of data emerging from large - scale stellar surveys .
[ sec : obs ] the observed dataset was taken from @xcite and it comprises the core of the sample used in ( * ? ? ?
* hereafter paper 1 ) as well .
these stars were originally selected for high - resolution observations based on data obtained by the radial velocity experiment survey ( rave ; @xcite ) in order to study the metal - poor thick disk of the milky way .
high signal - to - noise spectra for these stars were obtained using high - resolution echelle spectrographs ( @xmath17 and s / n @xmath18 per pixel ) at several facilities around the world . for the present work
, the final sample contains 254 stars in the metallicity range . with respect to the full sample
, we have excluded horizontal branch candidates because the grid of stellar models used here extends only up to the tip of the red giant branch ( rgb ) .
also , stars with have not been considered to avoid large extrapolations from the grid of nlte corrections that extends down to @xmath19 @xcite stellar parameters were determined using two different techniques , as described in detail in paper 1 and summarized below .
first , we apply the classical method of 1d lte excitation - ionization equilibrium of fe@xmath11i and fe@xmath11ii , which relies on the minimization of the abundance trends with reduced line equivalent widths and lower level excitation potential .
@xmath20 and microturbulence are optimized to establish excitation balance , and @xmath0 and [ fe / h ] are found by simultaneously requiring ionization balance .
notwithstanding the physical limitations due to the assumptions of 1d lte , the method is still in widespread use today because of its simplicity and easiness of implementation .
in the second approach , substantial efforts were made to improve the accuracy of basic stellar parameters in the attempt to minimize the above - mentioned systematic errors .
firstly , effective temperatures were determined from different methods available for cool stars ( such as the balmer lines , semi - empirically calibrated photometry ) .
the final , optimal , @xmath20 is that derived from the weighted averaging of different methods . in a second step , we recomputed gravities and metallicites relaxing the lte assumption and adopting the new @xmath20 scale .
the nlte corrections were applied to fe@xmath11i abundances using the new wrapper code to moog .
the typical errors are @xmath21 k , @xmath22 dex , and @xmath23 } ) \approx 0.1 $ ] dex .
note that these are the absolute errors ( systematic + internal ) on the accuracy of our stellar parameters .
similar errors bars are typically quoted in standard techniques .
however , usually those represent the internal precision of a method only .
the results obtained by the lte - fe method are affected by much larger systematic errors . in the paper 1
, we contrasted our results for stars with surface gravities and temperatures determined by independent techniques , i.e. , from accurate _
hipparcos _ parallaxes and interferometry .
we found that while the lte - fe surface gravities were systematically too low by @xmath24 dex , the nlte - opt surface gravities were in agreement with astrometric values to @xmath25 dex .
this supports the accuracy of the new method . the spectroscopic stellar parameters obtained using both methods described above are compared in fig .
[ fig : nltepar ] .
hereafter , these two sets of stellar parameters are referred to as @xmath26 ( lte - fe method ) and @xmath27 ( nlte - opt method ) .
the major conclusion to be drawn from these plots is that the classical method of 1d lte excitation - ionization balance of fe leads to systematic errors in all three basic stellar parameters : * @xmath20 ( fig .
[ fig : nltepar ] , top panel ) : the lte - fe @xmath20 values are too low by for all stars below [ fe / h ] @xmath28 ; for dwarfs above this metallicity , the effective temperatures are usually over - estimated .
note that the lte - fe @xmath20 values for very cool giants appear to be reasonably accurate for [ fe / h ] @xmath29 .
* @xmath0 ( fig .
[ fig : nltepar ] , middle panel ) : surface gravities are usually under - estimated . the largest effects is seen for warm giants , @xmath30 k , which are biased by up to @xmath14 dex at [ fe / h ] @xmath31 and by @xmath15 dex at [ fe / h ] @xmath32 ; for metal - poor dwarfs , gravities are under - estimated by @xmath33 dex . for stars with [ fe / h ] @xmath34
, the errors are within @xmath35 dex .
* ( fig .
[ fig : nltepar ] , bottom panel ) : the error is nearly a linear function of metallicity itself , such that for dwarfs and giants with [ fe / h ] @xmath34 the error in [ fe / h ] is @xmath36dex
. the error changes sign and increases to @xmath37 dex at [ fe / h ] @xmath32 , while at [ fe / h ] @xmath31 the lte metallicity is too low by @xmath15 dex for giants and @xmath38 dex for dwarfs .
is it seen that the systematic effects are metallicity- and gravity - dependent , which is reflected in a notable spread of @xmath39 and @xmath40 at a given @xmath0 and @xmath20 .
[ fig : tracks_data ] compares the location of the stars for the parameter sets @xmath26 ( left panel ) and @xmath27 ( right panel ) on the @xmath20 - @xmath0 plane together with reference stellar evolutionary tracks computed using the ` garstec ` code ( * ? ? ?
* see next section for a description of the grid of stellar models ) .
tracks are identical in the left and right plots .
symbol sizes are proportional to [ fe / h ] as indicated in the right figure . even though the tracks are shown only for visual guidance , it is clear that the @xmath27 parameters trace the morphology of the turn - off much better .
results for giants are also quite encouraging . for each star with @xmath41
k , we have searched the entire grid of stellar models the find the best fit to @xmath26 and @xmath27 parameters as defined by the minimum @xmath16 . in the case of @xmath26 parameters ,
the average minimum @xmath16 for these stars is 1.9 , whereas if @xmath27 parameters are used it is 0.4
. clearly , @xmath27 parameters lead to an improved overall agreement with stellar tracks .
[ sec : models ] stellar models have been computed with ` garstec ` and detailed information about numerical aspects of the code and the input physics available can be found in @xcite ; here we just provide a very brief summary of the most relevant input physics included in the stellar models .
regarding microphysics , we have used the freeeos equation of state @xcite and the @xmath42 reaction rate recommended by luna @xcite .
convection is accounted for by using the standard mixing length theory @xcite and the mixing length parameter @xmath43 is taken from the calibration of a solar model ( see below ) .
convective overshooting is modeled as a diffusive process following the formulation by ( @xcite and see also eq.3 in @xcite ) and the free parameter @xmath44 is fixed to 0.02 .
this value represents a moderate amount of overshooting on the main sequence comparable to the canonical 0.2 @xmath45 is the pressure scale height , where @xmath46 and @xmath47 are the pressure and gas density respectively . ] . however , overshooting has to be limited in small convective cores still has , at the present solar age , a convective core .
this is ruled out from helioseismic results . ] . in ` garstec
` this is now achieved by implementing a geometric cutoff where the extension of the overshooting region decreases quadratically with the ratio between the size of the convective core and the pressure scale height at its boundary ( see @xcite for details ) . in order to fix the mixing length parameter and the reference composition of our models
we have calibrated a standard solar model using the @xcite solar composition , from which the present - day surface @xmath48 and @xmath49 .
the latter defines @xmath50= 0}$ ] in our models . from the calibration we obtain @xmath51 and the initial solar helium abundance @xmath52 and metallicity @xmath53 . due to the effects of microscopic diffusion , the initial solar composition corresponds to @xmath50= + 0.06}$ ] . for the present work ,
we have computed a dense grid of stellar models in the mass range @xmath54 in steps of 0.01 m@xmath55 .
the initial composition of the models is computed assuming a cosmic helium - to - metal enrichment @xmath56 anchored to the initial solar composition .
the initial fe abundance in the grid spans the range @xmath57 \leq + 0.5}$ ] in steps of 0.1 dex for @xmath58 < 0.0}$ ] and 0.05 dex for @xmath58 > 0.0}$ ] . in the evolutionary tracks
is 0.06 dex higher than quoted on the label .
we keep this nomenclature in order to associate [ fe / h]= 0 evolutionary tracks to those with initial composition equal to initial solar composition . ]
the set of tracks on which we base the results presented in the following sections include microscopic diffusion and a 0.4 dex @xmath10-element enhancement for @xmath58 \leq -0.6}$ ] .
a maximum depletion of metals due to gravitational settling of about 0.4 dex is obtained for stars typical of the turn - off in metal poor ( [ fe / h ] @xmath59 ) globular clusters .
no additional extra mixing to those mentioned is included in the models .
a detailed discussion on this is of no relevance to this work because we are interested in assessing the effect of input spectroscopic parameters in the determination of stellar properties , and these results do not depend on the reference stellar tracks used .
models are evolved from pre - main sequence up to the tip of the rgb or the age @xmath60 gyr , whichever occurs first .
bayesian inference offers a powerful way to characterize a system by allowing the incorporation of prior knowledge into the statistical analysis in an objective way . in the context of our problem , i.e. determination of fundamental stellar parameters ,
bayesian methods have been discussed and developed by various authors .
of particular pedagogical interest are the articles by @xcite and @xcite and more recent and sophisticated analysis with applications to large stellar surveys can be found in @xcite , @xcite and @xcite . our implementation of bayesian inference of basic stellar parameters is described in the next section .
[ sec : method ] in bayesian analysis , the stellar parameters we want to determine , e.g mass and age , are random variables for which the probability distribution function ( pdf ) needs to be found .
in particular , the posterior pdf , that is the probability of a model being correct given the observed data and the prior knowledge , is expressed as @xmath61}~|~t_{\rm eff , obs } , ... ) & \propto & p ( m,\tau,{\rm [ fe / h ] } ) \\ & & \times \ l(t_{\rm eff , obs } , ... ~|~m,\tau,{\rm [ fe / h ] } ) \end{array } \label{eq : bayes}\ ] ] where the function @xmath62})$ ] on the right - hand side represents the prior probability of the stellar parameters and @xmath63})$ ] the likelihood , i.e. the probability of the observed parameters given a model characterized by a set of observables d@xmath64})$ ] . assuming independence and gaussian errors in the observed data , the likelihood is @xmath65})\right)^2 } { 2 \sigma_j^2 } \right)}.\ ] ] in our case , @xmath66 consists of three parameters : @xmath20 , @xmath0 , and [ fe / h ] . other parameters , like photometric information , can be readily implemented , but we do not consider them here . the prior probability density of model parameters can be written as @xmath67 } ) = p(\tau ) \times p({\rm [ fe / h]}~|~\tau ) \times p(m~|~{\rm [ fe / h]},\tau)\ ] ] for the prior on mass , we assume a standard salpeter initial mass function ( imf ) independent of age and metallicity . for metallicity we assume a flat prior , i.e. that at all ages all metallicities are _ a priori _ equally probable . this simple choice is justified in this work because we are not interested in studying age - metallicity relations for stellar populations and because sufficiently reliable estimates of metallicity are available . finally , for the age we also assume a flat prior , i.e. that star formation rate ( sfr ) is constant , but we set a cut at @xmath68 gyr .
therefore @xmath69 ~ m_{\odot } , \label{eq : imf}\ ] ] @xmath70 ~{\rm gyr}. \label{eq : sfr}\ ] ] more complex priors can be elaborated , but our main goal here is to investigate the differences in masses , ages , and distances due to the input spectroscopic parameters rather than to characterize stellar populations . finally , the proportionality constant in eq .
[ eq : bayes ] is the normalization of the posterior pdf such that @xmath71}~|~t_{\rm eff , obs } , ... ) \,{\rm dm\,d\tau\,d[fe / h]}=1 $ ] , and it is of no relevance for the estimation of parameters and uncertainties .
the pdf of a stellar parameter @xmath72 that takes the values @xmath73 along evolutionary tracks is given by @xmath74}\,|\,t_{\rm eff , obs } , ... ) \,dm\,d\tau\,d{\rm [ fe / h ] } \label{eq : ageprob } \end{array}\ ] ] where the integration is over all evolutionary tracks .
a key advantage of bayesian inference is that , as expressed before , it allows using prior knowledge about the problem being considered .
priors are particularly important when a quantity of interest , affected by the prior , has noticeable changes within the range of observational uncertainties where the likelihood does not change much .
a typical example , originally discussed by @xcite and further expanded by @xcite , is the different evolutionary speed of stellar models that populate the same region of observed parameters .
when this is not accounted for , as in methods purely based on maximum likelihood , the results are biased towards faster evolutionary phases , a bias known as the terminal - age bias @xcite . to illustrate here the importance of using information not contained in the likelihood , we present a similar example .
let us assume a star has the following observed parameters : and . for simplicity ,
we exclude metallicity in this example by fixing it to [ fe / h ] @xmath75 . in the top panel of fig .
[ fig : example_bayes ] we show three evolutionary tracks of models with different masses overlapping in the range of the observed parameters .
while the 1.1 m@xmath55 model is entering the subgiant branch ( sgb ) , the 1.2 m@xmath55 is on the main sequence ( ms ) and the 1.3 m@xmath55 model is still on the pre - main sequence ( pms ) .
first , we perform a standard maximum likelihood analysis ( hereafter , ml ) to determine the stellar mass and age based on monte - carlo sampling .
we create synthetic distributions of @xmath20 and @xmath0 assuming gaussian distributions of errors and search for the best - fit model in all evolutionary tracks of the assumed metallicity . the resulting mass and age distributions
are shown in the middle and bottom panels of fig .
[ fig : example_bayes ] , respectively .
the mass distribution shows a bimodality with a secondary peak and a tail extending towards larger masses ( @xmath76 ) . the age distribution shows a dominant peak at very low ages ( @xmath77 ) clearly seen in the inset that zooms in in this age range
there are two other local maxima at about 4.3 gyr and 4.8 gyr , but with much lower relative probabilities than the solution at very low age .
thus the ml results indicate a strong preference for the pms solution for the observed parameters . while it can not be completely ruled out that the star is on the pms , considerations of the shortness of this evolutionary phase render this possibility , a priori , not very likely .
a brute force way of taking this into consideration is to directly eliminate pms models from the analysis .
we impose this restriction by removing all stellar models with ages @xmath78 myr . in this way , while the best - fit model is still determined purely from its ml , we are already using prior information by assuming the chance of observing a pms star is negligible ( actually null , although this also is a wrong assumption ! ) . results for this exercise
are labelled ml no - pms in the middle and bottom panels of fig .
[ fig : example_bayes ] . in the mass distribution
, the tail at high masses vanished since its origin was purely from the pms models .
the same is true for the ages : the unrealistically large peak at @xmath79 myr disappeared , and the peaks at 4.3 gyr and 4.8 gyr are now the dominant features , but their relative weight is still the same as in the pure ml analysis and the older peak , associated with stars entering the sgb still dominates . the results of the bayesian analysis for our fiducial star are also shown in fig .
[ fig : example_bayes ] ( middle and bottom panels , solid black line ) . as in the case of the ml case ,
three peaks are present but the maximum probability in each of them reflects information not contained in the ml analysis . the pms solution , although possible , is now strongly disfavoured , whereas the ms solution is now the most likely one and it outweighs the sgb solution .
the bayesian mass distribution ( middle panel ) is similar to the ml no - pms case .
the vanishing contribution of the pms is now easy to understand from the eq .
[ eq : ageprob ] . the probability for a given mass
is obtained by marginalizing the posterior pdf over all possible ages . the stellar tracks with high masses , which contribute in the observational box with the pms ,
are strongly disfavoured because of the short duration of this phase .
however , the pms solution still has a non - negligible probability in the age distribution .
the widths of the peaks reflect evolutionary speed in each phase , with pms evolution being about @xmath80 times faster than late ms evolution and early sgb a few times faster than late ms .
there is one important aspect related to the interpetation of the height of the peaks in the bayesian age pdf . strictly speaking ,
when determining the probability of a given age @xmath81 using equation [ eq : ageprob ] , there is no information on the speed of stellar evolution because we are looking at a time snapshot .
as stated by @xcite , the relative heights of the peaks in the age distribution are determined by how closely packed evolutionary tracks of different masses are around the observational error box ( i.e. , the region over which likelihood is significant and varies little ) . where variations in stellar mass introduce small changes on the location of a stellar model ( for the same age )
, the integrand in equation [ eq : ageprob ] will vary slowly when integrating in mass .
thus , the interpretation of the height of the peaks as depending on the evolutionary speed of stars is not straightforward .
it only follows if slow evolutionary speeds are associated with regions on the observational plane where small changes in mass produce small changes in the observational quantities the determine the likelihood . as stated above , the unique direct manifestation of a slowly evolving phase in the age pdf is the width of the peaks in the distribution . band . on the left results are shown for a turn - off star and , on the right for a star on the red giant branch . in the latter case ,
neither mass or age can be determined . the absolute magnitude ( luminosity ) , on the other hand , can be derived precisely in both cases .
black solid lines : salpeter imf as a prior in the mass distribution ; red dashed lines : flat ifm .
[ fig : pdf],width=321 ] fig . [ fig : pdf ] shows in black thick lines the pdfs of stellar parameters of two stars from our sample . on the left ,
j112344.9 - 290327 has [ fe / h ] @xmath82 , @xmath83 k , and @xmath84 , i.e. it is a moderately metal - poor turn - off star . in this case ,
mass , age and absolute @xmath85 magnitude magnitudes only , because they are used to determine distances . ]
have pdfs that lead to well determined central values and uncertainties .
following @xcite , we compute stellar parameters and their uncertainty directly from the different moments of the pdfs . in this case , we obtain @xmath86 gyr , @xmath87 , and @xmath88 . as a second case we have selected j115337.3 - 020036 , with [ fe / h ] @xmath89 , @xmath90 k , and @xmath91 , a star on the lower part of the rgb .
as it is well known , evolutionary tracks of stars with different masses and similar metallicity are almost degenerate along the rgb in the @xmath92 plane ( fig .
[ fig : tracks_data ] ) .
this degeneracy precludes the determination of its mass and age as seen from the pdfs in fig .
[ fig : pdf ] .
the cut in the age distribution at 15 gyr simply reflects our age prior , and this corresponds to the mass cut at @xmath93 m@xmath55 .
the shapes of the pdfs are almost completely determined by the adopted priors , i.e. , imf and sfr , and by the fact that more massive stars evolve faster along the rgb than lower mass stars .
no actual information about the true stellar mass and age can be obtained in this case .
in fact , a pure likelihood analysis leads to age and mass distributions that are basically flat . on the other hand ,
the absolute magnitude of this star is still well - defined , and in fact yields @xmath94 , the same precision as for the turn - off star .
this conclusion remains valid for all stars along the rgb , the absolute magnitude is a well - defined quantity even if the underlying evolutionary tracks are highly degenerate , and we can robustly estimate stellar distances independently of the evolutionary stage of the stars .
finally , we have tested the impact of the imf prior by considering also a flat imf .
results for the two examples just discussed are shown in figure [ fig : pdf ] in red dashed line . clearly , the impact of changing the imf is minimal in those cases where stellar parameters are well defined . on the contrary ,
the large impact of priors in the age and mass distributions for the giant star are evidence that no actual information can be extracted in these cases : the posterior pdf is largely determined by the priors .
[ sec : results ] using the procedure described in sect .
[ sec : method ] and the two sets , @xmath26 and @xmath27 of stellar parameters ( section [ sec : obs ] ) , we computed ages , masses and absolute magnitudes for the stars in our sample .
to determine the distances , we used the observed 2mass @xmath95 magnitudes and the absolute @xmath95 magnitudes obtained from the bayesian analysis .
the synthetic magnitudes were computed from the marcs grid of synthetic spectra @xcite assuming plane - parallel radiative transfer for @xmath96 and spherically - symmetric mode otherwise .
the synthetic @xmath97 magnitudes have been computed adopting the absolute calibrations and zero - points discussed in @xcite .
reddening for each star was computed using @xcite maps following the iterative procedure described in @xcite . , width=321 ] , @xmath0 , [ fe / h ] ) are from the rave dr3 catalogue for both distance determinations .
[ fig : valid_dist],width=321 ] in order to assess the robustness of our bayesian scheme , we selected two sub - samples of stars with basic stellar parameters from @xcite and from @xcite .
for the former , we determine ages and compare them with the results from @xcite . for the latter ,
we focus on the distances . for stellar ages ,
the comparison only comprises stars for which @xcite derived ages from isochrone fittingk were excluded from the comparison because the grid of stellar evolution models is not complete above those temperatures and because we use the ` marcs ` model atmospheres @xcite , which are available for @xmath98 k only . ] in order to avoid biases derived from comparing ages obtained with methods not based on stellar evolutionary tracks . the very good agreement ( fig .
[ fig : lach ] ) implies that when the observational uncertainties are small , @xmath99 in @xmath20 and @xmath100 mag in @xmath101 , maximum likelihood methods return similar results to the bayesian analysis and age determinations are robust .
a similar agreement was reported by @xcite .
the uncertainties in the ages we derive are generally smaller compared both to @xcite and @xcite .
we have derived stellar distances using the input stellar parameters from @xcite for the 58 stars common with our sample , and show the results in comparison with rave distances @xcite in fig .
[ fig : valid_dist ] . results are an excellent agreement .
therefore , we conclude that , provided the same input stellar parameters are used , our bayesian scheme and the isochrone matching method by @xcite yield very similar results , with no obvious systematic offsets between methods .
this is in line with @xcite , who also applied a ( more sophisticated ) bayesian analysis to the rave stellar sample and found that , although some systematic differences were found , the overall agreement with @xcite was good . are included ( see text ) .
[ fig : agemass],title="fig:",width=321 ] are included ( see text ) .
[ fig : agemass],title="fig:",width=321 ] having verified the reliability of the bayesian method , we proceed with the analysis of our stellar sample .
the degeneracy of stellar tracks discussed in sec .
[ sec : realstars ] limits the ability to derive masses and ages on the rgb . for this reason
, we derived masses and ages for stars with @xmath102 , that is , for @xmath103 stars only . in fig .
[ fig : agemass ] we compare masses and ages derived from the two sets @xmath26 and @xmath27 of input stellar parameters .
stellar masses derived from the @xmath27 parameters are systematically lower than those derived from @xmath26 , by about @xmath104 m@xmath55 .
the primary reason for the difference lies in the larger gravities obtained with the @xmath27 method .
errors of the masses are comparable in both cases , and range between 0.03 m@xmath55 up to 0.10 m@xmath55 .
nevertheless , the systematic shift to lower masses caused by the use of more accurate @xmath27 stellar parameters can not be neglected . for stellar ages ,
the behavior is opposite , as expected .
the results based on the @xmath27 parameters are higher by @xmath8 to @xmath105 percent .
however , not in all cases a smaller mass implies a larger age .
close inspection of both panels in fig .
[ fig : agemass ] reveals only two stars with larger @xmath27 masses , whereas six of them have smaller ages derived from @xmath27 parameters .
typical uncertainties of our bayesian ages are of the order of 20% and , except for two cases , range between 10% and 30% .
we conclude that the lte - fe method underestimates stellar ages by @xmath106 gyr , and the errors are more pronounced for hotter and more metal - poor stars .
the main goal of this section is to show the dramatic impact that using the physically realistic @xmath27 stellar parameters has on the determination of stellar distances . as summarized in sect .
[ sec : obs ] , there are large differences between the @xmath26 and @xmath27 stellar parameters sets . the top panel of fig .
[ dist_rel ] shows the change in the absolute magnitude @xmath107 as a function of the @xmath108 differences .
the strong linear correlation simply reflects the fact that changes in @xmath0 are accompanied by a reciprocal change in luminosity due to their inverse functional dependence on stellar radius .
clearly , the @xmath27 stellar parameters indicate that all our stars , with few exceptions , are _ fainter _ than what would be inferred from using the lte - fe parameters . furthermore
, if , for the sake of discussion , we neglect the effect of reddening , then distance scales exactly as the square root of the luminosity of the star .
consequently , distances derived from the @xmath26 set of parameters are overestimated , as illustrated in fig .
[ dist_rel ] ( bottom panel ) where we show the ratio of distances derived using @xmath26 and @xmath27 stellar parameters .
similar information is provided in fig .
[ fig : dist_logg ] in which the ratio of distances , corrected for reddening , is plotted as a function of @xmath109 .
this illustrates clearly how systematic errors in distance determinations impact different stellar types .
stellar metallicities are encoded in the symbol sizes and colours , as indicated in the legend .
distances to metal - rich stars , [ fe / h ] @xmath110 , are , to first order , unbiased .
on the other hand , for metal - poor ( [ fe / h ] @xmath111 ) dwarfs and sub - giants ( @xmath112 ) , lte - fe parameters imply a systematic distance error of about @xmath113 .
the problems are exacerbated at low metallicity and low gravity .
metal - poor giants suffer from the largest fractional distance biases of up to @xmath3 .
magnitudes and distances caused by using the improved set of input basic stellar parameters , @xmath27 as a function of the change in @xmath0 , for our full sample . [ dist_rel],width=321 ] and @xmath27 stellar parameters as a function of surface gravity for our full sample . symbol sizes and colours
indicate the stellar metallicity .
[ fig : dist_logg],width=321 ] fig .
[ fig : dist_abs ] shows the final distribution of distances determined using the @xmath27 and @xmath26 stellar parameters for our full sample .
this plot summarizes what has already been discussed above , and allows us to put the results into the more general perspective of large - scale stellar surveys . in a magnitude - limited survey , for example
, where more metal - rich unevolved stars dominate the nearby sample and metal - poor luminous giants are predominantly observed at large distances , classical full lte - fe analysis , @xmath0 , and [ fe / h ] can be all determined simultaneously .
] will systematically overestimate distances , placing stars progressively further than they are .
this has important consequences in all cases when stellar metallicities , positions , and kinematics are used for studies of galactic structure and evolution @xcite .
we touch upon this in sect .
[ sec : summary ] .
stellar parameters and those from the rave dr3 catalogue for stars common in both samples .
the trend strongly resembles that of the comparison between @xmath26 and @xmath27 distances ( fig .
[ fig : dist_abs ] ) .
[ fig : rave_vs_us_dist],width=321 ] stellar parameters and those from the rave dr3 catalogue for stars common in both samples .
the trend strongly resembles that of the comparison between @xmath26 and @xmath27 distances ( fig .
[ fig : dist_abs ] ) .
[ fig : rave_vs_us_dist],width=321 ] stellar parameters and those from the rave dr3 catalogue for stars common in both samples .
the trend strongly resembles that of the comparison between @xmath26 and @xmath27 distances ( fig .
[ fig : dist_abs ] ) .
[ fig : rave_vs_us_dist],width=321,height=200 ] since our sample of stars was drawn from the internal rave catalogue , it is very important to investigate how our results , i.e. @xmath20 , @xmath0 , [ fe / h ] , and distances , compare with those obtained by the rave collaboration . in what follows
, we will compare our data with those by @xcite and @xmath0 for the 58 stars we use in this comparison are exactly the same as in @xcite ; only [ fe / h ] differ by about 0.1 to 0.2 dex , and this has a very minor impact on the determination of distances .
we thus prefer to use all data from @xcite , which has been used by rave to obtain distances .
] , with which we have @xmath114 stars in common .
we highlight the key methodological differences below . *
the quality of observations .
our parameters rely on high - resolution data , r @xmath115 , with a very broad wavelength coverage , @xmath116 ; the rave spectra have r @xmath117 and cover only a limited wavelength range between @xmath118 and @xmath119 . * the physics of stellar atmospheres and radiative transfer models .
in particular , our analysis relies on nlte radiative transfer , while the rave spectroscopic pipeline operates fully on 1d lte synthetic spectra . * the different approaches to determine masses and ages , and stellar evolution models . `
garstec ` tracks were used in this work , whereas @xcite adopted padova isochrones .
also , the approach followed in that study is esentially a nearest neighbor technique , however with the important difference that the isochrones were resampled to a uniform spacing in age to account for the variable speed of stellar evolution , and a weighing scheme to include , for example , the impact of the initial mass function . in this respect
, their analysis also makes use of priors . * @xcite neglected reddening in their distance determinations .
we used the reddening values following @xcite ; also comparison with the interstellar nad lines in our high - resolution spectra indicates that reddening is not negligible for many stars .
* the uncertainties on our input stellar parameters are typically smaller than those of @xcite .
their values are @xmath120 dex in [ fe / h ] and @xmath0 and range between @xmath18 to 200 k in most cases for @xmath20 , whereas our standard estimates are @xmath121 k in @xmath20 , and @xmath122 dex in [ fe / h ] and @xmath0 .
the differences between @xmath27 parameters and those from the rave dr3 pipeline are shown in fig .
[ fig : rave_vs_us_para ] . size and colour of the symbols indicate metallicity and gravity ( we used rave dr3 values for reference ) .
we also show the differences @xmath27 and @xmath26 for the same stars in grey empty squares .
rave dr3 parameters show a very large spread with respect to @xmath27 .
in fact , the spread is much larger than with respect to the @xmath26 parameters .
clearly it is not possible to single out specific factors which cause the spread ; however , some conclusions can be drawn . in most cases , the rave dr3 @xmath0 values
are clearly underestimated for giants with @xmath123 up to @xmath124 dex and one case as large as 1.3 dex .
this effect is more pronounced for the metal - poor stars , in qualitative agreement with the differences between the @xmath27-@xmath26 parameters for the same stars .
the same is true for subgiants ; in most cases the dr3 gravities are underestimated . for dwarfs , on the other hand
, we find dr3 gravities are overestimated by as much as 0.5 dex .
@xcite arrive at the same conclusion comparing dr3 gravities with stellar models . as a consequence of the differences in stellar parameters , particularly changes in @xmath0 , the distances we obtain from the @xmath27 parameters are systematically _ smaller _ than the rave dr3 data ( @xcite ) .
for the most distant metal - poor giants , the offset is of the order @xmath125 percent .
this is illustrated in figure [ fig : rave_vs_us_dist ] , which shows that beyond 1.5 kpc all @xmath27 distances but one are smaller than those derived in @xcite .
for a few dwarfs in the immediate solar neighborhood , the results are opposite , as expected from the sign of the @xmath0 corrections .
the overall trend is very clear and the qualitative similarity with the difference between @xmath27 and @xmath26 distances ( fig .
[ fig : dist_abs ] ) is striking .
these results provide alarming evidence that the rave dr3 stellar parameters are affected by the systematic error caused by the 1d lte assumption in spectroscopic parameter determinations .
test calculations for the caii near - ir triplet using the mean 3d model atmospheres from @xcite indicate that these spectral lines are strongly affected by the @xmath126 differences , while the near - ir fe i lines are mostly affected by nlte .
[ fig : spectra ] illustrates these effects for the two well - studied metal - poor stars , a subgiant hd 140283 ( top plot ) and a warm turn - off star hd 84937 ( bottom plot ) with [ fe / h ] @xmath127 and @xmath128 dex , respectively .
the spectra are from @xcite and shown in the top panels in their original resolution , @xmath129 . in the bottom panels
the spectra have been degraded to the quality of the rave spectra , i.e. @xmath130 and a typical s / n @xmath131 ( * ? ? ?
the @xmath132 nlte synthetic spectra were computed as in @xcite , but applying lte radiative transfer to caii .
the 1d lte radiative transfer models under - estimate the strength of the caii lines ( the effect of @xmath133 structure ) and over - estimate that of the fe i lines ( the effect of nlte ) . also , the near - ir lines of @xmath10-elements ( si i , mg i , and ti i ) , which along with fe i lines dominate the rave spectra @xcite all form in nlte @xcite and it is unlikely that the nlte effects cancel out to produce a spectrum close to lte
. the effects are clearly present irrespectively of the spectral resolution .
we close this section with a few comments related to the two different validations that @xcite presented for distance determination from rave data .
the first one is a comparison with hipparcos stars present in their catalogue with trigonometric errors smaller than 20% and spectroscopic distance errors smaller than 32% .
these results are shown in figure 3 in @xcite and a large spread is present in the comparison . even if no systematic difference was found , as noted by the authors , almost all these stars dwarfs with a solar - like metallicity ( [ m / h]=@xmath134 ) , exactly the case where we do not expect large nlte effects .
the second test consisted in comparing about 14 stars in 7 open clusters with known distances .
again , all these clusters have solar - like metallicity @xcite and we do not expect nlte effects to play a central role . in any case , we note that distances derived by ( * ? ? ?
* see their table 2 ) for the cluster stars deviate from the cluster distances by anything between -20 to + 40% .
these stars were reanalyzed by ( * ? ? ?
* see their table 3 ) and results show a similar spread , although in a few cases distance determinations are also quite different from those in @xcite .
it is simply not possible to try to establish confidence in spectroscopically derived parameters , particularly at low metallicities and gravities , by extrapolating confidence gained by means of tests performed in very different regimes , e.g. at solar metallicity .
this paper is the second in the series where we explore the influence of different methods on the determination of basic and fundamental stellar parameters for fgkm stars .
the main goal is to understand to what extent our knowledge about stars , their surface and interior structure , is biased by inferences derived from classical model atmospheres with their simplifying assumptions ( 1d hydrostatic and local thermodynamic equilibrium , lte ) , which have been used as a basis for spectroscopic characterization of stars over the past @xmath135 years . and @xmath27 parameters .
colours depict the stellar surface gravity as indicated in the plots . [
fig : sphere],title="fig:",width=340 ] and @xmath27 parameters .
colours depict the stellar surface gravity as indicated in the plots .
[ fig : sphere],title="fig:",width=340 ] in paper 1 , we developed a novel nlte - opt technique to determine accurate effective temperature , surface gravity , and metallicity from a stellar spectrum and then applied the method to a large sample of stars selected from the rave survey .
the principal components of the method are ( i ) the use of a robust effective temperature scale from the analysis of balmer line wings and , ( ii ) the determination of surface gravities and metallicities by solving physically - realistic nlte radiative transfer in fe lines .
this avoids substantial systematic uncertainties in parameters introduced by the commonly - used approximation of lte .
a comparison of the new nlte - opt stellar parameters to that obtained from the standard method of 1d lte excitation - ionization balance of fe , lte - fe , revealed important differences ranging from @xmath136 k to @xmath5 ( subgiants , rgb ) in @xmath137 , @xmath122 to @xmath14 dex in @xmath0 , and @xmath100 to @xmath15 dex in [ fe / h ] . in this paper , we focus on the determination of evolutionary parameters of stars , masses and ages , and their distances . we develop a robust bayesian technique using state - of - the - art evolutionary tracks computed with the ` garstec ` code @xcite .
the importance of the bayesian scheme in this context has been raised in several earlier studies @xcite .
this method has a principal advantage over standard maximum likelihood ( or minimum @xmath16 ) methods in that it allows a simple way of introducing additional ( prior ) knowledge in the statistical analysis .
test comparisons of results obtained with our bayesian scheme with other similar studies confirm its reliability and accuracy .
we then apply the bayesian method to both sets of stellar parameters , lte - fe and nlte - opt , for the sample of stars from our paper 1 .
the stars were selected from the rave catalogue .
they cover the full parameter space in @xmath138 ( @xmath139 to @xmath140 k ) , @xmath0 ( @xmath139 to @xmath140 ) , and [ fe / h ] ( @xmath141 to @xmath6 ) and sample the local volume within @xmath142 kpc .
spectroscopic distances were computed for all these stars using the 2mass photometry .
masses and ages were determined for the @xmath103 dwarf and subgiant stars ( @xmath112 ) only , the limit set by the degeneracy of stellar tracks of different mass along the rgb in the @xmath143 plane . from these calculations , the following conclusions can be drawn . *
the _ classical method _ of lte excitation - ionization balance of fe , as a rule , underestimates @xmath144 , @xmath0 , and [ fe / h ] of fgkm stars . as demonstrated in paper 1 , for the most metal - poor giants ,
[ fe / h ] @xmath31 , the errors are of the order @xmath145 dex in metallicity .
this has implications for the _ metallicity distribution function _ of stars in different galactic components . for our stellar sample , which is representative of the thick - disk , the metal - poor end of the mdf steepens and
is skewed towards the more metal - rich domain ( by @xmath146[fe / h ] @xmath147 dex ; @xcite ) .
* lte - fe stellar parameters result in overestimated distances compared to the nlte - opt results , and the bias depends on the metallicity and evolutionary stage of a star .
the changes in surface gravity are most critical for metal - poor rgb stars , for which the nlte - opt distances are smaller by up to @xmath148 .
the distances to metal - poor subgiant and turnoff stars are in error by a factor of @xmath149 .
only more metal - rich turn - off stars are barely affected by this bias .
[ fig : sphere ] shows the spatial heliocentric distribution of the stars in our sample determined with @xmath26 ( top panel ) and @xmath27 ( bottom panel ) parameters and makes evident the large impact of using inaccurate stellar parameters for deriving spectrophotometric distances , in particular for giants with @xmath150 .
the overall distribution of distances in our sample steepens towards smaller distances as shown in fig .
[ fig : dist_histo ] . as a consequence ,
significant errors may enter the determination of space velocities and orbital eccentricities , and differences in the population membership of stars ( i.e. thick disk or halo ) are expected .
* comparing our nlte - opt results with the rave dr3 distances @xcite for the stars common in both samples we find a pattern similar to the comparison between nlte - opt and lte - fe results .
this is not surprising : metal lines that dominate the near - ir rave spectra are known to be strongly affected by nlte effects ( sect .
[ sec : ravecomp ] ) and our test calculations with mean 3d models also show that the lines are sensitive to the @xmath151 atmospheric structure differences . as a result ,
more distant stars ( predominantly metal - poor giants ) have larger dr3 distances compared to our values ( fig .
[ fig : rave_vs_us_dist ] ) , by @xmath152 .
+ in this respect , it is very interesting to interpret our results in the light of the findings by @xcite where the authors concluded , based on simple analytical considerations that distances for sdss / segue metal - poor stars are systematically overestimated .
consistent with the arguments presented in that study , our calculations predict that the systematic error in @xmath0 and , thus , distance increases with decreasing metallicity ; [ fe / h ] is affected too , although at a lesser degree .
thus , there will be two effects .
first , more metal - poor stars will appear more luminous and thus more distant , which may be mis - interpreted for a very metal - poor outer stellar population .
this goes in the direction of explaining why this component would appear to be more metal poor than the inner halo component .
second , the systematic error in distance may propagate into the error in transverse velocity , causing the effect of a star on a retrograde orbit .
this strengthens the claim by @xcite that the available observational evidence for a counter - rotating galactic halo component @xcite is weak . *
masses and ages are likewise affected in a systematic way by the differences between nlte - opt and lte - fe , although there is not a one - to - one mapping of the change in mass to the change in age .
in the lte - fe method , the majority of stars appear younger and slightly more massive .
the difference with nlte - opt is of the order @xmath153 m@xmath154 .
the ages are inaccurate by anything between @xmath8 and @xmath105 percent . in this respect
, it will be interesting to re - investigate the _ age - metallicity _ relation in the solar neighborhood , which is usually drawn from stellar samples with spectroscopic metallicities and ages obtained by simple isochrone fitting .
stellar parameters to estimate distances .
[ fig : dist_histo],width=321 ] we emphasize that , contrary to a wide - spread impression , departures from lte and 1d hydrostatic equilibrium will affect low- and high - resolution spectra .
the reason is that these effects change equivalent widths ( or , equivalently , line indices ) , and consequently abundances inferred from a diagnostic feature .
this can not be compensated by decreasing resolution of a spectrum , which merely alters the line profile shape without affecting the total energy absorbed in a line . in summary
, we have shown that the accuracy of the stellar parameters ( @xmath155 , @xmath0 and [ fe / h ] ) is crucial to obtain reliable masses , ages , and spectroscopic distances for late - type stars .
our findings lead to the inevitable conclusion that using the adequate @xmath20 scale and accounting for the deviations from lte and 1d hydrostatic equilibrium in stellar atmospheres is of fundamental importance if we want to achieve an accurate characterization of stellar populations .
until now , most of the work in the community has gone into developing sophisticated statistical methods to extract the largest possible amount of information from large - scale stellar surveys .
although these efforts are undoubtedly valuable , they will not cure the major problem affecting the final distributions of masses , ages , metallicities , distances , which arises because systematic errors in basic stellar parameters dominate .
we thus advise particular caution when applying stellar parameters obtained from 1d lte models in studies of stellar populations .
we thank the anonymous referee for comments that have helped improving the presentation of results .
ams is partially supported by the european union international reintegration grant pirg - ga-2009 - 247732 , the micinn grant aya2011 - 24704 , by the esf eurocores program eurogenesis ( micinn grant eui2009 - 04170 ) , by sgr grants of the generalitat de catalunya and by the eu - feder funds .
grr acknowledges partial support through grants from esf eurogenesis and max planck society for the firststars collaboration .
y. s. , beers t. c. , sivarani t. , allende prieto c. , koesterke l. , wilhelm r. , re fiorentin p. , bailer - jones c. a. l. , norris j. e. , rockosi c. m. , yanny b. , newberg h. j. , covey k. r. , zhang h .- t . , luo a .- l .
, 2008 , aj , 136 , 2022 | for studies of galactic evolution , the accurate characterization of stars in terms of their evolutionary stage and population membership is of fundamental importance .
a standard approach relies on extracting this information from stellar evolution models but requires the effective temperature , surface gravity , and metallicity of a star obtained by independent means . in previous work
, we determined accurate effective temperatures and non - lte @xmath0 and [ fe / h ] ( nlte - opt ) for a large sample of metal - poor stars , @xmath1 [ fe / h ] @xmath2 , selected from the rave survey .
as a continuation of that work , we derive here their masses , ages , and distances using a bayesian scheme and garstec stellar tracks . for comparison , we also use stellar parameters determined from the widely - used 1d lte excitation - ionization balance of fe ( lte - fe ) .
we find that the latter leads to systematically underestimated stellar ages , by 10 - 30% , but overestimated masses and distances .
metal - poor giants suffer from the largest fractional distance biases of @xmath3 .
furthermore , we compare our results with those released by the rave collaboration for the stars in common ( dr3 , @xcite ) .
this reveals @xmath4 to @xmath5 k offsets in effective temperature , @xmath6 to @xmath7 dex offsets in surface gravity , and @xmath8 to @xmath3 in distances .
the systematic trends strongly resemble the correlation we find between the nlte - opt and lte - fe parameters , indicating that the rave dr3 data may be affected by the physical limitations of the 1d lte synthetic spectra .
our results bear on any study , where spectrophotometric distances underlie stellar kinematics .
in particular , they shed new light on the debated controversy about the galactic halo origin raised by the sdss / segue observations .
[ firstpage ] stars : fundamental parameters stars : distances stars : evolution stars : late - type galaxy : kinematics and dynamics |
hydroxylamine ( nh@xmath0oh ) has been suggested as a possible reactant precursor in the formation of interstellar amino acids @xcite .
the presence of amino acids in the gas - phase toward astronomical environments would have a profound impact on the effort to understand the origin of complex molecular material in space .
the recent discovery of the simplest amino acid glycine ( nh@xmath0ch@xmath0cooh ) in cometary samples collected by the stardust mission has provided new clues towards our knowledge of the delivery of prebiotic material to planetesimals @xcite yet not to their initial formation .
that is , are complex prebiotic molecules , such as glycine , formed via reactions of smaller precursors _ after _ their incorporation into cometary bodies , or do these complex molecules form first in the gas phase before accretion ?
the search for interstellar , gas - phase glycine has therefore attracted much attention , but has yet to be unambiguously confirmed in space @xcite .
recently , laboratory experiments have shown that ionized nh@xmath0oh , reacting in the gas phase with acetic acid ( ch@xmath1cooh ) and propanoic acid ( ch@xmath1ch@xmath0cooh ) , can lead to the formation of glycine and the amino acids @xmath2- and @xmath3-alanine ( ch@xmath1ch(nh@xmath0)cooh ) @xcite .
as acetic acid has already been observed in various environments ( shiao et al .
2010 and references therein ) , the detection of nh@xmath0oh would be of much interest to the astrochemistry community , helping to answer the question of how these large complex molecules form in astronomical environments .
protonated hydroxlyamine , nh@xmath1oh@xmath4 and nh@xmath0oh@xmath0@xmath4 , is also fundamentally interesting as a prebiotic molecule , having been shown to be a precursor to amino acid formation @xcite .
as shown in equations [ eq1]-[eq3 ] , protonated hydroxylamine can react with ch@xmath1cooh ( equation [ eq1 ] ) and ch@xmath1ch@xmath0cooh ( equations [ eq2 ] and [ eq3 ] ) to produce protonated glycine and protonated @xmath3- and @xmath2-alanine , respectively .
since ch@xmath1cooh is a well established interstellar molecule , the detection of nh@xmath1oh@xmath4 in the ism would greatly enhance our understanding of the possible formation route to glycine and possibly other simple amino acids in interstellar environments @xcite .
@xmath5 @xmath6 @xmath7 there are few laboratory studies of nh@xmath0oh or protonated nh@xmath0oh formation .
nishi et al . ( 1984 ) proposed a route for synthesis of nh@xmath0oh involving ice mixtures of water and ammonia where a radical recombination reaction ( equation [ eq4 ] ) just above the surface of the ice under irradiated conditions produces nh@xmath0oh . @xmath8 additionally ,
zheng & keiser ( 2010 ) have recently produced nh@xmath0oh through electron irradiation of water - ammonia ices .
they propose that nh@xmath0oh results from the radical recombination of nh@xmath0 and oh inside the ices . the results of both of these studies suggest that radical reactions within ice mantles on grain surfaces may be responsible for nh@xmath0oh production .
in fact , two recent gas - grain chemical models employ such reactions of radicals in their simulations .
charnley et al .
( 2001 ) assumed that nitrogen atoms will first react with oh in the gas - phase to produce large amounts of no ( equation [ eq5 ] ) .
a fraction of no ( @xmath910% ) is then accreted onto dust grains where it can then be converted to form species such as hno and nh@xmath0oh through h addition reactions .
this formation pathway is contingent upon the depletion of no onto dust grains in significant quantities in astronomical environments , though solid evidence for no on grain surfaces is limited .
observational data for the presence of no on grain mantles was first reported in the infrared by allamandola & norman ( 1978 ) via the fundamental rovibrational band at 5.3 @xmath10 m .
more recently , akyilmaz et al . ( 2007 ) have shown that gas - phase no is depleted towards the peak of dust emission in two sources , suggesting that no has accreted onto the grains in these regions .
modeling by garrod et al .
( 2008 ) employs a more expansive network of radical - radical reactions within the ice - mantle , incorporating large radicals formed from photolysis of the ice constituents already known to be present .
these radical fragments " go on to react , building more complex species as they become mobile on the grain surface through a gradual warm - up process before being liberated into the gas phase .
formation of nh@xmath0oh is predicted to start from the radical - radical reaction of nh + oh addition on grain surfaces followed by hydrogenation or directly by the reaction of oh + nh@xmath0 .
the model predicts an nh@xmath0oh column density as high as 10@xmath12 @xmath13 ; easily within the detectable limits of modern radio telescopes .
given the potential importance of nh@xmath0oh to prebiotic chemistry and the high predicted abundances , we conducted a search for nh@xmath0oh in the frequency range of 130 - 170 ghz towards seven sources : irc+10216 , orion kl , orion s , sgr b2(n ) , sgr b2(oh ) , w3irs5 , and w51 m . while these sources are known to contain copious amounts of complex molecular material , no definitive evidence was found for nh@xmath0oh toward any of these sources .
upper limits to the beam averaged column density of nh@xmath0oh were calculated based on the 1@xmath14 rms noise limit of the observed spectra and we discuss possible explanations for the lower than expected abundances .
a 2 mm spectral line survey of irc+10216 , orion kl , orion s , sgr b2(n ) , sgr b2(oh ) , w3irs5 , and w51 m ( hereafter , the turner 2 mm survey ) was conducted using the nrao 12 m telescope on kitt peak by b. e. turner between 1993 and 1995 ] .
table 1 lists the observing parameters for each source in the survey .
column 1 lists the source ; columns 2 and 3 list the reported pointing positions precessed to j2000 coordinates ] ; column 4 reports the observed spectral linewidth in km s@xmath15 and column 5 lists the assumed source local standard of rest velocity ( km s@xmath15 ) for each source . the frequency range covered by this survey was between 130 - 170 ghz and the half - power beam width ( hpbw ) varied from 38 46across the band .
the observations were taken using a dual channel , sis junction single side band receiver with typical receiver noise ranging from 75 - 100 k. the backend consisted of a 768 channel , 600 mhz bandwidth hybrid spectrometer with spectral resolution of 0.781 mhz per channel or @xmath91.3 km / s at 150 ghz .
the intensity scale at the nrao 12 m is given as @xmath16 and corrects for forward spillover loss .
the radiation temperature is defined in equation [ eq6 ] , where @xmath17 is the beam efficiency .
these data were mined for the all the available 2 mm lines of nh@xmath0oh listed in table 2 .
@xmath18 in total , 54 transitions of nh@xmath0oh are reported between 130 and 170 ghz from the published literature @xcite .
of these 54 transitions , 14 were selected , five _ a_-type transitions and nine _ c_-type transitions , in this search for having the largest line strength and lowest upper state energy level .
table 2 is a summary of each searched transition and lists the nh@xmath0oh transition rest frequency ( column 1 ) , the transition quantum numbers ( column 2 ) , @xmath19 is the telescope beam size at the observed frequency ( column 3 ) , @xmath20 is the upper state energy ( column 4 ) , the transition type ( column 5 ) , log@xmath21(@xmath22 ) is the einstein a coefficient ( column 6 ) and g@xmath23 is the @xmath24 degeneracy of the upper state ( column 7 ) .
other relevant spectroscopic parameters such as the nh@xmath0oh dipole moments , partition function and rotational constants are listed in the notes of table 2 .
figures [ atype]-[ctype2 ] show the observed spectra ( black trace ) for each source in the frequency range of the nh@xmath0oh target transitions . shown in red
is a simulated spectrum of the expected transition line strengths from nh@xmath0oh using the total column density predicted for sgr b2(n ) from the garrod et al .
( 2008 ) model and source - appropriate rotational temperatures and line widths ( equation 4 ) .
the simulated spectra in red has been scaled down by a factor of 100 for clarity . for illustrative purposes ,
an unscaled simulation is shown in blue for sgr b2(n ) in figure 1 .
while the total column density of nh@xmath0oh in sources other than sgr b2(n ) can be expected to vary based on chemical composition and physical environment , the simulations serve to show in a qualitative sense that the searched for transitions of nh@xmath0oh are not present toward these sources beyond the 1@xmath14 rms noise limit and certainly not present in the abundances predicted by the model . in several sources ,
emission features are present at a number of the appropriate center frequencies for nh@xmath0oh .
yet , the strongest transitions of nh@xmath0oh in this band are the @xmath24 = 3 - 2 manifold near 151 ghz and in no source are all of the expected transitions observed .
this indicates that the observed emission features are coincidental overlap with other molecular transitions and nh@xmath0oh is not observed in these sources . using equation [ eq7 ]
, upper limits to the beam averaged column density based on the 1@xmath14 rms noise limit were calculated and are reported in table [ cds ] . an approximate rotational temperature appropriate for each source
was used based on data available in the references shown in table [ cds ] . for the purposes of this work
, ch@xmath1oh was used as a primary source of temperature information when available .
ch@xmath1oh was chosen as it is a well known temperature probe and traces the molecular gas of the desired regions .
additionally , the wealth of observational data on ch@xmath1oh establishes confidence in temperatures derived from its observations .
it should be noted that the upper limits presented here are fairly insensitive to the relatively minor range of rotational temperatures observed in these sources .
fractional abundances with respect to molecular hydrogen were calculated for each source based on these upper limits . in the case of sgr b2(n ) , the garrod et al .
( 2008 ) model predicts a relative abundance for nh@xmath0oh of 3.5 x 10@xmath25 - 4.2 x 10@xmath26 , which is up to six orders of magnitude higher than the observed upper limits ( see table 3 ) .
in this paper we reported on the negative detection of hydroxylamine ( nh@xmath0oh ) towards several astronomical sources . upper limits to the beam averaged column density have also been determined for each source based on the 1@xmath14 rms noise level in each spectra .
recent chemical models introduced a new gas - grain chemical network utilizing radical - radical reactions as formation mechanisms ( garrod et al .
the model reproduces the beam averaged column densities of species such as methanol ( ch@xmath1oh ) , acetaldehyde ( ch@xmath1cho ) , and even glycolaldehyde ( ch@xmath0(oh)cho ) with excellent agreement with current observed abundances toward the sgr b2(n ) star - forming region @xcite .
however , for nh@xmath0oh , the predicted abundances are 3.5@xmath2810@xmath25 - 4.2@xmath2810@xmath26 , nearly six orders of magnitude higher than the observed upper limit of 8@xmath2810@xmath29 towards sgr b2(n ) reported in this study .
the following sections discuss possible explanations for this surprising difference , focusing on possible formation and destruction mechanisms .
the formation of nh@xmath0oh within ice grains was first proposed by nishi et al .
( 1984 ) as shown in equation [ eq1 ] .
previous experimental attempts to produce nh@xmath0oh within the gas phase through the reaction of hno@xmath4 with h@xmath0 have failed @xcite .
the garrod et al .
( 2008 ) grain chemistry model assumes two formation mechanisms for nh@xmath0oh , both of which assume radical - radical reactions within the grain mantle . in early times
, nh@xmath0oh is formed through the barrierless ( see figure 1 of garrod et al .
2008 ) reactions of the hydroxyl radical ( -oh ) with nh followed by hydrogenation ( equations [ eq8 ] and [ eq9 ] ) similar to the well studied hydrogenation reactions of co forming ch@xmath1oh @xcite . @xmath30 @xmath31 however , there is no current theoretical or experimental work to suggest this hydrogenation reactionleading to the final product of hydroxlyamine in equation ( 9 ) proceeds in a manner similar to co as the model assumes . in fact , given the different states of these molecules , co having a @xmath32 electronic configuration while no is a @xmath33 , it is likely that the hydrogenation of no will proceed quite differently from that of co. this difference could account for the higher abundances predicted by the garrod et al .
( 2008 ) model at lower temperatures .
it is also worth bearing in mind that the branching ratios found in the gas phase are not applicable to solid state processes as the surrounding ice can carry away the energy fairly efficiently .
as warming takes place and the hydroxyl radical becomes more mobile on the surface of the grain , the model predicts the barrierless reaction with nh@xmath0 to become dominant ( equation [ eq10 ] ) .
@xmath34 however , experimental studies have shown @xcite that in an isolated argon matrix , nh@xmath0 quickly combines with a free hydrogen radical to form nh@xmath1 when a temperature of 20 k was reached .
the question then becomes whether nh@xmath0 and oh have a higher probability to react to form nh@xmath0oh in interstellar ices before they recombine with free hydrogen to form nh@xmath1 and h@xmath0o , respectively .
interstellar ices are considerably more complex than the isolated matrices used in the schnepp & dressler ( 1960 ) laboratory study , as are the ices considered in the garrod et al .
( 2008 ) model which contain a number of other simple species ( e.g. ch@xmath35 , ch@xmath1oh , nh@xmath1 , co , co@xmath0 , hcooh , h@xmath0o ) . as such , nh@xmath0 and
oh are not the only species present to react with free hydrogen which may instead react with itself ( to form h@xmath0 ) or with other smaller species .
an examination of the rates of reaction of free hydrogen with these species might therefore be in order to help determine if nh@xmath0 and oh would be available in sufficient quantities to collide and form enough nh@xmath0oh to be detectable in the ism .
however , the diffusion and evaporation rates of atomic h and other radicals are of more importance .
radical - radical reactions will proceed at the diffusion rates of the reactants and evaporation dominates the loss channels of atomic hydrogen under almost all temperature regimes , which therefore makes it the most important route for determining overall hydrogen populations . unfortunately , these diffusion rates of reactants are difficult to ascertain experimentally . there is experimental evidence to support the radical - radical formation of nh@xmath0oh from nh@xmath0 and oh precursors in electron - irradiated ammonia - water ice samples @xcite . upon irradiation
, an nh@xmath1 species is found to undergo unimolecular decomposition to form the nh@xmath0 radical and a free hydrogen atom ( equation 11 ) .
water decomposes in a similar fashion , forming oh and h. @xmath36 after irradiation , a new absorption peak at @xmath91500 cm@xmath15 was observed and attributed to nh@xmath0 formation .
as the ice samples were warmed , the species released in the gas phase were monitored by ir and mass spectroscopy .
the presence of nh@xmath0oh was first noted in the ir measurements as the sample reached 174 k. as the temperature continued to rise , the abundance of nh@xmath0oh decreased until non - detection at 200 k. nh@xmath0oh was also observed in the mass spectroscopy measurements from 160 - 180 k. it is important to note that nh@xmath0oh is observed at temperatures above which most , if not all , of the water and ammonia are sublimed .
while this study does support the formation of nh@xmath0oh through radical - radical recombination within interstellar ices , it also provides potential evidence as to why nh@xmath0oh is not currently observed in the ism .
is there a possible temperature problem ?
the garrod et al .
( 2008 ) model predicts the high abundance of nh@xmath0oh with a temperature on the order of @xmath9130 k. zheng & keiser ( 2010 ) find that water , ammonia and hydroxlyamine co - desorb , while some nh@xmath0oh remains in the ice samples to somewhat higher temperatures . according to the experimental data
, nh@xmath0oh was not observed in the gas phase until temperatures exceeded 160 k. however , temperature programmed laboratory desorption experiments demonstrate behavior that is strongly dependent on the heating rate and this experimentally determined evaporation temperature does not correspond to the actual interstellar value . without running futher analyses of the laboratory data , such as fitting the data to a polanyl - wigner type expression ( see galway & brown 1999 and references therein ) or running chemical models , making a qualitative comparison between evaportation temperatures from the lab and in the interstellar medium is difficult . also
, strong water absorption bands obscure nh@xmath0oh absorption features , making its detection using infrared observations towards hot core regions impossible .
however , it is possible the nh@xmath0oh emission is confined to very compact ( @xmath385@xmath39 ) hot core regions such as the sgrb2(n - lmh ) where the temperatures are higher and hydroxlyamine would be completely released from the grain surfaces @xcite .
single dish observations from this survey would be too beam diluted to detect the emission from this compact region .
such an occurence has been noted before .
acetic acid ( ch@xmath1cooh ) has been confirmed in several hot core regions using interfermetric observations after searches with single dish observations resulted in negative detections @xcite . additionally , many of the single dish spectra presented here are very dense with molecular emission .
interferometric images have less line confusion than that of single dish .
many of features in single dish spectra arise from an extended region or from different locations within the source and this is demonstrated with the detection of amino acetonitrile by belloche et al . (
the single dish spectra of amino acetonitrile were heavily contaminated with other complex molecular species with only 88 of the 398 observed transitions being reported as relatively `` clean '' .
interferometric observations of sgr b2(n ) confirmed the compact emission of amino acetonitrile with a source size of 2 " @xcite .
as such , higher spatial resolution interferometric observations may be needed to more thoroughly couple to the higher temperature regions in order to detect hydroxlyamine .
alternatively , observations could be conducted towards molecular sources in shocked regions such as the bipolar outflow l1157(b ) . in these types of sources , molecules which are formed on grain surfaces but which are not liberated into the gas phase by thermal desorption due to low temperatures
are instead ejected into the gas phase by shocks @xcite .
detection of nh@xmath0oh in these sources would provide valuable insight into the mechanisms behind its formation pathways and eventual release into the gas phase .
next we examine possible pathways for the destruction of nh@xmath0oh once it enters the gas phase .
it is well known that ion - molecule reactions are important in gas phase interstellar chemistry and that protonated species play an important role in reaction mechanisms .
nh@xmath0oh , having a high proton affinity ( @xmath9193.5 kcal mol@xmath15 ) is particularly susceptible to protonation from other species such as h@xmath40 , hco@xmath4 , ch@xmath41 , h@xmath1o@xmath4 and ch@xmath1oh@xmath42 ( blagojevic et al .
2003 and references therein ) .
the energies of protonation of nh@xmath0oh by h@xmath4 and the possibility of proton transfer by h@xmath40 have been predicted by theory @xcite . as a result of protonation , two stable species
were reported : nh@xmath1oh@xmath4 and nh@xmath0oh@xmath42 , with nh@xmath1oh@xmath4 found to be more stable by @xmath9100 kj mol@xmath15 .
the reaction of nh@xmath0oh with either h@xmath4 or protonated methanol ( ch@xmath1oh@xmath42 ) was predicted to be very exothermic and it was proposed that the excess energy would either dissociate the species or could lead to the rearrangement of the species to the higher energy nh@xmath0oh@xmath42 .
this could result in an enhanced abundance of nh@xmath0oh@xmath42 in the ism .
once in the gas phase , recent theoretical work has shown that the reaction of ionized and protonated nh@xmath0oh with h@xmath0 , its most likley collision partner , is highly unfavorable @xcite .
these species , therefore , are likely to remain as reaction partners for further chemistry .
given these considerations , even if nh@xmath0oh is produced on ice grains through radical - radical reactions , upon its release into the gas phase it may quickly undergo protonation .
this would result in very low observed abundances of nh@xmath0oh in the ism .
once protonated , the reaction with h@xmath0 , by far the most likely collision partner , is highly unfavorable and the lifetimes of these species should therefore be greatly enhanced .
a search for nh@xmath0oh@xmath42 and nh@xmath1oh@xmath4 within these star forming regions might be prudent ; although , dissociative recombination reactions could result in lowered abundances of these species .
this would first require the acquisition of the rotational spectra of these species in the laboratory to enable astronomical searches .
we report the non - detection of nh@xmath0oh towards seven sources . calculated upper limits for the abundance of this molecule are as much as six orders of magnitude lower than those predicted for the species by recent models .
several factors could account for this discrepancy including the rapid removal of precursor molecules from ice mantles through reaction with free hydrogen or the rapid protonation ( and subsequent dissociative recombination ) of nh@xmath0oh by h@xmath4 , h@xmath40 , ch@xmath41 , and other efficient protonation mechanisms .
the single dish observations presented here are likely highly beam diluted .
higher resolution interferometric observations could provide the sensitivity required for detection and therefore allow better refinement of models which currently predict the presence of nh@xmath0oh in high abundance .
bam gratefully acknowledges g.a .
blake for his support , as well as funding by an nsf graduate research fellowship .
we also would like to thank the anonymous reviewer for the valuable comments and suggestions to improve the quality of the paper .
l r r c c orion kl & 05 35 14.5 & -05 22 32.6 & 4.5 & @xmath43 + orion s & 05 35 16.5 & -05 19 26.7 & 4.5 & @xmath44 + irc+10216 & 09 47 57.3 & + 13 16 43.0 & 9.0 & @xmath45 + sgr b2(oh ) & 17 47 20.8 & -28 23 32.2 & 9.0 & @xmath46 + sgr b2(n ) & 17 44 11.0 & -28 22 17.3 & 9.0 & @xmath47 + w51 m & 19 16 43.8 & + 14 30 07.5 & 9.0 & @xmath48 + w3irs5 & 02 27 04.1 & + 61 52 21.4 & 9.0 & @xmath49 + [ sources ] c c c c c c c 151020.70 & 3(1,3)-2(1,2 ) & 41.52 & 10.57 & a & -5.27587 & 7 + 151101.99 & 3(2,2)-2(2,1 ) & 41.50 & 27.16 & a & -5.47928 & 7 + 151102.32 & 3(2,1)-2(2,0 ) & 41.50 & 27.16 & a & -5.47928 & 7 + 151117.67 & 3(0,3)-2(0,2 ) & 41.49 & 5.04 & a & -5.22390 & 7 + 151207.01 & 3(1,2)-2(1,1 ) & 41.47 & 10.57 & a & -5.27422 & 7 + 164340.78 & 9(1,9)-9(0,9 ) & 38.15 & 75.59 & c & -7.02982 & 19 + 164627.49 & 8(1,8)-8(0,8 ) & 38.09 & 60.48 & c & -7.02783 & 17 + 164883.24 & 7(1,7)-7(0,7 ) & 38.03 & 47.04 & c & -7.02617 & 15 + 165107.71 & 6(1,6)-6(0,6 ) & 37.98 & 35.28 & c & -7.02472 & 13 + 165300.63 & 5(1,5)-5(0,5 ) & 37.93 & 25.2 & c & -7.02335 & 11 + 165461.76 & 4(1,4)-4(0,4 ) & 37.89 & 16.8 & c & -7.02237 & 9 + 165590.89 & 3(1,3)-3(0,3 ) & 37.87 & 10.08 & c & -7.02148 & 7 + 165687.89 & 2(1,2)-2(0,2 ) & 37.84 & 5.04 & c & -7.02079 & 5 + 165752.62 & 1(1,1)-1(0,1 ) & 37.83 & 1.68 & c & -7.02037 & 3 + [ transitions ] l c l l l c orion kl@xmath50 & 6.4 & @xmath382 x 10@xmath51 & 7.0 x 10@xmath52 & @xmath383 x 10@xmath53 & 120 + orion s@xmath54 & 4.0 & @xmath389 x 10@xmath55 & 1.0 x 10@xmath52 & @xmath389 x 10@xmath53 & 80 + irc+10216@xmath56 & 5.1 & @xmath388 x 10@xmath51 & 3.0 x 10@xmath57 & @xmath383 x 10@xmath58 & 200 + sgr b2(oh)@xmath59 & 7.0 & @xmath383 x 10@xmath51 & 1.0 x 10@xmath60 & @xmath383 x 10@xmath53 & 70 + sgr b2(n)@xmath61 & 6.6 & @xmath382 x 10@xmath51 & 3.0
x 10@xmath60 & @xmath388 x 10@xmath29 & 70 + w51m@xmath62 & 7.7 & @xmath384 x 10@xmath51 & 1.0 x 10@xmath60 & @xmath384 x 10@xmath53 & 100 + w3irs5@xmath63 & 4.3 & @xmath382 x 10@xmath51 & 5.0 x 10@xmath52 & @xmath383 x 10@xmath53 & 70 + [ cds ] | observations of 14 rotational transitions of hydroxylamine ( nh@xmath0oh ) using the nrao 12 m telescope on kitt peak are reported towards irc+10216 , orion kl , orion s , sgr b2(n ) , sgr b2(oh ) , w3irs5 , and w51 m .
although recent models suggest the presence of nh@xmath0oh in high abundance , these observations resulted in non - detection .
upper limits are calculated to be as much as six orders of magnitude lower than predicted by models .
possible explanations for the lower than expected abundance are explored . |
the distribution of brightness over the solar photosphere is not uniform , due to the presence of dark sunspots and bright faculae in active regions and to the granular convection and bright network elements , associated with slender flux tubes located in the intergranular lanes .
the baricentre of the brightness distribution over the solar disk , i.e. , the photocentre of the sun , is affected by the brightness inhomogeneities having the largest scales , that is by sunspots and faculae , because the averaged effect of the granulation and the network elements is negligible , thanks to their almost uniform distribution on spatial scales comparable with the solar radius .
the rotation of the sun and the evolution of the active regions make the perturbation of the position of the photocentre a function of the time .
sunspots and faculae are manifestations of the magnetic field , amplified and modulated by the solar dynamo , and are present also on the photospheres of late - type stars hosting hydromagnetic dynamos in their convective envelopes . therefore , the brightness inhomogeneities observed in the sun are also present on late - type stars during their main - sequence lifetime and are most prominent in young , rapidly rotating stars , given their more powerful dynamo action ( e.g. * ? ? ? * ) . the perturbation of the position of the photocentre of late - type stars due to their magnetic activity should be considered in some detail , in view of the forthcoming space missions gaia and sim .
they will obtain astrometric measurements of the position of late - type stars with @xmath0 with an accuracy between @xmath1 and @xmath2 @xmath3as ( micro arc - seconds ) ; see , e.g. , @xcite .
moreover , ground - based interferometric instruments , such as prima at vlti , will also reach an accuracy of @xmath4 @xmath3as in the determination of the photocentre position .
the feasibility of such a ground - based high - precision differential astrometry has been recently demonstrated by the phases ( palomar high - precision astrometric search for planets ) program ( see * ? ? ?
. in the present paper , a simple model for computing the astrometric effects of cool spots and bright solar - like faculae in a late - type star is introduced , and applied to derive the maximum expected amplitude of variation of the photocentre position as a function of the active region filling factor .
since the perturbation is cyclically modulated by the stellar rotation , it can be a source of spurious detections of close - by planetary companions by means of the astrometric technique .
we present some cases leading to possible confusion and propose methods to discriminate the effects of a planet from those due to stellar activity .
let us consider a cartesian reference frame with the @xmath5-@xmath6 plane in the plane of the sky , that is the origin o at the unperturbed photocentre of the star , the @xmath5-axis along the declination direction pointing to the north pole , the @xmath6-axis along the direction of the increasing right ascension and the @xmath7-axis pointing toward the observer .
the position @xmath8 of the photocentre at a given time @xmath9 depends on the distribution of the brightness on the surface of the apparent disk of the star at the isophotal wavelength @xmath10 of the observations , and is given by : @xmath11 where @xmath12 is the specific intensity at wavelength @xmath10 at the given position @xmath13 on the stellar disk , and the integration is extended over the apparent stellar disk @xmath14 .
equations ( [ pc_def ] ) define the photocentre as the baricentre of the flux distribution emerging from the stellar disk . for an unperturbed star , such a distribution is symmetric around the disk centre and the photocentre coincides with the projection of the baricentre of the star on the plane of the sky . in order to describe the distribution of the surface brightness when there are photospheric inhomogeneities , it is more convenient to adopt a spherical reference frame rotating with the star . specifically , a spherical reference frame is considered with the origin at the baricentre o@xmath15 of the star and the @xmath16-axis directed along its rotation axis , rotating with the stellar angular velocity @xmath17 with respect to an inertial reference frame .
we can express the position of a generic point on the stellar surface by means of its cartesian co - ordinates in that frame : @xmath18 where @xmath19 is the radius of the star , assumed to be spherically symmetric , and @xmath20 and @xmath21 are the colatitude and the longitude of the point on the stellar surface , respectively .
the transformation from the latter reference frame to the reference frame of the stellar disk can be achieved by means of three successive eulerian rotations . in order to define them ,
let us consider that the stellar equatorial plane intersects the plane of the sky along a line of nodes that makes an angle @xmath22 with respect to the direction of the @xmath5-axis on the plane of the sky .
the first rotation by an angle @xmath23 around the @xmath24-axis makes the fundamental stellar meridian coincide with its position at the time @xmath25 when it crosses the line of nodes .
the second rotation by the inclination @xmath26 of the stellar rotation axis , around the line of nodes , makes the @xmath24-axis coincide with the @xmath7 axis .
finally , a rotation by an angle @xmath22 , around the @xmath7 axis , reports the reference frame into that of the plane of the sky . in matrix form
, we can write the co - ordinate transformation as follows : @xmath27 = \tilde{a } \left [ \begin{array}{l } x_{0 } \\
z_{0 } \\ \end{array } \right ] , \label{transf}\ ] ] where @xmath28 is the transformation matrix that is given by : @xmath29 where the individual factor matrixes express the three rotations as specified above , respectively .
note that the angle @xmath30 is a linear function of the time , whereas @xmath26 and @xmath22 are constant .
we subdivide the photosphere of the star into @xmath31 elements , each of which is identified by the longitude @xmath21 and colatitude @xmath20 of its central point .
the status of the photosphere in each element can be either unperturbed , spotted , or of facular type .
the specific intensity of the unperturbed photosphere is assumed to be given by a linear limb - darkening law : @xmath32 where @xmath33 is the specific intensity at the centre of the disk , @xmath34 the linear limb - darkening coefficient at the isophotal wavelength @xmath10 , and @xmath35 , where @xmath36 is the angle between the normal to the given surface element and the line of sight .
it is given by : @xmath37 + \cos
\theta \cos i.\ ] ] the specific intensity of the spotted photosphere is given by : @xmath38 , where the spot contrast @xmath39 is a function of the wavelength @xmath10 and is assumed to be independent of @xmath3 ( cf .
the specific intensity of the facular photosphere is assumed to be equal to that of the unperturbed photosphere at the disk centre , while its contrast increases linearly toward the limb , i.e. , @xmath40 $ ] , with the constrast coefficient @xmath41 being a function of the wavelength ( see * ? ? ? * for details ) .
the value of the flux coming from the @xmath26-th surface element at the wavelength @xmath10 and the time @xmath42 is given by : @xmath43 \delta a_{i } \mu(t_{k } ) g[\mu(t_{k } ) ] & \mbox { for the unperturbed photosphere , } \\ i_{\rm s}[\lambda , \mu(t_{k } ) ] \delta a_{i } \mu(t_{k } ) g[\mu(t_{k } ) ] & \mbox { for the spotted photosphere , } \\ i_{\rm f}[\lambda , \mu(t_{k } ) ] \delta a_{i } \mu(t_{k } ) g[\mu(t_{k } ) ] & \mbox { for the facular photosphere,}\\ \end{array } \right.\ ] ] where @xmath44 is the visibility function of the surface element , i.e. , @xmath45 for @xmath46 , @xmath47 for @xmath48 .
the total flux from the stellar disk at time @xmath42 is : @xmath49 . the co - ordinates of the photocentre on the plane of the sky at time @xmath42 are given by : @xmath50 where the co - ordinates @xmath51 and @xmath52 depend on the spherical co - ordinates @xmath53 of the @xmath26-th surface element and the geometrical parameters of the star through eqs .
( [ solidal_coo ] ) and ( [ transf ] ) . from eq .
( [ ph_res ] ) , we see that @xmath54 and @xmath55 depend on the isophotal wavelength of the observations .
it can be assumed to be fixed for a given astrometric instrument , hence we shall not consider explicitly that dependence .
other effects affecting the position of the photocentre , e.g. , the oscillations due to a planet or a planetary system ( e.g. * ? ? ?
* ) , or the proper motion of the star , can be linearly added up to the effects of the surface brightness inhomogeneities computed with the scheme introduced above .
we have applied the model described in sect . [ model ] to compute the apparent motion of the photocentre of a star with a single active region on its surface , containing only cool spots and having a longitudinal extension of @xmath56 between latitudes @xmath56 and @xmath57 in the northern hemisphere .
the active region has an area of 0.83% of the whole stellar surface , i.e. , about @xmath58 times larger than the area of the largest sunspot groups at the maximum of the eleven - year cycle .
however , we believe this is more appropriate for late - type stars that are slightly more active than the sun ( see sect .
[ application ] . for the sake of simplicity , our model spotted region
is assumed to be uniform in contrast and squared in shape ( see fig .
[ figar ] ) .
however , as we shall see below , the maximum photocentre excursion depends primarily on the total area of the active region with its shape playing a secondary role . the limb - darkening coefficient of the unperturbed photosphere is assumed to be @xmath59 at the given isophotal wavelength and the spot contrast @xmath60 , which is appropriate to describe the bolometric flux perturbation in the case of sunspots , corresponding to an effective temperature of the spotted photosphere of @xmath61 k
, i.e. , @xmath62 k below that of the unperturbed photosphere , which accounts , on average , for the combined effects of umbrae and penumbrae ( e.g. , * ? ? ?
* ) . in fig .
[ phot_ex ] , we plot the position of the photocentre versus the rotational phase for different values of the inclination @xmath26 of the stellar rotation axis along the line of sight .
the photocentre excursion is measured in unit of the apparent radius of the stellar disk @xmath63 along the direction of the @xmath5-axis , adopting a position angle @xmath64 in our calculations .
the motion along the @xmath6-axis is similar , but shifted in phase by an amount that depends on the co - ordinates of the active region and the stellar parameters @xmath26 and @xmath22 .
the active region is assumed to be stable , so that stellar rotation is the only effect responsible for the motion of the photocentre .
the variations of the relative flux corresponding to the simulations shown in fig .
[ phot_ex ] are plotted in fig .
[ flux_var ] . when the inclination is lower than @xmath56 , the active region is always in view and produces a continuous modulation of the position of the photocentre and of the flux that depends on the variation of its position on the stellar disk and its projected area . for a pole - on view ,
the projected area is constant and the variation of the photocentre abscissa is purely sinusoidal , whereas the flux stays constant . when the inclination increases , the modulation of the photocentre abscissa is no longer sinusoidal .
when @xmath65 , there are intervals of phase during which the photocentre is unperturbed , i.e. , when the active region transits on the invisible hemisphere of the star .
considering the case with @xmath66 , we note that , when the active region first appears on the rising part of the disk , it produces a negative shift of the photocentre abscissa because the baricentre of the flux distribution falls in the opposite half of the disk
. then the shift reverses its sign , when the active region is carried by stellar rotation on the other half of the disk , and eventually it becomes zero when the active region comes out of view . for @xmath67 ,
the duration of the active region transit across the disk depends on its average latitude and the inclination .
moreover , the sign and the value of the photocentre excursion depend also on the position angle of the projection of the stellar rotation axis on the plane of the sky , as measured by the angle @xmath22 .
the combination of these different factors explains the different shapes and amplitudes of the curves plotted in fig .
[ phot_ex ] .
the flux variation produced by the active region increases with increasing inclination and reaches its maximum value when the active region transits through the centre of the stellar disk , i.e. , for @xmath66 , because in that case its projected area reaches its maximum value .
note that the excursion of the photocentre and the flux variations are correlated for @xmath68 , as shown in fig .
[ correl ] . when only a single active region is present on the stellar disk , the excursion of the photocentre reaches its maximum value for a given value of the total active region area . as a matter of fact ,
if the same total area is subdivided into two or more active regions , the excursion generally decreases because the simultaneous presence of more than one brightness inhomogeneity leads to a reduction of their combined effects .
therefore , we have investigated the variation of the maximum photocentre excursion @xmath69 as a function of the filling factor @xmath70 for a single active region by means of our model ( see fig . [ excur_l ] ) .
the stellar inclination is assumed fixed at @xmath66 , that is the most probable value for a random orientation of the stellar rotation axis .
the average latitude of the active region is @xmath71 to maximize its effect on the position of the photocentre . in those simulations ,
we consider also the effect of a facular component that surrounds the dark spot .
the area of the facular component is assumed to be ten times that of the cool spots , as suggested by solar observations and modelling of the solar irradiance variations ( see , e.g. , * ? ? ?
* ; * ? ? ?
* and references therein ) .
the spot constrast is assumed to be @xmath72 , while the facular contrast is @xmath73 ( see , e.g. , * ? ? ?
note that the effect of the faculae is small , in spite of their relative area being ten times that of cool spots .
this is due to the fact that their effect is significant only close to the limb , because their contrast is negligible close to the centre of the disk .
moreover , the projection effect disfavours the facular contribution , because their projected area becomes small close to the limb .
the situation changes if we consider hot spots instead of solar - like faculae , that is bright regions the constrast of which is significant also close to the disk centre , as is sometimes observed in the case of very active stars ( cf . ,
e.g. , the case of the active component of the rs cvn binary system hr 1099 reported by * ? ? ?
however , since we are here mainly interested in solar - like stars , we shall not consider further this possibility .
the variation of the amplitude of the rotational modulation of the stellar flux versus the filling factor is plotted in fig .
[ flux_l ] .
the correlations shown in figs . [ excur_l ] and [ flux_l ] can be well approximated by power laws up to a value of the filling factor @xmath74 , beyond which they deviate from such a simple dependence because the amplitude of the variation saturates when a large fraction of the stellar disk is covered with active regions . applying a linear best fit to the plots in figs .
[ excur_l ] and [ flux_l ] , we find the following regression expressions , valid for @xmath75 , with a correlation coefficient @xmath76 : @xmath77 and @xmath78 where the reported uncertainties correspond to one standard deviation .
the first of eqs .
( [ power_law ] ) confirms the results previously obtained by @xcite in the case of an active region consisting of dark spots only .
@xcite also estimated upper limits for the astrometric effect of dark spots assuming a simplified model with a fixed projected spot geometry and neglecting limb - darkening effects .
their model confirms the almost linear correlation between the variation of the stellar flux and the photocentre excursion that can be deduced combining together eqs .
( [ power_law ] ) and ( [ flux_law ] ) .
however , they predict an astrometric effect about 60% larger than ours for a given photometric amplitude , essentially because of the neglection of the foreshortening and limb - darkening effects .
note that the simultaneous presence of spots and faculae decreases the photocentre excursion in our model and makes the regression more steep , especially for @xmath79 .
this is due to the opposite effects of bright and dark regions spatially associated on the stellar disk . for a filling factor larger than 0.03
, the effect of the faculae becomes negligible because of their low contrast close to the centre of the disk that leads to a saturation of their contribution .
equations ( [ power_law ] ) and ( [ flux_law ] ) can be used to estimate the maximum relative photocentre excursion and flux variation , respectively , expected for a given spot coverage in a solar - like star .
this is useful to evaluate the possibility that stellar activity may account for an observed shift of the photocentre in a late - type star .
the effects of stellar magnetic activity on the position of the photocentre can be principally observed in nearby solar - like stars , because the maximum excursion can not exceed the apparent stellar radius . considering the one hundred nearest stars listed in the catalogue of @xcite ,
those with spectral type between f0v and m5v and apparent magnitude @xmath80 have been selected for a quantitative study .
they are within @xmath81 pc from the sun .
their radii have been estimated from their @xmath82 color indexes according to @xcite , whereas their distances are taken from the hipparcos catalogue .
seven stars turn out to have an apparent radius @xmath63 larger than 1000 @xmath3as and are listed in table [ table1 ] where the columns from the left to the right give the number in the hipparcos catalogue , the name , the spectral type , the apparent visual magnitude , the radius , the distance ( with its standard deviation ) , and the apparent radius of each star , respectively . moreover , 13 stars turn out to have an apparent radius between @xmath83 and @xmath84 @xmath3as , while 46 have an apparent radius @xmath85 @xmath3as .
.the seven stars with an apparent radius @xmath63 larger than 1000 @xmath3as and spectral type later than f5 , listed in order of decreasing @xmath63 . [ cols=">,^,^,>,^,^,^ " , ] the results obtained in sect . [ results ] show that for a filling factor of 0.2% , corresponding to the sun at the maximum of the eleven - year cycle , excursions as large as @xmath86 @xmath3as can be expected for @xmath87 cen a in the case of an active region consisting only of cool spots . in the case of an active region containing also solar - like faculae , those values are reduced by a factor of @xmath88 because the faculae tend to decrease the effect of the spots at such a low value of the filling factor .
a solar value of the filling factor is likely to be appropriate for @xmath87 cen a because the star appears to have a level of activity comparable or slightly lower than the sun , with a rotation period of about 29 days ( see , e.g. , * ? ? ?
note that the above estimates refer to the case of a single active region , thus the excursion can be reduced if a few active regions are simultaneously present on the stellar disk . in any case
, those amplitudes will be easily detectable by astrometric space - borne experiments like gaia and sim , and could be within the reach of ground - based interferometers , such as prima .
note that a periodic signal can be detected with high confidence in a sufficiently long time series , even if the errors on the single measurements are larger than its amplitude ( see , e.g. , * ? ? ?
* ) . in the sun ,
the average lifetime of the sunspot groups dominating the flux modulation is @xmath89 days ( cf .
, e.g. , * ? ? ?
* ; * ? ? ?
* ) , but they have a remarkable tendency to form in sequence within activity complexes , the lifetime of which is @xmath90 months ( e.g. , * ? ? ?
. therefore , if an activity complex on @xmath87 cen a is stable for a time interval longer than the rotation period , it can produce a cyclic modulation of the position of the photocentre with a semi - amplitude up to @xmath86 @xmath3as and a period of 29 days that can be erroneously attributed to the reflex motion induced by a planet of @xmath91 earth masses on an orbit of semi - major axis of @xmath92 au , assuming that the star has the same mass of the sun .
such a spurious detection is difficult to disprove by correlating the photocentre motion with the variation of the stellar optical flux , as measured by ground - based photometry , because the expected amplitude of variation is only @xmath93 mag ( cf .
[ flux_l ] and eq .
[ flux_law ] ) . on the average
, the level of activity is expected to be higher than solar for the stars of spectral types later than the sun listed in table [ table1 ] , because their convection zones are deeper and their rotation periods are comparable or shorter than the solar one ( cf . , e.g. , * ? ? ?
the expectation is confirmed in the case of @xmath94 eri , for which spot modelling of most photometry indicates a filling factor of at least 0.7% @xcite .
this implies a maximum excursion of @xmath95 @xmath3as for the photocentre position with the rotation period of the star , i.e. , @xmath96 days .
moreover , the observations indicate that the signal would maintain its coherence for several rotations because the timescale of spot evolution is longer than 35 days .
the observed amplitude of the optical flux variation of @xmath94 eri is only @xmath97 mag , that is at the detection limit of ground - based photometry .
it is interesting to note that @xmath94 eri has a planetary companion on a wide orbit with a period of 6.8 years ( e.g. , * ? ? ?
the astrometric orbit has a semi - major axis of @xmath98 mas ( milli arcseconds ) , that is much larger than the astrometric noise expected to be produced by stellar activity .
the inclination of the planetary orbit is @xmath99 . assuming that the stellar spin and the orbital angular momentum are aligned
, we can adopt the same inclination for the stellar rotation axis .
this implies that all the active regions located at latitudes higher than @xmath71 are circumpolar and produce a continuous modulation of the photocentre position with the stellar rotation period
. they can give rise to a spurious detection of further planetary companions around @xmath94 eri .
specifically , assuming a stellar mass of @xmath100 m@xmath101 , a photocentre oscillation with a period of @xmath102 days and an amplitude of @xmath103 @xmath3as can be attributed to a planet of @xmath104 jupiter masses on an orbit of semi - major axis @xmath105 au .
late - type stars with rotation periods of a few days have spot filling factors of the order of @xmath106 ( cf . , e. g. , * ? ? ?
they can produce astrometric signatures with amplitudes up to 0.1 stellar radii , i.e. , up to @xmath107 @xmath3as in the group of 66 nearby stars considered above .
if we extend our sample to the solar - type stars within 60 pc from the sun , at least 13,000 objects brighter than @xmath108 are expected to be observable by gaia with an astrometric precision between 5 and 10 @xmath3as @xcite . the apparent radius of the sun at a distance of 60 pc is @xmath109 @xmath3as , implying that the most active stars in the sample ( at least @xmath110%@xmath111% ) should display measurable astrometric effects at the level of @xmath112 @xmath3as .
the cyclic modulation of the signal with the rotation period is much shorter than the typical timescales of starspot evolution , that range typically between @xmath113 and @xmath114 months @xcite . as a consequence ,
the astrometric modulation is quasi - stable , increasing the probability of spurious planetary detections .
of course , the correlation of the astrometric signal with the photometric variation can be used to discriminate between activity - induced and planetary - induced photocentre motions .
such stars would have relative flux variations of the order of 0.1 mag , that is about @xmath115 times larger than the photometric precision of gaia .
however , the unknown value of the inclination of the stellar rotation axis leaves an ambiguity in those cases in which no flux variation can be detected .
we have pointed out the importance of taking into account the astrometric effects of surface brightness inhomogeneities in future astrometric search for extrasolar planets , in particular of planets with an orbital period of a few days or a few tens of days around late - type stars .
the simultaneous measurement of stellar wide - band fluxes offers a powerful tool to discriminate activity - induced effects from the reflex motion of the photocentre induced by an orbiting body .
however , the cases of nearby stars , such as @xmath87 cen or @xmath94 eri , or of rotators observed almost pole - on require some additional methods to investigate the possible effects of magnetic activity . the fraction of stars having an inclination lower than , say , @xmath56 is 6.0% , which implies that at least a few tens of ambiguous cases should be expected in the gaia sample of stars within 60 pc .
a further method of discrimination in those cases comes from the fact that the activity - induced motion has the same period of the rotation of the star .
if the inclination is not too low , it can be measured from the modulation of the chromospheric ca ii h & k or ca infrared triplet fluxes , even for stars with an activity level as low as the sun ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , or can be roughly estimated from the spectroscopic @xmath116 and an assumed stellar radius .
conversely , when the star is viewed almost pole - on , no rotational modulation is expected , but the level of activity can be estimated from those chromospheric indicators these considerations indicate the importance of a characterization of the level of activity and the determination of the rotation period for those stars that have been selected as potentially planet - harbouring objects by means of the astrometric method .
this is particularly important when the measurement samples consist of widely separated data , as will be the case for gaia that will observe a given star for about 80 times in five years , on the average .
dedicated simulations , based on the observed positions of solar active regions along an entire activity cycle and the time - sampling law of gaia , are required to assess the impact of stellar activity on gaia long - term astrometry for solar - like stars .
this will be the subject of a future work . here
we limit ourselves to some general considerations based on solar analogy .
individual spot groups will grow and decay on timescales of a few weeks , nevertheless the astrometric signal due to photospheric activity may show some phase coherence for timescales comparable with the lifetimes of complexes of activity . in the case of sunspot complexes , those timescales do not exceed @xmath117 solar rotations @xcite . however , @xcite reported the persistence of two chromospheric active regions separated by @xmath118 in longitude for time intervals longer than @xmath119 days during periods of moderate or high solar activity .
if a similar behaviour characterizes magnetic activity in other solar - like stars , this can give rise to spurious planetary detections given that the astrometric signal produced by two active regions separated by @xmath118 in longitude shows a sinusoidal oscillation with a period half that of stellar rotation .
stars with a level of activity significantly higher than our sun often show active longitudes of spot activity lasting for decades ( see , e.g. , * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) that can give rise to spurious astrometric detections of bodies orbiting around them .
modelling of the wide - band fluxes of highly active stars indicates that cool spots usually dominate their photometric variations .
moreover , extrapolations based on solar observations at different levels of activity along the 11-yr cycle , support the conclusion that their active regions are dominated by cool spots with bright faculae playing a secondary role @xcite .
this implies that their activity - induced astrometric signal can be estimated from the first of equations ( [ power_law ] ) ( or the corresponding plot in fig .
[ excur_l ] for @xmath120 ) .
afl and cdm wish to dedicate this work to marcello rodon , professor of astronomy in the university of catania and director of the astrophysical observatory of catania , who suddenly passed away during the preparation of the present paper .
the authors wish to thank the anonymous reviewers for their valuable comments .
afl wishes to thank drs .
m. g. lattanzi and a. spagna for useful discussions . active star research at inaf - catania astrophysical observatory and the department of physics and astronomy of catania university is funded by miur ( _ ministero delluniversit e della ricerca _ ) , and by _ regione siciliana _ , whose financial support is gratefully acknowledged . | using a simple model based on the characteristics of sunspots and faculae in solar active regions , the effects of surface brightness inhomogeneities on the position of the photocentre of the disk of a solar - like , magnetically active star , are studied .
a general law is introduced , giving the maximum amplitude of the photocentre excursion produced by a distribution of active regions with a given surface filling factor .
the consequences for the detection of extrasolar planets by means of the astrometric method are investigated with some applications to forthcoming space missions , such as gaia and sim , as well as to ground - based interferometric measurements .
spurious detections of extrasolar planets can indeed be caused by activity - induced photocentre oscillations , requiring a simultaneous monitoring of the optical flux and a determination of the rotation period and of the level of activity of the target stars for an appropriate discrimination .
, , astrometry , stars : activity , planetary systems , stars : spots 95.10.jk , 97.10.jb , 97.10.qh , 97.82.fs |
elliptic 3 surfaces have been the focus of much work in algebraic geometry and number theory over the last few decades .
for such a surface over a field of characteristic @xmath2 with at least one singular fiber , the rank of the mordell - weil group can be as large as @xmath3 .
cox @xcite proved in 1989 , using the surjectivity of the period map for 3 surfaces , that any integer between @xmath2 and @xmath3 actually occurs as the rank of the mordell - weil group of an elliptic 3 surface over @xmath4 .
however , the proof uses transcendental methods , and does not lead to an algebraic construction of examples of elliptic 3 surfaces of a given rank . in 2001
kuwata constructed explicit examples of elliptic 3 surfaces over @xmath5 whose mordell - weil group over @xmath6 has rank @xmath7 , where @xmath8 and @xmath9 ( @xcite ) .
the proof uses inose s theorem ( see proposition [ prop : inose ] ) , which relies on a transcendental argument . for this reason ,
it was not clear how to find explicit generators of the mordell - weil group of such elliptic surfaces until our recent work @xcite , where we found a systematic method to describe a set of generators ( see also @xcite for earlier progress on this question ) . meanwhile , there have been constructions of elliptic 3 surfaces with mordell - weil rank @xmath1 ( @xcite@xcite ) in order to fill the gap of @xcite . however , these constructions are based on somewhat different ideas from that of @xcite , to which the method of @xcite does not seem to apply to find generators of the mordell - weil group .
the goal of this article is to describe some new constructions of elliptic 3 surfaces with high mordell - weil rank using ideas closely related to the idea of @xcite .
the construction of @xcite is based on the kummer surfaces @xmath10 associated with the product of two elliptic curves .
for such a kummer surface shioda and inose @xcite constructed a double cover with certain properties ( now called a shioda - inose structure ) .
we say that a 3 surface @xmath11 has a shioda - inose structure if @xmath11 admits an involution fixing the global @xmath0-form on @xmath11 , such that the quotient is a kummer surface @xmath12 , and provided that the rational quotient map @xmath13 induces a hodge isometry @xmath14 of ( scaled ) transcendental lattices .
when @xmath15 is the product of two elliptic curves , @xmath11 was constructed in @xcite as an elliptic surface having two @xmath16 fibers .
shioda and inose used this construction to prove a beautiful theorem establishing a one - to - one correspondence between singular 3 surfaces and even positive definite binary quadratic forms .
later , inose @xcite constructed @xmath11 explicitly as a quartic surface in @xmath17 and as the quotient of a kummer surface by an involution .
thus , we have a chain of rational maps @xmath18 , which we now call a `` kummer sandwich '' ( @xcite ) .
the construction of @xmath19 in @xcite can be viewed as a base change of elliptic surfaces , and it is this fact that @xcite generalized to construct elliptic 3 surfaces of high rank .
elliptic 3 surfaces of various mordell - weil ranks are constructed as a base change from @xmath11 , which we call the inose surface .
our main idea in this article is to replace @xmath15 by the jacobian @xmath20 of a curve @xmath21 of genus @xmath0 .
we propose two different generalizations of inose s fibration on @xmath10 .
a detailed study of shioda - inose structures on 3 surfaces was carried out by morrison @xcite . for kummer surfaces of principally polarized abelian surfaces ,
double covers with shioda - inose structure were studied by naruki @xcite and dolgachev @xcite . in @xcite ,
kumar explicitly described these surfaces as a family of elliptic 3 surfaces with @xmath16 and @xmath22 fibers , whose weierstrass coefficients are related to invariants of the genus @xmath0 curve @xmath21 . in this construction ,
the rational map @xmath23 in @xcite is a 2-isogeny between elliptic surfaces .
the dual isogeny sets up the sandwich @xmath18 .
our first construction is based on the observation that @xmath24 can also be described as an inose - style base change .
more precisely , there is an elliptic fibration on the kummer surface ( fibration 13 in @xcite ) which can be obtained from the shioda - inose surface in @xcite as a base change by a simple change of variables . in 3 we give the details of this construction of elliptic 3 surfaces of various mordell - weil ranks .
in particular , we give two numerical examples that have mordell - weil rank @xmath1 .
note that fibration 13 in @xcite has one @xmath25 fiber and one @xmath26 fiber , while inose s fibration on @xmath10 has two @xmath25 fibers .
there does exist an elliptic fibration on @xmath27 with two @xmath25 fibers , and it can be constructed from known fibrations in @xcite using the `` 2-neighbor method '' of elkies .
however , this fibration does not generically have a section .
nevertheless , taking its jacobian fibration produces an elliptic fibration with section and the same fiber type .
taking a base change , we can construct elliptic 3 surfaces with mordell - weil rank up to @xmath3 . in
4 we give details of this construction , and give a numerical example with mordell - weil rank @xmath1 . in future work
, we hope to find explicit generators for the mordell - weil groups of the elliptic 3 surfaces constructed here , along the lines of @xcite .
we thank matthias schtt and the referees for helpful comments on an earlier version of the paper .
kumar was supported in part by nsf career grant dms-0952486 , and by a grant from the mit solomon buchsbaum research fund .
kuwata was partially supported by jsps grant - in - aid for scientific research ( c ) no . 26400023 , and by the chuo university grant for special research .
the computer algebra systems ` magma ` , ` sage ` , ` gp / pari ` , ` maxima ` and ` maple ` were used in the calculations for this paper .
by an elliptic surface we mean a smooth projective surface @xmath11 together with a relatively minimal elliptic fibration @xmath28 over a smooth projective curve @xmath21 , all defined over a field @xmath29 , which we assume to be a number field throughout .
we are interested in the case where @xmath11 is a 3 surface , @xmath21 is isomorphic to @xmath30 over @xmath29 , and @xmath31 admits a section @xmath32 also defined over @xmath29 . under these assumptions
the generic fiber @xmath33 of @xmath31 is an elliptic curve defined over the rational function field @xmath34 .
we also assume there is at least one singular fiber ( to ensure that the mordell - weil group is finitely generated ) .
denote by @xmath35 the mordell - weil lattice @xmath36 .
we refer the reader to @xcite for the theory of mordell - weil lattices and @xcite for a broader overview of elliptic surfaces .
a theorem of shioda and tate connects the mordell - weil @xmath37 group with the picard group or the nron - severi group @xmath38 of @xmath33 ( note that linear equivalence and algebraic equivalence coincide ) .
in particular , we have the shioda - tate formula ( cf .
shioda@xcite ) @xmath39 where @xmath40 is the number of ( geometrically ) irreducible components of the fiber @xmath41 .
we call @xmath42 the ( geometric ) picard number . in the absence of a section
we would say that @xmath31 is a genus one fibration . sometimes in the literature
these are called elliptic fibrations and those with section are labeled jacobian elliptic fibrations . the following theorem of inose is essential to our construction .
[ prop : inose ] let @xmath43 and @xmath44 be 3 surfaces defined over a number field @xmath29 and @xmath45 a rational map of finite degree .
then @xmath43 and @xmath44 have the same picard number .
inose proves this lemma for 3 surfaces over @xmath4 using a transcendental argument . since we know that @xmath46 , the conclusion is still valid for our case . due to the transcendental nature of inose s proof
we do not have a very good understanding of how the nron - severi lattices of @xmath43 and @xmath44 are related .
thus , knowledge of an explicit set of divisors generating the picard group of @xmath47 , or of the mordell - weil group if @xmath47 is elliptically fibered , does not easily translate to the same for @xmath48 .
let @xmath11 be a kummer surface associated with an abelian surface @xmath49 .
then , @xmath50 let @xmath49 be a simple abelian surface . then , the picard number @xmath51 is given as follows .
@xmath52 let @xmath49 be an abelian surface isogenous to the product of two elliptic curves @xmath53 . then ,
the picard number @xmath54 is given as follows .
let @xmath21 be a curve of genus @xmath0 given by the equation @xmath56 where @xmath57 is a squarefree polynomial .
in particular , we allow @xmath58 to be @xmath2 ( but in that case @xmath59 ) . in @xcite
it is shown that a generic elliptic 3 surface with @xmath16 and @xmath22 fibers has a shioda - inose structure such that the quotient is the kummer surface of the jacobian of a genus @xmath0 curve .
in fact , this correspondence sets up a birational map between the moduli space of such elliptic 3 surfaces and @xmath60 , the moduli space of genus @xmath0 curves .
it can also be viewed as an isomorphism between the moduli space of 3 surfaces lattice polarized by @xmath61 and @xmath62 , the moduli space of principally polarized abelian surfaces . for a genus @xmath0 curve @xmath21
, we can describe the unique associated elliptic 3 surface with @xmath16 and @xmath22 fibers as follows .
let @xmath63 be the igusa - clebsch invariants ( see for example @xcite@xcite ) .
then the corresponding 3 surface is given by the weierstrass equation @xmath64 it has a type @xmath16 fiber at @xmath65 and a type @xmath22 at @xmath66 .
we say that @xmath21 is `` general '' if the other singular fibers are all @xmath67 .
the surface defined by is isomorphic to @xmath68 which we call @xmath69 , and define the base change @xmath70 compare with the definition of @xmath71 in @xcite and @xcite .
straightforward calculations show that the kodaira - nron model of @xmath72 is a 3 surface for @xmath73 . in @xcite the map @xmath74
is given as a @xmath0-isogeny between two elliptic surfaces .
all jacobian elliptic fibrations ( elliptic fibrations with section ) on @xmath27 are classified in @xcite .
fibration 13 in @xcite is @xmath75 which is isomorphic to @xmath76 by @xmath77 .
thus , @xmath76 is isomorphic to @xmath27 , and we have a kummer sandwich diagram : @xmath78 . the surface @xmath69 can be realized as a quartic surface just as inose s suface in @xcite : @xmath79 therefore we can regard @xmath69 as a generalization of inose s surface to the case of jacobian of the genus @xmath0 curve @xmath21 .
this point of view was also studied in @xcite .
[ thm : g^(n ) ] suppose @xmath21 is general .
then the rank of the mordell - weil group @xmath80 is given by the table below .
here , @xmath81 is the picard number of @xmath20 . @xmath82
determination of the types of singular fibers is a straightforward application of tate s algorithm . by proposition [ prop : inose ] , the picard number of each of the 3 surfaces @xmath83 is equal to that of @xmath27 , namely , @xmath84 .
now , calculation of the rank is a straightforward application of the shioda - tate formula . in order to obtain elliptic 3 surfaces
whose mordell - weil rank @xmath1 , it now suffices to find curves @xmath21 such that the picard number of its jacobian @xmath20 is @xmath85 .
we give two different examples below .
as shown in @xcite , a set of five sections forms a basis of @xmath86 if @xmath87 .
these sections are defined over the field @xmath88)$ ] . in the case where the jacobian @xmath20 is simple , @xmath89 if and only if it has quaternionic multiplication .
writing down an explicit equation of such a curve @xmath21 is a delicate number theoretical problem . here
, we use an example of @xcite which depends on a result of hashimoto and tsunogai @xcite ( see also @xcite ) .
take the curve in ( * ? ? ?
* lemma 4.5 ) and let @xmath90
. then we obtain @xmath91 it has quaternionic multiplication by a maximal order in the quaternion algebra of discriminant @xmath92 , which is the smallest possible discriminant for a non - split quaternion algebra .
calculating igusa - clebsh invariants , and simplifying the coefficients by a change of variables , we obtain the elliptic surface @xmath93 simple calculations show that the singular fibers are irreducible except at @xmath65 where it has a fiber of type @xmath94 .
this shows that the rank of @xmath95 equals @xmath1 .
if we take the quadratic twist of @xmath96 by @xmath97 , the elliptic surface becomes simpler : @xmath98 however , it is possible that the field of definition of the mordell - weil group of the original surface may be easier to describe .
as a second example , we consider the case where @xmath20 is isogenous to the product of two elliptic curves . since we want @xmath99 to be equal to @xmath85 , we need to consider the case where @xmath100 with @xmath101 without complex multiplication .
take two elliptic curves @xmath102 these both have rational @xmath92-torsion points , and are @xmath0-isogenous to each other .
moreover , we have @xmath103\simeq e_{2}[3]\simeq{\mathbb{z}}/3{\mathbb{z}}\times\mu_{3}$ ] as galois modules .
take the quadratic twist @xmath104 of @xmath105 by @xmath106 .
then , the galois isomorphism @xmath103\to e_{2}[3]$ ] isometric with respect to the weil pairing induces an anti - isometry @xmath103\to e_{2}^{(-3)}[3]$ ] .
therefore , by frey - kani @xcite , there exists a curve @xmath21 of genus @xmath0 admitting morphisms @xmath107 and @xmath108 of degree @xmath85 . to find an explicit equation , we use shaska s result @xcite . using the fact that @xmath109 and @xmath110 , we find that the curve @xmath21 corresponds to the values @xmath111 , @xmath112 in equation ( 8) in @xcite . after making some changes of variables ,
we obtain the following weierstrass equation for @xmath21 : @xmath113 the maps @xmath114 and @xmath115 are given by @xmath116 where we have used the following weierstrass forms for @xmath117 and @xmath104 : @xmath118 the elliptic 3 surface @xmath72 corresponding to this @xmath21 is given by @xmath119 ( here , we have scaled @xmath120 , @xmath121 , and @xmath122 differently from the definition of @xmath72 to clear denominators . )
it is readily verified that the singular fibers other than the @xmath94 fiber at @xmath65 are all irreducible .
since @xmath117 and @xmath104 are isogenous over @xmath123 and do not have complex multiplication , the rank of @xmath95 is exactly @xmath1 .
the reader may wonder why we did not use a curve @xmath21 such that @xmath20 is @xmath124-isogenous to the product @xmath53 .
the reason is that the resulting elliptic surface @xmath69 has at least one @xmath125 fiber other than @xmath126 and @xmath127 fibers .
thus , the rank of @xmath95 can not be @xmath1 .
similarly , example 1 in @xcite yields @xmath69 with two @xmath125 fibers , and example 2 yields one with a @xmath128 fiber .
a kummer surface @xmath27 can be realized as a complete intersection of three quadrics in @xmath130 .
it contains two sets of sixteen lines intersecting each other as shown in ( * ? ? ?
* , figure 21 ) .
classically , these sets are labeled _ nodes _ and _ tropes _ ; in the model of the kummer surface as a quartic surface in @xmath131 with sixteen singular points , the ( blown - down ) nodes are the singular points , while the tropes correspond to the sixteen planes tangent to the kummer surface along plane conics ( which are the transforms of the tropes ) . as indicated in figure 1
, we find a configuration of a pair of divisors of type @xmath129 .
we thus have an elliptic fibration on @xmath27 having these two @xmath129 fibers .
it turns out , however , that such an elliptic fibration does not have a section .
thus , it is not in the list of elliptic fibrations studied in @xcite . to obtain an explicit equation of this fibration
, we start with fibration 1 in @xcite , two @xmath26 fibers and six @xmath135 fibers , with mordell - weil group @xmath136 . the zero section in @xcite
is taken to be @xmath137 .
however , we will take @xmath138 to be the zero section ; this does not affect the weierstrass equation , however changes the identity components of the reducible fibers , for instance .
we can now take a @xmath0-neighbor step to the elliptic fibration corresponding to the fiber @xmath139 ( this fiber class is in the orbit of fibration @xmath140 of @xcite ) .
we omit the details of the neighbor - step , apart from giving the elliptic parameter @xmath141 where @xmath142 is the equation of the section @xmath143 .
the new elliptic fibration has weierstrass equation . @xmath144
the fiber @xmath145 is at @xmath146 , while there is another fiber of type @xmath147 at @xmath148 , given by @xmath149 .
there are also @xmath135 fibers at @xmath150 and @xmath151 .
the first is @xmath152 ( in the notation of fibration @xmath153 of @xcite ) and the second has @xmath154 as a component .
we can now move by another @xmath0-neighbor step to the elliptic fibration with @xmath155 as a fiber .
the elliptic parameter is @xmath156 we omit the remaining details .
the jacobian of this new fibration is given by @xmath157 where @xmath158 and @xmath159 it has @xmath25 fibers at @xmath160 and @xmath146 . since @xmath166 is the jacobian of an elliptic fibration on @xmath27 ,
there is a rational map @xmath167 of finite degree .
thus , by proposition [ prop : inose ] , the picard number of @xmath166 is equal to @xmath84 .
the rest of the proof is similar to that of theorem [ thm : g^(n ) ] .
thus , @xmath168 for any simple abelian surface without extra endomorphisms gives an elliptic 3 surface of rank @xmath1 .
however , although a very general abelian surface ( in the moduli space of principally polarized abelian surfaces ) has this property , and we expect a `` random '' numerical example to have @xmath169 , we must verify it for any putative example explicitly . to do so , we first use leprvost s criterion ( * ? ? ?
* lemme 3.1.2 ) : a genus @xmath0 curve @xmath21 over @xmath5 has simple jacobian if there is a prime @xmath170 of good reduction such that the galois group of the characteristic polynomial of frobenius of the reduced curve is the 8-element dihedral group @xmath171 .
we can compute the characteristic polynomial of frobenius in magma or sage ; the latter has an optimized implementation of the @xmath172-adic harvey - kedlaya algorithm for computing the matrix of frobenius for a curve over a finite field .
[ e6e6example ] let @xmath173 , @xmath174 , and @xmath175 .
then , we can verify that the jacobian @xmath20 of the curve @xmath176 is simple and @xmath177 by the lemma below .
after a certain change of variables , we have @xmath178 the mordell - weil rank of @xmath168 is @xmath1 .
the characteristic polynomial of @xmath179 and @xmath180 on @xmath181 are given by @xmath182 we can directly verify ( for instance in gp / pari or magma ) that @xmath183 has galois group @xmath184 .
therefore , the jacobian is simple .
next , since @xmath183 and @xmath185 are irreducible , it follows by ( * ? ? ?
* theorem 2 ) that the endomorphism rings of the reductions @xmath186 and @xmath187 are orders in quartic fields defined by @xmath183 and @xmath185 .
these fields are easily checked to be linearly disjoint .
since @xmath188 is a subfield of each , it must be @xmath5 , and @xmath189 must be @xmath190 .
, _ defining equations of singular @xmath194 surfaces and a notion of isogeny _ , proceedings of the international symposium on algebraic geometry ( kyoto univ .
, kyoto , 1977 ) ( tokyo ) , kinokuniya book store , 1978 , 495502 . | we describe two constructions of elliptic 3 surfaces starting from the kummer surface of the jacobian of a genus @xmath0 curve .
these parallel the base - change constructions of kuwata for the kummer surface of a product of two elliptic curves .
one of these also involves the analogue of an inose fibration .
we use these methods to provide explicit examples of elliptic 3 surfaces over the rationals of geometric mordell - weil rank @xmath1 . |
observations in recent years have revealed a strong correlation in the local universe between the central supermassive black holes ( bhs ) and their host galaxies as manifested in the relation between the bh mass and the bulge velocity dispersion , @xmath0 ( e.g. @xcite ; ) , the bulge stellar mass @xmath1 ( e.g. @xcite ; @xcite ) , the concentration of light in the galaxy ( e.g. @xcite ) and the bulge binding energy , @xmath11 ( e.g. @xcite ) . the evolution of these relations with redshift is still unclear with some studies finding evolution in the @xmath0 @xcite and @xmath1 ( @xcite ) relations , with the high redshift bhs being overmassive for a fixed @xmath12 and @xmath13 compared to the local relations , whereas other studies are consistent with no redshift evolution in the observed correlations @xcite .
a possible explanation for this discrepancy lies in the uncertainties in observational selection biases and in the evolution in the intrinsic scatter that is typically stronger for larger bh masses @xcite . the observed correlations are typically explained using theoretical models relying on some form of self - regulated bh mass growth , in which gas is fed to the central black hole until the black hole releases sufficient energy to unbind the gas and blow it away in momentum- or pressure - driven winds ( e.g. ) .
the observed correlations and their evolution with redshift have been reproduced in semi - analytic models ( e.g. @xcite ) , in self - consistent numerical simulations of both isolated galaxies and galaxy mergers ( @xcite ) as well as in galaxies simulated in a full cosmological setting ( @xcite ) .
the key assumption in these models is that the galaxies undergo a brief radiatively - efficient quasar phase triggered by gas - rich galaxy mergers ( e.g. @xcite ) during which the bulk of the bh growth is taking place and the observed correlations are established . in a previous paper ( @xcite , hereafter j09 )
, we showed that the merger remnants of both equal- and unequal - mass mergers of disk and elliptical galaxies satisfy the observed @xmath0 and @xmath1 correlations . in this letter , we study for the first time in detail the evolution of the bh scaling relations during disk galaxy mergers using a new sample of high resolution simulations including a self - consistent formulation for bh feedback .
we seed the bhs initially with masses corresponding to locations below , on and above the observed relations .
thus , in contrast to previous studies we study here for the first time also overmassive bhs lying initially above the observed relations .
the bh scaling relations are defined for merger remnants that have reached their final dynamical state and it is not obvious if the scaling relations are valid during the merging process and at what stage the galaxies evolve onto the relations . , @xmath12,@xmath13 ) are evaluated in figs .
[ mbh_evo_new_bbh]-[mbh_evo_new_abh].,width=279 ]
the simulations were performed using the treesph - code gadget-2 @xcite on the local altix 3700 bx2 machine .
the code includes radiative cooling for a primordial composition of hydrogen and helium .
star formation and the associated supernova feedback is included using the sub - resolution model of @xcite . in this model
the ism is treated as a two - phase medium @xcite in which cold clouds are embedded in a tenuous hot gas at pressure equilibrium .
bh feedback is modeled using the @xcite effective model in which the bh sink particles accrete gas from the surrounding medium according to a bondi - hoyle - lyttleton parametrization @xmath14 with the maximum accretion rate set by the eddington limit . here
@xmath15 and @xmath16 are the density and sound speed of the surrounding gas , respectively , @xmath17 is the velocity of the bh relative to the surrounding gas and @xmath18 is a dimensionless efficiency parameter . following j09 a total of 0.5% of accreted the rest mass energy is then injected as thermal energy into the gas surrounding the bh particle . throughout this paper
we use the bh repositioning method that ensures rapid merging of the bhs once the two bhs reside in the same parent galaxy .
we assume that the bhs merge instantly if they enter the smoothing length of each other and if their relative velocity is below the local sound speed .
the progenitor galaxies are setup following the method of @xcite and as detailed in j09 with all primary galaxies having @xmath19,@xmath20 corresponding to a virial mass of @xmath21 is defined such that @xmath22 analogous to the milky way .
the dark matter profiles in all model galaxies are described by a @xcite profile constructed using the concentration parameter @xmath23 of the corresponding nfw halo @xcite .
the dark matter halos are populated with exponential disks with a baryonic mass fraction of @xmath24 and a stellar hernquist bulge with a fraction @xmath25 @xmath26 of the total virial mass , @xmath27 .
the disk is setup with a fractional gas content of @xmath28 , with the rest being disk stars , with the scale lengths of the disk and bulge computed as detailed in j09 .
the secondary models in the unequal - mass mergers are identical to the primary models , except for the fact that all components are scaled down in mass by a factor of three . in setting up the primary galaxy disks we use a relatively high numerical resolution of 300,000 disk particles of which a fraction @xmath28 are gaseous , 100,000 bulge particles and 400,000 dark matter particles .
each gas particle can only spawn a single stellar particle resulting in a constant particle mass resolution of @xmath29 for all baryonic particles and @xmath30 for the dark matter particles .
the gravitational softening length was set to @xmath31 for all baryonic particles and the bhs and to @xmath32 for the dark matter particles .
finally we insert a seed black hole at rest in the center of each galaxy model , where we vary the bh mass depending on if the models should lie initially below ( b - series ) , on ( o - series ) or above ( a - series ) the observed local bh scaling relations .
we adopt orbital geometry g13 @xcite for all simulations in this paper .
this geometry corresponds to the inclinations @xmath33 and the arguments of the pericenter @xmath34 for the primary and secondary galaxies , respectively .
the galaxies approach each other on a parabolic orbit where the initial separation of the progenitors is @xmath35 and the pericentric distance is @xmath36 , where @xmath37 , @xmath38 and @xmath39 , @xmath40 are the virial and disk scale radii for the primary and secondary galaxies , respectively .
all simulations were evolved for a total of @xmath41 with the merger typically taking place at @xmath42 .
the main numerical parameters and final properties of all 10 simulations are summarized in table [ sims ] .
the parameters governing the multi - phase feedback are as detailed in j09 resulting in a star formation rate ( sfr ) of @xmath43 for our primary galaxy .
the accretion parameter @xmath18 ( eq . [ bondi ] ) can essentially be seen as an empirical correction factor that translates from the resolvable low mean density to the time - averaged small - scale density at the location of the bh .
we found that for @xmath44 the bhs reach eddington limited mass growth during the merger .
this is not the case for the value @xmath45 used in j09 ( see fig .
[ dist_sfr_bh - evo ] ) .
the corresponding evolution on the @xmath0 plane is such that the 3b2bh1 @xmath46 simulation does not evolve onto the relation , whereas the 3b2bh @xmath47 evolves onto the relations reproducing the lower resolution results from j09 ( see fig .
[ mbh_evo_new_bbh ] ) .
we conclude that as the spatial resolution in the simulations is increased the corresponding @xmath18 value needs to be decreased in order to reproduce the lower resolution runs .
thus , we adopt @xmath44 for all subsequent simulations .
finally , we note that our results do not depend critically on @xmath18 as long as the condition of eddington limited mass growth during the merger is fulfilled .
we plot in fig .
[ dist_sfr_bh - evo ] the relative distance between the bhs , the total sfr and the total bh accretion rate for a subsample of six simulations . in addition , we overplot filled circles at time intervals of @xmath48 throughout the 3 gyr simulation indicating the time at which the galaxies are extracted for analysis .
this gives a total of 16 snapshots during the merger .
we perform the analysis using all stellar particles found in a sphere with a radius of @xmath49 centered on the bh of the corresponding galaxy .
the repositioning technique employed in this study ensures that the bhs always trace the highest density central structure of their parent halo .
the inclusion of a massive bh initially has a marked effect on the star formation and bh accretion histories of the mergers ( see fig .
[ dist_sfr_bh - evo ] ) .
the higher the initial bh masses the lower the resulting sfrs are throughout the merger .
the sfrs are suppressed by a factor of @xmath50 in the o - series and a factor of @xmath51 in the a - series simulations compared to the b - merger series .
in addition , massive bhs suppress the peak bh accretion rate during final coalescence by a factor of @xmath52 .
following observational estimates we calculate from each snapshot the line - of - sight stellar velocity dispersion @xmath12 inside the effective radius using 50 randomly projected realizations of the galaxies . together with the extracted bh masses
we then plot the evolution of the bhs during the mergers on the @xmath0 plane ( top left panels in figs . [ mbh_evo_new_bbh]-[mbh_evo_new_abh ] ) .
here we also overplot the observed local relation ( @xmath53 , @xmath54 , @xmath55 , @xmath56 , @xcite ) as a solid line with the dashed lines indicating the @xmath57 errors . in the top right - hand panels of figs .
[ mbh_evo_new_bbh]-[mbh_evo_new_abh ] we calculate as a function of time the relative offset of the simulated bh masses from the ones predicted by the relation defined as , @xmath58 where a positive @xmath59 indicates an overmassive and a negative @xmath59 an undermassive bh with respect to the observed correlation ( the @xmath0 relation in this instance ) .
the dashed lines in the plots give the mean @xmath57 error in the velocity dispersion range probed by our simulations . in the bottom panels of figs .
[ mbh_evo_new_bbh]-[mbh_evo_new_abh ] we plot the evolution of the bhs on the @xmath60 plane , with the total bulge masses derived following the method outlined in @xcite .
artificial images smoothed with a gaussian filter were created of every snapshot seen from 50 random projections by binning the central 30 kpc into 128x128 pixels .
we then fit both a pure srsic @xmath61 as well as an exponential disk and srsic bulge component simultaneously @xmath62 . here
@xmath63 are the effective radii and @xmath64 the srsic indices of the bulge component , @xmath65 is the scale length of the disk , @xmath66 are the central surface densities and @xmath67 is chosen so that @xmath68 encloses half the total mass .
we then solve for the best fit bulge - to - total ( b / t ) ratio for each projection and give the total stellar bulge mass as the mean of the best fit values .
in addition , we overplot the observed local @xmath60 relation with errors @xmath69 , @xmath70 , @xmath71 , @xcite ) together with the corresponding evolution of @xmath59 as a function of time in the bottom right panels .
we plot in fig . [ mbh_evo_new_bbh ]
the evolution of our b - merger sample on the planes of the two bh scaling relations . for this sample
the initial bh mass is always set to @xmath73 @xmath74 similarly to j09 and thus rendering the bh initially dynamically unimportant .
all models show initially relatively modest evolution towards the observed relations , with the first passage at @xmath75 primarily only seen as a mild increase in @xmath12 and @xmath13 . in the phase @xmath76 when the galaxies are falling back towards the second encounter both @xmath12 and @xmath13
are increasing more rapidly , due to stronger merger - induced tidal disturbances .
after the second encounter the galaxies rapidly merge within @xmath77 , with the shorter timescales corresponding to the equal - mass merger and the merger with the higher initial gas fraction of @xmath78 ( see fig .
[ dist_sfr_bh - evo ] ) .
this evolution is qualitatively similar to the early results of @xcite , who found that the velocity dispersion increases primarily in two phases , with mild increase after the first encounter and very rapid evolution at the time of the coalescence . in @xcite
the authors also conjectured that the bhs would fall on the @xmath0 already after the first encounter if a constant fraction of 1% of the central gas supply would be accreted onto the bh .
however , these early studies were purely hydrodynamical , lacking self - consistent bh growth and feedback .
we find that although the bhs evolve towards the relation during the merger , the final relations are only established at the time of the final coalescence of the bhs .
typically the velocity dispersions of the remnants are somewhat too large immediately after the merger .
the effective velocity dispersion then slowly decreases due to some low - level residual star formation with the stars preferentially settling in a disk - like structure , with this effect being stronger in high - gas fraction mergers .
however , this effect is less strong in low - gas fraction and low mass ratio mergers , as star formation and bh mass growth are more strongly terminated in these mergers after final coalescence ( fig .
[ dist_sfr_bh - evo ] , see also @xcite ) . on the other hand , we find that the bulge mass is established on a short timescale immediately after coalescence with the pre - coalescence bulge mass typically being markedly lower . the prime driver for the scatter is caused by the initial gas fraction @xmath28 of the progenitor galaxies , with the scatter produced by the evolutionary stage of the merger remnants after the final coalescence being only a secondary effect , as can be seen in the relatively modest evolution between coalescence ( star symbols ) and the final merger remnant state ( square symbols ) . in fig
[ mbh_evo_new_obh ] we study the evolution of the o - merger sample , for which we set the initial seed bh mass such that the model galaxies lie on the observed relations @xmath3 .
both the primary and secondary galaxies in the mergers typically evolve initially along the observed relations , but after the first encounter before final coalescence the secondary galaxy typically evolves horizontally off the relations , again due to the tidal disturbances after the first passage ( fig .
[ mbh_evo_new_obh ] ) .
however , after the merging of the bhs all remnants again evolve rapidly onto the relations , with the exception of the high gas fraction merger , which evolves off the relation .
thus for progenitors with initially low gas fractions ( @xmath5 ) the bh feedback is able to self - regulate the growth of the bhs maintaining them on the relation .
for very high gas fractions the bh growth is too efficient with the self - regulation being unable to maintain the galaxies on the observed relations . finally , motivated by the observations of @xcite we run simulations in which the bhs were initially overmassive by a factor of three @xmath79 . the evolution of this a - merger sample is shown in fig .
[ mbh_evo_new_abh ] .
again we see initially weak evolution with the initial offset maintained , but after the first encounter the secondary bh evolves horizontally towards the observed relation and in some instances even below the relations . after the merger all remnants again evolve above the relations . however , the final merger remnants of the low gas fraction ( @xmath5 ) simulations have a lower offset @xmath80 by @xmath81 dex compared to the initial state . this effect is stronger in the 1:1 merger , which is experiencing a more violent encounter , with more gas channeled to the center and thus stronger self - regulation of the bh .
the high gas fraction merger ( @xmath82 ) on the other hand evolves even further away from the observed relations increasing the offset typically by @xmath81 dex .
in this letter , we have studied the evolution of galaxy mergers on the @xmath83 and @xmath1 planes as a function of time .
we have shown that progenitors with low initial gas fractions ( @xmath5 ) starting below the relations evolve onto the relations , progenitors on the relations stay there and progenitors above the relations evolve towards the relations , but still remaining above them .
progenitors with high initial gas fractions ( @xmath9 ) evolve above the relations in all cases , with the initial gas fraction thus being the prime source of scatter in the observed relations ( see also @xcite ) .
the evolution for all mergers is initially slow with the observed relations typically being rapidly established during a relatively short phase centered at the time of final coalescence of the bhs .
the observations of @xcite indicate that the bhs at high redshifts were typically overmassive by a factor of @xmath84 for a fixed @xmath12 and @xmath13 compared to the local relation .
these bhs could plausibly have been formed during a brief quasar phase ( e.g. @xcite ) triggered during the mergers of very gas - rich galaxies , thus resulting in overmassive bh masses for their given @xmath12 and @xmath85 , as seen in our @xmath9 simulation series . however , this being the case it is not obvious how these galaxies would evolve onto the local observed relations until the present day .
another binary merger with massive bhs in place does not bring the galaxies onto the relations ( o- and a - series ) , with high gas fractions mergers moving them even further away from the relations .
potentially , the bulge mass could be increased by dry minor mergers ( e.g. @xcite ) , whereas internal secular processes could be responsible for increasing the velocity dispersion .
however , it is not obvious how both the velocity dispersion and the bulge mass could be increased at a fixed bh mass .
finally , another possibility could be that some of the bhs undergoing mergers are ejected in a sling - shot effect @xcite , thus decreasing the total bh mass .
the fact that the bhs starting below the relations evolve onto the relation , whereas the ones above do not , indicates that our simulations rule out the scenario in which overmassive bhs evolve onto the relations through gas - rich mergers , whereas undermassive bhs can evolve onto the relations in galaxy mergers .
thus , given the numerical limitations inherit in our simulations , our results disfavor the picture in which supermassive bhs develop significantly before their parent bulges . finally , our current bh accretion and feedback prescription seems to describe the growth of bh during the merger phase adequately .
however , it is not obvious that the present description is also valid during the secularly driven low accretion phase after the merger is completed .
some initial steps to improve the prescription have been taken by developing an entirely new momentum driven feedback prescription @xcite .
recent observations @xcite have indicated that there is a time lag of @xmath86 between the peak of star formation and the onset of agn activity .
our current model has difficulties in reproducing this and thus developing models that shed light on this discrepancy might also ultimately help us understanding in how , when and why the bhs in galaxies evolve onto the observed relations . | we study the evolution of black holes ( bhs ) on the @xmath0 and @xmath1 planes as a function of time in disk galaxies undergoing mergers . we begin the simulations with the progenitor black hole masses being initially below @xmath2 , on @xmath3 and above @xmath4 the observed local relations .
the final relations are rapidly established after the final coalescence of the galaxies and their bhs .
progenitors with low initial gas fractions ( @xmath5 ) starting below the relations evolve onto the relations @xmath6 , progenitors on the relations stay there @xmath7 and finally progenitors above the relations evolve towards the relations , but still remaining above them @xmath8 .
mergers in which the progenitors have high initial gas fractions ( @xmath9 ) evolve above the relations in all cases @xmath10 .
we find that the initial gas fraction is the prime source of scatter in the observed relations , dominating over the scatter arising from the evolutionary stage of the merger remnants .
the fact that bhs starting above the relations do not evolve onto the relations , indicates that our simulations rule out the scenario in which overmassive bhs evolve onto the relations through gas - rich mergers . by implication
our simulations thus disfavor the picture in which supermassive bhs develop significantly before their parent bulges . |
the suggestion of lowering the fundamental scale of quantum gravity to as low as tev offers a promising scenario in addressing important cosmological issues .
recently randall and sundrum ( herein refered to as the rs model ) [ 1 ] demonstrated that can be achieved without the need for large extra dimensions but instead through curved 5-dimensional space - time @xmath1 that generates an exponential suppression of scales at the boundary where the tev brane is located .
brane cosmology of the rs model deviates from the usual frw universe .
friedman equation for brane expansion produces an unconventional relation for the hubble constant : @xmath2 instead of @xmath3 [ 10 ] .
various authors suggested ways of obtaining the usual relation @xmath4 for cosmology .
however many phenomenological and astrophysical constraints like inflation and the moduli problem , the cosmological constant , bbn , reheat temperature @xmath5 , sufficient e - folding , density perturbation and large scale structure , etc .
make this a formidable task .
the original rs model was extended by i. oda [ 2 ] to include many static branes , while t. nikei [ 3 ] found solutions in the case of two branes inflating in their transverse space .
based on the rs model , the framework of this paper is a generalized scenario of _ many branes that are inflating in their transverse space_. tev branes can have positive as well as negative tensions . in sec.2
we review the exact solution in closed form of [ 6 ] .
a useful result of [ 6 ] is that each brane expands according to its canonical , effective expansion rate @xmath6 , that depends on the position of the brane in the fifth dimension . the 5th dimension is required to remain static during and after inflation . in sec.3 we allow for matter in bulk and consider rolling branes . in the effective @xmath7 picture , the energy of the brane is replaced with a slow - roll potential ( inflation ) which comes to a steep end near the minimum , resulting in reheating . as expected , this requires a potential for stabilizing the radion during and after inflation .
radion potential induces inflation in the branes and decays when branes roll to the stable minimum of the potential .
this confirms dvali and tye s view [ 4 ] that although inflaton is a brane mode in the ground state it behaves as an `` inter - brane '' mode that describes the relative separation of branes in the 5th dimension .
it is this relative separation that induces inflation on the branes with the slow - roll properties .
it is assummed that the inflaton is very weakly coupled to the bulk modes because the radion is heavy .
this fact avoids reheating of the extra dimension as well as overclosure of the universe from the kk excitations .
it also preserves the success of bbn by ensuring that the radion has settled at its minimum and that the size of the extra dimension has not evolved since . for this to be the case , two requirements need be satisfied ( [ 5 ] ) : the radion stabilized by an effective potential @xmath8 must have a mass @xmath9 larger than the expansion rate @xmath10 ; and the brane must have negligible thickness otherwise the universe can not be treated as @xmath11 dimensional at distances of @xmath12 i.e. @xmath13 , where @xmath14 is the size of the extra dimension .
we consider a potential for the radion that satisfies the above requirements in sect.3 . and join the slow roll solution of [ 6 ] reviewed in sect.2 to the solution found in [ 7 ] that is valid during reheating , when branes reach the steep end in the potential .the number of e - foldings and the reheating temperature are calculated .
the issue of the effective @xmath7 cosmological constant @xmath0 and a summary of results are presented in sec.4 .
the radion potential induces a canonical cosmological constant on the branes .
@xmath0 is a dynamic quantity . when branes are displaced away from the minimum , in the slow - roll region , @xmath0 results in inflation of the branes .
however when branes roll towards the minimum of the bulk potential , the reheating regime , @xmath0 very fast takes the value @xmath15 before bbn time .
that results in a recovery of the usual friedman equation for an frw radiation dominated universe .
density perturbation constraints put an upper bound on the value of @xmath16 .
the lower bound @xmath17 is the familiar fine - tuning problem of the cosmological constant .
solutions found in [ 6 ] for a five dimensional model with many branes along the fifth compact direction @xmath18 , that inflate in their transverse space @xmath19 , extend the recent models of rs , i.oda and t.nihei [ 1,2,3 ] .
the 5d metric ansatz is @xmath20 with @xmath18 ranging between @xmath21 and @xmath22 and @xmath23 running over the five dimensions . the 5d einstein - hilbert action is : @xmath24 where the 4d metric @xmath25 is obtained from the 5d one as follows : @xmath26 ; and @xmath27 is the lagrangian of the matter fields , ( the constant vacuum energy ) , in the i - th brane .
the 5d gravitational coupling constant @xmath28 is related to the 4d newton constant @xmath29 through @xmath30 @xmath31 .
the dynamics of the boundaries with energy density denoted by @xmath32 ( where we have separated a constant vacuum energy contribution , @xmath33 ) are based on the slow roll assumption .
the following equations result from varying the action in eqn.(2.2 ) with respect to the metric , [ 1,2,3,6 ] @xmath34 - \frac{1}{f^2 } \left[\frac{f''}{f } + ( \frac{f'}{f})^2 \right ] & = & - \frac{\kappa^2}{3 f}[\sum_i({\cal l}_i + v_i)\delta(z - l_i ) ] + \frac{\lambda_5}{3}\nonumber \\ & & \nonumber\\ \frac{1}{f^2}\left [ ( 2\frac{\ddot{v}}{v})+ ( \frac{\dot{v}}{v})^2\right ] - \frac{3}{f^2}\left [ \frac{f''}{f } + ( \frac{f'}{f})^2\right ] & = & - \frac{\kappa^2}{f}\left [ \sum_i({\cal l}_i + v_i)\delta(z - l_i)\right ] + { \lambda_{5}}\nonumber\\ & & \nonumber\\ \frac{1}{f^2}\left [ ( \frac{\ddot{v}}{v})+ ( \frac{\dot{v}}{v})^2\right ] - \frac{2}{f^2}(\frac{f'}{f})^2 & = & \frac{\lambda_5}{3 } \end{aligned}\ ] ] where prime and dot denote differentiation with respect to z and t respectively , and @xmath35 satisfies @xmath36\ ] ] eqns .
( 2.3 ) become @xmath37 & = & - \frac{\kappa^2}{3f } [ \sum_i({\cal l}_i + v_i)]\delta(z - l_i ) + \frac{\lambda_5}{3 } \\ & & \\ \frac{h^2}{f^2 } - \frac{1}{f^2}(\frac{f'}{f})^2 = \frac{\lambda_5}{6 } \end{array}\ ] ] the effective expansion rate is defined as @xmath38 .
each brane , depending on its position in the fifth dimension , feels a different expansion rate resulting from scaling of the canonical proper time with the position of the brane , @xmath39 .
thus the effective expansion rate for brane - bound observers , of each brane located at @xmath40 in the extra dimension , in their canonical coordinates , is @xmath41 .
denote the absolute value of @xmath42 by @xmath43 .
since we consider an @xmath44 geometry , @xmath43 is positive and satisfies the following relation @xmath45 the solutions of eqn s . (
2.5 ) are , [ 6 ] @xmath46^{-1 } } \\
g(z ) & = & \left(\displaystyle{\sum^{n-1}_{i=1}}(-1)^{i+1}|z - l_i|+l\right)(-\beta)\\ g''(z)&= & 2\left(\displaystyle{\sum^{n-1}_{i=1}}(-1)^{i+1 } \delta |z - l_i| \right ) ( -\beta ) \end{aligned}\ ] ] * figure 1.0 * the even @xmath47 branes are located at @xmath48 with alternating positive and negative tensions such that the number of positive energy branes equals the number of negative energy branes ( fig.1.0 ) .
+ the constants in ( 2.6 - 2.8 ) are given in terms of the parameters of the model by ; @xmath49 and @xmath50 .
the value of @xmath51 is determined by @xmath43 and @xmath52 .
the junction conditions , with a static fifth dimension , on the energy densities @xmath32 of each brane , and their canonical expansion rates @xmath53 , ( fig.2.0 ) , become @xmath54 ) \\
h_{ieff } & = & \sqrt { \frac { \lambda}{6 } } ( \displaystyle \sinh[g(l_i)])\end{aligned}\ ] ] * fig .
2.0 * the solution for the warp factor @xmath55 , eqns.(2.6 - 2.7 ) , fig.1.0 , has a minimum at @xmath56 and a maximum at @xmath57 . and
the overlap of branes located at @xmath56 see [ 2 ] . ] the physical length , @xmath58 , depends on the energies of the branes because they are gravitational sources for the curvature of the fifth dimension .
it is clear from eqns.(2.9 , 2.10 ) that branes positioned near the minimum of the warp factor , @xmath56 ( the tev branes ) in the fifth dimension , have the maximum energy and effective expansion rate allocated to them .
branes near the position of the maximum warp factor @xmath59 ( planck branes ) , are almost devoid of energy and have the lowest canonical expansion rate . at @xmath59
is almost but not identically zero ( fig.2.0).similar result have also been obtained by [ 9,11 ] ] junction conditions of eqn.(2.9,2.10 ) give the following friedmann equation for the expansion rate and the energy density for the branes @xmath60 notice the unusual friedman relation between the energy density and the expansion rate for each individual brane [ 8,9,10 ] .
an important result of the eqn .
( 2.11 ) is that each brane has a different expansion rate depending on its position in @xmath61 and that the tev branes do not have to be negative energy branes in this model .
it was noted in [ 7 - 12 ] that the @xmath62 metric of eqn.(2.1 ) has a gauge freedom in the ( @xmath63 ) plane .
we use this property in what follows . consider having a bulk field @xmath64 with mass @xmath65 , which is pinned down to the branes through an interaction term @xmath66 [ 11 ] .
the wavefunction is taken to be @xmath67 and @xmath68 are the vev eigenvalues of the bulk field at the brane location @xmath69 . by varying the 5-dimensional action @xmath70 with @xmath18 , and assuming that the values of the field @xmath71 on the branes , @xmath68 , are large , one gets the following equation of motion for the scalar field @xmath64 can be as high as @xmath72 , hence there is a lower bound of @xmath73 gev to it from inflation , [ 5,17 ] ] @xmath74 \right\ } \psi(z ) = 0\,.\ ] ] the conformal factor multiplies the @xmath75-term due to @xmath18 translation invariance of the metric .
the induced effective energy density for the branes [ 4,5 ] is @xmath76 where @xmath77 , obtained from the interaction term in the lagrangian , @xmath78 , is the induced potential to the @xmath40 brane from radion coupling , and @xmath79 is the trace of the stress - energy tensor of the brane . by branes tension ] the kinetic term in the effective lagrangian ( dilaton - type coupling , first term in eqn.(3.2 ) ) comes from varying the brane action with respect to its position @xmath80 relative to other branes and describes the relative motion of the i@xmath81-brane with respect to the other branes ( [ 4,5 ] ) .
the effective radion potential @xmath82 is the sum of two contributions , the energy of the bulk field ( @xmath83 ) and the energy of pointlike brane sources , @xmath68 and @xmath79 , which are the branes energy densities weighed by the conformal factor . in a very interesting paper , authors of the first reference in [ 5 ]
explicitly show the analogue of a gauss law for stabilizing the radion , namely ; in order to have a static fifth dimension , consistent with einstein equations and junction conditions near the branes translation invariance .
matter and radiation energy densities of the branes are related to first - order metric fluctuations near the branes through junction conditions for the energy density @xmath84 and pressure @xmath85 , @xmath86 , @xmath87 , treated in detail in [ 5,7,9,10,11 ] . ]
, the effective potential of the radion should be zero , @xmath88 .
after putting the bulk field solution back into the action , [ 11 ] , and integrating out the fifth dimension , the effective radion potential becomes @xmath89 , for the @xmath90 patch with the slow - roll regime , where the conditions @xmath91 are satisfied .
the 4-dimensional potential induced to the @xmath80-brane , @xmath77 is @xmath92\ ] ] the energy density on the @xmath93-brane is given by eqn.(3.2 ) .
* figure 3.0 * the @xmath94 potential has the slow - roll regime for the most part except around @xmath95 ( fig.3.0 ) where it becomes very steep .
we can therefore use the previously found solution of ref.[6 ] , eqn.(2.11 ) for the slow - roll part in which @xmath77 is nearly a constant .
@xmath96 the region @xmath97 where the slow roll conditions , @xmath98 and @xmath99 , break down is studied under the sudden approximation . the point @xmath100 , where the slow roll regime ends , is taken to correspond to the last 60 e - foldings of the slow - rolling brane .
let us promote the location of the brane @xmath80 to a scalar field , @xmath101 . with this notation
, we can calculate the e - foldings and find the location @xmath100 of the last one , @xmath102 , as follows @xmath103}{ch[g(\phi_l)]}\right]\ ] ] where the slow roll ends and the brane rolls down the steep part of the potential is given by : @xmath104}{ch(g \phi_l)}\right ] \simeq l_n - l \left . \right)$]]note
that this potential also satisfies the thin brane requirements for the effective expansion rate at the end of inflation @xmath105 ( eqn .
( 2.10 ) , fig . 2.0 ) .
we ignore effects of backreaction or quantum corrections on the geometry . +
* 3.2 * _ brane dynamics _
the kinetic term @xmath106 is ignored for the slow - roll part up to @xmath107 .
the radion dumps most of its energy to the tev branes which are in the slow - roll regime , located far from @xmath108 since its coupling has the lowest suppression to the branes in that regime , eqn.(2.9,2.10 ) .
the highest suppression is in the planck brane at @xmath109 because the maximum of @xmath55 is at @xmath109 and minimum at the tev brane @xmath110 , eqn.(2.6 ) , fig.1.0 .
this confirms dvali and tye s view in [ 4 ] that inflation is an inter - brane mode , and that the radion is the inflaton of each brane .
the physical reason why tev brane - bound observers feel the highest expansion rate as indicated by eqn .
( 2.7 , 2.10 - 2.11 ) is : tev branes get the maximum inflation rate and consequently the highest vacuum energy because most of the radion energy is dumped in them , while observers bound to the planck branes have the lowest vacuum energy and expansion rate @xmath111 as they get very little energy induced by the radion due to the suppression with @xmath72-scales . allowing a bulk potential for stabilizing the radion , also induces a potential for the branes that makes them exit inflation and recover late - time cosmology .
the volcano type potential @xmath94 , which at the brane location appears as @xmath112 , eqn.(3.3 ) , is almost flat for most of the rolling time up to a critical distance @xmath113 where it becomes very steep and brane inflation ends rapidly ( fig.3.0 ) . in the first approximation [ 4,5,6,12 ] the extra radii is frozen both during and after inflation as the radion is heavy and stabilized by the potential in sect .
the mass of the radion @xmath9 satisfies @xmath114 .\ ] ] at the minimum of radion potential its energy is released into radiation on the branes .
the inflaton remains very weakly coupled to the bulk modes .
this means only the brane is reheating thus avoiding the problem of overclosure of the universe .
the reason is that the mass of radion @xmath9 is greater than the effective hubble constant @xmath115 during the last stages of inflation and also the coupling of branes to the bulk modes are at maximum suppression at @xmath59 ( the location of the planck brane ) .
it was shown in [ 5 ] that if this is the case , the production of kk modes is greatly suppressed .
these requirements are important as they preserve the success of bbn .
the size of extra dimension does not evolve from the time of nucleosynthesis and any shift in newton s constant @xmath29 , related to the shift of the radion minima , [ 5 ] , is negligible .
the slow - roll conditions break at @xmath116 and the branes roll quickly down to the bottom of the potential . for the steep part of the potential , @xmath117 , we use the @xmath118 @xmath119 .
it is a good approximation as the roll down time @xmath120 from @xmath100 to @xmath14 goes as @xmath121 thus it is very small .
authors of [ 7 ] found the following solution for the case when the brane near @xmath122 is in a mixed state of inflation and radiation @xmath123 where @xmath124 @xmath125 and references therein for eqns.(3.7 - 3.10 ) ] the index @xmath126 refers to the point @xmath122 where slow - roll approximation breaks down and @xmath127 to the point @xmath59 where the brane expansion rate is minimum .
@xmath128 now denotes brane tension ] is the brane tension and @xmath129 denote the energy density and pressure for the branes at @xmath122 .
@xmath130 is the part of the energy density that decays to radiation , with decay rate @xmath131 .
it was shown in [ 5,7,8,9,10,11 ] ) that one recovers the radiation - dominated friedman cosmology at @xmath132 under the condition : @xmath133 the procedure is straightforward ; one solves the eqns .
( 3.7 - 3.8 ) to find the inflaton @xmath134 , and radiation energy density @xmath130 , and plugs the result in eqn .
( 3.6 ) to get the hubble expansion rate .
the field @xmath135 during inflation obeys @xmath136 at @xmath137 our solution eqn .
( 3.4 ) for the slow - roll inflation is joined to the solution of [ 7 ] , eqn ( 3.6 ) .
for details of the case of the inflaton oscillating thereby reheating the universe , i.e , a radiation - dominated regime near the minimum of the potential ] note that the switch from inflation to radiation happens during a very short time @xmath138 when the brane rolls down the steep potential .
therefore we use the sudden approximation when joining our solution [ 6 ] to that of [ 7 ] at the end of the slow roll regime , @xmath113 .
the brane evolves to a radiation dominated friedmann cosmology when rolling fast towards @xmath109 , the stable minimum of the radion .
although the kinetic term becomes important in the intermediate stage during the steep roll from @xmath113 to @xmath109 we consider that the inflaton decays very fast , i.e. this regime occurs in a very short time ( from the validity of the sudden approximation , the roll time @xmath121 is very small as @xmath9 is very large ) and therefore can be neglected .
instead lets concentrate only on the two regimes of : slow - roll inflation , just before reaching @xmath100 , and the reheating / radiation dominated regime around and at @xmath109 , where the radion stabilizes .
let us join the slow - roll solution given in eqn.(3.4 ) , ref.[6 ] to the solution found in [ 7 ] , eqn.(3.6 ) , near @xmath113 with the notation @xmath139 and tension @xmath128 @xmath140 ^ 2 - \frac{\lambda}{6 } + ce^{-4h_{eff}\tau_n } \ , , \\
\rho^n_m & = & \frac{1}{2 } \dot{\phi_n}^2 + u_n(\phi ) + \rho_r\ , , u_{min}(z = l ) = e_{l } - m^2f(z = l)\ , , \rho^n_m - \rho_{l } = u_n + \rho_r - u_{min } = -c[1 - e^{-4h_{eff}\delta\tau } ] \simeq -c \ , .
\end{aligned}\ ] ] joining the two solutions introduces a constraint on the integration constant @xmath141 ( eqn.3.14 ) that depends on the initial conditions [ 7 ] .
the constraint is @xmath142 and represents the ammount of energy from initial ( @xmath100 ) to final ( @xmath14 ) state that has been converted to radiation . in order to recover friedman cosmology around @xmath109 from eqn.(3.6 ) with a radiation dominated brane world and the radion stabilized at its minimum @xmath143 ^ 2 - \frac{\lambda}{6}\ ] ] we must require [ 9,10,11,13 ] : @xmath144 i.e @xmath145 with an upper bound such that it satisfies the constraint of the density perturbations ( weinberg s window ) .
the induced vacuum energy @xmath146 on the brane comes to an almost zero value during radiation - dominated time and conventional cosmology at late times is recovered .
at this point @xmath147 reaches also its minimum value ( see fig.2 ) .
thus when the brane was displaced from this minimum , it had a large vacuum energy @xmath0 and expansion rate @xmath148 induced from the radion during the slow - roll regime .
however , due to the dynamics of the brane and radion stability ( going to its minimum ) , the vacuum energy felt by brane - bound observers @xmath0 decreased while the brane rolled towards its minimum .
the minimum of the induced potential on the brane @xmath149 is required to be normalized around @xmath150 in order to recover friedman cosmology , eqn.(3.15 ) , hence does not have to be exactly zero .
, i.e. @xmath151 , through careful cancellation of the brane and bulk vacuum energies , introduces a particle horizon at @xmath59 , as discussed in [ 6,12 ] . in this case the integration constant @xmath141 is related to the area of the horizon . ]
the lifetime of radion on the brane is very short and it decays before the bbn era into radiation , thus can not serve as a dark- matter candidate , ( details in [ 15,17 ] ) .
however , the bulk with mass @xmath65 is a good candidate,(as first suggested by [ 5 ] ) , because it has a small annihilation cross - section .
this can be shown through the following estimates .
the cross - section @xmath152 goes as @xmath153 where @xmath65 is of order @xmath154 or higher and it is stable due to the fact that bulk modes have the highest suppression in coupling to the branes located at @xmath59 where the radion stabilizies , thus the condition @xmath155 is satisfied for reheat temperature as low as @xmath156 [ 5,17 ] .
note that due to the small size of the extra dimension and the warp factor , i.e. the relation @xmath157 and the lower bound on the radion mass @xmath9 , @xmath158-gev , ( [ 5,17 ] ) , the reheat temperature given by @xmath159 [ 17 ] can easily take values @xmath160 gev .
there is also the constraint of density perturbations at the end of inflation .
@xmath161 in order for this astrophysical constraint to be satisfied it imposes an upper bound on the value of @xmath162 .
this translates into a constraint on the size @xmath58 of the @xmath163 curvature . through eqns.(2.10 - 2.11 , 3.19 ) , the constraint eqn.(3.19 ) on the energy @xmath164 , and @xmath148 at the end of inflation are such that @xmath165 the short lifetime of the radion ( which is seen by the brane as the effective @xmath7 inflaton field ) means that it decays very fast and reheats the universe very fast . +
phenomenological models must reproduce conventional cosmology features and satisfy astrophysical constraints in order to become sussessful theories . in this paper
we investigate the cosmology of a generalized rs model in a dynamic setting .
a bulk field introduces a potential which stabilizies the radion .
branes roll because their relative separation adjustes to the radion settling towards its stable minimum .
meanwhile , the radion appears as the inflaton in the effective @xmath7 description of the brane world .
it confirms dvali and tye s point of view [ 4 ] that although the inflaton is a brane mode in the ground state , it behaves as an inter - brane mode that describes a relative separation of the branes in the extra dimension .
this approach clarifies the physics of brane dynamics during inflation .
the radion which depends on the inter - brane separation , in the effective 4d description , drives inflation in the branes .
the tev branes have minimum suppression to the radion modes as compared to the planck branes , thus they will have the maximum inflation confirmed by eqn .
( 2.6 - 2.11 ) .
the coupling of planck branes to the radion and bulk modes have maximum suppression , thereby almost zero inflation rate and vacuum energy since the energy dumped into them by the radion is highly suppressed by planck mass scales @xmath166 .
branes inflate with different canonical expansion rates because their coupling to the radion depends on their position / angle in the @xmath61 .
one of the motivations to consider this model with many branes inflating in their transverse space , is that it allows tev branes to have positive energy and tension , as compared to the case of two branes only where the geometry forces the tev brane to have negative tension .
the dynamics of the model ( rolling of the branes and radion stability ) ensure the recovery of coventional cosmology features at late times .
the radion decays very fast in the @xmath7 brane world , reheats the universe and drives a decaying cosmological constant while reaching its stable minimum .
thus , the cosmological constant issue becomes a dynamic problem , where the branes adjust their positions towards a static solution with a stabilized radion .
however the final value of the vacuum energy @xmath0 felt by brane - bound observers , despite being very small , needs to be tuned to zero .
it has an upper bound in order to satisfy the astrophysical constraint of density perturbations , eqn.(3.19 - 3.21 ) .
this constraint puts a bound on the bulk curvature ( eqn.3.21 ) . due to their final positioning in the warp factor , coupling of the branes
to bulk modes are at maximum suppresion after inflation when the radion has stabilized .
this avoids the reheating of the bulk .
the brane becomes a frw radiation - dominated universe .
there is no moduli problem because the radion is heavy , @xmath167 , and decays very fast thus overclosure of the universe and production of dangerous relics do not arise in this model .
cosmological and astrophysical constraints such as : reheat temperature , sufficient number of e - folding , negligible changes in the constants of nature , are satisfied .
bulk modes can provide a good candidate for dark matter as they are cold , stable ( almost decoupled from the branes ) and the conditions of the smallness of their annihilation cross - section and the reheat temperature are satisfied ( see eqns .
( 3.17 - 3.21 ) ) . for the most part ,
the potential is nearly flat at the slow roll regime and ends steeply to a fast decaying,/reheating regime , near the stable minimum , @xmath59 .
the sudden approximation employed in solving the equations in the two regimes ( i.e. ignoring the intermediate steep part of the fast roll regime , located between the nearly flat part and the stable minimum , where the kinetic term is important ) , is thus justified .
the time taken to roll down the steep part of the potential is proportional to the inverse of the radion mass and therefore very small .
the backreaction of brane matter on the geometry and their quantum corrections have been ignored . * l. randall and r.sundrum , a large mass hierarchy from a small extra dimension , * phys.rev.lett.83:3370-3373 * , @xmath168 * ichiro oda , * phys.lett.b480:305-311 * , _ @xmath169 _ ; * phys.lett.b472:59-66 * , _ @xmath170 _ * t.nihei , * phys.lett.b465:81-85 * , @xmath171 ; n.kaloper , * phys.rev.d60:123506 * , @xmath172 * g. dvali and h. tye , * phys.lett.b450:72-82 * , @xmath173 * c. csaki , m.graesser , l. randall and j. terning , * phys.rev.d62:045015 * , @xmath174 ; c. csaki and y. shirman , * phys.rev.d61:024008 * , @xmath175 ; g. felder , l. kofman and a. linde , * jhep 0002:027*,@xmath176 * l. mersini,cosmological constant of the inflating branes in the randall - sundrum - oda model
, @xmath177 * e.e .
flanagan , s - h .
henry tye and ira wasserman , * phys.rev.d62:044039 * , @xmath178 * a.lukas , b.a.ovrut and d.waldram , * phys.rev.d61:023506 * , @xmath179 * c.csaki , m.graesser,c.kolda,j.terning , * phys.lett.b462:34-40 * , @xmath180 ; j.cline , c.grojean,g.servant , * phys.rev.lett.83:4245,1999 * , @xmath181 ; j. cline , * phys.rev.d61:023513 * , @xmath182 * p. binetruy , c.deffayet , d.langlois , * nucl.phys.b565:269-287 * , @xmath183 , ; p.kanti , i.i kogan , k.a .
olive and m.pospelov , * phys.lett.b468:31-39 * , @xmath184 ; p. krauss , * jhep 9912:011 * , @xmath185 * w.d.goldberger and m.b.wise , * phys.rev.lett.83:4922-4925 * , @xmath186 , * phys.lett.b475:275-279 * , @xmath187 ; h.b.kim and h.d.kim , * phys.rev.d61:0604003 * , @xmath188 ; p.steinhardt , * phys.lett.b.462:41-47 * , @xmath189 ; l.mersini , * mod.phys.lett.a14:2393-2402 * , @xmath190 * a.chamblin , s.w.hawking , h.s.reall , * phys.rev.d61:065007 * , @xmath191 ; b.bajc and g.gabadadze , * phys.lett.b474:282-291 * , @xmath192 ; s.m.carroll , s.hellerman,m.trodden , * phys.rev.d61:065001 * , @xmath193 * n. arkani - hamed , s , dimopoulos , g.dvali and n.kaloper , * phys.rev.lett.84:586-589 * , @xmath194 * n.kaloper and a.linde , * phys.rev.d59:043508 * , @xmath195 * j. lykken and l.randall , * jhep 0006:014 * , @xmath196 * h. verlinde , * nucl.phys.b580:264-274 * , @xmath197 * daniel j.h .
chung , edward w. kolb and antonio riotto , * phys.rev.d59:023501 * , @xmath198 * a.albrecht , p.j .
steinhardt , m.s .
turner and f.wilczek , * phys.rev.lett.48 * ( 1982 ) * fig .
3.0 the radion potential @xmath94 vs.@xmath18 .
tev branes are located on the flat part ( note maximum inflation rate ) and roll down the steep end towards the minimum @xmath59 . at this point ,
the radion has stabilizied , branes exit inflation , and reheat the universe * | we examine the cosmology of randall - sundrum model in a dynamic setting where scalar fields are present in the bulk as well as the branes .
this generates a mechanism similar to that of goldberger - wise for radion stabilization and the recovery of late - time cosmology features on the branes . due to the induced radion dynamics , the inflating branes roll towards the minimum of the radion potential , thereby exiting inflation and reheating the universe . in the slow roll part of the potential , the tev branes have maximum inflation rate and energy as their coupling to the radion and bulk modes have minimum suppresion .
hence , when rolling down the steep end of the potential towards the stable point , the radion field ( which appears as the inflaton of the effective 4d theory in the branes ) decays very fast and reheats the universe .
this process results in a decrease of the branes canonical vacuum energy , @xmath0 .
however , at the minimum of the potential @xmath0 is small but not neccessarily zero and the fine - tuning issue remains .
density perturbation constraints introduce an upper bound on @xmath0 . due to the large radion mass and strong suppression to the bulk modes , moduli problems and bulk reheating
do not occur .
the reheat temperature and a sufficient number of e - folding constraints for the brane - universe are also satisfied .
the model therefore recovers the radiation dominated frw universe . |
the conservation of angular momenta between itinerant electrons and localized magnetizations in magnetic materials leads to the fascinating concept of spin - transfer torque ( stt)@xcite ; the spin angular momentum of the electrons can be transferred to the magnetization via their mutual exchange coupling , which enables to drive the dynamics of magnetization by charge current .
the stt in ferromagnets ( fms ) , providing a vital information - writing technology , has been driving the explosive growth of the field of spintronics up until now.@xcite in textured fms , the efficiency of the stt ( in the unit of velocity ) can be defined by @xmath0 with @xmath1 the g factor , @xmath2 the bohr magneton , @xmath3 the elementary charge , @xmath4 the saturation magnetization , @xmath5 the charge current density , and @xmath6 the net spin polarization carried by the charge current .
recently , antiferromagnets ( afms ) are generating more attention due to their potential to become a key player in technological applications where afms play active roles.@xcite if eq .
( [ u ] ) is directly applied to afms , one would conclude that there can be no stt in afms where @xmath6 becomes zero or vanishingly small ; recent research has been confirming that this is of course not the case .
the study of stts involving amf materials was started by investigation of current - driven effects in spin valves or multi - layer systems where each afm layer carries a single domain.@xcite theoretical studies have unveiled an important role of the stt also in textured afms as in textured fms@xcite ; xu _
et al._@xcite examined the current - driven dynamics of a domain wall ( dw ) in a two - sublattice afm metal by _ ab initio _ calculations .
swaving and duine@xcite formulated a stt in a one - dimensional bipartite afm , based on the landau - lifshitz ( ll ) equations for the sublattice - magnetizations in the continuous limit .
et al._@xcite derived the possible forms of stts that are allowed by symmetry argument .
the dynamics of afm textures driven by spin - polarized current has been also studied@xcite thus far , however , it still remains an open question how the stt efficiency , the counterpart of eq .
( [ u ] ) , can be defined for general afm magnetic textures.@xcite finding the stt efficiency would guide us to how to control the stts in afms for designing more prominent stt effects . in this work
, we develop a formalism of current - driven dynamics of two - sublattice afm textures , where the stt efficiency is provided in terms of unambiguous material parameters . a challenge in deriving the stt in the afms comes from the fact that the electron spin dynamics is not as obvious as in fms at all , because in the afm there are two exchange fields corresponding to the two sublattice - magnetizations , that the electron spin can respect .
we formulate the stt in two regimes where the analytical expressions for the electron spin are available ; when the inter - sublattice electron dynamics is dominant over the electron - magnetization exchange coupling , and the opposite .
we find that the stt mechanism that governs its efficiency can quite differ in those two regimes . in the limit of large exchange coupling
, the stt can be generated due to spatial variation of the antiferromagnetic order . in the opposite limit ,
on the other hand , the stt requires a sufficiently large canting between the sublattice - magnetizations .
these predictions can be quantified by studying the response of afm spin waves to the charge current . in the limit of large exchange coupling ,
the charge current inevitably causes the spin - wave doppler shift , whereas , in the opposite limit , it can modify the spin - wave spectrum only when there exists a pronounced sublattice - canting .
our results demonstrate quantitatively that the stt effects in an afm highly depend on which class of afm we consider .
we consider an itinerant afm composed of two sublattices ( 1 and 2 ) with equal saturation magnetization @xmath4 . in order to treat the magnetization classically ,
the coarse graining for the magnetic channel is performed.@xcite the classical vector @xmath7 @xmath8 is a continuous function in space that represents the local magnetization direction in the sublattice 1 , with a similar definition for @xmath9 ; here the lattice structure is smeared out and the magnetizations of both sublattices are defined at every point in space .
this classical treatment is allowed when the spatial variation of each magnetization is sufficiently slow compared to the atomistic length scale .
the dynamics of the magnetizations are assumed to obey the coupled ll equations with the gilbert - type damping term@xcite ; @xmath10 where @xmath11 is the gyromagnetic ratio and @xmath12 is the damping constant , which are assumed for simplicity to be sublattice independent .
@xmath13 are the effective magnetic fields with @xmath14 being the magnetic energy density , and @xmath15 are the stts to be determined . for the conduction electron channel we employ the following four - band hamiltonian density@xcite ; @xmath16 a derivation of which starting from an atomistic tight - binding model is discussed in appendix a. the upper - left ( bottom - right ) bands correspond to the sublattice 1 ( 2 ) . in the first equality of eq .
( [ h ] ) , the first matrix is the kinetic energy tensor where the diagonal and off - diagonal components describe the intra- and inter - sublattice electron dynamics , respectively , with @xmath17 being the momentum operator of the electron , whereas the second matrix represents the exchange interaction with @xmath18 being the exchange coupling energy and @xmath19 the pauli matrices indicating the electron spin operator . in the second equality , we set @xmath20 and @xmath21 reflecting the sublattice symmetry , use the tensor product representation of the sublattice and spin spaces with the dirac matrices @xmath22 and define the net moment and the nel - order vector by @xmath23 the afm coupling between @xmath24 and @xmath25 is the leading energy scale so that @xmath26 and @xmath27 .
we regard @xmath18 and @xmath28 as parameters , where @xmath29 denotes the expectation value at the fermi surface .
the expressions for @xmath15 are to be derived in the two limiting cases ; the parameter regimes where @xmath30 ( the exchange - dominant regime hereafter ) and where @xmath31 ( the mixing - dominant regime hereafter ) .
the explicit forms of @xmath32 and @xmath33 can be determined based on an atomistic tight - binding model , as discussed in appendix a. the condition @xmath34 can be met in afms where the inter - sublattice electron dynamics is relatively unfavorable ; e.g. , layered afms with the c axis being longer than the other axes ( fig .
2b in appendix a ) . to expand @xmath35 in powers of @xmath36
, we perform the unitary transformation@xcite @xmath37 with @xmath38 because the kinetic energy operators , @xmath32 and @xmath33 , in general do not commute with @xmath39 and @xmath40 , there appear in eq .
( [ hj ] ) terms that contain their commutators .
these terms and the last term in eq .
( [ hj ] ) can be ignored when the spatiotemporal variations of the magnetizations are sufficiently slow ( see appendix b for quantitatively more accurate discussion ) . with this condition
the expression for @xmath41 can be reduced to @xmath42 eq .
( [ hj2 ] ) proves that the inter - sublattice band - mixing can be neglected up to the order of @xmath36 in the certain condition . in this rotated frame ,
the conduction electrons only couple to either @xmath24 or @xmath25 , whereas it is important to note that these sublattice moments are mutually coupled .
therefore , the spin gauge fields for the itinerant electrons that reside in the @xmath43-th sublattice are determined by @xmath44 and the stts @xmath15 in eq .
( [ llg ] ) are derived as @xmath45 where @xmath46 is a dimensionless parameter@xcite and the stt efficiency @xmath47 is given by @xmath48 with @xmath49 representing the spin polarization of the conduction electrons in each sublattice .
we remark here that eqs .
( [ hj2 ] ) and ( [ stt_j ] ) can not be obtained just by assuming the condition @xmath30 ; if the magnetizations change their directions in time and space fast enough , it can cause considerable inter - sublattice band mixing even in the exchange - dominant regime ( see appendix b ) .
but still , one should point out that this formal result does justify ignoring the interband hopping and translating things as the stt in each sublattice to be fairly independent .
the inter - sublattice electron dynamics may be predominant as @xmath31 in , e.g. , bipartite afms where the nearest - neighbor atomic sites connect the different sublattices ( fig .
2a in appendix a ) .
we show here that the expressions for stts in this parameter regime quite differ from eq .
( [ stt_j ] ) .
let us first perform the following unitary transformation on the sublattice space of eq .
( [ h ] ) ; @xmath50 with @xmath51 in the new framework @xmath33 comes in the diagonal components , while @xmath52 is in the off - diagonal components .
the upper - left ( bottom - right ) part of @xmath53 corresponds to the anti - bonding ( bonding ) electron states formed by the two sublattice - states
. then we perform another unitary transformation to expand @xmath53 in powers of the operator @xmath54 ; @xmath55 with @xmath56 assuming the sufficiently slow and smooth variation in the directions of magnetizations , we can express @xmath57 as ( see appendix c for quantitatively more accurate discussion ) @xmath58 here we have succeeded in block - diagonalizing @xmath53 up to the order of @xmath54 . in the mixing - dominant regime with the hamiltonian ( [ ht3 ] ) , the conduction electron spins only see the net moment @xmath39 regardless of the sublattice degree of freedom . in afms
, @xmath39 can emerge due to several origins such as external magnetic fields , the dzyaloshinsky - moriya interaction ( dmi ) , and the spatiotemporal variations in the magnetizations.@xcite the magnitude of the net moment @xmath59 ( @xmath60 ) generally varies in both time and space .
this fact makes it difficult to obtain analytical expressions for the stts for general cases . in the perfect compensation ,
i.e. , when @xmath61 , the electron - magnetization interaction in eq .
( [ ht3 ] ) vanishes and no stts arise . when @xmath62 becomes as large as @xmath63 and @xmath64 over the relevant sample region
, it can induce the net spin polarization where the majority ( minority ) electron spins adiabatically follow the direction of @xmath65 ( @xmath66 ) . in this latter case , @xmath15 in eq .
( [ llg ] ) are given by ( see appendix d for a derivation ) @xmath67 , \label{stt_t}\ ] ] where @xmath68 , @xmath69 is a phenomenological parameter , and the stt efficiency @xmath70 is defined by @xmath71 here , @xmath72 is the net spin polarization of the conduction electrons with respect to @xmath39 . notice that eq .
( [ stt_t ] ) clearly differs from eq .
( [ stt_j ] ) ; both @xmath24 and @xmath25 enter the stts @xmath15 in eq .
( [ stt_t ] ) , in the contrast that each of @xmath44 appears in eq .
( [ stt_j ] ) .
here let us study the effects of the stts on the spin - wave dispersions of the two systems shown in fig .
1 ; an easy - axis ( ea ) afm with external dc magnetic field applied along the easy axis , and an easy - plane ( ep ) afm with external dc field applied in the easy - plane .
we take the magnetic energy density as@xcite @xmath73 - k ( m_{1z}^2 + m_{2z}^2 ) + \mu_0 { \bm h } \cdot ( { \bm m}_1 + { \bm m}_2 ) , \label{w}\ ] ] where @xmath74 and @xmath75 characterize the homogeneous and inhomogeneous exchange couplings , @xmath76 is the uniaxial anisotropy constant along the @xmath77 axis , and @xmath78 is the external magnetic field . in the case of ea - afm ( @xmath79 ) , both @xmath24 and @xmath25 lie in the @xmath77 direction at equilibrium ( fig .
1a ) when the external dc field @xmath80 is in the range of @xmath81@xcite , where @xmath82 in the absence of charge current , the low - energy spin - wave dispersions of this ea - afm are given by @xmath83 where @xmath84 for the ep - afm ( @xmath85 ) , the parallel component of the magnetizations with respect to the dc field is determined by@xcite ( fig .
1c ) @xmath86 the low - energy spin - wave dispersions are @xmath87 ) and presence ( @xmath88 ) of charge current .
the horizontal axis indicates the parallel component of @xmath89 with respect to the charge current .
these modes are affected by the charge current only in the exchange - dominant regime .
the inset magnifies the area indicated by the dotted box , clearly showing the shift of the spectrum around the @xmath90 point . *
c*. schematic of the easy - plane ( ep ) amf .
* d*. the spin wave dispersions of the ep - afm in the absence ( @xmath91 and @xmath92 ) and presence ( @xmath93 , @xmath94 , and @xmath95 ) of charge current . in the mixing - dominant regime ,
only the @xmath96-mode couples to the charge current . in * a * and * c * , the arrows @xmath97 and @xmath98 indicate the equilibrium configurations of @xmath24 and @xmath25 . , width=302 ] let us examine the stt effects on the above eigenfrequencies . in the exchange - dominant regime , eq .
( [ stt_j ] ) indicates that , in the small dissipation limit with @xmath99 and @xmath100 , applying the charge current is to replace the partial derivative @xmath101 in eq .
( [ llg ] ) by the lagrange derivative ; @xmath102 implying the galilean invariance of the system with respect to the electron flow . in systems with the galilean invariance being respected , the spin - wave spectrum exhibits the current - induced doppler shift@xcite ; the spin - wave dispersions change as @xmath103 in the mixing - dominant regime , the analytic form for the stts in eq .
( [ stt_t ] ) requires the net moment @xmath39 to be large enough to satisfy the condition discussed in the previous section . for the @xmath104-modes of the ea - afm
, @xmath39 is mostly vanishingly small and thus tangible stt effects can not be expected . for the ep - afm , on the other hand , there exists the canting moment @xmath39 that can satisfies the above - mentioned condition when the dc field is sufficiently large .
it is shown from eqs .
( [ llg ] ) and ( [ stt_t ] ) that the spin - wave doppler shift takes place in the @xmath91-mode but not in the @xmath92-mode ; @xmath105 this distinct feature arises because in the mixing - dominant regime the charge current couples only to @xmath39 ; the excitation in the @xmath106-mode is the precession of @xmath107 , whereas it is the precession of @xmath108 in the @xmath96-mode . in fig .
1b and d compared are the spin - wave dispersions of the ea- and ep - afms with and without charge current . for the material parameters , values in the typical range for afms
are employed@xcite : @xmath109 j / m@xmath110 , @xmath111 j / m , @xmath112
j / m@xmath110 , @xmath113 a / m , and @xmath114 s@xmath115/(a / m ) .
the magnitude of dc field is set to @xmath116 a / m and @xmath117 a / m for the ea- and ep - afms , respectively . for the ratio of the stt efficiencies ,
@xmath118 is assumed for simplicity , with @xmath119 m / s .
these results demonstrate the important role played by the inter - sublattice electron dynamics ; the reaction of an afm to the charge current qualitatively differs depending on the ratio @xmath120 .
the spin - wave doppler shift offers a way to quantify the stt in the stationary condition in both time and space.@xcite
let us compare our results with existing literature .
for this purpose , we rewrite eq .
( [ llg ] ) in terms of @xmath121 . in the exchange - dominant regime , this leads to the closed equation of motion for @xmath40 ; @xmath122= 0 , \label{n}\end{aligned}\ ] ] while @xmath39 is determined as a slave function of @xmath40 ; @xmath123 where the condition @xmath26 has been used .
the charge current enters eqs .
( [ n ] ) and ( [ m ] ) through the lagrange derivative @xmath124 ( except for the dissipation part ) , being consistent with the previous discussion regarding the galilean invariance . in the absence of charge
current , eqs .
( [ n ] ) and ( [ m ] ) reproduce the well - known equations of motion for @xmath40 and @xmath39 under magnetic fields.@xcite in the mixing - dominant regime , the @xmath121-representation of eq .
( [ llg ] ) is generally not as compact as eqs .
( [ n ] ) and ( [ m ] ) .
when we limit ourselves to the special case where @xmath39 and @xmath40 are always in a single plane , however , the galilean invariance is restored in the strict manner , and eqs .
( [ n ] ) and ( [ m ] ) hold with @xmath47 and @xmath46 replaced by @xmath70 and @xmath69 , respectively .
this condition can be met when , e.g. , a dw is formed in a nanowire that possesses a homogeneous dmi with its dmi vector pointing out - of - plane.@xcite the dw motion predicted by eq .
( [ n ] ) is consistent with the results in literature ( see appendix e).@xcite eq . ( [ n ] ) contains the stt terms predicted in ref . [ ] by symmetry argument , whereas the phenomenologically introduced coefficients are now explicitly given by the stt efficiency @xmath47 . since the newly - added terms in eq .
( [ n ] ) are higher order in terms of the field and derivatives , they were discarded in the previous work .
( [ n ] ) not only makes clear that there is the galilean - invariant nature in the afms , but also predicts the cross terms of magnetic field and charge current , which we will investigate elsewhere . the main focus of ref .
[ ] is on the one - dimensional bipartite afm where @xmath31 .
they conjectured the nonequilibrium electron spin density proportional to @xmath125 with @xmath126 being a parameter in the unit of velocity .
we found that , however , the electron spins predominantly couple to @xmath39 in this parameter regime .
this fact leads to the difference in the results obtained by the two approaches . while our stt in the mixing - dominant regime is of the first order of @xmath127 , their stt in ref .
[ ] is of higher order as @xmath128 . in conclusion
, we have derived the stt efficiency in the two - sublattice afms when the inter - sublattice kinetic energy of the conduction electrons is dominant / negligible compared to the exchange coupling energy . in reality
, many of afm materials should be somewhere in between the two extremes , where numerical approaches will become more powerful .
our theory demonstrates quantitatively that the stts in afms can , in contrast to in fms , highly depend on the nature of kinetic energy of the electrons .
these predictions may be tested by studying the spin - wave doppler shift in the presence of charge current .
the authors are grateful to k. yamamoto , k. kubo , and m. mori , for fruitful discussions .
this research was supported by research fellowship for young scientists from japan society for the promotion of science , grant - in - aid for scientific research ( no . 24740247 , 26247063 , 16k05424 ) from mext , japan , alexander von humboldt foundation , the ministry of education of the czech republic ( grant no . lm2011026 ) and the grant agency of the czech republic ( grant no .
14 - 37427 ) .
-th unit cells , where the sublattice 1 ( 2 ) contributes the magnetization @xmath129 ( @xmath130 ) .
@xmath131 and @xmath132 represent the nearest - neighbor hopping parameters between intra- and inter - sublattice sites , respectively . * c*. schematic of the coarse - grained model in eq .
( [ h ] ) , where both of the sublattice - magnetizations @xmath24 and @xmath25 are continuous and defined at every point in space . such as the atomistic lattice structures and the electron hopping natures
are reflected in the kinetic energy tensor of the electron .
, width=302 ] here let us start from an atomistic model for the afm metal .
the presence of two sublattices leads to unit cells that contain two sites ; the @xmath133-th unit cell consists of the @xmath134 site that belongs to the first sublattice and the @xmath135 site from the second sublattice , on which the magnetizations @xmath129 and @xmath130 are located , respectively ( fig .
2a and b ) .
the tight - binding hamiltonian for the conduction electron is given by @xmath136 here , @xmath137
( @xmath138 ) is the annihilation ( creation ) operator of an electron with spin @xmath139 at the @xmath134 site , and similarly for @xmath140 ( @xmath141 ) .
the first and second terms are the kinetic energies , where @xmath131 ( @xmath132 ) represents the nearest - neighbor hopping parameter between intra-(inter-)sublattice sites ( fig .
2a and b ) .
the sum @xmath142 in the first terms only takes into account the nearest - neighbor intra - sublattice pairs , i.e. , @xmath143 , whereas the second terms include pairs within unit cells , i.e. , @xmath144 .
the third terms describe the on - site exchange coupling . introducing the four - component field operator @xmath145 and its fourier transformation by @xmath146 , with @xmath147 being the sample volume and @xmath148 indicating the position vector of the @xmath133-th unit cell , eq .
( [ h_tight ] ) can be rewritten into @xmath149 fashion as @xmath150 here , @xmath151 and @xmath152 , where @xmath153 and @xmath154 denote , respectively , the vectors connecting the intra- and inter - sublattice nearest - neighbor sites .
the explicit forms of @xmath155 and @xmath156 are given , e.g. , in the bipartite afm by @xmath157 and in the layered amf by @xmath158 here @xmath159 is the lattice constant in the bipartite afm , and @xmath160 and @xmath161 are , respectively , the lattice constants within and between the fm - ordered layers in the layered afm .
the @xmath17-independent terms have been neglected in eqs .
( [ t_bip ] ) and ( [ t_lay ] ) . by taking the continuous limit in the real space for the second term of eq .
( [ h_tight2 ] ) and moving to the first - quantized representation , we arrive at eq .
( [ h ] ) where @xmath162 and @xmath163 are identified with @xmath155 and @xmath156 , respectively , with @xmath17 read as the quantum operator . to expand @xmath35 in eq .
( [ h ] ) in powers of @xmath36 , we perform the unitary transformation in eq .
( [ hj ] ) or @xmath164 + \frac { [ s_j , [ s_j , { \cal h } ] ] } { 2 } + ...
+ i \hbar \partial_t s_j + ... , \label{hj_ap } \end{aligned}\ ] ] with @xmath165 given in eq .
( [ sj ] ) .
the terms in eq .
( [ hj_ap ] ) are computed up to the first order of @xmath36 as @xmath166 & = & - \gamma_5 \left ( t_{12 } + \frac { [ t_{12 } , { \bm \sigma } \cdot { \bm n } ] { \bm \sigma } \cdot { \bm n } } { 2 } \right ) + \frac { \gamma_0 } { j } \left ( t_{12}^2 { \bm \sigma } \cdot { \bm n } - \frac { t_{12 } [ t_{12 } , { \bm \sigma } \cdot { \bm n } ] } { 2 } \right ) \nonumber \\ & & + \gamma_0 \gamma_5 \left ( - \frac { t_{12 } [ t_{11 } , { \bm \sigma } \cdot { \bm n } ] } { 2 j } + \frac { [ t_{12 } , { \bm \sigma } \cdot { \bm m } ] { \bm \sigma } \cdot { \bm n } + 2 i t_{12 } { \bm \sigma } \cdot ( { \bm n } \times { \bm m } ) } { 2 } \right ) , \end{aligned}\ ] ] @xmath167 ] } { 2 } & = & - \frac { \gamma_0 } { 2 j } \left ( t_{12}^2 { \bm \sigma } \cdot { \bm n } - \frac { t_{12 } [ t_{12 } , { \bm \sigma } \cdot { \bm n } ] } { 2 } + \frac { \left\ { t_{12 } { \bm \sigma } \cdot { \bm n } , [ t_{12 } , { \bm \sigma } \cdot { \bm n } ] { \bm \sigma } \cdot { \bm n } \right\ } } { 4 } \right ) \nonumber \\ & & - \frac { { \bm 1 } } { 2 j } \left ( t_{12}^2 { \bm \sigma}\cdot{\bm m } + \frac { \left [ t_{12 } { \bm \sigma}\cdot{\bm n } , [ t_{12 } , { \bm \sigma}\cdot{\bm m } ] { \bm \sigma}\cdot{\bm n } \right ] } { 4 } + i t_{12 } \frac { [ t_{12 } , { \bm \sigma } \cdot ( { \bm n } \times { \bm m } ) ] { \bm \sigma } \cdot { \bm n } - [ t_{12 } , { \bm \sigma } \cdot { \bm n } ] { \bm \sigma } \cdot ( { \bm n } \times { \bm m } ) } { 2 } \right ) \nonumber \\ & & + { \cal o } \left ( j^ { - 2 } \right ) .\end{aligned}\ ] ] the expression for @xmath41 is thus given by @xmath168 with @xmath169 + \frac { \left\ { t_{12 } { \bm \sigma } \cdot { \bm
n } , \left [ t_{12 } , { \bm \sigma } \cdot { \bm n } \right ] { \bm \sigma } \cdot { \bm n } \right\ } } { 2 } \right ) , \\ { \cal f}_2 & = & - \frac { 1 } { 4 j } \left ( \frac { \left [ t_{12 } { \bm \sigma } \cdot { \bm n } , \left [ t_{12 } , { \bm \sigma } \cdot { \bm m } \right ] { \bm \sigma } \cdot { \bm n } \right ] } { 2 } + i t_{12 } \left\ { \left [ t_{12 } , { \bm \sigma } \cdot \left ( { \bm n } \times { \bm m } \right ) \right ] { \bm \sigma } \cdot { \bm n } - \left [ t_{12 } , { \bm \sigma } \cdot { \bm
n } \right ] { \bm \sigma } \cdot \left ( { \bm n } \times { \bm m } \right ) \right\ } \right ) , \\ { \cal f}_3 & = & \frac { \left [ t_{12 } , { \bm \sigma } \cdot { \bm n } \right ] { \bm \sigma } \cdot { \bm n } } { 2 } , \\ { \cal f}_4 & = & \frac { t_{12 } } { 2 j } \left ( - \left [ t_{11 } , { \bm \sigma } \cdot { \bm
n } \right ] + i { \bm \sigma } \cdot \hbar \partial_t { \bm n } \right ) + \frac { \left [ t_{12 } , { \bm \sigma } \cdot { \bm m } \right ] { \bm \sigma } \cdot { \bm n } + 2 i t_{12 } { \bm \sigma } \cdot \left ( { \bm n } \times { \bm m } \right ) } { 2 } .\end{aligned}\ ] ] @xmath170 in eq .
( [ hj2_ap ] ) can be neglected up to the first order of @xmath36 when the spatiotemporal variations of the magnetizations are as slow as @xmath171 \rangle \ll \langle t_{12
} \rangle^2 / j$ ] , @xmath172 \rangle \ll \langle t_{12 } \rangle^2 / j$ ] , @xmath173 \rangle \ll \langle t_{12 } \rangle$ ] , and @xmath174 .
approximating both @xmath175 and @xmath176 by @xmath18 , we arrive at eq .
( [ hj2 ] ) .
the commutators of the kinetic energy terms , @xmath32 and @xmath33 , and @xmath40 and/or @xmath39 give rise to a spatial derivative of @xmath40 and/or @xmath39 , because the kinetic energies are functions of @xmath177 .
the expressions of the commutators are accessible by assuming the forms of @xmath178 and @xmath179 , where the effective masses @xmath180 and @xmath181 can be deduced from , e.g. , eqs .
( [ t_lay ] ) in the case of the layered afm . setting @xmath182 and @xmath183 ev , and employing the typical value for the fermi wave number @xmath184 of the conduction electrons in metals as @xmath185 m@xmath115 , the above mentioned conditions
are well met with the spatiotemporal variation of the magnetizations considered in fig . 1 .
expand @xmath53 in powers of the operator @xmath54 by the unitary transformation in eq .
( [ ht2 ] ) or @xmath186 + \frac { [ s_t , [ s_t , { \cal h}_t ] ] } { 2 }
+ ... + i \hbar \partial_t s_t + ... , \label{ht'_ap } \end{aligned}\ ] ] with @xmath187 given in eq . (
each term in eq .
( [ ht_ap ] ) is computed up to the first order of @xmath54 as @xmath188 & = & - \gamma_5 j \left ( { \bm \sigma } \cdot { \bm n } + \frac { [ t^{-1}_{12 } , { \bm \sigma } \cdot { \bm n } ] t_{12 } } { 2 } \right ) + \gamma_0 j^2 \left ( t_{12}^{-1 } - \frac { \left [ t_{12}^{-1 } , { \bm \sigma } \cdot { \bm
n } \right ] { \bm \sigma } \cdot { \bm n } } { 2 } \right ) \nonumber \\ & & - \gamma_0 \gamma_5 \frac { j } { 2 } \left\ { t_{12}^{-1 } [ t_{11 } , { \bm \sigma } \cdot { \bm n } ] - j [ t_{12}^{-1 } , { \bm \sigma } \cdot { \bm m } ] { \bm \sigma } \cdot { \bm
n } - 2 i t_{12}^{-1 } j { \bm \sigma } \cdot ( { \bm n } \times { \bm m } ) \right\ } , \end{aligned}\ ] ] @xmath189 ] } { 2 } & = & - \gamma_0 \frac { j^2 } { 2 } \left ( t_{12}^{-1 } + \frac { { \bm \sigma } \cdot { \bm
n } [ t_{12}^{-1 } , { \bm \sigma } \cdot { \bm n } ] } { 2 } + \frac { \left\ { t_{12}^{-1 } { \bm \sigma } \cdot { \bm n } , \left [ t_{12}^{-1 } , { \bm \sigma } \cdot { \bm n } \right ] t_{12 } \right\ } } { 4 } \right ) + { \cal o}(t_{12}^{-2}).\end{aligned}\ ] ] the expression for @xmath190 is thus given by @xmath191 where @xmath192 { \bm \sigma } \cdot { \bm n } + \frac { \left\ { t_{12}^{-1 } { \bm \sigma } \cdot { \bm n } , \left [ t_{12}^{-1 } , { \bm \sigma } \cdot { \bm n } \right ] t_{12 } \right\ } } { 2 } \right ) , \\
{ \cal y}_2 & = & - \frac { j \left [ t_{12}^{-1 } , { \bm \sigma } \cdot { \bm n } \right ] t_{12 } } { 2 } , \\ { \cal y}_3 & = & t_{12}^{-1 } j \left ( \frac { - \left [ t_{11 } , { \bm \sigma } \cdot { \bm
n } \right ] + { \bm \sigma } \cdot i \hbar \partial_t { \bm n } } { 2 } + i j { \bm \sigma } \cdot \left ( { \bm n } \times { \bm m } \right ) \right ) + \frac { j^2 \left [ t_{12}^{-1 } , { \bm \sigma } \cdot { \bm m } \right ] { \bm \sigma } \cdot { \bm n } } { 2 } .\end{aligned}\ ] ] @xmath193s is eq .
( [ ht2_ap ] ) can be neglected up to the first order of @xmath54 when the spatiotemporal variations of the magnetizations are sufficiently slow that @xmath194 t_{12 } \rangle \ll t_{12}^{-2 } j^2 $ ] , @xmath195 \rangle \ll t_{12}^{-1 } j^2 $ ] , and @xmath196 .
it can be shown by the similar discussion as in the appendix b and setting @xmath197 and @xmath198 ev that these conditions are well satisfied in the systems considered in fig . 1 . approximating @xmath199 by @xmath33 , we arrive at eq .
( [ ht3 ] ) . under the adiabatic approximation where the majority ( minority ) electron spin adiabatically follows the direction of @xmath200 ( @xmath201 ) , the ( normalized ) expectation value @xmath202 of the conduction electron spin can be represented by @xmath203 where the upper ( lower ) sign corresponds to the majority ( minority ) electron , and @xmath204 is the slight deviation from @xmath205 .
assume that the electron spin obeys the following continuity equation ; @xmath206 where @xmath207 denotes the average electron velocity , @xmath208 , and @xmath209 is the relaxation time for the electron - spin flip . by substituting eq .
( [ e - spin ] ) into ( [ eom - s ] ) , the expression for @xmath210 is obtained as @xmath211 , \ ] ] where @xmath212 . the torques @xmath213 that the electron spins exert on @xmath44 are given by @xmath214 \nonumber \\ & = & - \frac { g \mu_b \left ( n_+ - n_- \right ) } { 2 \tau_{\rm ex } m_{\rm s } | { \bm m } | } { \bm m}_i \times { \bm m}_{j\neq i } - { \bm m}_i \times \left [ \hat{{\bm m } } \times \left ( { \bm u}_t \cdot \nabla \right ) \hat{{\bm m } } + \beta_t \left ( { \bm u}_t \cdot \nabla \right ) \hat{{\bm m } } \right ] , \label{stt - t_ap } \end{aligned}\ ] ] where @xmath215 is the majority ( minority ) electron density , and the stt efficiency @xmath70 is given in eq .
( [ ut ] ) . in the second equality of eq .
( [ stt - t_ap ] ) , @xmath216 and @xmath217 have been used . here , the spin polarization @xmath72 of the conduction electrons with respect to @xmath39 is defined by @xmath218 . the first term in the second equality of eq .
( [ stt - t_ap ] ) contributes the modulation to the afm coupling between @xmath24 and @xmath25 ; we absorb this first term into the definition of the afm exchange coupling .
the second terms in the second equality of eq .
( [ stt - t_ap ] ) , which are eq .
( [ stt_t ] ) , are the stts @xmath15 that act on textured afms .
consider a one - dimensional afm nanowire stretching in the @xmath77-axis with easy - axis anisotropy ( @xmath219 ) along it .
an equilibrium afm texture is determined by @xmath40 and @xmath39 at which the magnetic energy density @xmath14 , which is given in eq .
( [ w ] ) , takes an extremal value . in the absence of external field ,
a static dw solution satisfying the boundary condition @xmath220 is @xmath221 , \\
\varphi & = & 0 , \end{aligned}\ ] ] where the polar angles are defined by @xmath222 , @xmath223 represents the dw center position , and @xmath224 .
apply a dc charge current in the @xmath77 direction , and examine the current - driven dynamics of the dw by using eq .
( [ n ] ) for the exchange - dominant regime . to obtain an analytical solution for the dw dynamics ,
we make the steady - motion approximation , where the dw maintains the equilibrium profile with @xmath223 being time dependent ; the dw exhibits a translational motion described by time evolution of the collective coordinate @xmath223 . by rewriting eq .
( [ n ] ) into the equation of motion for @xmath223 by preforming the volume integral of the equation,@xcite one obtains @xmath225 as the dynamics of afm textures in general has an inertia,@xcite the above equation provides with the terminal velocity of the dw . the dissipative process described by @xmath46
is required to drive the dw by the charge current . as pointed in the sec .
iv , the same argument applies to the mixing - dominant regime with @xmath46 and @xmath226 replaced by @xmath69 and @xmath227 , respectively , when there exists a sufficiently large @xmath39 over the relevant sample region and @xmath40 and @xmath39 lie in a single plane .
when , in the mixing - dominant regime , the magnitude and direction of @xmath39 have some significant dependence on time and space that does not meet the above - mentioned conditions , it can make it difficult to obtain analytical expressions for the stt effects , which is beyond the scope of the present paper .
99 j. c. slonczewski , j. magn .
magn . mater . *
159 * , l1 ( 1996 ) ; l. berger , phys .
b * 54 * , 9353 ( 1996 ) .
d. c. ralph and m. d. stiles , j. magn .
magn . mater . *
320 * , 1190 ( 2008 ) ; a. brataas , a. d. kent , and h. ohno , nature mater . *
11 * , 372 ( 2012 ) .
t. jungwirth , x. marti , p. wadley , and j. wunderlich , nat . nanotechnol . * 11 * , 231 ( 2016 ) .
a. s. nez , r. a. duine , p. haney , and a. h. macdonald , phys .
b * 73 * , 214426 ( 2006 ) ; r. a. duine , p. m. haney , a. s. nez , and a. h. macdonald , phys . rev .
b * 75 * , 014433 ( 2007 ) ; z. wei , a. sharma , a. s. nez , p. m. haney , r. a. duine , j. bass , a. h. macdonald , and m. tsoi , phys . rev
. lett . * 98 * , 116603 ( 2007 ) ; p. m. haney , d. waldron , r. a. duine , a. s. nez , h. guo , and a. h. macdonald , phys .
b * 75 * , 174428 ( 2007 ) . p. m. haney and a. h. macdonald , phys
lett . * 100 * , 196801 ( 2008 ) ; s. urazhdin and n. anthony , phys .
lett . * 99 * , 046602 ( 2007 ) .
h. v. gomonay and v. m. loktev , phys .
b * 81 * , 144427 ( 2010 ) ; h. v. gomonay , r. v. kunitsyn , and v. m. loktev , phys . rev .
b * 85 * , 134446 ( 2012 ) .
j. linder , phys .
b * 84 * , 094404 ( 2011 ) .
h. ben mohamed saidaoui , a. manchon , and x. waintal , phys . rev .
b * 89 * , 174430 ( 2014 ) .
r. cheng , j. xiao , q. niu , and a. brataas , phys .
lett . * 113 * , 057601 ( 2014 ) ; r. cheng , m. w. daniels , j .-
zhu , and d. xiao , phys .
b * 91 * , 064423 ( 2015 )
. y. xu , s. wang , and k. xia , phys .
. lett . * 100 * , 226602 ( 2008 ) .
a. c. swaving and r. a. duine , phys .
b * 83 * , 054428 ( 2011 ) : j. phys . : condens .
matter * 24 * , 024223 ( 2012 ) .
k. m. d. hals , y. tserkovnyak , and a. brataas , phys .
lett . * 106 * , 107206 ( 2011 ) .
r. cheng and q. niu , phys .
b * 86 * , 245118 ( 2012 ) : _ ibid_. * 89 * , 081105(r ) , ( 2014 ) . e. g. tveten , a. qaiumzadeh , o. a. tretiakov , and a. brataas , phys
lett . * 110 * , 127208 ( 2013 ) . j. barker and o. a. tretiakov , phys .
lett . * 116 * , 147203 ( 2016 ) .
l. nel , ann .
( paris ) , * 3 * , no . 2 , 137 ( 1948 ) .
a. g. gurevich and g. a. melkov , _ magnetization oscillations and wave _ , ( crc press , 1996 ) .
y. yamane , j , ieda , and j. sinova , phys .
b * 93 * , 180408(r ) ( 2016 ) .
s. zhang and z. li , phys .
lett . * 93 * , 127204 ( 2004 ) .
a. f. andreev and v. i. marchenko , phys .
* 23 * , 21 ( 1980 ) .
n. papanicolaou , phys .
b * 51 * , 15062 ( 1995 ) ; _ ibid . _ * 55 * , 12290 ( 1997 )
. a. n. bogdanov , u. k. rssler , m. wolf , and k .- h .
mller , phys .
b * 66 * , 214410 ( 2002 ) . v. vlaminck and m. bailleul , science * 322 * , 410 ( 2008 ) . v. g. baryakhtar , b. a. ivanov , and m. v. chetkin , sov
* 28 * , 7 ( 1985 ) . | we formulate a theory of spin - transfer torques in antiferromagnets , which covers the small to large limits of the exchange coupling energy relative to the kinetic energy of the inter - sublattice electron dynamics .
our theory suggests a natural definition of the efficiency of spin - transfer torques in antiferromagnets in terms of well - defined material parameters , revealing that the charge current couples predominantly to the antiferromagnetic order parameter and the sublattice - canting moment in , respectively , the limits of large and small exchange coupling . the effects can be quantified by analyzing the antiferromagnetic spin - wave dispersions in the presence of charge current : in the limit of large exchange coupling
the spin - wave doppler shift always occurs , whereas , in the opposite limit , the only spin - wave modes to react to the charge current are ones that carry a pronounced sublattice - canting moment .
the findings offer a framework for understanding and designing spin - transfer torques in antiferromagnets belonging to different classes of sublattice structures such as , e.g. , bipartite and layered antiferromagnets . |
understanding the origin of the cosmic ray ( cr ) energy spectrum is central to astrophysics and has long been the focus of intensive study .
the spectrum of cr nuclei at @xmath010@xmath110@xmath2gev / nucleon is thought to be the result of diffusive shock acceleration ( dsa ) mechanisms in supernova remnants ( snr ) , followed by diffusive propagation in the interstellar medium ( ism ) @xcite .
the conventional descriptions predict source spectra such as @xmath3 for primary nuclei ( _ e.g. _ , , ) which are steepened as @xmath4 by diffusion .
the data constrain @xmath52.7 ( depending on the element ) and @xmath60.20.7 , whereby @xmath72.02.5 . in dsa calculations ,
the source spectral slope should be @xmath82.2 , which implies @xmath90.5 .
on the other hand , large values for @xmath10 are at odds with the anisotropy observations at tev energies , which favor @xmath110.3 . on top of that , recent experiments atic-2 , cream , and pamela , reported a remarkable spectral hardening at energies above @xmath0100gev / nucleon which can not be explained by conventional mechanisms @xcite .
proposed explanations of this effect deal with acceleration mechanisms @xcite , nearby snrs @xcite or multi - source populations @xcite .
i propose here that the hardening originates from a spatial change of the cr diffusion properties in the different regions of the galactic halo .
the key hypotesis is that the diffusion coefficient @xmath12 is not separable into energy and space terms as usually assumed .
as i will show , this hypotesis leads to a pronounced change in slope for the energy spectra of cr nuclei at earth , giving a good description of the present data .
remarkably , the model proposed here has also positive impacts on several open problems in the cr acceleration / propagation physics .
i use a one - dimensional inhomogeneus diffusion model for cr transport and interactions @xcite .
the galaxy is modeled to be a disc ( half - thickness @xmath13100pc ) containing the gas ( number density @xmath141@xmath15 ) and the cr sources .
the disc is surrounded by a diffusive halo ( half - thickness @xmath165kpc ) with zero matter density . for each stable cr nucleus , the transport equation is : @xmath17 where @xmath18 is its number density as function of the @xmath19-coordinate , @xmath20 is the position - dependent diffusion coefficient and @xmath21 is the destruction rate in the ism at velocity @xmath22 and cross section @xmath23 .
the source term @xmath24 is split into a primary term @xmath25 , from snrs , and a secondary production term @xmath26 , from spallation of heavier ( @xmath27 ) nuclei with rate @xmath28 .
the quantities @xmath29 , @xmath12 , @xmath24 and @xmath30 depend on energy too .
equation[eq::transport1d ] can be solved in steady - state conditions ( @xmath31 ) with the boundary conditions @xmath32 @xcite . for each cr nucleus
, the differential energy spectrum is given by : @xmath33 \,,\ ] ] where @xmath34@xmath35@xmath36 , @xmath37 , and @xmath38@xmath35@xmath39 . in cr propagation studies ,
the diffusion coefficient is usually assumed to be separable as @xmath40 ( in homogeneus models @xmath411 , whereby @xmath42 ) . in these models ,
the functions @xmath43 and @xmath44 are independent on energy , so that the predictions at earth ( @xmath19@xmath45@xmath46 ) are spectrally uninfluenced by the choice of @xmath47 .
conversely , i employ a non - separable @xmath48 , which reflects the existence of different cr transport properties in the propagation volume , depending on the nature and scale distribution of the magnetic - field irregularities .
in fact , while snr explosions may generate large irregularities in the region near the galactic plane , the situation in the outer halo is different because that medium is undisturbed by snrs . from these considerations , the authors of ref .
@xcite found that the turbulence spectrum in the halo should be flatter than that in the galactic plane .
this implies a strong latitudinal dependence for the parameter @xmath10 , which suggests spatial variations of the cr energy spectra .
noticeably , new data reported by _
fermi_/lat on the diffuse @xmath49-ray emission at @xmath010100gev of energy seem to support these suggestions : the @xmath49-ray spectra observed near the galactic plane ( latitude @xmath508@xmath51 ) are found to be harder than those at higher latitudes @xcite . following the above arguments ,
i adopt a simple _ two - halo model _ consisting in two diffusive zones .
the _ inner halo _ is taken to surround the disk for a typical size @xmath52 of a few hundred pc ( @xmath530.1 ) .
its medium properties are influenced by snrs , which produce a steep turbulence spectrum in terms of energy density per wave number , @xmath54 , presumably close to the kolmogorov regime @xmath551/3 .
the _ outer halo _ represents a wider region , @xmath56 , which is undisturbed by snrs .
its turbulence spectrum is driven by crs and should be flatter .
for instance , crs with rigidity spectrum @xmath57 ( where @xmath58 ) excite turbulent modes of wave - numbers @xmath59 , giving a spectrum @xmath0@xmath60 , where @xmath61 is of the order of the unity .
this situation can be realized by a rigidity dependent diffusion coefficient of the type @xmath62 where @xmath630.04kpc / myr specifies its normalization at the reference rigidity @xmath645gv . for @xmath65 , the diffusion coefficient of eq.[eq::twohalodiffcoeff ]
produces a higher cr confinement in the inner halo ( with @xmath551/3 ) , whereas the outer halo ( with @xmath660.5 1 ) represents a _ reservoir _ from which crs leak out rapidly and can re - enter the inner halo .
the non - separability of @xmath12 has a remarkable consequence on the model predictions at @xmath67 .
it can be understood if one neglects the term @xmath30 and takes a source term @xmath68 . from eq.[eq::solutionvsz ] and using eq.[eq::twohalodiffcoeff ] , one finds @xmath69 which describes the cr spectrum as a result of two components .
its differential log - slope as a function of rigidity reads @xmath70 which shows a clear transition between two regimes . in practice the low - energy regime ( @xmath71 )
is never reached due to spallation , that becomes relevant and even dominant over escape ( @xmath72 ) . in this case
the log - slope is better approximated by @xmath73 @xcite .
the hard high - energy regime ( @xmath74 ) is determined by the diffusion properties of the inner halo only . in this limit
one has @xmath75 .
the effect vanishes at all rigidities when passing to the homogeneus limit of @xmath76 ( one - halo ) or @xmath77 ( identical halos ) , where one recovers the usual relation @xmath78 . from eq.[eq::solutionvsz ] , it can be seen that the intensity of the harder component diminishes gradually with increasing @xmath79 , _
i.e. _ , the cr spectra at high energies are steeper in the outer halo , as suggested by the new
_ fermi_/lat data . .
, title="fig:",scaledwidth=45.0% ] . , title="fig:",scaledwidth=45.0% ] figure[fig::ccresults ] shows the and spectra at earth , from eq.[eq::solutionvsz ] , at kinetic energies above 10gev .
the implementation of the model follows refs.@xcite .
the source spectra are taken as @xmath80 , with @xmath812.3 for @xmath82 and @xmath812.16 for @xcite .
the two halos are defined by @xmath165kpc and @xmath830.5kpc .
a kolmogorov - type diffusion is adopted for the inner halo ( @xmath841/3 ) , while @xmath85 is taken as 0.55 , consistently with ref.@xcite .
results from the two - halo model ( solid lines ) are compared with those from the homogeneus model ( dashed lines ) , which uses @xmath840.6 , and with recent cr data .
below @xmath010gev / nucleon , the solar modulation is apparent and it is described using a _ force - field _ modulation potential @xmath86400mv @xcite . note that my model may be inadequate in this energy region due to approximations .
remarkably , my model reproduce well the observed changes in slope at @xmath87100gev , in good agreement with the trend indicated by the data .
it should be noted , however , that the sharp spectral structures suggested by the pamela data at @xmath0300gev can not be recovered .
figure[fig::ccresults]c shows the mean logarithmic mass , @xmath88 , which is described well by the two - halo model .
figure[fig::ccresults]d shows the ratio from the two - halo model , which is predicted to harden as @xmath89 , while its low - energy behavior is similar to that of the homogeneus model ( with @xmath900.6 in the whole halo ) .
interestingly , a slight flattening for the ratio is also suggested by recent data , _
e.g. _ , from tracer . a multi - channel study of the data will allow to resolutely test these features .
the ratio hardening is also connected with the global large - scale anisotropy amplitude , which should increases as @xmath91 @xcite .
thus , the two - halo model predicts a gradual flattening of @xmath92 , as indicated by the present data .
it is also interesting noticing that the anisotropy may be furtherly reduced at all energies if one accounts for a proper radial dependence for @xmath12 @xcite . in summary ,
the model i proposed is potentially able to reconcile a weak energy dependent of the anisotropy amplitude ( as suggested by observations ) with relatively hard source spectra @xmath932.2 ( as preferred by the dsa theory ) , giving good fits to the ratio .
i recall that plain diffusion models employ @xmath932.2 and @xmath940.5 in the whole halo ( which gives a too strong energy dependence for @xmath92 ) , whereas diffusive - reacceleration models use @xmath932.4 and @xmath941/3 ( which are too challenging for the dsa and require strong reacceleration to match the data ) .
i have shown that the spectral hardening observed in crs may be consequence of a galactic diffusion coefficient that is not separable in energy and space terms . from the model presented here ,
the hardening arises naturally as a local effect and vanishes gradually in the outer halo , where the cr spectra are also predicted to be steeper .
this effect must be experienced by all cr nuclei as well as by secondary - to - primary ratios .
recent data from _ fermi_/lat and tracer seem to support this scenario , but the change in slope predicted by this model is more gradual than that suggested by pamela data .
all these features can be tested with the data forthcoming by , so that the present model will be discriminated against other interpretations .
this scenario has an interesting impact on the cr physics .
as shown , a kolmogorov diffusion for the inner halo ( @xmath551/3 ) is consistent with relatively hard source spectra ( @xmath72.2 ) , giving good descriptions of both the ratio and the high - energy trend of the anisotropy amplitude .
+ + i acknowledge the support of the _ italian space agency _ under contract asi - infn i/075/09/0 . | recent data from atic , cream and pamela revealed that the energy spectra of cosmic ray ( cr ) nuclei above 100 gev / nucleon experience a remarkable hardening with increasing energy .
this effect can not be recovered by the conventional descriptions of cr acceleration and diffusive propagation processes . using analytical calculations ,
i show that the hardening effect can be consequence of a spatial change of the cr diffusion properties in different regions of the galaxy .
i discuss the implications of this scenario for the main cr observables and its connections with the open issues of the cr physics . |
one of the most thoroughly explored subjects in all of science is that of thermal relaxation of magnetisation , in magnetic systems of all shapes and sizes .
investigations in this area ( which go back many centuries @xcite ) have revealed many subtleties , and even today there are many unsolved puzzles ( e.g. , the physics of `` magnetic avalanches '' ) .
very recently a whole new set of questions in this area has arisen , with the advent of well - characterised `` nanoscopic '' magnetic structures @xcite .
such structures include `` made - to - order '' magnetic grains , magnetic wires and superlattice arrays , as well as thin films and spin chains .
there are very obvious applications of such nanomagnets in , e.g. , the recording industry , as well as in information transmission and computing .
theoretical activity has been particularly intense on 1-dimensional spin systems @xcite , and on the tunneling and coherence phenomenon which should exist in both grains ( involving `` giant spin '' @xcite dynamics ) and in the dynamics of domain walls @xcite .
both the theory and the experimental activity in the latter field have been reviewed recently @xcite . in this paper
we focus on something rather different - the `` quantum relaxation '' of magnetisation that must occur in any small magnetic particle , once thermally - activated relaxation has ceased .
although there does not seem to have been a serious theoretical analysis of such relaxation before , the conventional view has been that it must proceed via a simple tunneling process ( dissipative or otherwise ) .
the picture we shall derive here is more subtle , and contains some rather fascinating new physics .
tunneling is certainly involved ; but we find that what really controls relaxation is the nuclear spin system inside the magnetic particle , often with the help of phonons or electrons .
one can put the result in the following way : the nuclear spins provide the essential `` switch '' that turns on ( or turns off ) the quantum transitions .
as far as we are aware , there has been no previous recognition that nuclei had any role whatsoever to play in magnetic relaxation , despite the enormous number of experiments in this area . nor
do we know of any theoretical analysis of the influence of electrons on magnetic relaxation .
there has certainly been no attempt at a theory dealing with the combined effects of nuclear spins , phonons , and electrons , which is what we do here .
as we shall see , it is impossible to give a realistic analysis without putting them all together .
we shall not try here to address experiments in any detail , for two reasons .
first , each different experimental system has its own idiosyncracies , and we shall see that our theory has many different parameters - it is simply not possible to give detailed results for all possible cases , and we feel that it would be preferable to examine particular systems for which parameters are well known , as test cases .
this we intend to do elsewhere .
second , in most experiments on magnetic grains , the grains are rather close together .
as we shall briefly explain in this paper , this makes it probable that even the low - t relaxation of an experimental ensemble of grains proceeds via dipolar interactions between the grains . the theory in this paper is closely related to previous work on quantum coherence in magnetic grains @xcite . in that work
a very detailed analysis was given of the way nuclear spins control the dynamics of `` giant spins '' , i.e. , spins with quantum number @xmath1 .
we also analysed the effect of phonons and electrons in `` decohering '' the quantum motion of @xmath2 .
the problem of quantum relaxation is different , in that even if coherence is destroyed , relaxation may still proceed incoherently .
what we shall find is that in the absence of nuclear spins , phonons can not relax @xmath2 in a small biasing field , because they couple too weakly to it ; on the other hand electrons couple so strongly to @xmath2 that they freeze it completely .
the nuclear spins can then _ liberate _ @xmath2 - amusingly , the mechanisms by which they do this are the same as those by which they _ destroy _ the quantum coherent motion of @xmath2 . the bias field @xmath3 acting on @xmath2 , generated by the combined hyperfine fields of all
the nuclei can easily be greater than @xmath4 ( e.g. , for tbfe@xmath5 grains ) , and depending on the size of the grain may be equivalent to an external field from @xmath6 up to well over @xmath7 ! this bias changes with time in a diffusive way , at a rate governed by inter - nuclear dipolar interactions ( in the absence of phonons and electrons ) , and also nuclear spin - phonon or spin - electron interactions , when the latter are present .
every so often , as @xmath8 shunts around , it brings quantum 2 states corresponding to different orientations of @xmath2 into near degeneracy . at this point
the system can tunnel , provided ( a ) the nuclear bias field @xmath9 gives it time to do so coherently , ( b ) the usual dissipative effects of the electrons and phonons are not too destructive .
on the other hand , we shall also see that the nuclear spins can also further _ inhibit _ the relaxation of @xmath2 , at low @xmath10 .
if the nuclear spins are allowed to come to equilibrium , then at temperature below the hyperfine coupling energy , they exert a large negative bias field on @xmath2 , which essentially traps @xmath2 for astronomical times .
thus , depending on @xmath10 and on the nuclear polarisation , the nuclei can enhance or suppress relaxation .
although our purpose in this paper is not to deal with the implications for magnetic technology involving quantum relaxation , we feel we should at least make some remarks here about what they might be .
fig.1 shows the way in which magnetic computer memory elements have decreased exponentially in size in the last 40 years - such plots have figured large in recent discussions of the need for new kinds of computer design , incorporating elements which operate quantum - mechanically . at present computers , as well as magnetic tapes , use elements which behave classically , and are stable over long periods of time .
this stability over decades exists because they are _ big _ - the energy barrier between 2 states of the element , proportional to its size , usually exceeds @xmath11 even at room temperature . the hypothesized `` quantum threshold '' ,
below which tunneling is important even at @xmath12 , is usually supposed to be for grains containing roughly @xmath13 spins , depending on the material involved .
however we shall find that the threshold between stable behaviour and quantum relaxation can be moved around a lot , depending on how the magnetic system is coupled to the nuclear spins , and whether or not it is coupled to electrons .
we shall see that in some cases , one can freeze the dynamics of even microscopic magnetic systems ( containing only 10 spins ) , for times of thousands of years or more .
conversely , we shall see how it is possible for much larger systems ( containing @xmath14 electronic spins or more ) to relax very quickly ( in @xmath15 ) .
these theoretical insights may have some relevance to the design of future nanoscopic magnetic devices . in
what follows we shall analyse the role of electrons ( section [ sec:2 ] ) , phonons ( section [ sec:3 ] ) , and nuclei ( section [ sec:4 ] ) , acting in isolation on @xmath2 . then
in section [ sec:5 ] we shall put them together , to give the final rather complex picture . as noted above
, we save detailed discussion of experiments for other papers ; but in section [ sec:6 ] we will indicate the general nature of our predictions .
we briefly describe the giant spin model here , and the various energy scales relevant to the physics .
the basic idea behind this model has been explained in detail elsewhere @xcite .
the exchange couplings @xmath16 between electronic spins in the grain are enormous compared to the anisotropy couplings @xmath17 ( typically @xmath18 , whereas @xmath19 ) .
the mesoscopic grains can be treated as giant rotators , keeping only states @xmath20 , with @xmath21 .
the spectrum of this giant spin is shown in fig.1 in a small bias field @xmath22 . sometimes , as a model example
, we shall use the easy - axis / easy - plane system , having hamiltonian @xmath23 -\gamma \vec{s } \cdot \vec{h}_o \;. \label{1.1}\ ] ] we will usually assume @xmath24 , so @xmath25 . the low energy physics of ( [ 1.1 ] ) can be understood in terms of the truncated hamiltonian @xmath26 where the tunneling splitting @xmath27 is @xmath28 and the bare parameters are @xmath29 @xmath30 the `` bounce frequency '' @xmath31 is also roughly the distance to the next pair of excited levels ; the total barrier height in ( [ 1.1 ] ) , between the 2 semiclassical minima at @xmath32 , is @xmath33 . thus if , say , @xmath34 , we might easily have @xmath35 , a barrier height @xmath36 , but a splitting @xmath37 , or less .
this is certainly not the whole story .
magnetic grains having more complicated symmetries are not so easily described using wkb / instanton methods .
the `` internal '' magnon modes of the giant spin are also neglected , except for those uniform precession modes corresponding to @xmath20 .
there will also be phonons ( spectrally very weak at these low energies ) and possibly electrons ( which will mix strongly with all states of the grain ) .
finally and most importantly , the nuclei in the grain ( and also possibly outside ) will couple via hyperfine interactions to each level shown in fig.2 .
the net result of this is shown in fig.3 for the 2 lowest levels in the presence of a bias field .
typically @xmath38 , where @xmath39 is an _
individual _ hyperfine coupling ; however @xmath40 nuclei will spread out a single grain level into a mass of @xmath41 levels , of gaussian half - width @xmath42 , where @xmath43 is the principal hyperfine splitting . for rare earths ,
where @xmath44 or greater , this half - width can easily exceed @xmath31 for mesoscopic grains , and the whole `` giant spin '' model begins to get rather complex . in this paper
we will ignore all physics at energies @xmath45 or greater , simply including it into renormalisation parameters of a low-@xmath10 effective hamiltonian .
for such a treatment to be valid , we then require @xmath46 we will also assume that any `` loose spins '' on the surface of the grain still have couplings @xmath47 to @xmath2 . if not , they will be treated as part of the `` spin bath '' environment , along with nuclei and paramagnetic impurity spins . before beginning , it is useful to bear in mind 2 simple points
. note first that in the presence of any bias @xmath22 which is greater than @xmath48 in ( [ 1.2 ] ) , the system is essentially `` frozen '' in one or other of the states @xmath49 or @xmath50 .
suppose , e.g. , the system starts in the higher - energy state @xmath49 at @xmath51 .
then the probability @xmath52 of finding it @xmath49 at time @xmath53 , in the absence of any other couplings , is given by @xmath54 where @xmath55 .
thus the coupling to phonons , electrons , and nuclei is necessary if we are to have any relaxation at all .
usually one thinks of this as a kind of spontaneous or stimulated emission , perhaps in the presence of dissipation , in which the requisite energy is taken up by a bath of oscillators .
in fact this conventional picture of tunneling of a biased 2-level system , coupled dissipatively to a bath of oscillators @xcite , works reasonably well for a giant spin coupled to either electrons or phonons .
this we shall see in detail in the next two sections .
however as soon as we introduce the spin bath of nuclei , etc . , it breakes down completely ; this is basically why our final results are so surprising .
the second point arises from the finite - dimensional nature of the giant spin hilbert space .
as emphasized by van hemmen and suto @xcite this means that we must use wkb calculations of tunneling rates with great care when discussing the motion of @xmath2 .
this point is especially clear in recent work of politi et al .
@xcite ( see also villain et al .
@xcite ) , analysing the mn@xmath56o@xmath56 acetate systems ; here tunneling appears to proceed via a 4th - order term @xmath57 in @xmath58 .
consequently transitions may only proceed between states differing by @xmath59 .
to correctly handle such selection rules in the instanton calculus is not a trivial problem .
even more important , there is no continuum of final states in this tunneling problem , and tunneling may only proceed if there is near resonance between initial and final states .
this means that in the presence of a bath of , e.g. , phonons , one may _ not _ in general simply apply the caldeira - leggett tunneling formalism @xcite to calculate tunneling , since this formalism assumes a continuum of final states . for this reason calculations such as those of garg and kim @xcite , of the tunneling rate of a strongly biased grain interacting with phonons ,
can only be correct when @xmath60 . for finite @xmath61
one must take account of the discrete nature of the spectrum of @xmath61 , at least when the level spacing is greater than the ( very small ) linewidth caused by phonon damping ( note further that calculations of grain tunneling in the presence of a nuclear spin bath should _ never _ use the caldeira - leggett method , since the spin bath can not be mapped onto an oscillator bath of the feynman - vernon @xcite type - thus calculations of this kind @xcite are not valid for any @xmath61 ) .
we now turn to the detailed treatment of the different relaxation mechanisms .
readers wishing to see only a summary of the results should proceed directly to section [ sec:6 ] .
in this section we develop a model for the coupling between the giant spin and the electron bath only , and use it to calculate relaxation rates for the grain magnetisation .
since , as far as we are aware , no attempt has ever been made to deal with the coupled giant spin / electron system before , it is clear that we must begin from first principles .
the full development of our model , including detailed discussion of the multiple - scattering of the electrons by the individual spins inside the grain , the effects due to surface spins , to the discreteness of the electronic spectrum inside the grain if it is on an insulating substrate , the interaction between electrons and the internal degrees of freedom ( magnons ) in the grain , etc .
, will not appear in this paper .
we shall also set aside questions about how conduction electron dynamics _ outside _ the grain , in a conducting substrate , will be influenced by the giant spin dynamics .
this is a very important question for any experiment ( which can use the substrate magnetoresistance as a probe of the grain dynamics ) , but is too complicated to deal with here .
there are 3 physical situations which we will address here .
the first involves conducting electrons both inside and outside the grain ; we shall see that this problem can be analysed fairly clearly .
likewise the problem of an insulating grain interacting with conducting substrate can be modelled in a reasonably treatable way .
the problem of a conducting grain and an insulating substrate is more messy , however we can fairly easily extract the essential results .
our essential result is that unless the substrate is insulating , electrons simply freeze @xmath2 , unless @xmath61 is very small .
we emphasize that in reality , nuclear spins must be included ( sections [ sec:4 ] and [ sec:5 ] ) .
we shall start by assuming that the giant spin , contained within a volume @xmath62 , interacts with an electron fluid which permeates freely through the boundary between grain and substrate ; i.e. , a hamiltonian @xmath63 describing short - range exchange interaction between the localised individual spins @xmath64 , located at positions @xmath65 inside the grain , and conduction electrons in momentum @xmath66 .
a more refined model would couple grain electrons in eigenstates @xmath67 to substrate electrons @xmath66 , via a transfer matrix @xmath68 across the boundary .
however it is intuitively obvious that this is unnecessary once @xmath69 , where @xmath70 is the typical level spacing between internal electron states of the grain . if the electronic bandwidth is @xmath71 , then @xmath72 , and for most metals @xmath73 ; thus if @xmath74 , @xmath75 , so that an insulating layer at least @xmath76 thick would be required to make such a model necessary .
we assume in using ( [ 2.1 ] ) that the internal grain spins are locked together to form a giant spin , and that the @xmath77 already take account of the virtual mixing between internal excited states of the magnetic ions and of the giant spin ( magnons , surface modes ) and the electron bath .
thus , as usual , we shall be working at low @xmath10 and small @xmath78 .
however we shall assume that the @xmath77 can vary ; near the surface of the grain boundary they may be a little weaker .
the present model does not include `` loose spins '' , i.e. , surface spins which are more weakly coupled to the motion of @xmath2 ( the reader should not confuse the kondo coupling @xmath77 here with the inter - spin exchange @xmath16 between spins inside the grain ) .
these will be discussed in section [ sec:4 ] onwards .
typically the kondo couplings @xmath79 .
we now rewrite ( [ 2.1 ] ) , incorporating the giant spin hypothesis , to get a volume averaged interaction @xmath80 @xmath81 where the form factor @xmath82 integrates the number density @xmath83 of microscopic spins over the grain volume , and @xmath84 is the mean value of the @xmath77 .
@xmath85 is a sort of `` giant spin kondo hamiltonian '' , and as such contains a lot of interesting physics , much of which we will ignore in this paper .
recall that the kondo coupling for microscopic spins is just ( [ 2.2 ] ) with @xmath83 in ( [ 2.3 ] ) restricted to a single lattice cell ; for this problem one conventionally defines a dimensionless coupling @xmath86 for most metals , where @xmath87 is the fermi surface density of states .
the kondo problem can be mapped onto an ohmic spin - boson problem of the kind discussed by leggett _ et al_. @xcite ; even for this spin-@xmath88 system , there is a large dimensionless ohmic coupling to the electron bath ; @xmath89 .
it is possible to start with ( [ 2.1 ] ) and build up an analogous description by considering the individual kondo scattering from each spin , in many different orbital channels .
here we shall finesse entirely this move by instead integrating out the electrons directly , starting from ( [ 2.2 ] ) .
we first truncate directly to a biased spin - boson hamiltonian of form ( for @xmath90 in ( [ 1.1 ] ) ) : @xmath91 where the bias field is parallel to the easy axis ( we shall ignore transverse fields here ) . the caldeira - leggett spectral function for this problem
is @xmath92 and the ohmic coupling constant @xmath93 is determined by a standard fermi surface average @xcite over electrons , i.e. , @xmath94 the naive guess that @xmath95 , apparently confirmed by ( [ 2.6 ] ) , is however wrong ; carrying out the average , one gets @xmath96 this at first surprising result is a consequence of interference between kondo scattering at different sites - notice that @xmath97 , so that the problem is a multiple scattering one .
a related point is that the estimate of @xmath93 in ( [ 2.7 ] ) looks perturbative - it is not clear it should work for large @xmath93 .
however for a grain of radius @xmath98 some @xmath99 orbital channels are playing role in electron scattering . if @xmath100 , the phase shifts in each channel are small ( one has @xmath101 ) , making it possible to apply a perturbation theory , yet the sum over all the channels can be rather large .
the case of large phase shifts is more complicated , and may result in a giant spin dynamics different from that described by the standard ohmic model .
we address this possibility in appendix a. for mesoscopic spins ( @xmath102 ) the coupling in ( [ 2.7 ] ) is very large ; we may take over standard results from the spin - boson problem to discuss its effect on the dynamics of @xmath2 .
since @xmath103 , coherence at zero bias is destroyed at any temperature ; at @xmath12 the grain dynamics are rigidly locked , and the spin @xmath2 can not move . at finite @xmath10 and @xmath104 , the system fluctuates incoherently between the states @xmath105 and @xmath106 at a rate @xmath107 ; on the other hand if @xmath108 , one has incoherent relaxation at a rate @xmath109 .
the general result for the relaxation rate is @xcite @xmath110 \mid^2 \over \gamma [ 2\alpha_b ] } \ ; , \label{2.8}\ ] ] where @xmath111 $ ] is the gamma function ; for @xmath112 , this gives @xmath113 \over \gamma [ 2\alpha_b ] } \left ( { 2 \pi t \over \omega_o } \right)^{2\alpha_b-1 } \;.
\label{2.9}\ ] ] on the other hand for very low @xmath10 , such that @xmath114 , it gives @xmath115 } \left
( { \epsilon \over \omega_o } \right)^{2\alpha_b-1 } \;. \label{2.10}\ ] ] the crucial thing to notice here is that for microscopic spins ( @xmath116 ) , there will be easily observable relaxation of @xmath2 . on the other hand
if @xmath61 is much greater than @xmath117 , the giant spin will be frozen completely , even over astronomical times , unless either bias or temperature is large ( @xmath118 ) . moreover , in this model we would expect that applying either a large bias or temperature would lead to a sudden `` switching on '' of the relaxation at some critical bias or temperature ( however recall that our model is not strictly valid for @xmath118 , since higher levels of @xmath2 will then enter into the picture ) . to see this , consider 2 examples .
first , suppose that @xmath119 and @xmath120 , so @xmath121 .
then ( [ 2.9 ] ) shows that for bias @xmath104 , we have @xmath122 , and if @xmath123 , we have @xmath124 . since @xmath125 might be something like @xmath126 , this implies low @xmath10 or low @xmath8 relaxation times of fractions of a second . on the other hand , consider a second example where @xmath119 but @xmath127 , so @xmath128 .
then if @xmath104 , we find that @xmath129 , and if @xmath12 , we find @xmath130.then if @xmath104 we see that the relaxation suddenly switches on at a temperature @xmath131 ( if we are interested in a relaxation time of order seconds ) ; for temperatures below this , relaxation times increase catastrophically ( if @xmath132 , we find @xmath133 years ! ) .
the same happens as a function of bias when @xmath134 .
obviously if we go to even larger grains , with @xmath135 , this switching process becomes almost a step function of temperature .
although our 2-level model starts to break down once @xmath136 , this will not affect the validity of this switching result , or our calculation of the switching temperature .
we have not included nuclear spins in these calculations - in section [ sec:5 ] we shall see how a combination of nuclear spins and electrons affects the dynamics of @xmath2 .
the insulating grain can be analysed in a similar way , provided we gloss over some of the complications arising from the variation of both the electron density and the spin properties in the vicinity of the surface .
thus we naively extend the previous model by writing an effective interaction hamiltonian @xmath137 @xmath138 where now the form factor takes account of the decay of the electron density @xmath139 as one penetrates into the sample . by the same manoeuvres as before this leads to an ohmic coupling @xmath140 where @xmath141 , but is multiplied by some `` scale factor '' which describes the depth to which the electrons penetrate into the grain ; clearly it can vary widely .
the reduction to only near surface spins makes @xmath142 for large @xmath61 , but from ( [ 2.13 ] ) we see that @xmath143 if @xmath144 . for smaller grains ,
i.e. , @xmath145 , one can have @xmath146 , and then results different from the `` frozen spin '' behaviour of ( [ 2.8])-([2.10 ] ) are obtained .
we may define a renormalised splitting @xmath48 , and a damping rate @xmath147 , by @xmath148 @xmath149 at @xmath12 one can get @xcite coherent motion if @xmath150 ; however any such coherence is destroyed once @xmath151 , which even for microscopic values of @xmath61 ( i.e. , @xmath152 ) will occur at extremely low temperatures , usually much less then @xmath153 . for higher @xmath10 we find , when @xmath104 and @xmath146 , that @xmath154 is still given by ( [ 2.9 ] ) , which in terms of @xmath48 is @xmath155 \over \gamma [ 2\alpha_s ] } \left ( { 2 \pi t \over \delta_s }
\right)^{2\alpha_s-1 } \;. \label{2.16}\ ] ] for finite @xmath3 and @xmath10 , the form ( [ 2.8 ] ) is still valid , as is ( [ 2.10 ] ) when @xmath108 . from all of these results
we conclude that for low @xmath10 and @xmath3 , the relaxation dynamics of mesoscopic grains will be extremely slow in the presence of a conducting substrate .
thus , although our models have certainly not included all possible nuances of the surface physics , one conclusion is very clear and unlikely to be affected by further refinements , i.e. , that for @xmath156 , mesoscopic magnetic grains on conducting substrates will have their spin dynamics completely frozen . however once either @xmath10 or @xmath3 is of order @xmath31 or greater , these dynamics will rapidly be liberated .
again , these statements will be modified if nuclear spin effects are important ( section [ sec:5 ] ) . in the opposite extreme when the substrate is insulating
, the character of the electron states is quite different - they form a set of discrete states @xmath157 inside the grain , with mean spacing @xmath158 for bandwidth @xmath71 .
we describe these states via creation and annihilation operators @xmath159,@xmath160 ; these states will be rather complicated , even in the absence of spin - orbit coupling .
it is very useful to separate the exchange scattering interaction into bulk and surface parts , where now the bulk part sums over the whole sample , assuming that all the @xmath77 are the same , whilst the surface part describes the _ difference _ between @xmath161 and the true interaction hamiltonian @xmath162 ; thus if @xmath163 with the @xmath164 creating electron on site @xmath165 , we write @xmath166 and the surface term @xmath167 we notice that the bulk term is diagonal in both the site representation and the exact eigenstate representation , because of completeness of states .
this is important , because the condition @xmath168 means that the electron spins inside the system will rotate rigidly with @xmath2 , and from the last form in ( [ 2.18 ] ) we see that there is no way this adiabatic rotation can excite the low energy orbital states @xmath169 of the system .
thus the effect of @xmath161 is simply to renormalise the moment of inertia of the giant spin .
on the other hand the surface term does have off - diagonal matrix elements @xmath170 and this may result in infrared electron - hole pair production when @xmath2 flips .
the calculation of the corresponding coupling parameter @xmath171 proceeds in exactly the same way as in the previous subsection ( the corrections to the plane - wave calculation are small in a large grain with @xmath172 ) .
thus at high temperatures the internal conduction electrons constitute an ohmic bath with @xmath173 ( here @xmath174 is given by @xmath175 averaged over the surface ) .
now we notice that this result holds only at @xmath176 . at lower temperature the effective coupling constant goes to zero exponentially with @xmath10 : @xmath177 and the infrared renormalisation of the tunneling rate stops at @xmath178 if @xmath179 one has @xmath180 , and we do not have to worry about conduction electrons in the grain at all because they will rotate adiabatically with @xmath2
. the onset of coherent motion in zero bias can be found by comparing the damping rate @xmath181 with @xmath48 .
expressing all the parameters in terms of @xmath61 we find the criterion of coherence in the form @xmath182 for experiments at @xmath183 temperatures and small bias @xmath184 , this means that electronic dissipation effects can be ignored in such grains provided @xmath185 .
on the other hand for @xmath10 and/or @xmath186 , we return to the case of strong ohmic dissipation which looks very much the same as that for an insulating grain on a metallic substrate .
in the absence of electrons ( and of nuclear spins ) relaxation in a biased system proceeds via phonons .
this is a very slow process - typical relaxation times are of order months .
a number of previous studies have appeared ; garg & kim @xcite and chudnovsky @xcite have discussed grain tunneling within a straightforward caldeira - leggett approach , assuming a continuum of final states available for the tunneling - we have already explained problems with this approach in the introduction .
et al_. @xcite have given a very thorough discussion of the effects of phonons on tunneling in the mn@xmath56o@xmath56 acetate system , including the effects of discrete initial and final states , and the non - diagonal couplings ( to be defined below ) .
they also take proper account of the symmetries and selection rules in the tunneling transitions .
whether their theory explains the relaxation measurements in mn@xmath56o@xmath56 acetate @xcite is another matter - in our opinion the explanation may have more to do with dipolar interactions between the molecules @xcite .
the treatment of phonon effects we shall give is more general although quite close to theirs .
we first discuss the form the low - energy effective hamiltonian should take , depending on the symmetries of @xmath187 and the direction of the applied bias field , and determine the relaxation of the giant spin dynamics for the different possible couplings .
the values of the relevant couplings are then estimated - no attempt is made at exact calculations , since these would depend so much on the detailed nature of the system in question .
our main result is that these couplings are so weak that phonons by themselves are incapable of causing anything but very slow relaxation of @xmath2 .
we start by assuming , for ease of discussion , a simple easy - axis , easy - plane bare giant spin hamiltonian in a bias ( eqs.([1.1 ] ) , ( [ 1.2 ] ) ) .
the phonons are described by @xmath188 , with @xmath189 ( we take account of polarization later , and ignore optical modes ) , and the general form of the truncated interaction is then @xmath190 \ ; ; \label{3.1}\ ] ] @xmath191 where @xmath192 is the transverse bias .
it is unusual to consider non - diagonal couplings ( i.e. , couplings in @xmath193 ) , since in general their effects are reduced , compared to those of diagonal properties , by a factor @xmath194 .
however we shall see that in this system the diagonal couplings are zero unless we apply a field perpendicular to the easy axis .
the reason for this is very simple - time reversal symmetry implies that the magnetoacoustic interaction can not distinguish between grain states @xmath195 and @xmath196 when @xmath197 ( a point which also arises in the interaction of phonons with domain walls @xcite ) , and diagonal couplings describe interactions which do distinguish such states . applying a transverse field projects a component of @xmath195 on @xmath196 , and enables such a coupling ; from this argument we see that @xmath198 is proportional to the transverse field @xmath199 .
we shall later see that @xmath200 for a field applied in the easy plane . before deriving ( [ 3.1 ] ) and evaluating the couplings ,
we first give the results for relaxation rates that derive from it .
just as for the electron bath , these rates are very low , but for the opposite reason - instead of being frozen by a very strong coupling , @xmath2 is frozen because the effective coupling is so weak ! the relaxation rates are derived by first defining caldeira - leggett spectral functions @xmath201 where @xmath202 ; these describe the phonon effect on grain dynamics via @xmath203nd order perturbation theory .
it has of course been known for decades that these couplings are very weak for @xmath204 , the debye frequency ; in fact @xmath205 with @xmath206 in @xmath207 dimensions ; this is the fundamental reason for the coherent motion of polarons and defects in insulators at low temperatures @xcite .
we will see below that @xmath208 in our case .
we start with the case where there is no transverse field @xmath199 , so that @xmath209 in ( [ 3.1 ] ) .
the relaxation dynamics for this case are described as damped motion between @xmath210 with a relaxation rate ( see , e.g. , @xcite ) @xmath211 @xmath212 where the longitudinal bias @xmath213 .
this rate is of course very small ( @xmath214 ) .
notice that @xmath215 is quite different from that rate which would be calculated via a naive application of the spin - boson results @xcite ; in fact it is smaller by a factor @xmath216 . on the other hand application of a transverse field @xmath217
gives a diagonal coupling , according to ( [ 3.3 ] ) , which can be incorporated into a spin - boson model ; in the absence of any bias @xmath8 one finds a relaxation rate @xmath218 or , in a non - zero bias , @xmath219 which is independent of @xmath3 as long as @xmath220 . using ( [ 3.3 ] ) we may express @xmath221 in terms of @xmath222 , and write @xmath223 these rates are very long , because there are so few phonons available to absorb the tunneling energy at these frequencies - we shall give typical numbers once the coupling constant @xmath222 has been determined .
the magnetoacoustic coupling is of course very well understood @xcite ; its effect on the wkb tunneling action of a grain has been analysed by garg & kim @xcite and by chudnovsky @xcite .
there has been some confusion about the relation between chudnovsky s `` angular momentum coupling '' and the usual magnetoacoustic coupling , which we try to resolve here .
the essential points we wish to make are that the couplings @xmath224 , in ( [ 3.1 ] ) , and @xmath222 in ( [ 3.5 ] ) , are given to within dimensionless constants @xmath225 by @xmath226 @xmath227 so that the relaxation rate in a longitudinal field @xmath228 is finally @xmath229 whereas the relaxation rate in a non - zero transverse field @xmath199 has an additional term @xmath230 by comparing ( [ 3.13 ] ) and ( [ 3.14 ] ) we find their ratio to be @xmath231 .
this ratio is large unless the external field is almost parallel to @xmath232 .
to get some ideas of the relaxation rates implied by these formulae , let us consider a situation where @xmath233 , @xmath234 , @xmath235 , @xmath236 , and @xmath237
. then we get @xmath238 ( i.e. , @xmath207 hours ) .
thus , if we ignore nuclear spin effects , the relaxation times at @xmath239 are already very long ; and @xmath240 is actually a large value for @xmath27 .
if the bias energy is further reduced to the @xmath27 scale we get astronomically large @xmath241 . to get ( [ 3.11 ] ) and ( [ 3.12 ] )
we start by noting that chudnovsky s angular momentum coupling @xmath242 integrated over the grain volume , is nothing but that set of transverse terms in the magnetoacoustic coupling which is responsible for transporting angular momentum to the transverse phonons .
this magnetoacoustic coupling has the general form @xmath243 where the coefficients @xmath244 are energy densities , and in fact are simply combinations of the magnetic anisotropy constants ( i.e. , of @xmath245 and @xmath246 in the case of @xmath247 ) , since this is the only energy scale that can be involved . from this
we see that for the grain , @xmath248 .
however we also have @xmath249 \sim \omega_o s \ ; , \label{3.17}\ ] ] and so @xmath250 also ; the direct relation to the @xmath251 can be easily determined for any particular @xmath252 and lattice type . from this
it follows that @xmath253 in ( [ 3.1 ] ) .
it is instructive to rederive this by estimating the correction @xmath254 to the bare tunneling action in a given strain field @xmath255 , i.e. , finding @xmath256 this is determined trivially from the interaction @xmath257 by integrating over time ( assuming a negligible change in the instanton path ) ; writing @xmath258 we get @xmath259 where @xmath260 is the form factor , and @xmath261 is the polarization of the relevant phonon ; for long wavelength phonons ( @xmath262 ) this gives us equation ( [ 3.11 ] ) ; and @xmath222 in ( [ 3.12 ] ) immediately follows using ( [ 3.4 ] ) .
notice that in the case of the easy - axis / easy - plane hamiltonian @xmath263 , the same argument gives @xmath264 it is sometimes useful to write all these couplings in terms of the relevant sound velocity , using @xmath265 ; usually this will mean the transverse sound velocity since it is usually lower than the longitudinal one .
then without specifying exactly which of the many combinations of the @xmath251 enter into our expression for some particular symmetry , but just calling this combination @xmath266 , we write @xmath267 @xmath268 finally , we remind the reader that none of the analysis in this section takes account of nuclear spins .
we now turn to a very different kind of bath from the phonon and electron bath , viz . , the `` spin bath '' made up from nuclear spins both inside and outside the grain , paramagnetic electronic impurity spins , and any loose spins in or near the grain surface .
as noted in the introduction , this spin bath can not be mapped onto an oscillator bath , and so a completely different kind of theoretical framework is necessary to describe its effects on mesoscopic or macroscopic quantum objects .
the main point of this section is to show how the spin bath , interacting in isolation with a giant spin @xmath2 , will basically block any relaxation at all , unless either the grain happens to be in a total field ( produced by the sum of the external field and the `` internal bias field '' generated by the nuclei ) which is very small , less than roughly @xmath27 ; _ or _ the nuclear spin diffusion mechanism of energy bias diffusion allows the system to find the resonance window . in what follows we first briefly recall the 4 mechanisms by which the spin bath controls the giant spin dynamics , and set up the formalism required to analyse magnetic relaxation in the presence of nuclear spins .
we then derive the relaxation dynamics of @xmath2 , first ignoring the internuclear spin diffusion ( to see how @xmath2 is frozen ) , then including it ( to show how @xmath2 can then relax ) . in an earlier series of papers @xcite we have constructed a theoretical description of a giant spin interacting with a more or less arbitrary spin bath , and solved this model for the quantum dynamics of @xmath2 in the absence of any external bias .
the same sort of model has also been set up to describe an arbitrary unbiased macroscopic coordinate interacting with a spin bath .
elsewhere we have used this model to deal with coherence experiments in squid s @xcite , and also shown how it leads to some remarkable conclusions about the quantum theory of measurement ( particularly concerning the role that nuclear spins play in making the mesoscopic world behave classically @xcite ) .
here we generalise this work to deal with a _ biased _ giant spin . as before , we shall assume that the giant spin can be truncated to 2 levels , and that @xmath156 ; all effects of higher levels will be absorbed into the parameters of an effective hamiltonian .
this hamiltonian was derived in refs.@xcite , and is written for the biased case as @xmath269 + h.c . \big\ } -\gamma_e \vec{s } \cdot \vec{h}_o + { { \hat \tau } _ z \over 2 } \sum_{k=1}^n \omega_k^{\parallel } \ : { { \vec l}_k \cdot { \hat { \vec \sigma } } _ k } + \sum_{k=1}^n \omega_k^{\perp}\ : { { \vec m}_k \cdot { \hat { \vec \sigma } } _ k } \;.
\label{4.1}\ ] ] there is also an inter - nuclear interaction of dipolar form ; thus in general we have to add to ( [ 4.1 ] ) a term @xmath270 we shall not need to write down the matrix elements @xmath271 in detail ; their strength is @xmath272 , the transverse nuclear relaxation rate .
typically we shall assume that @xmath78 is parallel to the easy axis of our easy - axis / easy - plane system , so that the total effect of @xmath78 will be written as @xmath273 where @xmath274 ( we change notation - the symbol `` @xmath3 '' will be used to denote the `` internal '' bias coming from the nuclei ) .
the spin bath is described in ( [ 4.1 ] ) by spin-@xmath88 variables @xmath275 .
the vectors @xmath276 , @xmath277 , @xmath278 , and @xmath279 are unit vectors ; the `` static '' couplings between @xmath2 and the @xmath275 are described by the last two terms , and the dynamic effects are given by the terms in the curly brackets .
the longitudinal static coupling @xmath280 tells us the change in energy of @xmath281 when @xmath2 flips between @xmath282 ; thus the vector @xmath283 represents the difference between the effective field acting on @xmath281 before @xmath2 flips and that field after @xmath2 flips .
if the @xmath275 describe nuclei , this field will be almost entirely hyperfine in origin , unless the @xmath275 are produced by truncating the levels of a higher - spin nucleus ( i.e. , a nuclear moment @xmath284 with @xmath285 ) .
in this latter case the total field acting on the @xmath275 will also include , e.g. , quadrupolar contributions ; it then follows that the field @xmath286 will be non - zero .
this transverse field is the _ sum _ of the fields acting on @xmath281 before and after @xmath2 flips .
the dynamic terms in ( [ 4.1 ] ) originate as follows .
the phase @xmath287 is the sum of the `` kramers phase '' @xmath288 ( see introduction - notice that in ( [ 4.1 ] ) we have not yet included @xmath289 into the definition of @xmath27 ) and a renormalisation @xmath290 of this phase , caused by the @xmath275 .
the term @xmath291 describes the amplitude for @xmath281 to flip , under the influence of @xmath2 , when @xmath2 flips ; both @xmath292 and @xmath291 partly derive from the unitary `` transfer matrix '' @xmath293 describing the effect of a tunneling @xmath2 on @xmath281 : @xmath294 @xmath295 here @xmath296 describes the microscopic interaction between @xmath2 and @xmath281 ; @xmath297 is the environment state before @xmath2 flips , and @xmath298 after @xmath2 flips .
the signs @xmath299 refer to the path traced by @xmath2 whilst flipping .
the term in @xmath300 and further contributions to @xmath292 and @xmath291 are necessary because environmental spins also have an effect on the tunneling action for @xmath2 ( in exact analogy with the nondiagonal terms discussed in section [ sec:3 ] ) .
the derivation of ( [ 4.1 ] ) has been described in detail elsewhere @xcite . here
we shall simply quote the values of the parameters in ( [ 4.1 ] ) that arise if we start with the microscopic hyperfine hamiltonian @xmath301 \ ; , \label{4.5}\end{aligned}\ ] ] i.e. , our usual easy - axis / easy - plane model coupled to nuclei via hyperfine couplings @xmath302 ; we get @xmath303 @xmath304 the assumption in ( [ 4.6 ] ) , that the hyperfine coupling @xmath305 , is almost always true . in any case @xmath291 , @xmath300 , and
@xmath292 can be calculated as general functions of @xmath306 , for any initial microscopic hamiltonian .
it is useful to bear in mind the physical meaning of the parameter @xmath291 in this regime ; @xmath291 is the amplitude for @xmath281 to coflip with @xmath2 .
note also that more general interactions than ( [ 4.5 ] ) will produce further renormalisations of @xmath27 .
working with ( [ 4.1 ] ) ( which is of course nothing but a low - energy effective hamiltonian , in the usual spirit which also led to ( [ 2.4 ] ) and ( [ 3.1 ] ) ) allows us to bring out the 4 essential physical mechanisms operating in the dynamics of @xmath2 .
these are \(i ) even when @xmath307 , an internal field bias @xmath308 acts on @xmath2 . typically for nuclei
we expect the hyperfine couplings to be tightly clustered around a principal value @xmath43 , with @xmath309 .
there will however be a spread @xmath310 , due not only to inter - nuclear couplings ( principally via dipolar or nakamura - suhl interactions , with @xmath311 ) but also other couplings such as transfer hyperfine couplings .
one may then define a `` density of states '' @xmath312 for bias @xmath3 ( fig.3 ) ; if @xmath313 , this consists of gaussian peaks @xmath314 of width @xmath315 around @xmath316 , where @xmath317 is the total nuclear polarisation , inside a gaussian envelope of width @xmath318 ; this envelope extends out to a maximum bias @xmath319 : @xmath320 ( here we assume that @xmath321even integer ; otherwise @xmath322 is impossible !
the modifications required for @xmath321odd are trivial and irrelevant to the ensuing discussion ) . in most cases , however , @xmath323 , and the different `` polarisation groups '' ( i.e. , the different peaks ) completely overlap , and we end up with a gaussian envelope for @xmath312 , i.e. , @xmath324 where @xmath325 . now for tunneling to occur at all , the total bias @xmath326 must be @xmath327 , otherwise the grain is simply trapped .
the fraction of grains in an ensemble having bias @xmath327 is roughly @xmath328 if @xmath329 , and @xmath330 if @xmath331 ; the result for @xmath329 arises because only a fraction @xmath332 of that portion @xmath325 of grains having @xmath333 can flip when the different polarisation groups do not overlap .
thus only a small fraction @xmath334 of spins are not `` degeneracy blocked '' from tunneling ( in the absence of nuclear dynamics , the effects of which are discussed below and in a section [ sec:5 ] ) .
this degeneracy blocking mechanism operates just as easily whether there is an external bias or not - in both cases only grains in the small window of bias can make transitions .
\(ii ) even in the absence of an external thermostat , the nuclear system can change the internal bias @xmath3 .
this occurs via dipolar interactions between the nuclei , at a rate @xmath335 .
this pairwise flipping of nuclei conserves @xmath336 but allows the nuclear bath to `` wander in bias space '' over the full range of bias energy associated with a particular polarization group , i.e. , over an energy range @xmath337 . if the spread @xmath338 is due entirely to dipolar interactions , so @xmath339 , then the fluctuating bias covers the whole energy range in a time @xmath340 .
this fluctuating bias @xmath9 can destroy coherence , but it can help magnetic relaxation , by helping the system find the small bias window .
\(iii ) the third mechanism , called `` orthogonality blocking '' , arises because the transverse fields @xmath341 acting on the environmental spins ( which are typically @xmath342 ) cause a mismatch between the initial and final nuclear wave - functions , in a way reminiscent of anderson s catastrophe @xcite . defining @xmath343 ( assuming @xmath344 )
, this blocking effect can be parametrised by @xmath345 , where @xmath346 so that @xmath347 .
orthogonality blocking inhibits both coherence and relaxation of @xmath2 ; it spreads out the high - frequency ( @xmath348 ) response of the grain to lower frequencies .
\(iv ) a final interaction mechanism arises simply because the giant spin can flip the nuclear spins .
this process is parametrised by the @xmath291 in ( [ 4.4 ] ) ; for the whole spin bath one finds that on average a number @xmath349 of spins will be flipped each time @xmath2 flips , where @xmath350 for @xmath351 .
this causes phase decoherence ; we have called this mechanism `` topological decoherence '' because it adds a random winding number to the effective action for @xmath2 .
it has a decoherence effect on tunneling , and thus inhibits magnetic relaxation .
the formal discussion of all these mechanisms has appeared in our previous work . in the next 2 sub - sections we will deal first with the dynamics of @xmath2 in the absence of spin diffusion between the nuclei , to show how the nuclei keep all but a tiny fraction of grains frozen ; then we show how spin diffusion changes this picture .
we assume that neither the grain nor the spin bath is connected to an external thermostat except at time @xmath352 ( in reality both are coupled to phonons and possibly electrons - see next section ) .
we also neglect nuclear spin diffusion here , and estimate its role later . thus relaxation of @xmath2 , in a bias , can only occur if the bias field due to the @xmath275 allows states @xmath353 , with the @xmath354 in some polarisation state @xmath355 , to overlap in energy ( within @xmath356 ) with some states @xmath357 , where the polarisation state @xmath358 in general .
thus we expect some nuclear spins to flip , if @xmath2 is to relax , although if @xmath359 is large , and the applied bias is small one can even have energy overlap between initial and final states having the same @xmath360 ( this does not , of course , mean that no spins are flipped during the transition , but only that none have to be flipped ) .
the formal treatment of this problem is a generalisation of our treatment of the unbiased case .
we write , for the time correlation function @xmath361 , the form @xcite @xmath362 \;. \label{4.10}\ ] ] this formula is crucial , and so we now spend a little time explaining it .
the average @xmath363 is over initial bias , with both thermal and density of states weighting .
@xmath364 is a function defined in ref.@xcite ; it describes all those grains in an ensemble having bias @xmath3 and for which , every time @xmath2 flips , the polarisation state of the @xmath365 changes by @xmath366 : @xmath367 in which @xmath368 describes a simple biased 2-level system ( cf . eq.([1.7 ] ) ) @xmath369 @xmath370 @xmath371 with @xmath372 , @xmath373 ( so that @xmath374 ) .
eq.([4.11 ] ) for @xmath375 can be understood as combining a `` phase average '' @xmath376 over the phase @xmath377 and winding number @xmath378 ) , with an `` orthogonality average '' @xmath379 these averages are performed over the biased 2-level correlation function @xmath380 , in which the tunneling amplitude @xmath381 depends on @xmath382 , @xmath383 , and @xmath384 via ( [ 4.14 ] ) .
eq.([4.10 ] ) for @xmath385 may seem a little odd , since it sums over apparently independent processes in which the polarisation state of the nuclei changes by @xmath366 each time @xmath2 flips .
why should the polarisation state change by @xmath366 each time ?
after all , when @xmath2 flips , the number of nuclei which flip is _ random _ ( with an average number @xmath386 ) !
the answer to this question comes from energy conservation considerations .
consider a single grain that at @xmath51 has its nuclear environment in the state @xmath387 corresponding to the polarisation group @xmath336 and the bias energy @xmath388 ( see fig.4 ) .
now @xmath2 flips - what are the possible final states for the combined system , which have energies in resonance with @xmath388 ?
there are three possibilities : \(1 ) none of the environmental spins flip ; then the final state is @xmath389 with @xmath390 , since @xmath391 changes sign .
\(2 ) some small number of environmental spins , say @xmath392 , flip together with @xmath2 , so that the net change in the nuclear spin polarisation is @xmath366 ( i.e. , @xmath393 up and @xmath394 down spins are flipped ) . as a first approximation
the final state energy will be @xmath395 ; assuming that the spread in nuclear frequencies is small , we may neglect for the moment the correction @xmath396 .
we denote these states as @xmath397 , and @xmath398 in fig.4 .
\(3 ) an enormous number of nuclear spins flip together with @xmath2 , with @xmath399 as large as @xmath40 ; in this case we can not neglect the correction @xmath400 any more , because it is comparable or even larger ( for @xmath323 ) than the typical difference @xmath401
. then we can use @xmath402 to `` fine tune '' a resonance ( which is impossible , in general , for the case ( 2 ) ) .
thus , we find in this case @xmath403 .
the correction strongly depends , of course , on the particular set of nuclei flipped ; a `` fine tuned '' situation is shown as state @xmath404 in fig.4 .
it seems at first , that the latter possibility is the best we can do for tunneling in resonance .
recall , however , that the probability that any given environmental spin flips during the transition is very small ( @xmath405 ) .
now , if we are going to tunnel between @xmath406 and @xmath404 then the amplitude of such a transition will be @xmath407 , and for large @xmath40 will be even less then @xmath408 ! thus such a resonance is simply impossible to realise .
in fact , only a small number of environmental spins may be flipped with a reasonable probability .
since the probability to flip none is just @xmath409 , we find the probability to flip exactly @xmath399 spins to be @xmath410 , which peaks at @xmath411 .
the parameter @xmath349 may be large in some systems , but still @xmath412 .
one immediately recognizes that the case ( 1 ) plays a role at small @xmath349 , while the case ( 2 ) will dominate when @xmath413 ( formally , we could include ( 1 ) in ( 2 ) as a particular transition with @xmath414 ) .
now we make use of the inequality @xmath309 to notice that among all possible transitions with different @xmath382 , the energy mismatch between the initial and final states @xmath415 is either much larger then @xmath27 for all @xmath382 , or is close to the resonance @xmath416 for only one specific value @xmath417 , with all the other transitions being @xmath418 away in energy .
once we have a resonance the system will make transitions from @xmath419 to @xmath420 by changing the nuclear spin polarisation by exactly @xmath421 each time to maintain this resonance . as for all the other transitions with @xmath422 , they give only very small corrections , of order @xmath423 or less ( cf .
eqs.([1.7 ] ) and ( [ 4.12 ] ) ) to the main contribution to the dynamics of @xmath2 , coming from the @xmath424 term .
of course eqs.([4.10 ] ) and ( [ 4.11 ] ) simply sum over all processes , and the resonance value of @xmath382 is included automatically it is worth noting that nothing depend explicitly on the initial polarisation state .
this point is further illustrated in fig.4 .
for given @xmath388 the initial polarisation state can be @xmath382 , and if @xmath388 is close to the center of the gaussian @xmath425 , then the resonance value is @xmath426 .
if however @xmath388 is , say , the down - tail state of another polarisation group @xmath427 , then transition to the up - tail state of @xmath428 has exactly the same energy mismatch and polarisation change @xmath366 .
that is why the _ only relevant statistical average is over the bias energy_. of course , if @xmath329 , and different polarisation groups do not overlap , the bias energy is related to the specific group and eq.([4.10 ] ) may be written as ( @xmath429 ) @xmath430 it is easy to understand now that external bias , which is indistinguishable from the internal one as far as the giant spin is concerned , will simply shift the resonance condition for given @xmath3 to some other value @xmath431 , as is clearly seen from the diagram in fig.5 , which is similar to fig.4 , but now with @xmath432 .
we may further simplify eq.([4.10 ] ) by changing variables from @xmath433 , and write it as @xmath434 \;.
\label{4.20}\ ] ] then because the function @xmath435 \sim ( \delta_o /\epsilon ) ^2 $ ] for large bias , and in most cases @xmath312 is a smooth function on a @xmath27 scale ( for @xmath436 ) , we have @xmath437 ( we assumed here that @xmath438 ) .
in the opposite limiting case @xmath439 , when the gaussian in ( [ 4.19 ] ) is a @xmath440-function , we have @xmath441 to summarize , we may understand ( [ 4.10 ] ) in a fairly simple way as including all effects of the spin bath on the dynamics of @xmath2 .
these come from the averages over phase ( eq.([4.14b ] ) , over the orthogonality mismatch between initial and final spin bath states ( eq.([4.14c ] ) ) , and over the internal bias @xmath3 , acting on @xmath2 , caused by the spin bath .
it also comes from summing over all possible changes in the polarisation state of the bath when @xmath382 flips .
it is perhaps worth noting , for those who may be used to the theory of `` oscillator bath '' environments @xcite , that the averages appearing here for the spin bath environment are very different in form , for the simple reason that in the case of the spin bath , _ most of the dynamics of the spins comes from their coupling to the macroscopic system itself _ ( in our case , to @xmath2 ) .
by contrast in the oscillator bath models , the coupling to the macroscopic system is weak ( @xmath442 , for each oscillator ) , and the dynamics of the individual oscillators is only weakly perturbed by the system .
this is why , in the theory of the oscillator bath , one can first calculate the weakly perturbed oscillator motion as a function of the system coordinates , and then integrate out the oscillators by functional averaging .
no such manoeuvre is possible for the spin bath ( or indeed any other environment where the couplings are not weak ) , and it is _ not possible _ to map the spin bath onto an oscillator bath ( see also ref.@xcite ) .
any attempt to do so ( by , e.g. , writing a spectral function @xmath443 , for a set of `` nuclear spin oscillators '' , as in ref.@xcite ) , can not be expected to give meaningful results . from these remarks
we can also see what has been left out of eq.([4.11 ] ) for @xmath444 ; we have left out everything coming from the independent dynamics of the spin bath ( independent , that is , from @xmath2 ) .
formally this can be put in by including these dynamics in @xmath445 in ( [ 4.12 ] ) . as we argue below this
corresponds to allowing a _ time - dependent bias _
@xmath9 in ( [ 4.10 ] ) and ( [ 4.11 ] ) .
we neglect the effect of coupling of the nuclei to phonons which is actually an incredibly small effect in most cases . in reality
the nuclear spin dynamics comes almost entirely from nuclear spin diffusion , parametrised by the transverse relaxation time @xmath340 , and caused by the dipolar interaction in ( [ 4.2 ] ) . as we shall see presently , if the spin bath is mainly composed of nuclei , this spin diffusion is crucial , since without it @xmath2 can not relax at all .
spin diffusion processes , in which @xmath336 remains unchanged but pairs of nuclei flip , cause @xmath9 to fluctuate in time because of the variation @xmath446 in the coupling of each nuclei to @xmath2 ( thus , for a process @xmath447 involving 2 nuclei , the total change in internal bias @xmath448 ; if @xmath40 nuclei flip , the change in @xmath8 is @xmath449 ) .
now this fluctuation is _ fast _ , in the sense that for large values of @xmath382 the time it takes @xmath9 to change by an energy @xmath450 is usually much less than the time @xmath451 required for @xmath2 to flip if it is in the coherence window ( of energy width @xmath450 ) ; this will be demonstrated below .
consequently it is not necessary ( or even useful ) to go through the elaborate detour of recalculating @xmath445 in eq.([4.12 ] ) , including a coupling to some effective `` oscillator bath '' , intended to model the effects of spin diffusion .
we simply write instead that for this fast diffusion problem , where @xmath9 varies over an energy range @xmath452 , one has : @xmath453 \ ; , \label{4.23}\ ] ] where we introduced the notation @xmath454 with @xmath455=integer and @xmath456 to define the shift in the polarisation change enforced by applying the external bias . here
( as before ) @xmath457 @xmath458 @xmath459 and @xmath381 is given by ( [ 4.14 ] ) .
in the usual case where @xmath323 , or if the external bias is close to a multiple of @xmath43 , the expression for the effective relaxation rate simplifies to @xmath460 this concludes our formal discussion of the problem of quantum relaxation of @xmath2 when it is coupled to a spin bath .
we now use the results to find @xmath461 for a few interesting cases .
we start by ignoring spin diffusion , in order to demonstrate the way in which relaxation is blocked in its absence .
let us first consider a really pathological case , in which there is no degeneracy blocking at all , i.e. , @xmath462 .
then the function @xmath312 simply becomes a set of sharp lines , and _ all states in a grain ensemble may resonantly tunnel _ , if @xmath463 .
if @xmath464 , then no resonance is possible .
initially there will be fast relaxation , involving processes where @xmath382 is not large . at longer times
the higher-@xmath382 processes take over - recall that for large @xmath382 , the transition amplitude @xmath450 is very small , since @xmath465 , which collapses when @xmath466 .
to get some idea of the resulting relaxation , consider what happens if @xmath429 , and let us ignore topological decoherence for simplicity ( which makes no difference for the long-@xmath53 asymptotics of the result ) .
then we have @xmath467\ ; , \label{4.29}\end{aligned}\ ] ] where @xmath468 , assuming @xmath469 ( note if @xmath469 , then the topological phase average ( [ 4.14b ] ) collapses to a @xmath440-function @xmath470 and we get ( [ 4.29 ] ) anyway ) .
the resulting curves are shown for @xmath471 in fig.6 , for various @xmath444 .
it is easy to then find the behaviour of @xmath385 by substituting ( [ 4.29 ] ) into ( [ 4.22 ] ) .
the steepest - descent integral over @xmath382 gives an accurate answer for long times .
however we more quickly derive this long time behaviour from dimensional arguments .
the sum over @xmath382 contains some @xmath472 polarisation groups inside the broad envelope function @xmath473 ; thus the crossover from @xmath474 to the equilibrium @xmath475 occurs around a time @xmath476 such that @xmath477 , i.e. , @xmath478 for shorter times the relaxation is roughly logarithmic , viz .
@xmath479 } \ ; ; \;\;\;\;(t\ll t_c ) \ ; , \label{4.31}\ ] ] whereas for longer times one has @xmath480 with logarithmic accuracy . the numerical evaluation of @xmath385 is shown in fig.7 .
however even the tiniest degeneracy blocking will upset these results - in fact if @xmath481 , the result ( [ 4.29 ] ) fails completely .
since the proverbial `` thunderstorm on jupiter '' is enough to give a @xmath310 exceeding @xmath482 for the large values of @xmath382 governing long time relaxation , we see that for relaxation ( just as for coherence @xcite ) , it is the limit of strong degeneracy blocking that is experimentally meaningful .
let us again consider the zero bias case in this limit , where @xmath483 .
we go back to ( [ 4.21 ] ) , again assume @xmath469 for simplicity , and , noting that the oscillatory @xmath484 term in @xmath445 ( eq.([4.12 ] ) ) gives an integrated contribution @xmath485 = { \pi \delta_m(y ) \over 2 } \int_0^{2\delta_m(y)t } dz j_0(z ) \ ; ; \label{4.33}\ ] ] we find @xmath486 this result describes an ensemble of grains in which a fraction @xmath334 of grains relaxes , leaving a fraction @xmath487 completely unrelaxed in the infinite - time limit ; a is given by @xmath488 since terms for @xmath489 make almost no contribution to @xmath334 due to the collapse of @xmath450 , it can be approximated for large @xmath386 by @xmath490 where @xmath325 is the number of states in the @xmath491 polarisation group ( in ref.@xcite we showed that individual terms in the sum over @xmath382 decreased as @xmath492 ; but the sum over all @xmath382 gives the factor @xmath493 ) .
thus we see that apart from the small number of grains in near resonance , most grains will be frozen for eternity by the nuclear bias field , even in the absence of an external bias ( again , ignoring nuclear spin diffusion ) . adding an external bias makes essentially no difference to this ( since this bias is physically indistinguishable from the internal bias ) .
in fact the bias will have no effect at all on ( [ 4.34 ] ) until @xmath494 the energy scale of variation of @xmath312 .
for @xmath495 , this means that we will see oscillations in the decay rate as a function of @xmath22 , with period @xmath43 . for @xmath323 , the decay rate and decay function @xmath334 will change very little until @xmath496 ; for @xmath497 the relaxed fraction @xmath498 since not even the nuclei can bring the system into resonance .
we derive also a useful formula for the bias and temperature effects by noting that only small @xmath499 contribute to the sum in eq.([4.21 ] ) and thus @xmath500 giving @xmath501 none of the results ( [ 4.29])-([4.36f ] ) is physical at very long time , because we have ignored the time variation of @xmath8 caused by nuclear spin diffusion .
let us now demonstrate that nuclear spin diffusion is _ fast _ , so that equations ( [ 4.24])-([4.26 ] ) are justified .
notice first that since dipolar pairwise flips occur at a rate @xmath335 for a given pair , and assuming short - ranged pair flips , we have roughly @xmath502 nuclear flips per second in the sample .
then in a time @xmath451 we have @xmath503 such flips , and in this time the bias will change by @xmath504 .
the condition for fast diffusion is @xmath505 ( so the system has no time to tunnel ) , i.e. , fast diffusion requires @xcite
@xmath506 if the spread @xmath507 arises from the dipolar interactions between the same kind of nuclei only , then @xmath508 ( but usually @xmath509 , because of other nuclei and fields ) , then this criterion becomes @xmath510 .
now in fact @xmath335 will be typical @xmath511 ( depending on the isotopic concentration of nuclear spins , etc . ) .
for any mesoscopic spin @xmath512 ( at least for most bare hamiltonians ) , so that even if ( [ 4.37 ] ) is violated for @xmath513 ( i.e. , the zero polarisation change process ) , it is obeyed for large values of @xmath382 . since the bulk of the relaxation involves values of @xmath514 , we see that ( [ 4.37 ] ) will always be the relevant condition . in this case eqs.([4.23])-([4.27 ] ) follow immediately .
let us now therefore evaluate @xmath385 for the coupled grain / spin bath system , taking full account of the dipolar interactions between the spins , and assuming ( [ 4.37 ] ) to be satisfied ; using ( [ 4.23])-([4.25 ] ) , with @xmath515 , this means we must evaluate , for @xmath516 @xmath517 the integral is very easily evaluated ; calling it @xmath518 , we have @xmath519 where the step - function @xmath520 , with @xmath521 arises because values of @xmath382 range up to roughly @xmath472 , and are thus typically large ; thus @xmath522 \bigg\ } \;.
\label{4.41}\ ] ] noting again that if @xmath323 , the dependence of @xmath385 on @xmath22 in ( [ 4.41 ] ) will be rather small until @xmath523 , we begin by analysing this result for @xmath429 .
this zero bias case has of course been previously analyzed in our coherence papers @xcite , but there we were only interested in the very short time behaviour of @xmath385 , i.e. , that part involving frequencies @xmath348 , connected with possible coherent oscillations .
coherent behaviour can only arise in the @xmath513 polarisation sector of ( [ 4.29 ] ) , which ignores spin diffusion .
however in analysing the quantum relaxation properties of @xmath2 , we need to sum over all @xmath472 important terms in ( [ 4.41 ] ) ; the effect of the spin diffusion will be to unblock the long - time relaxation . in evaluating ( [ 4.38 ] ) or ( [ 4.41 ] ) , a steepest - descent integration over @xmath382 is possible , but just as with the case of no spin diffusion , dimensional analysis is sufficient ; the crossover time from unrelaxed to almost relaxed behaviour occurs at a time @xmath476 given by @xmath524 ( compare ( [ 4.30 ] ) ) , and the short - time relaxation is roughly logarithmic , as in ( [ 4.31 ] ) , viz .
, @xmath525 } \ ; ; \;\;\;\;(t\ll t_c ) \ ; , \label{4.43}\ ] ] the long - time behaviour is , analogously to ( [ 4.32 ] ) , given by @xmath526 with @xmath527 corrections in the exponent .
thus , amusingly , the results with spin diffusion included look just like those _ without _ spin diffusion , provided @xmath439 in the latter .
the physical reason for this is simple - rapid spin diffusion basically eliminates the effects of degeneracy blocking , by allowing the bias @xmath9 to cover the whole energy range of each polarisation group .
this is then the fundamental reason why spin diffusion `` unlocks '' @xmath2 and allows it to relax .
in fig.7 we showed a plot of @xmath385 against @xmath528 , which clearly brings out the 3 relaxation regimes .
for very short times @xmath529 , there is a sudden relaxation involving only @xmath530 processes - the theory of this was given entirely in our coherence papers .
for times @xmath53 such that @xmath531 , we have roughly logarithmic relaxation .
then for @xmath532 , we have the behaviour in ( [ 4.32 ] ) or ( [ 4.44 ] ) . at this point
it is important to realise that for all but microscopic spins , @xmath476 will be astronomically long . from ( [ 4.42 ] ) we have @xmath533 consider now , as in fig.7 , the case of a mesoscopic spin where , e.g. , @xmath534 , @xmath535 , @xmath536 , and @xmath537 .
we then find that @xmath538 , i.e. , more than @xmath539 times the age of the universe !
thus for mesoscopic grains we _ always _ get logarithmic relaxation at long times , from nuclear spins , in the quantum regime .
readers familiar with the experimental measurements of magnetic relaxation in grains may well be quite surprised at this result , since it is almost a central dogma of magnetism that logarithmic relaxation must arise from a distribution of energy barriers @xcite .
nevertheless we see that not only is this not so , but that the logarithmic relaxation has been derived here in a very general way , with essentially no approximations ( but see below ) .
on the other hand for microscopic spins ( @xmath152 ) , one finds that @xmath476 can be short .
for example , consider a situation where @xmath540 , @xmath541 , @xmath542 , @xmath543 , @xmath544 , and @xmath545 ( this would roughly describe a particle containing @xmath546 tb atoms ) . then one finds @xmath547 , and any direct relaxation experiment ( typically conducted over time periods @xmath548 ) would be mostly in the long - time relaxation regime .
these results are both surprising and interesting , in view of the very long time ( often logarithmic ) relaxation observed in experiments .
however we immediately point out that the results ( [ 4.43])-([4.45 ] ) are by no means complete - although they are correct to within @xmath527 corrections for a coupled grain / spin bath system , a realistic calculation must incorporate _ all three _ of the relaxation mechanisms we have discussed , in sections [ sec:2 ] , [ sec:3 ] , and this one . since this changes the results yet again , we will not waste time here analysing the case of large bias ( @xmath549 in ( [ 4.23])-([4.26 ] ) ) for the grain / spin bath system , but proceed directly to the general case .
we now come to the crux of the paper , which involves putting together the various mechanisms discussed in sections [ sec:2]-[sec:4 ] to give a physically realistic picture of quantum relaxation in magnetic particles .
as mentioned in the introduction , we do not attempt a comprehensive discussion of all possible cases .
such a discussion would involve consideration of a large class of `` giant spin '' bare hamiltonians , with widely varying behaviour , and with each being considered throughout the whole range of field strength , field orientation , and temperature .
this would only be the beginning - we would then have to go on to examine the large variety of couplings to nuclear spins ( with hyperfine couplings ranging over 3 orders of magnitude , plus quadrupolar couplings ) , the variety of different magnetoacoustic couplings , etc .
we adopt the tactic here of showing how the mechanisms combine , and giving some details for one model .
in other papers , in preparation , we analyse some specific experimental systems in much greater detail .
we emphasize once again here that , in our view , magnetic relaxation in most experiments involves the dipolar interactions _ between _ grains in an essential way .
we begin by showing how phonons and nuclear spins combine to give a rather surprising form for the relaxation at low @xmath10 .
this analysis is appropriate to the case where both grain and substrate arre insulating .
we then go on to include electrons - in this case phonons become irrelevant , and only the combination of electrons and nuclei are important . before discussing the details ,
let us first state the method we shall use .
we have already seen how the dynamics of @xmath2 is given in the presence of a spin bath ( eqs.([4.23 ] ) or ( [ 4.38 ] ) ) .
when we couple in a bath of oscillators which allow transitions between states @xmath550 and @xmath551 of the combined grain / spin bath system , in which there is no restriction on the difference @xmath366 between the polarisations of the spin bath states @xmath552 or @xmath553 , or on their biases @xmath388 and @xmath554 ( apart from those imposed by energy conservation ) , the form of @xmath385 must change . in this paper
we will not attempt to give a general expression covering all mechanisms simultaneously ; this is more than we need .
instead we will use 2 expressions , which apply in the 2 limiting cases of interest .
these are \(a ) for short times , when the spin - mediate relaxation dominates ; then we ignore the oscillator bath , and use eq.([4.10 ] ) .
\(b ) for longer times the oscillator bath - mediated relaxation will take over . in this regime , each time @xmath2 flips , some of the bias energy is taken up by an oscillator mode , and the rest by some number @xmath399 of flipped spins in the spin bath .
there is no restriction on either @xmath399 or the polarisation change @xmath366 , for this process to work .
the complete calculation of @xmath385 here is very complicated , but we notice that as soon as we reach times where the typical oscillator - mediated transition rate , at a typical bias , is faster than the spin bath - mediated transition rate ( for those grains in an ensemble which have not yet relaxed ) , then a reasonable approximation for @xmath385 will be @xmath555 e^{-t/\tau(\epsilon , t ) } \bigg\ } \ ; , \label{5.1}\ ] ] where @xmath556 is the oscillator - mediated relaxation rate , and @xmath557 is the equilibrium population of the state @xmath195 in a given bias ( in our previous discussion of the nuclear spin effects we assumed that this bias was actually much less than @xmath10 ) .
we note that this expression is _ incoherent _ , because it applies to the majority of grains in a large bias , which can not relax via the spin bath as discussed in the previous section .
the relaxation rate in this incoherent case can be calculated as a second order perturbation theory expression in the tunneling amplitude @xmath27 .
thus the only effect of the spin bath which is left is the distribution over the bias .
the formal proof of the above statement is easy to follow for the case of pure othogonality blocking . in a typical bias @xmath558 the second order perturbation theory expression for the tunneling rate
is given by @xmath559 where the sums are over the initial and final states of the spin bath @xmath560 and oscillator bath @xmath561 with the equilibrium density matrices @xmath562 .
here @xmath563 describes the interaction with the phonons ( we do not even need its explicit form to prove the point ) , and the orthogonality rotation operator @xmath564 is given by @xmath565 the crucial point here is that in a large bias one may drop the energy @xmath566 transferred to the spin bath from the @xmath567-function argument in ( [ 5.1a ] ) , because only transitions for which @xmath568 contribute to the answer .
after that we have the sum over the complete set of states @xmath569 , so that @xmath570 because the of the unitarity of @xmath564 , and the final answer is that for the oscillator bath alone . including topological decoherence changes the result in a minor way
indeed , with nonzero @xmath291 , we have @xmath571 and the result for the relaxation rate in the oscillator bath is renormalized by the factor @xmath572 unless @xmath573 and the topological phase @xmath574 is a multiple of @xmath575 , this factor is of order @xmath576 , and clearly has no essential effect on the results .
thus , apart from the renormalisation factor @xmath71 , the relaxation of a single grain in a nuclear bias field @xmath3 is that already given in sections [ sec:2 ] and [ sec:3 ] , and all that remains for a grain ensemble is to average over the bias field .
the case of insulating grains and an insulating substrate is the one where there is most obvious competition and interplay , between the nuclear spin and phonon relaxation mechanisms .
they are both slow .
we have already seen ( fig.6 , and eq.([4.43 ] ) ) what the short - time relaxation will look like .
now consider the implications of ( [ 5.1 ] ) for the long - time relaxation .
we worked out @xmath215 for phonon relaxation in the absence of nuclear spins ; for a longitudinal bias this was given by ( [ 3.13 ] ) , and will be roughly the same in the presence of the spin bath ( i.e. , with the renormalisation factor ( [ 5.1d ] ) above ) .
thus we write @xmath577 in the form @xmath578 where @xmath579 is a `` typical '' normalising relaxation rate , defined as @xmath580 and @xmath581 is the width of the gaussian peak in @xmath312 , i.e. , @xmath582 notice that for a fixed bias @xmath583 , the actual relaxation rate @xmath584 still depends on temperature as @xmath585 . at low @xmath10
this scales down to the constant value @xmath579 , whereas at high @xmath10 it increases linearly , i.e. , @xmath586 .
it is useful to get some preliminary idea of typical time scales for @xmath587 .
let us consider 3 examples , viz . , ( i ) a tb oxide grain , with @xmath588 , containing @xmath589 nuclei ; ( ii ) a particle of er@xmath590as ( such particles are apparently insulating @xcite , containing @xmath589 nuclei , with @xmath591 ; and ( iii ) ni@xmath590o grains with @xmath591 , but @xmath592 only ( coming from the @xmath593 of ni@xmath594 nuclear spins ) .
let us also assume that @xmath595 and @xmath234 for each example .
however @xmath581 varies dramatically - for the tb@xmath590o grain , @xmath596 , and @xmath597 ; for the er@xmath590as grain , @xmath44 , and @xmath598 ; whilst for the ni@xmath590o grain , @xmath599 , and @xmath600 .
we thus get a wide range of time scales ; for the tb@xmath590o grains one has @xmath601 , for the er@xmath590as grains , one has @xmath602 , whilst for the ni@xmath590o grains , one has @xmath603years !
we now consider the general behaviour of @xmath604 .
we start by noticing that the weighting function for the initial states , of bias @xmath3 , contributing to ( [ 5.1 ] ) , will be @xmath605 \bigg\ } \;.
\label{5.9}\end{aligned}\ ] ] ignoring @xmath22 for the moment , we see that in zero external field this distribution peaks at _ negative _ bias @xmath606 .
this is natural ; as we lower @xmath10 , more and more of the nuclei align with @xmath2 , thereby lowering the energy of the combined system in the initial state .
this process continues until @xmath607 nuclei are aligned with @xmath2 , i.e. , when @xmath608 , at a temperature @xmath609 .
we have then reached the bottom edge of the distribution @xmath312 , way out in the wings . if @xmath432 , the distribution peaks at @xmath610 .
thus we write @xmath611 e^{-t/\tau(\epsilon , t ) } \bigg\ } \nonumber \\ & = & \int { dx \over \sqrt{2 \pi } } e^{-{1 \over 2}(x-\overline{\xi}_h+1/\overline{t})^2 } { e^{-x/\overline{t}}+e^{x/\overline{t } } e^{-(t/\tau_o)\:x^3 \coth ( x/2\overline{t } ) } \over 2\cosh ( x/\overline{t } ) } \ ; , \label{5.10}\end{aligned}\ ] ] where we have normalised all energies by @xmath581 : @xmath612 let us first look at the high-@xmath10 limit of ( [ 5.10 ] ) , i.e. , for @xmath613 when @xmath614 .
then ( [ 5.10 ] ) is easily evaluated ; one finds @xmath615 ( we recall that this result is correct only when @xmath616 - otherwise higher levels of the grain will come into play . )
this surprising power - law behaviour is actually easily understood .
the initial fast relaxation ( at @xmath617 ) comes from grains with large bias ; the slower relaxation ( @xmath618 ) comes from the grains with smaller bias .
the typical relaxation occurs at a rate @xmath619 , i.e. , it is faster for higher temperatures , in accordance with the remarks just after eq.([5.8 ] ) .
we may think of the power law as a `` grain - ensemble '' sum of a lot of different exponential decays , or as a funny kind of `` stretched exponential '' .
note however that the decay is _ not _ logarithmic in time .
now consider the low-@xmath10 limit , with @xmath620 ; this is a little more complicated . from (
[ 5.10 ] ) we now have @xmath621 let us consider first the behaviour of ( [ 5.14 ] ) when @xmath622 .
using the fact that @xmath620 one may further simplify this expression to @xmath623 \ ; , \label{5.15}\ ] ] this clearly defines a temperature - dependent relaxation rate @xmath624 which goes to zero at low @xmath10 .
we note however that the validity of ( [ 5.15 ] ) requires that the calculation is not affected by the edges of @xmath312 , which means that at a temperature @xmath625 , eq.([5.16 ] ) crosses over to a constant value , i.e. , @xmath626 but the magnitude of the relaxing component at this temperature is already extremely small ; in fact @xmath627 .
it is again useful to consider what is the the maximum relaxation time at low @xmath10 . for the examples
previously mentioned , we have ( i ) for the tb@xmath590o grain , @xmath628 minutes ; ( ii ) for the er@xmath590as grain , @xmath629 days ; and ( iii ) for the ni@xmath590o grain , @xmath630 years . defining the exponentially small total amplitude of the relaxing component as @xmath631 we find the long time asymptotics of @xmath385 as @xmath632
the reason for the @xmath633 decrease in the relaxation rate at low @xmath10 is simply that as we lower the temperature , the typical bias energy decreases linearly with @xmath634 , because only a small fraction of grains in the gaussian tail which have very small bias @xmath635 contribute to the relaxation .
the majority of the grains are actually _ `` trapped '' _ by the hyperfine interaction in a negative bias energy @xmath636 . thus at this low temperature even phonons will not help to liberate the giant spin .
consider now eq.([5.14 ] ) with nonzero external bias .
for @xmath637 the previous answer ( [ 5.19 ] ) hardly changes except that the amplitude of the relaxing component is given now by @xmath638 we observe that already for rather small bias @xmath639 we have an exponential dependence on @xmath22 .
the negative bias will simply further suppress an already exponentially small fraction of relaxing grains and we shall not go into more details here .
the case of positive @xmath22 is much more intriguing and surprising .
after an exponential increase in the amplitude up to @xmath640 for @xmath641 , the answer changes drastically . for @xmath642
one finds the magnetisation function to be @xmath643 which has different behaviour depending on whether we look at short times , where @xmath644 or long times , where @xmath645 note that eqs.([5.21])-([5.24 ] ) are derived in the approximation that @xmath646 . as before ,
the bias is restricted to be much less than @xmath31 .
finally , at @xmath647 , this behaviour changes yet again to @xmath648 .
we note the anomalous temperature and bias dependence of the effective relaxation rate ( [ 5.23 ] ) near the crossover . from this analysis
we conclude that unless the external bias is larger than @xmath581 ( or @xmath649 at low temperature ) the ensemble averaged results for the magnetisation relaxation are not described by the naive theory of a two level system coupled to a bath of oscillators , and spin bath interactions essentially modify the answer .
it is worth noting three effects which have to be kept in mind when considering the evolution of the experimental data in external bias : \(i ) since the dependence on @xmath22 starts when @xmath650 or @xmath651 , one _ can not _ derive the value of the tunneling amplitude from this dependence , as it would be in the case of isolated two - level system ; \(ii ) at low temperature the giant spin is `` trapped '' in its initial state by the large and negative bias produced by the spin bath .
\(iii ) only large and positive bias can liberate the giant spin and to allow complete magnetisation relaxation .
the temperature and bias dependence of the relaxation rate is anomalous near the crossover region @xmath652 . of course , when @xmath653 we recover back the pure case of oscillator bath relaxation . at this point
we would like to comment on our starting assumption that the initial state of the spin bath is equilibrated with the giant spin direction @xmath654 . experimentally this could be arranged by applying a very strong negative bias @xmath22 during a time period much longer than the longitudinal nmr relaxation time @xmath655 , and switching it off at @xmath51 . in some systems
however this procedure may not work because of an astronomically long @xmath655 ( in fact , low temperature longitudinal nmr is still something of an unsolved mystery ; spin - lattice relaxation times at @xmath183 temperatures ought to be many years in insulating crystals , unless some gapless magnetic excitations are involved , e.g. , on the sample surface or crystal defects ) .
it may be useful then to introduce two _ different _ temperatures - one for the oscillator bath @xmath656 and the other for the spin bath @xmath657 ( which may even be negative , i.e. , @xmath658 ! ) .
our basic equation ( [ 5.1 ] ) is still valid in this more general case , but now the distribution over the initial bias is defined by the spin bath temperature , whereas the giant spin evolution toward equilibrium is governed by the crystal temperature , i.e. , @xmath659 e^{-t/\tau(\epsilon , t_{ph } ) } \bigg\ } \;. \label{5.25}\ ] ] one may proceed with the analysis of this expression as before . at this point we feel that considering more cases in this paper will not add much to the physical picture . depending on the particular experimental system and sample preparation ,
the necessary formulae can be easily derived from ( [ 5.25 ] ) .
summarising what we have found for insulating systems , we see that at short times , the relaxation proceeds entirely via the nuclear spin bath , and is logarithmic in time ( eq.([4.43 ] ) ) . at longer times phonons
take over , and we get power law decay in time ; at low @xmath10 this goes as @xmath660 , with @xmath661 ( eqs.([5.16 ] ) and ( [ 5.19 ] ) ) , but only a small fraction of grains relax unless the bias @xmath662 or larger .
this bias is necessary to counteract the nuclear bias field . at higher temperatures we still get power - law relaxation ( eq.([5.13 ] ) ) .
the crossover between nuclear spin - mediated and phonon - mediated relaxation can be understood by matching eq.([4.43 ] ) with the relevant phonon expression .
we now turn to the case where either the grain or the substrate is _
conducting_. we use the giant kondo model of section [ sec:2 ] to describe interaction of @xmath2 with electrons , with a dimensionless coupling @xmath663 .
if both grain and substrate are conducting , then @xmath664 ( eq.([2.7 ] ) ) ; if only the substrate conducts , then @xmath665 ( eq.([2.13 ] ) ) ; and if only the grain conducts , @xmath666 ( eq.([2.21 ] ) ) .
the formal analysis is almost identical to that just used for phonons - we start again from ( [ 5.1 ] ) , now using the general result ( [ 2.8 ] ) for the electronic relaxation rate .
the high temperature limit is most transparent because @xmath667 in eq.([2.9 ] ) is independent of the bias .
thus we find a pure exponential relaxation @xmath668 at high temperature .
for very small grains or conducting grains on an insulating substrate at @xmath669 , we have a peculiar relaxation regime when the internal bias is much less than temperature , but the electronic damping rate @xmath670 is already small , that is for @xmath671 there is the temperature range where @xmath672 . now
the electronic relaxation rate is inversely proportional to the bias energy @xmath673 substituting this expression to ( [ 5.1 ] ) we find the time correlation function as @xmath674 which after easy evaluation yields @xmath675 the temperature dependence of the effective relaxation rate is @xmath676 , and again in the small external bias we have an unusual decay law - it is neither simple exponential nor power - law .
let us deal now with the low - temperature behaviour , @xmath677 . using ( [ 2.10 ] ) we then get @xmath678\mid^2 \ ; \nonumber\end{aligned}\ ] ] ( compare ( [ 5.14 ] ) ) .
if the bias @xmath679 , this defines a temperature dependent relaxation rate @xmath680 } \delta_o \left ( { \delta_o \over \omega_o } \right ) \left ( { 2\pi t \over \omega_o } \right)^{2\alpha-1 } \;.
\label{5.32}\ ] ] the answers crucially depend on the parameter @xmath663 .
we consider three limiting cases : \(i ) for @xmath671 we approximate the function @xmath681 by @xmath682 and obtain @xmath683 \(ii ) for @xmath684 we have @xmath685 , and , as in the high temperature limit , the relaxation is given by the simple exponential law @xmath686 \(iii ) finally we consider the case of large @xmath663 , which has a power - law asymptotic at long times @xmath687 \left ( { \tau_{eff } \over \pi t } \right)^{1/(2\alpha -1 ) } \label{5.37}\ ] ] in all these expressions @xmath688 is given by ( [ 5.20 ] ) .
the interpretation of these results also depends on the value of @xmath663 . if both grain and substrate are conducting , then the relaxation time in ( [ 5.32 ] ) will be greater than the age of the universe unless @xmath145 ( depending on the value of @xmath689 ) .
this shows the astonishing power of the electron bath to `` freeze '' the dynamics of @xmath2 , unless it is quite microscopic in size .
we shall not analyze here the case of very small @xmath61 , since once @xmath690 , the detailed structure of @xmath691 will no longer be gaussian - there will be lots of fine structure , depending on the particular systen involved .
such studies are best done on a case - by - case basis .
if only the substrate is conducting , then @xmath692 will be small until @xmath74 . for larger @xmath61
then the giant spin is again frozen .
for @xmath100 the results ( [ 5.33])-([5.36 ] ) can be applied if @xmath637 . again , as @xmath693 approaches @xmath694 , there will be an amusing crossover , with a relaxation rate given for finite @xmath695 by @xmath696 finally , if only the grain is conducting , @xmath663 can be exponantially small , if the spacing @xmath697 between the internal electron levels is large enough - this typically requires that @xmath698 , so that @xmath185 ( cf .
eq.([2.23 ] ) ) . in this case , unlike the cases where the substrate is conducting , the short - time relaxation will be again dominated by nuclear spin - mediated transitions , and the weak electron - mediated transitions only enter at later times in possible competition with the phonon - mediated transitions ; the typical electron - mediated rate becomes @xmath699 note that for low enough temperature , @xmath700 in ( [ 5.39 ] ) becomes so long that electrons in the grain become irrelevant , and we return to the calculations for the nuclear spin / phonon - mediated problem , i.e. , the grain behaves as an insulator . thus to summarize , we see that depending on whether the grain and/or substrate are conducting , we get behaviour ranging from grains frozen for all eternity ( if @xmath701 ) , to grains behaving essentially as insulators , with short - time relaxation controlled by the nuclei ( when @xmath702 ) . in the former case
we see an extraordinary alliance between the electrons ( with their strong dissipative suppression of tunneling ) and nuclear spins ( which trap @xmath2 in a negative bias @xmath703 ) to block any motion of @xmath2 even at low @xmath10 , where traditionally one expects tunneling . in fact , as @xmath704 , our calculations show that @xmath2 is blocked for eternity when @xmath61 takes microscopic values ( e.g. , @xmath705 ) , because the negative bias becomes very large .
this is a quite astonishing demonstration of the control that nuclear spins exert over the dynamics of @xmath2 - the spin is only unblocked when @xmath706 , when @xmath671 again ( and the whole wkb framework breaks down ! ) .
let us first summarize our main results . at temperatures @xmath10 such that @xmath707 , we may employ a 2-level model with bias to describe the grain . coupling to electrons and/or phonons in this model
can be treated by the usual oscillator bath methods .
however the crucial coupling turns out to be to the nuclear spin bath , to which oscillator bath models do not apply .
the main effect of this coupling is to spread each of the 2 levels for @xmath2 into a gaussian `` band '' of half - width @xmath708 , and total width @xmath709 , where @xmath43 is the hyperfine coupling and @xmath40 the number of nuclei in the grain .
if @xmath40 and @xmath43 are small , @xmath581 may only be a few @xmath183 ( as in ni or fe grains ) .
if @xmath40 and @xmath43 are large ( as in mesoscopic rare earth grains ) , @xmath581 may be hundreds of kelvin or more ( so that @xmath710 ) . at low temperatures
the nuclei begin to line up with the grain vector @xmath2 , and then @xmath2 finds itself in a _ negative _ internal nuclear bias field , with a mean bias @xmath3 of roughly @xmath711
. then at short times the only way that the grain can relax is by taking energy from the nuclear system .
the inter - nuclear dipolar interactions allow this to happen , by causing the bias @xmath3 to become time - dependent - the result is a slow relaxation , which for an ensemble of grains gives a fraction of relaxed grains going roughly logarithmically in time ( section [ sec:4 ] ) : @xmath712 } \;. \label{6.1}\ ] ] this logarithmic relaxation , shown in fig.7 , has nothing to do with a distribution of sizes of the grains ( we assume all grains have the _ same _ parameters ) . at longer times
the grains can relax very slowly via either electron- or phonon - mediated transitions , again in the time - varying nuclear field . if the grain is insulating ( no electrons ) , we get power - law relaxation ; a fraction @xmath334 of grains can relax , where @xmath713 in a bias field @xmath22 , and these grains relax as @xmath714 where @xmath715 if @xmath716 , and @xmath717 with @xmath718 once @xmath719 .
thus very few grains can relax at until the external bias @xmath22 compensates the internal bias @xmath3 and `` untraps '' @xmath2 . if there are electrons around , the results depend on whether the substrate is conducting or not .
if it is , then all but microscopic spins ( @xmath116 ) will be trapped in states of frozen magnetisation for astronomical times , once the temperature goes below @xmath31 .
this remarkable result is a combination of `` degeneracy blocking '' caused by the nuclear bias field , and the very strong electronic dissipation .
however if only the grain is conducting , and the substrate is insulating , then for @xmath720 , the electronic relaxation becomes negligible , and we go back to the fully insulating case ; the full results for all values of electronic coupling appear in section [ sec:5].@xmath590b .
it is clear that these results have implications both for future magnetic device design , and for the thousands of experiments that have been done on magnetic grain relaxation over the years . as far as devices are concerned ,
perhaps the most interesting result is the freezing of @xmath2 when both grain and substrate are conducting unless @xmath116 .
this may have far - reaching implications for future computers and information storage , since it implies that if we are prepared to go to low temperatures , one should be able to use even magnetic molecules as permanent memory storage elements , perhaps of the `` m.r.a.m . ''
type ( cf .
ref.@xcite ) .
even more tempting is the possibility that we may be able to manipulate @xmath2 in such molecules indirectly , by controlling the nuclear spin polarisation , since it is the nuclear bias field that controls the dynamics of @xmath2 . as far as experiments on grain relaxation
are concerned , we believe that it is quite urgent that experimental tests of our results be done at low @xmath10 .
unfortunately most of the many relaxation measurements that have already been done are at higher temperatures , where relaxation is dominated by thermal activation .
what we have found here is that the low-@xmath10 behaviour does not look like the conventional tunneling picture at all - there is no crossover to a temperature - independent `` tunneling relaxation '' at temperatures @xmath707 .
the low-@xmath10 behaviour of magnetic particles is not governed by the external bias at all , unless it is very large ; instead it is governed by the _ random distribution _ of internal nuclear bias fields .
this renders previous theories of the tunneling of magnetic particles somewhat meaningless , regardless of the external field .
thus we believe that it would be useful to look again at those experiments that _ have _ been done at low @xmath10 on ensembles of relaxing grains @xcite , particularly in an attempt to understand the `` plateau '' that often appears at low @xmath10 in the relaxation rate ( the `` magnetic viscosity plateau '' ) .
however an important cautionary note is necessary . in virtually all of these experiments , _ inter - grain interactions _
( mediated by the grain dipolar fields ) are very important .
we believe this to be the reason for the common occurence of `` avalanche '' magnetisation reversal in grain ensembles ( see , e.g. , ref.@xcite ) ; these avalanches are _ prima facie _ evidence for the importance of inter - grain interactions ( such avalanches also occur in multi - domain magnets , because of dipolar interactions between the domains and walls @xcite ) . thus to interpret most experiments ( including the remarkable recent results on mn@xmath56o@xmath56 molecules @xcite ) , we must incorporate these interactions - this we do elsewhere @xcite
. however some experiments may not suffer this problem .
the recent results of coppinger _ et al_. @xcite , on eras grains , with @xmath74 , may be a case in point .
it is not yet clear what is the effective coupling to the electron bath for this system , but we believe that the theory in the present paper should be applicable to this case .
this work was supported by nserc in canada , by the international science foundation ( maa300 ) , and by the russian foundation for basic research ( 95 - 02 - 06191a ) .
we would also like to thank i. affleck , m. bloom , b.g .
turrell , and w. hardy for discussions of the theory , and b. barbara , f. coppinger , d. maude , c. paulsen , and j. portal for discussions concerning experiments .
more general discussions concerning the spin bath , with s. coleman , a.j .
leggett , r. penrose , and w. unruh , were also very useful .
in truncating the initial hamiltonian ( [ 2.2 ] ) to the 2-level system for the giant spin one generates both the diagonal , i.e. , proportional to @xmath721 term , and nondiagonal terms @xmath722 .
the latter are usually not present in the spin - boson hamiltonian because they are subdominant to the diagonal ones . in general
we may write @xmath723 here @xmath724 and @xmath725 are the orbital channels for electron scattering off the grain ( @xmath726 ) , and we write the hamiltonian to be diagonal in @xmath724 for @xmath727 coupling .
we observe that initially ( before scaling ) the transverse coupling @xmath728 is exponentially small as compared with @xmath729 .
in fact @xmath730 because transverse matrix elements flip the giant spin as well and are therefore proportional to the overlap between the @xmath731 and @xmath732 states .
formally ( [ a.1 ] ) is a _ strongly anisotropic _ kondo model with multi - channel scattering in a _
magnetic field _ @xmath27 applied along the weak - scattering direction . since the @xmath728 are small we may ignore the renormalization of the @xmath729 and study the renormalization group equation for the @xmath728 first .
introducing dimensionless couplings @xmath733 and the same for @xmath734 , we find @xmath735 \ ; , \label{a.2}\ ] ] @xmath736 where @xmath737 is the scattering phase shift due to @xmath738 .
the crucial point is that for a large grain , when many scattering channels are equally coupled to the giant spin , the anderson orthogonality @xcite term @xmath171 will always dominate in the scaling of @xmath739 .
thus , quite generally , we never face the strong coupling regime when the @xmath739 are of order @xmath740 and large .
naively one would ignore these transverse couplings completely , but we have to remember that the giant spin dynamics is governed by the _
magnetic field @xmath27 which is also subject to renormalization due to anderson orthogonality .
the relevant renormalization group variable is fugacity @xmath741 , and @xmath742 ( the solution of this equation is explicitly used in our equations ( [ 2.8])-([2.10 ] ) )
. thus the question of whether we may neglect conduction electron spin flips or not depends on the parameter @xmath743
if @xmath744 is less than unity we are back to ( [ 2.8])-([2.10 ] ) , and electron spin flips are irrelevant . if we find @xmath745 then the correct expression for the giant spin relaxation time will be @xmath746 note that for @xmath745 we definitely have @xmath747 , so the @xmath2 dynamics are definitely incoherent , and the above refinement is hardly different from ( [ 2.8])-([2.10 ] ) ( it is equivalent to going from @xmath748 to @xmath749 ) . | at temperatures below the magnetic anisotropy energy , monodomain magnetic systems ( small particles , nanomagnetic devices , etc . ) must relax quantum mechanically - thermal activation is ineffective .
the discrete nature of the spectrum is important .
this quantum relaxation must be mediated by the coupling to both nuclear spins and phonons ( and electrons if either particle or substrate is conducting ) .
we analyze the effect of each of these couplings , and then combine them .
conducting systems can be modelled by a `` giant kondo '' hamiltonian , with nuclear spins added in as well . at low temperatures , even microscopic particles on a conducting substrate ( containing only @xmath0 spins ) will have their magnetisation frozen over millenia by a combination of electronic dissipation and the `` degeneracy blocking '' caused by nuclear spins .
raising the temperature leads to a sudden unblocking of the spin dynamics at a well defined temperature .
insulating systems are quite different . the relaxation is strongly enhanced by the coupling to nuclear spins . at short times
the magnetisation of an ensemble of particles relaxes logarithmically in time , after an initial very fast decay - this relaxation proceeds entirely via the nuclear spins . at longer times phonons
take over , but the decay rate is still governed by the temperature - dependent nuclear bias field acting on the particles - decay may be exponential or power - law depending on the temperature .
the most surprising feature of the results is the pivotal role played by the nuclear spins .
the results are relevant to any experiments on magnetic particles in which interparticle dipolar interactions are unimportant .
they are also relevant to future magnetic device technology .
pacs numbers : 75.10.jm , 75.60.jp |
given two independent multivariate iid samples @xmath6 with corresponding lebesgue densities @xmath2 and @xmath3 respectively , we are interested in identifying simultaneously subregions of the densities support where @xmath2 deviates significantly from @xmath3 at prespecified but arbitrarily chosen level @xmath7 . for this aim a multiple test of the composite hypothesis @xmath8 versus @xmath9 is proposed , built from a suitable combination of randomized nearest - neighbor statistics .
the procedure does not require any preliminary information about the multivariate densities such as compact support , strict positivity or smoothness and shape properties , and it is valid for arbitrary finite sample sizes @xmath10 and @xmath11 .
the hierarchical structure of p - values for subsets of deviation between @xmath2 and @xmath3 provides insight into the local power of nearest - neighbor classifiers , based on the training set @xmath12 .
thus our method is of interest in particular if the classification error depends strongly on the value of the feature vector , related to recent literature on classification procedures by belomestny and spokoiny ( 2007 ) .
there is an extensive amount on literature concerning two - sample problems .
most of it is devoted to the one - dimensional case as there exists the simple but powerful `` quantile transformation '' , allowing for distribution - freeness under the null hypothesis of several test statistics .
starting from the classical univariate mean shift problem ( see e.g. hjek and @xmath13idk 1967 ) , more flexible alternatives as stochastically larger or omnibus alternatives have been investigated for instance by behnen , neuhaus and ruymgaart ( 1983 ) , neuhaus ( 1982 , 1987 ) , fan ( 1996 ) , janic - wrblewska and ledwina ( 2000 ) , and ducharme and ledwina ( 2003 ) .
our approach is different in that it aims at spatially adaptive and simultaneous identification of local rather than global deviations . in the above cited literature asymptotic power
is discussed against single directional alternatives tending to zero at a prespecified rate , typically formulated by means of the densities @xmath14 and @xmath15 corresponding to the transformed observations @xmath16 , where @xmath17 denotes the mixed distribution function with density @xmath18 .
note that the mapping @xmath17 coincides with the inverse quantile transformation under the null . for power investigation of our procedure
a specific two - sample minimax set - up is introduced .
it is based on a reparametrization of @xmath19 to a couple @xmath20 , reducing the composite hypothesis `` @xmath4 '' to the simple one `` @xmath21 '' with the multivariate mixed density @xmath22 as infinite dimensional nuisance parameter .
the reparametrization conceptionally differs from the above described transformation for the univariate situation as it can not rely on the inverse mixed distribution function .
nevertheless it leads under moderate additional assumptions in that case to the same notion of efficiency . in order to explore the power of our method , the alternative is assumed to be of the form @xmath23 for fixed but unknown @xmath22 ,
some suitably chosen ( semi-)norm @xmath24 , a constant @xmath25 and a given smoothness class @xmath26 .
for any @xmath7 the quality of a statistical level-@xmath27-test @xmath28 is then quantified by its minimal power @xmath29 where the infimum is running over all couples @xmath19 which are contained in the set ( [ menge ] ) .
it is a general problem that an optimal solution @xmath28 may depend on @xmath26 and @xmath22 .
since the smoothness and shape of a potential difference @xmath30 are rarely known in practice , it is of interest to come up with a procedure which does not depend on these properties but is ( almost ) as good as if they were known , leading to the notion of minimax adaptive testing as introduced in spokoiny ( 1996 ) .
note that here we have however @xmath22 as an additional infinite dimensional nuisance parameter . the problem of data - driven testing a simple hypothesis is further investigated for instance by ingster ( 1987 ) , eubank and hart ( 1992 ) , ledwina ( 1994 ) , ledwina and kallenberg ( 1995 ) , fan ( 1996 ) and dmbgen and spokoiny ( 2001 ) among others , the two - sample context by butucea and tribouley ( 2006 ) .
the idea in common is to combine a family of test statistics corresponding to different values of the smoothing parameters , respectively ; see , for instance , rufibach and walther ( 2008 ) for a general criterion of multiscale inference .
the closest in spirit to ours is the procedure developed in dmbgen and spokoiny ( 2001 ) within the continuous time gaussian white noise model and further explored by dmbgen ( 2002 ) , dmbgen and walther ( 2008 ) and rohde ( 2008 ) , all concerned with univariate problems .
walther ( 2010 ) treats the problem of spatial cluster analysis in two dimensions .
the paper is organized as follows . in the subsequent section , a multiple randomization test
is introduced , built from a combination of suitably standardized nearest - neighbor statistics .
its calibration relies on a new coupling exponential bound and an appropriate extension of the multiscale empirical process theory .
asymptotic power investigations and adaptivity properties are studied in section 3 , where the reparametrized minimax set - up is introduced .
it is shown that our procedure is sharply asymptotically adaptive with respect to @xmath31-norm @xmath32 on isotropic hlder classes @xmath26 , i.e. minimax efficient over a broad range of hlder smoothness classes simultaneously . the application to local classification
is discussed in section 4 .
the one - dimensional situation is considered separately in section 5 where an alternative approach based on local pooled order statistics is proposed . in that case
the statistic does not depend on the observations explicitly but only on their order which in contrast to nearest - neighbor relations is invariant under the quantile transformation .
section 6.1 is concerned with a decoupling inequality and the coupling exponential bounds which are essential for our construction .
both results are of independent theoretical interest .
all proofs and auxiliary results about empirical processes are deferred to section 6.2 and section 6.3 .
the procedure below is mainly designed for dimension @xmath33 .
the univariate case contains a few special features and is considered separately in section [ sec : d=1 ] .
let @xmath34 and denote by @xmath35 the pooled set of observations .
for any @xmath36 , the @xmath37th nearest - neighbor of @xmath38 with respect to the _
euclidean distance _ is denoted by @xmath39 ; we define @xmath40 .
note that the nearest - neighbors are unique a.s .
the weighted labels are defined as follows @xmath41 in order to judge about some possible deviation of @xmath2 from @xmath3 on a given set @xmath42 , a natural statistic to look at is a standardized version of @xmath43 or more sophisticated , @xmath44 for some kernel @xmath45 supported by @xmath46 , where @xmath47 and @xmath48 denote the empirical measures corresponding to the first and second sample , respectively . note that the statistic is not distribution - free , and in order to build up a multiple testing procedure several statistics corresponding to different sets @xmath46 have to be combined in a certain way .
let @xmath49 denote any kernel of bounded total variation with @xmath50 and @xmath51 for @xmath52 .
we introduce the local test statistics @xmath53 where @xmath54 ^ 2.\\\ ] ] every @xmath55 is some in a certain sense standardized weighted average of the nearest - neighbor s labels and its absolute value should tend to be large whenever @xmath2 is clearly larger or smaller than @xmath3 within the random euclidean ball with center @xmath56 and radius @xmath57 .
the idea is to build up a multiple test , combining all possible local statistics @xmath55 .
the typical way is to consider the distribution of the supremum @xmath58 , see , e.g. heckmann and gijbels ( 2004 ) .
the problem is that the distribution is driven by small scales with a corresponding loss of power at larger scales , as there are many more small scales which contribute to the supremum . here
, we aim at a supremum type test statistic @xmath59 where the constants @xmath60 are appropriately chosen correction terms ( independent of the label vector @xmath61 ) for adjustment of multiple testing within every `` scale '' @xmath37 of @xmath37-nearest - neighbor statistics .
these correction terms in the calibration aim to treat all the scales roughly equally .
although the distribution of @xmath62 under the null hypothesis depends on the unknown underlying distribution @xmath4 , the conditional distribution @xmath63 of the above statistic is invariant under permutation of the components of the label vector @xmath64 . here and subsequently , the index `` @xmath65 '' indicates the null hypothesis , i.e. any couple @xmath19 with @xmath4 .
precisely , let the random variable @xmath66 be uniformly distributed on the symmetric group @xmath67 of order @xmath68 , independent of @xmath69 .
then @xmath70 where @xmath71 .
elementary calculation entails that @xmath72 thus the null hypothesis is satisfied if , and only if , the hypothesis of permutation invariance ( or complete randomness ) conditional on @xmath35 is satisfied .
an adequate calibration of the randomized nearest - neighbor statistics , i.e. the choice of smallest possible constants @xmath60 , requires both , an exact understanding of their tail behavior and their dependency structure .
note that the randomized nearest - neighbor statistics have a geometrically involved dependency structure . even in case of the rectangular kernel @xmath28
it depends explicitly on the `` random design '' @xmath35 which incomplicates the sharp - optimal calibration for multiple testing compared to univariate problems , where the dependency of the single test statistics remains typically invariant under monotone transformation of the design points .
also , the optimal correction originally designed for gaussian tails in dmbgen and spokoiny ( 2001 ) does not carry over as only the subsequent bernstein type exponential tail bound is available . [ [ a - coupling - exponential - inequality ] ] * a coupling exponential inequality * + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + based on an explicit coupling , the following proposition extends and tightens the exponential bounds derived in serfling ( 1974 ) for a combinatorial process in the present framework . if not stated otherwise , the random variable @xmath66 is uniformly distributed on @xmath67 , independent of @xmath69 .
] let @xmath55 be as introduced above and define @xmath73 then @xmath74 where @xmath75 [ [ remark ] ] remark + + + + + + the expression @xmath76 is the payment for decoupling which appears by replacing the tail probability of a hypergeometric ensemble by that of the binomial analogon .
for details we refer to section [ sec : decoupling ] . in the typical case @xmath77
we obtain @xmath78 .
compared to results obtained for weighted averages of standardized , independent bernoullis , the above bernstein type appears to be nearly optimal , i.e. subgaussian tail behavior ( with leading constant @xmath79 ) is actually not present . via inversion of the above exponential inequality , additive correction terms @xmath60 for adjustment of multiple testing are constructed .
the next theorem motivates our approach .
the construction is designed for typical arrangements of the observation values which appear with probability close to one . to avoid technical expenditure ,
we restrict our attention to compactly supported densities .
@xmath80 denotes the dual bounded lipschitz metric ( see , e.g. van der vaart and wellner 1996 ) which generates the topology of weak convergence .
`` @xmath81 '' refers to convergence in probability along the sequence of distributions @xmath82 .
[ thm : limit distribution ] define the test statistic @xmath83 with @xmath84 where @xmath85 and @xmath86 .
assume that the sequence of mixed densities @xmath87 on @xmath88^d$ ] is equicontinuous and uniformly bounded away from zero , while @xmath89 .
then the sequence @xmath90 of conditional distributions is tight in @xmath91-probability .
additionally , @xmath92 where @xmath93^d,\\ 0<r\leq{\underset{x\in[0,1]^d}{\max } } \arrowvert x - t\arrowvert_2 } } \left\{\ \frac{\big\arrowvert\int_{[0,1]^d } { \phi}_{rt , n}(x)\,dw(x)\big\arrowvert}{\gamma_{rt , n}}\ -\ \sqrt{2\log\bigl(1/{\gamma_{rt , n}}^2\bigr)}\ \right\}\ ] ] with @xmath94 a standard brownian sheet in @xmath88^d$ ] , @xmath95^d } { \phi}_{rt , n}(x)^2d x\big)^{1/2}\ $ ] and @xmath96^d}\psi\bigl(\frac{\arrowvert z - t\arrowvert_2}{r}\bigr)h_n(z)dz\bigg]\sqrt{h_n(x)}.\ ] ] the extra - term @xmath97 in the constant @xmath60 results from the exponential inequality in proposition [ ? ? ] and can be viewed as an additional penalty for non - subgaussianity .
the theorem entails in particular that the sequence @xmath98 is weakly approximated in probability by a tight sequence of _ non - degenerate _
distributions @xmath99 which indicates that our corrections @xmath60 are appropriately defined and can not be chosen essentially smaller .
note that the approximation @xmath99 depends on the unknown mixed distribution even under the null hypothesis .
let @xmath100 denote the generalized @xmath101-quantile of @xmath102 .
then we propose the conditional test @xmath103 our method can be viewed as a multiple testing procedure . for a given set of observations @xmath12
, the corresponding test statistic exceeds the @xmath101-quantile if , and only if , the random set @xmath104 is nonempty , where @xmath105 denotes the euclidean ball in @xmath106 with center @xmath107 and radius @xmath108 .
since the test is valid conditional on the set of observations , we may conclude that @xmath2 deviates from @xmath3 at significance level @xmath27 on _ every _ euclidean ball @xmath109 . in order to reduce the computational expenditure and to increase sensitivity on smaller scales ,
one may restrict one s attention to pairs @xmath110 for @xmath111 for some integer @xmath112 . note the validity of the test
does not require any assumption about the densities - even not lebesgue continuity .
recently , walther ( 2010 ) proposes a multiple test for cluster analysis in two dimensions based on a suitable combination of local @xmath113-likelihood ratio statistics , evaluated on a fixed choice of axis - parallel rectangles .
these statistics are not linear in @xmath114 and @xmath115 , respectively , but result in a subgaussian tail - behavior .
in this section we show that the above introduced multiple testing procedure possesses optimality properties in a certain minimax sense .
nonparametric comparison of different samples was recently investigated in the minimax approach by butucea and tribouley ( 2006 ) , in a rate - adaptive way and of a different sense from our results here .
we focus mainly on the considerably more involved problem of efficient adaptivity .
let us first introduce some notation .
for any set @xmath116^d$ ] and function @xmath117 from @xmath88^d\rightarrow \r$ ] , @xmath118 .
for any convex @xmath119 let @xmath120 denote the isotropic hlder smoothness class , which for @xmath121 equals @xmath122 let @xmath123 denote the largest integer strictly smaller than @xmath124 . for @xmath125 , @xmath120 consists of all functions @xmath126 that are @xmath127 times continuously differentiable such that the following property is satisfied : if @xmath128 denotes the taylor polynomial of @xmath117 at the point @xmath129 up to the @xmath127th order , @xmath130 in particular the definition entails that @xmath131 implies @xmath132 for every orthonormal transformation @xmath133 . for any pair of densities @xmath19 on @xmath88^d$ ] ,
let @xmath134 denote the corresponding mixed density @xmath5 .
fix a continuous density @xmath135 and define @xmath136 to be the set of pairs of densities such that @xmath137^d\big)\ \ \text{and}\ \
h(m , n , p , q)=h.\ ] ] [ [ reparametrizing - the - composite - hypothesis ] ] * reparametrizing the composite hypothesis * + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + with the notation above , @xmath138 consequently `` @xmath4 '' is equivalent to `` @xmath21 '' , and if @xmath139 is kept fixed , the composite hypothesis `` @xmath4 '' reduces to the simple hypothesis `` @xmath21 '' . in order to develop a meaningful notion of minimax - efficiency for the two - sample problem we treat subsequently the mixed density @xmath140 as fixed but unknown infinite dimensional nuisance parameter for testing the hypothesis @xmath141 note that in case that @xmath22 is uniformly bounded away from zero and @xmath2 is close to @xmath3 , @xmath142 coincides approximately with the difference @xmath143 , see also the explanation subsequent to theorem [ thm : lower bound ] .
[ [ remark-1 ] ] remark + + + + + + it is worth being noticed that the optimal statistic for testing @xmath144 against any fixed alternative @xmath142 equals the likelihood ratio statistic @xmath145 whose distribution still depends on @xmath22 under the null .
here and subsequently , the subscript @xmath146 indicates the distribution with density @xmath147 .
the rational behind the reparametrization is to eliminate the dependency on the nuisance parameter @xmath22 in the expectation under the null of the first and second order term of the @xmath113-likelihood expansion , resulting in asymptotic independence of @xmath22 for its distribution under the hypothesis for any local sequence @xmath148 .
the subsequent theorem is about the lower bound of hypothesis testing within the above defined classes of densities .
[ thm : lower bound ] let @xmath149 where @xmath150 defines the solution to the optimal recovery problem ( [ eq : recovery ] ) below .
assume that the sequence of mixed densities @xmath151 on @xmath88^d$ ] is equicontinuous and uniformly bounded away from zero .
then for any fixed @xmath25 and every nondegenerate rectangle @xmath116^d$ ] , @xmath152 for arbitrary tests @xmath153 at significance level @xmath154 .
note that @xmath153 may depend on @xmath155 and even on the nuisance parameter @xmath156 as already does the neyman - pearson test for testing @xmath144 against any one - point alternative .
we now turn to the investigation of the test introduced in section [ sec : nearest neighbors ] . to motivate the choice of an optimal kernel for our test statistics and its relation to the optimal recovery problem ,
let us restrict our consideration to the gaussian white noise context , leading in case of univariate hlder continuous densities on @xmath88 $ ] with @xmath157 to locally asymptotically equivalent experiments @xmath158 for two independent brownian motions @xmath159 and @xmath160 on the unit interval ( nussbaum 1996 , theorem 2.7 with @xmath161 and remark 2.8 ) . a multiscale statistic built from standardized differences of kernel estimates @xmath162 ( which is actually not admissible since @xmath156 is unknown in general )
then yields a distribution under the null close to ours in theorem [ thm : limit distribution ] , up to the fact that our local integrals in dimension one are taken with respect to a brownian bridge , reformulated to a wiener process integrand by change of the kernel .
concerning the optimization of @xmath28 , the quantity to be maximized within this gaussian white noise context appears to be the expectation of the single test statistics under the least favorable alternatives as their variances do not depend on the mean . in case @xmath163
this expression equals @xmath164):\\ \arrowvert { \phi}\arrowvert_{j}\geq \delta}}\,\frac{\int{\phi}(t)\,\psi(t)\,dt}{\arrowvert\psi\arrowvert_2},\]]leading to the dual representation of the optimal recovery problem ( see donoho 1994a ) .
[ [ the - optimal - recovery - problem - in - higher - dimension ] ] * the optimal recovery problem in higher dimension * + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in the framework of isotropic hlder balls , the optimal recovery problem leads to the solution @xmath165 of the optimization problem @xmath166 the closedness of @xmath167 in @xmath168 entails that the solution exists , its convexity implies furthermore uniqueness whence by isotropy of the functional class @xmath169 it must be radially symmetric . in case @xmath121 , one easily verifies that @xmath170 . in its generality
, the optimal recovery problem in higher dimension has not yet been investigated . considering the partial derivatives of @xmath150 along the coordinate axes
entails that @xmath171 is necessarily contained in @xmath172 .
however , the transfered optimization problem @xmath173 does not coincide with the univariate optimal recovery problem due to the additional weighting by @xmath174 which comes into play by polar coordinate transformation . whether the solution of ( [ eq : rec 2 ] ) for @xmath125 is compactly
supported or not is still open . for the case of univariate densities , it is known that the solution of the optimal recovery problem has compact support for any @xmath175 ( leonov 1997 ) , but an explicit solution in case @xmath125 is known for @xmath176 only . concerning details and advice on its construction ,
see donoho ( 1994b ) and leonov ( 1999 ) . for dimension @xmath177 , see klemel and tsybakov ( 2001 ) .
the next theorem is about the asymptotic power of the multiple test developed in section [ sec : nearest neighbors ] .
we restrict our attention to compact rectangles of @xmath178 to avoid boundary effects .
this restriction may be relaxed by the use of suitable boundary kernels , extending those of lepski and tsybakov ( 2000 ) for the univariate regression case to higher dimension .
[ thm : effizienz2 ] let @xmath179 denote the multiple rerandomization test at significance level @xmath27 , based on the kernel @xmath180 rescaled to @xmath88 $ ] . in case of unbounded support of @xmath171 , we may use a truncated solution @xmath181 .
let @xmath182 .
assume that @xmath183 is equicontinuous and uniformly bounded away from zero .
then for any fixed @xmath25 , there exists a @xmath184 such that @xmath185 for any nondegenerate compact rectangle @xmath186 .
in particular , the test is sharp - optimal adaptive with respect to the second hlder parameter @xmath187 .
while in view of the results in ingster ( 1987 ) the optimal rate of testing may be expected , some technical effort had to be done to propose a calibration achieving even sharp minimax - optimality .
[ [ remark-2 ] ] remark + + + + + + it is worth being noticed that the procedure achieves the upper bound uniformly over a large class of possible mixed densities .
the intrinsic reason is that conditioning on @xmath35 is actually equivalent to conditioning on @xmath188 , which indeed is a sufficient and complete statistic for the nuisance functional @xmath189 .
[ [ remark - sharp - adaptivity - with - respect - to - beta - and - l ] ] remark ( sharp adaptivity with respect to @xmath124 and @xmath187 ) + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + our construction , including the procedure especially designed for the one - dimensional situation , involves one kernel , shifted and rescaled depending on location and volume of the nearest - neighbor cluster under consideration . due to the dependency of the optimal recovery solution @xmath150 on @xmath124 , the corresponding test statistic @xmath190 achieves sharp adaptivity with respect to the second hlder parameter @xmath187 only .
taking in addition the supremum @xmath191}t_n(\beta ) $ ] over all kernels @xmath150 within a compact range @xmath192\subset ( 0,\infty)$ ] , sharp adaptivity with respect to both hlder parameters may be attained , provided that the above supremum statistic still defines a tight sequence ( in probability ) , i.e. the corresponding sequence of @xmath193-quantiles @xmath194 was stochastically bounded .
then the convergence @xmath195 for any random couple @xmath196 and any choice of @xmath124 could be be extracted from the proof of theorem [ thm : effizienz2 ] . at least for @xmath197 $ ] this tightness
may be deduced from the fact that the unimodal and symmetric @xmath171 depends continuously on @xmath124 in the sup - norm in particular @xmath198 as defined in theorem 2 with @xmath199 depends continuously on @xmath124 in the weak topology .
a general investigation especially for @xmath125 is beyond the scope of this article .
the next theorem shows however that our procedure simply based on the rectangular kernel is rate - adaptive with respect to both hlder parameters @xmath155 . due to the fact that it combines locally all nearest - neighbor scales at the same time
, it even adapts to inhomogeneous smoothness of @xmath30 , i.e. achieves _
spatial adaptivity_. [ thm : spatial adaptivity ] let @xmath179 denote the multiple rerandomization test based on the rectangular kernel .
assume that @xmath182 .
then for any fixed @xmath200 and parameters @xmath201 , @xmath184 and any collection of disjoint compact rectangles @xmath202^d$ ] , @xmath203 , there exist constants @xmath204 with @xmath205 the results from the previous paragraph deal with small scales of different ( arbitrary ) order depending on the smoothness classes under consideration . in particular , the minimax lower bound is concerned with scales tending to zero as @xmath206 , and it is not yet clear that there is no substantial loss at rather large scales .
the size of possible deviation @xmath207 and the scale ( here @xmath208 ) are linked in a specific way depending on the smoothness class under consideration , because the smoothness assumptions do not allow for arbitrarily fast decay to zero .
the next theorem is different in spirit .
we do not focus on smoothness classes but on stylized situations with @xmath142 being lower bounded by a `` plateau '' of absolute value @xmath209 within a ball @xmath210 . with @xmath211 denoting the lebesgue measure on @xmath212^d)$ ] , define @xmath213^d:\ \phi(m , n , p , q)(z ) \,\geq\ , \frac{c}{\sqrt{n\delta^d}}\ \forall\ , z\in b_x(\delta),\ , 0 < c\leq\sqrt{n\delta^d } \bigg\ } , \end{aligned}\ ] ] @xmath214^d:\ \phi(m , n , p , q)(z ) \,\leq\ , \frac{-c}{\sqrt{n\delta^d}}\ \forall\ , z\in b_x(\delta),\ , 0<c\leq\sqrt{n\delta^d}\bigg\}\end{aligned}\ ] ] and @xmath215 [ thm : local alternatives ] assume that @xmath182 .
\(i ) if @xmath153 is any sequence of tests at significance level @xmath216 , then @xmath217 implies that @xmath218 and @xmath219 .
\(ii ) if @xmath220 decribes the multiple rerandomization test based on the rectangular kernel at significance level @xmath221 , @xmath222 provided that @xmath218 and @xmath223 .
in particular our test is also consitent against local alternatives of the type @xmath224 for @xmath225 .
comparing ( i ) and ( ii ) demonstrates that the adaptive search for the location of deviations costs an additional logarithm of its inverse scale .
one may read out of the proof that the restriction for the sequence @xmath226 in ( ii ) can be slightly refined .
suppose we are given an iid sample @xmath227 , @xmath228 , where the marginal distribution of @xmath229 is assumed to be lebesgue - continuous with density @xmath22 on @xmath106 , and @xmath230 takes values in @xmath231 with @xmath232 then @xmath233 with @xmath234 .
assuming @xmath235 to be known , the question of local classification is to identify simultaneously subregions in @xmath106 where @xmath236 deviates significantly from @xmath211 which results in local testing the hypothesis @xmath237 imitating our procedure introduced in section [ sec : nearest neighbors ] , we may combine suitably standardized local weighted averages of labels , but the standardization differs due to the fact that the sum of ( strictly ) positive labels is random and not fixed , in particular @xmath238 are stochastically independent .
consequently , we may then rely the procedure on the classical bernstein exponential inequality for weighted averages of standardized bernoullis .
of course , the optimal separation constant for testing `` @xmath239 '' within some euclidean ball @xmath105 and its complement depends on the amount of observations in @xmath105 , whence analogously to the consideration above for the two - sample problem we may use the reparametrization of @xmath240 to @xmath241 with @xmath242 the power optimality results carry over to the classification context with similar arguments as used in the proof of theorem [ thm : effizienz2 ] .
we omit its explicit formulation at this point .
the one - dimensional situation allows for an alternative and more elegant approach based on order relations . for let @xmath243 denote the order statistic built from the pooled sample and define for any @xmath244 the local test statistics @xmath245 where @xmath246 compared to the procedure described in the previous section , we omit the explicit dependence of the weights on the observed values .
note that in contrast to nearest - neighbor relations , the order remains invariant under quantile transformation , i.e. @xmath247 , resulting in distribution - freeness of the corresponding multiscale statistic under the null .
suppose the null hypothesis is satisfied for some lebesgue continuous distribution on the real line .
then conditional on the order statistics as well as unconditional , the label vector is uniformly distributed on the set @xmath248 the described test statistics are local versions of classical wilcoxon rank sum statistics .
we omit any further investigation as the calibration for multiple testing can be done analogously to that proved in theorem [ thm : limit distribution ] but keep in mind that the approximating gaussian multiscale statistic under the null hypothesis will be independent of the nuisance functional @xmath189 due to the quantile transformation .
note that the use of typical mathematical tools for power investigation of rank statistics like hoeffding s decomposition is getting involved because the kernel @xmath171 for @xmath121 is not differentiable .
this section contains the coupling exponential bounds , i.e. in this context for weighted averages from a hypergeometric ensemble . using a different technique , namely an explicit coupling construction
, the subsequent proposition extends results of hoeffding ( 1963 ) on decoupling of expectations of convex functions in the arithmetic mean of a sample without replacement .
whereas in the latter case decoupling with constant @xmath249 is actually correct , a simple counterexample for an ensemble of two elements already shows that the result does not extend to arbitrary weighted averages , and some payment for decoupling appears to be necessary .
[ prop : decoupling ] let @xmath250 be iid with @xmath251 let @xmath252 with @xmath253 and @xmath254 be convex . then @xmath255 with @xmath256 in particular , @xmath257 for @xmath258 .
[ [ proof ] ] proof + + + + + let @xmath259 be uniformly distributed on the set @xmath260 and let @xmath261 such that @xmath259 and @xmath262 are independent . define @xmath263 conditional on @xmath259 and @xmath262 , the random vector @xmath264 is constructed as follows : if @xmath265 , let @xmath266 for all @xmath267 and let @xmath268 be uniformly distributed on the set @xmath269 for @xmath270 , let @xmath271 for all @xmath272 and let @xmath273 be uniformly distributed on @xmath274 note that @xmath275 are iid @xmath276
. then @xmath277\\ & \geq\ \e\,\psi\biggl(\e\big\{\min\big(\frac{s}{m},\frac{n - s}{n - m}\big)\big\}\sum_{i=1}^n a_ix_i\biggr)\ \ \ \ \ \textrm{(jensen inequality)}.\end{aligned}\ ] ] furthermore , @xmath278 with @xmath279 , which is uniformly bounded for @xmath280 .
@xmath281@xmath282 using the decoupling above , the next proposition presents the exponential bounds for the combinatorial process which are essential for our construction .
it implies proposition [ ? ? ] and improves in particular exponential tail bounds for the hypergeometric distribution of serfling ( 1974 ) in the coefficient in front of @xmath283 for @xmath284 close to zero or one , moderate @xmath285 and large @xmath68 .
note that this coefficient is crucial for the efficiency of the testing procedure .
the results may also be compared with the decoupling based exponential tail bounds in de la pe@xmath286a ( 1994 , 1999 ) .
[ prop : bernstein ] let @xmath275 be iid with @xmath287 let @xmath288 real valued numbers with @xmath289 its arithmetic mean and denote @xmath290 then in case of @xmath291 , @xmath292 where @xmath293 [ [ proof-1 ] ] proof + + + + + with @xmath294 we obtain for any @xmath295 @xmath296 whereby the last inequality follows from the fact that for any random variable @xmath297 with @xmath298 , @xmath299 and @xmath300 , @xmath301 elementary algebra shows that ( [ bennett ] ) is minimized with the choice @xmath302 which yields first a bennett ( 1962 ) exponential bound and because of @xmath303 consequently the bernstein type @xmath304 a symmetry argument provides the same bound for @xmath305 replaced by @xmath306 , which completes the proof of the first inequality . using that @xmath307 , we obtain the second asserted inequality from @xmath308 @xmath281@xmath282 this section collects results in the context of empirical processes which are essential for the next section . for any totally - bounded pseudo - metric space @xmath309
, we define the covering number @xmath310 let @xmath311 denote the borel-@xmath312-field on @xmath313 induced by the pseudo - metric @xmath236 ( which induces a topology in the usual sense , although without the hausdorff - property if it is not a metric ) and let @xmath314^{\mathcal{t}}$ ] be a family of measurable functions . for any probability measure @xmath315 on @xmath311 , consider the pseudo - distance @xmath316 for @xmath317 .
then for any @xmath318 , the uniform covering numbers of @xmath319 are defined as @xmath320 , where the supremum is running over all probability measures @xmath315 on @xmath321 .
( dmbgen and walther ( 2008 , technical report ) ) [ chaining ] let @xmath322 be a stochastic process on a totally bounded pseudo - metric space @xmath323 .
let @xmath324 be some positive constant , and for @xmath325 let @xmath326 a nondecreasing function on @xmath327 such that for all @xmath328 and @xmath329 , [ chaining.1 ] \
{ > g ( , ) } k ( - ) ( s , t ) .
then for arbitrary @xmath325 and @xmath330 , @xmath331 where @xmath332 is a dense subset of @xmath313 , and j(,a ) & : = & _ 0^g((a d(u)^2 / u ) , u ) du , + d(u ) = d(u , , ) & : = & \ { # _ o : _ o , ( s , t ) > u s , t _ o } . * remark .
* suppose that @xmath333 for some constants @xmath334 .
in addition let @xmath335 with constants @xmath336 and @xmath337 .
then elementary calculations show that for @xmath338 and @xmath330 , @xmath339 with @xmath340 . for the proof of theorem [ thm : limit distribution ] the subsequent extension of the chaining lemma vii.9 in pollard ( 1984 ) and theorem 8 in the technical report to dmbgen and walther ( 2008 ) will be used .
it complements in particular the existing multiscale theory by a uniform tightness result and to a situation where only a sufficiently sharp _ uniform stochastic _ bound on local covering numbers is available , which typically involves additional logarithmic terms .
the situation arises for example in the multivariate random design case where a non - stochastic bound obtained via uniform covering numbers and vc - theory may be too rough . [ levy ]
let @xmath341 be a sequence of random variables such that @xmath342 takes values in some polish space @xmath343 .
for any @xmath344 , let @xmath345 be a stochastic process on some countable , metric space @xmath346 , where @xmath347 .
suppose that the following conditions are satisfied : * ( i ) * there are measurable functions @xmath348 $ ] and @xmath349 such that for arbitrary @xmath350 , @xmath328 and @xmath325 , @xmath351 @xmath352 @xmath353 * ( ii ) * there exists a sequence @xmath354 of measurable sets and positive constants @xmath355 such that @xmath356\ ] ] whenever @xmath357 . for constants
@xmath358 define @xmath359 then there exists a constant @xmath360 such that for @xmath361 @xmath362 is at least @xmath363 whenever @xmath357 . if in particular @xmath364 and @xmath365 a.s . , then the sequence @xmath366 is tight in @xmath367-probability , provided that @xmath368 a.s .
[ [ remark-3 ] ] remark + + + + + + note that in case of @xmath369 with @xmath370 , + & = & ( w ( 1/))^1/ + o ( ( e/ ) ( e)^1/- 1 ) + & = & ( w ( 1/))^1/ + o(1 ) 0 .
[ [ proof-2 ] ] proof + + + + + due to the factorization lemma , the conditional probability and expectation factorize under the above conditions , i.e. we may consider a sequence @xmath371 and work with the sequence of conditional laws @xmath372 , but note that we do not assume equality of @xmath373 and @xmath374 in general . the first part of the proof is a modification of the chaining in dmbgen and walther ( 2008 , technical report ) applied to the conditional distribution @xmath372 for @xmath375 . here
we need however to define their additive correction function @xmath376 ( @xmath377 subsequently due to the dependence on @xmath68 and @xmath342 in our setting ) in a different way , taking into account the additional logarithmic terms in the bound of the covering numbers . lining up with their arguments , a suitable choice for the correction function @xmath378 appears to be @xmath379
this term is essential for our proof of efficiency .
it is important that the constant @xmath27 does not influence the leading term . concerning the tightness in probability as stated in the second part of theorem [ levy ] ,
notice that it does not follow by an immediate continuity argument because the metric ( and the metric space ) change with both , @xmath342 and @xmath68 , hence some additional uniformity is required .
for @xmath380 let @xmath381 be defined by @xmath382\\t\in\tt_n}}\bigg\{\frac{|z_n(t;y_n)|}{\sig_n(t;y_n ) } - g_n \biggl ( w \log\big(1/\sig_n(t;y_n)\big ) + c \log\log\big(e/\sigma_n(t;y_n)\big),\ , \sigma_n(t;y_n ) \biggr)\bigg\}.\end{aligned}\ ] ] first observe that for any fixed @xmath184 , @xmath383 the first part of theorem [ levy ] implies that the first term on the right - hand - side in ( [ eq : ende ] ) is bounded by @xmath384 for @xmath385 whenever @xmath357 . concerning the second term in ( [ eq : ende ] ) , note that @xmath386}\ h_n(\delta';y_n)\ + \
\frac{1}{\delta}\sup_{\substack{t\in\tt_{y_n}:\\ \sigma_n(t;y_n)\geq\delta}}\big\arrowvert z_n(t;y_n)\big\arrowvert.\ ] ] then the conclusion follows if we establish that @xmath387 for @xmath388 and @xmath375 , let @xmath389 be a maximal subset of @xmath390 with @xmath391 for arbitrary different indices @xmath392 . note that @xmath393 by assumption ( ii )
. then condition ( i ) implies that @xmath394 on the other hand , we have on the set @xmath395 the bound @xmath396 with @xmath397 tending to zero sufficiently slowly , ( [ eq : 9 ] ) and ( [ eq : 9 ] ) show together with the stochastic equicontinuity condition @xmath365 a.s .
@xmath398 since the assumption @xmath368 a.s .
guarantees @xmath399 the tightness in @xmath400-probability is proved.@xmath282 [ [ proof - of - theorem - thm - limit - distribution ] ] proof of theorem [ thm : limit distribution ] + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + let @xmath401 . in view of the @xmath55 s , the behavior of the process @xmath402 conditional on @xmath35 needs to be investigated , where @xmath403 is uniformly distributed on the set @xmath404 for notational convenience it seems useful to redefine the process on the random index set @xmath405 via the map @xmath406 and extend it to a process @xmath407 with @xmath408^d , \ , 0<r\leq\max_{x\in[0,1]^d}\arrowvert x - t\arrowvert_2\big\}$ ] by the definition @xmath409 where @xmath410 and @xmath411 denote the empirical measures based on the permutated variables @xmath412 and @xmath413 , respectively .
let @xmath414 ^ 2d\hat{\h}_n(x),\end{aligned}\ ] ] with @xmath188 the empirical measure of the observations @xmath415 . in the sequel we make use of the results in the previous section twice - in order to prove the tightness and weak approximation in probability of the sequence of conditional test statistics and within the `` loop '' we use the chaining arguments again to establish a sufficiently tightened uniform stochastic bound for the covering numbers below .
i. ( subexponential increments and bernstein type tail behavior ) the inversion of the conditional bernstein type exponential inequality in proposition [ prop : bernstein ] shows that for any @xmath416 , @xmath417 where @xmath418 with @xmath419 let the random pseudo - metric @xmath420 on @xmath313 be defined by @xmath421,\end{aligned}\ ] ] with @xmath422 then the application of the second exponential inequality of proposition [ prop : bernstein ] implies for any fixed @xmath423 that @xmath424 where @xmath425 \ii .
( random local covering numbers ) we need a bound for the local random covering numbers @xmath426 this is the most involved part of the proof .
in contrast to previous work we aim at a uniform stochastic bound . in order to establish a sufficiently sharp upper bound , the following two claims are established : \(i ) let @xmath427 and define @xmath428 for arbitrary different points in @xmath429 via @xmath430\bigg(1+c\log\big(4\,e\big/\max\big[\e\,{\hat{\rho}_{2,n}}^2,4/n\big]\big)\bigg),\ ] ] with @xmath431 a positive constant to be chosen later .
note that the map @xmath432 is subadditive for @xmath433 $ ] , hence @xmath428 defines a metric .
furthermore let @xmath434 , where @xmath435 then there exist a constant @xmath436 and a sequence @xmath354 of measurable sets with @xmath437 , such that for any @xmath25 , @xmath438 $ ] with @xmath439 and any realization @xmath440 @xmath441 if @xmath28 is not rectangular . in case of the rectangular kernel ,
the set @xmath442 in the covering number has to be replaced by @xmath443 ( ii ) there exists a constant @xmath444 , independent of @xmath445 and @xmath68 , such that whenever @xmath439 , the upper bound given in ( i ) is again bounded from above by @xmath446 moreover , the latter bound remains valid with @xmath313 in place of @xmath429 . note that we can not rely our bound directly on uniform covering numbers and vapnik - cervonenkis ( vc ) theory as the envelope @xmath447 only allows for a bound of order @xmath448 , which would result in the loss of efficiency of the procedure , and a pre - partitioning of @xmath429 as used in the proof of ( ii ) seems to be rather involved .
_ proof of ( i ) : _ we first derive a uniform stochastic bound for the random metric @xmath449 .
recall that every function @xmath28 of bounded total variation is representable as a difference of isotonic functions @xmath450 and @xmath451 . with the definition of the subgraphs @xmath452^d\times
\r:\ , y\leq \psi^{(i)}_{tr}(x)\big\},\ \ i=1,2,\ ] ] the set @xmath453 has a vc - dimension bounded by @xmath454 ( van der vaart and wellner 1996 ) with envelope @xmath455 .
consequently , the uniform covering numbers @xmath456 with @xmath457 is bounded by @xmath458 for some real - valued @xmath459 and some constant @xmath460 .
the boundedness of @xmath28 shows that @xmath26 is uniform glivenko - cantelli in particular ( see dudley , gin and zinn 1991 , for instance ) . as an immediate consequence , @xmath461 for any @xmath25
however such a bound is not sufficient for our purposes . because of @xmath462
, the squared random metric @xmath463 is @xmath464 times the sum of @xmath68 independent random variables with absolute values @xmath465 , hence @xmath466 now the application of bernstein s exponential inequality ( see shorack and wellner 1986 ) entails @xmath467}\bigg\arrowvert\ > \ \eta\bigg)\ & \leq\ 2\exp\bigg(-\frac{\eta^2/2}{1+\eta/3}\bigg)\\ & \leq\ 2\exp\bigg(-\frac{3}{2}\eta+\frac{9}{2}\bigg)\end{aligned}\ ] ] for arbitrary points
i.e. @xmath469 , standardized by @xmath470 , has ( uniformly ) subexponential tails .
analogously , the process @xmath469 has subexponential increments with respect to the metric @xmath471 given by @xmath472i\big\{a\not = b\big\},\ \ a , b\in\tt\times\tt.\ ] ] note that @xmath473 $ ] is lipschitz continuous with respect to @xmath471 .
theorem [ chaining ] shows that the above ingredients imply that @xmath474 for some adequately chosen @xmath475 , where we use the definition of @xmath476 from theorem [ levy ] with @xmath477 and @xmath478 .
now we may apply the latter to conclude that there exists some universal constant @xmath460 such that the probability of the event @xmath479\log\big(4\,e\big/\max\big[4/n,\e\hat{\rho}_{2,n}\big((t , r),(t',r')\big)^2\big]\big)\nonumber\\ & \text{~~for some~}(t , r),(t',r')\ \text{with~}\e\hat{\rho}_{2,n}\big((t , r),(t',r')\big)^2\leq\delta'\bigg\}\nonumber\end{aligned}\ ] ] is bounded by some function @xmath480 independent of @xmath68 with @xmath481 . since the probability in ( [ eq : unif .
] ) is antitonic in @xmath482 for any fixed @xmath68 with limes @xmath65 as @xmath483 for any fixed @xmath482 , there exists a sequence @xmath484 along which the result of ( [ eq : unif .
conv . ] ) still holds true .
thus , combining ( [ eq : unif . conv . ] ) and ( [ eq : unif .
2 ] ) for a sequence @xmath485 sufficiently slowly implies the existence of a sequence of sets @xmath486 with @xmath487 such that @xmath488^{1/2}\big(1+c\log\big(4\,e\big/\max\big[4/n,\e\hat{\rho}_{2,n}^2]\big)\big)^{1/2 } \ \
\text{whenever $ \underbar{x}\in\aa_n$.}\ ] ] the treatment of the random set @xmath489 is similar in spirit but more involved because the random quantity @xmath490 is not representable as a sum of independent variables .
however we can use the decomposition @xmath491 { \hat{\gamma}_n}^2={\hat{\gamma}_{2,n}}^2-{\hat{\gamma}_{1,n}}^2 $ ] . before deriving a stochastic bound , we notice the following : if @xmath28 describes the rectangular kernel , we have @xmath492 , i.e. @xmath493 in this case , the random set @xmath494 is consequently contained in the union @xmath495 consider the general case . using that @xmath496 and @xmath497 we may apply the above chain of arguments for @xmath498 to @xmath499 and @xmath500 together with the upper bounds in ( [ eq : varianz 1 ] ) and ( [ eq : varianz 3 ] ) for the standardization respectively and obtain the existence of a constant @xmath501 such that @xmath502^{1/2}}{\sqrt{n}}\log\big(e\sqrt{n}\big/\max\big[1/n,\gamma_{2,n}^2\big]^{1/2}\big)\\ & \leq\ \hat{\gamma}_{1,n}\ \leq\ \gamma_{1,n}\ + \ \frac{c_1\max\big[1/n,\gamma_{2,n}^2\big]^{1/2}}{\sqrt{n}}\log\big(e\sqrt{n}\big/\max\big[1/n,\gamma_{2,n}^2\big]^{1/2}\big)\end{aligned}\ ] ] whenever @xmath503 for some sequence @xmath504 with asymptotic probability @xmath249 , uniformly evaluated at @xmath505 . note that @xmath506 , @xmath507 for all @xmath505 .
the same holds true with a constant @xmath508 and a sequence @xmath509 with asymptotic probability @xmath249 and @xmath499 and @xmath510 replaced by @xmath500 and @xmath511 . using the lower bound for @xmath500 and the upper bound for @xmath499 , a bit of algebra yields @xmath512^{1/2}\frac{k}{\sqrt{n}}\log\big(e\sqrt{n}\big/\max\big[1/n,\gamma_{2,n}^2\big]^{1/2}\big)^2\bigg\}\end{aligned}\ ] ] whenever @xmath513 . here and from now on , @xmath324 denotes some universal constant , not dependent on @xmath68 and @xmath514 .
its value may be different in different expressions .
now we first consider the case @xmath515 then the above condition shows that @xmath516^{1/2}\frac{k}{\sqrt{n}}\log\big(e\sqrt{n}\big/\max\big[1/n,\gamma_{2,n}^2\big]^{1/2}\big)^2\\ & \leq\ 2\max\bigg\{\delta,\ \max\big[1/n,\gamma_{2,n}^2\big]^{1/2}\frac{k}{\sqrt{n}}\log\big(e\sqrt{n}\big/\max\big[1/n,\gamma_{2,n}^2\big]^{1/2}\big)^2\bigg\},\end{aligned}\ ] ] which entails that @xmath517 for @xmath518 by the isotonicity of @xmath519 on @xmath520 $ ] . on the other hand ,
the case @xmath521 implies already that @xmath28 is equal to the rectangular kernel : if the sup is attained it is obvious .
the equicontinuity of @xmath522 and its uniformly bounded @xmath523-norm @xmath524 imply its uniform boundedness , hence relative compactness in the topology of uniform convergence by the arzel@xmath525-ascoli - theorem .
there therefore exists at least a uniformly convergent subsequence @xmath526 with ( uniformly ) continuous limit , say @xmath22 , along this result holds true as well , because @xmath527 depends continuously on the mixed density .
this however implies that @xmath28 describes the rectangular kernel , because the uniform limit @xmath22 of that subsequence is bounded away from zero .
hence in case of ( [ eq : aa ] ) , we consequently obtain by ( [ eq : union ] ) @xmath528 _ proof of ( ii ) : _ since @xmath28 is of bounded total variation , there exists some finite measure @xmath529 such that for any @xmath530 , @xmath531 $ ] . with @xmath532\delta\bigg[0,\frac{\arrowvert t'-x\arrowvert_2}{r'}\bigg]\ ] ] we obtain @xmath533}\int i\big\ { y\in m_x(t , t',r , r')\big\}\,d\h_n(x).\label{eq : a}\end{aligned}\ ] ] then @xmath534 implies that @xmath535 . since @xmath156 is uniformly bounded from above , we obtain that ( [ eq : a ] ) is not greater than @xmath536 .
consequently , @xmath537 if @xmath538 with the metric @xmath539 defined below in ( [ metric ] ) , due to the isotonicity of @xmath540 for @xmath541 $ ] , @xmath460 .
@xmath28 attains its maximum @xmath249 at @xmath65 , hence there exists some @xmath542 such that @xmath543 whenever @xmath544 . using in addition the uniform boundedness of @xmath156 away from zero we obtain @xmath545 we now start bounding the covering numbers @xmath546 where the metric @xmath539 on @xmath547 is pointwise defined by @xmath548\bigg)\label{metric}\ ] ] with @xmath549 the volume of the @xmath539-dimensional euclidean ball with radius @xmath550 . again by the isotonicity of @xmath551 for @xmath541 $ ] , the inequality @xmath552 implies that @xmath553 is not greater than @xmath554 .
thus in order to finish claim ( ii ) , it is sufficient to bound @xmath555 first note that there exists a finite collection of at most @xmath556 points @xmath557 such that the set @xmath558 is contained in the union @xmath559 with @xmath560^{1/d}\big)\bigg\}\ ] ] for some universal @xmath561 .
the rotation and translation invariance of the lebesgue measure leads to the rescaling invariance for the covering numbers @xmath562 but a minimal @xmath563-@xmath564-net of the set @xmath565 for some _ fixed _
@xmath566 contains not more than @xmath567^d$ ] elements @xmath568 with @xmath324 uniformly in @xmath569 $ ] , noticing that @xmath570 and @xmath571 .
now fix a @xmath572-net @xmath573 with respect to @xmath574 and observe that @xmath575 for @xmath576 , @xmath577 , which shows that the quantity ( [ cn ] ) is bounded by @xmath578 ( with @xmath324 uniformly in @xmath397 and @xmath579 ) .
correspondingly , this holds true for @xmath580)^{1/2},\aa_i , \tilde{d}\big)$ ] , hence the covering number ( [ eq : covering number ] ) is bounded by @xmath581 for some universal constant @xmath444 .
an analogous bound holds for @xmath429 in place of @xmath313 ( and @xmath439 ) : if @xmath582 denotes an @xmath397-net with respect to @xmath539 in @xmath583 , we may define a @xmath584-net @xmath585 in @xmath586 via the definition @xmath587 . the corresponding covering numbers in case of the rectangular kernel for the sets @xmath588
can be treated with similar arguments , which concludes the proof of ( ii ) . in order to line up with the requirements of theorem [ levy ] , let us remark that the proof of that chaining requires only the special choice @xmath589 for some exponent @xmath590 , which entails that @xmath591 for some @xmath459 in case @xmath592 .
but for any @xmath593 , @xmath594 , and with the same arguments as used in ( i ) we obtain for @xmath595 that the inequality @xmath596 holds , uniformly in @xmath597^d$ ] , with asymptotic probability @xmath249 , which entails @xmath598 for some @xmath599 .
( tightness and weak approximation in probability ) as a consequence of the above exponential inequalities in step i and the bound for the uniform covering numbers @xmath600 , theorem [ chaining ] shows @xmath601 where the @xmath31 within the brackets is even running over elements of @xmath602 . now
the application of theorem [ levy ] entails that @xmath603 is tight in @xmath604-probability .
what remains being proved is the weak approximation .
starting from ( [ eq : ec ] ) , the uniform convergence ( [ eq : unif .
conv . ] ) implies in particular the asymptotic stochastic equicontinuity @xmath605 since to any subsequence of the metric @xmath606 there exists some uniformly convergent subsubsequence as a consequence of the relative compactness of @xmath522 in the uniform topology , it suffices ( via proof of contradiction ) for the weak approximation in probability @xmath607 to establish the convergence of finite dimensional distributions . here
, @xmath80 is defined via the outer expectations @xmath608 .
for let @xmath609 be a collection of points from @xmath610 . denote furthermore @xmath611 then @xmath612 with @xmath613 the @xmath614th nearest - neighbor of @xmath107 within @xmath35 .
let @xmath615 be pointwise be defined by @xmath616 using that @xmath617 equals @xmath618 for two random variables @xmath619 and @xmath620 , one finds that @xmath491\mathrm{cov}\,\big(y_n(t , r),y_n(t',r')\big\arrowvert\xx_n\big)$ ] equals @xmath621 replacing the empirical measure @xmath188 by its expectation @xmath189 , the above six expressions in ( [ covarianz ] ) coincide with the covariance @xmath622 of the limiting process @xmath623 .
define @xmath624 , @xmath625 .
since @xmath626 and @xmath627 by an application of the weak law of large numbers for triangular arrays to each of the expressions in ( [ covarianz ] ) separately , hjek s central limit theorem for permutation statistics extended for the multivariate setting yields the desired weak convergence in probability of the finite dimensional distributions . for notational convenience ,
define @xmath628 and @xmath629 since @xmath630 and @xmath631 as @xmath483 , it follows from the above established results that @xmath632for any fixed @xmath633 $ ] .
an application of theorem [ levy ] as well as its subsequent remark imply that @xmath634 for any @xmath635 .
thus , because obviously @xmath636 we obtain @xmath637 [ [ proof - of - theorem - thm - lower - bound ] ] proof of theorem [ thm : lower bound ] + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + let @xmath638 be some compact rectangle of @xmath639 .
fix @xmath640 .
for any integer @xmath641 let @xmath642 be some maximal subset of points such that @xmath643 and @xmath644 for arbitrary different points @xmath645 .
then @xmath646 .
now let @xmath647 be the solution of the subsequent optimization problem : ( @xmath648 ) minimize @xmath649 under the constraints @xmath650 these constraints define a closed and convex set in @xmath651^d\big)$ ] which is non - empty for @xmath37 sufficiently large ( and uniformly in @xmath68 due to the equicontinuity of @xmath183 and the rescaling property , see subsequently to ( [ eq : orr ] ) below ) .
consequently in the latter case , the argmin @xmath647 exists and is unique .
the resulting density candidates @xmath652 are non - negative and thus contained in @xmath653 as soon as additionally @xmath654 this is guaranteed for sufficiently large @xmath68 when sequence @xmath655 tends to zero . for any statistical level-@xmath27-test @xmath656 $ ] for testing the hypothesis `` @xmath657 '' it holds true that @xmath658 for short we write @xmath659 for @xmath660 in the sequel .
note that the test is allowed to depend on the nuisance functional @xmath156 ( in fact the @xmath113-likelihood and its distribution do ) .
now we aim at determining @xmath661 such that the right - hand - side tends to zero as @xmath68 goes to infinity .
although @xmath662 for any different @xmath645 , the likelihood - ratios @xmath663 are not independent . however , they are independent conditional on the random vector @xmath664 with entries @xmath665 note that @xmath666 . following at this point standard truncation arguments as , for instance , in dmbgen and walther ( 2009 ) , proof of lemma 7.4 , it turns out to be sufficient for the convergence to zero of ( [ eq : likelihood ] ) to find @xmath661 and @xmath667 $ ] such that the ratio @xmath668 tends to zero as @xmath68 goes to infinity . but
@xmath669 using the bound @xmath670 for @xmath671 .
now let @xmath672 be the solution to the following optimization problem ( @xmath673 ) minimize @xmath649 subject to @xmath674 notice the rescaling property @xmath675 with @xmath676 and @xmath677 @xmath678 @xmath679 with @xmath680 and @xmath681 . due to the equicontinuity of @xmath522 , @xmath682 whence @xmath683 because the minimum in ( @xmath648 ) depends continuously on the mixed density @xmath156 as can be seen using a lagrange multiplier for the centering constraint .
note that the @xmath684-term is uniformly in @xmath685 .
now the combination of ( [ eq : 1 ] ) and ( [ eq : 2 ] ) shows that for @xmath661 sufficiently small , ( [ eq : ee ] ) is bounded by @xmath686 by construction , @xmath687 for some constant @xmath688 .
now fix @xmath25 and define @xmath689 observe that the sequence @xmath690 is increasing in @xmath37 .
we need to check that @xmath691 .
note that in contrast to ( [ eq : orr ] ) , the solution of ( [ eq : recovery ] ) does not integrate to zero in general and it remains still open if @xmath150 is compactly supported for @xmath33 and @xmath125 .
starting from @xmath150 , it is sufficient to construct a sequence @xmath692 satisfying the constraints of the optimization problem ( @xmath673 ) such that @xmath693 . then the equality @xmath694 follows from @xmath695 .
the existence is sketched in the appendix of the extended version of this article . as a consequence there exists some @xmath696 such that @xmath697 .
now one verifies that the lower bound is established with the choice @xmath698 and some sequence @xmath699 with @xmath700 .
@xmath282 [ [ proof - of - theorem - thm - effizienz2 ] ] proof of theorem [ thm : effizienz2 ] + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + by virtue of theorem [ thm : limit distribution ] , the sequence @xmath102 is tight in @xmath701-probability , resulting in stochastic boundedness of the sequence of random quantiles @xmath702 .
the bounded total variation of the kernel for @xmath703 is a consequence of its monotonicity , for @xmath704 it results from the continuous differentiability of @xmath705 and its compact support . for notational convenience the dependency on @xmath124 and @xmath324
is suppressed .
they are arbitrary but fixed unless stated otherwise .
first note that for any random couple @xmath196 it holds true that @xmath706 hence it is sufficient to prove that for any sequence @xmath707 of admissible alternatives there exists a random sequence of @xmath708 with @xmath709 as in the proof of theorem [ thm : limit distribution ] define @xmath710 , @xmath711 .
let @xmath712 and @xmath713 .
define @xmath714 with @xmath715 now let the process @xmath716 on @xmath313 pointwise be defined by @xmath717 furthermore , let us introduce the random variables @xmath718 , based on the indices @xmath719 which are defined analogously to @xmath196 but with the minimum running over the set @xmath720 only .
then , recalling the definition @xmath721 , @xmath722 whereby we used for the first term in the last inequality that @xmath718 differs from @xmath723 for at most two indices @xmath724 ; the second term follows by including and evaluating the conditional expectation given @xmath718 as @xmath229 is independent of @xmath718 . replacing again @xmath718 by @xmath723 , the second expression behind the inequality in formula ( [ expr ] )
is bounded by @xmath725\bigg\arrowvert .
\label{expr2}\end{aligned}\ ] ] now we can make use of the fact that @xmath726 with @xmath727 . recall that @xmath728 is uniformly bounded due to the equicontinuity assumption on @xmath522 and the constraint on the @xmath523-norm @xmath524 , whence the term in ( [ expr2 ] ) is not greater than @xmath729 using the bounded total variation @xmath455 of @xmath28 and @xmath730 and @xmath529 as defined in the proof of theorem [ thm : limit distribution ] , the integral which appears in ( [ expr3 ] ) can be bounded by @xmath731}\bigg(\int i\big\{y\in m_x(t_n , r_n,\hat{t}_{n},\hat{r}_{n})\big\}dx\bigg)\nonumber\\ & = \
tv(\psi)\,\e\,\lambda\big(b_{t_n}(r_n)\delta b_{\hat{t}_{n}}(\hat{r}_{n})\big)\nonumber\\ & = \ o\big(r_n^{d-1}n^{-1/d}\big),\label{eq : rate}\end{aligned}\ ] ] using in the last bound besides the stochastic convergence rate @xmath732 the uniform integrability of the sequences @xmath733 , @xmath734 which result from @xmath735^d)\big)^{n}\ ( = ( 1-vx^d)^n\text{~if~ } b_{t_n}(x)\subset[0,1]^d)$ ] and @xmath736 . here
, @xmath737 denotes the volume of the @xmath539-dimensional euclidean unit ball , i.e. @xmath738 . together with ( [ expr ] ) - ( [ expr3 ] ) this shows that for any sequence of admissible alternatives @xmath707 @xmath739 if in particular @xmath740 , the term in ( [ eq : e - approximation ] ) is of order @xmath741 we need to check that @xmath742 for this we use the decomposition @xmath743\hat{\gamma}_n(t , r)^2\ = \ \hat{\gamma}_{2,n}(t , r)^2\ -\
\hat{\gamma}_{1,n}(t , r)^2 $ ] to this end note first that @xmath744 the `` @xmath745''-term results from the hlder continuity of @xmath28 ( for @xmath125 the first derivative of @xmath28 is uniformly bounded on @xmath746 $ ] ) , @xmath747 and the fact that @xmath748 while @xmath749 .
the case @xmath750 is done analogously ( taking the square ) . to verify ( [ eq : convergence variance ] )
it remains to be shown that @xmath751 which however is a simple consequence of chebychef s inequality since for any @xmath175 and any sequence of admissible alternatives @xmath707 , the sequence @xmath752 or some subsequence decreases ( if it decreases ) at a slower rate than @xmath753 .
the above considerations show in particular that @xmath754 using in addition that @xmath78 .
consequently , @xmath755 and it has to be verified that the latter quantity goes to infinity . recall that @xmath756^d } \psi_{t_nr_n}(x)^2h_n(x)dx\ -\ \bigg(\int_{[0,1]^d}\psi_{t_nr_n}(x)h_n(x)dx\bigg)^2\nonumber\\ & = \
\big(1+o(r_n^d)\big)\int_{[0,1]^d } \psi_{t_nr_n}(x)^2h_n(x)dx.\label{appp}\end{aligned}\ ] ] we first assume that @xmath757 , i.e. @xmath758 .
using that @xmath759^d}\sup_{x\in b_{t}(\delta)}\big\arrowvert h_n(x)-h_n(t)\big\arrowvert\ = \ 0,\ ] ] which follows by the same argument as used in theorem [ thm : lower bound ] and the fact that any sequence of centers @xmath760 has a convergent subsequence by the compactness of @xmath88^d$ ] , @xmath761^d}\psi_{t_nr_n}(x)\phi_n(x)dx}{\big[\int_{[0,1]^d}\psi_{t_nr_n}(x)^2dx\big]^{1/2}}\,\big(1+o(1)\big).\ ] ] using the approximation in ( [ appp ] ) we obtain analogously @xmath762^d}\psi_{t_nr_n}(x)^2dx\bigg)\bigg]^{1/2}.\ ] ] recall that @xmath763 with @xmath324 the bound of the support .
standard calculation shows that the bounded @xmath168-norm of @xmath150 implies @xmath764^{1/2}}\ = \ \frac{\big\arrowvert\int\psi_{t_nr_n;\beta}(x)\phi_n(x)dx\,\big\arrowvert}{\big[\int\psi_{t_nr_n;\beta}(x)^2dx\big]^{1/2}}\big(1+c_k\big)\ \
\text{with $ c_k\rightarrow 0 $ as~$k\rightarrow\infty$,}\ ] ] but note that the total variation @xmath765 is increasing in @xmath324 .
define now @xmath766 . then by its construction , @xmath767 .
moreover , by the closedness in @xmath168 and the convexity of the sets @xmath768 and @xmath769 , it results finally from convex analysis and the definition of @xmath150 that @xmath770^{1/2}}\
\geq\ \frac{\delta_n^{-1}\arrowvert \delta_n\psi_{t_nr_n;\beta}\arrowvert_2 ^ 2}{\arrowvert\psi_{t_nr_n;\beta}\arrowvert_2}\ = \ \delta_nr_n^{d/2}\arrowvert\gamma_{\beta}\arrowvert_2.\ ] ] combining ( [ appp ] ) ( [ eq : apppp ] ) , one verifies for the expression of the right hand side in ( [ term ] ) that it possesses the approximation @xmath771 which goes to infinity for @xmath324 sufficiently large . if there exists a sequence @xmath707 of admissible alternatives such that @xmath772 there exists by the considerations above a subsequence ( for simplicity also denoted by @xmath773 ) along which @xmath774 stays uniformly bounded away from zero . but the bounds ( [ eq : e - approximation ] ) and ( [ eq : convergence variance ] ) show that @xmath775 as well as the logarithmic correction term @xmath776 are in this case of smaller order than @xmath777 , which concludes the proof by contradiction .
@xmath282 [ [ proof - of - theorem - thm - spatial - adaptivity ] ] proof of theorem [ thm : spatial adaptivity ] + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + following the considerations of the proof of theorem [ thm : effizienz2 ] , it has to be established that there exist random sequences @xmath778 with @xmath779 , @xmath203 , such that for any sequence of alternatives as formulated in theorem [ thm : spatial adaptivity ] and any fixed @xmath184 @xmath780 then the result follows because the finite intersection of sets with asymptotic probability equal to @xmath249 has asymptotically mass @xmath249 as well . inspired by the arguments in rohde ( 2008 ) for the univariate regression context , we first establish the following : for @xmath781^d\big)$ ] with @xmath782 and @xmath783^d}\arrowvert \phi_n(x)\arrowvert$ ] , there exists some constant @xmath784 and a compact ball @xmath785 with center @xmath786 such that @xmath787^d\big ) \geq\
c\arrowvert \phi_n(x^*)\arrowvert^{d/\beta } \text{~~and~~ } \big\arrowvert\phi_n(x)\big\arrowvert\ \geq\ \frac{1}{2}\big\arrowvert \phi_n(x^*)\big\arrowvert\ \ \text{for all $ x\in b\cap[0,1]^d$}.\ ] ] assume that @xmath125 ( the above inequality is trivial in case @xmath121 ) .
with @xmath788 we denote subsequently some multi - index , where @xmath789 defines its length , @xmath790 and @xmath791 $ ] the partial differential operator .
let @xmath792^d\big)$ ] with @xmath793 . by the definition of the isotropic hlder class we have @xmath794 , which entails that @xmath795^d}\leq d+l\sqrt{d}^{\beta}$ ] .
in order to establish ( [ eq : function ] ) , note that for any polynomial @xmath796 , the topology induced by the metrics corresponding to the two norms @xmath797^d}\arrowvert p(x)\arrowvert$ ] and @xmath798 respectively on the ring of polynomials of total degree at most @xmath123 on @xmath88^d$ ] is the topology of uniform convergence , hence these two norms are equivalent . consequently , the boundedness of the polynomial @xmath799 by @xmath800 uniformly in @xmath801 implies that there exists some constant @xmath802 such that @xmath803 for all multi - indices @xmath804 with @xmath805 .
now the mean value theorem implies for some intermediate point @xmath806 @xmath807 thus , @xmath808^d.\ ] ] if @xmath792^d\big)$ ] with @xmath809 , then the function @xmath810 , for @xmath811^d$ ] pointwise defined by @xmath812^d\big\}$ ] is element of @xmath813 with @xmath814 . note that @xmath815 is a convex set .
therefore , the above considerations imply that @xmath816 on @xmath817^d.\ ] ] but then its lebesgue measure is always greater than @xmath818 for some constant @xmath819 , independent of @xmath482 and @xmath786 , hence ( [ eq : function ] ) is established .
let now @xmath820 fixed but arbitrary , @xmath202^d$ ] some nondegenerate rectangle , @xmath821 a sequence of functions with @xmath822 .
it has to be shown that there exists a universal constant @xmath823 such that @xmath824 whenever @xmath825 .
first , we choose a compact ball @xmath826 with center @xmath827 satisfying @xmath828 and ( [ eq : function ] ) .
let the couple @xmath829 be defined by @xmath830 consulting the proof of theorem [ thm : effizienz2 ] , this definition of @xmath723 allows for an approximation as in ( [ term ] ) .
since @xmath831 for all @xmath832 , @xmath833^d\big)\big]^{1/2}\\ & \geq\ c\arrowvert\tilde{\phi}_n\arrowvert_{j_i}^{(\beta+d/2)/\beta}\sqrt{n}\big(1+o(1)\big ) \end{aligned}\ ] ] for some universal constant @xmath834 . now the asserted result is easily deduced for a sufficiently large constant @xmath835.@xmath282 [ [ proof - of - theorem - thm - local - alternatives . ] ] proof of theorem [ thm : local alternatives ] .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + \(i ) let @xmath836 be such that @xmath837^d}$ ] and @xmath838 the sequence of piecewise constant functions on @xmath88^d$ ] with @xmath839 for @xmath840 , @xmath841 for @xmath842 and equals zero otherwise , where @xmath843 is such that @xmath844 and @xmath845 .
then @xmath846 with @xmath847 iid uniformly distributed on @xmath88^d$ ] .
note that @xmath848\\ & \sim\ \sum_{i=1}^{n_m}\log\big(1+(1-m / n)\frac{c_n}{\sqrt{n\delta_n^d}}\,r_i\big)\ + \ \sum_{j=1}^{n_{n - m}}\log\big(1-(m / n)\frac{c_n}{\sqrt{n\delta_n^d}}\,r_j'\big)\end{aligned}\ ] ] with @xmath849 and @xmath850 two independent sequences of iid rademacher variables , @xmath851 and @xmath852 independent with @xmath853 suppose first that @xmath854 . by extracting a subsequence
if necessary we may assume that @xmath855 , @xmath856 $ ] and @xmath857 .
then , with @xmath858 denoting weak convergence , @xmath859 \longrightarrow_w\ \q\label{eq : limit}\ ] ] with the convolution @xmath860 and the poisson weights @xmath861 . since @xmath862
, we can apply le cam s notion of contiguity ( le cam and yang 2000 , chapter 3 ) to conclude that @xmath863 consequently @xmath218 .
now assume that @xmath218 but @xmath864 . without loss of generality
we may assume that @xmath865
. then lindeberg s clt entails that ( [ eq : limit ] ) holds true with @xmath866 again , the limiting distribution satisfies @xmath862 , whence @xmath219 .
\(ii ) we begin as in the proof of theorem [ thm : effizienz2 ] , but with @xmath867 , @xmath868 and @xmath869 . adjusting ( [ expr ] ) ( [ eq : e - approximation ] ) yields
@xmath870 the arguments of the proof of theorem [ thm : effizienz2 ] apply again and lead to the expansion @xmath871 while with the same reasoning as in the proof of theorem [ thm : spatial adaptivity ] @xmath872 for some constant @xmath873 and @xmath874 .
thus , if @xmath223 and @xmath218 , ( [ ea ] ) goes to infinity and the result follows.@xmath282
we start with a basic but useful property of the solution to ( [ eq : rec 2 ] ) .
suppose @xmath171 has only finitely many extremal points .
from the last extremal point on the function @xmath171 is monotoneous and the integral over @xmath876 can only be finite if both envelopes are vanishing in @xmath875 .
now consider the case of infinitely many extremal points . since the @xmath168-norm of the solution ( [ eq : rec 2 ] ) is finite and if there exists a sequence of local extrema of @xmath171 which stays uniformly bounded away from zero , their width must be bounded by a zero sequence .
but now the result follows via contradiction of ( [ eq : function ] ) , which , of course , is also applicable for local extrema.@xmath282 let @xmath388 be fixed . define @xmath877 to be a positive real number such that the following conditions are satisfied : @xmath877 is a local extremal point , @xmath878 , @xmath879 ( doable by lemma [ lemm1 ] ) .
now extend the function @xmath880 to a compactly supported function @xmath881 such that @xmath882 , @xmath883 and @xmath884 smaller than @xmath885 ; this is possible for @xmath877 sufficiently large ( because the uniform boundedness from @xmath877 yields the boundedness of all partial derivatives by a multiple of @xmath397 with the same argument as used in the proof of theorem [ thm : spatial adaptivity ] ; so one may extend the function first to a compactly supported one in @xmath886 and then extend it close to zero such that its integral vanishes ) - we omit an explicit construction at this point . with @xmath397
sufficiently small , this construction leads to what was required in the proof of theorem [ thm : lower bound ] .
lutz dmbgen s contribution to the decoupling subsequent to an extended discussion in bern is gratefully acknowledged .
furthermore , i want to thank three unknown referees and an associate editor for their valuable comments and careful reading of the manuscript .
: : behnen , k. , neuhaus , g. and ruymgaart , f. ( 1983 ) .
two sample rank estimators of optimal nonparametric score - functions and corresponding adaptive rank statistics .
, 588599 .
: : belomestny , d. and spokoiny , v. ( 2007 ) . spatial aggregation of local likelihood estimates with application to classification .
, 22872311 .
: : bennett , g. ( 1962 ) .
probability inequalities for sums of independent random variables . , 3345 . : : butucea , c. and tribouley , k. ( 2006 ) .
nonparametric homogeneity tests . _
j. statist .
inference * 136 * _ , 597639 .
: : donoho , d. ( 1994a ) . statistical estimation and optimal recovery
, 238270 .
: : donoho , d. ( 1994b ) .
asymptotic minimax risk for sup - norm loss
solution via optimal recovery .
, 145170 .
: : ducharme , g.r . and ledwina . t. ( 2003 ) .
efficient and adaptive nonparametric test for the two - sample problem .
, 20362058 .
: : dudley , r.m .
, gin , e. and zinn , j. ( 1991 ) uniform and universal glivenko - cantelli classes . _
j. theoretical probability . * 4 * _ , 485510 .
: : dmbgen , l. ( 2002 ) .
application of local rank tests to nonparametric regression .
, 511537 .
: : dmbgen , l. and spokoiny , v.g .
multiscale testing of qualitative hypotheses .
, 124152 .
: : dmbgen , l. and walther , g. ( 2008 ) .
multiscale inference about a density .
, 17581785 ; _ accompagnying technical report available at _
http://arxiv.org/abs/0706.3968 : : eubank , r.l . and hart , j.d .
testing goodness - of - fit in regression via order selection criteria .
, 14121425 .
: : fan , j. ( 1996 ) .
test of significance based on wavelet thresholding and neyman s truncation .
, 674688 .
: : hjek , j. and @xmath887idak , z. ( 1967 ) . _ theory of rank tests .
_ academic press . : : gijbels , i. and heckmann , n. ( 2004 ) .
nonparametric testing for a monotone hazard function via normalized spacings . _
j. nonpar .
statist .
* 16 * _ , 463477 .
: : hoeffding , w. ( 1963 ) .
probability inequalities for sums of bounded random variables . , 1330 .
: : ingster , y. ( 1987 ) .
asymptotically minimax testing of nonparametric hypotheses .
, 553574 .
: : janic - wrblewska , a. and ledwina , t. ( 2000 ) . data driven rank test for two - sample problem .
_ scand .
j. statist .
* 27 * _ , 281297 .
: : klemel , j. and tsybakov , a. ( 2001 ) . sharp adaptive estimation of linear functionals . , 15671600 .
: : le cam , l. and yang , g. ( 2000 ) .
_ asymptotics in statistics : some basic concepts . _ springer , new york .
: : ledwina , t. and kallenberg , w.c.m .
consistency and monte carlo simulation of a data - driven version of smooth goodness - of - fit tests .
, 15941608 .
: : ledwina , t. ( 1994 ) .
data - driven version of neyman s smooth test of fit .
, 10001005 .
: : leonov , s.l .
( 1997 ) . on the solution of an optimal recovery problem and its applications in nonparametric statistics .
, 476490 .
: : leonov , s.l .
remarks on extremal problems in nonparametric curve estimation . , 169178 .
: : lepski , o. and tsybakov , a. ( 2000 ) .
asymptotically exact nonparametric hypothesis testing in sup - norm and at a fixed point .
_ probab .
theory rel .
fields * 117 * _ , 1748 . : : neuhaus , g. ( 1982 ) .
@xmath144-contiguity in nonparametric testing problems and sample pitman efficiency .
, 575582 .
: : neuhaus , g. ( 1987 ) .
local asymptotics for linear rank statistics with estimated score functions . , 491512 .
: : nussbaum , m. ( 1996 ) .
asymptotic equivalence of density estimation and gaussian white noise . , 2399 - 2430 .
: : de la pe@xmath286a , v.h .
a bound on the moment generating function of a sum of dependent variables with an application to simple sampling without replacement .
, 197211 .
: : de la pe@xmath286a , v.h .
( 1999 ) . a general class of exponential inequalities for martingales and ratios .
, 537564 .
: : pollard , d. ( 1984 ) .
_ convergence of stochastic processes .
_ springer .
: : rohde , a. ( 2008 ) .
adaptive goodness - of - fit tests based on signed ranks .
, 13461374 .
: : rufibach , k. and walther , g. ( 2009 ) . a block criterion for multiscale inference about a density , with applications to other multiscale problems . , to appear .
: : serfling , r.j .
probability inequalities for the sum of sampling without replacement . , 39 - 48 . : : shorack , g.r . and
wellner , j.a .
( 1986 ) . _ empirical processes with applications to statistics . _
wiley , new york .
: : spokoiny , v. ( 1996 ) .
adaptive hypothesis testing using wavelets .
, 24772498 .
: : van der vaart , a.w . and wellner , j.a .
_ weak convergence and empirical processes .
_ springer .
: : walther , g. ( 2010 ) .
optimal and fast detection of spatial clusters with scan statistics . , to appear . | based on two independent samples @xmath0 and @xmath1 drawn from multivariate distributions with unknown lebesgue densities @xmath2 and @xmath3 respectively , we propose an exact multiple test in order to identify simultaneously regions of significant deviations between @xmath2 and @xmath3 . the construction is built from randomized nearest - neighbor statistics . it does not require any preliminary information about the multivariate densities such as compact support , strict positivity or smoothness and shape properties .
the properly adjusted multiple testing procedure is shown to be sharp - optimal for typical arrangements of the observation values which appear with probability close to one .
the proof relies on a new coupling bernstein type exponential inequality , reflecting the non - subgaussian tail behavior of a combinatorial process . for power investigation of the proposed method a reparametrized minimax set - up
is introduced , reducing the composite hypothesis `` @xmath4 '' to a simple one with the multivariate mixed density @xmath5 as infinite dimensional nuisance parameter . within this framework ,
the test is shown to be spatially and sharply asymptotically adaptive with respect to uniform loss on isotropic hlder classes .
the exact minimax risk asymptotics are obtained in terms of solutions of the optimal recovery . |
graphene is an easily accessible , truly two - dimensional system which is attractive from the point of view of both the basic physics and possible applications . in the past decade
, it has attracted immense investigations.@xcite among these works , the time - resolved optical pump - probe measurement is widely used to investigate the dynamics of electrons in graphene.@xcite with this method , a pump pulse is used to excite electrons from the valence band to the conduction one and then a time - delayed probe pulse with the probe - photon energy @xmath0 is applied to detect the differential transmission ( dt ) .
the experimentally obtained dt under a probe - photon energy much higher than twice of the fermi energy is often positive with its fast relaxation of several hundred femtoseconds followed by a slower picosecond relaxation.@xcite from the dt , the conductivity @xmath1 is extracted . moreover , if @xmath0 is also much larger than the scattering rate and the electrons are considered to be free , the conductivity is given by ( @xmath2)@xcite @xmath3 , \label{conductivityinter}\ ] ] in which @xmath4 , @xmath5 is the electron distribution with @xmath6 @xmath7 representing the conduction ( valence ) band and @xmath8 stands for the resonant absorption state . with this equation and the temporal evolution of @xmath9 , the evolution of the electron distribution difference at the resonant absorption state @xmath10
is directly detected .
in addition to the positive dt , the negative dt , which indicates the increase of the conductivity after the pumping ( positive differential conductivity ) , has also been observed in different experiments.@xcite sun _
et al . _ reported a negative dt in a system of very high equilibrium fermi energy , with the probe - photon energy being lower than twice of the fermi energy.@xcite soon after that , the negative dt was also observed by george _
_ in the case where the probe - photon energy is as low as tens of mev.@xcite these two kinds of negative dt can be well understood theoretically . by considering the inter - band conductivity of free electrons [ eq . ( [ conductivityinter ] ) ] , the negative dt reported by sun _ et al._@xcite is shown to come from the weakening of the pauli blocking due to the heating of electrons by the pump pulse.@xcite as for the case with low probe - photon energy,@xcite the heating of the electrons can still contribute to the negative dt.@xcite nevertheless , with the scattering taken into consideration , the intra - band conductivity was found to be also important for the negative dt.@xcite in contrast to the previous two cases with moderate or small probe - photon energies , the negative dt has also been observed with the probe - photon energy being very high ( e.g. , @xmath11
ev).@xcite the mechanisms lead to this kind of negative dt is still unclear . at such high probe - photon energy ,
the fermi surface is less possible to be as high as half of the probe - photon energy and hence the weakening of the pauli blocking is less important . as for the intra - band conductivity , its strength is largely suppressed and hence whether it is strong enough to lead to the negative dt is unclear .
moreover , in previous works investigating problems of the negative dt , the scattering strength used was assumed to be constant.@xcite with constant scattering strength , the inter - band conductivity can not lead to the negative dt .
nevertheless , with the variation of the scattering strength taken into consideration , whether it can lead to negative dt is still unknown .
in addition , the variation of the coulomb self - energy after the pumping was also suggested to be the mechanism leading to the increase of the inter - band conductivity and hence the negative dt.@xcite however , whether it is strong enough to lead to a negative dt has not been fully theoretically demonstrated and it is even claimed that the negative dt itself may come from external factors.@xcite in this work , we calculate the optical conductivity based on the linear response theory with the keldysh green function approach@xcite by taking into account the contribution of the electron - phonon and the electron - impurity scatterings as well as the coulomb self - energy explicitly .
we show that for graphene with low equilibrium electron density without the coulomb - hole ( ch ) self - energy,@xcite the negative dt can appear when the probe - photon energy @xmath0 is low ( in the order of the scattering rate ) or when @xmath0 is high ( much larger than the scattering rate ) . for the low @xmath0 case
, the negative dt mainly comes from the large increase of the intra - band conductivity after the pumping due to the large variation of the electron distribution . as for the high @xmath0 case ,
the negative dt is shown to tend to appear when the hot - electron temperature @xmath12 is close to the equilibrium one and the chemical potential is higher than the equilibrium chemical potential but considerably lower than @xmath13 .
this negative dt is found to come from the variation of both the intra- and the inter - band components of the conductivity .
especially , for the inter - band conductivity , we find that both the scattering and the hartree - fock ( hf ) self - energy can cause the negative dt.@xcite in addition , the influence of ch self - energy is also addressed .
this paper is organized as follows . in sec . 2 , we set up the model and lay out the formalism . in sec .
3 the results obtained numerically are presented .
we summarize and discuss in sec . 4 .
the effective hamiltonian of graphene near the @xmath14 and @xmath15 points can be described by@xcite @xmath16 here , @xmath17
@xmath7 stands for @xmath14 ( @xmath15 ) valley , @xmath6 @xmath7 represents conduction ( valence ) band , @xmath18 denotes spin and @xmath19 with @xmath20 being the fermi velocity . in this work ,
we only investigate the linearly polarized normal incident light with the electric field along @xmath21 direction .
then , based on the effective hamiltonian , the current operator reads ( the in - plane photon momentum @xmath22)@xcite @xmath23 with the dipole matrices being @xmath24/2.\end{aligned}\ ] ] it has been shown that the hot - electron fermi distribution is quickly established after the pumping with the hot electron temperatures in the conduction and valence bands being the same while the chemical potentials being different.@xcite based on this result , the electron density matrix without interactions is set to be @xmath25 $ ] with @xmath26 being the chemical potential in band @xmath27 , @xmath28 and @xmath12 standing for the electron temperature .
then , from the kubo formula with the keldysh green function approach , the optical conductivity is expressed as@xcite @xmath29/(16\pi^3\omega ) .
\label{conductivity}\end{aligned}\ ] ] here , @xmath30 comes from the valley and spin degeneracies and @xmath31 + 1\}^{-1}$ ] . in this equation
, the two band indices @xmath27 and @xmath32 in @xmath33 $ ] can be used to distinguish the contributions of the intra- and inter - band conductivities .
the intra - band conductivity is defined as terms with @xmath34 and the inter - band one with @xmath35 . with this definition of the inter- and intra - band conductivities ,
their differential conductivities are defined as the difference of the corresponding conductivities after and before the pumping .
the electron spectral function is given by @xmath36^{-1}\}.\ ] ] here , @xmath37 represents the retarded self - energy from the electron - electron , electron - phonon and electron - impurity interactions . the self - energy from the electron - impurity scattering
is calculated from@xcite @xmath38[1+\cos(\theta_{\bf k}-\theta_{\bf k^\prime } ) ] .
\label{imselfenergy}\end{aligned}\ ] ] here , @xmath39 is the dielectric function calculated as scharf _
et al._@xcite and @xmath40 represents the two - dimensional bare coulomb potentials with @xmath41 standing for the background dielectric constant.@xcite for the self - energy from the electron - phonon scattering , although the temperatures of electrons and phonons can be different , the feynman rules and diagrammatic technique for the keldysh green function approach are still valid.@xcite with this approach , the retarded self - energy is obtained as @xmath42 . \label{phononselfenergy}\end{aligned}\ ] ] here , @xmath43 are the electron - phonon scattering matrices for phonons in branch @xmath44 with their detailed forms given by scharf _
et al._;@xcite @xmath45 - 1\}^{-1}$ ] is the distribution for phonons with energy @xmath46 and temperature @xmath47 . in the following ,
we ignore the effect of polaronic shifts , i.e. , the real part of the electron - phonon self - energy and compensate such approximation by renormalizing the spectral function to fulfill @xmath48 as done by scharf _
et al._.@xcite as for the self - energy from the coulomb interaction , we take into account the hf self - energy@xcite @xmath49 here , @xmath50 and @xmath51 are the electron and hole distributions , respectively .
this self - energy comes from the coulomb exchange self - energy@xcite @xmath52 \nonumber\\ & & \hspace{-1.5cm}\mbox{}\times f(\eta^\prime v_f{\bf k^\prime},\eta^\prime){v_{{\bf k^\prime}-{\bf k}}}/\epsilon_{{\bf k^\prime}-{\bf k } } , \label{hfselfenergy}\end{aligned}\ ] ] with the exchange term of the full valence band subtracted as it is already included in the original single - particle energy.@xcite however , it is known in semiconductor optics that there is a debye shift or the ch self - energy in addition to the hf self - energy , which is expressed as@xcite @xmath53 with @xmath54 representing the dielectric function at @xmath55 k and @xmath56 mev .
the physics of this ch self - energy is as follows .
after the pumping , the pump - excited electrons change the screening strength . as a result ,
the subtracted exchange term of the full valence band also varies [ eq . ( [ chselfenergy ] ) ] .
it has been shown in semiconductor optics that the ch self - energy leads to a strong renormalization of the band structure ( band - gap shrinkage).@xcite due to the similar renormalization of the band structure in graphene , it was speculated that this variation of the coulomb self - energy is responsible for the negative dt.@xcite in this work , we also calculate the influence of the coulomb self - energy to check this speculation .
from the relation between the transmission and conductivity@xcite @xmath57 one finds that the negative dt emerges when the optical conductivity increases after the pumping .
therefore , instead of calculating the dt , we investigate the variation of the conductivity before and after the pulse ( differential conductivity ) @xmath58 . here , for electrons before the pumping , we choose their temperature @xmath59 k. we concentrate on the case with the photon - excited electron density much larger than the equilibrium one and hence choose the chemical potential @xmath60 for convenience .
then , due to the symmetry between electrons and holes , one always has @xmath61 after the establishment of the fermi distribution . in our calculation , the material parameters are the same as in ref . and the impurity density @xmath62 @xmath63 .
the temperature of phonons @xmath47 is chosen to be the same as electrons and the substrate is chosen to be sio@xmath64 unless otherwise specified . in this section ,
we first give an analytical analysis on the intra- and inter - band components of the conductivity and then show their contributions to the total differential conductivity numerically .
the increase of the total conductivity after pumping may come from both the intra- and the inter - band components of the conductivity .
for the intra - band component in eq .
( [ conductivity ] ) , it can be simplified by assuming the self - energy @xmath65 .
further by using the condition @xmath66 and taking the limit @xmath67 , one has @xmath68 .
\label{intrasigma } \end{aligned}\ ] ] here , @xmath69 is the real part of the self - energy which comes from the hf and ch self - energies and @xmath70\nonumber\\ & & \hspace{-1.5 cm}=2\int^\infty_{\frac{-\mu^\prime}{k_bt_e}}dx{(k_bt_ex+\mu^\prime)}\frac{e^x}{(e^x+1)^2 } , \label{dform}\end{aligned}\ ] ] with @xmath71 . from eq .
( [ intrasigma ] ) , one finds that the increase of the scattering strength , i.e. , the decrease of @xmath72 , leads to the increase ( decrease ) of the conductivity when the probe - photon energy @xmath0 is larger ( smaller ) than @xmath73 .
moreover , from eqs .
( [ intrasigma ] ) and ( [ dform ] ) , it is found that the increase of the electron temperature @xmath12 and chemical potential @xmath74 as well as the decrease of the self - energy @xmath75 lead to the increase of @xmath76 and hence @xmath77 . as for the inter - band conductivity ,
its increase may come from the variations of the scattering ( the imaginary part of the self - energy ) , the hf self - energy and the ch self - energy ( the real part of the self - energy ) .
for the scattering , it can be shown that the inter - band conductivity increases with the decrease of the scattering strength.@xcite therefore , the positive differential conductivity may appear when the scattering strength decreases .
on the other hand , for the hf and ch self - energies , their influences are understood from the simplified eq .
( [ conductivity ] ) by neglecting the electron - impurity and electron - phonon scatterings : @xmath78\nonumber \\ & & \hspace{-0.5cm}=[f(-\varepsilon_{{\bf k}_\omega}+\sigma^{\rm ee}_{{\bf k}_\omega,-1},-1)-f(\omega-\varepsilon_{{\bf k}_\omega}+\sigma^{\rm ee}_{{\bf k}_\omega,-1},1)]\nonumber\\ & & \hspace{-0.cm}\times{g_d\sigma_0v^2_fk_\omega}/\{\omega[2v_f-\partial_k(\sigma^{\rm ee}_{{\bf k}_\omega,-1}-\sigma^{\rm ee}_{{\bf k}_\omega,1})]\}. \label{interch}\end{aligned}\ ] ] here , @xmath79 satisfies @xmath80 and @xmath81 stands for the corresponding self - energy from the coulomb interaction ( hf and ch self - energies ) . after the pumping , the difference of the hf self - energy @xmath82 may increase due to the increase of the carrier density and that of the ch self - energy @xmath83 can also increase due to the enhancement of the screening strength . as a result , @xmath84 may increase , which tends to enhance the inter - band conductivity . .
the states after the pumping are chosen to be ( a ) @xmath85 k and @xmath86 mev , and ( b ) @xmath87 k and @xmath86 mev .
the total conductivity without the ch self - energy in ( a ) is fitted for @xmath88 mev by eq .
( [ scfit ] ) with the fitted results plotted as the green dotted curve .
the inset zooms the range at high @xmath0 with the inter - band differential conductivity without the ch and hf self - energies plotted as brown solid triangles . the black chain line in the figure marks @xmath89.,width=328 ]
we then numerically calculate the probe - photon energy dependence of the total differential conductivity .
the case without the ch self - energy is first investigated .
it is noted that the chemical potential @xmath74 and temperature @xmath12 of hot - electrons vary with the temporal evolution after the pumping .
here , we take @xmath85 k and @xmath86 mev as an example to investigate the possible mechanisms discussed in sec .
the total differential conductivity as function of probe - photon energy is plotted in fig .
[ figsw1](a ) ( red solid curve ) . to identify the individual contributions of inter- and intra - band components
, we also plot the differential conductivity with only the inter - band component in the figure ( open squares ) .
one finds that , at very low probe - photon energy @xmath0 ( @xmath90 mev ) , the total differential conductivity is positive and far away from zero while the inter - band differential conductivity is negative .
this indicts that the large positive differential conductivity at low @xmath0 comes from the intra - band component . as analyzed in sec .
iiia1 , this large positive intra - band differential conductivity at low probe - photon energy @xmath0 comes from the variation of @xmath76 and @xmath72 after the pumping .
the variation of @xmath76 can be obtained from eq .
( [ dform ] ) if the electron distribution ( i.e. , @xmath12 and @xmath74 ) and the hf self - energy @xmath75 are known .
it is noted that @xmath75 for the temperatures and chemical potentials calculated here is always less than 3 mev , much smaller than the chemical potential after the pumping ( 300 mev ) .
hence , the influence of the hf self - energy on @xmath76 can be neglected and one obtains @xmath91 and @xmath92 mev for the cases before and after the pumping , respectively . to further identify the contribution of @xmath72 , we fit the total conductivity in the region @xmath93 mev with @xmath94 here , @xmath9 is included to take into account the influence of the inter - band conductivity .
@xmath72 is fitted to be 0.07 mev@xmath95 and 0.17 mev@xmath95 for the cases before and after pumping , respectively [ the obtained differential conductivity is plotted in fig .
[ figsw1](a ) as the green dotted curve ] .
one finds that the relative change of @xmath72 is much smaller than that of @xmath76 .
hence the variation of the intra - band conductivity mainly comes from the variation of the electron distribution , which leads to the huge increase of @xmath76 .
moreover , since the influence of the hf self - energy on @xmath76 can be neglected , its influence on the intra - band component of the differential conductivity is also marginal .
it is further noted from the figure that , with the increase of the probe - photon energy @xmath0 from zero , @xmath96 decreases fast and becomes negative around @xmath97 mev .
this quick decrease of @xmath96 can be understood from eq .
( [ intrasigma ] ) , which infers that the influence of the intra - band component of the differential conductivity is significant only when @xmath0 is in the order of @xmath73 .
in addition to the large positive differential conductivity shown in the case of low probe - photon energy @xmath0 , another small positive total differential conductivity also appears when @xmath0 is high ( larger than 1200 mev ) as shown by the red solid curve in the inset .
one also finds from the inset that the inter - band differential conductivity ( open squares ) is also positive .
moreover , it is smaller than the total one with their difference being in the same order ( @xmath98 ) as the inter - band differential conductivity .
this means that the positive total differential conductivity comes from both the inter- and the intra - band components and both contributions are important at such high @xmath0 . for the positive contribution from the intra - band component at such high @xmath0
, it still mainly comes from the large variation of the electron distribution after the pumping as the case with low probe - photon energy @xmath0 .
this is confirmed by calculating the intra - band component in eq .
( [ conductivity ] ) with the distribution [ @xmath99 fixed with the pumping .
we find that the obtained intra - band conductivity decreases more than 75% for all @xmath0 shown in the inset ( not plotted in the figure ) . as for the positive contribution from the inter - band component
, it may come from the variations of both the hf self - energy and the scattering strength after pumping as discussed in sec .
although the influence of the hf self - energy on the intra - band component is marginal , it may have non - negligible influence on the inter - band component . to further identify its contribution
, we also plot the inter - band differential contribution without the hf self - energy in the inset ( brown solid triangles ) .
one finds that the one with the hf self - energy is about @xmath100 times of that without the hf self - energy .
this means that the variations of both the hf self - energy and the scattering strength have important contributions to the positive inter - band differential conductivity at such high probe - photon energy .
we further investigate the influence of the ch self - energy on the differential conductivity . with the ch self - energy , the total differential conductivity and the one with only inter - band component
are plotted as the blue dashed curve and the blue solid dots in the figure , respectively .
one finds that their qualitative behaviors are similar to those without the ch self - energy , i.e. , the large positive differential conductivity from the intra - band component at low probe - photon energy @xmath0 as well as the emergence of positive differential conductivity at high @xmath0 .
nevertheless , it is seen that the differential conductivity with the ch self - energy included varies markedly compared with the one without the ch self - energy .
this means that the ch self - energy has pronounced influence on the differential conductivity .
moreover , it is also seen that with the ch self - energy , the total and the inter - band differential conductivities are very close at high @xmath0 as shown in the inset .
this indicates that the large influence of the ch self - energy at high @xmath0 is mainly on the inter - band component . to further show the influence of @xmath12
, we plot the probe - photon energy @xmath0 dependence of the differential conductivities at the electron temperature @xmath87 k and chemical potential @xmath101 mev in fig .
[ figsw1](b ) . for the case without the ch self - energy , the total differential conductivity ( red solid curve ) at high @xmath0 ( @xmath102 mev as shown in the inset ) becomes negative , which is different from the case at @xmath85 k. to show the origin of this negative differential conductivity , we plot the the inter - band one with ( red open squares ) and without ( brown solid triangles ) the hf self - energy in the inset .
it is shown that they are also negative and the total differential conductivity is close to the inter - band one without the hf self - energy .
this indicates that the negative total differential conductivity comes from the negative inter - band differential conductivity without the ch and hf self - energies under high temperature at high @xmath0 . to see the influence of the ch self - energy , we plot in the inset both the total ( blue dashed curve ) and the inter - band ( blue solid dots ) differential conductivities .
one finds both become positive for the high @xmath0 case .
this again shows that the ch self - energy has a large influence on the differential conductivity .
-@xmath74 space with @xmath103 k. ( b ) chemical potential dependence of the total differential conductivity at @xmath104 mev ( blue solid dots ) and 1700 mev ( blue solid curve ) as well as the inter - band one at @xmath104 mev ( red open squares ) and 1700 mev ( red dashed curve ) .
the chemical potential dependence of the inter - band differential conductivity without the hf self - energy is also plotted as the green dashed curve . here , @xmath105
k. the red arrows mark the peak of the inter - band differential conductivities .
( c ) total differential conductivity in the @xmath0-@xmath74 parameter space with @xmath47 being 200 k lower than @xmath12 .
( d ) chemical potential dependence of the total differential conductivity at @xmath106 mev as well as that for the inter - band one with and without the hf self - energy . here ,
@xmath107 k and @xmath108 k. ( e ) total differential conductivities in the @xmath12-@xmath74 parameter space .
here , @xmath104 mev and @xmath109 .
( f ) temperature dependence of the total differential conductivity as well as that for the inter - band one with and without the hf self - energy . here
, @xmath74 is chosen to be 320 mev .
for clarity , only the positive differential conductivity is plotted in ( a ) , ( c ) and ( e ) .
the chain lines in ( b ) , ( d ) and ( f ) mark @xmath89 .
, width=328,height=436 ] in the previous section , the mechanisms and their contributions to the positive differential conductivity are investigated under specific electron temperatures @xmath12 and chemical potential @xmath74 .
however , the hot - electron temperature and the chemical potential after the pumping vary with time . here , we investigate the conditions of @xmath12 and @xmath74 as well as the probe - photon energy @xmath0 for the emergence of the positive differential conductivity ( i.e. , negative dt ) .
since the positive differential conductivity in the low @xmath0 case has been shown to come from the intra - band component of the conductivity , in this section , we only investigate the case with high @xmath0 , where the negative dt was speculated to come from the scattering and the coulomb self - energy.@xcite the cases without the ch self - energy are first studied with the total differential conductivity in the @xmath0-@xmath74 parameter space plotted in fig .
[ figsw2](a ) . here
, the electron temperature is chosen to be 700 k , which is among those estimated in the experiment and the phonon temperature is chosen to be the same as that of electrons .
we only plot the region with positive differential conductivity for clarity .
it is found that if the probe - photon energy is larger than 1250 mev , the positive differential conductivity emerges with the corresponding range of @xmath74 depending on @xmath0 . to further show the contributions of the inter- and intra - band components , we plot the chemical potential dependence of the total and the inter - band differential conductivities in fig . [ figsw2](b ) with @xmath104 and 1700 mev . from the figure , one finds that the inter - band differential conductivities ( open squares and dashed curve ) are negative but the total conductivity is positive when @xmath110 mev @xmath111 mev for @xmath106 mev ( solid curve ) and @xmath112 mev @xmath113 mev for @xmath104 mev ( dots ) .
this indicates that the positive differential conductivity comes from the intra - band component .
this result is consistent with previous ones that the intra - band differential conductivity increases with @xmath12 and @xmath74 [ eq . ( [ dform ] ) ] while the inter - band one is negative if the hot - electron temperature is much larger than the equilibrium one .
moreover , one also finds that peaks appear in the chemical potential dependences of both the total and the inter - band differential conductivities with the corresponding @xmath74 considerablely smaller than @xmath13 . considering that the intra - band differential conductivity increases with @xmath74 monotonously , the peak shown in the total differential conductivity is understood to come from the inter - band one .
in addition , it is also shown in the figure that with the increase of @xmath0 , the peak of the inter - band differential conductivity appears at higher @xmath74 ( pointed by the arrows ) , this explains that the positive total differential conductivity tends to appear at higher @xmath74 with the increases of @xmath0 as shown in fig .
[ figsw2](a ) . to further show the influence of the hf self - energy on the inter - band differential conductivity
, we plot the inter - band differential conductivity at @xmath106 mev without the hf self - energy ( green dashed curve ) in fig .
[ figsw2](b ) .
it is shown that the inter - band differential conductivities with and without the hf self - energy are close at small @xmath74 ( @xmath114 mev ) .
this is understood since the hf self - energy is small when the excited electron density is small ( i.e. , @xmath74 is small ) . on the other hand ,
when the excited electron density is much larger than the equilibrium one ( i.e. , @xmath115 ) , the hf self - energy also becomes large and it leads to an obvious increase of the inter - band differential conductivity as shown in the figure .
one also finds from fig .
[ figsw2](a ) that the positive differential conductivity ( negative dt ) emerges only when the probe - photon energy @xmath0 is high enough .
this is supported by the experimental data by shang _
et al._.@xcite they showed that the negative dt appears when the probe - photon energy @xmath0 is larger than 2000 mev but disappears when @xmath0 is smaller than 1835 mev under very strong pumping strength . from the numerical results shown in fig .
[ figsw2](a ) ( where @xmath107 k ) , this property can be shown by electrons with the chemical potential @xmath74 around 490 mev .
one finds that the corresponding differential conductivity ( dt ) changes from negative ( positive ) to positive ( negative ) when @xmath116 mev .
it is noted that the electron density under these @xmath12 and @xmath74 is @xmath117 @xmath63 .
this electron density is also reasonable since the pumping strength used in the experiment is very strong .
it is noted that in our calculation , all modes of phonons are assumed to be at the same temperature as electrons .
nevertheless , in genuine situation , the phonon - phonon scattering is not strong enough to buildup such phonon distribution.@xcite to further investigate the influence of the phonon temperature , we also show the results with the phonon temperature being 200 k lower than the electron temperature @xmath12 in fig .
[ figsw2](c ) .
it is shown that with the decrease of the phonon temperature , the region of the positive differential conductivity is enlarged .
this is understood since the electron - phonon scattering strength decreases with the decrease of the phonon temperature , which increases the inter - band differential conductivity.@xcite to confirm this , we also plot the chemical potential dependence of the total and the inter - band differential conductivities with @xmath108 k and @xmath106 mev in fig .
[ figsw2](d ) .
compared to the one with @xmath118 k [ fig . [ figsw2](b ) ] , it is found that the inter - band differential conductivity without the hf self - energy ( green dashed curve ) increases markedly and it can even lead to positive differential conductivity at @xmath119 mev@xmath120 mev .
we further give a detailed investigation on the total differential conductivity in the electron - temperature chemical - potential ( @xmath12-@xmath74 ) space at a fixed probe - photon energy @xmath104 mev .
the phonon temperature is chosen to be the same as the electrons and the results are plotted in fig .
[ figsw2](e ) .
one finds that the positive differential conductivity can appear when @xmath121 k. to further identify the contributions of the inter- and intra - band components , we plot the temperature dependence of the differential conductivity in fig .
[ figsw2](f ) with @xmath122 mev .
one finds that the total differential conductivity ( solid curve ) is obviously larger than the inter - band one ( dotted curve ) .
this indicates the important positive contribution of the intra - band component .
moreover , it is also shown that the difference between the inter - band differential conductivities with ( red dotted curve ) and without ( green dashed curve ) the hf self - energy varies slowly with the temperature .
this means that in the parameter range investigated here , the influence of temperature on the hf self - energy is small .
in addition , it is also found that the inter - band differential conductivity with only scattering ( i.e. , without the hf self - energy ) can also be positive when @xmath12 is close to the equilibrium one ( 300 k ) .
however , it decreases with the increase of the temperature quickly and becomes negative when @xmath123 k. now , we can lay out the conditions for the emergence of the negative dt ( or equivalently , positive differential conductivity ) in graphene with low equilibrium electron density .
the negative dt can appear both in the low probe - photon energy @xmath0 ( in the order of the scattering rate ) case and in the high @xmath0 case ( much larger than the scattering rate ) .
for the low @xmath0 case , this negative dt mainly comes from the increase of the intra - band conductivity .
this increase is mainly due to the large variation of the electron distribution after the pumping and is almost not influenced by the hf self - energy in the temperature range investigated here . as for the high @xmath0 case , although the dt for free electrons [ i.e. , the one with the conductivity calculated from eq .
( [ conductivityinter ] ) ] can never be negative , the negative dt can appear if the scattering is taken into consideration . with the scattering
, the inter - band conductivity tends to lead to a negative dt when the hot - electron temperature @xmath12 is close to the equilibrium one and the chemical potential @xmath74 is higher than the equilibrium chemical potential but considerably smaller than @xmath13 .
moreover , this region of the negative dt determined by the inter - band conductivity can be further expanded by the intra - band one , which always tends to decrease the dt if @xmath12 and @xmath74 increase after the pumping .
furthermore , the hf self - energy can also contribute to the negative dt by increasing the inter - band differential conductivity when @xmath74 is much larger than the equilibrium chemical potential ( about several hundred milli - electron volts larger ) in the temperature range usually estimated in the experiment ( around 1000 k ) .
this further expands the region of the negative dt at high @xmath0 .
we further note that if the phonon temperature is considered to be smaller than that of electrons , the inter - band differential conductivity increases , which also expands the region of the negative dt .
( a ) and sic ( b ) in @xmath12-@xmath74 parameter space . for clarity ,
only the positive differential conductivity is plotted.,width=328 ] finally , we address the influence of the ch self - energy on the emergence of the negative dt ( positive differential conductivity ) . the total differential conductivity with the ch self - energy for graphene on sio@xmath64 substrate in the electron - temperature chemical - potential ( @xmath12-@xmath74 ) parameter spaces is plotted in fig . [ figsw3](a ) .
here , the probe - photon energy @xmath0 is chosen to be 1500 mev . comparing with the results without the ch self - energy [ fig .
[ figsw2](e ) ] , one finds that the ch self - energy enhances both the region and the magnitude of the positive differential conductivity markedly .
this means that the variation of the ch self - energy after the pumping is the dominant mechanism leading to the positive differential conductivity at such high @xmath0 if it is taken into consideration .
it is worth noting that , the shape for the region of the positive differential conductivity with the ch self - energy is similar to the one without it [ fig .
[ figsw2](e ) ] ( this is also true for @xmath0-@xmath74 parameter space ) .
this means that the tendency for the emergence of the total differential conductivity summarized in sec .
iiib is still valid , although the variation of the ch self - energy after the pumping becomes the dominant mechanism leading to the positive differential conductivity .
it is noted that since the ch self - energy mainly comes from the variation of the screening , its influence should be smaller if the graphene layer is on a substrate with higher background dielectric consistent . to confirm this
, we show the results for graphene on sic [ fig .
[ figsw3](b ) ] , whose background dielectric constant is 2.2 times of that for sio@xmath64 .
one finds that both the magnitude and the region of the positive differential conductivity decrease obviously compared with the one on sio@xmath64 substrate . nevertheless , the positive differential conductivity ( negative dt ) region shown on sic substrate is still very large .
actually , the negative dt region shown on sio@xmath64 and sic substrates with the ch self - energy is so large that it disagrees with some of the existing experiments.@xcite dawlaty _
et al._@xcite have shown experimentally for graphene on sic substrate that the dt is positive when the optical excited electron density is about @xmath124 @xmath63 with the corresponding @xmath12 estimated to be around 900 k and @xmath74 around zero under the probe - photon energy close to 1500 mev.@xcite however , the numerically obtained dt with these parameters is negative as shown in fig .
[ figsw3](b ) .
moreover , ruzicka _
et al._@xcite and brida _
et al._@xcite also showed in graphene on sio@xmath64 substrate that the dt is positive when the excited electron density is about @xmath125 @xmath63 with the corresponding @xmath12 estimated to be around 1000 k and @xmath74 around @xmath126 mev.@xcite however , the numerically obtained dt with the ch self - energy is also negative as shown in fig .
[ figsw3](a ) .
nevertheless , for the case without the ch self - energy , the dts under these parameters are all positive [ fig .
[ figsw2](e ) for the sio@xmath64 substrate and that for the sic substrate is not shown ] .
further considering the fact that the result obtained without the ch self - energy is consistent with the experiment with the negative dt ( sec .
iiib),@xcite one concludes that although the calculation of the ch self - energy used here is very successful in semiconductors,@xcite it overestimates the coulomb self - energy in the gapless graphene system .
this may come from the approximations used to obtain eq .
( [ chselfenergy ] ) .
for example , the background dielectric constant @xmath41 is taken to be constant although it is shown to increase with @xmath127 [ ref . ] and the vertex correction@xcite for the coulomb screening is neglected .
a better description on the influence of the coulomb self - energy is beyond the scope of this investigation .
in summary , we have investigated the emergence of the negative dt by calculating the optical conductivity based on the linear response theory with the keldysh green function approach . in our calculation
, the influences of the electron - phonon and electron - impurity scatterings as well as the coulomb self - energy are explicitly included .
we investigate the system with the photo - excited electron density much larger than the equilibrium one and show that the negative dt ( or equivalently , the positive differential conductivity ) can appear at low probe - photon energy @xmath0 ( in the order of the scattering rate ) or at high @xmath0 ( much larger than the scattering rate ) . for the low @xmath0 case ,
the negative dt is shown to mainly come from the large increase of the intra - band conductivity due to the large variation of the electron distribution after the pumping .
as for the high @xmath0 case , it is shown that the inter - band conductivity with only scattering can lead to the negative dt when the hot - electron temperature @xmath12 is close to the equilibrium one and the chemical potential @xmath74 is higher than the equilibrium chemical potential but considerably lower than @xmath13 . moreover
, the intra - band conductivity can further expand this region of the negative dt because it always tends to decrease dt if @xmath12 and @xmath74 increase after the pumping .
furthermore , in the temperature range usually estimated in the experiments ( around 1000 k ) , the hf self - energy is also shown to be able to contribute to the negative dt by increasing the inter - band differential conductivity considerably when @xmath74 is much higher than the equilibrium one .
this further expands the region of the negative dt at high @xmath0 .
nevertheless , the hf self - energy has little influence on the intra - band conductivity and hence its effect on the negative dt at low @xmath0 can be neglected .
in addition , we also show that if the phonon temperature is smaller than that of electrons , the inter - band differential conductivity is increased which also expands the region of the negative dt .
our numerical results are consistent with the @xmath0 dependence of the negative dt obtained experimentally by shang _
et al._.@xcite the influence of the ch self - energy is also investigated .
we find that it markedly expands the region of the negative dt .
however , this expansion is too large which makes the obtained region of the negative dt become incompatible with the existing experiments.@xcite this means that the ch self - energy calculated with the well established approach in semiconductors@xcite overestimates the coulomb self - energy in the gapless graphene system .
more investigation to take into account the high order correlation is needed .
the authors would like to thank t. s. lai for valuable discussions .
this work was supported by the national basic research program of china under grant no .
2012cb922002 and the strategic priority research program of the chinese academy of sciences under grant no . xdb01000000 . | by using the kubo linear response theory with the keldysh green function approach , we investigate the mechanism leading to the negative differential transmission in system with the equilibrium electron density much smaller than the photon - excited one .
it is shown that the negative differential transmission can appear at low probe - photon energy ( in the order of the scattering rate ) or at high energy ( much larger than the scattering rate ) .
for the low probe - photon energy case , the negative differential transmission is found to come from the increase of the intra - band conductivity due to the large variation of electron distribution after the pumping . as for the high probe - photon energy case ,
the negative differential transmission is shown to tend to appear with the hot - electron temperature being closer to the equilibrium one and the chemical potential higher than the equilibrium one but considerably smaller than half of the probe - photon energy .
we also show that this negative differential transmission can come from both the inter- and the intra - band components of the conductivity .
especially , for the inter - band component , its contribution to the negative differential transmission is shown to come from both the hartree - fock self - energy and the scattering .
furthermore , the influence of the coulomb - hole self - energy is also addressed . |
the aim of this paper is to demonstrate that reactions of exclusive vector meson photo- and electroproduction can be described in the framework of a regge - eikonal scheme ( the general regge - eikonal approach was developed in @xcite ) similar to that as for high - energy nucleon - nucleon scattering .
one of the distinctive features of the proposed model is the use of nonlinear parametrizations for regge trajectories in the diffraction region in which their qcd asymptotic behavior is taken into account explicitly .
in essence , the behavior of regge trajectories at large transferred momenta is a fact of fundamental importance : qcd does not permit linear trajectories .
hence , there emerges the problem of compatibility of nonlinearity of regge trajectories following from qcd with available set of experimental data .
the scheme proposed in @xcite and applied in this paper partially solves this problem since in the framework of our approach phenomenological regge trajectories in the diffraction region satisfy the qcd asymptotic relations explicitly .
the detailed discussion of the fundamental properties of regge trajectories ( such as their essential nonlinearity in the euclidean domain ) can be found in @xcite .
here we limit ourselves by substantiating several useful relations .
our viewpoint ( by no means very original ) is based on the conviction that qcd is the fundamental theory of strong interaction .
so , in the limit of large negative values of the argument ( equivalent to short distances ) the exchange by any reggeon must turn into exchange by some colour - singlet parton combination and the asymptotic behavior of regge poles corresponding to exchanges by definite parton combinations can be determined .
for example , if some reggeon corresponds to exchange by the quark - antiquark pair ( @xmath6 reggeon , etc . ) in the range of the perturbative qcd validity we come to @xcite @xmath7 in the case of multigluon exchange one obtains @xcite @xmath8 and , in particular , for two - gluon exchange @xcite @xmath9 where @xmath10 is the qcd running coupling .
besides , if we assume that @xmath11 increases slowly enough at @xmath12 ( for example , not faster than @xmath13 , @xmath14 ) so that the dispersion relations with not more than one subtraction take place , i.e. @xmath15 and @xmath16 at @xmath17 we obtain @xmath18 it is usually assumed that for the true regge trajectories relations ( [ gerg ] ) are fulfilled @xcite . in @xcite , it was demonstrated that taking into account physical restrictions ( [ meson ] ) , ( [ gluon ] ) , ( [ gerg ] ) allows us to describe the diffractive pattern of high - energy nucleon - nucleon scattering in the framework of the minimal phenomenological regge - eikonal model with only three ( nonlinear ) regge trajectories ( soft pomeron and two secondary reggeons ) .
behavior is so much important for diffractive scattering which occurs mainly at quite small transferred momenta .
is nt it more economical to deal with much more simple and convenient in many respects linear approximation @xmath19 ? in ref .
@xcite , we demonstrated that such an approximation is valid only in quite a narrow interval of @xmath20 ( e.g. @xmath21 for high - energy @xmath22 scattering ) . ] below we will show that such an approach also allows us to describe available data on high - energy ( @xmath23 ) exclusive vector meson photo- and electroproduction in the framework of the similar model , using only two @xmath5-even vacuum regge trajectories , namely , soft and hard pomerons ( here we follow the pattern of donnachie and landshoff @xcite ) .
the soft pomeron trajectory is the same as for the diffractive nucleon - nucleon scattering ( in accordance with the requirement of universality of regge trajectories ) and the hard pomeron trajectory , lying much higher than the soft one in the euclidean domain ( see fig . [ trav ] ) , has the asymptotic behaviour ( [ hard ] ) ( in accordance with the fact that at high transfers the high - energy scattering amplitude is dominated by two - gluon exchange ) .
the `` invisibility '' of the hard pomeron in the elastic diffraction may be explained by the fact that at available energies its contribution to the on shell eikonal is suppressed by the soft pomeron due to the disparity in on shell residues @xmath24 ( subscripts `` @xmath25 '' and `` @xmath26 '' correspond to hard and soft ) in the nonperturbative regime .
we will treat the processes of exclusive vector meson production from the standpoint of the vector - dominance model ( vdm ) @xcite in which the incoming photon fluctuates into a virtual vector meson which , in turn , scatters from the target proton . according to the so - called hypothesis of @xmath27-channel helicity conservation ( the adequacy of this approximation
is confirmed by numerous experimental data on helicity effects @xcite ) the cross section of reaction @xmath28 is dominated by the nonflip helicity amplitude which can be represented in the form @xmath29 where the sum is taken over all neutral vector mesons .
@xmath30 is the coefficient of vector dominance dependent on the type of vector meson @xmath31 , helicity @xmath32 and virtuality @xmath33 of the incoming photon .
@xmath34 is the diffractive hadronic amplitude with regge - eikonal structure ( the applicability of the regge - eikonal approach to the hadronic reactions with off shell particles is grounded in @xcite ) .
it is known that for small enough values of @xmath20 the diagonal coupling of vacuum reggeons to hadrons is much stronger than the off diagonal coupling ( for example , elastic proton - proton scattering has a considerably larger cross section than diffractive excitation of n(1470 ) in the proton - proton collisions ) .
so , in the high - energy range ( @xmath23 ) where vacuum exchanges exceed the nonvacuum ones ( the main indication of such an exceeding is the manifest growth of integrated cross sections with the collision energy increase ) the so - called `` diagonal approximation '' of ( [ genvdm ] ) may be used @xmath35 ( conversely , for a consistent description of the data at @xmath36 the contribution from nonvacuum reggeons must be taken into account and approximation ( [ diavdm ] ) becomes invalid ) . here
we must note that recent measurements by zeus collaboration @xcite did not reveal patent dependence on @xmath20 and @xmath37 of the ratio of differential cross sections for the cases of longitudinally and transversely polarized incoming photon and this is the reason why we have omitted the superscript @xmath32 for @xmath38 in ( [ diavdm ] ) .
in other words , we presume that spin phenomena related to the dependence of @xmath38 on helicities of incoming and outgoing particles are much more fine effects than diffractive scattering itself and further deal only with quantities averaged over spin states . according to ( [ diavdm ] ) and regge theory @xcite the pole contribution of any @xmath5-even reggeon @xmath39 to the amplitude @xmath40 has the form ( the use of the designation @xmath41 for the pole contribution is caused by the subsequent use of the regge - eikonal scheme ; see below ) @xmath42 where @xmath43 , @xmath44 is the reggeon trajectory , and @xmath45 , @xmath46 are the reggeon form factors of participating particles .
the vdm in its simplest form gives @xcite @xmath47 where @xmath48 is the vector meson mass , @xmath49 is the electromagnetic coupling , and @xmath50 is the width of the vector meson decay to the electron - positron pair .
however , these relations contradict both the experimental data on @xmath33 dependence of the ratio @xmath51 @xcite and the dimensional counting rules @xcite according to which functions @xmath52 , @xmath53 , and , hence , @xmath54 must exhibit moderate ( nonpowerlike ) @xmath33 behavior . to resolve this conflict we have to modify the vdm coefficients @xmath55 and thus @xmath56 @xmath57 where , for any virtual photon spin state @xmath32 , @xmath58 , and function @xmath59 exhibits nonsteep @xmath33 behavior . further on
, we will concentrate on the diffractive pattern ( @xmath20 dependence ) of exclusive vector meson production and its evolution with collision energy ( @xmath37 dependence ) . since the second factor in the right - hand side of ( [ diavdm ] )
nearly does not depend on @xmath32 we may consider only cross sections averaged over helicities of external particles . the averaged pole contribution of a reggeon @xmath39 to the amplitude takes the form @xmath60 @xmath61 where @xmath62 @xmath63 in a well - known paper @xcite , it was argued that @xmath64 where @xmath65 is the vector meson `` bare '' mass . at infinite values of @xmath65
the usual form of the current - field relation takes place .
an interesting feature of this variant is a possibility to extend the vdm applicability for arbitrarily high @xmath33 .
we do not fix the concrete functional form of @xmath66 ( it does not help to determine @xmath33 behavior of @xmath67 because of the absence of detailed information on @xmath33 dependence of @xmath45 ) . according to the dimensional counting rules @xcite @xmath68 at high @xmath33 ( we point out that such a behavior implies the violation of bjorken scaling @xcite ) , so @xmath67 must exhibit slow evolution with @xmath33 . after these general remarks we can proceed to construct the scattering amplitude .
it is well known that for soft reactions such as elastic nucleon - nucleon scattering or light vector meson photoproduction the introduction of only one trajectory with intercept higher than unity is necessary ( the soft pomeron ) .
but in the processes with a hard scale ( for example , photon virtuality or / and heavy vector meson mass ) the rise of the cross sections with the collision energy becomes noticeably faster than in the soft ones and so there emerges the question if there exists one or several extra reggeons lying higher than the soft pomeron but with entirely suppressed residue(s ) in the nonperturbative regime @xcite . residues for secondary trajectories ( @xmath6 reggeon etc . ) .
the last circumstance looks too counter - intuitive and a hard pomeron seems inevitable . ] below we will use the minimal phenomenological scheme based on the idea of donnachie and landshoff @xcite that only one extra reggeon with the intercept much higher than the soft pomeron s one is needed .
the difference between our realization of this idea and the donnachie - landshoff model lies in a choice of the functional form of regge trajectories since the authors @xcite insist on the strict linearity of the trajectories while we use essentially nonlinear parametrizations with asymptotic behavior following from qcd ( cf .
the footnote 2 ) .
we assume that at collision energies @xmath23 we may neglect contributions from secondary reggeons . in this case
the minimal vacuum exchange off shell eikonal ( the sum of one - reggeon - exchange amplitudes @xcite ) takes the form @xmath69
@xmath70 @xmath71\,,\ ] ] where @xmath72 , @xmath73 are regge trajectories of the soft and hard pomerons , @xmath74 and @xmath75 are the soft pomeron form factors of outgoing vector meson and proton , and @xmath76 is the regge residue of the hard pomeron .
factors @xmath77 and @xmath78 are absorbed into an overall hard residue @xmath79 since the hard pomeron gives no noticeable contribution to the eikonal of nucleon - nucleon scattering at accessible energies and so we can not separately fix the form factor @xmath78 from the data on high - energy proton-(anti)proton diffraction .
the corresponding on - shell eikonal of elastic @xmath80 scattering is of the form : @xmath81 @xmath82 since regge trajectories are universal functions of one argument , i.e. depend neither on the type of the process in which the corresponding reggeons contribute nor on the virtualities of the incoming particles , we will use , for the description of exclusive diffractive production of vector mesons , the same phenomenological expressions for the soft pomeron trajectory and the corresponding form factor of the proton as we did for high - energy nucleon - nucleon scattering @xcite : @xmath83\,,\ ] ] @xmath84 ( the values of free parameters obtained by fitting the data on proton-(anti)proton angular distributions are represented in table [ tab1 ] ) .
.parameters for @xmath72 and @xmath75 obtained by fitting the data on proton-(anti)proton angular distributions . [ cols="<,<,<,<",options="header " , ] at last , after determining the hard pomeron trajectory we can describe the phenomenology of exclusive production of @xmath0 @xcite .
for the case of photoproduction , @xmath85 , we can neglect the contribution from the hard pomeron .
however , high quality of the data on integrated cross sections at nonzero @xmath33 does not allow us to do this for the deeply virtual electroproduction .
besides , determination of @xmath20 slopes of @xmath86 and @xmath87 is quite a hard task because of little amount and rather poor - quality data on the electroproduction angular distributions .
so , for an unambiguous description we put the values of these slopes at nonzero @xmath33 equal to those for @xmath2-photoproduction , i.e. @xmath88 and @xmath89 .
the results of the fitting data are represented in table [ tab4 ] and figs .
[ rho0][elarho ] . at the end of this section
we would like to emphasize that the main success of the used model consists in a satisfactory description of electroproduction of @xmath1 meson at @xmath90 ( fig .
[ phi5 ] : 1 free parameter versus 35 points ) and photoproduction of @xmath2 meson ( figs .
[ jps01 ] , [ jps02 ] , [ jps03 ] : 6 free parameters versus 114 points ) .
all other sets of data at fixed @xmath33 do not , in fact , allow unambiguous verification of validity of our regge trajectories because of the insufficient amount of data on angular distributions ( the data on electroproduction of @xmath0 and @xmath2 do not even allow us to fit parameters @xmath91 and @xmath92 unambiguously ) .
two main features of our approach are the use of an eikonalized ( unitarized ) expression for the scattering amplitude and the exploitation of the essentially nonlinear parametrizations of regge trajectories with explicit qcd asymptotic behavior .
although there exists a whole sea of papers devoted to the description of @xmath37 and @xmath33 evolution of integrated cross sections and structure functions , only few of them deal with diffractive pattern ( @xmath20 dependence ) of the scattering amplitude .
however , angular distributions contain very valuable information about the geometry of the interaction region and so they are more important quantities than integrated cross sections and structure functions .
the papers which deal with this subject can be conventionally sorted into two groups .
the first group consists of the purely phenomenological works . in @xcite ,
the so - called `` dipole pomeron model '' is used with the soft pomeron as a double regge pole with a square root threshold singularity . in @xcite , the model of dual amplitudes with mandelstam analyticity ( dama ) with a square root leading singularity
is used for the description of the @xmath2 photoproduction .
these two approaches , in fact , do not imply the concrete functional dependence of regge trajectories ( the authors of these papers use nonlinear parametrizations for regge trajectories ) and so are not in contradiction of principle with the requirement for reggeons to have a qcd asymptotic behavior . in @xcite , the usual born approximation for the scattering amplitude is used with postulated strictly linear regge trajectories .
we do not share this point of view since we consider the consistency with qcd ( which implies nonlinearity ) more important than quite an arbitrary hypothesis about strict linearity .
the second group consists of two papers @xcite exploiting the `` colour dipole model '' with a main hypothesis that if the incoming photon is of high virtuality or the outgoing vector meson is of high mass then the photon fluctuates into a quark - antiquark pair which scatters elastically off the proton and , at last , recombines into the vector meson .
the main advantage of this approach is that it allows us to get a concrete functional form of @xmath33 dependence for reggeon form factors in the perturbative region .
hence , the only functional uncertainties in the scattering amplitudes is the on - shell reggeon form factors and regge trajectories .
however , these functions must be the same as for the corresponding elastic scattering and , so , the concrete applications of the colour dipole approach to the vector meson electroproduction must be supplemented by the simultaneous description of the high - energy proton-(anti)proton diffractive scattering with the same regge trajectories and form factors of the proton .
this has not been done in @xcite . nevertheless , the colour dipole approach is still the only existing method for the quantitative estimation of the @xmath33 evolution of the off shell reggeon form factors . by no means
is it in contradiction with the phenomenological approach used in our paper ( the colour dipole model allows us to determine the @xmath33 dependence while the regge - eikonal model allows us to determine @xmath37 and @xmath20 dependence ) and a synthesis of these two approaches seems promising for better understanding of the diffraction mechanism . after obtaining the explicit @xmath33 dependence of the reggeon form factors in the framework of the colour dipole approach the verification of the proposed minimal regge - eikonal model
will become possible not only by fitting to the hera data at separate values of @xmath33 but also by the simultaneous fit to the whole massive of hera data on @xmath93 and integrated cross sections .
moreover , the model will be useful for predictions of cross sections of the exclusive vector meson photo- and electroproduction in proton-(anti)proton collisions at rhic , tevatron , and cern lhc energies @xcite ( the absorptive corrections due to the soft pomeron exchanges between incoming and between outgoing nucleons may be obtained automatically since we know the soft pomeron trajectory and the corresponding form factor of the nucleon ) .
now we can summarize the results of our analysis .
the successful simultaneous description of the data on elastic proton-(anti)proton scattering @xcite and the data on electroproduction of @xmath1 meson at low values of @xmath33 especially at @xmath90 in the framework of the one - pomeron model is the evidence that the soft pomeron absolutely dominates not only in the high - energy nucleon - nucleon scattering but also in the low @xmath33 electroproduction of light vector mesons at available energies .
the results on the description of the high - quality data on integrated cross sections of @xmath0 electroproduction confirm that for electroproduction of light mesons at low @xmath33 ( about several @xmath94 ) the contribution of the hard pomeron to the eikonal ( born term ) is several times less than the soft pomeron s one . the successful description of the @xmath2 photoproduction in the framework of a two - pomeron model with the hard pomeron that has the kirschner - lipatov asymptotic behavior @xcite in the deeply perturbative domain points to the fact that the hard pomeron has a perturbative nature ( contrary to the soft pomeron ) and dominates over the soft pomeron at high @xmath33 and high @xmath37 in accord with expectations of the parton model ( the smaller residue slope than the soft pomeron s one results in the domination of the hard pomeron also at larger @xmath20 ) . among other things , we explicitly demonstrated that neglecting the contribution of regge cuts is not justified _ a priori _ and so in any phenomenological model the validity of the born approximation must be grounded by explicit estimation of absorptive corrections .
also it was shown that when considering the processes of exclusive vector meson production there is no need to introduce the `` effective dependence '' of vacuum regge trajectories on the virtuality of the incoming photon ( this idea is very popular ; see , for example , @xcite ) .
in fact , such a dependence contradicts general principles since the true regge trajectories are universal analytical functions of one argument and do not depend on the properties of external particles ( such as virtuality ) .
this hypothesis is disguised by the approximation @xmath95 which may be valid only in a limited kinematical domain .
thus , it was shown that at collision energies higher than @xmath96 available data on exclusive diffractive reactions @xmath28 ( @xmath97 ) , @xmath98 , @xmath99 may be described in the framework of a simple phenomenological regge - eikonal model .
this was achieved by using nonlinear regge trajectories in which their qcd asymptotic behavior was taken into account explicitly .
we would like to point out that linear regge trajectories not only are in contradiction with qcd but even a satisfactory phenomenological description of the available data on @xmath20 dependence of exclusive production of vector mesons has not been attained yet within approaches using linear trajectories ( other phenomenological models with @xmath33-independent regge trajectories and acceptable value of @xmath100dof @xcite , @xcite also use nonlinear approximations to them ) .
j. kwiecinski , phys.rev .
d * 26 * ( 1982 ) 3293 + r. kirschner , z.phys .
c * 67 * ( 1995 ) 459 h. cheng , t.t .
wu , phys.rev.lett .
* 22 * ( 1969 ) 666 + f.e .
low , phys.rev .
d * 12 * ( 1975 ) 163 + s. nussinov , phys.rev .
d * 14 * ( 1976 ) 246 + p.d.b .
collins , p.j .
kearney , z.phys .
c * 22 * ( 1984 ) 277 r. kirschner , l.n .
lipatov , z.phys .
c * 45 * ( 1990 ) 477 e. martynov , e. predazzi , a. prokudin , phys.rev .
d * 67 * ( 2003 ) 074023 + r. fiore , l.l . jenkovszky , f. paccanoni , a. prokudin , phys.rev .
d * 68 * ( 2003 ) 014005 r. fiore , l.l .
jenkovszky , v.k .
magas , f. paccanoni , a. prokudin , + phys.rev .
d * 75 * ( 2007 ) 116005 h. kowalski , l. motyka , g. watt , phys.rev .
d * 74 * ( 2006 ) 074016 c. marquet , r. peschanski , g. soyez , phys.rev .
d * 76 * ( 2007 ) 034011 zeus collaboration ( m. derrick et al . ) , phys.lett .
b 377 ( 1996 ) 259 zeus collaboration ( m. derrick et al . ) , z.phys .
c * 63 * ( 1994 ) 391 + zeus collaboration ( m. derrick et al . ) , z.phys .
c * 69 * ( 1995 ) 39 + h1 collaboration ( s. aid et al . ) , nucl.phys .
b * 463 * ( 1996 ) 3 + zeus collaboration ( m. derrick et al . ) , z.phys .
c * 73 * ( 1997 ) 253 + zeus collaboration ( j. breitweg et al . ) , eur.phys.j .
c * 2 * ( 1998 ) 247 + zeus collaboration ( j. breitweg et al . ) , eur.phys.j .
c * 14 * ( 2000 ) 213 | the processes of exclusive photo- and electroproduction of vector mesons @xmath0(770 ) , @xmath1(1020 ) and @xmath2(3096 ) at collision energies @xmath3 and transferred momenta squared @xmath4 are considered in the framework of a phenomenological regge - eikonal scheme with nonlinear regge trajectories in which their qcd asymptotic behavior is taken into account explicitly . by comparison of available experimental data from zeus and h1 collaborations with the model predictions it is demonstrated that corresponding angular distributions and integrated cross sections in the above - mentioned kinematical range can be quantitatively described with use of two @xmath5-even vacuum regge trajectories .
these are the `` soft '' pomeron dominating the high - energy reactions without a hard scale and the `` hard '' pomeron giving an essential contribution to photo- and electroproduction of heavy vector mesons and deeply virtual electroproduction of light vector mesons .
-1.0 cm |
nuclear physics has always been plagued with difficulties and ambiguities related to off - shellness of two - body forces and the inclusion of many - body forces deduced from the existence of equivalent hamiltonians @xcite . in a remarkable paper glckle and polyzou @xcite point out that both problems are actually intertwined since `` different off - shell extensions of two - body forces can be equivalently realized as three - body interactions '' and while `` there are no experiments measuring only three - body binding energies and phase shifts that can determine if there are no three - body forces in a three - body system '' it is still likely that `` there may be some systems for which it is possible to find a representation in which three - body forces are not needed '' .
however , triton and @xmath1-particle _ ab initio _ calculations ( see e.g. @xcite for a review ) using 2-body high quality interactions , i.e. constrained to fit nn scattering data with @xmath2 display exceedingly simple regularities , such as the tjon line , a linear correlation between the binding energies , @xmath3 within a relatively wide range of energies ( see fig . [
fig : tjon ] ) .
this suggests the onset of some scale invariance @xcite and the understanding of this correlation should also provide a credible value for the slope of the tjon line .
actually , there is much freedom in making unitary transformations keeping the two body bound and continuum spectrum while generating a wide range of three- and four - body properties .
a practical way to generate these transformations is by means of the similarity renormalization group ( srg ) proposed by gazek and wilson @xcite and independently by wegner @xcite .
the srg was first used within nuclear physics by bogner , furnstahl and perry @xcite .
we discuss how the above mentioned regularities could be understood within a srg context by analysing its infrared fixed points as we discussed previously @xcite .
the proposed scenario is free of ambiguites but indicates a predominant role played by 3-body forces even in triton and alpha nuclei .
the general srg equation corresponds to a one - parameter operator evolution dynamics given by @xcite , @xmath4,h_s ] \label{eq : srg}\end{aligned}\ ] ] and supplemented with an initial condition at @xmath5 , @xmath6 . with the exception of some few cases @xcite
these equations are mostly solved numerically @xcite , by implementing a high momentum uv cut - off , @xmath7 , and an infrared momentum cut - off @xmath8 , which reduces the analysis to the finite @xmath9-dimensional case with @xmath10 .
the isospectrality of the srg becomes evident from the trace invariance @xmath11 .
fixed points of eq .
( [ eq : srg ] ) are given by stationary solutions , @xmath12,h_s]=0 $ ] requiring @xmath13 = f(h_s)$ ] .
thus there exists a basis where both @xmath14 and @xmath15 can be simultaneously block - diagonal for different energy subspaces with dimension equal the degeneracy .
the question is what choices of @xmath14 actually drive the solution to this block - diagonal form
. we will assume the usual separation @xmath16 . for generators which have the property @xmath17 , and using cyclic properties of the trace and the invariance of @xmath18 one gets @xmath19)^\dagger ( i[g_s , h_s ] ) \le 0 \ , .
\end{aligned}\ ] ] because @xmath20 is positive but its derivative is negative the limit for @xmath21 exists and corresponds to the fixed points , i.e. any starting @xmath6 gets indeed diagonalized by the srg equations ( [ eq : srg ] ) .
thus , the srg equation are just a continuous way of diagonalizing @xmath6 .
the key point is that there are many ways to diagonalize a hamiltonian .
for an n - dimensional space there are @xmath22 possible permutations regarding the final _ ordering _ of states .
there are two complementary generator choices : @xmath23 by wegner @xcite and by gazeck and wilson @xcite @xmath24 .
more generally @xmath25 @xcite , describes a different trajectory but identical fixed points .
a choice in between is the block - diagonal generator @xmath26 in two complementary subspaces with @xmath27 @xcite . denoting by @xmath28 the hamiltonian and @xmath29 the kinetic energy eigenvalues ,
the stability analysis @xcite can be extended to show that ( @xmath30 is some suitable matrix encoding the initial @xmath6 ) @xmath31 therefore , while _ all _ @xmath22 fixed points are stable in the wegner case , in the wilson case there is a unique fixed point where the hamiltonian preserves the original ordering of states according to the kinetic energy since @xmath32 .
the emerging crossing structure will be discussed in more detail in ref .
@xcite .
the previous considerations answer the question of _ how _ similar are two given hamiltonians in terms of the frobenius scalar product , norm and induced metric , latexmath:[\ ] ] which can be interpreted in terms of the number of pairs in the @xmath45 and @xmath57 states being @xmath60 for triton and @xmath61 for @xmath51he . within this shell model scheme @xmath62 . for @xmath63
the deuteron is unbound by @xmath64 , and the binding energy obtained for the triton is @xmath65 to be compared with the exact faddeev equation result @xmath66 whereas the @xmath1-particle yields @xmath67 .
it is interesting to analyze the srg evolution from these @xmath63 values down to @xmath68 _ without _ 3-body or 4-body forces , i.e. taking @xmath48 .
one can show @xcite that for the variational shell model @xmath69 which is checked by the numerical calculation ( see fig .
[ fig : binding ] ) .
we also find linear correlations in two regimes @xmath70 if we now switch on the srg induced 3-body and 4-body forces , the difference of the on - shell result to the original value corresponds to the off - shellnes of the @xmath63 two body potential , @xmath71 neglecting @xmath72 and taking @xmath73 corresponding to 4 triplets in the @xmath1-particle ( see also @xcite ) we get @xmath74 which gives @xmath75 .
of course , one may say the @xmath50h = pnn contain 2 deuterons and @xmath51he = ppnn contains 4 deuterons , so that @xmath76 . in general , we may write @xmath77 and @xmath78 and hence @xmath79 most calculations in nuclear physics use a mean field reference state ( often ho shell model ) upon which correlations are built .
for instance , in ref . @xcite it is found that the number of _ correlated pairs _ is given by @xmath80,@xmath81 , @xmath82 and @xmath83 with @xmath84 which , from our eq .
( [ eq : pair - triplet ] ) , requires @xmath85 and then @xmath86 .
these tjon lines are plotted in fig .
[ fig : tjon ] using the different choices in eq .
( [ eq : pair - triplet ] ) and compared with several accurate calculations using realistic 2-body forces ( without 3- or 4-body forces ) @xcite . in our interpretation , the slope of the tjon line is just the number of _ correlated triplets _ in @xmath51he , @xmath87 for _ on - shell _ interactions .
these intriguing results suggest that @xmath88 and will be analysed in more detail elsewhere @xcite .
srg methods allow to reduce the two - body off - shell ambiguity _
completely _ when the infrared limit is taken and thus only _ measurable _ two - body information is needed .
the same observation applies to multi - body interactions , and while current calculations fix the three body force from the triton binding energy , more work is required to pin down how much three - body measurable input would actually be demanded within such an on - shell scheme .
however , using these ideas a simple explanation of the observed linear tjon correlations between the @xmath1-particle and triton binding energies emerges from purely combinatorial arguments . in this scheme
the srg evolved 3-body forces are large in the infrared limit whereas 4-body forces remain moderate .
s. szpigel , v.s .
timoteo , f.d.o .
duraes : similarity renormalization group evolution of chiral effective nucleon - nucleon potentials in the subtracted kernel method approach .
annals phys . *
326 * , 364 ( 2010 ) e. d. jurgenson , p. maris , r. j. furnstahl , p. navratil , w. e. ormand and j. p. vary : p - shell nuclei using similarity renormalization group evolved three - nucleon interactions , phys .
c * 87 * , 054312 ( 2013 ) k. a. wendt : similarity renormalization group evolution of three - nucleon forces in a hyperspherical momentum representation .
c * 87 * , 061001 ( 2013 ) | the similarity renormalization group reduces the off - shellness by driving the evolved interaction towards a diagonal band . we analyze the infrared limit and the corresponding on - shell interactions and its consequences for light nuclei . using a harmonic oscillator shell model
we obtain a tjon line @xmath0 which can be understood from a combinatorics counting of nucleon pairs and triplets in the triton and @xmath1-particle and compares favorably with realistic calculations . |
the following is a list of parts used to build the cover slip cavity laser .
1 . sanyo laser diode : 670 nm , 7 mw , dl3149 - 057 , $ 14.14 , and socket s7060 , $ 3.98 , thorlabs 2 .
collimation tube with lens , lt230p - b , $ 108.00 , and spannor wrench , spw301 , $ 13.00 , thorlabs 3 .
kinematic mount , kms , $ 33.50 , thorlabs 4 .
piezoelectric elements ( 2 ) , ae0203d04 , $ 72.00 , thorlabs 5
. piezoelectric controller , mdt694a , $ 680.00 , thorlabs 6 .
function generator , gfg8020h , $ 193.50 , mouser electronics 7 .
sorbathane pad , sb12b , $ 49.00 , thorlabs 8 .
power supply ( 2 ) , 680-wm113-d5 , $ 42.20 , mouser electronics 9 .
temperature sensor , ad590 , $ 12.80 , thorlabs 10 .
thermocouple probe , tjc36cass020k6smpm , $ 30.00 , omega inc .
11 . thermocouple controller , hh74k , $ 55.00 , omega inc .
additional stock electronics components that were used to construct the laser ( op amps , breadboards , bud boxes , potentiometers ) , the microscope cover slip , and laboratory supplies such as expoxy all totaling $ 120 , are not itemized .
+ the lithium heat - pipe oven was constructed from the following parts .
1 . optical viewports ( 2 ) , fvg0150 , $ 95.26 , varian inc .
2 . 1.5 `` x 16 '' stainless steel tube w/ cf flanges and nupro inc .
ss 4bg tsw valve , $ 180.00 , custom welded 3 .
heater wire , aerocoax 1hn080b-3.9 , $ 3.50/ft , ari industries inc .
high temp cement , ob600 , $ 18.00 , omega engineering inc .
variac voltage supply tdgc2 - 0.5 , $ 50.00 , circuit specialists inc .
fibrafax thermal insulation , 93435k26 , $ 24.75/ft , mcmaster - carr 7 . thermocouple probe , tjc36cass020k6smpm , $ 30.00 , omega engineering inc
. thermocouple controller , hh74k , $ 55.00 , omega inc
li pellets ( 50 g ) , alfa aesar # 41829 , $ 81.70 , sigma - aldrich chemical the li heat pipe oven operates with the central region near 400 and tube ends ( and viewports ) at 80 .
heater wire was wound around the central 5 of the 16 long stainless steel tube .
the central 8 were then covered in 2 thick fibrafax thermal insulation .
the high temperature cement was used to bond the heater wire to standard hookup wire outside the thermal insulation surrounding the oven .
mounts for the heat pipe oven were machined out of stock aluminum .
five grams of li were loaded in the ss tube , and the assembly was then baked under 50 mt argon buffer gas at 200 for 2 hours .
the oven tube was then sealed with the valve and operated at 400 for over a year .
the ad590 temperature sensor produces @xmath17 . in the circuit shown in figure [ temp_circuit ] ,
the transimpedance gain and offset makes @xmath18 equal to 0 v at 20 and 10 v at 80 .
the difference between the adjustable @xmath19 and @xmath18 makes @xmath20 .
the error signal @xmath20 is used with feedback to regulate @xmath21 .
the correction signal , @xmath21 is then used to regulate the current supplied to the heating elements on the laser block ( shown within dotted line box in figure [ temp_circuit ] ) .
temperature stabilization circuits with similar function are found in @xcite .
resistor and 2 transistors serve as heating elements mounted to the aluminum mounting block .
the dashed box indicates which elements are physically mounted onto the aluminum mounting block .
the buffer circuit is duplicated four times and utilized where the double - arrows indicate connections to test - points where voltages can be monitored.,width=432 ] the piezo elements were controlled with the piezo gain and offset circuit shown in figure [ piezo_circuit ] .
this simple circuit was put in line before the high voltage amplifier ( piezoelectric controller , mdt694a , thorlabs ) .
these steps are a guide for tuning the cover slip cavity laser frequency to an atomic resonance . 1 .
adjust the laser wavelength to within 0.1 nm of 671 nm by modifying the mounting block temperature while using a grating spectrometer to monitor the wavelength .
2 . ramp the laser current using a 100 hz triangle function .
make the ramp extend from threshold to the maximum safe current ( 45 ma in the case of the sanyo model dl3149 - 057 ) .
3 . tip and tilt the cover slip by using the kinematic mount to provide optical feedback to the laser cavity .
do this first by getting the dim secondary laser beam ( caused from the back reflection from the cover slip ) to overlap with the primary ( brighter ) laser beam .
while observing the transmitted power through a low - finesse fabry - perot etalon ( the windows of the heat - pipe oven can serve this purpose ) , verify that the piezo - voltage offset and gain can modify the frequencies at which mode - hops occur .
re - align the cover slip using control of the laser modes as an indication of good coupling back into the laser cavity .
5 . observe the transmission through the heat - pipe oven with the laser current driven by the 100 hz ramp . slowly ramp the piezo voltage ( with the potentiometer ) to deliberately hop between different modes , and scan the temperature in order to see any hint of an atomic absorption ( or fluorescence ) signal .
once any atomic absorption ( or fluorescence ) has been observed , then obtain a larger mode - hop - free range by following these steps : ( 1 ) adjust the piezo - voltage gain to suppress mode hops , and ( 2 ) adjust the piezo - voltage offset to select the laser mode resonant with the atoms . 7 . for the largest mode - hop - free tuning range it is often required to iterate steps 1 and 2 and fine tune the laser temperature and the cover slip alignment .
this work was supported with nsf grant no .
we thank dr .
w. bickel and dr .
j. jones for diagnostic equipment , k. guerin for assistance with mechanical drawings , and m. parker of rincon research inc . for optics components .
the sanyo dl3149 - 057 , the mitsubishi ml101-j8 and the missubishi ml101-j27 each demonstrated resonant absorption by li vapor .
the sanyo 4039 - 11 and hitachi models hl6720 , hl6714 g , and hl6724 mg were also tested but did not demonstrate resonant absorption by li vapor due to unsufficient wavelength tuning range .
at least 3 units of each model were tested over a range of temperature from 0 to 80 .
the sanyo dl3149 - 057 gave the largest mode - hop - free tuning range , so results from this model are discussed here . | the design of a 671 nm diode laser with a mode - hop - free tuning range of 40 ghz is described .
this long tuning range is achieved by simultaneously ramping the external cavity length with the laser injection current .
the external cavity consists of a microscope cover slip mounted on piezoelectric actuators .
in such a configuration the laser output pointing remains fixed , independent of its frequency . using a diode with an output power of 5 - 7 mw
, the laser linewidth was found to be smaller than 30 mhz .
this cover slip cavity and feedforward laser current control system is simple , economical , robust , and easy to use for spectroscopy , as we demonstrate with lithium vapor and lithium atom beam experiments .
external cavities are often used to control the frequency of a laser .
early investigations with a plane mirror reflecting light back into a laser ( the design in figure 1 ) showed that optical feedback can pull " the laser frequency and influence which laser - cavity mode is active @xcite . instead of a mirror a reflective grating
can be used , as in the popular littrow @xcite or littman - metcalf @xcite configurations with diode lasers . however , scanning the cavity alone offers a limited mode - hop - free tuning range with red diode lasers .
this was reported by @xcite , and confirmed in our laboratory .
we found that 2 ghz was the maximum single - mode tuning range we could achieve when using only optical feedback to control the frequency of a 671 nm diode laser .
we tried seven different diode laser models @xcite , and found the same limited tuning range with either a grating or a mirror forming the external cavity .
adjusting only the diode laser current also resulted in a limited ( 2 - 3 ghz ) mode - hop - free tuning range .
this scan range is limited by the overlap of the modes for the laser diode and external cavity . a substantial increase in the tuning range
was caused by modulating _ both _ the external cavity and the diode laser current together .
here we describe a diode laser system in which the laser current and the external cavity length are simultaneously ramped to achieve a 40 ghz mode - hop - free tuning range operating near the lithium atom 2s-2p transition wavelength of 671 nm .
a microscope cover slip mounted on a piezoelectric transducer serves as the external cavity and output coupler , hence we refer to this as a _
cover slip cavity _
@xcite . surprisingly , the modest reflectivity of the glass cover slip surface with no coating ( r @xmath0% ) is sufficient to force single - mode operation while the laser frequency scans continuously over 40 ghz . since either end of this mode - hop - free tuning range
can be extended by adding ( or subtracting ) an offset to the cavity length , this mode - hop - free tuning range appears to be limited by non - linearity in the required diode laser current or cavity length adjustments .
it was found empirically that the laser current affects the frequency with the slope for the sanyo model dl3149 - 057 diode laser .
the cavity length affects the frequency with the slope .
this is consistent with the relationship @xmath1 for our cavity ( length @xmath210 mm ) and red light ( @xmath3=447 thz ) .
therefore , as the cavity length increases by 22 nm , the laser current should be increased by 250 @xmath4a to anticipate the -1 ghz shift in laser frequency .
this `` feedforward '' procedure increases the tuning range of the cover slip cavity laser . as a feedforward system
, there is no feedback loop connecting the laser current and cavity length adjustments ; they are just commensurately scanned in a predictable fashion using analog electronics .
the temperature of the laser mounting block was also stabilized to within 0.1 because temperature was seen to affect the laser frequency with the slope @xmath5
= -0.03/ghz ( i.e. a wavelength shift of 1 nm per 20 ) .
the laser current is modulated by an input voltage @xmath6 that can be provided by a function generator .
a resistor in series with the laser keeps the current linearly related to the input voltage [ @xmath7 to first order above the diode knee ] .
the same input voltage is also used to control the cavity length . to accomplish this ,
@xmath6 is added to an offset ( @xmath8 ) , amplified with a gain factor @xmath9 , and then directly applied to the piezoelectric actuators to adjust the cover slip cavity length .
hence the cavity length [ @xmath10 is also linearly related to the input voltage .
the length - to - voltage multiplier ( @xmath11=4 @xmath4m/150 v ) of the piezoelectric actuators @xcite makes the required piezo - voltage to laser current slope -3.3v / ma for mode - hop - free laser scanning as illustrated in figure 2 . ) .
the same voltage is shifted ( by adding @xmath8 ) then amplified by gain ( g ) and applied to the piezoelectric actuator ( p ) .
not shown : laser protection circuitry .
( right ) the predicted linear relation between the laser current and the cavity actuator voltage and the resulting shift in laser frequency .
( the laser frequency is centered on 447,000 ghz for the wavelength 671 nm ) .
nonlinearities in the laser current as a function of @xmath6 , nonlinearities in the piezo - voltage as a function of @xmath6 or nonlinearities in the relationship of these quantities to laser frequency limit the mode - hop - free scan range of the system.,title="fig:",width=408 ] ) .
the same voltage is shifted ( by adding @xmath8 ) then amplified by gain ( g ) and applied to the piezoelectric actuator ( p ) .
not shown : laser protection circuitry .
( right ) the predicted linear relation between the laser current and the cavity actuator voltage and the resulting shift in laser frequency .
( the laser frequency is centered on 447,000 ghz for the wavelength 671 nm ) .
nonlinearities in the laser current as a function of @xmath6 , nonlinearities in the piezo - voltage as a function of @xmath6 or nonlinearities in the relationship of these quantities to laser frequency limit the mode - hop - free scan range of the system.,title="fig:",width=216 ] ramping the laser current simultaneously with the cavity length is not a novel idea .
this was also used to enhance the tuning range for the various diode laser systems which utilized an external cavity grating @xcite . however , our cover slip cavity with feedforward current control has the advantages of simplicity , economy , and stability .
we discuss these advantages after explaining the mechanical construction .
mechanical views of the cover slip cavity , collimating optics , and temperature - stabilized laser mount are shown in figure 3 .
the physical layout has two principle parts : a mounting block , and a kinematic mount .
the laser diode is mounted in a 0.6-diameter tube with collimating optics @xcite and the tube is clamped in a 2x1x1 aluminum block pictured in figure [ layout ] . a temperature sensor ( ad590 ) and heaters ( a 5 ohm resistor and 2 transistors )
are also mounted on the block .
the stock 1-square kinematic mount @xcite is disassembled and a 0.7-square hole is milled in one plate ( as shown in figure 3 ) .
a 0.3-diameter hole is drilled in the other plate to transmit the output beam .
two piezoelectric actuators are mounted on either side of this output hole , and a 0.5 diameter glass cover slip is attached to the actuators with epoxy .
the cover slip is thus positioned within the kinematic mount so that it can form a short external cavity .
the kinematic mount is attached to the aluminum block with epoxy so that the cover slip and output hole are centered on the laser beam .
the mount enables coarse alignment of the cover slip after the initial construction . during feedforward operation , the system s only moving parts are the cover slip and the piezoelectric actuators
. the cavity length can be coarsely adjusted ( from 10 mm to 20 mm ) by relocating the collimating tube , and with care the kinematic mount can be taken apart ( to replace the cover slip ) and reassembled while the plate with the 0.7 hole remains attached to the mounting block .
the block and cavity assembly is mounted on a standard optics post and positioned in a thermally stable enclosure ( with a hole for the output beam ) .
to reduce vibrations , the entire system is mounted on a 1 kg brass plate and placed on top of sorbathane pads .
one advantage of the cover - slip cavity is simplicity .
if a grating were used for the external cavity there are two mechanical design criteria that must be considered because both the cavity length as well as the grating angle affect the laser frequency . in comparison ,
our cover slip cavity has only one important dimension : cavity length .
the laser beam always points in the same direction with the cover slip cavity ( whereas some grating cavities shift the beam direction as the frequency is scanned ) .
the cover slip transmits @xmath12 of the laser beam so it offers a high output efficiency .
we avoided anti - reflection coating the diode ( as was done in @xcite ) , we avoided removing the diode can ( which has glass surfaces ) and we avoided coating either side of the cover slip ( either to increase or decrease reflectivity ) in order to keep the design simple .
the entire laser system was constructed from parts that cost $ 1500 , and a lithium heat - pipe vapor cell was constructed for a cost of $ 800 .
the parts are listed in appendix i. if the temperature of the mounting block drifts by even 0.1 then the offset voltage @xmath8 needs to be adjusted , but this is the only intervention needed to keep the system scanning across the lithium atom resonance features for days at a time .
this stability and long tuning range makes the red light particularly useful for instructional laboratories and atomic physics research .
we operated individual sanyo dl3149 - 057 lasers for over 6 months with the current ramping in the range 35 to 45 ma .
the laser is typically scanned by applying a triangle waveform to @xmath6 that varies from 8 to 10 volts and cycles at 100 hz .
for some applications we also locked the laser frequency to a saturated absorption feature by connecting @xmath6 to the output of a locking circuit .
a number of experiments were done to test the performance of the cover slip cavity system .
lithium atom absorption spectra demonstrating a 40-ghz mode - hop - free frequency scan are shown in figure 4 . the vapor cell absorption features ( fig 4 upper left ) each have a 3 ghz doppler width due to thermal motion of the lithium atoms . in this plot ,
data are shown ( lower left ) for two vapor cell temperatures with correspondingly different optical depths .
the flat - bottomed absorption features with the higher optical depth allowed us to measure the amount of off - mode light in the laser output beam .
this was seen to be 7 @xmath13 2 % .
the laser intensity is modulated from 5 to 7 mw in accord with the laser current , but the laser beam pointing remains stable during the scan .
a fabry - perot transmission spectrum is shown in figure 4 ( lower left ) confirming the scanning range of 40 ghz .
saturated absorption spectra shown in figure 4 ( upper right ) were obtained using the two - probe method described in @xcite .
also shown is an atom beam deflection signal ( bottom right ) obtained by shining the cover slip cavity laser beam at a highly collimated atom beam .
( see @xcite for a description of this technique ) .
both types of doppler - free spectra show the large 803 mhz hyperfine splitting due to the f=1,2 levels of the ground @xmath14 state of @xmath15li .
the atom beam deflection spectrum also shows a smaller but clearly resolved splitting due to the f=1,2 levels in the excited @xmath16 state of @xmath15li .
radians , the deflection spectrum has essentially zero doppler width .
power broadening , transit time broadening , and the laser linewidth all contribute to the four observed 55-mhz ( fwhm ) dips due to deflection .
in addition to the 803 mhz spacing due to the ground state hyperfine splitting , the atom beam deflection signal shows the 92 mhz hyperfine splitting due to the @xmath16 excited state of @xmath15li.,title="fig:",width=336 ] radians , the deflection spectrum has essentially zero doppler width .
power broadening , transit time broadening , and the laser linewidth all contribute to the four observed 55-mhz ( fwhm ) dips due to deflection . in addition to the 803 mhz spacing due to the ground state hyperfine splitting , the atom beam deflection signal shows the 92 mhz hyperfine splitting due to the @xmath16 excited state of @xmath15li.,title="fig:",width=336 ] radians , the deflection spectrum has essentially zero doppler width .
power broadening , transit time broadening , and the laser linewidth all contribute to the four observed 55-mhz ( fwhm ) dips due to deflection . in addition to the 803 mhz spacing due to the ground state hyperfine splitting , the atom beam deflection signal shows the 92 mhz hyperfine splitting due to the @xmath16 excited state of @xmath15li.,title="fig:",width=336 ] radians , the deflection spectrum has essentially zero doppler width .
power broadening , transit time broadening , and the laser linewidth all contribute to the four observed 55-mhz ( fwhm ) dips due to deflection .
in addition to the 803 mhz spacing due to the ground state hyperfine splitting , the atom beam deflection signal shows the 92 mhz hyperfine splitting due to the @xmath16 excited state of @xmath15li.,title="fig:",width=336 ] the laser line - width was smaller than 30 mhz , as found with a fabry - perot interferometer .
this upper limit on the linewidth is consistent with the atomic deflection spectra and the saturated absorption spectra shown in figure [ 4fig ] . in additional tests we modulated the laser current at 800 mhz ( to add sidebands to the laser spectrum ) as described in @xcite
this did not reduce the tuning range .
we also used this laser to study faraday rotation in vapor @xcite , the hanle effect in vapor @xcite , and optical pumping of an atom beam .
we thus demonstrated that the cover slip cavity diode laser serves for several atomic physics experiments . in conclusion ,
we have described a diode laser system that uses a cover slip cavity and feedforward operation to achieve a mode - hop - free scanning range of 40 ghz .
the frequency can be scanned over the lithium atom 2s-2p resonances near 671 nm , with the entire system being simple , economical , stable and robust . |
the standard model ( sm ) are known to have three families of charged and neutral fermions .
there is however no upper limit on the number of fermion families .
a fourth family ( sm4 ) @xcite could be in fact the key to many unsolved puzzles , such as the hierarchies of the fermion mass spectrum including neutrino masses and mixing , electroweak symmetry breaking , baryogenesis , and a variety of interesting phenomena in cp and flavor physics @xcite .
fourth family leptons and quarks have been searched for previously by the lep and tevatron experiments and now by the lhc experiments .
the most stringent lower mass limits at 95% cl are @xcite @xmath2 the mass bound on the heavy neutrino depends on the type of neutrino ( dirac or majorana ) and whether one considers a coupling of the heavy neutrino to @xmath3 or @xmath4 .
it should also be noted that assumptions about the coupling of the fourth family and the decay mode were made in deriving the quark mass limits .
the limits can be weaker without these assumptions .
on the other hand , the triviality bound from unitarity of the @xmath5 @xmath6-wave scattering @xcite indicates a maximum @xmath7 mass of around 500gev , although this estimate is based on tree - level expressions . with the presence of fourth family quarks , the dominant higgs production process , the gluon - gluon fusion process ,
is further enhanced ( see sec .
[ sec : sigma ] and also @xcite ) . using the large enhancement of @xmath8 in sm4 , cdf , d0 and cms have been able to exclude the sm4 higgs boson at 95% cl for @xmath9gev @xcite and @xmath10gev @xcite . in @xcite
( see also @xcite ) , it was pointed out that the mass limit on @xmath11 ( eq.([eq : nu4_limit ] ) ) was derived when the mixing angle between the fourth family neutrino and at least one of the neutrinos in the sm is assumed to be larger than @xmath12 . for smaller mixing angles , @xmath13 is quasistable and the mass of @xmath13 is bounded only from the analysis of @xmath14 boson decay at @xmath15gev @xcite . in this case , the decay mode @xmath16 becomes the dominant one at low higgs mass and the tevatron lower mass limits of 131gev would be increased to 155gev @xcite . in the following
, we will not consider this possibility .
in the sm , the dominant higgs production process is the gluon - gluon fusion process where gluons from colliding beams couple to a heavy quark loop from which the higgs boson is emitted .
the cross section at the leading - order ( lo ) can be written as @xcite @xmath17 with @xmath18 ^ 2|i|^2\,,\hspace{5 mm } i=\sum_q i_q\ ] ] and @xmath19 is the gluon distribution evaluated at @xmath20 and @xmath21 , @xmath22 ( @xmath23 ) is the fermi ( strong ) coupling constant .
the quantity @xmath24 is given in terms of @xmath25 : @xmath26\,,\ ] ] where @xmath27 with @xmath28 . in the heavy quark @xmath29 limit , @xmath30 , @xmath31 , whereas in the light quark @xmath29 limit @xmath32 , @xmath33 .
this is why in the sm , the top quark is by far the dominant contribution . in a fourth generation with two additional heavier quark @xmath7 and @xmath34 ,
the higgs production cross section is enhanced with that of the sm by a factor @xmath35 the dependence as a function of the higgs boson mass @xmath36 is shown in fig .
[ fig : rsigma ] for two different @xmath37 values of infinite mass and 400gev .
the @xmath7 mass is fixed as @xcite @xmath38}\right)\,,\ ] ] to be consistent with the constraint of electroweak precision data . the maximum enhancement factor of about 9 is reached at the small higgs boson mass @xmath36 value where @xmath39 and @xmath40 . for two scenarios with @xmath41 and @xmath42 . ] at low @xmath36 , the enhancement is independent of the quark mass value of the fourth generation . at higher higgs mass values , the heavier the quark mass , the smaller the enhancement factor .
the heavy @xmath43 scenario may not be physical as when it is beyond about @xmath44gev the weak interaction among heavy particles becomes strong and perturbation theory breaks down .
however , since the enhancement is the smallest , the resulting exclusion limits would be more conservative .
this is the scenario used by cms in their recent publication @xcite .
the other scenario with @xmath45gev corresponds to one of the scenarios used by the tevatron experiments @xcite . in fig .
[ fig : rsigma ] , the enhancement based on lo cross sections is also compared with the corresponding factor in nlo calculated with a modified higlu program @xcite .
the cross sections @xmath46 are calculated for @xmath47 collisions at 7tev center - of - mass energy has essentially no dependence on it . ] .
as far as the ratio @xmath48 is concerned , the difference between nlo and lo is small .
this means that one may use this ratio ( e.g. in nlo ) and precisely predicted sm cross section values in higher orders to derive the corresponding higher - order cross section in a fourth generation model : @xmath49 indeed , use this relation and take the nnlo cross section value @xmath50 pb at @xmath51gev as an example ( table 1 in @xcite ) and the corresponding enhancement factor of 9.223 ( see the linked web page below ) , we obtain 182.79pb for the low mass scenario .
this derived cross section value @xmath52 is in excellent agreement with the independent prediction of 182.51pb given in table 1 in @xcite .
the sm higgs decay branching fractions calculated using hdecay @xcite is displayed in fig . [
fig : br ] .
the decay mode @xmath53 is the reverse of the gluon - gluon fusion process @xmath54 .
the partial decay width of the decay mode @xmath53 in a fourth generation model with respect to that of the sm is thus enhanced by the same factor as the corresponding higgs production cross section in the gluon - gluon fusion process ( eq.([eq : rsigma ] ) ) .
the decay mode @xmath55 is similar to @xmath53 except that charged leptons , the @xmath56-boson and charged higgs bosons also contribute to the loop .
the partial decay width is @xcite @xmath57 where @xmath58 is the fine - structure coupling constant .
the quark , lepton , @xmath56-boson and colorless charged scalar contributions are @xcite : @xmath59 where @xmath60 denotes the charge of fermion @xmath61 in units of @xmath62 and @xmath63\,,\\ & & i_l=2\lambda_l+\lambda_l(4\lambda_l-1)f(\lambda_l)\,,\\ & & i_w=3\lambda_w(1 - 2\lambda_w)f(\lambda_w)-3\lambda_w-\frac{1}{2}\,,\\ & & i_s=-\lambda_s\left[1 + 2\lambda_s f(\lambda_s)\right]\,,\end{aligned}\ ] ] and @xmath64 . in a fourth generation model , both @xmath65 and @xmath66 terms receive additional contributions from fourth generation quarks @xmath67 and lepton @xmath68 . finally the partial decay width @xmath69 is also affected by additional fourth generation quark loops . therefore the branching fractions ( fig .
[ fig : br_sm4 ] ) in a fourth generation model look different from that of the sm in particular for @xmath70 , @xmath71 and @xmath1 modes at low higgs mass values .
tev ( left ) and @xmath72gev ( right).,title="fig : " ] tev ( left ) and @xmath72gev ( right).,title="fig : " ] in fig . [ fig : br_sm4 ] , the two mass scenarios are compared .
the @xmath68 mass is further assumed to be the same as that of @xmath34 .
but as far as these fourth generation quark and lepton masses are heavy enough ( @xmath73 ) , the difference in branching fractions is hardly visible .
the branching fractions in a fourth generation model are calculated without applying the electroweak and higher - order qcd corrections as they do not apply to the fourth generation quarks .
for the relevant decay modes @xmath71 and @xmath74 , the ratio of the branching fractions in a fourth generation model over that in the sm @xmath75 is compared in the two mass scenarios in fig .
[ fig : rbr](left ) , which shows that the effect of the mass scenarios is indeed small .
the effect of the electroweak and higher - order qcd corrections on the branching fraction ratio is also small as it is illustrated in fig .
[ fig : rbr](right ) .
the overall enhancement of the product of the higgs production cross section and the higgs branching fraction in a fourth family over that in the sm @xmath76_{\rm sm4}}{\left[\sigma(gg\rightarrow h)\times b(h\rightarrow x)\right]_{\rm sm}}\ ] ] is shown in fig .
[ fig : xsbr ] as a function of the higgs mass for decay modes @xmath77 , @xmath78 and @xmath79 and for the two mass scenarios . the numerical values for @xmath48 , @xmath80 and @xmath81 for 59 higgs mass points ranging from 110gev to 600gev are given in linked web pages @xcite .
one advantage of reporting @xmath48 instead of @xmath82 that we mentioned in sec .
[ sec : sigma ] is that the ratio is less sensitive to higher order corrections .
it is also less sensitive to other theoretical uncertainties .
one example is the dependence on the choice of parton distribution functions ( pdfs ) . in fig .
[ fig : pdf ] , the variation of choosing two different pdfs ( mstw2008nlo @xcite and herapdf1.0 @xcite ) has been compared with the default choice of cteq6 m @xcite . in the considered higgs mass range ,
the difference on @xmath48 is well within 0.2% . shown for the high mass quark scenario as a function of the higgs mass @xmath36 . ]
we have discussed the implication in the higgs production cross sections in the gluon - gluon fusion process and the higgs decay branching fractions in modes @xmath70 , @xmath71 and @xmath1 when including quarks and leptons of a fourth generation model .
the enhancement / modification on the gluon - gluon fusion cross section ( @xmath48 ) and the higgs decay branching fractions ( @xmath83 ) of a fourth generation model over the sm as a function of the higgs mass @xmath36 in the range of @xmath84gev has been calculated and shown for two mass scenarios .
these ratios are found to have little sensitive to theoretical variations such as the higher order corrections for both @xmath48 and @xmath83 and pdfs for @xmath48 .
the authors wish to thank m. spira for help and advice in using the higlu and hdecay codes .
z. zhang is also grateful to f. richard and e. kou for discussions .
t. aaltonen _ et al .
_ , ( the cdf collaboration ) , _
search for heavy bottom - like quarks decaying to an electron or muon and jets in @xmath85 collisions at @xmath86tev _ , phys .
106 ( 2011 ) 141803 [ arxiv:1101.5728 [ hep - ex ] ] .
kribs , t. plehn , m. spannowsky and t.m.p .
tait , _ four generations and higgs physics _ , phys .
d76 ( 2007 ) 075016 [ arxiv:0706.3718 [ hep - ph ] ] .
n. becerici schmidt , s.a .
etin , s. istin and s. sultansoy , _ the fourth standard model family and the competition in standard model higgs boson search at tevatron and lhc _ , eur .
j. c66 ( 2010 ) 119 [ arxiv:0908.2653 [ hep - ph ] ] .
t. aaltonen _ et al . _ , ( the cdf and d0 collaborations ) , _ combined tevatron upper limit on @xmath88 and constraints on the higgs boson mass in fourth - generation fermion models _ , ( 2010 ) [ arxiv:1005.3216 [ hep - ex ] ] . s. chatrchyan _ et al . _ ( the cms collaboration ) , _ measurement of @xmath89 production and search for the higgs boson in @xmath47 collisions at @xmath90tev _ , phys . lett . b699 ( 2011 ) 25 [ arxiv:1102.5429 [ hep - ex ] ] .
aaron _ et al .
_ , ( the h1 and zeus collaborations ) , _ combined measurement and qcd analysis of the inclusive @xmath91 scattering cross sections at hera _ , jhep 1001 ( 2010 ) 109 [ arxiv:0911.0884 [ hep - ex ] ] . | in a fourth generation model with heavy quarks , the production cross section of the higgs boson in the gluon - gluon fusion process is significantly increased due to additional quark loops . in a similar way ,
the partial decay width of the decay channels @xmath0 and @xmath1 is modified . these changes and their impact on the higgs search
are discussed .
lal 11 - 109 fourth generation , higgs production , higgs decays |
the study of the intersection theory on the moduli space of riemann surfaces with boundary ( often viewed , with the boundary removed , as open riemann surfaces ) was recently initiated in @xcite .
the authors constructed a descendent theory in genus @xmath0 and obtained a complete description of it . in all genera , they conjectured that the generating series of the descendent integrals satisfies the open kdv equations .
this conjecture can be considered as an open analog of the famous witten s conjecture @xcite .
a higher genus construction of the open descendent theory will appear in @xcite .
a combinatorial formula for the open intersection numbers in all genera was found in @xcite . in this paper ,
using the combinatorial formula from @xcite , we present a matrix integral for the generating series of the open intersection numbers . then applying some analytical tools to this matrix integral ,
we prove the main conjecture from @xcite .
the introduction is organized as follows . in section [ wittc ]
we briefly recall the original conjecture of e. witten from @xcite . in section [ konts ]
we recall kontsevich s combinatorial formula and kontsevich s proof ( @xcite ) of witten s conjecture .
section [ okdv ] contains a short account of the main constructions and conjectures in the open intersection theory from @xcite .
section [ ocomb ] describes the combinatorial formula of @xcite for the open intersection numbers of @xcite . throughout this text
@xmath1 $ ] will denote the set @xmath2 a compact riemann surface is a compact connected smooth complex curve . given a fixed genus @xmath3 and a non - negative integer @xmath4 the moduli space of all compact riemann surfaces of genus @xmath3 with @xmath5 marked points
is denoted by @xmath6 p. deligne and d. mumford defined a natural compactification of it via stable curves in @xcite in 1969 . given @xmath7 as above , a stable curve is a compact connected complex curve with @xmath5 marked points and finitely many singularities , all of which are simple nodes .
the collection of marked points and nodes is the set of special points of the curve .
we require that all the special points are distinct and that the automorphism group of the curve is finite .
the moduli of stable marked curves of genus @xmath3 with @xmath5 marked points is denoted by @xmath8 and is a compactification of @xmath6 it is known that this space is a non - singular complex orbifold of complex dimension @xmath9 for the basic theory the reader is referred to @xcite . in his seminal paper @xcite , e. witten , motivated by theories of @xmath10-dimensional quantum gravity , initiated new directions in the study of @xmath8 . for each marking index @xmath11 he considered the tautological line bundles @xmath12 whose fiber over a point @xmath13\in { { \overline{\mathcal{m}}}}_{g , l}\ ] ] is the complex cotangent space @xmath14 of @xmath15 at @xmath16 .
let @xmath17 denote the first chern class of @xmath18 , and write @xmath19 the integral on the right - hand side of is well - defined , when the stability condition @xmath20 is satisfied , all the @xmath21 s are non - negative integers , and the dimension constraint @xmath22 holds . in all other cases
@xmath23 is defined to be zero .
the intersection products are often called _ descendent integrals _ or _ intersection numbers_. note that the genus is uniquely determined by the exponents @xmath24 .
let @xmath25 ( for @xmath26 ) and @xmath27 be formal variables , and put @xmath28 let @xmath29 be the generating function of the genus @xmath3 descendent integrals .
the bracket @xmath30 is defined by the monomial expansion and the multilinearity in the variables @xmath25 .
concretely , @xmath31 where the sum is over all sequences of non - negative integers @xmath32 with finitely many non - zero terms .
the generating series @xmath33 is called the _ ( closed ) free energy_. the exponent @xmath34 is called the _ ( closed ) partition function_. put @xmath35 . witten s conjecture ( @xcite ) says that the closed partition function @xmath36 becomes a tau - function of the kdv hierarchy after the change of variables @xmath37 .
in particular , it implies that the closed free energy @xmath38 satisfies the following system of partial differential equations : @xmath39 these equations are known in mathematical physics as the kdv equations .
e. witten ( @xcite ) proved that the intersection numbers satisfy the string equation @xmath40 for @xmath41 .
this equation can be rewritten as the following differential equation : @xmath42 e. witten also showed that the kdv equations together with the string equation actually determine the closed free energy @xmath38 completely .
there was a later reformulation of witten s conjecture due to r. dijkgraaf , e. verlinde and h. verlinde ( @xcite ) in terms of the virasoro algebra .
define differential operators @xmath43 , @xmath44 , by @xmath45 while for @xmath46 @xmath47 the virasoro equations say that the operators @xmath43 , @xmath48 , annihilate the closed partition function @xmath36 : @xmath49 it is easy to see that the virasoro equations completely determine all intersection numbers .
r. dijkgraaf , e. verlinde and h. verlinde ( @xcite ) proved that this description is equivalent to the one given by the kdv equations and the string equation .
witten s conjecture was proven by m. kontsevich @xcite .
see @xcite for other proofs .
kontsevich s proof @xcite of witten s conjecture consisted of two parts .
the first part was to prove a combinatorial formula for the gravitational descendents .
let @xmath50 be the set of isomorphism classes of trivalent ribbon graphs of genus @xmath3 with @xmath51 faces and together with a numbering @xmath52 $ ] .
denote by @xmath53 the set of vertices of a graph @xmath54 .
let us introduce formal variables @xmath55 , @xmath56 $ ] .
for an edge @xmath57 let @xmath58 where @xmath11 and @xmath59 are the numbers of faces adjacent to @xmath60 .
then we have @xmath61 the second step of kontsevich s proof was to translate the combinatorial formula into a matrix integral .
then , by using non - trivial analytical tools and the theory of tau - functions of the kdv hierarchy , he was able to prove that @xmath36 is a tau - function of the kdv hierarchy and , hence , the free energy @xmath38 satisfies the kdv equations . in @xcite r. pandharipande , j. solomon and r.t .
constructed an intersection theory on the moduli space of stable marked disks .
let @xmath62 be the moduli space of stable marked disks with @xmath63 boundary marked points and @xmath5 internal marked points .
this space carries a natural structure of a compact smooth oriented manifold with corners .
one can easily define the tautological line bundles @xmath64 for @xmath65 $ ] , as in the closed case . in order to define gravitational descendents , as in ,
we must specify boundary conditions . indeed , given a smooth compact connected oriented orbifold with boundary , @xmath66 of dimension @xmath51
, the poincar - lefschetz duality shows that @xmath67 thus , given a vector bundle on a manifold with boundary , only _ relative euler class _ , relative to nowhere vanishing boundary conditions , can be integrated to give a number .
the main construction in @xcite is a construction of boundary conditions for @xmath68 in @xcite , vector spaces @xmath69 of _ multisections _ of @xmath70 which satisfy the following requirements , were defined .
suppose @xmath71 are non - negative integers with @xmath72 then 1 .
[ it : canonic1 ] for any generic choice of multisections @xmath73 for @xmath74 the multisection @xmath75\\1\le
j\le a_i } } s_{ij}\ ] ] vanishes nowhere on @xmath76 2 .
[ it : canonic2 ] for any two such choices @xmath77 and @xmath78 we have @xmath79 where @xmath80 and @xmath81 is the relative euler class .
the multisections @xmath82 , as above , are called _
canonical_. with this construction the open gravitational descendents in genus @xmath0 are defined by @xmath83 where @xmath84 is as above and @xmath77 is canonical .
in a forthcoming paper @xcite , j. solomon and r.t . define a generalization for all genera . in @xcite a moduli space @xmath85 which classifies real stable curves with some extra structure
is constructed .
the moduli space @xmath85 is a smooth oriented compact orbifold with corners , of real dimension @xmath86 the stability condition is @xmath87 note that naively , without adding an extra structure , the moduli of real stable curves of positive genus is non - orientable . on
@xmath85 one defines vector spaces @xmath88 , for @xmath65,$ ] for which the genus @xmath3 analogs of requirements [ it : canonic1],[it : canonic2 ] from above hold .
write @xmath89 for the corresponding higher genus descendents .
introduce one more formal variable @xmath77 . the _ open free energy _ is the generating function @xmath90 where @xmath91 , @xmath92 , and again we use the monomial expansion and the multilinearity in the variables @xmath93 the following initial condition follows easily from the definitions ( @xcite ) : @xmath94 in @xcite the authors conjectured the following equations : @xmath95 they were called the open string and the open dilaton equation correspondingly .
put @xmath96 the main conjectures in @xcite are [ conjecture : open kdv ] the following system of equations is satisfied : @xmath97 in @xcite equations were called the open kdv equations .
it is easy to see that @xmath98 is fully determined by the open kdv equations , the initial condition and the closed free energy @xmath38 .
let @xmath99 be the _
open partition function_. in @xcite the authors introduced the following modified operators : @xmath100 where the operators @xmath43 were defined in section [ subsubsection : closed virasoro ] .
[ conjecture : open virasoro ] the operators @xmath101 , @xmath44 , annihilate the open partition function : @xmath102 in @xcite equations were called the open virasoro equations .
again it is easy to see that the open free energy @xmath98 is fully determined by the open virasoro equations , the initial condition @xmath103 and the closed free energy @xmath38 . from the closed string equation
it immediately follows that the open string equation is equivalent to , for @xmath104 .
moreover , from the equation @xmath105 it follows that the open dilaton equation @xmath106 is equivalent to , for @xmath107 .
more precisely , in @xcite it was conjectured that there exists a definition of open intersection numbers for @xmath108 , for which the open kdv and open virasoro equations hold .
the definition was later given in @xcite .
although it was not clear at all that the open kdv and the open virasoro equations are compatible , in @xcite it was proved that they indeed have a common solution .
for @xmath109 the conjectures were proved in @xcite . in @xcite
the conjectures are proved for @xmath110 and the open string and the open dilaton equations are proved for all @xmath111 the main result of this paper is the following theorem .
[ theorem : main ] conjectures [ conjecture : open kdv ] and [ conjecture : open virasoro ] are true .
let @xmath112 be a power series in the variables @xmath113 with the coefficients from @xmath114 $ ] . in @xcite the following system of equations
was introduced : @xmath115 it was called the half of the burgers - kdv hierarchy .
this system is obviously stronger than the system of the open kdv equations . in
@xcite it was actually shown that the half of the burgers - kdv hierarchy is equivalent to the open kdv equations together with equation .
denote by @xmath116 a unique solution of system - specified by the initial condition @xmath117 in @xcite it was shown that @xmath116 satisfies the open kdv equations , the initial condition and the open virasoro equations .
this proved the equivalence of the open analog of witten s conjecture and the open virasoro conjecture .
this also shows that theorem [ theorem : main ] immediately implies the following corollary .
the open free energy @xmath98 satisfies the half of the burgers - kdv hierarchy .
consider more variables @xmath118 and let @xmath119 .
let @xmath112 be a power series in the variables @xmath120 with the coefficients from @xmath114 $ ] .
let us extend the half of the burgers - kdv hierarchy by the following equations : @xmath121 in @xcite the extended system - was called the ( full ) burgers - kdv hierarchy .
let @xmath122 be a unique solution of it specified by the initial condition @xmath123 we obviously have @xmath124 . in
@xcite it was proved that the function @xmath125 satisfies the following extended virasoro equations : @xmath126 where @xmath127 here we , by definition , put @xmath128 . in @xcite it was conjectured that , by adding descendents for boundary marked points , one can geometrically define intersection numbers which will be the coefficients of @xmath122 . in @xcite j. solomon an r.t . give a complete proposal for the construction in all genera , and
it is proved to produce the correct intersection numbers in genus @xmath129 moreover , with this proposal the extended open free energy @xmath130 is defined , as well as the extended open partition function @xmath131 it is proved in @xcite that the following equations hold @xmath132 in @xcite , section 5.2 , it is shown that @xmath133 satisfies these equations as well .
thus , theorem [ theorem : main ] implies that @xmath134 and we immediately obtain the following generalization of theorem [ theorem : main ] .
[ theorem : extended main ] the extended open free energy @xmath130 is a solution of the full burgers - kdv hierarchy . from the recent result of a. alexandrov @xcite it also follows that the extended open partition function @xmath135 becomes a tau - function of the kp hierarchy after the change of variables @xmath37 and @xmath136 . in @xcite r.t .
proved a combinatorial formula for the geometric models which were defined in @xcite .
he showed that all these intersection numbers can be calculated as sums of amplitudes of diagrams which will be described below . in this paper a matrix model is constructed out of this combinatorial formula .
using this matrix model we prove our main theorem [ theorem : main ] .
a topological @xmath137-surface with boundary @xmath15 , is a topological connected oriented surface with non - empty boundary , @xmath63 boundary marked points @xmath138},$ ] and @xmath5 internal marked points @xmath139},$ ] and genus @xmath111 by genus we mean , as usual in the open theory , the doubled genus , that is , the genus of the doubled surface obtained by gluing two copies of @xmath15 along @xmath140 we require the stability condition @xmath141 let @xmath142 be non - negative integers such that @xmath143 be a finite set and @xmath144\to a$ ] a map .
@xmath145 will be implicit in the definition .
a _ @xmath137-ribbon graph with boundary _ is an embedding @xmath146 of a connected graph @xmath147 into a @xmath137-surface with boundary @xmath15 such that * @xmath148}\subseteq \iota(v(g))$ ] , where @xmath53 is the set of vertices of @xmath147 .
we henceforth consider @xmath149 as vertices .
* the degree of any vertex @xmath150 is at least @xmath151 .
* @xmath152 . * if @xmath153 then @xmath154 } d_i,\ ] ] where each @xmath155 is a topological open disk , with @xmath156 .
we call the disks @xmath155 faces . * if @xmath157 , then @xmath158 .
the genus @xmath159 of the graph @xmath147 is the genus of @xmath15 .
the number of the boundary components of @xmath147 or @xmath15 is denoted by @xmath160 and @xmath161 stands for the number of the internal vertices .
we denote by @xmath162 the set of faces of the graph @xmath163 and we consider @xmath164 as a map @xmath165 by defining for @xmath166 where @xmath16 is the unique internal marked point in @xmath167 the map @xmath164 is called the labeling of @xmath168 denote by @xmath169 the set of boundary marked points @xmath148}.$ ] two ribbon graphs with boundary @xmath170 are isomorphic , if there is an orientation preserving homeomorphism @xmath171 and an isomorphism of graphs @xmath172 , such that 1 . @xmath173 2 .
@xmath174 for all @xmath175.$ ] 3 .
@xmath176 where @xmath177 are the labelings of @xmath178 respectively and @xmath179 is any face of the graph @xmath168 note that in this definition we do not require the map @xmath180 to preserve the numbering of the internal marked points .
a ribbon graph is _ critical _ , if * boundary marked points have degree @xmath10 . *
all other vertices have degree @xmath151 . *
if @xmath181 then @xmath109 and @xmath182 such a component is called _ a ghost_. a _ nodal ribbon graph with boundary _ is @xmath183 , where * @xmath184 are ribbon graphs with boundary . * @xmath185 is a set of _ ordered _ pairs of boundary marked points @xmath186 of the @xmath187 s which we identify . after the identification of the vertices @xmath188 and @xmath189 the corresponding point in the graph is called a node .
the vertex @xmath188 is called the legal side of the node and the vertex @xmath189 is called the illegal side of the node .
* we also require the graph @xmath147 to be connected .
we require that elements of @xmath190 are disjoint as sets ( without ordering ) .
the set of edges @xmath191 is composed of the internal edges of the @xmath187 s and of the boundary edges .
the boundary edges are the boundary segments between successive vertices which are not the illegal sides of nodes . for any boundary edge @xmath60
we denote by @xmath192 the number of the illegal sides of nodes lying on it .
the boundary marked points of @xmath147 are the boundary marked points of @xmath187 s , which are not nodes .
the set of boundary marked points of @xmath147 will be denoted by @xmath169 also in the nodal case .
a nodal graph @xmath183 is _ critical _ , if * all of its components @xmath187 are critical . *
any boundary component of @xmath187 has an odd number of points that are the boundary marked points or the legal sides of nodes .
* ghost components do not contain the illegal sides of nodes . a nodal ribbon graph with boundary is naturally embedded into the nodal surface @xmath193 .
the ( doubled ) genus of @xmath15 is called the genus of the graph .
the notion of an isomorphism is also as in the non - nodal case . in figure [ nodal ]
there is a nodal graph of genus @xmath0 , with @xmath194 boundary marked points , @xmath195 internal marked points , three components , one of them is a ghost , two nodes , where a plus sign is drawn next to the legal side of a node and a minus sign is drawn next to the illegal side .
denote by @xmath196 the set of isomorphism classes of critical nodal ribbon graphs with boundary of genus @xmath3 , with @xmath63 boundary marked points , @xmath5 faces and together with a bijective labeling @xmath197,$ ] the combinatorial formula in @xcite is fix @xmath198 such that @xmath199 .
let @xmath200 be formal variables .
then we have @xmath201 where @xmath202 we thank a. alexandrov , l. chekhov , d. kazhdan , o. g. louidor , a. okounkov , r. pandharipande , p. rossi , s. shadrin , j. p. solomon and d. zvonkine for discussions related to the work presented here .
a. b. was supported by grant erc-2012-adg-320368-mcsk in the group of r. pandharipande at eth zurich , the grants rffi 13 - 01 - 00755 and nsh-4850.2012.1 .
r.t . was supported by isf grant 1747/13 and erc starting grant 337560 in the group of j. p. solomon at the hebrew university of jerusalem .
part of the work was completed during the visit of a.b to the einstein institute of mathematics of the hebrew university of jerusalem in 2014 and during the visits of r.t . to the forschungsinstitut fr mathematik at eth zrich in 2013 and 2014 .
in this section we present a matrix integral that is a starting point in our proof of theorem [ theorem : main ] . instead of deriving a matrix integral for the open partition function @xmath203 directly from the combinatorial formula ,
we write a matrix model for an auxiliary function @xmath204 that is a sum over non - nodal ribbon graphs with boundary with some additional structure .
we then relate the function @xmath204 to the open partition function @xmath203 by an action of the exponent of some quadratic differential operator .
the section is organized as follows . in section [
subsection : reformulation ] we give a slight reformulation of the combinatorial formula . in section [
subsection : sum over non - nodal graphs ] we introduce an auxiliary function @xmath204 and relate it to the open partition function @xmath203 . this relation is given by lemma [ lemma : relation ] .
section [ subsection : brief recall ] contains a brief review of basic facts about the integration over the space of hermitian matrices . in section [
subsection : matrix integral ] we give a matrix integral for the function @xmath204 .
this is the subject of proposition [ proposition : matrix integral ] .
the matrix integral in this proposition is understood in the sense of formal matrix integration . in section [
subsection : convergent matrix integral ] we discuss how to make sense of it as a convergent integral .
we fix an integer @xmath205 throughout this section and we set the genus parameter @xmath27 to be equal to @xmath206 here we reformulate the combinatorial formula .
this step is completely analogous to what m. kontsevich did in @xcite ( see the proof of theorem 1.1 there ) .
denote by @xmath207 the set of isomorphism classes of critical nodal ribbon graphs with boundary together with a labeling @xmath208 $ ] ( a coloring of faces in @xmath190 colors ) . for a graph @xmath209 and an edge @xmath210 ,
let @xmath211 introduce formal variables @xmath212 and consider the diagonal @xmath213 matrix @xmath214 from the combinatorial formula it follows that @xmath215 where @xmath216 and @xmath217 here @xmath218 stands for the number of the internal edges of the graph @xmath187 .
here we introduce an auxiliary function @xmath204 and relate it to the open partition function @xmath203 .
we denote by @xmath219 the ring of formal series of the form @xmath220 where @xmath221 is a rational function in @xmath212 homogeneous of degree @xmath222 .
we denote by @xmath223 the subspace that consists of series of the form @xmath224 , where @xmath221 is a rational function in @xmath212 homogeneous of degree @xmath222 .
note that the ring @xmath225 $ ] can be naturally considered as a subring of @xmath219 .
let us introduce the following auxiliary set of graphs .
denote by @xmath226 the set of isomorphism classes of critical non - nodal ribbon graphs with boundary , that are not ghosts , together with * a labeling @xmath227 $ ] , * a map @xmath228 $ ] ; such that on each boundary component of @xmath147 the number of the boundary marked points @xmath229 with @xmath230 is odd .
vertices @xmath229 with @xmath230 will be called legal and vertices @xmath229 with @xmath231 will be called illegal .
the boundary edges of @xmath147 are , by definition , the boundary segments between successive vertices of @xmath147 which are not the illegal boundary marked points .
we will use the following notations : @xmath232 let @xmath233 introduce an auxiliary formal variable @xmath234 and define the following series of rational functions : @xmath235 where @xmath236 by definition , @xmath204 is an element of @xmath237 $ ] .
we do not know if @xmath204 belongs to @xmath225[[s , s_-]]$ ] or not .
several computations in low degrees motivate us to conjecture that @xmath238[[s , s_-]]$ ] .
moreover , we conjecture that there exists a power series @xmath239[[s , s_-]]$ ] such that for any @xmath205 we have @xmath240 however , we do not need this statement in the paper .
let @xmath241 [ lemma : relation ] we have @xmath242 the lemma easily follows from equation , the definition of @xmath204 and the standard result about the graphical representation of the action of the exponent of a quadratic differential operator .
we recommend the book ( * ? ? ?
* sections 3 , 4 ) as a good introduction to this subject .
denote by @xmath243 the space of hermitian @xmath213 matrices .
for @xmath244 let @xmath245 and @xmath246 .
introduce a volume form @xmath247 by @xmath248 consider positive real numbers @xmath249 .
recall that @xmath250 . introduce a gaussian measure @xmath251 on the space @xmath243 by @xmath252 where the normalization @xmath253 is determined by the constraint @xmath254 for any polynomial @xmath255_{1\le i , j\le n}$ ] let @xmath256 the integrals @xmath257 are described by the following result ( see e.g. ( * ? ? ?
* section 3.2.3 ) ) .
[ lemma : wick formula ] _
1_. if @xmath258 is homogeneous of odd degree , then @xmath259 . + _ 2_. for any indices @xmath260 , we have @xmath261 where the sum is taken over all permutations @xmath262 of the set of indices @xmath263 such that @xmath264 and @xmath265 . + _ 3_. we have @xmath266 in this section we present a matrix integral representation for the function @xmath204 .
let @xmath267.\end{aligned}\ ] ] note that @xmath268 therefore , the exponent @xmath269 can be represented in the following way @xmath270 [ proposition : matrix integral ] we have @xmath271 here the integral is understood in the sense of formal matrix integration .
it means the following .
we express @xmath272 as a series of the form @xmath273 where @xmath274 is a polynomial of degree @xmath275 in @xmath276 and expressions of the form @xmath277 here the degree is introduced by putting @xmath278 from lemma [ lemma : wick formula ] it follows that @xmath279 is zero , if @xmath275 is odd , and is a rational function in @xmath212 of degree @xmath280 , if @xmath275 is even .
then the integral on the right - hand side of is defined by @xmath281 we begin by recalling the derivation of the matrix integral for @xmath282 from kontsevich s combinatorial formula ( @xcite ) .
consider an arbitrary trivalent ribbon graph without boundary with a coloring of faces in @xmath190 colors .
it can be obtained by gluing trivalent stars ( see fig . [
fig : internal vertex ] ) in such a way that corresponding indices on glued edges coincide .
( 0.5,0 )
( 0.5,-2.5 ) ; ( -0.5,0 )
( -0.5,-2.5 ) ; ( 0.5,0 )
( 2.67,1.25 ) ; ( 0,0.87 )
( 2.17,2.12 ) ; ( -0.5,0 ) ( -2.67,1.25 ) ; ( 0,0.87 ) ( -2.17,2.12 ) ; ( 0.5,-2.5 ) ( -0.5,-2.5 ) ; ( 2.67,1.25 )
( 2.17,2.12 ) ; ( -2.67,1.25 ) ( -2.17,2.12 ) ; ( 4.5,0 )
( 6,0 ) ; ( 4.5,0.2 )
( 4.5,-0.2 ) ; ( 6,0 )
( 5.7,0.2 ) ; ( 6,0 )
( 5.7,-0.2 ) ; \(b ) at ( 0.5,-1.25 ) ; ( b ) at ( -0.5,-1.25 ) ; ( b ) at ( 1.5,0.75 ) ; ( b ) at ( 1.30,1.45 ) ; ( b ) at ( -1.5,0.75 ) ; ( b ) at ( -1.3,1.45 ) ; ( b ) at ( 8.5,0 ) ; ( 0,0.5 ) ( 2.5,0.5 ) ; ( 0,-0.5 )
( 2.5,-0.5 ) ; ( 0,0.5 ) ( -1.25,2.67 ) ; ( -0.87,0 ) ( -2.12,2.17 ) ; ( 0,-0.5 ) ( -1.25,-2.67 ) ; ( -0.87,0 ) ( -2.12,-2.17 ) ; ( 2.5,0.5 )
( 2.5,-0.5 ) ; ( -1.25,2.67 ) ( -2.12,2.17 ) ; ( -1.25,-2.67 ) ( -2.12,-2.17 ) ; \(b ) at ( 1.25,0.5 ) ; ( b ) at ( 1.25,-0.5 ) ; ( 0,0.5 ) ( -2.5,0.5 ) ; ( 0,-0.5 ) ( -2.5,-0.5 ) ; ( 0,0.5 )
( 1.25,2.67 ) ; ( 0.87,0 )
( 2.12,2.17 ) ; ( 0,-0.5 )
( 1.25,-2.67 ) ; ( 0.87,0 )
( 2.12,-2.17 ) ; ( -2.5,0.5 ) ( -2.5,-0.5 ) ; ( 1.25,2.67 )
( 2.12,2.17 ) ; ( 1.25,-2.67 ) ( 2.12,-2.17 ) ; \(b ) at ( -1.25,0.5 ) ; ( b ) at ( -1.25,-0.5 ) ; to a trivalent star we associate the polynomial @xmath283 $ ] . then the wick formula ( lemma [ lemma : wick formula ] ) and kontsevich s formula
imply that @xmath284 this is the famous kontsevich s integral ( @xcite ) .
we recommend the reader the book ( * ? ? ?
* sections 3 and 4 ) for a more detailed explanation of this techique .
consider now a non - nodal critical ribbon graph with boundary @xmath285 .
each boundary component is a circle with a configuration of vertices of degrees @xmath10 and @xmath151 .
vertices of degree @xmath10 are boundary marked points and they are of two types : legal and illegal . in fig .
[ pic2 ] we draw an example of a boundary component .
legal boundary marked points are marked by @xmath286 and illegal ones by @xmath287 .
each graph from @xmath226 can be obtained by gluing trivalent stars and some number of boundary components .
( 2.5,0.5 ) arc ( 11.31:78.69:2.55 ) ; ( 2.5,0.5 )
( 5,0.5 ) ; ( 2.5,-0.5 ) arc ( -11.31:-78.69:2.55 ) ; ( 2.5,-0.5 ) ( 5,-0.5 ) ; ( 2.5,-0.5 ) ( 2.5,0.5 ) ; ( -2.5,0.5 ) arc ( 168.69:101.31:2.55 ) ; ( -2.5,0.5 )
( -5,0.5 ) ; ( -2.5,-0.5 ) arc ( -168.69:-101.31:2.55 ) ; ( -2.5,-0.5 ) ( -5,-0.5 ) ; ( -2.5,-0.5 ) ( -2.5,0.5 ) ; ( 0.5,2.5 )
( 0.5,5 ) ; ( -0.5,2.5 ) ( -0.5,5 ) ; ( -0.5,2.5 ) ( 0.5,2.5 ) ; ( 0.5,-2.5 )
( 0.5,-5 ) ; ( -0.5,-2.5 ) ( -0.5,-5 ) ; ( -0.5,-2.5 ) ( 0.5,-2.5 ) ; ( 0.5,5 ) ( -0.5,5 ) ; ( -5,0.5 ) ( -5,-0.5 ) ; ( 0.5,-5 ) ( -0.5,-5 ) ; ( 5,0.5 ) ( 5,-0.5 ) ; \(b ) at ( 2.21,1.27 ) ; ( b ) at ( 2.21,1.27 ) ; ( b ) at ( 1.27,2.21 ) ; ( b ) at ( 1.27,2.21 ) ; ( b ) at ( 1.80,-1.80 ) ; ( b ) at ( 1.80,-1.80 ) ; ( b ) at ( -1.27,-2.21 ) ; ( b ) at ( -1.27,-2.21 ) ; ( b ) at ( -2.21,-1.27 ) ; ( b ) at ( -2.21,-1.27 ) ; ( 6,0 ) ( 7.5,0 ) ; ( 6,0.2 )
( 6,-0.2 ) ; ( 7.2,0.2)(7.5,0)(7.2,-0.2)cycle ; \(b ) at ( 8,0 ) ; consider a boundary component of some non - nodal critical ribbon graph with boundary @xmath288 .
recall that , by definition , the boundary edges are the boundary segments between successive vertices which are not illegal boundary marked points .
let @xmath289 be the edges of the boundary component , ordered in the clockwise direction .
moreover we orient each edge in the clockwise direction , so that any boundary edge points from a source vertex to a target vertex .
let @xmath290 the combinatorial formula suggests that to the boundary component we should assign the following expression : @xmath291\in{\mathbb c}[h_{i , j}],\end{gathered}\ ] ] where @xmath292 denotes the automorphism group of the boundary component .
an example is shown in fig .
[ pic2 ] . for a moment
we ignore the combinatorial coefficient .
if we sum expressions over all possible choices of a boundary component , we get @xmath293.\end{gathered}\ ] ] let us deal more carefully with the combinatorial coefficient .
denote by @xmath294 the number of the boundary vertices of degree @xmath151 of the graph @xmath147 and by @xmath295 the number of the boundary edges .
since @xmath296 , we have @xmath297 therefore , we have to rescale expression in the following way : @xmath298.\end{gathered}\ ] ] we immediately recognize here the function @xmath299 .
again , the wick formula together with kontsevich s formula and the combinatorial formula imply that @xmath300 the proposition is proved .
one can show that the integral @xmath301 is absolutely convergent and determines a well - defined function of @xmath302 .
it is not hard to show that the asymptotic expansion of it , when @xmath303 , is given by @xmath282 ( @xcite ) . in our case
we can also make sense of the integral in as a convergent integral .
suppose that @xmath55 are positive real numbers , @xmath234 is a real number such that @xmath304 , @xmath305 , and @xmath77 is a purely imaginary complex number .
then the integral @xmath306 is absolutely convergent and determines a well - defined function of @xmath307 and @xmath234 .
in this section we introduce a certain version of the fourier transform .
it happens that after this transformation formula becomes much simpler .
the main result of this section is proposition [ proposition : convergent fourier transform ] .
we again fix an integer @xmath205 throughout this section and set the genus parameter @xmath27 to be equal to @xmath308 . here
we define our version of the fourier transform and describe its main properties .
section [ subsubsection : preliminary ] is preliminary . the main definition is contained in section [ subsubsection : first fourier transform ] . in section [ subsubsection : second fourier transform ] there is a slightly different version of it that will also be useful .
section [ subsubsection : properties ] is devoted to the properties of our fourier transform . for an arbitrary positive real @xmath310 and a non - negative integer @xmath311
, we have the following classical formula : @xmath312 a power series @xmath313 $ ] is called admissible if it has the form @xmath314 , where @xmath315 $ ] .
the space of admissible power series is denoted by @xmath309^{adm}\subset{\mathbb c}[[s]]$ ] . for
an admissible power series @xmath313^{adm}$ ] , @xmath316 , the formal fourier transform @xmath317(z)\in{\mathbb c}[[z^{-1},z]$ ] is defined by @xmath318(z):=\sqrt\frac{z}{2\pi}\int^{form}_{{\mathbb r}}{\widetilde f}(-is+z)e^{\frac{i}{6}s^3-{\frac{1}{2}}s^2 z}ds:=\\ = \sum_{k\ge 0}\sqrt\frac{z}{2\pi}\int_{{\mathbb r}}\frac{{\widetilde f}(-is+z)}{k!}\left(\frac{i}{6}s^3\right)^ke^{-{\frac{1}{2}}s^2 z}ds\in{\mathbb c}[[z^{-1},z].\end{gathered}\ ] ] the reader can see that the formal fourier transform of @xmath319 is the asymptotic expansion of the integral @xmath320 when @xmath321 goes to @xmath322 . an element @xmath324 $ ] will be called admissible , if the series @xmath325 , @xmath326 , satisfies the following property .
there exists a sequence @xmath327 , @xmath328 , of positive integers such that @xmath329 , when @xmath330 , and @xmath331 .
the space of all admissible elements will be denoted by @xmath323^{adm}$ ] . for an admissible element @xmath332^{adm}$ ] , it can be seen that the element @xmath333 can be expressed in the following way : @xmath334 where @xmath335 are homogeneous linearly independent rational functions and @xmath336 $ ] are polynomials in @xmath77 .
the formal fourier transform @xmath317(\lambda , z)\in{\mathcal r}_n[[z^{-1},z]]$ ] is defined by @xmath337(\lambda , z):=\sum_m g_m(\lambda)\sqrt{\frac{z}{2\pi}}\int^{form}_{\mathbb r}e^{\frac{is^3}{6}-\frac{1}{2}s^2z}p_m(-is+z)ds\in{\mathcal r}_n[[z^{-1},z]].\end{gathered}\ ] ] we again see that the formal fourier transform can be considered as the asymptotic expansion of the integral @xmath338 when @xmath321 goes to @xmath322 .
a formal fourier transform for power series from @xmath340[[s]]$ ] is introduced completely analogously .
we introduce a grading in the ring @xmath340 $ ] assigning to @xmath25 the degree @xmath341 . for @xmath342 let @xmath340^d$ ] be the subspace of @xmath340 $ ] that consists of power series of the form @xmath343 , where @xmath344 $ ] is a homogeneous polynomial of degree @xmath275 .
the subspace @xmath340[[s]]^{adm}\subset { \mathbb c}[[t_0,t_1,\ldots]][[s]]$ ] is defined similarly to the previous section , using the filtration @xmath345={\mathbb c}[[t_0,t_1,\ldots]]^0\supset{\mathbb c}[[t_0,t_1,\ldots]]^1\supset{\mathbb c}[[t_0,t_1,\ldots]]^2\supset\ldots.\ ] ] it is easy to see that if @xmath346[[s]]$ ] is admissible , then @xmath347\ ] ] is also admissible . for an admissible element @xmath346[[s]]^{adm}$ ]
the formal fourier transform is defined by @xmath337(t_*,z):=\sqrt{\frac{z}{2\pi}}\int^{form}_{\mathbb r}e^{\frac{is^3}{6}-\frac{1}{2}s^2z}{\widetilde f}(t_*,-is+z)ds\in{\mathbb c}[[t_0,t_1,\ldots]][[z^{-1},z]],\end{gathered}\ ] ] where @xmath348 .
the relation to the fourier transform from the previous section is given by @xmath349\right|_{t_i=-(2i-1)!!{\mathop{\mathrm{tr}}\nolimits}\lambda^{-2i-1}}=\phi^{form}_s\left[\left.f\right|_{t_i=-(2i-1)!!{\mathop{\mathrm{tr}}\nolimits}\lambda^{-2i-1}}\right].\end{gathered}\ ] ] let us describe three basic properties of our formal fourier transform .
the symbol @xmath350 will stand for @xmath219 or @xmath340 $ ] .
[ lemma : injectivity ] the map @xmath351^{adm}\to{\mathcal a}[[z^{-1},z]]$ ] is injective .
suppose that @xmath352 .
the proof for @xmath353 $ ] is the same .
suppose we have @xmath317=0 $ ] for some admissible @xmath332^{adm}$ ] .
let @xmath333 .
we have @xmath354 where @xmath355 are linearly independent homogeneous rational functions and @xmath336 $ ] are polynomials in @xmath77 . since @xmath317=0 $ ] , we get @xmath356 for all @xmath275 .
let us choose a non - zero @xmath357 and let @xmath358 , where @xmath359 .
a direct calculation shows that @xmath360 therefore this integral is non - zero . this contradiction proves the lemma .
[ lemma : fourier of derivative ] for any admissible element @xmath361^{adm}$ ] , the derivative @xmath362 is also admissible and @xmath363=\frac{z^2}{2}\phi^{form}_s[f]$ ] . the proof is again presented in the case @xmath352 . if @xmath316 , then @xmath364 . therefore , @xmath362 is also admissible .
now we compute @xmath365(\lambda , z)=\sqrt\frac{z}{2\pi}\int^{form}_{{\mathbb r}}e^{\frac{is^3}{6}-\frac{1}{2}s^2z}\left(\frac{(-is+z)^2}{2}{\widetilde f}(\lambda ,- is+z)+i\frac{{\partial}}{{\partial}s}{\widetilde f}(\lambda ,-
is+z)\right)ds=\\ = \frac{z^2}{2}\phi^{form}_s[f](\lambda , z).\end{gathered}\ ] ] the last equality follows from integration by parts .
the lemma is proved .
[ lemma : fourier and multiplication ] for any admissible element @xmath361^{adm}$ ] , the product @xmath366 is also admissible and @xmath367=-\left(\frac{1}{z}\frac{{\partial}}{{\partial}z}-\frac{1}{2z^2}-z\right)\phi^{form}_s[f].\ ] ] the admissibility of the product @xmath368 is obvious .
we compute @xmath369=\sqrt\frac{z}{2\pi}\frac{1}{z}\int^{form}_{\mathbb r}\left[\left(i\frac{{\partial}}{{\partial}s}-\frac{s^2}{2}\right){\widetilde f}(-is+z)\right]e^{\frac{i}{6}s^3-\frac{1}{2}s^2 z}ds=\\ = \sqrt\frac{z}{2\pi}\int^{form}_{\mathbb r}is{\widetilde f}(-is+z)e^{\frac{i}{6}s^3-\frac{1}{2}s^2 z}ds.\end{gathered}\ ] ] therefore , @xmath370=\sqrt\frac{z}{2\pi}\int^{form}_{\mathbb r}(is - z){\widetilde f}(-is+z)e^{\frac{i}{6}s^3-\frac{1}{2}s^2 z}ds=\phi_s^{form}[sf].\ ] ] the lemma is proved .
recall that @xmath371[[s]].\ ] ] from formula it is easy to see that the series @xmath372 is admissible .
it is also easy to see that @xmath373 , where @xmath374 .
thus , for any @xmath375 , the series @xmath376 is admissible . we conclude that the fourier transform @xmath377(\lambda , z , s_-)$ ] with respect to the variable @xmath77 is well - defined .
[ proposition : formal fourier transform ] we have @xmath378(\lambda , z)=\phi^{form}_s\left[e^{\frac{s^3}{6}}f^o_n\right]\left(\lambda , z,\frac{z^2}{4}\right).\ ] ] by lemma [ lemma : relation ] , we have @xmath379 from lemma [ lemma : fourier of derivative ] it follows that @xmath380(\lambda , z , s_-)=e^{\frac{z^2}{4}\frac{{\partial}}{{\partial}s_-}}\left(\phi^{form}_s\left[f^o_n e^{\frac{s^3}{6}}\right](\lambda , z , s_-)\right)=\\ = \phi^{form}_s\left[f^o_n e^{\frac{s^3}{6}}\right]\left(\lambda , z , s_-+\frac{z^2}{4}\right).\end{gathered}\ ] ] setting @xmath381 , we get the statement of the proposition
. propositions [ proposition : formal fourier transform ] and [ proposition : matrix integral ] together with section [ subsection : convergent matrix integral ] imply the following proposition .
[ proposition : convergent fourier transform ] the series @xmath382(\lambda , z)$ ] is the asymptotic expansion of the integral @xmath383 when @xmath384 and @xmath385 .
let @xmath190 be an arbitrary positive integer . recall that the series @xmath386 $ ] is defined by @xmath233
let @xmath387 .
we set the genus parameter @xmath27 to be equal to @xmath206 the following proposition is the key step in the proof of theorem [ theorem : main ] .
[ proposition : main ] we have @xmath388\left(\lambda , z\right)=\tau^c_{n+1}(\lambda_z)\sqrt{\det\left(\frac{1-z\lambda^{-1}}{1+z\lambda^{-1}}\right)}.\end{gathered}\ ] ] before proving the proposition let us make a remark about the right - hand side of equation . we see that @xmath389[[z^{-1}]]$ ] and @xmath390[[z]]$ ] .
so , on the right - hand side of we multiply a power series in @xmath391 and a power series in @xmath321 . in general , the multiplication of two such series may not be well - defined . in our case ,
the issue is resolved as follows .
we have @xmath392 therefore , @xmath393 has the form @xmath394 , where @xmath395 $ ] is homogeneous of degree @xmath222 .
thus , the product on the right - hand side of is well - defined .
our proof of proposition [ proposition : main ] uses a famous technique , that is sometimes called the averaging procedure over the unitary group .
we recall it in section [ subsection : averaging ] . after that , in section [ subsection : proof of the main proposition ] , we prove proposition [ proposition : main ] .
it is well - known that an arbitrary hermitian matrix @xmath396 admits a polar decomposition , @xmath397 , where @xmath398 is a unitary matrix and @xmath399 is a diagonal matrix with real entries .
denote by @xmath400 the group of unitary @xmath213 matrices . given a vector @xmath401 ,
define a map @xmath402 by @xmath403 for a hermitian matrix @xmath404 let @xmath405 be the vector of its eigenvalues .
it is defined up to a permutation of the coordinates . for a subset @xmath406 , that is invariant under permutations of the coordinates ,
let @xmath407 suppose moreover that @xmath406 is compact and measurable .
let @xmath408 be an arbitrary smooth function .
we denote by @xmath409 the haar probability measure on @xmath400 .
we have the following result ( see e.g. ( * ? ? ?
* sections 3 and 4 ) ) .
[ lemma : polar ] we have @xmath410 let @xmath411 be arbitrary real numbers such that @xmath412 and @xmath413 , for @xmath414 .
let @xmath415 and @xmath416 .
let @xmath417 be an arbitrary non - zero complex parameter .
the harish - chandra - itzykson - zuber ( hciz ) formula says that @xmath418 the right - hand has no poles on the diagonals @xmath419 or @xmath420 , and so it defines a smooth function of @xmath421 real parameters @xmath411 .
formula was originally found in @xcite and then was rediscovered in @xcite .
we will apply the hciz formula in the following way .
again , let @xmath422 be a compact measurable subset of @xmath423 invariant under permutations of the coordinates .
let @xmath408 be a smooth unitary invariant function : @xmath424 , for any @xmath425 .
finally , consider pairwise distinct real numbers @xmath426 and set @xmath415 .
[ lemma : averaging ] we have @xmath427 where @xmath428 and @xmath429 .
the lemma follows from combining lemma [ lemma : polar ] and the hciz formula .
suppose that @xmath212 and @xmath321 are positive real numbers .
consider kontsevich s integral @xmath430 as we already recalled in section [ subsection : convergent matrix integral ] , the integral is absolutely convergent and its asymptotic expansion , when @xmath431 , is given by the series @xmath432 ( see @xcite ) .
suppose moreover that @xmath433 .
using proposition [ proposition : convergent fourier transform ] we see that equation is a consequence of the following equation : @xmath434 the integral is still absolutely convergent , when @xmath55 s and @xmath321 are complex numbers with positive real parts .
moreover , the integral is semi - convergent , if @xmath435 are purely imaginary non - zero complex numbers , and the asymptotic expansion of it , when @xmath436 is still given by the power series @xmath432 .
see the discussion of these subtle questions in @xcite .
let @xmath55 and @xmath321 be purely imaginary complex numbers with positive imaginary parts .
assume moreover that @xmath437 where we choose a particular value of the square root , such that all entries of the matrix @xmath438 have positive imaginary parts .
let @xmath426 be the diagonal elements of @xmath438 .
let @xmath439 let us perform the change of variables @xmath440 on the left - hand side of equation .
we get @xmath441 where @xmath442 .
define @xmath443 to be equal to @xmath0 . by lemma [ lemma :
averaging ] , the last integral is equal to @xmath444 where @xmath445 and @xmath446 . note that @xmath447 where @xmath448 .
therefore , we can rewrite as follows : @xmath449,\end{gathered}\ ] ] where @xmath450 and @xmath451 . by lemma [ lemma : averaging ] ,
the expression in the square brackets is equal to @xmath452 redenoting @xmath453 by @xmath77 , expression is equal to @xmath454.\end{gathered}\ ] ] now we make the shift @xmath455 in the integral in the square brackets . using also that @xmath456 , we get that is equal to @xmath457.\end{gathered}\ ] ] finally , making the shift @xmath458 , we come to @xmath459 equation is proved , and , thus , also proposition [ proposition : main ] .
from the dimension constraint it follows that , if an open intersection number is non - zero , then the genus @xmath3 is uniquely determined by @xmath21 s and @xmath63 . using this observation , it is easy to show that , if the open kdv equations hold for @xmath460 , then they are also true for an arbitrary @xmath27 .
the same is true for the open virasoro equations .
therefore , without loss of generality , we can assume that @xmath460 .
let @xmath461 be the shift operator which acts on a series @xmath462 $ ] by @xmath463(t_0,t_1,\ldots):=f\left(t_0-\frac{k_0}{z},t_1-\frac{k_1}{z^3},t_2-\frac{k_2}{z^5},\ldots\right),\ ] ] where @xmath464 and , by definition , @xmath465 .
let @xmath466 .
proposition [ proposition : main ] implies the following corollary .
[ corollary : main ] we have @xmath467=g_z[\tau^c]e^{\xi^c}.\end{gathered}\ ] ] it is easy to check that for an arbitrary @xmath205 we have @xmath468\right|_{t_i=-(2i-1)!!{\mathop{\mathrm{tr}}\nolimits}\lambda^{-2i-1}}=\tau^c_{n+1}(\lambda_z)\quad\text{and}\quad \left.e^{\xi^c}\right|_{t_i=-(2i-1)!!{\mathop{\mathrm{tr}}\nolimits}\lambda^{-2i-1}}=\sqrt{\det\left(\frac{1-z\lambda^{-1}}{1+z\lambda^{-1}}.\right)}.\ ] ] therefore , by proposition [ proposition : main ] and equation , we have @xmath469\right)\right|_{t_i=-(2i-1)!!{\mathop{\mathrm{tr}}\nolimits}\lambda^{-2i-1}}=\left.\left(g_z[\tau^c]e^{\xi^c}\right)\right|_{t_i=-(2i-1)!!{\mathop{\mathrm{tr}}\nolimits}\lambda^{-2i-1}}.\end{gathered}\ ] ] for an arbitrary @xmath470 the functions @xmath471 become algebraically independent , when @xmath190 is sufficiently large .
therefore , equation implies that @xmath472=g_z[\tau^c]e^{\xi^c}$ ] . the corollary is proved .
let us prove that the open free energy @xmath98 satisfies the half of the burgers - kdv hierarchy - . in
@xcite it was obtained that the half of the burgers - kdv hierarchy can be written in a very convenient form using the lax formalism .
let us briefly recall this result .
a pseudo - differential operator @xmath438 is a laurent series @xmath473 where @xmath275 is an arbitrary integer and @xmath474 are formal power series in @xmath475 .
we will always identify @xmath476 with @xmath477 .
let @xmath478 the product of pseudo - differential operators is defined by the following commutation rule : @xmath479 where @xmath480 and @xmath346 $ ] . for any @xmath481 and a pseudo - differential operator @xmath438 of the form @xmath482 there exists a unique pseudo - differential operator @xmath483 of the form @xmath484 such that @xmath485 .
consider the pseudo - differential operator @xmath486 . in
@xcite it was proved that the half of the burgers - kdv system - is equivalent to the following system : @xmath487 let us prove that the open free energy @xmath98 satisfies equations and .
let @xmath488 from the fact that the closed partition function @xmath36 is a tau - function of the kdv hierarchy it follows that the series @xmath489 is the wave function of the kdv hierarchy ( see e.g. @xcite ) , that is , it satisfies the following equations : @xmath490 from corollary [ corollary : main ] it follows that @xmath491=\psi|_{s_*=0}.\end{gathered}\ ] ] let us prove .
let @xmath492 .
we have @xmath493=o_{t_n}\phi^{form}_s\left[e^{f^o}\right]\stackrel{\text{by~\eqref{eq : eq2}}}{=}o_{t_n}\left(\psi|_{s_*=0}\right)=\left(o_{t_n}\psi\right)|_{s_*=0}\stackrel{\text{by~\eqref{eq : first for the wave}}}{=}0.\ ] ] by lemma [ lemma : injectivity ] , we get @xmath494 and equation is proved .
let us prove .
let @xmath495 .
we have @xmath496\stackrel{\text{by lemma~\ref{lemma : fourier of derivative}}}{=}\left(\frac{z^2}{2}-\frac{1}{2}l\right)\phi^{form}_s\left[e^{f^o}\right]\stackrel{\text{by~\eqref{eq : eq2}}}{=}\left(\frac{z^2}{2}-\frac{1}{2}l\right)\left(\psi|_{s_*=0}\right)=\left(o_s\psi\right)|_{s_*=0}\stackrel{\text{by~\eqref{eq : second for the wave}}}{=}0.\ ] ] by lemma [ lemma : injectivity ] , we get @xmath497 and equation is proved . in
@xcite it was shown that the open kdv equations follow from the half of the burgers - kdv hierarchy . therefore , conjecture [ conjecture : open kdv ] is proved .
we also see that the open free energy @xmath98 satisfies the initial condition @xmath498 . in
@xcite it was proved that such a solution of the half of the burgers - kdv hierarchy satisfies the open virasoro equations .
thus , conjecture [ conjecture : open virasoro ] is also proved .
the derivation of the open virasoro equations from the open kdv equations was given in @xcite .
another derivation was obtained in @xcite . in this section
we want to show that the open virasoro equations can be very easily obtained from corollary [ corollary : main ] .
again we can assume that @xmath460 . by corollary [ corollary : main ] , @xmath499=g_z[\tau^c]e^{\xi^c}.\ ] ] from the closed virasoro equations it is easy to derive
that ( see e.g. @xcite ) @xmath500e^{\xi^c}\right)=\left(\frac{z^{2n+2}}{2^{n+1}}\left(\frac{1}{z}\frac{{\partial}}{{\partial}z}-\frac{1}{2z^2}-z\right)+\frac{n+1}{2^{n+2}}z^{2n}\right)\left(g_z[\tau^c]e^{\xi^c}\right).\ ] ] by lemmas [ lemma : fourier of derivative ] and [ lemma : fourier and multiplication ] , we have @xmath501&=\phi_s^{form}\left[\left(-\frac{{\partial}^{n+1}}{{\partial}s^{n+1}}\circ s+\frac{n+1}{4}\frac{{\partial}^n}{{\partial}s^n}\right)\tau^o\right]=\\ & \quad\quad=-\phi_s^{form}\left[\left(s\frac{{\partial}^{n+1}}{{\partial}s^{n+1}}+\frac{3n+3}{4}\frac{{\partial}^n}{{\partial}s^n}\right)\tau^o\right].\end{aligned}\ ] ] here , as above , the circle @xmath502 means the composition of operators . in the last operator inside the fourier
transform we immediately recognize the difference between the open virasoro operator @xmath503 and the closed virasoro operator @xmath43 ( see ) .
therefore , we get @xmath504=0.\ ] ] lemma [ lemma : injectivity ] implies that @xmath505 .
the open virasoro equations are proved .
s. k. lando , a. k. zvonkin .
_ graphs on surfaces and their applications_. with an appendix by don b. zagier .
encyclopaedia of mathematical sciences , 141 .
low - dimensional topology , ii .
springer - verlag , berlin , 2004 .
xvi+455 pp .
a. okounkov , r. pandharipande .
_ gromov - witten theory , hurwitz numbers , and matrix models_. algebraic geometry - seattle 2005 .
part 1 , 325 - 414 , proc .
pure math .
, 80 , part 1 , amer .
soc . , providence , ri , 2009 . | in a recent work , r. pandharipande , j. p. solomon and the second author have initiated a study of the intersection theory on the moduli space of riemann surfaces with boundary .
they conjectured that the generating series of the intersection numbers satisfies the open kdv equations . in this paper
we prove this conjecture .
our proof goes through a matrix model and is based on a kontsevich type combinatorial formula for the intersection numbers that was found by the second author . |
detection of sparse mixtures is an important problem that arises in many scientific applications such as signal processing @xcite , biostatistics @xcite , and astrophysics @xcite , where the goal is to determine the existence of a signal which only appears in a small fraction of the noisy data .
for example , topological defects and doppler effects manifest themselves as non - gaussian convolution component in the cosmic microwave background ( cmb ) temperature fluctuations .
detection of non - gaussian signatures are important to identify cosmological origins of many phenomena @xcite .
another example is disease surveillance where it is critical to discover an outbreak when the infected population is small @xcite .
the detection problem is of significant interest also because it is closely connected to a number of other important problems including estimation , screening , large - scale multiple testing , and classification .
see , for example , @xcite , @xcite , @xcite , @xcite , and @xcite .
one of the earliest work on sparse mixture detection dates back to dobrushin @xcite , who considered the following problem originating from multi - channel detection in radiolocation .
let @xmath0 denote the rayleigh distribution with the density @xmath1 .
let @xmath2 be independently distributed according to @xmath3 , representing the random voltages observed on the @xmath4 channels . in the absence of noise , @xmath5 s are all equal to one , the nominal value ; while in the presence of signal , exactly one of the @xmath5 s becomes a known value @xmath6 . denoting the uniform distribution on @xmath7 $ ] by @xmath8 , the goal is to test the following competing hypotheses @xmath9 , \quad \mbox{versus}\quad h_{1}^{(n ) } : \alpha_i = 1 + ( \alpha -1 ) { { \mathbf{1}_{\left\{{i = j}\right\ } } } } , \quad j \sim u_n \ , .
\label{eq : ht.dobrushin}\ ] ] since the signal only appears once out of the @xmath4 samples , in order for the signal to be distinguishable from noise , it is necessary for the amplitude @xmath10 to grow with the sample size @xmath4 ( in fact , at least logarithmically ) . by proving that the log - likelihood ratio converges to a stable distribution in the large-@xmath4 limit ,
dobrushin @xcite obtained sharp asymptotics of the smallest @xmath10 in order to achieve the desired false alarm and miss detection probabilities .
similar results are obtained in the continuous - time gaussian setting by burnashev and begmatov @xcite .
subsequent important work include ingster @xcite and donoho and jin @xcite , which focused on detecting a sparse binary vector in the presence of gaussian observation noise .
the problem can be formulated as follows .
given a random sample @xmath11 , one wishes to test the hypotheses @xmath12 \quad \mbox{versus}\quad h_{1}^{(n ) } : y_i \ , { { \stackrel{\text{{i.i.d.\xspace}}}{\sim}}}\ , ( 1-\epsilon_n ) { { \mathcal{n}}}(0,1 ) + \epsilon_n { { \mathcal{n}}}(\mu_n , 1 ) , i \in [ n ] \label{eq : ht.idj}\ ] ] where the non - null proportion @xmath13 is calibrated according to @xmath14 and the non - null effect @xmath15 grows with the sample size according to @xmath16 equivalently , one can write @xmath17 where @xmath18 is the observation noise . under the null hypothesis ,
the mean vector @xmath19 is equal to zero ; under the alternative , @xmath20 is a non - zero sparse binary vector with @xmath21 , where @xmath22 denotes the point mass at @xmath23 .
_ the detection boundary _ , which gives the smallest possible signal strength , @xmath24 , such that reliable detection is possible , is given by the following function in terms of the sparsity parameter @xmath25 : @xmath26 see ingster @xcite and donoho and jin @xcite .
therefore , the hypotheses in can be tested with vanishing probability of error if and only if the pair @xmath27 lies in the strict epigraph @xmath28 which is called the _
detectable region_. furthermore , because the fraction of the non - zero mean is very small , most tests based on the empirical moments have no power in detection .
donoho and jin @xcite proposed an adaptive testing procedure based on tukey s higher criticism statistic and showed that it attains the optimal detection boundary without requiring the knowledge of the unknown parameters @xmath27 .
the above results have been generalized along various directions within the framework of two - component gaussian mixtures .
jager and wellner @xcite proposed a family of goodness - of - fit tests based on the rnyidivergences @xcite , including the higher criticism test as a special case , which achieve the optimal detection boundary adaptively .
the detection boundary with correlated noise was established in @xcite which also proposed a modified version of the higher criticism that achieves the corresponding optimal boundary . in a related setup ,
@xcite considered detecting a signal with a known geometric shape in gaussian noise .
minimax estimation of the non - null proportion @xmath13 was studied in cai , jin and low @xcite .
the setup of @xcite and @xcite specifically focuses on the two - point gaussian mixtures .
although @xcite and @xcite provide insightful results for sparse signal detection , the setting is highly restrictive and idealized . in particular , it has the limitation that the signal strength must be a constant under the alternative , i.e. , the mean vector @xmath20 takes constant value @xmath15 on its support . in many applications ,
the signal itself varies among the non - null portion of the samples .
a natural question is the following : what is the detection boundary if @xmath15 varies under the alternative , say with a distribution @xmath29 ? motivated by these considerations ,
the following heteroscedastic gaussian mixture model was considered in cai , jeng and jin @xcite : @xmath30 in this case , ( * ? ? ?
* theorems 2.1 and 2.2 ) showed that reliable detection is possible if and only if @xmath31 where @xmath32 is given by @xmath33 where @xmath34 .
it was also shown that the optimal detection boundary can be achieved by a double - sided version of the higher criticism test . although the setup in cai , jeng and jin @xcite is more general than that considered in @xcite and @xcite , it is still restricted to the two - component gaussian mixtures .
in many applications such as the aforementioned multi - channel detection @xcite and astrophysical problems @xcite , the sparse signal may not be binary and the distribution may not be gaussian . in the present paper , we consider the problem of sparse mixture detection in a general framework where the distributions are not necessarily gaussian and the non - null effects are not necessarily a binary vector .
more specifically , given a random sample @xmath35 , we wish to test the following hypotheses @xmath36 where @xmath37 is the null distribution and @xmath38 is a distribution modeling the statistical variations of the non - null effects . the non - null proportion @xmath39 is calibrated according to .
in this paper we obtain an explicit formula for the fundamental limit of the general testing problem under mild technical conditions on the mixture .
we also establish the adaptive optimality of the higher criticism procedure across all sparse mixtures satisfying certain mild regularity conditions .
in particular , the general results obtained in this paper recover and extend all the previously known results mentioned earlier in a unified manner .
the results also generalize the optimality and adaptivity of the higher criticism procedure far beyond the original equal - signal - strength gaussian setup in @xcite and the heteroscedastic extension in @xcite . in the most general case ,
it turns out that the detectability of the sparse mixture is governed by the behavior of the log - likelihood ratio evaluated at an appropriate quantile of the null distribution .
although our general approach does not rely on the gaussianity of the model , it is however instructive to begin by considering the special case of _ sparse normal mixture _ with @xmath40 , i.e. , @xmath41 it is of special interest to consider the
_ convolution model _ , where @xmath42 is a standard normal mixture and @xmath43 denotes the convolution of two distributions . in this case
the hypotheses can be equivalently expressed via the additive - noise model , where @xmath44 under the null and @xmath45 under the alternative .
based on the noisy observation @xmath46 , the goal is to determine whether @xmath20 is the zero vector or a _ sparse _ vector , whose support size is approximately @xmath47 and non - zero entries are distributed according to @xmath29 .
therefore , the distribution @xmath29 represents the prior knowledge of the signal .
the case of @xmath29 being a point mass is treated in @xcite . the case of rademacher @xmath29 in covered in (
* chapter 8) . the heteroscedastic case where @xmath29 is gaussian is considered in @xcite .
these results can be recovered by particularizing the general conclusion in the present paper .
moreover , our results also shed light on what governs detectability in gaussian noise when the signal does not necessarily have equal strength .
for example , consider the classical setup where the signal strength @xmath15 is now a random variable .
if we have @xmath48 for some random variable @xmath49 , then the resulting detectable region is given by the ingster - donoho - jin expression scaled by the @xmath50-norm of @xmath49 .
on the other hand , it is also possible that certain distributions of @xmath15 induces different shapes of detectable region than .
see sections [ sec : char ] and [ sec : ex.dilate ] for further discussions .
the rest of the paper is organized as follows .
states the setup , defines the fundamental limit of sparse mixture detection and reviews some previously known results .
the main results of the paper are presented in sections [ sec : main ] and [ sec : hc ] , where we provide an explicit characterization of the optimal detection boundary under mild technical conditions .
moreover , it is shown in that the higher criticism test achieves the optimal performance adaptively .
particularizes the general result to various special cases to give explicit formulae of the fundamental limits .
discussions of generalizations and open problems are presented in .
the main theorems are proven in , while the proofs of the technical lemmas are relegated to the appendices . throughout the paper ,
@xmath51 and @xmath52 denote the cumulative distribution function ( cdf ) and the density of the standard normal distribution respectively .
let @xmath53 .
let @xmath54 denote the @xmath4-fold product measure of @xmath55 .
we say @xmath55 is absolutely continuous with respect to @xmath56 , denoted by @xmath57 , if @xmath58 for any measurable set @xmath59 such that @xmath60 .
we say @xmath55 is singular with respect to @xmath56 , denoted by @xmath61 , if there exists a measurable @xmath59 such that @xmath62 and @xmath60 .
we denote @xmath63 if @xmath64 , @xmath65 if @xmath66 , @xmath67 if @xmath68 and @xmath69 if @xmath70 .
these asymptotic notations extend naturally to probabilistic setups , denoted by @xmath71 , etc . , where limits are in the sense of convergence in probability .
in this section we define the fundamental limits for testing the hypotheses in terms of the sparsity parameter @xmath25 .
an equivalent characterization in terms of the hellinger distance is also given .
it is easy to see that as the non - null proportion @xmath13 decreases , the signal is more sparse and the testing problem in becomes more difficult .
recall that @xmath13 is given by where @xmath72 parametrizes the sparsity level .
thus , the question of detectability boils down to characterizing the smallest ( resp .
largest ) @xmath25 such that the hypotheses in can be distinguished with probability tending to one ( resp .
zero ) , when the sample size @xmath4 is large .
for testing between two probability measures @xmath55 and @xmath56 , denote the optimal sum of type - i and type - ii error probabilities by @xmath73 where the infimum is over all measurable sets @xmath59 . by the neyman - pearson lemma @xcite
, @xmath74 is achieved by the likelihood ratio test : declare @xmath55 if and only if @xmath75 .
moreover , @xmath74 can be expressed in terms of the _ total variation distance _ @xmath76 as @xmath77 for a fixed sequence @xmath78 , denote the total variation between the null and alternative by @xmath79 which takes values in the unit interval . in view of ,
the fundamental limits of testing the hypothesis are defined as follows .
@xmath80 if @xmath81 , the common value is denoted by @xmath82 .
[ def : betas ] as illustrated by , the operational meaning of @xmath83 and @xmath84 are as follows : for any @xmath85 , all sequences of tests have vanishing probability of success ; for any @xmath86 , there exists a sequence of tests with vanishing probability of error . in information - theoretic parlance , if @xmath87 , we say _ strong converse _ holds , in the sense that if @xmath88 , all tests fail with probability tending to one ; if @xmath89 , there exists a sequence of tests with vanishing error probability . and regimes of ( in)distinguishability of the hypotheses in the large-@xmath4 limit .
] clearly , @xmath84 and @xmath83 only depend on the sequence @xmath90 .
the following lemma , proved in , shows that it is always sufficient to restrict the range of @xmath25 to the unit interval .
@xmath91 [ lmm : beta01 ] \(o ) at ( -1,0 ) ; ( bl ) at ( 2,0 ) ; ( bu ) at ( 3,0 ) ; ( one ) at ( 6,0 ) ; ( b ) at ( 7,0 ) ; ( d2 ) [ right of = b , node distance=.2 cm ] @xmath25 ; ; ; ; ; ( o ) ( b ) ; ( o ) circle[radius=1pt ] ; ( @xmath92 ) ( bu ) ; ( @xmath93 ) ( bl ) ; ( one ) circle[radius=1pt ] ; at ( .5,.3 ) @xmath94 ; at ( .5,-.3 ) @xmath95 ; at ( 4.5,.3 ) @xmath96 ; at ( 4.5,-.3 ) @xmath97 ; in the gaussian mixture model with @xmath40 , if the sequence @xmath98 is parametrized by some parameter @xmath24 , the fundamental limit @xmath82 in is a function of @xmath24 , denoted by @xmath99 . for example , in the ingster - donoho - jin setup where @xmath100 , @xmath82 , denoted by @xmath101 , can be obtained by inverting : @xmath102 in terms of , the detectable region is given by the strict hypograph @xmath103 .
the function @xmath101 , plotted in , plays an important role in our later derivations .
similarly , for the heteroscedastic mixture , inverting gives @xmath104 as shown in , all the above results can be obtained in a unified manner as a consequence of the general results in .
closely related to the total variation distance is the hellinger distance ( * ? ? ?
* chapter 2 ) @xmath105 which takes values in the interval @xmath106 $ ] and satisfies the following relationship : @xmath107 therefore , the total variation distance converges to zero ( resp .
one ) is equivalent to the squared hellinger distance converges to zero ( resp .
we will be focusing on the hellinger distance partly due to the fact that it tensorizes nicely under the product measures : @xmath108 denote the hellinger distance between the null and the alternative by @xmath109 in view of and , the fundamental limits @xmath84 and @xmath83 can be equivalently defined as follows in terms of the asymptotic squared hellinger distance : @xmath110
in this section we characterize the detectable region explicitly by analyzing the exact asymptotics of the hellinger distance induced by the sequence of distributions @xmath78 . this subsection
we focus on the case of sparse normal mixture with @xmath111 and @xmath38 absolutely continuous .
we will argue in that by performing the lebesgue decomposition on @xmath38 if necessary , we can reduce the general problem to the absolutely continuous case .
we first note that the _ essential supremum _ of a measurable function @xmath112 with respect to a measure @xmath113 is defined as @xmath114 we omit mentioning @xmath113 if @xmath113 is the lebesgue measure .
now we are ready to state the main result of this section .
let @xmath40 .
assume that @xmath38 has a density @xmath115 with respect to the lebesgue measure .
denote the log - likelihood ratio by @xmath116 let @xmath117 be a measurable function and define @xmath118 1 . if @xmath119 _ uniformly _ in @xmath120 , where @xmath121 on a set of positive lebesgue measure , then @xmath122 .
2 . if @xmath123 _ uniformly _ in @xmath120 , then @xmath124 . consequently , if the limits in and agree and @xmath121 on a set of positive measure , then @xmath125 .
[ thm : main ] . assuming the setup of
, we ask the following question in the reverse direction : what kind of function @xmath10 can arise in equations and ?
the following lemma ( proved in ) gives a necessary and sufficient condition for @xmath10 .
however , in the special case of convolutional models , the function @xmath10 needs to satisfy more stringent conditions , which we also discuss below .
suppose @xmath126 holds _ uniformly _ in @xmath120 for some measurable function @xmath117 .
then @xmath127 in particular , @xmath128 lebesgue - a.e .
conversely , for all measurable @xmath10 that satisfies , there exists a sequence of @xmath98 , such that holds . additionally , if the model is convolutional , i.e. , @xmath129 , then @xmath10 is convex .
[ lmm : alpha ] in many applications , we want to know how fast the optimal error probability decays if @xmath25 lies in the detectable region .
the following result gives the precise asymptotics for the hellinger distance , which also gives upper bounds on the total variation , in view of .
assume that holds .
for any @xmath130 , the exponent of the hellinger distance is given by @xmath131 where @xmath132 which satisfies @xmath133 ( resp .
@xmath134 ) if and only if @xmath135 ( resp .
@xmath136 ) .
[ thm : ebeta ] as an application of , the following result relates the fundamental limit @xmath82 of the convolutional models to the classical ingster - donoho - jin detection boundary : let @xmath129 .
assume that @xmath29 has a density @xmath137 which satisfies that @xmath138 uniformly in @xmath139 for some measurable @xmath140 .
then @xmath141 where @xmath101 is the ingster - donoho - jin detection boundary defined in .
[ cor : conv ] it should be noted that the convolutional case of the normal mixture detection problem is briefly discussed in ( * ? ? ?
* section 6.1 ) , where inner and outer bounds on the detection boundary are given but do not meet . here completely settles this question .
see for more examples .
we conclude this subsection with a few remarks on . under the assumption that the function @xmath121 on a set of positive lebesgue measure , the formula shows that the fundamental limit @xmath82 lies in the very sparse regime ( @xmath142 ) .
we discuss the two extremal cases as follows : 1 .
_ weak signal _ : note that @xmath143 if and only if @xmath144 almost everywhere . in this case
the non - null effect is too weak to be detected for any @xmath145 .
one example is the zero - mean heteroscedastic case @xmath146 with @xmath147 .
then we have @xmath148 .
2 . _ strong signal _ : note that @xmath149 if and only if there exists @xmath150 , such that @xmath151 and @xmath152 at this particular @xmath150 , the density of the signal satisfies @xmath153 , which implies that there exists significant mass beyond @xmath154 , the extremal value under the null hypothesis @xcite .
this suggests the possibility of constructing test procedures based on the _ sample maximum_. indeed , to understand the implication of more quantitatively , let us look at an even weaker condition : there exists @xmath150 such that @xmath151 and @xmath155 which , as shown in , implies that @xmath149 .
[ rmk : extreme ] in general @xmath82 need not exist .
based on , it is easy to construct a gaussian mixture where @xmath84 and @xmath83 do not coincide .
for example , let @xmath156 and @xmath157 be two measurable functions which satisfy and give rise to different values of @xmath158 in , which we denote by @xmath159 . then there exist sequences of distributions @xmath160 and @xmath161 which satisfy for @xmath156 and @xmath157 respectively .
now define @xmath98 by @xmath162 and @xmath163 . then by , we have @xmath164 .
the detection boundary in @xcite is obtained by deriving the limiting distribution of the log - likelihood ratio which relies on the normality of the null hypothesis .
in contrast , our approach is based on analyzing the sharp asymptotics of the hellinger distance .
this method enables us to generalize the result of to sparse non - gaussian mixtures , where we even allow the null distribution @xmath37 to vary with the sample size @xmath4 .
consider the hypothesis testing problem .
let @xmath165 .
denote by @xmath166 and @xmath167 the cdf and the quantile function of @xmath38 , respectively , i.e. , @xmath168 if the log - likelihood ratio @xmath169 satisfies @xmath170 as @xmath171 _ uniformly _ in @xmath172 for some measurable function @xmath173 . if @xmath174 on a set of positive lebesgue measure , then @xmath175 [ thm : ng ] the function @xmath176 appearing in satisfies the same condition as in . comparing with
, we see that the uniform convergence condition is naturally replaced by the uniform convergence of the log - likelihood ratio evaluated at the null quantile .
using the fact that @xmath177 for all @xmath178 @xcite , which implies that @xmath179 uniformly as @xmath180 , we can recover from by setting @xmath181 .
the results in and are obtained under the assumption that the non - null effect @xmath38 is absolutely continuous with respect to the null distribution @xmath37 .
next we show that it does not lose generality to focus our attention on this case . using the hahn - lebesgue decomposition ( * ? ? ?
* theorem 1.6.3 ) , we can write @xmath182 for some @xmath183 $ ] , where @xmath184 and @xmath185 . put @xmath186 which satisfies @xmath187
. then @xmath188 by , @xmath189 therefore the asymptotic hellinger distance of the original problem is completely determined by @xmath190 and the square - hellinger distance @xmath191 , which is also of a _
sparse mixture _ form , with @xmath192 replaced by @xmath193 given in .
in particular , we note the following special cases : 1 . if @xmath194 , then @xmath195 ( resp .
@xmath196 ) if and only if @xmath197 ( resp .
@xmath196 ) , which means that detectability of the original sparse mixture coincide with the new mixture .
2 . if @xmath198 , then @xmath199 , which means that the original sparse mixture can be detected reliably .
in fact , a trivial optimal test is to reject the null hypothesis if there exists one sample lying in the support of the singular component @xmath200 .
as discussed in , the fundamental limit @xmath82 of testing sparse normal mixtures can be achieved by the likelihood ratio test . however , in general the likelihood ratio test requires the knowledge of the alternative distribution , which is typically not accessible in practice . to overcome this limitation , it is desirable to construct _ adaptive _ testing procedures to achieve the optimal performance simultaneously for a collection of alternatives .
this problem is also known as _
universal hypothesis testing_. see , e.g. , @xcite and the references therein , for results on discrete alphabets .
the basic idea of adaptive procedures usually involves comparing the empirical distribution of the data to the null distribution , which is assumed to be known . for the problem of detecting sparse normal mixtures , it is especially relevant to construct adaptive procedures , since in practice the underlying sparsity level and the non - zero priors are usually unknown .
toward this end , donoho and jin @xcite introduced an adaptive test based on tukey s _ higher criticism _ statistic .
for the special case of , i.e. , @xmath201 , it is shown that the higher criticism test achieves the optimal detection boundary while being adaptive to the unknown non - null parameters @xmath27 . following the generalization by jager and wellner @xcite via rnyidivergence , next we explain briefly the gist of the higher criticism test .
given the data @xmath202 , denote the empirical cdf by @xmath203 respectively .
similar to the kolmogorov - smirnov statistic @xcite which computes the @xmath50-distance ( maximal absolute difference ) between the empirical cdf and the null cdf , the higher criticism statistic is the maximal pointwise @xmath204-divergence between the null and the empirical cdf .
we first introduce a few auxiliary notations .
recall that the @xmath204-divergence between two probability measures is defined as @xmath205 in particular , the binary @xmath204-divergence function ( i.e. , the @xmath204-divergence between bernoulli distributions ) is given by @xmath206 where @xmath207 denotes the bernoulli distribution with bias @xmath208 .
the higher criticism statistic is defined by @xmath209 based on the statistics , the higher criticism test declares @xmath210 if and only if @xmath211 where @xmath212 is an arbitrary fixed constant . the next result shows that the higher criticism test achieves the fundamental limit @xmath82 characterized by while being adaptive to all sequences of distributions @xmath98 which satisfy the regularity condition .
this result generalizes the adaptivity of the higher criticism procedure far beyond the original equal - signal - strength setup in @xcite and the heteroscedastic extension in @xcite . under the same assumption of , for any @xmath88 , the sum of type - i and type - ii error of the higher criticism test vanishes as @xmath213 .
[ thm : hc ]
in this section we particularize the general result in to several interesting special cases to obtain explicit detection boundaries .
we derive the classical detection boundary from for the equal - signal - strength setup , which is a convolutional model with signal distribution @xmath214 and @xmath15 in .
the log - likelihood ratio is given by @xmath215 plugging in @xmath216 , we have @xmath217 .
consequently , the condition is fulfilled uniformly in @xmath120 with @xmath218 straightforward calculation yields that @xmath219 applying , we obtain the desired expression for @xmath99 . as a variation of , the symmetrized version of @xmath220 was considered in ( * ? ? ?
* section 8.1.6 ) , whose detection boundary is shown to be identical to .
indeed , for binary - valued signal distributed according to , we have @xmath221 which gives rise to @xmath222 comparing with and , we conclude that the detection boundary still applies . generalizing both the unary and binary signal distributions in
, we consider @xmath29 that is the distribution of the random variable @xmath223 where @xmath224 is a sequence of positive numbers and @xmath49 is distributed according to a fixed distribution @xmath55 , parameterizing the shape of the signal . in other words , @xmath29 is the dilation of @xmath55 by @xmath15 .
we ask the following question : by choosing the sequence @xmath15 and the random variable @xmath49 , is it possible to have detection boundaries which are shaped differently than the classical ingster - donoho - jin detection boundary ?
it turns out that for @xmath225 , the answer to the above question is negative .
as the next theorem shows , the detection boundary is given by that of the classical setup rescaled by the @xmath50-norm of @xmath49 .
note that and corresponds to @xmath226 and @xmath227 , respectively .
consider the convolutional model @xmath129 , where @xmath29 is the distribution of @xmath228 . then @xmath229 [ cor : dilate ] recall that @xmath230 denotes the ingster - donoho - jin detection boundary defined in . since the log - likelihood ratio is given by @xmath231}$ ]
, we have @xmath232 } \nonumber \\ = & ~ { \mathop{\mathrm{ess \ , sup}}}_x { \left\ { -x^2 + 2 u x \right\ } } \log n ( 1+o(1 ) ) \label{eq : alphadilate } , \end{aligned}\ ] ] where we have applied and the essential supremum in is with respect to @xmath55 , the distribution of @xmath49 . therefore @xmath233 . applying yields the existence of @xmath82 , given by @xmath234 where follows from the facts that @xmath230 is increasing and that @xmath235
. tightens the bounds given at the end of ( * ? ? ?
* section 6.1 ) based on the interval containing the signal support . from we see that the detection boundary coincides with the classical case with @xmath236 replaced by @xmath50-norm of @xmath49 . therefore , as far as the detection boundary is concerned , only the support of @xmath49 matters and the detection problem is driven by the maximal signal strength .
in particular , for @xmath237 or non - compactly supported @xmath49 , we obtain the degenerate case @xmath149 ( see also about the strong - signal regime ) . however , it is possible that the density of @xmath49 plays a role in finer asymptotics of the testing problem , e.g. , the convergence rate of the error probability and the limiting distribution of the log - likelihood ratio at the detection boundary .
one of the consequences of is the following : as long as @xmath225 , non - compactly supported @xmath49 results in the degenerate case of @xmath238 , since the signal is too strong to go undetected .
however , this conclusion need not be true if @xmath15 behaves differently .
we conclude this subsection by constructing a family of distributions of @xmath49 with unbounded support and an appropriately chosen sequence @xmath239 , such that the detection boundary is non - degenerate : let @xmath49 be distributed according to the following _ generalized gaussian _ ( subbotin ) distribution @xmath240 @xcite with shape parameter @xmath241 , whose density is @xmath242 put @xmath243 .
then the density of @xmath244 is given by @xmath245 .
hence @xmath246 which satisfies the condition with @xmath247 applying , we obtain the detection boundary @xmath82 ( a two - dimensional _ surface
_ parametrized by @xmath248 shown in ) as follows @xmath249 where is the ingster - donoho - jin detection boundary .
dilate.pdf ( 8,60)@xmath82 ( 102,3)@xmath24 ( 60,57)@xmath250 ( 60,46)@xmath251 ( 60,26)@xmath252 ( 60,15)@xmath253 equation can be further simplified for the following special cases . * @xmath252 ( laplace ) : plugging into , straightforward computation yields @xmath254 * @xmath250 ( gaussian ) : in this case we have @xmath255 and @xmath256 .
this is a special case of the heteroscedastic case in @xcite , which will be discussed in detail in .
simplifying we obtain @xmath257 which coincides with .
the heteroscedastic normal mixtures considered in corresponds to @xmath258 with @xmath15 given in and @xmath259 .
in particular , if @xmath260 , @xmath38 is given by the convolution @xmath261 , where the gaussian component @xmath262 models the variation in the signal amplitude . for any @xmath120 , @xmath263 where @xmath264 similar to the calculation in , we have . ] @xmath265 and @xmath266 note that @xmath267 if @xmath268 . assembling and applying , we have @xmath269 solving the equation @xmath270 in @xmath24 yields the equivalent detection boundary in terms of @xmath24 . in the special case of @xmath271 , where the signal is distributed according to @xmath272 , we have @xmath273 therefore , as long as the signal variance exceeds that of the noise , reliable detection is possible in the very sparse regime @xmath145 , even if the average signal strength does not tend to infinity .
we consider the detection boundary of the following generalized gaussian location mixture which was studied in ( * ? ?
* section 5.2 ) : @xmath274 where @xmath240 is defined in , and @xmath275 . since @xmath276
uniformly in @xmath277 , is fulfilled with @xmath278 .
applying , we have @xmath279 it is easy to verify that agrees with the results in ( * ? ? ?
* theorem 5.1 ) .
similarly , the detection boundary for exponential-@xmath280 mixture in ( * ? ? ?
* theorem 1.7 ) can also be derived from .
we conclude the paper with a few discussions and open problems . our main results in only concern the _ very sparse _ regime @xmath282 .
this is because under the assumption in that @xmath121 on a set of positive lebesgue measure , we always have @xmath283 . one of the major distinctions between the very sparse and moderately sparse regimes is the effect of symmetrization . to illustrate this point ,
consider the sparse normal mixture model .
given any @xmath38 , replacing it by its symmetrized version @xmath284 always increases the difficulty of testing .
this follows from the inequality @xmath285 , a consequence of the convexity of the squared hellinger distance and the symmetry of @xmath51 .
a natural question is : does symmetrization always have an impact on the detection boundary ? in the very sparse regime , it turns out that under the regularity conditions imposed in , symmetrization does not affect the fundamental limit @xmath82 , because both @xmath38 and @xmath286 give rise to the same function @xmath10 .
it is unclear whether @xmath84 and @xmath83 remain unchanged if an arbitrary sequence @xmath98 is symmetrized .
however , in the moderately sparse regime , an asymmetric non - null effect can be much more detectable than its symmetrized version .
for instance , direct calculation ( see for example ( * ? ? ?
* section 2.2 ) ) shows that @xmath287 for @xmath288 , but @xmath289 for @xmath290 .
moreover , unlike in the very sparse regime , moment - based tests can be powerful in the moderately sparse regime , which guarantee that @xmath291 .
for instance , in the above examples @xmath288 or @xmath290 , the detection boundary can be obtained by thresholding the sample mean or sample variance respectively .
more sophisticated moment - based tests such as the excess kurtosis tests have been studied in the context of sparse mixtures @xcite .
it is unclear whether they are always optimal when @xmath292 .
while establishes the adaptive optimality of the higher criticism test in the very sparse regime @xmath145 , the optimality of the higher criticism test in the moderately sparse case @xmath292 remains an open question .
note that in the classical setup , it has been shown @xcite that the higher criticism test achieves adaptive optimality for @xmath293 $ ] and @xmath294 . in this case since @xmath295 , we have @xmath296 and thus does not apply .
it is possible to obtain a counterpart of and an analogous expression for @xmath82 for the moderately sparse regime if one assumes a similar uniform approximation property of the log - likelihood ratio , for example , @xmath297 for some function @xmath10 .
another interesting problem is to investigate the optimality of procedures introduced in @xcite based on rnyidivergence under the same setup of .
laplace s method ( see , e.g. , ( * ? ? ?
* section 2.4 ) ) is a technique for analyzing the asymptotics of integrals of the form @xmath298 when @xmath299 is large .
the proof of uses the following first - order version of the laplace s method .
since we are only interested in the exponent ( i.e. , the leading term ) , we do not use saddle - point approximation in the usual laplace s method and impose _ no _ regularity conditions on the function @xmath112 except for the finiteness of the integral .
moreover , the exponent only depends on the _ essential supremum _ of @xmath112 with respect to @xmath300 , which is invariant if @xmath112 is modified on a @xmath300-negligible set .
let @xmath301 be a measure space .
let @xmath302 be measurable .
assume that @xmath303 holds uniformly in @xmath304 for some measurable @xmath305 .
if @xmath306 for some @xmath307 , then @xmath308 [ lmm : laplace ] first we deal with the case of @xmath309 , which implies that @xmath310 for all @xmath311 . moreover , by chernoff bound , @xmath312 . by , for any @xmath313 , there exists @xmath314 such that @xmath315 for all @xmath304 and @xmath316 .
therefore , @xmath317 for any @xmath318 and @xmath311 . then @xmath319 . by the arbitrariness of @xmath23 and @xmath320 , we have @xmath321 .
next we assume that @xmath322 . by replacing @xmath112 with @xmath323
, we can assume that @xmath324 without loss of any generality
. then @xmath325 @xmath300-a.e .
hence , by , @xmath326 holds for all @xmath316 . by the arbitrariness of @xmath320
, we have @xmath327 for the lower bound , note that , by the definition of @xmath324 , @xmath328 for all @xmath212 .
therefore , by , we have @xmath329 for any @xmath318 and @xmath212 . first sending @xmath330 then @xmath331 and @xmath332 , we have @xmath333 completing the proof of .
let @xmath301 be a measure space .
let @xmath302 be measurable .
let @xmath334 satisfy @xmath335 @xmath300-a.e .
for some @xmath336 .
if @xmath337 holds uniformly in @xmath304 for some measurable @xmath338 , then @xmath339 conversely , if @xmath340 holds uniformly in @xmath304 for some measurable @xmath341 such that @xmath342 for some @xmath307 , then @xmath343 [ lmm : laplace2 ] by , for any @xmath313 , there exists @xmath344 such that @xmath345 for all @xmath304 and @xmath346 . by definition of @xmath347 ,
therefore , by , we have @xmath349 for all sufficiently large @xmath299 . first sending @xmath330
then @xmath332 , we obtain the desired .
next we assume that @xmath350 . by replacing @xmath112 with @xmath351
, we can assume that @xmath352 without loss of any generality .
then @xmath325 @xmath300-a.e . on @xmath59
hence , by , @xmath353 holds for all @xmath316 . by the arbitrariness of @xmath320 ,
we have @xmath354 the following lemma is useful for analyzing the asymptotics of hellinger distance : 1 .
[ sqrt1 ] for any @xmath355 , the function @xmath356 is strictly convex on @xmath357 and strictly decreasing and increasing on @xmath358 $ ] and @xmath359 , respectively .
[ sqrt2 ] for any @xmath360 , @xmath361 [ lmm : sqrt ] 1 .
since @xmath362 is strictly concave , @xmath363 is strictly convex .
solving for the stationary point yields the minimum at @xmath364 .
first we consider @xmath365 .
since @xmath366 is convex , @xmath367 is increasing .
consequently , we have @xmath368 for all @xmath369 .
+ next we consider @xmath370 . by the concavity of @xmath362 , @xmath371 is decreasing .
hence @xmath372 for all @xmath373 $ ] . assembling the above two cases yields .
the following lemmas are useful in proving : let @xmath140 be measurable and @xmath113 be any measure on @xmath374 .
the function @xmath375 defined by @xmath376 is decreasing and lower - semicontinuous , where the essential supremum is with respect to @xmath113 .
[ lmm : gs ] the monotonicity is obvious .
we only prove lower - semicontinuity , which , in particular , also implies right - continuity .
let @xmath377 . by definition of the essential supremum , for any @xmath378
, we have @xmath379 . by the dominated convergence theorem , @xmath380 .
hence there exists @xmath381 such that @xmath382 for all @xmath383 , which implies that @xmath384 for all @xmath383 . by the arbitrariness of @xmath378
, we have @xmath385 , completing the proof of the lower semi - continuity . under the conditions of , for any @xmath386 , @xmath387 [ lmm : gntail ]
first assume that @xmath388 .
then @xmath389 where the last equality follows from .
the proof for @xmath390 is completely analogous .
let @xmath391 . put @xmath392 .
since @xmath393 by assumption , we also have @xmath394 .
denote the likelihood ratio by @xmath395 .
then @xmath396 _ ( direct part ) _ recall the notation @xmath158 defined in , which can be equivalently written as @xmath397 assuming , we show that @xmath122 by lower bounding the hellinger distance .
to this end , fix an arbitrary @xmath212 .
let @xmath398 .
denote by @xmath399 the lebesgue measure on the real line . by definition of the essential supremum , @xmath400 since @xmath401 for all @xmath150 and @xmath402 , we must have @xmath403 since , by assumption , @xmath404 , there exists @xmath405 , such that @xmath406 by assumption , there exists @xmath407 such that @xmath408 holds for all @xmath120 and all @xmath409 . from , we have either @xmath410 or @xmath411
next we discuss these two cases separately : _ case i _ : assume .
let @xmath412 the square hellinger distance can be lower bounded as follows : @xmath413 } \label{eq : acc0}\\ \geq & ~ { \mathbb{e}\left [ { \left ( \sqrt{1 + n^{-\beta } ( \exp(\ell_n(u \sqrt{2 \log n } ) ) - 1 ) } - 1 \right)}^2 { { \mathbf{1}_{\left\{{|u| \leq 1 , \alpha(u)- \frac{u^2}{2 } \geq \beta + \delta - \frac{1}{2 } , \alpha(u ) \geq 2 \epsilon}\right\ } } } } \right ] } \nonumber \\ \geq & ~ { \mathbb{e}\left [ { \left ( \sqrt{1 + n^{-\beta } ( n^{\alpha(u ) - \epsilon } - 1)}-1 \right)}^2 { { \mathbf{1}_{\left\{{|u| \leq 1 , \alpha(u)- \frac{u^2}{2 } \geq \beta + \delta - \frac{1}{2 } , \alpha(u ) \geq 2 \epsilon}\right\ } } } } \right ] } \label{eq : acc2}\\ \geq & ~ \frac{(\sqrt{2}-1)^2}{4 } { \mathbb{e}\left [ n^{(\alpha(u ) - \epsilon-\beta ) \wedge 2(\alpha(u ) - \epsilon-\beta ) } { { \mathbf{1}_{\left\{{|u| \leq 1 , \alpha(u)- \frac{u^2}{2 } \geq \beta + \delta - \frac{1}{2 } , \alpha(u ) \geq 2 \epsilon}\right\ } } } } \right ] } \label{eq : acc3}\\ = & ~ \frac{(\sqrt{2}-1)^2 \sqrt{\log n}}{4\sqrt{\pi } } \int n^{(\alpha(u ) - \epsilon-\beta ) \wedge 2(\alpha(u ) - \epsilon-\beta ) - u^2 } { { \mathbf{1}_{\left\{{|u| \leq 1 , \alpha(u)- \frac{u^2}{2 } \geq \beta + \delta - \frac{1}{2 } , \alpha(u ) \geq 2 \epsilon}\right\ } } } } { { \rm d}}u \label{eq : acc4 } \\
\geq & ~ \frac{(\sqrt{2}-1)^2 \sqrt{\log n}}{4\sqrt{\pi } } \lambda{\left\ { |u| \leq 1 , \alpha(u)- \frac{u^2}{2 } \geq \beta + \delta - \frac{1}{2 } , \alpha(u ) \geq 2 \epsilon \right\ } } n^{-1+\frac{\delta}{2 } } \label{eq : acc5}\end{aligned}\ ] ] where * : by .
* : by .[sqrt1 ] and .
* : without loss of generality , we can assume that @xmath414 . then applying the lower bound in .[sqrt2 ] yields the desired inequality .
* : we used the density of @xmath415 defined in . * : given that @xmath416 and @xmath417 , we have both @xmath418 and @xmath419 .
_ case ii _ : now we assume .
following analogous steps as in the previous case , we have @xmath420 where is due to the following : since @xmath151 and @xmath421 , we have both @xmath422 and @xmath423 . combining and
we conclude that @xmath424 . by the arbitrariness of @xmath212 and the alternative definition of @xmath83 in
, the proof of @xmath122 is completed . _
( converse part ) _ fix an arbitrary @xmath212 .
let @xmath425 we upper bound the hellinger integral as follows : first note that @xmath426 } + { \mathbb{e}\left [ { \left ( \sqrt{1 + n^{-\beta } ( l_n - 1 ) } - 1 \right)}^2 { {
\mathbf{1}_{\left\{{l_n \leq 1}\right\ } } } } \right]}. \label{eq : h3}\ ] ] applying .[sqrt1 ] , we have @xmath427 } \leq ( \sqrt{1-n^{-\beta}}-1)^2 \leq n^{-2\beta } = o(n^{-1 } ) , \label{eq : h33}\ ] ] since @xmath428 by .
consequently , the asymptotics of the hellinger integral @xmath429 is dominated by the first term in , denoted by @xmath430 , which we analyze below using the laplace method . by
, there exists @xmath431 such that @xmath432 holds for all @xmath120 and all @xmath433 . then @xmath434 } \nonumber \\ = & ~ { \mathbb{e}\left [ { \left ( \sqrt{1 + n^{-\beta } ( \exp(\ell_n(u \sqrt{2 \log n } ) ) - 1 ) } - 1 \right)}^2 { { \mathbf{1}_{\left\{{l_n \geq 1}\right\ } } } } \right ] } \nonumber \\ \leq & ~ { \mathbb{e}\left [ { \left ( \sqrt{1 + n^{-\beta } ( n^{\alpha(u ) + \delta } - 1 ) } - 1 \right)}^2 { { \mathbf{1}_{\left\{{\alpha(u ) \geq -\delta}\right\ } } } } \right ] } \label{eq : cv0}\\ \leq & ~ { \mathbb{e}\left [ { \left ( \sqrt{1 + n^{\alpha(u ) + \delta-\beta } } - 1 \right)}^2 \right ] } \nonumber \\ \leq & ~ { \mathbb{e}\left [ n^{(2(\alpha(u ) + \delta-\beta ) ) \wedge ( \alpha(u ) + \delta-\beta ) } \right ] } \label{eq : cv1}\\ = & ~ \sqrt{\frac{\log n}{\pi } } \int n^{(2(\alpha(u ) + \delta-\beta))\wedge ( \alpha(u ) + \delta-\beta ) - u^2 } { { \rm d}}u \label{eq : cv2}\end{aligned}\ ] ] where and are due to and .[sqrt2 ] , respectively .
next we apply to analyze the exponent of .
first we verify the integrability condition : @xmath435 in view of . applying to
, we have @xmath436 by , @xmath437 holds a.e .
consequently , @xmath438 holds for almost every @xmath439 \cup [ 1,\infty)$ ] and @xmath440 holds for almost every @xmath441 $ ] . these conditions immediately imply that @xmath442 holds a.e . assembling and , we conclude that @xmath443 . by the arbitrariness of @xmath212 and the alternative definition of @xmath84 in
, the proof of @xmath124 is completed . in view of the proof of ,
the desired readily follows from combining , , and .
let @xmath391 .
write the hellinger integral in terms of the cdf of the likelihood ratio as follows : @xmath444 } \nonumber \\ = & ~ \int_{{{\mathbb{r}}}_+ } { { \mathbb{p}\left\ { { \left ( \sqrt{1 + n^{-\beta } ( l_n(w ) - 1 ) } - 1 \right)}^2 \geq t \right\ } } } { { \rm d}}t \nonumber \\ = & ~ \int_{{{\mathbb{r}}}_+ } { { \mathbb{p}\left\ { \sqrt{1 + n^{-\beta } ( l_n(w ) - 1 ) } \geq 1+\sqrt{t } \right\ } } } { { \rm d}}t + \int_0 ^ 1 { { \mathbb{p}\left\ { \sqrt{1 + n^{-\beta } ( l_n(w ) - 1 ) } \leq 1-\sqrt{t } \right\ } } } { { \rm d}}t \nonumber \\ = & ~ \underbrace{\int_{{{\mathbb{r}}}_+ } { { \mathbb{p}\left\ { l_n(w ) \geq 1 + n^{\beta}((1+\sqrt{t})^2 - 1 ) \right\ } } } { { \rm d}}t}_{\triangleq a_n } + \underbrace{\int_0 ^ 1 { { \mathbb{p}\left\ { l_n(w ) \leq 1 + n^{\beta}((1-\sqrt{t})^2 - 1 ) \right\ } } } { { \rm d}}t}_{\triangleq b_n } \label{eq : h2}\end{aligned}\ ] ] where @xmath445 due to the following reasoning : since @xmath446 , the integrand in @xmath447 is zero whenever @xmath448 , i.e. , @xmath449 .
therefore @xmath450 .
therefore @xmath429 is dominated by @xmath430 , which we analyze below : @xmath451 using the cdf of @xmath452 under the null hypothesis , therefore the exponent of @xmath430 coincides with the exponent of the integral @xmath453 , which can be analyzed via the laplace method .
[ rmk : lncdf ] put @xmath454 _
( necessity ) _ since @xmath455 , it is sufficient to prove @xmath456 since @xmath457 , we have @xmath458 . by assumption , @xmath459 uniformly in @xmath150 .
then for all @xmath212 , @xmath460 holds for sufficiently large @xmath4 .
in particular , @xmath461 . for
general @xmath462 , let @xmath463 , @xmath464 and @xmath465 .
put @xmath466 and @xmath467 . then @xmath468 $ ] .
hlder s inequality yields @xmath469 , which gives the desired .
it then follows from that @xmath470 , i.e. , @xmath128 a.e . _
( sufficiency ) _ let @xmath10 be a measurable function satisfying .
let @xmath38 be a probability measure with the density @xmath471 which is a legitimate density function in view of .
then the log - likelihood ratio satisfies @xmath472 , which fulfills uniformly . for convolutional models ,
the convexity of @xmath10 is inherited from the geometric properties of the log - likelihood ratio in the normal location model : since @xmath473}}{\varphi(y)}$ ] is convex for any random variable @xmath49 ( see , e.g. , ( * ? ? ?
* property 3 ) and @xcite ) , we have @xmath474 for any @xmath373 $ ] and @xmath475 . dividing both sides by @xmath476 and sending @xmath171
, we have @xmath477 . since @xmath478
, we have @xmath479 where the last equality follows from . plugging the above asymptotics into @xmath480
, we see that is fulfilled uniformly in @xmath481 with @xmath482 applying , we obtain @xmath483 where the last step follows from the .
let @xmath484 . put @xmath485 . since @xmath165 by assumption , we also have @xmath486 .
denote the likelihood ratio ( radon - nikodym derivative ) by @xmath487 .
then @xmath488 instead of introducing the random variable @xmath415 in for the gaussian case , we apply the quantile transformation to generate the distribution of @xmath489 : let @xmath415 be uniformly distributed on the unit interval .
then @xmath490 which is exponentially distributed . putting @xmath491 we have @xmath492 set @xmath493 and @xmath494 , which satisfy @xmath495 for all sufficiently large @xmath4 .
for the converse proof , we can write the square hellinger distance as an expectation with respect to @xmath496 : @xmath497 } \\ & ~ + { \mathbb{e}\left [ { \left ( \sqrt{1 + n^{-\beta } ( \exp(\ell_n(z_n(1-u ) ) ) - 1 ) } - 1 \right)}^2 { { \mathbf{1}_{\left\{{0 < u \leq \frac{1}{2}}\right\ } } } } \right ] } .\end{aligned}\ ] ] analogous to , by truncating the log - likelihood ratio at zero , we can show that the hellinger distance is dominated by the following : @xmath498 } \nonumber \\ & ~ + { \mathbb{e}\left [ { \left ( \sqrt{1 + n^{-\beta } ( \exp(\ell_n(z_n(1-u ) ) ) - 1 ) } - 1 \right)}^2 { { \mathbf{1}_{\left\{{0
< u \leq \frac{1}{2 } , t_n(s_n ) \geq 0}\right\ } } } } \right ] } \nonumber \\ = & ~ { \mathbb{e}\left [ { \left ( \sqrt{1 + n^{-\beta } ( \exp(r_n(s_n ) ) - 1 ) } - 1 \right)}^2 { { \mathbf{1}_{\left\{{s_n > \log_n 2 , r_n(s_n ) \geq 0}\right\ } } } } \right ] } \label{eq : jg1 } \\ & ~ + { \mathbb{e}\left [ { \left ( \sqrt{1 + n^{-\beta } ( \exp(t_n(s_n ) ) - 1 ) } - 1 \right)}^2 { { \mathbf{1}_{\left\{{s_n \geq \log_n 2 , t_n(s_n ) \geq 0}\right\ } } } } \right ] } \nonumber \\ \leq & ~ { \mathbb{e}\left [ { \left ( \sqrt{1 + n^{-\beta } ( n^{\alpha_0(s_n)+\delta } - 1 ) } - 1 \right)}^2 + { \left ( \sqrt{1 + n^{-\beta } ( n^{\alpha_1(s_n)+\delta } - 1 ) } - 1 \right)}^2 \right ] } \label{eq : jg2 } \\ \leq & ~ 2 \ , { \mathbb{e}\left [ n^{2(\alpha_0 \vee \alpha_1(u ) + \delta-\beta ) \wedge ( \alpha_0 \vee \alpha_1(u ) + \delta-\beta ) } \right ] } \\ \leq & ~ n^{-1-\delta } \label{eq : jg3 } \end{aligned}\ ] ] where follows from , from and from . the direct part of the proof is completely analogous to that of by lower bounding the integral in .
let @xmath499 , which is uniformly distributed on @xmath358 $ ] under the null hypothesis . with a change of variable
, we have @xmath500 which satisfies that @xmath501 @xcite .
therefore the type - i error probability of the test vanishes for any choice of @xmath212 .
it remains to show that @xmath502 under the alternative . to this end , fix @xmath503 and put @xmath504 and @xmath505 . by
, we have @xmath506 where @xmath507 is binomially distributed with sample size @xmath4 and success probability @xmath508 .
therefore @xmath509 } = \sqrt{n } \frac{\rho_{n , s } - r_{n , s}}{\sqrt{r_{n , s}(1-r_{n , s } ) } } = n^{\frac{1}{2}-\beta } \frac{{g_n}(\sqrt{2s\log n } ) - r_{n , s}}{\sqrt{r_{n , s}(1-r_{n , s } ) } } .
\label{eq : evn}\ ] ] and @xmath510 by chebyshev s inequality , @xmath511 } \right\ } } } \leq \frac{4 \ , { \mathsf{var}}v_n(s)}{{\mathbb{e}\left [ v_n(s ) \right]}^2 } = \frac{4 \rho_{n , s}(1- \rho_{n , s})}{n ( \rho_{n , s } - r_{n , s})^2}.\ ] ] by , @xmath512 where @xmath513 . plugging into and yields @xmath509 } = n^{\frac{1+s}{2}-\beta+v(s)+o(1 ) } \label{eq : evns}\ ] ] and @xmath511 } \right\ } } } \leq n^{2\beta - s-1 - 2v(s)+o(1)}+ n^{\beta-1-v(s)+o(1)}. \label{eq : pvns}\ ] ] suppose that @xmath514
. then @xmath515 } = \omega(\sqrt{\log \log n})$ ] .
moreover , we have @xmath516 and @xmath517 since @xmath518 . combining , and
, we obtain @xmath519 that is , the type - ii error probability also vanishes .
consequently , a sufficient condition for the higher criticism test to succeed is @xmath520 where follows from the following reasoning : by ( * ? ? ?
* proposition 3.5 ) , the supremum and the essential supremum ( with respect to the lebesgue measure ) coincide for all lower semi - continuous functions .
indeed , @xmath521 is lower semi - continuous by , and so is @xmath522 .
it remains to show that the right - hand side of coincides with the expression of @xmath82 in .
indeed , we have @xmath523 note that the second equality follows from interchanging the essential supremums : for any bi - measurable function @xmath524 , @xmath525 where the last essential supremum is with respect to the product measure . thus the proof of the theorem is completed .
this appendix collects a few properties of total variation and hellinger distances for mixture distributions .
let @xmath526 and @xmath527 .
then @xmath528 which satisfies @xmath529 [ lmm : h2.mix ] since @xmath527 , there exists a measurable set @xmath530 such that @xmath531 and @xmath532
. then @xmath533 the inequalities in follow from and the facts that @xmath534 and @xmath535 . for any probability measures
@xmath536 , @xmath537 is decreasing on @xmath358 $ ] .
[ lmm : hdec ] fix @xmath538 .
since @xmath539 , the convexity of @xmath540 yields @xmath541 we conclude this appendix by proving presented in : by , the function @xmath542 is decreasing , which , in view of the characterization , implies that @xmath543 .
thus it only remains to establish the rightmost inequality in . to this end , we show that as soon as @xmath25 exceeds @xmath544 , @xmath545 becomes @xmath546 regardless of the choice of @xmath98 : fix @xmath547 . then @xmath548 where follows from the data - processing inequality , which is satisfied for all @xmath112-divergences @xcite , in particular , the total variation : @xmath549 , where @xmath550 is any probability transition kernel .
while is sufficient for our purpose in proving , it is unclear whether the monotonicity carries over to @xmath551 , since product measures do not form a convex set .
it is however easy to see that @xmath551 is decreasing , which follows from the proof of with @xmath552 replaced by @xmath553 .
it is also clear that @xmath554 is decreasing in view of .
[ rmk : hdec ]
in this appendix we show that implies that @xmath555 , i.e. , for any @xmath556 , the hypotheses in can be tested reliably . without loss of generality ,
we assume that @xmath557 .
then @xmath558 we show that the total variation distance between the product measures converge to one .
put @xmath559^n$ ] . in view of the first inequality in
, the total variation distance can be lower bounded as follows : @xmath560 using , we have @xmath561 on the other hand , @xmath562 where the last equality is due to @xmath563 .
therefore @xmath564 for any @xmath556 , which proves that @xmath555 .
in fact , the above derivation also shows that the following _ maximum test _ achieves vanishing probability of error : declare @xmath210 if and only if @xmath565 .
in general the maximum test is suboptimal .
for example , in the classical setting where @xmath566 , ( * ? ? ?
* theorem 1.3 ) shows that the maximum test does not attain the ingster - donoho - jin detection boundary for @xmath567 $ ] . | detection of sparse signals arises in a wide range of modern scientific studies .
the focus so far has been mainly on gaussian mixture models . in this paper , we consider the detection problem under a general sparse mixture model and obtain an explicit expression for the detection boundary .
it is shown that the fundamental limits of detection is governed by the behavior of the log - likelihood ratio evaluated at an appropriate quantile of the null distribution .
we also establish the adaptive optimality of the higher criticism procedure across all sparse mixtures satisfying certain mild regularity conditions .
in particular , the general results obtained in this paper recover and extend in a unified manner the previously known results on sparse detection far beyond the conventional gaussian model and other exponential families .
* keywords : * hypothesis testing , high - dimensional statistics , sparse mixture , higher criticism , adaptive tests , total variation , hellinger distance . |
the integration by part identities @xcite are by now a standard tool for obtaining relations between the many integrals associated to any feynman graph or , equivalently , for working out recurrence relations for expressing the generic integral in terms of the `` master integrals '' or master amplitudes " of the considered graph .
a good example of the use of the integration by part identities is given in @xcite , where the recurrence relations for all the 2-loop self - mass amplitudes are established in the arbitrary masses case .
+ it has been shown in @xcite that by that same technique one can obtain a set of linear first order differential equations for the master integrals themselves ; the coefficients of the equations are ratio of polynomials with integer coefficients in all the variables ; the equations are further non homogeneous , with the non homogeneous terms given by the master integrals of the simpler graphs obtained from the considered graph by removing one or more internal propagators .
+ restricting ourselves for simplicity to the self - mass case , for any feynman graph the related integrals can in general be written in the form @xmath5 in more detail , @xmath6 stands for the @xmath7-continuous integration on an arbitrary number @xmath8 of loops and @xmath9 stands for the set of the corresponding loop momenta , so that there are altogether @xmath10 different scalar products , including @xmath11 ; @xmath12 is the product of any power of the scalar products in the numerators divided by any power of the propagators occurring in the graph ( all masses will always be taken as different , unless otherwise stated ) ; as the propagators are also simple combinations of the scalar products , simplifications might occur between numerator and denominator and as a consequence one expects quite in general @xmath13 different factors altogether in the numerator and denominator , independently of the actual number of propagators present in the graph ( graphs with less propagators have more factors in the numerator and _ viceversa _ ) ; therefore the symbol @xmath14 in eq.([1 ] ) stands in fact for a set of @xmath13 indices the ( integer ) powers of the @xmath13 factors .
+ the integration by parts corresponding to the amplitudes of eq.([1 ] ) are @xmath15 = 0 \ , \hskip 1 cm i=1, ...
,l { \label{2}}\ ] ] where @xmath16 stands for any of the @xmath17 vectors @xmath9 and @xmath18 ; there are therefore @xmath19 identities for each set of indices @xmath14 . the identity is easily established for small @xmath7 the integral of the divergence vanishes
when the derivatives are explicitly carried out , one obtains the sum of a number of terms , all equal to a simple coefficient ( an integer number or , occasionally , @xmath7 ) , times an integrand of the form @xmath20 , with the set of indices @xmath21 differing at most by a unity in two places from the set @xmath14 . +
that set of identities is infinite ; even if they are not all independent , they can be used for obtaining the recurrence relations , by which one can express each integral in terms of a few already mentioned
master amplitudes " , through a relation of the form @xmath22 where the set of indices @xmath23 takes the very few values corresponding to the master amplitudes , @xmath24 refers to simpler master integrals in which one or more denominators are missing , and the coefficients @xmath25 are ratios of polynomials in @xmath7 , masses and @xmath11 .
+ let us consider now one of the master amplitudes themselves , say the master amplitude identified by the set of indices @xmath23 ; according to eq.([1 ] ) we can write @xmath26 by acting with @xmath27 on both sides we get @xmath28 according to the discussion following eq.([2 ] ) , the _ r.h.s .
_ is a combination of integrands ; as eq.([3 ] ) applies to each of the corresponding integrals , one obtains the relations @xmath29 which are the required master equations . as in eq.([3 ] ) , @xmath24 refers to simpler master integrals ( in which one or more denominators are missing ; they constitute the non - homogeneous part of the master equations ) , to be considered as known when studying the @xmath30 .
+ it is obvious from the derivation that the master equations can be established regardless of the number of loops .
it is equally clear that for graphs depending on several external momenta ( such as vertex or 4-body scattering graphs ) one has simply to replace the single operator @xmath27 of eq.([5 ] ) by the set of operators @xmath31 , where @xmath32 run on all the external momenta , and with some more algebra one can obtain master equations in any desired mandelstam variable .
+ the master equations are a powerful tool for the study and the evaluation of the master amplitudes ; among other things : * they provide information on the values of the master amplitudes at special kinematical points ( such as @xmath33 in eq.([6 ] ) ; the _ l.h.s .
_ vanishes , as @xmath33 is a regular point , so that the _ r.h.s .
_ is a relation among master amplitudes at @xmath33 , usually sufficient to fix their values at that point ) ; * the master equations are valid identically in @xmath7 , so that they can be expanded in @xmath34 and solved recursively for the various terms of the expansion in @xmath34 , starting from the most singular ( with 2-loop amplitudes one expects at most a double pole in @xmath34 ) ; * when the initial value at @xmath35 has been obtained , the equations can be integrated by means of fast and precise numerical methods ( for instance with a runge - kutta routine ) , so providing a convenient approach to their numerical evaluation ; note that the numerical approach can be used both for arbitrary @xmath7 or for @xmath36 , once the expansion has been properly carried out ; * the equations can be used to work out virtually any kind of expansion , in particular the large @xmath11 expansion , as will be shown in some detail ; * in particularly simple cases ( for instance when most of the masses vanish and only one or two scales are left ) the analytic quadrature of the equations can lead to the analytic evaluation of the master amplitudes .
the use of the master equations for studying the 1-loop self - mass and the 2-loop sunrise self - mass graph has been already discussed , @xcite , @xcite ; we will describe here its application to the 2-loop 4-propagator self - mass graph shown in fig.1 .
+ the corresponding amplitude is defined as @xmath37\ ( k_2 ^ 2+m_2 ^ 2)\ [ ( p - k_1-k_2)^2+m_3 ^ 2 ] } ; { \label{6a } } \end{aligned}\ ] ] the conventional factor @xmath38 is defined in @xcite , it is sufficient to know that at @xmath36 its value is 1 ; if all momenta are euclidean the @xmath39 in the propagators is not needed .
+ skipping details , the master equation is + where the @xmath40 are the 2-loop self - mass sunrise master amplitudes @xcite@xcite , @xmath41 } ; { \label{6b } } \end{aligned}\ ] ] and @xmath42 while @xmath43 corresponds to the 2-loop vacuum amplitude , @xmath44 } { \label{6d } } \end{aligned}\ ] ] and , as usual , @xmath45 the value at @xmath35 is ( almost trivially ) found to be @xmath46 the expansion in @xmath34 reads @xmath47 by expanding in the same way all the other amplitudes occurring in the master equation eq.([7 ] ) and using the results of @xcite , the first values are found to be @xmath48 where @xmath49 is the finite part at @xmath36 of the 1-loop self mass ; more exactly , defining @xmath50 } \ , \ ] ] and expanding in @xmath34 , one finds @xcite @xmath51 with @xmath52 { \label{13 } } \end{aligned}\ ] ] where @xmath53 [ p^2+(m_1-m_2)^2 ] } \ ] ] @xmath54
quite in general , if + 5pt @xmath55 , the large @xmath11 expansion is @xmath56 any @xmath8-loop amplitude , indeed , at large @xmath11 develops @xmath8 terms with fractional powers " in @xmath11 , with exponents @xmath57 , besides the regular " term containing integer powers only . as any 2-loop amplitude has fractional dimension " equal to 2 ( in square mass units ) , on dimensional grounds the coefficients @xmath58 and @xmath59 must have fractional dimension " equal to 0 , 1 and 2 respectively .
+ similar expansions are valid for the sunrise amplitudes appearing in the _ r.h.s .
_ of eq.([7 ] ) such as @xmath60 as well as for the 1-loop self - mass amplitude ( whose fractional power " is just @xmath61 ) @xmath62 when the above expansions are inserted into the master equations and in the recurrence relations ( to be regarded , in this context , as differential equations in the masses ) , one obtains a number of equations providing important relations between the coefficients of the expansions . + as a result , one finds @xcite @xmath63 where @xmath64 is a dimensionless function of @xmath7 only ; an explicit calculation ( most easily performed by putting @xmath65 ) gives for its expansion in @xmath66 @xmath67 all the other @xmath68 can then obtained explicitly and are found to be proportional to @xmath64 .
+ one further finds @xmath69 and similar expressions for all the other @xmath70 .
+ note here that in the massless limit one has the exact relation @xmath71 likewise , one obtains ( @xcite ; the results are reported here with minor changes of notation )
@xmath72 @xmath73 @xmath74 by expanding as usual around @xmath36 @xmath75 one finds @xmath76 when the large @xmath11 expansions , eq.s([14],[15],[16 ] ) , are substituted into eq.([7 ] ) , one can express the @xmath77 in terms of the other coefficients , already known .
+ one finds @xmath78 and @xmath79 while @xmath80 depends on the combination @xmath81 when the combination is expanded around @xmath36 as in eq.([24 ] ) and the explicit values of eq.([25 ] ) are used , the first 3 terms of the expansion namely the double pole , the simple pole and the term constant in @xmath34 are all found to vanish , suggesting the existence of the exact relation @xmath82 the above result seems to be confirmed by a preliminary investigation of the large @xmath11 behaviour of the 5-propagator 2-loop self - mass graph ( in progress ) . when eq.([28 ] ) is taken as valid , one finds @xmath83 and @xmath84 { \label{30 } } \end{aligned}\ ] ] 2 truecm * acknowledgements . * as in previous work , the algebra needed in all the steps of the work has been processed by means of the computer program form @xcite by j. vermaseren .
one of the authors ( e.r . ) is glad to acknowledge an interesting discussion with k.g .
chetyrkin on the universality of the coefficients of the large @xmath11 expansions
. 2 truecm 99 k.g .
chetyrkin and f.v .
tkachov , , 159 ( 1981 ) ; f.v .
tkachov , , 65 ( 1981 ) .
tarasov , , 455 ( 1997 ) .
e. remiddi , , 1435 ( 1997 ) , hep - th/9711188 ( 1997 ) .
similar equations , for the one - loop box amplitudes in 4 dimensions , was proposed by v. de alfaro , b. jaki and t. regge in their paper _ differential properties of feynman amplitudes _ to the volume _ high - energy physics and elementary particles _
, iaea , vienna 1965 . a differential equations method , based on mass derivatives , has been also proposed by a.v .
kotikov , , 158 ( 1991 ) , and used in a number of subsequent papers ( see in a.v .
kotikov , hep - ph/9807440 ) . in that approach , amplitudes with a single non vanishing mass @xmath23 are expressed as a suitable integral of the corresponding massless amplitudes , which are taken as known .
m. caffo , h. czy , s. laporta and e. remiddi , hep - th/9805118 ( 1998 ) , ( to be published ) .
vermaseren , _ symbolic manipulation with form _ , computer algebra nederland , amsterdam ( 1991 ) . | the general lines of the derivation and the main properties of the master equations for the master amplitudes associated to a given feynman graph are recalled . some results for the 2-loop self - mass graph with 4 propagators
are then presented . * master equations for master amplitudes @xmath0 * m. caffo@xmath1 , h. czy @xmath2 , s. laporta@xmath3 and e. remiddi@xmath4 + * _ infn , sezione di bologna , i-40126 bologna , italy _ * _ dipartimento di fisica , universit di bologna , i-40126 bologna , italy _ * _ institute of physics , university of silesia , pl-40007 katowice , poland _
e - mail : [email protected] + [email protected] + [email protected] + [email protected] + - + pacs 11.10.-z field theory + pacs 11.10.kk field theories in dimensions other than four + pacs 11.15.bt general properties of perturbation theory + @xmath0presented at the zeuthen workshop on elementary particle physics - loops and legs in gauge theories - rheinsberg , 19 - 24 april 1998 . |
the acceleration of high - energy particles is an important problem in astrophysics .
this paper examines one acceleration mechanism , `` surfing acceleration , '' as it applies to relativistic particles at perpendicular shocks .
an illustration of an ion undergoing surfing acceleration at a perpendicular shock is given in figure [ fig : f1 ] .
in this picture an ion of velocity @xmath3 arrives just upstream of the shock with small @xmath4 and is reflected by the jump in the electrostatic potential @xmath5 at the shock , which is illustrated schematically in figure [ fig : f2 ] .
the lorentz force then causes the ion to return to the shock where it is again reflected and brought back to the shock front by the lorentz force .
this process continues , confining the particle to the vicinity of the shock , where it is accelerated in the @xmath6 direction by the motional electric field , @xmath7 , where @xmath8 is the plasma velocity .
if one thinks of @xmath9 in figure [ fig : f2 ] as analogous to the profile of an ocean wave , the accelerating ion in figure [ fig : f1 ] is analogous to a surfer . in this paper
, we will focus on ions .
however , shock surfing acceleration is believed to be important for electrons as well ( mcclements et al 2001 , hoshino & shimada 2002 ) , and it is trivial to modify the analysis of sections [ sec : analytic ] and [ sec : num ] to treat the electron case .
an ion s @xmath10-oscillations in front of the shock can be approximately described in terms of an effective potential energy ( lee et al 1996 , ucer & shapiro 2001 ) , which can be derived as follows .
we start with the @xmath10-component of the equation of motion , @xmath11 where @xmath12 is the relativistic lorentz factor , and we work in shock wave frame .
we assume that @xmath13 with @xmath14 a function of @xmath10 alone , that @xmath15 , that latexmath:[$|v_x| \ll scale that is much shorter than the time required for @xmath12 or @xmath17 to change appreciably .
we also treat @xmath18 as spatially uniform .
we then multiply equation ( [ eq : dpxdt ] ) by @xmath19 and integrate in time , treating @xmath12 and @xmath17 as constant , to obtain @xmath20 where @xmath21 , \label{eq : ueffdef0}\ ] ] @xmath22 is the energy of @xmath10-oscillations , and @xmath23 is the value of @xmath24 at its local minimum at @xmath25 .
we define @xmath26 where @xmath27 is the value of @xmath28 at its local maximum .
surfing ions , which are trapped in the effective potential well , have @xmath29 .
the effective potential energy is plotted in figure [ fig : f2 ] for @xmath30 .
the turning points of an ion of given @xmath22 , obtained by solving equation ( [ eq : xenergy ] ) with @xmath31 , are denoted @xmath32 and @xmath33 . over many bounces ,
@xmath12 and @xmath17 vary slowly , causing @xmath22 to change in a way that can be calculated by noting that the adiabatic invariant @xmath34 is approximately conserved ( ucer & shapiro 2001 ) .
surfing can only occur if there is a local minimum in @xmath24 , which requires that @xmath35 over some interval of @xmath10 [ see , e.g. , sagdeev & shapiro ( 1973 ) , katsouleas & dawson ( 1983 ) , ucer & shapiro ( 2001 ) ] . in order for particles with @xmath36 to surf
, @xmath37 must locally exceed @xmath18 , a condition that may arise at large alfvnic mach number ( hoshino & shimada 2002 ) .
the escape of a surfing particle from the effective potential well determines how far the particle can travel along the @xmath6 direction , which in turn determines the amount of energy the particle gains .
several escape mechanisms have been considered in the literature .
first , if the maximum value of @xmath38 , denoted @xmath39 , is less than @xmath18 ( assumed constant ) and a surfing particle s @xmath40 increases to the point that @xmath41 , then the local minimum of @xmath24 disappears and the particle escapes downstream ( sagdeev & shapiro 1973 ) .
second , as a non - relativistic particle s @xmath40 increases , the slope of @xmath24 at large @xmath10 increases as illustrated in figure [ fig : f2 ] , causing the width @xmath42 of a particle s bounce oscillations to decrease . since @xmath43 is approximately conserved , where @xmath44 is a typical value of @xmath45 , decreasing @xmath42 increases @xmath44 , thereby increasing @xmath22 ( ucer & shapiro 2001 ) . if @xmath22 increases above @xmath46 , then the particle escapes downstream ( lee et al 1996 , ucer & shapiro 2001 ) .
third , in an oblique shock the velocity vector of a surfing particle gradually rotates in the @xmath47-plane due to the lorentz force associated with @xmath48 , as depicted in figure [ fig : f3 ] , which is an adaptation of figure 4 of lee et al ( 1996 ) .
as this happens , @xmath40 is eventually reduced to zero and the particle escapes upstream since the @xmath49 component of the lorentz force ceases to turn the particle back towards the shock ( lee et al 1996 ) .
fourth , wave - particle interactions can reduce a surfing particle s @xmath40 to zero , again allowing the particle to escape upstream ( shapiro , lee , & quest 2001 ) .
fifth , as noted by lee , shapiro , & sagdeev ( 1996 ) and hoshino & shimada ( 2002 ) , some shocks are intrinsically nonstationary and can periodically `` break , '' allowing surfing particles to escape downstream . for perpendicular shocks and relativistic particles with @xmath36 , the first three mechanisms described above are absent .
this has led previous authors to consider scenarios of `` unlimited acceleration '' when @xmath50 over some interval in @xmath10 ( katsouleas & dawson 1983 , ucer & shapiro 2001 ) .
the purpose of this paper is to show that shock surfing of relativistic particles at perpendicular shocks leads to a power - law energy spectrum of accelerated particles , provided fluctuations in the shock electric and magnetic fields are not too large , and to calculate the power - law index of this spectrum .
we ignore diffusion in a particle s @xmath51-momentum , @xmath52 , and assume that changes in @xmath52 are dominated by the motional electric field .
however , we take into account moderate fluctuations in the electric and magnetic fields in the shock s vicinity , and in particular the effects of these fluctuations on the @xmath10 component of a particle s momentum , @xmath53 .
an ion propagating primarily in the @xmath6 direction encounters a series of random forces in the @xmath10 direction ( denoted @xmath54 ) due to fluctuations in @xmath38 and @xmath18 , as illustrated in figure [ fig : f4 ] .
these forces induce a series of random increments in @xmath53 , causing diffusion in @xmath53 .
for example , suppose the typical electric and/or magnetic field structures have a length @xmath55 along the @xmath51 direction and a typical amplitude @xmath56 , that an ion moves primarily in the @xmath57 direction at a speed @xmath58 , and that these structures fluctuate on a time scale that is longer than the time @xmath59 for a surfing particle to traverse one of the structures .
the random increment in @xmath53 when an ion moves through one structure then has a magnitude @xmath60 if the field structures are space filling , the ion undergoes one such momentum kick during each time @xmath61 , leading to a coefficient of diffusion in @xmath53 , @xmath62 that is independent of particle energy .
the energy spectrum of accelerated particles can be understood in terms of two competing effects .
on the one hand , diffusion in @xmath53 stochastically de - traps particles by increasing @xmath22 above @xmath46 , thereby allowing particles to escape downstream . on the other hand ,
the gradual increase in @xmath12 due to acceleration by @xmath63 focuses a surfing particle with @xmath36 into the bottom of the effective potential well , as follows from conservation of @xmath64 ( ucer & shapiro 2001 ) .
this combination of factors leads to a power - law energy spectrum of accelerated particles , @xmath65 .
the value of @xmath2 depends upon @xmath66 where @xmath67 is the time to double a surfing particle s energy and @xmath68 is the momentum diffusion time scale . as @xmath69
is decreased , the acceleration time is reduced relative to the momentum diffusion time scale , particles experience larger energy increases before escaping the surfing mechanism , and one expects the energy spectrum @xmath70 to become harder .
this expectation is born out by the calculations and simulations to be presented in sections [ sec : analytic ] and [ sec : num ] , which show that in the strong - scattering limit ( @xmath71 ) @xmath72 , in the moderate - scattering limit ( @xmath73 ) @xmath74 , and in the weak - scattering limit ( @xmath75 ) @xmath76 .
if , contrary to what we assume , the field fluctuations were so large that the minimum of the effective potential energy disappeared at locations separated by a typical distance @xmath77 along a surfing particle s trajectory , one would not obtain a simple power - law energy spectrum ; the spectrum would instead steepen above an energy @xmath78 corresponding to the energy a particle gains by surfing a distance @xmath77 ( hoshino & shimada 2002 ) . as mentioned previously ,
if a particle is assumed to undergo diffusion in @xmath52 with diffusion coefficient @xmath79 as well as `` advection '' in @xmath52 due to @xmath63 , then a particle can in principle diffuse to @xmath80 and escape into the upstream region .
the condition under which this escape mechanism can be neglected can be understood by considering an analogy to first - order fermi acceleration at a non - relativistic shock , as follows .
suppose a particle is a distance @xmath10 downstream of a shock , that the downstream fluid velocity is @xmath81 , and that the particle diffuses in space with diffusion coefficient @xmath82 .
the time for a particle to return to the shock in the absence of advection is @xmath83 , and the time for the particle s distance from the shock to be doubled by advection in the absence of diffusion is @xmath84 .
the probability that the particle returns to the shock is @xmath85 ( see , e.g. , bicout 1997 ) .
this advection - diffusion problem is analogous to the advection and diffusion in @xmath52 that would occur in shock surfing with nonzero @xmath79 .
the probability that a particle escapes the surfing mechanism by diffusing in momentum space to @xmath80 is again @xmath86 , where @xmath87 is the time to diffuse to @xmath80 in the absence of advection in @xmath52 , and @xmath88 is the time for a particle s @xmath51-momentum to double due to @xmath63 .
escape through @xmath52 diffusion is thus negligible provided @xmath89 , or latexmath:[$d_{p_y } \ll p_y^2
q we note that it is not clear how efficiently particles are injected into the surfing acceleration mechanism .
injection appears to be natural for pickup ions accelerated by the heliospheric termination shock . in the frame of the solar wind ,
pickup ions are born with particle speeds comparable to the solar - wind velocity , and thus pickup ions that arrive at the heliospheric termination shock at the right point in their gyromotion have very small @xmath4 , which allows them to be trapped in front of the shock ( lee et al 1996 , zank et al 1996 , lever et al 2001 ) .
surfing acceleration in other cases , however , may be more difficult to initiate
. it may be that particles can be injected at the time that a shock is formed if the shock originates from an explosive event .
it is also possible that turbulent conditions in the shock transition region can feed particles into the surfing mechanism .
hoshino & shimada ( 2002 ) have found electron surfing in pic simulations of perpendicular shocks , demonstrating that injection does occur , at least for electrons . additional work on injection , however , is needed .
although our analysis is restricted to the case of perpendicular shocks , some of the results may apply to the more general case of superluminal shocks , for which the point of intersection between a field line and the shock plane moves faster than @xmath91 . for such shocks
, there exists a `` perpendicular shock frame '' in which the shock is stationary and the upstream electric and magnetic fields lie in the shock plane ( begelman & kirk 1990 ) .
there are , however , two issues that would have to be addressed before our results could be generalized to this case . first , the condition for @xmath38 to exceed @xmath18 in more general superluminal shocks must be found .
second , in general the upstream flow velocity in the perpendicular shock frame has a component along the upstream magnetic field .
if particles start to surf with a significant @xmath92 velocity , they would only approach the state that we analyze with @xmath93 after significant acceleration in the @xmath6 direction . as a particle approaches this state ,
@xmath40 gradually increases , and the effects of such a gradual increase are not included in our analysis .
the remainder of this paper is organized as follows . in section [ sec :
analytic ] we show analytically that surfing ions that diffuse in @xmath53 are described by a power - law energy spectrum .
we calculate analytically the power - law index for the case of weak momentum diffusion ( @xmath75 ) , and derive a rough estimate for the power - law index in the case of strong momentum diffusion ( @xmath71 ) . in section [ sec : num ] we present monte carlo numerical calculations of surfing in the presence of diffusion in @xmath53 for the weak , moderate , and strong momentum - diffusion cases .
we summarize our results in section [ sec : disc ] .
we begin with the relativistic vlasov equation with a term added to model diffusion in @xmath53 , @xmath94 where @xmath95 is taken to be independent of particle energy as in equation ( [ eq : dpxest ] ) .
we consider a perpendicular shock and ignore spatial variations in the magnetic field and motional electric field : @xmath96 @xmath97 and @xmath98 we also take @xmath99 @xmath100 and @xmath101 we confine our analysis to relativistic surfing ions , for which we assume @xmath102 @xmath103 @xmath104 @xmath105 @xmath106 and @xmath107 equation ( [ eq : vl1 ] ) can then be rewritten as @xmath108 we now change variables , defining @xmath109 @xmath110 @xmath111 and @xmath112 or , equivalently , @xmath113 where @xmath114 is the effective potential energy for the particle s motion in @xmath10 illustrated in figure [ fig : f2 ] , @xmath22 is the energy associated with this motion , and @xmath23 is the value of @xmath24 at its local minimum at @xmath25 . as discussed in the introduction , surfing particles satisfy @xmath115 , where @xmath116 and @xmath27 is the value of @xmath24 at its local maximum . in terms of @xmath117 , equation ( [ eq : vl2 ] ) becomes @xmath118 where @xmath119 ( @xmath120 ) is the distribution function for particles with @xmath121 ( @xmath122 ) .
we assume that the time @xmath123 for a particle to bounce in the effective potential well is much shorter than the acceleration time or the time for a particle to escape from the potential well due to diffusion in @xmath53 .
we thus take @xmath124 and @xmath125 with @xmath126 we then set @xmath127 and expand equation ( [ eq : vl3 ] ) in powers of @xmath128 . upon collecting all terms of order @xmath129 (
there is only one ) and dividing by @xmath4 , we find that @xmath130 as in the introduction , we take @xmath32 and @xmath33 to be the bounce points of a particle of `` energy '' @xmath22 , with @xmath131 . since @xmath132 and @xmath133 give the distribution function at the same point in phase space , one has @xmath134 .
similarly , @xmath135 .
equation ( [ eq : eps0 ] ) thus implies that @xmath136 upon collecting all terms in equation ( [ eq : vl3 ] ) of order @xmath137 and dividing by @xmath4 , we find that @xmath138 we integrate equation ( [ eq : eps1 ] ) from @xmath32 to @xmath33 for @xmath119 , then integrate equation ( [ eq : eps1 ] ) from @xmath32 to @xmath33 for @xmath120 , and then add the two resulting equations to annihilate the terms involving @xmath139 , thereby obtaining @xmath140 where @xmath141 and @xmath142 we note the behavior of @xmath143 near @xmath144 for future reference .
near @xmath25 , @xmath145 using the technique of asymptotic matching , one can show that as @xmath146 , @xmath143 has the asymptotic expansion @xmath147 .
\label{eq : gexp}\ ] ] since @xmath148 exists and is integrable , the asymptotic expansion of @xmath148 near @xmath144 is given by term - wise differentiation of equation ( [ eq : gexp ] ) : @xmath149 .
\label{eq : dgdwexp}\ ] ] we now recast equation ( [ eq : solv ] ) into the standard fokker - planck form for the number of particles per unit @xmath22 per unit @xmath150 , @xmath151 we start by noting from equations ( [ eq : valgam ] ) , ( [ eq : deftp ] ) and ( [ eq : defpy0 ] ) that @xmath152 where both @xmath153 and @xmath63 are negative .
equations ( [ eq : taub2 ] ) and ( [ eq : valgam2 ] ) give @xmath154 next , we define @xmath155 and note that @xmath156 .
\label{eq : id2}\ ] ] throughout the rest of this section we drop the primes , setting @xmath157 . substituting @xmath158 into equation ( [ eq : solv ] ) and making use of equations ( [ eq : id1 ] ) and ( [ eq : id2 ] ) ,
we find @xmath159 where @xmath160 is the flux of particles in @xmath22-space .
the quantity in the first set of parentheses on the right - hand side of equation ( [ eq : defgamma ] ) is @xmath161 , while the coefficient of @xmath162 within the second set of parentheses is @xmath163 , where @xmath164 is the change in @xmath22 during a small time interval @xmath165 , and the angle brackets denote an average over the stochastic scattering that gives rise to diffusion in @xmath53 .
the term @xmath166 is the rate of change of @xmath22 caused by the approximate conservation of the adiabatic invariant @xmath167 that is , setting @xmath168 yields @xmath169 , which confirms the statement made in the introduction that @xmath64-conservation drives a relativistic surfing particle towards the bottom of the effective potential well ( ucer & shapiro 2001 ) . the terms on the right - hand side of equation ( [ eq : defgamma ] ) proportional to @xmath95 give the advection and diffusion in @xmath22-space arising from a time average of a particle s diffusion in @xmath53 .
if diffusion in @xmath53 causes a particle s @xmath22 to increase to @xmath46 , the particle escapes downstream .
we thus seek a solution to equation ( [ eq : fp1 ] ) subject to the boundary condition @xmath170 we also require that @xmath162 be finite and differentiable at @xmath171 . since there are no @xmath153 derivatives in equation ( [ eq : fp1 ] ) , we can treat particles with different values of @xmath153 independently .
we will henceforth focus on particles with a single value of @xmath153 , and assume that a large number of such particles are injected into the surfing mechanism at @xmath172 .
we rewrite equation ( [ eq : valgam2 ] ) , dropping the prime on @xmath173 , to obtain @xmath174 where @xmath175 is the time required to double the particle s initial @xmath51-momentum .
because a surfing particle s energy increases linearly in time , the number of particles that are accelerated to a final lorentz factor in the interval @xmath176 , denoted @xmath177 , is equal to the number of particles that escape the effective potential well in the time interval @xmath178 , where @xmath179 and @xmath180 from equation ( [ eq : gam1 ] ) .
thus , @xmath181 using equation ( [ eq : gam1 ] ) , we can rewrite equation ( [ eq : fp1 ] ) as @xmath182 + \frac{\partial ^2}{\partial u^2 } \left ( \frac{\alpha u p}{2h}\right ) \equiv{\cal l}(p ) \label{eq : fp2}\ ] ] where @xmath183 and @xmath184 the eigenvalue equation @xmath185 can be written in sturm - liouville form .
we consider eigenfunctions that are finite and differentiable at @xmath186 and satisfy @xmath187 for these boundary conditions , and since @xmath188 , the eigenvalues are real and the eigenfunctions of @xmath189 form a complete set on the interval @xmath190 . the solution for @xmath191 can thus be expanded as @xmath192 from equation ( [ eq : fp2 ] ) , @xmath193 where @xmath194 is determined from the initial conditions on @xmath162 .
we order the eigenvalues and their corresponding eigenfunctions so that @xmath195 etc . for @xmath196 , @xmath191
is dominated by the first term in the sum in equation ( [ eq : expp ] ) , @xmath197 and , from equations ( [ eq : defgamma ] ) , ( [ eq : dndg ] ) , and ( [ eq : longtime ] ) , @xmath198 where @xmath199 we note that @xmath200 , \label{eq : l}\ ] ] where @xmath201 multiplying equation ( [ eq : eig ] ) by @xmath202 , integrating from @xmath186 to @xmath203 , and integrating by parts , we find that @xmath204 we now calculate @xmath205 to leading order in @xmath69 when @xmath206 for any @xmath28 with a single potential well as in figure [ fig : f2 ] .
we take @xmath207 from sturm - liouville theory , we know that @xmath205 is the unique value of @xmath208 in equation ( [ eq : eig ] ) for which @xmath209 satisfies the boundary conditions and has no zeroes in the interval @xmath210 .
we use this fact to calculate @xmath205 in the following four steps : ( 1 ) we assume an order of magnitude for @xmath205 , ( 2 ) we derive the resulting solution for @xmath211 , ( 3 ) we determine the value of @xmath205 by requiring that @xmath212 and @xmath213 , and finally ( 4 ) we verify that @xmath211 has no zeroes for @xmath190 . starting with step 1
, we set @xmath214 and assume that @xmath215 is of order unity .
this value is close to @xmath216 s lower limit of 0 .
the corresponding particle energy spectrum is hard ( @xmath217 ) , which is expected in the small-@xmath69 limit , since the acceleration time is much less than the momentum - diffusion time [ see equation ( [ eq : intro_alpha ] ) ] .
we now carry out step 2 .
we set @xmath218 , \label{eq : deff}\ ] ] and rewrite @xmath219 as @xmath220 = - \frac{2\theta g^\prime}{\alpha^2 } \left ( e^{-u/\alpha } + e^{-1/\alpha}f\right ) .
\label{eq : fh}\ ] ] equations ( [ eq : pn0 ] ) and ( [ eq : pn1 ] ) give @xmath221 and @xmath222 we assume that @xmath223 decreases monotonically from 0 to @xmath224 as @xmath225 increases from 0 to 1 , an assumption that we verify at the end . then for @xmath226 , we can take @xmath227 on the right - hand side of equation ( [ eq : fh ] ) .
in addition , when @xmath228 , the entire right - hand side of equation ( [ eq : fh ] ) is of order @xmath229 and can be replaced by 0 to an excellent approximation .
thus , throughout the interval @xmath190 , we can neglect the second term in parentheses on the right - hand side of equation ( [ eq : fh ] ) , and write @xmath220 = -\frac{2\theta g^\prime e^{-u/\alpha}}{\alpha^2 } .
\label{eq : fh2}\ ] ] integrating twice , using @xmath230 from equation ( [ eq : gexp ] ) , and imposing equation ( [ eq : fu0 ] ) , we find that @xmath231 we now carry out step 3 .
imposing equation ( [ eq : fu1 ] ) and evaluating the integral in equation ( [ eq : fouter ] ) to lowest order in @xmath69 , we find @xmath232 for step 4 , we note that for @xmath190 , @xmath233 and @xmath234 are positive .
thus , @xmath223 decreases monotonically from @xmath235 to @xmath224 as @xmath225 is increased from 0 to 1 , as assumed .
this implies that @xmath162 has no zeroes for @xmath190 , and that the value of @xmath205 given by equations ( [ eq : deftheta ] ) and ( [ eq : theta ] ) is indeed ( a good approximation of ) the smallest eigenvalue .
equations ( [ eq : dgdwexp ] ) , ( [ eq : deftheta ] ) , and ( [ eq : theta ] ) give @xmath236 where @xmath237 is defined by equation ( [ eq : uexp ] ) .
the power - law index of the energy spectrum of accelerated particles is then @xmath238 from equation ( [ eq : deff ] ) , it can be seen that @xmath211 is strongly peaked near @xmath186 .
this means that if a large number of particles start surfing at @xmath172 , the majority of those remaining at large times will be concentrated near the bottom of the effective potential well ( with @xmath239 since @xmath240 ) . for reference ,
the higher eigenvalues can be found by expanding equation ( [ eq : eig ] ) in powers of @xmath69 , solving for @xmath241 in an inner region with @xmath242 and an outer region with @xmath243 , and then matching the inner and outer solutions in the region @xmath244 .
one finds that @xmath241 satisfies kummer s equation in the inner region regardless of the precise form of @xmath28 .
this is because @xmath245 in the inner region regardless of the exact form of @xmath28 .
the inner solution can be matched to the outer solution only if @xmath246 for @xmath247 , provided @xmath248 is of order unity ( e.g.,@xmath249 ) .
the @xmath248 zeroes of @xmath209 , denoted @xmath250 , occur in the inner region for @xmath248 of order unity , with each of the @xmath251 of order @xmath69 . when @xmath71 , we can write equation ( [ eq : eig ] ) to lowest order in @xmath252 as @xmath253 where @xmath254 we again seek a solution for @xmath255 that satisfies @xmath256 and @xmath257 . in going from equation ( [ eq : eig ] ) to equation ( [ eq : sd ] )
we have retained the highest order derivative .
we thus have a regular perturbation problem and do not expect a boundary layer . since there is no small parameter in equation ( [ eq : sd ] ) , we expect the smallest value of @xmath258 to be of order unity
. then , to lowest order in @xmath252 , @xmath259 to within a factor of order unity .
in this section , we confirm and extend the analytic results of section [ sec : analytic ] with the use of monte carlo test - particle calculations .
we follow a set of test particles ( ions ) with @xmath260 that obey the equations of motion @xmath261 and @xmath262 where @xmath263 is a stochastic function of time that increments @xmath53 by @xmath264 ( equal chance for + or - ) during each time step @xmath61 , where @xmath265 and @xmath95 is a constant .
we work in the shock wave frame , with @xmath13 , where @xmath266 we set @xmath267 , where @xmath268 , @xmath269 , and @xmath270 are the shock speed , the lorentz factor for the shock speed , and the upstream magnetic field in the laboratory frame , respectively .
although @xmath18 is in general not constant , we assume a constant value , @xmath271 .
we set @xmath272 , @xmath273 , and @xmath274 .
the minimum of the approximate effective potential @xmath275 occurs at @xmath276 .
we start the particles at @xmath277 , with @xmath278 , @xmath279 , and @xmath280 . by varying @xmath95 ,
we are able to consider different values of @xmath69 in equation ( [ eq : defalpha ] ) . to improve the statistics , we implement particle splitting as follows .
we start each simulation with 1000 particles . at each time step , we update @xmath281 and @xmath282 for each particle . when a particle escapes from the effective potential well the calculation for that particle is stopped and @xmath12 at that time is recorded .
when half of the original particles have escaped , each of the remaining particles is `` split''that is , a copy of each particle is created with the same position and momentum , and both the original and the copy are subsequently tracked .
when we calculate the energy spectrum , the weightings of split particles are halved .
the splitting process is repeated each time that the particle number drops to 500 until the end of the simulation , when every particle s @xmath12 exceeds a threshold value , @xmath283 . in the weak scattering case ,
@xmath284 , a difficulty for calculating the spectrum is that almost all of the particles reach the threshold @xmath285 before escaping .
we therefore use another method to obtain the spectrum when @xmath284 .
we assume that the spectrum is given by a power law , @xmath286 with @xmath287 .
the number of particles with @xmath12 between @xmath288 and @xmath289 is given by @xmath290 where @xmath291 is a constant .
the number of particles with @xmath292 is @xmath293 from eqs .
( [ nu1 ] ) and ( [ nu2 ] ) , we obtain the spectral index @xmath2 as @xmath294 we take @xmath295 and @xmath296 and track 5000 particles for each value of @xmath69 .
our numerical results for @xmath297 are plotted in figure [ fig : f5 ] along with our analytic result [ equation ( [ eq : bweak ] ) ] for the same shock parameters .
the numerical results and analytic result converge for @xmath75 .
our numerical results for @xmath298 are plotted in figure [ fig : f6 ] , and are consistent with equation ( [ eq : bstrong ] ) when @xmath71 .
in this paper we model surfing acceleration of relativistic particles at perpendicular shocks as a process in which particles undergo diffusion in @xmath53 ( @xmath10 being the direction of the shock normal ) while propagating in steady - state electric and magnetic field profiles . for shocks with @xmath50 over some interval in @xmath10 ,
where @xmath92 is the direction of the magnetic field , and for relativistic ions propagating primarily in the @xmath57 direction , surfing produces a power - law energy spectrum of accelerated particles , provided the momentum diffusion coefficient @xmath95 is independent of energy .
we have calculated analytically the power - law index for the case of weak momentum diffusion , given in equation ( [ eq : bweak ] ) , and have carried out monte carlo test - particle calculations to determine the power - law index for the weak , moderate , and strong momentum diffusion cases ( section [ sec : num ] ) .
we thank eric blackman , steve cowley , and the anonymous referee for helpful comments . this work was supported by nsf grant ast-0098086 and doe grants de - fg02 - 01er54658 and de - fc02 - 01er54651 at the university of iowa . | we study relativistic particles undergoing surfing acceleration at perpendicular shocks .
we assume that particles undergo diffusion in the component of momentum perpendicular to the shock plane due to moderate fluctuations in the shock electric and magnetic fields .
we show that @xmath0 , the number of surfing - accelerated particles per unit energy , attains a power - law form , @xmath1 .
we calculate @xmath2 analytically in the limit of weak momentum diffusion , and use monte carlo test - particle calculations to evaluate @xmath2 in the weak , moderate , and strong momentum - diffusion limits . |
there is a category @xmath0 whose objects are finite - dimensional symplectic manifolds @xmath1 and whose morphisms are symplectomorphisms @xmath2 . in attempting to understand the quantization procedure of physicists from a mathematical perspective , one may think of it as a functor from this symplectic category , where classical mechanics takes place , into the category of hilbert spaces and unitary operators , which is the realm of quantum mechanics .
it is well known that the category @xmath0 is too large , since there are `` no - go '' theorems that show that the group of all symplectomorphisms on @xmath1 does not act in a physically meaningful way on a corresponding hilbert space .
one standard remedy for this is to replace @xmath0 with a smaller category , replacing the symplectomorphism groups with certain finite - dimensional subgroups .
another is to replace the hilbert spaces and operators by objects depending on a formal parameter .
but there is also a sense in which the category @xmath0 is too _ small _ , since it does not contain morphisms corresponding to operators such as projectors and the self - adjoint ( or skew - adjoint ) operators that play the role of observables in quantum mechanics , nor can it encode the algebra structure itself on the space of observables .
( this collection of observables is not actually a hilbert space , but certain sets of operators do carry a vector space structure , with the inner product associated to the hilbert - schmidt norm . ) to enlarge the symplectic category , we look at the `` dictionary '' of quantization , following , for example , @xcite . in this dictionary , the cartesian product of symplectic manifolds corresponds to the tensor product of hilbert spaces , and replacing a symplectic manifold @xmath1 with @xmath3 ( which we denote by @xmath4 when we omit the symplectic structure from the notation for a given symplectic manifold ) corresponds to replacing a hilbert space @xmath5 by its conjugate , or dual , space @xmath6 .
thus , if symplectic manifolds @xmath7 and @xmath8 correspond to hilbert spaces @xmath9 and @xmath10 , the product @xmath11 corresponds to @xmath12 , which , with a suitable definition of the tensor product , is a space @xmath13 of linear operators from @xmath9 to @xmath10 .
another entry in the dictionary says that lagrangian submanifolds ( perhaps carrying half - densities ) in symplectic manifolds correspond to vectors or lines in hilbert space .
combining this idea with the one in the previous paragraph , we conclude that lagrangian submanifolds in @xmath11 should correspond to linear operators from @xmath9 to @xmath10.this suggests that , if the space of observables @xmath5 for a quantum system corresponds to a symplectic manifold @xmath14 , then the algebra structure on @xmath5 should be given by a lagrangian submanifold @xmath15 in @xmath16 the algebra axioms of unitality and associativity should be encoded by monoidal properties of @xmath15 in an extended symplectic category , @xmath17 , where the morphisms from @xmath14 to @xmath18 are the canonical relations ; i.e. , all the lagrangian submanifolds of @xmath11 ( not just those which are the graphs of symplectomorphisms ) and where the morphism composition is the usual composition of relations .
however , a problem immediately occurs : the composition of canonical relations may yield relations that are not submanifolds any more , and thus are not canonical relations !
@xmath17 is then not a true category , as the morphisms can not always be composed .
it is rather awkward to speak about a quantization functor in this context .
there have already been several approaches to remedy this defect .
one approach , developed by guillemin and sternberg in @xcite ( see @xcite for a recent version ) , is to consider only symplectic vector spaces and linear canonical relations .
another , suggested by wehrheim and woodward in @xcite , is to enlarge the category still further by allowing arbitrary `` formal '' products of canonical relations and equating them to actual products when the latter exist as manifolds . in this paper , we take yet another approach .
we construct a version of the extended symplectic `` category , '' which is a true category , by localizing it around lagrangian submanifolds .
its objects , called symplectic microfolds in the spirit of milnor s microbundles ( @xcite ) , are equivalence classes @xmath19 $ ] of pairs consisting of a symplectic manifold @xmath14 and a lagrangian submanifold @xmath20 , called the core .
the equivalence reflects the fact that these objects really describe the geometry of a neighborhood @xmath21 or a `` micro '' neighborhood @xmath21 of @xmath22 in @xmath14 .
in this `` micro '' setting , there is also a notion of canonical `` micro '' relations between two symplectic microfolds : they are lagrangian submicrofolds @xmath23 $ ] of the symplectic microfold product @xmath24\times[n , b]$ ] .
their composition is generally as ill behaved as it is for regular canonical relations .
one of the main points of this paper is to identify a certain subset of canonical `` micro '' relations satisfying a new transversality condition which ensures the composition is always well defined .
we consider these transverse `` micro '' canonical relations as morphisms between symplectic microfolds ; in this way , we obtain a new symmetric monoidal category : the extended microsymplectic category .
the extended symplectic `` category '' has been used as a sort of heuristic guideline in an attempt to quantize poisson manifolds ( see @xcite ) in a geometric way .
these attempts have been only partially successful due in part to the existence of nonintegrable poisson manifolds ( hence restricting the class of poisson manifold one can quantize ) , as well as to the ill - defined composition of canonical relations ( limiting thus the functorial properties of these geometric quantization methods ) .
the replacement of the extended symplectic `` category '' by its `` micro '' version provides new ways of dealing with both issues .
this paper lays the foundation for a series of work that revolves around two main themes : the categorification of poisson geometry and its functorial quantization as explained below .
since @xmath25 is a monoidal category , it is natural to consider its category of algebras @xmath26 .
the main statement we are aiming at here is the equivalence between this latter category and the category of poisson manifolds and poisson maps .
future research directions will include the study of a weakened version of @xmath26 whose algebra maps are replaced by bimodules .
this should correspond to a `` micro '' morita theory for poisson manifolds .
our second line of work will focus on constructing a monoidal functor from the extended microsymplectic category ( enhanced with half - density germs on the morphisms ) to the category of vector spaces .
since monoidal functors between monoidal categories induce functors between their respective categories of monoid objects , we obtain in this way a `` quantization '' functor from the category of poisson manifolds to the category of algebras in @xmath27 .
we thank giovanni felder , domenico fiorenza , jim stasheff and chenchang zhu .
a.s.c . acknowledges partial support of snf grant 200020 - 121640/1 , of the european union through the fp6 marie curie rtn enigma ( contract number mrtn - ct-2004 - 5652 ) , and of the european science foundation through the misgam program .
b. d. acknowledges partial support from snf grant pa002 - 113136 and from the netherlands organisation for scientific research ( nwo ) , and thanks wendy l. taylor for proofreading the manuscript .
acknowledges partial support from nsf grant dms-0707137 .
a @xmath28 @xmath29 @xmath30 @xmath31 consists of a manifold @xmath14 and a submanifold @xmath20 , called the @xmath32 .
two manifold pairs @xmath31 and @xmath33 are said to be equivalent if @xmath34 and if there is a third manifold pair @xmath35 such that @xmath36 is an open subset in both @xmath14 and @xmath18 simultaneously .
a map between local pairs is a smooth map from @xmath14 to @xmath18 that sends @xmath22 to @xmath37 .
note that we require equality of neighborhoods and not merely diffeomorphism .
a @xmath38 is an equivalence class of a local pairs @xmath31 .
we denote these equivalence classes either by @xmath19 $ ] or by @xmath39,a)$ ] .
sometimes , @xmath40 $ ] will be referred to as a @xmath29 @xmath41 around @xmath22 .
we define an equivalence relation on the maps of local pairs that send a representative of @xmath19 $ ] to a representative of @xmath42 $ ] by declaring two such maps equivalent if there is a common neighborhood of @xmath22 where they coincide .
the equivalence classes , written as@xmath43:[m , a ] & \longrightarrow & [ n , b],\end{aligned}\ ] ] are the maps between microfolds .
we say that @xmath44 $ ] is a germ above @xmath45 if , for @xmath46,$ ] we have that @xmath47 .
a @xmath48 of a microfold @xmath19 $ ] is a microfold @xmath42 $ ] such that @xmath49 and @xmath50 .
we define the graph of a microfold map @xmath44 $ ] as the submicrofold@xmath51 & : = & \big([{\operatorname{gr}}\psi],{\operatorname{gr}}\psi_{|a}\big)\end{aligned}\ ] ] of the product microfold @xmath52\times[n , b ] & : = & [ m\times n , a\times b].\end{aligned}\ ] ] microfolds and microfold maps form a category .
fibered products of microfolds are defined in the obvious way .
there are micro counterparts of symplectic manifolds .
a @xmath53 @xmath38 is a microfold @xmath54 $ ] where @xmath14 is a symplectic manifold and @xmath55 a lagrangian submanifold .
we call @xmath56 @xmath57 the symplectic microfolds of the form @xmath58.$ ] we will write @xmath59 to denote the zero section of a vector bundle @xmath60 . in the previous definition , we abused notation by writing @xmath58 $ ] instead of @xmath61 $ ] . a symplectomorphism between symplectic microfolds is a microfold map for which there is a representative that is a symplectomorphism . to avoid an explosion in the use of the prefix `` micro '' , we will keep the usual `` manifold terminology '' when available and assume that we are talking about the `` micro '' version when microfolds are around and no confusion is possible . for examples ,
we choose to use `` symplectomorphism '' instead of `` microsymplectomorphism '' , and so on .
symplectic microfolds and their symplectomorphisms form a category , which we denote by @xmath62 .
many special submanifolds of symplectic geometry have their corresponding `` micro '' versions .
a microsubmanifold @xmath63 $ ] of a symplectic microfold @xmath54 $ ] will be called @xmath64 , @xmath65 , or @xmath66 if there are representatives of @xmath67 $ ] and @xmath40 $ ] which are isotropic , lagrangian or coisotropic , respectively .
the language of microfolds is useful to express local geometric properties ; that is , properties that are true for all neighborhoods of some submanifold .
for instance , the lagrangian embedding theorem can be phrased as follows .
for any symplectic microfold @xmath54 $ ] , there exists a symplectomorphism @xmath68:[m , l ] & \longrightarrow & [ { { t}^*}l , l],\label{thm : lagrangina embedding}\end{aligned}\ ] ] above the identity .
actually , this theorem was first stated and proved using the language of local manifold pairs and their equivalences ( see @xcite ) .
an important notion in symplectic geometry is that of a @xmath69 @xmath70 ; i.e. , a lagrangian submanifold @xmath71 , where @xmath72 and @xmath73 are symplectic manifolds and @xmath4 is the symplectic manifold @xmath74 .
canonical relations are usually thought of as `` generalized symplectomorphisms '' and written as @xmath75 instead of @xmath71 .
the rationale behind this is twofold : the graph of a symplectomorphism is a canonical relation , and it is formally possible to extend the composition of symplectomorphisms to canonical relations .
namely , the composition of @xmath71 and @xmath76 is the subset of @xmath77 defined by@xmath78 where @xmath79 is the `` reduction '' map that projects @xmath80 to @xmath81 .
the major issue here is that composition of canonical relations is generally ill defined : @xmath82 may fail to be a submanifold , although , when it is , it is a lagrangian one and @xmath83 are respectively coisotropic and lagrangian in @xmath84 .
the composition @xmath82 is exactly the quotient of @xmath83 by the characteristic foliation of @xmath85 .
this ensures that @xmath82 is a lagrangian submanifold whenever it is a submanifold . ] . there is a well known criterion which limits the wildness of the composition and which we will need later .
[ thm : clean intersection]the composition @xmath82 of the canonical relations @xmath86 and @xmath87 as above is an immersed lagrangian submanifold of @xmath77 if the submanifolds @xmath83 and @xmath85 intersect cleanly . nevertheless , it is standard to think of symplectic manifolds and canonical relations as a category .
it is called the @xmath88 @xmath53 `` @xmath89 '' and will be denoted by @xmath17 .
many constructions in @xmath0 extend to @xmath17 .
for instance , we define the image of a point @xmath90 by a canonical relation @xmath71 as the subset of @xmath18 given by@xmath91 where @xmath92 is the projection on the second factor of @xmath93 .
the tangent relation @xmath94 to a canonical relation @xmath75 as the subset @xmath95 of @xmath96 given by the set of tangent vectors to @xmath86 .
the notion of canonical relation between symplectic manifolds can be transported to symplectic microfolds .
a @xmath97 @xmath98,k)$ ] between the symplectic microfolds @xmath19 $ ] and @xmath42 $ ] is a lagrangian submicrofold @xmath98,k)$ ] of @xmath99 $ ] .
we will often prefer the notation @xmath98,k)$ ] for canonical relations and reserve the notation @xmath19 $ ] for symplectic microfolds in order to distinguish between objects and morphisms .
we will also use the arrow notation@xmath100,k\big):[m , a ] & \longrightarrow & [ n , b]\end{aligned}\ ] ] to represent canonical relations between symplectic microfolds .
we interpret the core @xmath101 as a submanifold of @xmath102 and not as a submanifold of @xmath103 .
in other words , we decide to regard the core as a generalized morphism from @xmath37 to @xmath22 . the reason for this contraintuitive interpretation will become apparent later on . the composition of canonical relations in the microworld @xmath104 & \overset{([v],k)}{\longrightarrow } & [ n , b ] & \overset{([w],l)}{\longrightarrow } & [ p , c]\end{array}\ ] ] is given by the binary relation composition of their `` components '' @xmath105,l)\circ([v],k ) & : = & ( [ w\circ v],k\circ l).\label{eq : composition}\end{aligned}\ ] ] at this point , we still have the same kind of ill defined composition for canonical relations between symplectic microfolds as we had for symplectic manifolds .
however , we may consider a special type of canonical relations between symplectic microfolds that always compose well .
this is what we do next .
our starting point is the cotangent lift @xmath106 of a diffeomorphism @xmath107 : t^*a & & t^*b + & & _ _ b + a & _ & b it induces a canonical relation of the form@xmath108,{\operatorname{gr}}\phi\big):[{{t}^*}a , a ] & \longrightarrow & [ { { t}^*}b , b].\end{aligned}\ ] ] the fact that the canonical relation @xmath109,{\operatorname{gr}}\phi\big)$ ] comes from a map implies the identities , seen as a generalized morphism from @xmath110 to @xmath111 , satisfies @xmath112 for @xmath113 . ] @xmath114 for all @xmath115 in the zero section of @xmath116 and all tangent vectors @xmath117 to the zero section of @xmath116 . obviously , canonical relations coming from cotangent lifts compose well .
it turns out that the identities and are the key to this nice composability .
we therefore make the following definition .
a @xmath53 @xmath118 is a canonical relation of the form@xmath100,{\operatorname{gr}}\phi\big):[m , a ] & \longrightarrow & [ n , b],\end{aligned}\ ] ] where @xmath107 is a smooth map from @xmath37 to @xmath22 and such that there exists a representative @xmath119 $ ] for which@xmath120 where @xmath121 is the image of @xmath122 under the relation @xmath86 as defined by and @xmath123 is the image of @xmath117 under the tangent relation @xmath124 .
we will usually write @xmath125,\phi\big)$ ] instead of @xmath125,{\operatorname{gr}}\phi\big)$ ] .
the next proposition gives various characterizations of symplectic micromorphisms .
recall that a submanifold @xmath111 of a manifold @xmath14 is transverse to a subbundle @xmath126 of @xmath127 along a submanifold @xmath128 if @xmath129 in this case we write @xmath130 .
in particular , a submanifold @xmath111 is transverse to a submanifold @xmath110 along @xmath128 if @xmath111 is transverse to @xmath131 along @xmath132 ; we write this @xmath133
. a @xmath134 of a symplectic microfold @xmath19 $ ] is a lagrangian subbundle @xmath101 of the tangent bundle @xmath135 of @xmath14 restricted to @xmath22 such that , for all @xmath136 , we have instead of @xmath137 and @xmath138 instead of @xmath139 . ]
@xmath140 ( i.e. @xmath22 is transverse to @xmath101 along @xmath22 ) . a canonical relation of the form@xmath100,\phi\big):[m , a ] & \longrightarrow & [ n , b]\end{aligned}\ ] ] is said to be @xmath141
@xmath142 @xmath143 @xmath144 @xmath101 of @xmath42 $ ] if there is a @xmath119 $ ] such that@xmath145 in this case , we will abuse notation slightly and write@xmath146\pitchfork_{{\operatorname{gr}}\phi}\big({{t}}a\times k\big).\ ] ] @xmath147if @xmath98,\phi)$ ] is transverse to all splittings , we will say that it is @xmath148 @xmath141 .
the following proposition offers alternative descriptions of symplectic micromorphisms in terms of transverse intersections as pictured below :
@xmath149 at 10 180 @xmath98,\phi)$ ] at 10 150 @xmath150,\phi)$ ] at 10 60 @xmath151 at 140 30 @xmath152 at 140 60 @xmath37 at 187 105 @xmath18 at 187 160 @xmath22 at 247 40 @xmath14 at 289 40 @xmath151 at 249 160 @xmath152 at 237 100 @xmath98,\phi)$ ] at 280 65 [ pro : tranversality]consider a canonical relation of the form@xmath100,\phi\big):[m , a ] & \longrightarrow & [ n , b].\end{aligned}\ ] ] then , the following statements are equivalent : \1 .
@xmath98,\phi)$ ] is a symplectic micromorphism , \2 .
@xmath98,\phi)$ ] is strongly transverse , \3 .
there is a @xmath119 $ ] such that @xmath153 is transverse .
@xmath154 . in general
, we have that@xmath155 which yields @xmath156 because of .
similarly , one obtains @xmath157 from .
now , for any splitting @xmath101 of @xmath42 $ ] , we have that@xmath158 and , therefore , that@xmath159 along @xmath152 . since @xmath160 and since @xmath161
, we have that@xmath162 if and only if @xmath163 . by definition , we have that @xmath164 and ,
therefore , we can conclude that@xmath165 along @xmath152 . a dimension count yields@xmath166 which completes the proof that @xmath95 intersects @xmath167 transversally along @xmath152 .
first , note that the condition @xmath153 is part of both statements .
let @xmath101 be a splitting of @xmath42 $ ] . by hypothesis
, we have that@xmath169 along @xmath152 , which implies in particular that@xmath170 meaning that @xmath171 is a transverse intersection .
the fact that there exists @xmath119 $ ] such that@xmath173 implies .
namely , this gives immediately that @xmath174 and , by definition of @xmath121 ( see ) , we obtain that @xmath175 . since the intersection of @xmath86 with @xmath151 is transverse , it is also clean , i.e. , we have that@xmath176 using this equation and the same argument as above , one obtains . in step 1 of the proof of proposition [ pro : tranversality ]
, we showed that if a canonical relation @xmath98,\phi)$ ] is a symplectic micromorphism then there exists @xmath119 $ ] , such that @xmath86 intersects @xmath151 cleanly in @xmath152 : @xmath177 in turn , we proved that that conditions and imply that @xmath98,\phi)$ ] is strongly transverse .
this makes conditions and equivalent to @xmath98,\phi)$ ] being a symplectic micromorphism .
therefore , if we compare this with statement 3 of proposition [ pro : tranversality ] , we see that clean intersection is enough . in step 2
, we see that it is enough to have a single splitting @xmath101 of @xmath42 $ ] transverse to @xmath98,\phi)$ ] to show that it is a symplectic micromorphism .
we obtain thus the following , apparently weaker , version of proposition [ pro : tranversality ] .
[ cor : tranversality]consider a canonical relation of the form@xmath100,\phi\big):[m , a ] & \longrightarrow & [ n , b].\end{aligned}\ ] ] then , the following statements are equivalent : \1 .
@xmath98,\phi)$ ] is a symplectic micromorphism , \2 .
@xmath98,\phi)$ ] is transverse to a splitting of @xmath42 $ ] .
there is a @xmath119 $ ] such that @xmath153 is clean .
let us denote by @xmath178 the cotangent bundle of the one point manifold as the manifold containing only the singleton @xmath179 . ]
@xmath180 , which we regard as a symplectic microfold ; i.e.,@xmath181.\end{aligned}\ ] ] for any microsymplectic manifold @xmath54 $ ] , there is a unique symplectic micromorphism@xmath182}:{\textbf{e } } & \longrightarrow & [ m , l]\end{aligned}\ ] ] given by@xmath182 } : & = & \big(\big[\{(0,\star)\}\times l\big],{\operatorname{pr}}_{l}\big),\end{aligned}\ ] ] where @xmath183 is the unique map from @xmath184 to @xmath180 . on the other hand , symplectic micromorphisms @xmath185 & \longrightarrow & { \textbf{e}}\end{aligned}\ ] ] are in bijection with lagrangian submanifold germs @xmath186 $ ] around a given point @xmath187 that are transverse to @xmath184 at @xmath115 .
namely , the core map@xmath188 is specified by the image @xmath90 of the unique point @xmath189 ; hence@xmath190,\{x\}\big).\end{aligned}\ ] ] conditions and read @xmath191 and @xmath192 ; i.e. , such that @xmath193 and @xmath184 are transverse .
recall the standard identification@xmath194 via the `` schwartz transform '' ( see @xcite)@xmath195 for any smooth map @xmath196 , the transform of the conormal bundle @xmath197 is a symplectic micromorphism from @xmath58 $ ] to @xmath198 $ ] given by @xmath199,\phi\big).\end{aligned}\ ] ] we call it the @xmath56 @xmath200 of @xmath107 and denote it again by @xmath106 .
note that , whenever @xmath107 is a diffeomorphism , we slightly abuse notation since in this case@xmath201,\phi\big).\end{aligned}\ ] ] as in the macroworld , the graph@xmath51 & = & \big([{\operatorname{gr}}\psi],\psi_{|b}^{-1}\big)\end{aligned}\ ] ] of symplectomorphism germ @xmath44:[m , a]\rightarrow[n , b]$ ] is a symplectic micromorphism : obviously , we have that@xmath202 the following proposition tells us that every symplectic micromorphisms whose core map is a diffeomorphism is the graph of a symplectomorphism germ . [ prop : germinv ] if the core @xmath107 of a symplectic micromorphism@xmath203,\phi):[m , a ] & \rightarrow & [ n , b]\end{aligned}\ ] ] is a diffeomorphism , then there exists a symplectomorphism germ@xmath43:[m , a ] & \rightarrow & [ n , b]\end{aligned}\ ] ] such that @xmath204=([v],\phi)$ ] . for each @xmath205 , we have that the intersection@xmath206 is transverse .
we see this by counting the dimensions and by remarking that tangent space intersection @xmath207 is contained in @xmath208 , which implies that this intersection must be zero since@xmath209 this transverse intersection guarantees that , for each @xmath205 , there is a neighborhood @xmath210 of @xmath211 in @xmath93 and a neighborhood @xmath212 of @xmath122 in @xmath14 such that the restriction of the first factor projection,@xmath213 is a diffeomorphism .
let us denote by @xmath214 its inverse and set @xmath215 .
by construction , we have that@xmath216 since , for two overlapping neighborhoods @xmath212 and @xmath217 as above , the corresponding maps @xmath218 and @xmath219 coincide on @xmath220 , we obtain a germ@xmath43:[m , a ] & \longrightarrow & [
n , b]\end{aligned}\ ] ] such that @xmath204=([v],\phi)$ ] .
because @xmath221 $ ] is a lagrangian submanifold germ , @xmath44 $ ] is a symplectomorphism germ .
we now prove that the composition of symplectic micromorphisms [ m , a ] & & [ n , b ] & & [ p , c ] .
is always well defined and that@xmath222,\phi\circ\psi\big):[m , a ] & \longrightarrow & [ p , c]\end{aligned}\ ] ] is again a symplectic micromorphism .
we first prove a cleanness result in order to apply theorem [ thm : clean intersection ] to our micro setting .
[ lem : transvers]let @xmath125,\phi\big)$ ] and @xmath223,\psi\big)$ ] be symplectic micromorphisms as above . for all @xmath119 $ ] and @xmath224 $ ] , @xmath83 intersects @xmath225 transversally along @xmath226 .
we need to show that@xmath227 at all points@xmath228 in @xmath229 in the symplectic vector space @xmath230 we have that @xmath231 using the relation @xmath232 , which holds for any subspaces @xmath22 and @xmath37 of a symplectic vector space , one sees that the transversality equation @xmath233 becomes equivalent to @xmath234 we shall now prove that this last equation holds . by assumption , we have that @xmath235 moreover , we may rewrite @xmath36 as@xmath236 equation tells us that @xmath237 . using this and equation ,
we can then write @xmath238 since @xmath239 we see that @xmath240 and finally , that @xmath241 , as desired . the composition of two symplectic micromorphisms @xmath98,\phi)$ ] and @xmath150,\psi)$ ] via is well defined and yields a symplectic micromorphism again .
lemma [ lem : transvers ] together with a continuity argument yield that there is a neighborhood @xmath36 of @xmath226 where @xmath83 and @xmath85 still intersect transversally .
therefore , the map@xmath242 restricted to this neighborhood , is a immersion according to theorem [ thm : clean intersection ] . at this point ,
recall that a proper immersion @xmath243 that is injective on a closed submanifold @xmath244 is a embedding on a neighborhood of @xmath22 .
since the maps @xmath107 and @xmath245 are smooth , @xmath226 is closed .
moreover , on this submanifold , we have that@xmath246 meaning that @xmath79 maps @xmath226 diffeomorphically to @xmath247 .
therefore , there is a neighborhood @xmath248 of @xmath226 such that @xmath249 is a lagrangian submanifold containing @xmath247 .
this proves that the lagrangian submanifold germ @xmath250,\phi\circ\psi)$ ] is well defined .
we need to show that it is a symplectic micromorphism ; i.e. conditions and hold . to begin with ,
notice that @xmath251 now , since , by assumption , @xmath252 , we see that @xmath253 therefore , we obtain for @xmath82 , namely @xmath254 set @xmath255 with @xmath256 . realizing that @xmath257 a similar computation on the tangent space level yields for @xmath82 .
in this section , we reinterpret the results obtained so far in the language of monoidal categories .
we refer the reader to @xcite for an exposition on monoidal categories .
we will sometimes write @xmath258 for the objects and @xmath259 for the morphisms of a category @xmath260 . accordingly , given a functor @xmath261 , we denote by @xmath262 the object component of @xmath263 and by @xmath264 its morphism component .
* @xmath0 , the usual symplectic category of symplectic manifolds and symplectomorphisms , * @xmath17 , the extended symplectic `` category '' , where symplectomorphisms are replaced by canonical relations , and which is not a category .
* @xmath62 , the @xmath265 @xmath89 ; i.e. the category of symplectic microfolds and symplectomorphism germs , * @xmath25 , the @xmath88 @xmath265 @xmath89 ; i.e. , the category of symplectic microfolds and symplectic micromorphisms .
[ def : extension]a category @xmath266 is said to be an @xmath267 of a category @xmath260 if there is a functor @xmath261 such that @xmath268 is a bijection and such that @xmath269 is injective and bijective on the isomorphisms ; i.e.@xmath270 for all objects @xmath271 . in this case
, we also call the functor an @xmath267 .
the extended microsymplectic category is a symmetric monoidal category .
the tensor product of symplectic microfolds is simply given by@xmath52\otimes[n , b ] & : = & [ m\times n , a\times b].\end{aligned}\ ] ] given two symplectic micromorphisms@xmath273,\phi_{i}):[m_{i},a_{i } ] & \longrightarrow & [ n_{i},b_{i}],\quad i=1,2,\end{aligned}\ ] ] we define their tensor product as@xmath274,\phi_{1})\otimes([v_{2}],\phi_{2 } ) & : = & \bigg(\big[\big({\operatorname{id}}_{m_{1}}\times\epsilon_{n_{1},m_{2}}\times{\operatorname{id}}_{n_{2}}\big)(v_{1}\times v_{2})\big],\phi_{1}\times\phi_{2}\bigg),\end{aligned}\ ] ] where @xmath275 is the usual factor permutation .
the unit object @xmath178 is the cotangent bundle of the one - point manifold @xmath180 . as shown in section [ exa : unit object and morphism ]
, @xmath178 is initial .
the symmetry isomorphisms are given by@xmath276,[n , b ] } & : = & \big(\big[{\operatorname{gr}}\epsilon_{m , n}\big],\epsilon_{b , a}\big).\end{aligned}\ ] ] note that the opposite symplectic manifold @xmath277 has its natural micro version @xmath278}:=[\overline{m},l]$ ] .
it is straightforward , although cumbersome , to verify the following : an @xmath281 is a collection @xmath282 of sets together with composition laws@xmath283 for each @xmath284 , satisfying the associativity equations @xmath285 and unit @xmath286 such that @xmath287 for all @xmath288 . for any object @xmath111 in a monoidal category @xmath289 , one defines the endomorphism operad @xmath290 of @xmath111 to be the collection@xmath291 with the usual convention that @xmath292 for @xmath293 .
the composition laws are given by the tensor product and the usual composition in the category:@xmath294 the unit is the identity morphism @xmath295 .
since @xmath25 is a monoidal category , it makes sense to consider the endomorphism operad @xmath296)$ ] of a symplectic microfold @xmath19 $ ] .
there are two special operads sitting inside of it .
first , the cotangent lifts of the @xmath297-diagonal maps @xmath298 , @xmath299 , form an operad@xmath300)(n ) & : = & \big\{{{t}^*}\delta^{n}\big\},\quad n\geq1,\\ \mathcal{l}_{\delta}([m , a])(0 ) & : = & \big\{{\textbf{e}}_{[m , a]}\big\},\end{aligned}\ ] ] thanks to the properties@xmath301}\\ \delta^{k_{1}+\cdots+k_{n } } & = & \delta^{n}\circ(\delta^{k_{1}}\times\cdots\times\delta^{k_{n}})\\ { { t}^*}\delta^{n-1 } & = & { { t}^*}\delta^{n}\circ({\operatorname{id}}_{[m , a]}\otimes\cdots\otimes{\textbf{e}}_{[m , a]}\otimes\cdots\otimes{\operatorname{id}}_{[m , a]}).\end{aligned}\ ] ] now , @xmath302)$ ] sits in the suboperad @xmath303)$ ] of @xmath296)$ ] defined as follows . for @xmath299 , @xmath303)(n)$ ] is the set of symplectic micromorphisms @xmath19^{\otimes n}\rightarrow[m , a]$ ] whose core map is the @xmath297-diagonal @xmath304 . for @xmath293 , we set @xmath303)(0)=\{{\textbf{e}}_{[m , a]}\}$ ] .
note that the first degree of this suboperad is interesting : @xmath303)(1)$ ] is the group of symplectorphism germs @xmath305:[m , a]\rightarrow[m , a]$ ] fixing @xmath22 . in @xcite , @xmath306)$ ] was called the cotangent lagrangian operad over @xmath307 and @xmath308)$ ] the local lagrangian operad over @xmath307 .
they were introduced ad hoc in terms of generating functions of lagrangian submanifold germs . | we introduce the notion of symplectic microfolds and symplectic micromorphisms between them .
they form a monoidal category , which is a version of the `` category '' of symplectic manifolds and canonical relations obtained by localizing them around lagrangian submanifolds in the spirit of milnor s microbundles . |
in the standard model ( sm ) , the higgs mechanism requires only one higgs doublet to generate masses for fermions and gauge bosons .
it leads to the appearance of a neutral cp - even higgs scalar after spontaneous electroweak symmetry breaking .
the lep2 experiments have established a lower bound of 114.4 gev @xcite for the sm higgs boson mass at 95% confidence level . a two higgs doublet model ( 2hdm )
@xcite has higgs doublets @xmath12 and @xmath13 with the vacuum expectation values @xmath14 and @xmath15 .
there are five physical higgs bosons : a pair of singly charged higgs bosons @xmath16 , two neutral cp - even scalars @xmath6 ( heavier ) and @xmath17 ( lighter ) , and a neutral cp - odd pseudoscalar @xmath0 . the couplings of the higgs bosons to fermions and gauge bosons depend on the ratio of vacuum expectation values ( @xmath18 ) and a mixing angle ( @xmath19 ) between the weak and mass eigenstates of the neutral scalars .
the supersymmetry between a boson and a fermion preserves the elementary nature of higgs bosons .
the minimal supersymmetric standard model ( mssm ) @xcite requires two higgs doublets @xmath12 and @xmath13 coupling to fermions with weak isospin @xmath20 and @xmath21 respectively , to generate masses for fermions and gauge bosons and to cancel triangle anomalies associated with the fermionic partners of the higgs bosons . the higgs potential is constrained by supersymmetry such that all tree - level higgs boson masses and couplings are determined by just two independent parameters , commonly chosen to be the mass of the cp - odd pseudoscalar ( @xmath22 ) and @xmath23 .
the mixing angle @xmath19 between the neutral scalars is often chosen to be negative ( @xmath24 ) .
the lep2 collaborations have set a lower bound of 91 gev and 91.9 gev @xcite for the @xmath25 and the @xmath22 , respectively .
extensive studies have been made for the detection of a heavier mssm higgs boson ( @xmath26 or @xmath0 ) at the cern lhc @xcite .
for @xmath27 , @xmath28 , @xmath29 , and @xmath30 are possible discovery channels . in the sm
, @xmath31 offers great promise for 160 gev @xmath32 tev @xcite with physics background from @xmath33 @xcite and @xmath34 @xcite .
however , the heavy scalar in the mssm is observable via @xmath35 only for @xmath10 and @xmath36 gev @xcite .
the detection modes @xmath37 @xcite or @xmath38 @xcite and @xmath39 @xcite may be promising channels for simultaneous discovery of two higgs bosons in the mssm . for large values of @xmath23 ,
@xmath40 @xcite , and @xmath41 @xcite , are promising discovery channels for the @xmath0 and the @xmath6 . in some regions of parameter space ,
the rates for higgs boson decays to neutralinos ( @xmath42 ) are dominant and they might open up new promising modes for higgs detection @xcite . in two higgs doublet models ,
there are two complementary channels to search for a higgs scalar and a higgs pseudoscalar simultaneously : ( i ) @xmath43 @xcite with a coupling proportional to @xmath44 and ( ii ) @xmath7 with a coupling proportional to @xmath45 . at the lhc
, gluon fusion can be a significant source to produce a higgs pseudoscalar ( @xmath0 ) and a @xmath1 boson ( @xmath8 ) via triangle and box diagrams with the third generation quarks @xcite .
the top quark loop diagrams make dominant contribution to @xmath8 for @xmath10 @xcite . in the mssm and a 2hdm with model ii yukawa interactions
@xcite for the higgs bosons and fermions , one higgs doublet ( @xmath12 ) couples to down - type quarks and charged leptons while another doublet ( @xmath13 ) couples to up - type quarks and neutrinos .
the @xmath46 coupling is proportional to @xmath23 and this coupling can be greatly enhanced by a large value of @xmath23 .
in addition , the higgs pseudoscalar does not couple to gauge boson pairs at the tree level .
therefore , @xmath47 is the dominant decay channel for @xmath11 or for @xmath48 gev with @xmath49 .
recently , we demonstrated that gluon fusion could be a promising production mechanism to detect @xmath50 via @xmath8 for @xmath49 and @xmath51 gev @xcite . in this article , the prospects of the search for a higgs pseudoscalar ( @xmath0 ) associated with a @xmath1 boson produced are investigated .
we study the discovery potential of @xmath52 ) via bottom quark fusion ( @xmath2 ) at the lhc .
the production cross sections of @xmath9 at the lhc in a two higgs doublet model and the mssm are discussed in section ii .
the dominant physics backgrounds from production of @xmath53 and @xmath54 are presented in section iii .
the discovery potential of @xmath55 is discussed in section iv .
conclusions are drawn in section v.
we calculate the cross section for @xmath56 via @xmath2 in a two higgs doublet model and the minimal supersymmetric standard model with model ii yukawa interactions for the higgs bosons and fermions .
the parton distribution functions of cteq6l1 @xcite are employed to evaluate the cross section for @xmath52 with the higgs production cross section @xmath57 multiplied by the branching fractions of @xmath58 and @xmath47 . in the yukawa couplings of @xmath59 ( @xmath60 ) ,
the bottom quark mass is chosen to be the next - to - leading order ( nlo ) running mass @xmath61 @xcite , which is calculated with @xmath62 gev and the nlo evolution of the strong coupling @xcite . in figure 1 , we present the feynman diagrams for @xmath2 .
the s - channel diagrams contain the heavy higgs scalar @xmath6 [ @xmath63 and the light higgs scalar @xmath17 [ @xmath64 in the intermediate state .
therefore , this discovery channel might provide a good opportunity to measure the couplings of @xmath65 and @xmath66 .
the @xmath67 and @xmath68 channel diagrams are proportional to @xmath23 . when the @xmath67 and @xmath68 channel diagrams dominate , the cross section of @xmath2 can be enhanced by a large value of @xmath69 .
we have checked the relevant couplings with the unitarity condition for the total amplitude . at very high energy , the longitudinal polarization vector ( @xmath70 ) of the @xmath1 boson
can be expressed as the momentum vector over @xmath1 mass ( @xmath71 )
. then unitarity requires cancellation among the @xmath72 , @xmath67 , and @xmath68 channel diagrams at high energy .
figure 2 presents the cross section of @xmath52 via @xmath2 as a function of @xmath23 in a general two higgs doublet model and the minimal supersymmetric standard model .
we have chosen @xmath73 gev , @xmath74 gev , and @xmath75 for the general 2hdm .
it is clear that the cross section in a 2hdm can be significantly larger than that in the the mssm when the @xmath6 can decay into @xmath9 with @xmath76 since @xmath77 and @xmath19 are free parameters in a 2hdm . in the mssm with @xmath11 , @xmath22 and
@xmath25 are very close to each other for @xmath78 125 gev , while @xmath22 and @xmath77 are almost degenerate when @xmath79 125 gev @xcite .
therefore , the decay @xmath7 is kinematically inaccessible in the mssm .
the cross section for gluon fusion alone is also presented in this figure . while gluon fusion is the major source for producing @xmath9 with @xmath10 , bottom quark fusion can make dominant contribution for @xmath11 .
the higgs production rate via bottom quark fusion is very sensitive to the choice of factorization scale ( @xmath80 ) @xcite . in table
i we present the cross section of @xmath52 via @xmath2 in a two higgs doublet model with @xmath81 and several values of @xmath22 for @xmath82 . in our analysis
, we have chosen @xmath83 with the running bottom quark mass @xmath61 in the yukawa couplings .
our numerical values of cross section are comparable to the recent next - to - leading order qcd predictions for @xmath84 associated production @xcite .
= 1.5em .
the cross section in fb without cuts for @xmath85 via bottom quark fusion ( @xmath86 ) at @xmath87 tev .
we choose three values of the factorization scale ( @xmath80 ) for @xmath81 and several values of @xmath22 in a two higgs doublet model with @xmath74 gev , @xmath88 gev and @xmath75 . [ cols="^,^,^,^,^,^",options="header " , ] the discovery contours for @xmath52 via @xmath2 at the lhc are presented in figure 5 for an integrated luminosity of ( a ) @xmath89 fb@xmath90 and ( b ) @xmath91 fb@xmath90 .
we show 5@xmath92 contours in the @xmath93 plane for @xmath94 and 400 gev in a general two higgs doublet model with @xmath74 gev and @xmath88 gev . in addition
, we present the curve for the decoupling limit with @xmath95 . for @xmath89 fb@xmath90
, it will be possible to discover the @xmath9 signal for @xmath96 and @xmath97 gev .
the higher luminosity ( @xmath91 fb@xmath90 ) greatly improves the reach for @xmath22 up to 400 gev in a large region of the parameter space with @xmath98 .
bottom quark fusion and gluon fusion are complementary processes to produce a higgs pseudoscalar ( @xmath0 ) and a @xmath1 boson at the lhc . while gluon fusion is the major source of @xmath9 for @xmath10 , bottom quark fusion can make dominant contributions to the production of @xmath9 at the lhc for @xmath11 .
we have found promising results for @xmath99 via @xmath2 in two higgs doublet models at the lhc with @xmath91 fb@xmath90 for @xmath5 gev , @xmath100 , @xmath98 , and @xmath88 gev . in the mssm with @xmath101 gev , @xmath102 , and
the production cross section of @xmath8 is usually small .
gluon fusion ( @xmath8 ) offers great promise for @xmath51 gev and @xmath49 @xcite .
the production rate of @xmath9 from gluon fusion at the lhc is suppressed by the destructive interference between the triangle and the box diagrams as well as the negative interference between the top quark and the bottom quark loops , especially when they are comparable with @xmath103 @xcite . in a general two higgs doublet model ,
the cross section of @xmath2 and @xmath8 can be greatly enhanced when the heavier higgs scalar ( @xmath6 ) can decay into the higgs pseudoscalar and a @xmath1 boson .
if we take @xmath104 in a 2hdm , the higgs signal will be reduced to the level of the mssm .
this discovery channel might provide a good opportunity to discover two higgs bosons simultaneously if the heavier higgs scalar ( @xmath6 ) can decay into a @xmath1 boson and a higgs pseudoscalar ( @xmath0 ) .
we are grateful to sally dawson and michael spira for beneficial discussions .
this research was supported in part by the u.s .
department of energy under grants no .
de - fg02 - 04er41305 and no .
de - fg02 - 03er46040 .
the lep higgs working group , http://lephiggs.web.cern.ch/lephiggs/www/ ; lhwg note 2002 - 01
. j. gunion , h. haber , g. kane and s. dawson , _ the higgs hunter s guide _ ( addison - wesley , redwood city , ca , 1990 ) . h.p .
nilles , phys . rep . * 110 * ( 1984 ) 1 ; h. haber and g. kane , phys . rep . * 117 * ( 1985 ) 75 .
the lep higgs working group , http://lephiggs.web.cern.ch/lephiggs/www/ ; lhwg note 2001 - 04 . h. baer , m. bisset , c. kao and x. tata , phys . rev .
d * 46 * , 1067 ( 1992 ) . v. d. barger , m. s. berger , a. l. stange and r. j. phillips , phys .
d * 45 * , 4128 ( 1992 ) ; j. f. gunion and l. h. orr , phys . rev .
d * 46 * , 2052 ( 1992 ) .
j. f. gunion , r. bork , h. e. haber and a. seiden , phys .
d * 46 * , 2040 ( 1992 ) .
z. kunszt and f. zwirner , nucl .
b * 385 * , 3 ( 1992 ) .
h. baer , m. bisset , d. dicus , c. kao and x. tata , phys .
d * 47 * , 1062 ( 1993 ) ; h. baer , m. bisset , c. kao and x. tata , phys .
d * 50 * , 316 ( 1994 ) .
h. baer , c. kao and x. tata , phys .
b * 303 * , 284 ( 1993 ) ;
s. abdullin , h. baer , c. kao , n. stepanov and x. tata , phys . rev .
d * 54 * , 6728 ( 1996 ) . c. kao and n. stepanov , phys .
d * 52 * , 5025 ( 1995 ) .
s. dawson , d. dicus and c. kao , phys .
b * 545 * , 132 ( 2002 ) .
s. dawson , d. dicus , c. kao and r. malhotra , phys .
lett . * 92 * , 241801 ( 2004 ) .
cms technical proposal , cern / lhcc 94 - 38 ( 1994 ) . atlas technical proposal , cern / lhcc 94 - 43 ( 1994 ) .
e. richter - was , d. froidevaux , f. gianotti , l. poggioli , d. cavalli and s. resconi , int .
j. mod .
a * 13 * , 1371 ( 1998 ) ; atlas detector and physics performance technical design report , cern / lhcc 99 - 14/15 ( 1999 ) .
r. w. brown and k. o. mikaelian , phys .
d * 19 * , 922 ( 1979 ) .
d. a. dicus , c. kao and w. w. repko , phys . rev .
d * 36 * , 1570 ( 1987 ) .
e. w. n. glover and j. j. van der bij , nucl .
b * 321 * , 561 ( 1989 ) . c. kao , phys .
d * 46 * , 4907 ( 1992 ) .
j. yin , w. g. ma , r. y. zhang and h. s. hou , phys .
d * 66 * , 095008 ( 2002 ) .
j. f. donoghue and l. f. li , phys .
d * 19 * , 945 ( 1979 ) ; l. j. hall and m. b. wise , nucl .
b * 187 * , 397 ( 1981 ) . c. kao
, g. lovelace and l. h. orr , phys .
b * 567 * , 259 ( 2003 ) .
j. pumplin , d. r. stump , j. huston , h. l. lai , p. nadolsky and w. k. tung , jhep * 0207 * , 012 ( 2002 ) .
j. a. m. vermaseren , s. a. larin and t. van ritbergen , phys .
b * 405 * , 327 ( 1997 ) .
w. j. marciano , phys .
d * 29 * , 580 ( 1984 ) .
f. maltoni , z. sullivan and s. willenbrock , phys .
d * 67 * , 093005 ( 2003 ) .
r. v. harlander and w. b. kilgore , phys .
d * 68 * , 013001 ( 2003 ) .
j. campbell _ et al .
_ , arxiv : hep - ph/0405302 . q. li , c. s. li , j. j. liu , l. g. jin and c. p. yuan , arxiv : hep - ph/0501070 .
madgraph , by t. stelzer and w.f .
long , comput .
. commun . * 81 * , 357 ( 1994 ) .
helas , by h. murayama , i. watanabe and k. hagiwara , kek report kek-91 - 11 ( 1992 )
. n. brown , z. phys .
* c49 * ( 1991 ) 657 .
j. f. gunion and h. e. haber , phys .
d * 67 * , 075019 ( 2003 ) . | we investigate the prospects of detecting a higgs pseudoscalar ( @xmath0 ) in association with a @xmath1 gauge boson produced from bottom quark fusion ( @xmath2 ) at the cern large hadron collider ( lhc ) .
a general two higgs doublet model and the minimal supersymmetric standard model are adopted to study the discovery potential of @xmath3 or @xmath4 , via @xmath2 with physics backgrounds and realistic cuts .
promising results are found for @xmath5 gev in a general two higgs doublet model when the heavier higgs scalar ( @xmath6 ) can decay into a @xmath1 boson and a higgs pseudoscalar ( @xmath7 ) .
we compare the production rates from bottom quark fusion ( @xmath2 ) and gluon fusion ( @xmath8 ) , and find that they are complementary processes to produce @xmath9 in hadron collisions . while gluon fusion is the major source for producing a higgs pseudoscalar associated with a @xmath1 boson at the lhc for @xmath10 , bottom quark fusion can make dominant contributions for @xmath11 . |
solar prominences or filaments are cool and dense plasmas embedded in the million - kelvin corona @xcite . the plasmas originate from the direct injection of chromospheric materials into a preexisting filament channel , levitation of chromospheric mass into the corona , or condensation of hot plasmas from the chromospheric evaporation due to the thermal instability @xcite .
prominences are generally believed to be supported by the magnetic tension force of the dips in sheared arcades @xcite or twisted magnetic flux ropes ( mfrs ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
they can keep stable for several weeks or even months , but may get unstable after being disturbed .
large - amplitude and long - term filament oscillations before eruption have been observed by spaceborne telescopes @xcite and reproduced by numerical simulations @xcite , which makes filament oscillation another precursor for coronal mass ejections ( cmes ; * ? ? ?
* ) and the accompanying flares .
when the twist of a flux rope supporting a filament exceeds the threshold value ( 2.5@[email protected]@xmath4 ) , it will also become unstable and erupt due to the ideal kink instability ( ki ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . however , whether the eruption of the kink - unstable flux rope becomes failed or ejective depends on how fast the overlying magnetic field declines with height @xcite .
when the decay rate of the background field exceeds a critical value , the flux rope will lose equilibrium and erupt via the so - called torus instability ( ti ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
on the other hand , if the confinement from the background field is strong enough , the filament will decelerate to reach the maximum height before falling back to the solar surface , which means the eruption is failed @xcite .
in addition to the successful and failed eruptions , there are partial filament eruptions @xcite . after examining 54 h@xmath3 prominence activities
, @xcite found that a majority of the eruptive prominences show separation of escaping material from the bulk of the prominence ; the latter initially lifted away from and then fell back to the solar surface . to explain the partial filament eruptions , the authors proposed a cartoon model in which magnetic reconnection occurs inside an inverse - polarity flux rope , leading to the separation of the escaping portion of the prominence and the formation of a second x - type neutral line in the upper portion of the prominence . the inner splitting and subsequent partial prominence eruption
is also observed by @xcite .
@xcite interpreted an active prominence with the process of vertical reconnection between an inverse - polarity flux rope and an underlying magnetic arcade .
@xcite reported a partial filament eruption characterised by a quasi - static , slow phase and a rapid kinking phase showing a bifurcation of the filament .
the separation of the filament , the extreme - ultravoilet ( euv ) brightening at the separation location , and the surviving sigmoidal structure provide convincing evidences that magnetic reconnection occurs within the body of filament @xcite . @xcite carried out three - dimensional ( 3d ) numerical simulations to model the partial expulsion of a mfr .
after multiple reconnections at current sheets that form during the eruption , the rope breaks in an upper , escaping rope and a lower , surviving rope .
the `` partially - expelled flux rope '' ( pefr ) model has been justified observationally @xcite .
@xcite observed a distinct coronal downflow following a curved path at the speed of @xmath5150 km s@xmath1 during a cme - associated prominence eruption .
their observation provides support for the pinching off of the field lines drawn - out by the erupting prominences and the contraction of the arcade formed by the reconnection .
similar multithermal downflow at the speed of @xmath2380 km s@xmath1 starting at the cusp - shaped structures where magnetic reconnection occurred inside the erupting flux rope that led to its bifurcation was reported by @xcite .
@xcite studied a flare - associated partial eruption of a double - decker filament .
@xcite found that a stable double - decker mfr system existed for hours prior to the eruption on 2012 july 12 .
after entering the domain of instability , the high - lying mfr impulsively erupted to generate a fast cme and _ goes _
x1.4 class flare ; while the low - lying mfr remained behind and continuously maintained the sigmoidicity of the active region ( ar ) . from the previous literatures , we can conclude that magnetic reconnection and the release of free energy involve in most of the partial filament eruptions .
however , the exact mechanism of partial eruptions , which is of great importance to understanding the origin of solar eruptions and forecasting space weather , remain unclear and controversial . in this paper
, we report multiwavelength observations of a partial filament eruption and the associated cme and m6.7 flare in noaa ar 11283 on 2011 september 8 .
the ar emerged from the eastern solar limb on 2011 august 30 and lasted for 14 days .
owing to its extreme complexity , it produced a couple of giant flares and cmes during its lifetime @xcite . in section [ s - data ] , we describe the data analysis using observations from the big bear solar observatory ( bbso ) , _ soho _ , _ solar dynamics observatory _
( _ sdo _ ) , _ solar terrestrial relation observatory _ ( _ stereo _ ; * ? ? ?
* ) , _ goes _ , _ reuven ramaty high - energy solar spectroscopic imager _
( _ rhessi _ ; * ? ? ?
* ) , and _
wind_. results and discussions are presented in section [ s - result ] and section [ s - disc ] .
finally , we draw our conclusion in section [ s - sum ] .
on september 8 , the dark filament residing in the ar was most clearly observed at h@xmath3 line center ( @xmath26563 ) by the ground - based telescope in bbso . during 15:30@xmath016:30 ut ,
the filament rose and split into two parts .
the major part lifted away and returned to the solar surface , while the runaway part separated from and escaped the major part , resulting in a very faint cme recorded by the _ soho _ large angle spectroscopic coronagraph ( lasco ; * ? ? ? * ) cme catalog .
the white light ( wl ) images observed by the lasco / c2 with field - of - view ( fov ) of 2@xmath06 solar radii ( @xmath6 ) were calibrated using the _ c2_calibrate.pro _ in the _ solar software _ ( _ ssw _ ) .
the partial filament eruption was clearly observed by the atmospheric imaging assembly ( aia ; * ? ? ?
* ) aboard _ sdo _ with high cadences and resolutions
. there are seven euv filters ( 94 , 131 , 171 , 193 , 211 , 304 , and 335 ) and two uv filters ( 1600 and 1700 ) aboard aia to achieve a wide temperature coverage ( @xmath7 ) .
the aia level_1 fits data were calibrated using the standard program _
aia_prep.pro_. the images observed in different wavelengths were coaligned carefully using the cross - correlation method .
to investigate the 3d magnetic configurations before and after the eruption , we employed the line - of - sight ( los ) and vector magnetograms from the helioseismic and magnetic imager ( hmi ; * ? ? ?
* ) aboard _
sdo_. the 180@xmath8 ambiguity of the transverse field was removed by assuming that the field changes smoothly at the photosphere @xcite .
we also performed magnetic potential field and non - linear force free field ( nlfff ) extrapolations using the optimization method as proposed by @xcite and as implemented by @xcite .
the fov for extrapolation was 558@xmath95@xmath10466@xmath92 to cover the whole ar and make sure the magnetic flux was balanced , and the data were binned by 2@xmath102 so that the resolution became 2@xmath11 .
the eruption was also captured from different perspectives by the extreme - ultraviolet imager ( euvi ) and cor1 coronagraph of the sun earth connection coronal and heliospheric investigation ( secchi ; * ? ? ?
* ) instrument aboard the ahead satellite ( _ sta _ hereafter ) and behind satellite ( _ stb _ hereafter ) of _ stereo_. the cor1 has a smaller fov of [email protected] @xmath6 compared with lasco / c2 , which is favorable for the detection of early propagation of cmes . on september 8 ,
the twin satellites ( _ sta _ and _ stb _ ) had separation angles of 103@xmath8 and 95@xmath8 with the earth .
the presence of open magnetic field lines within the ar was confirmed indirectly by the evidence of type slowromancap3@ burst in the radio dynamic spectra .
the spectra were obtained by the s / waves @xcite on board _ stereo _ and the waves instrument @xcite on board the _ wind _ spacecraft .
the frequency of s / waves ranges from 2.5 khz to 16.025 mhz .
the waves has two radio detectors : rad1 ( [email protected] mhz ) and rad2 ( [email protected] mhz ) .
the accompanying m6.7 flare was obviously identified in the _ goes _ soft x - ray ( sxr ) light curves in [email protected] and 1@xmath08 . to figure out where the accelerated nonthermal particles precipitate , we also made hard x - ray ( hxr ) images and light curves at different energy bands ( 3@xmath06 , 6@xmath012 , 12@xmath025 , 25@xmath050 , and 50@xmath0100 kev ) using the observations of _ rhessi_. the hxr images were generated using the clean method with integration time of 10 s. the observing parameters are summarized in table [ tbl-1 ] .
figure [ fig1 ] shows eight snapshots of the h@xmath3 images to illustrate the whole evolution of the filament ( see also the online movie animation1.mpg ) .
figure [ fig1](a ) displays the h@xmath3 image at 15:30:54 ut before eruption .
it is overlaid with the contours of the los magnetic field , where green ( blue ) lines stand for positive ( negative ) polarities .
the dark filament that is @xmath239 mm long resides along the polarity inversion line ( pil ) .
the top panels of figure [ fig2 ] demonstrate the top - view of the 3d magnetic configuration above the ar at the beginning and after eruption , with the los magnetograms located at the bottom boundary . using the same method described in @xcite
, we found a magnetic null point and the corresponding spine and separatrix surface .
the normal magnetic field lines are denoted by green lines .
the magnetic field lines around the outer / inner spine and the separatrix surface ( or arcade ) are represented by red / blue lines . beneath the null point ,
the sheared arcades supporting the filament are represented by orange lines .
the spine is rooted in the positive polarity ( p1 ) that is surrounded by the negative polarities ( n1 and pb ) .
it extends in the northeast direction and connects the null point with a remote place on the solar surface .
such magnetic configuration is quite similar to those reported by @xcite , @xcite , and @xcite .
as time goes on , the filament rose and expanded slowly ( figure [ fig1](b ) ) .
the initiation process is clearly revealed by the aia 304 observation ( see the online movie animation2.mpg ) .
figure [ fig3 ] shows eight snapshots of the 304 images .
initial brigtenings ( ib1 , ib2 , and ib3 ) appeared near the ends and center of the sigmoidal filament , implying that magnetic reconnection took place and the filament got unstable ( figure [ fig3](b)-(d ) ) .
such initial brightenings were evident in all the euv wavelengths .
with the intensities of the brigtenings increasing , the dark filament rose and expanded slowly , squeezing the overlying arcade field lines .
null - point magnetic reconnection might be triggered when the filament reached the initial height of the null point ( @xmath215 mm ) , leading to impulsive brightenings in h@xmath3 ( figure [ fig1](c)-(d ) ) and euv ( figure [ fig3](e)-(h ) ) wavelengths and increases in sxr and hxr fluxes ( figure [ fig4 ] ) .
the m6.7 flare entered the impulsive phase .
the bright and compact flare kernel pointed by the white arrow in figure [ fig1](c ) extended first westward and then northward , forming a quasi - circular ribbon at @xmath215:42 ut ( figure [ fig1](d ) ) , with the intensity contours of the hxr emissions at 12@xmath025 kev superposed .
there was only one hxr source associated with the flare , and the source was located along the flare ribbon with the strongest h@xmath3 emission , which is compatible with the fact that the footpoint hxr emissions come from the nonthermal bremsstrahlung of the accelerated high - energy electrons after penetrating into the chromosphere .
the flare demonstrates itself not only around the filament but also at the point - like brightening ( pb hereafter ) and the v - shape ribbon to the left of the quasi - circular ribbon . since the separatrix surface intersects with the photosphere at pb to the north and the outer spine intersects with the photosphere to the east ( figure [ fig2](a ) ) , it is believed that nonthermal electrons accelerated by the null - point magnetic reconnection penetrated into the lower atmosphere not only at the quasi - circular ribbon , but also at pb and the v - shape ribbon .
figure [ fig4 ] shows the sxr ( black solid and dashed lines ) and hxr ( colored solid lines ) light curves of the flare .
the sxr fluxes started to rise rapidly at @xmath215:32 ut and peaked at 15:45:53 ut for 1@xmath08 and 15:44:21 ut for [email protected] .
the hxr fluxes below 25 kev varied smoothly like the sxr fluxes , except for earlier peak times at @xmath215:43:10 ut .
the hxr fluxes above 25 kev , however , experienced two small peaks that imply precursor release of magnetic energy and particle acceleration at @xmath215:38:36 ut and @xmath215:41:24 ut and a major peak at @xmath215:43:10 ut . the time delay between the sxr and hxr peak times implies the possible neupert effect for this event @xcite .
the main phase of the flare sustained until @xmath217:00 ut , indicating that the flare is a long - duration event . during the flare , the filament continued to rise and split into two branches at the eastern leg around 15:46 ut ( figure [ fig1](e ) ) , the right of which is thicker and darker than the left one .
such a process is most clearly revealed by the aia 335 observation ( see the online movie animation3.mpg ) .
figure [ fig5 ] displays eight snapshots of the 335 images .
it is seen that the dark filament broadened from @xmath215:42:30 ut and completely split into two branches around 15:45:51 ut .
we define the left and right branches as the runaway part and major part of the filament .
the two interwinding parts also underwent rotation ( panels ( d)-(h ) ) .
meanwhile , the plasma of the runaway part moved in the northwest direction and escaped . to illustrate the rotation , we derived the time - slice diagrams of the two slices ( s4 and s5 in panel ( f ) ) that are plotted in figure [ fig6 ] .
the upper ( lower ) panels represent the diagrams of s4 ( s5 ) , and the left ( right ) panels represent the diagrams for 211 ( 335 ) .
@xmath12 in the diagrams stands for the southwest endpoints of the slices .
the filament began to split into two parts around 15:42:30 ut , with the runaway part rotating round the eastern leg of the major part for @xmath21 turn until @xmath215:55 ut . during the eruption , the runaway branch of the filament disappeared ( figure [ fig1](f ) ) .
the major part , however , fell back to the solar surface after reaching the maximum height around 15:51 ut , suggesting that the eruption of the major part of the filament was failed .
the remaining filament after the flare was evident in the h@xmath3 image ( figure [ fig1](h ) ) .
nlfff modelling shows that the magnetic topology was analogous to that before the flare , with the height of the null point slightly increased by 0.4 mm ( figure [ fig2](b ) ) .
figure [ fig7 ] shows six snapshots of the 171 images .
the rising and expanding filament triggered the m - class flare and the kink - mode oscillation of the adjacent large - scale coronal loops within the same ar ( see the online movie animation4.mpg ) . with the filament increasing in height ,
part of its material was ejected in the northwest direction represented by `` s1 '' in panel ( c ) . after reaching the maximum height at @xmath215:51:12 ut , the major part of the filament returned to the solar surface . the bright cusp - like post - flare loops ( pfls ) in the main phase of the flare
are clearly observed in all the euv filters , see also figure [ fig7](f ) . to illustrate the eruption and
loop oscillation more clearly , we extracted four slices .
the first slice , s0 in figure [ fig1](f ) and figure [ fig7](d ) , is 170 mm in length .
it starts from the flare site and passes through the apex of the major part of the filament .
the time - slice diagram of s0 in h@xmath3 is displayed in figure [ fig8](a ) .
the filament started to rise rapidly at @xmath215:34:30 ut with a constant speed of @xmath2117 km s@xmath1 .
after reaching the peak height ( @xmath13 ) of @xmath2115 mm at @xmath215:51 ut , it fell back to the solar surface in a bumpy way until @xmath216:30 ut .
using a linear fitting , we derived the average falling speed ( @xmath222 km s@xmath1 ) of the filament in h@xmath3 wavelength .
the time - slice diagram of s0 in uv and euv passbands are presented in figure [ fig9 ] .
we selected two relatively hot filters ( 335 and 211 in the top panels ) , two warm filters ( 171 and 304 in the middle panels ) , and two cool filters ( 1600 and 1700 in the bottom panels ) , respectively .
similar to the time - slice diagram in h@xmath3 ( figure [ fig8](a ) ) , the filament rose at apparent speeds of 92@xmath0151 km s@xmath1 before felling back in an oscillatory way at the speeds of 34@xmath046 km s@xmath1 during 15:51@xmath016:10 ut and @xmath271 km s@xmath1 during 16:18@xmath016:30 ut ( figure [ fig9](a)-(d ) ) .
the falling speeds in uv wavelengths are @xmath278 km s@xmath1 during 15:51@xmath016:10 ut ( figure [ fig9](e)-(f ) ) .
the times when the major part of the filament reached maximum height in uv and euv passbands , @xmath215:51 ut , are consistent with that in h@xmath3 .
the later falling phase during 16:18@xmath016:30 ut is most obvious in the warm filters .
the downflow of the surviving filament in an oscillatory way was also observed during a sympathetic filament eruption @xcite . owing to the lower time cadence of bbso than aia
, the escaping process of the runaway part of the filament in figure [ fig1](e)-(f ) was detected by aia .
we extracted another slice s1 that is 177 mm in length along the direction of ejection ( figure [ fig7](c ) ) .
@xmath12 mm and @xmath14 mm represent the southeast and northwest endpoints of the slice .
the time - slice diagram of s1 in 171 is displayed in figure [ fig8](b ) .
contrary to the major part , the runaway part of the filament escaped successfully from the corona at the speeds of 125@xmath0255 km s@xmath1 without returning to the solar surface . the intermittent runaway process during 15:45@xmath016:05 ut
was obviously observed in most of the euv filters .
we extracted another slice s2 that also starts from the flare site and passes through both parts of the filament ( figure [ fig7](d ) ) .
the time - slice diagram of s2 in 171 is drawn in figure [ fig8](c ) . as expected ,
the diagram features the bifurcation of the filament as pointed by the white arrow , i.e. , the runaway part escaped forever while the major part moved on after bifurcation and finally fell back .
the eruption of the filament triggered transverse kink oscillation of the adjacent coronal loops ( ol in figure [ fig7](a ) ) .
the direction of oscillation is perpendicular to the initial loop plane ( see the online movie animation4.mpg ) .
we extracted another slice s3 that is 80 mm in length across the oscillating loops ( figure [ fig7](b ) ) .
@xmath12 mm and @xmath15 mm represent the northwest and southeast endpoints of the slice . the time - slice diagram of s3 in 171
is shown in figure [ fig8](d ) , where the oscillation pattern during 15:38@xmath015:47 ut is evidently demonstrated .
the ol moved away from the flare site during 15:38@xmath015:41 ut before returning to the initial position and oscillating back and forth for @xmath22 cycles . by fitting the pattern with a sinusoidal function as marked by the white dashed line ,
the resulting amplitude and period of the kink oscillation were @xmath21.6 mm and @xmath2225 s. we also extracted several slices across the ol and derived the time - slice diagrams , finding that the coronal loops oscillated in phase and the mode was fundamental .
the initial velocity amplitude of the oscillation was @xmath244.7 km s@xmath1 .
the speed of propagation of the mode @xmath16 , where @xmath17 is the loop length , @xmath18 is the period , @xmath19 is the alfvn wave speed , and @xmath20 and @xmath21 are the plasma densities inside and outside the loop @xcite . in figure [ fig7](a ) , we denote the footpoints of the ol with black crosses that are 106.1 mm away . assuming a semi - circular shape , the length of the loop @xmath22 mm and @xmath23 km s@xmath1 .
using the same value of @xmath24 , we derived @xmath25 km s@xmath1 .
in addition , we estimated the electron number density of the ol to be @xmath22.5@xmath26 @xmath27 based on the results of nlfff extrapolation in figure [ fig2](a ) .
the kink - mode oscillation of the loops was best observed in 171 , indicating that the temperatures of loops were @xmath20.8 mk .
the escaping part of the filament was also clearly observed by _ sta_/euvi .
figure [ fig10 ] shows six snapshots of the 304 images , where the white arrows point to the escaping filament . during 15:46@xmath016:30 ut ,
the material moved outwards in the northeast direction without returning to the solar surface . the bright m6.7 flare pointed by the black arrows
is also quite clear .
the runaway part of the filament resulted in a very faint cme observed by the wl coronagraphs .
figure [ fig11](a)-(d ) show the running - difference images of _
sta_/cor1 during 16:00@xmath016:15 ut . as pointed by the arrows , the cme first appeared in the fov of _ sta_/cor1 at @xmath216:00 ut and propagated outwards at a nearly constant speed , with the contrast between cme and the background decreasing as time goes on .
the propagation direction of the cme is consistent with that of the runaway filament in figure [ fig10 ] .
figure [ fig11](e)-(f ) show the running - difference images of lasco / c2 during 16:36@xmath016:48 ut .
the faint blob - like cme first appeared in the fov of c2 at @xmath216:36 ut and propagated in the same direction as that of the escaping filament observed by aia in figure [ fig7](c ) .
the central position angle and angular width of the cme observed by c2 are 311@xmath8 and 37@xmath8 .
the linear velocity of the cme is @xmath2214 km s@xmath1 .
the time - height profiles of the runaway filament observed by _
sta_/euvi ( _ boxes _ ) and the corresponding cme observed by _ sta_/cor1 ( _ diamonds _ ) and lasco / c2 ( _ stars _ ) are displayed in figure [ fig12 ] .
the apparent propagating velocities represented by the slopes of the lines are 60 , 358 , and 214 km s@xmath1 , respectively . taking the projection effect into account ,
the start times of the filament eruption and the cme observed by lasco / c2 and _ sta_/cor1 from the lower corona ( @xmath28 ) are approximately coincident with each other . in the cdaw catalog ,
the preceding and succeeding cmes occurred at 06:12 ut and 18:36 ut on september 8 . in the cor1 cme catalog , the preceding and succeeding cmes occurred slightly earlier at 05:45 ut and 18:05 ut on the same day , which is due to the smaller fov of cor1 than lasco / c2 .
therefore , the runaway part of the filament was uniquely associated with the cme during 16:00@xmath018:00 ut .
then , a question is raised : how can the runaway part of the filament successfully escape from the corona and give rise to a cme ?
we speculate that open magnetic field lines provide a channel . in order to justify the speculation
, we turn to the large - scale magnetic field calculated by the potential field source surface ( pfss ; * ? ? ?
* ; * ? ? ?
* ) modelling and the radio dynamic spectra from the s / waves and waves instruments . in figure
[ fig13 ] , we show the magnetic field lines whose footpoints are located in ar 11283 at 12:04 ut before the onset of flare / cme event .
the open and closed field lines are represented by the purple and white lines .
it is clear that open field lines do exist in the ar and their configuration accords with the directions of the escaping part of filament observed by aia and the cme observed by c2 .
the radio dynamic spectra from s / waves and waves are displayed in panels ( a)@xmath0(b ) and ( c)@xmath0(d ) of figure [ fig14 ] , respectively .
there are clear signatures of type slowromancap3@ radio burst in the spectra . for _ sta
_ , the burst started at @xmath215:38:30 ut and ended at @xmath216:00 ut , during which the frequency drifted rapidly from 16 mhz to @xmath20.3 mhz . for _ stb _ that was @xmath20.07 au further than _ sta _ from the sun , the burst started slightly later by @xmath22 minutes with the frequency drifting from @xmath24.1 mhz to @xmath20.3 mhz since the early propagation of the filament was blocked by the sun . for waves ,
the burst started at @xmath215:39:30 ut and ended at @xmath216:00 ut with the frequency drifting from 13.8 mhz to @xmath20.03 mhz .
the starting times of the radio burst were consistent with the hxr peak times of the flare . since the type slowromancap3@ radio emissions result from the cyclotron maser instability of the nonthermal electron beams that are accelerated and ejected into the interplanetary space along open magnetic field lines during the flare @xcite ,
the type slowromancap3@ radio burst observed by _
stereo _ and _ wind _ provides indirect and supplementary evidence that open magnetic field lines exist near the flare site .
it is widely accepted that the solar eruptions result from the release of magnetic free energy . for this event
, we studied how the energy is accumulated by investigating the magnetic evolution of the ar using the hmi los magnetograms ( see the online movie animation5.mpg ) .
figure [ fig15 ] displays four snapshots of the magnetograms , where the ar is dominated by negative polarity ( n1 ) .
a preexisting positive polarity ( p1 ) is located in the northeast direction . from the movie
, we found continuous shearing motion along the highly fragmented and complex pil between n1 and p1 .
for example , the small negative region n2 at the boundary of the sunspot was dragged westward and became elongated ( figure [ fig15](b)-(d ) ) . to better illustrate the motion
, we derived the transverse velocity field ( @xmath29 , @xmath30 ) at the photosphere using the differential affine velocity estimator ( dave ) method @xcite .
the cadence of the hmi los magnetograms was lowered from 45 s to 180 s. figure [ fig16 ] displays six snapshots of the magnetograms overlaid with the transverse velocity field represented by the white arrows .
the velocity field is clearly characterized by the shearing motions along the pil .
the regions within the green and blue elliptical lines are dominated by eastward and westward motions at the speeds of @xmath21.5 km s@xmath1 . from the online movie ( animation6.mpg )
, we can see that the continuous shearing motions were evident before the flare , implying that the magnetic free energy and helicity were accumulated and stored before the impulsive release .
once the free energy of the ar is accumulated to a critical value , chances are that the filament constrained by the overlying magnetic field lines undergoes an eruption .
several types of triggering mechanism have been proposed .
one type of processes where magnetic reconnection is involved include the flux emergence model @xcite , catastrophic model @xcite , tether - cutting model @xcite , and breakout model @xcite , to name a few .
another type is the ideal magnetohydrodynamic ( mhd ) processes as a result of ki @xcite and/or ti @xcite . from figure [ fig15 ] and the movie ( animation5.mpg )
, we can see that before the flare there was continuous magnetic flux emergence ( p2 , p3 , and p4 ) and subsequent magnetic cancellation along the fragmented pil .
we extracted a large region within the white dashed box of figure [ fig15](d ) and calculated the total positive ( @xmath31 ) and negative ( @xmath32 ) magnetic fluxes within the box . in figure
[ fig17 ] , the temporal evolutions of the fluxes during 11:00@xmath016:30 ut are plotted , with the evolution of @xmath31 divided into five phases ( i@xmath0v ) separated by the dotted lines . the first four phases before the onset of flare at 15:32 ut are characterized by quasi - periodic and small - amplitude magnetic flux emergence and cancellation , implying that the large - scale magnetic field was undergoing rearrangement before the flare .
the intensity contours of the 304 images in figure [ fig3](b ) and ( d ) are overlaid on the magnetograms in figure [ fig15](b ) and ( c ) , respectively .
it is clear that the initial brightenings ib1 and ib2 are very close to the small positive polarities p4 and p3 .
there is no significant magnetic flux emergence around ib3 . in the emerging - flux - induced - eruption model @xcite ,
when reconnection - favorable magnetic bipole emerges from beneath the photosphere into the filament channel , it reconnects with the preexisting magnetic field lines that compress the inverse - polarity mfr .
the small - scale magnetic reconnection and flux cancellation serve as the precursor for the upcoming filament eruption and flare . during the flare
when magnetic reconnection occurred between 15:32 ut and 16:10 ut , both the positive and negative magnetic field experienced impulsive and irreversible changes . despite that the flux emergences are plausible to interpret the triggering mechanism
, there is another possibility . in the tether - cutting model @xcite , a pair of @xmath33-shape sheared arcades that comprise a sigmoid reconnect when the two elbows come into contact , forming a short loop and a long mfr .
whether the mfr experiences a failed or ejective eruption depends on the strength of compression from the large - scale background field .
the initial brightenings ( ib1 , ib2 , and ib3 ) around the sigmoidal filament might be the precursor brightenings as a result of internal tether - cutting reconnection due to the continuous shearing motion along the pil .
after onset , the whole flux system erupted and produced the m - class flare . considering that the magnetic configuration could not be modelled during the flare , we are not sure whether a coherent mfr was formed after the initiation @xcite . compared to the flux emergences
, the internal tether - cutting seems more believable to interpret how the filament eruption was triggered for the following reasons .
firstly , the filament was supported by sheared arcade .
secondly , there were continuous shearing motions along the pil , and the directions were favorable for the tether - cutting reconnection . finally , the initial brightenings ( ib1 , ib2 , and ib3 ) around the filament in figure [ fig3 ] fairly match the internal tether - cutting reconnection with the presence of multiple bright patches of flare emission in the chromosphere at the feet of reconnected field lines , while there was no flux emergence around ib3 .
nlfff modelling shows that the twist number ( @xmath21 ) of the sheared arcades supporting the filament is less than the threshold value ( @xmath21.5 ) , implying that the filament eruption may not be triggered by ideal ki .
the photospheric magnetic field of the ar features a bipole ( p1 and n1 ) and a couple of mini - polarities ( e.g. , p2 , p3 , p4 , and n2 ) .
therefore , the filament eruption could not be explained by the breakout model that requires quadrupolar magnetic field , although null - point magnetic reconnection took place above the filament during the eruption . after the onset of eruption , the filament split into two parts as described in section [ s - result ] .
how the filament split is still unclear . in the previous literatures , magnetic reconnection
is involved in the split in most cases @xcite . in this study ,
the split occurred during the impulsive phase of the flare at the eastern leg that was closer to the flare site than the western one , implying that the split was associated with the release of magnetic energy .
the subsequent rotation or unwinding motion implies the release of magnetic helicity stored in the filament before the flare , presumably due to the shearing motion in the photosphere . nevertheless , it is still elusive whether the filament existed as a whole or was composed of two interwinding parts before splitting .
the way of splitting seems difficult to be explained by any of the previous models and requires in - depth investigations .
though the runaway part escaped out of the corona , the major part failed .
it returned to the solar surface after reaching the apex .
such kind of failed eruptions have been frequently observed and explained by the strapping effect of the overlying arcade @xcite or asymmetry of the background magnetic fields with respect to the location of the filament @xcite . in order to figure out the cause of failed eruption of the major part
, we turn to the large - scale magnetic configurations displayed in the bottom panels of figure [ fig2 ] .
it is revealed that the overlying magnetic arcades above ar 11283 are asymmetric to a great extent , i.e. , the magnetic field to the west of ar is much stronger than that to the east , which is similar to the case of @xcite . according to the analysis of @xcite , the confinements of the large - scale arcade acted on the filament are strong enough to prevent it from escaping .
we also performed magnetic potential - field extrapolation using the same boundary and derived the distributions of @xmath34 above the pil .
it is found that the maximum height of the major part considerably exceeds the critical height ( @xmath280@xmath11 ) of ti where the decay index ( @xmath35 ) of the background potential field reaches @xmath21.5 .
the major part would have escaped from the corona successfully after entering the instability domain if ti had worked .
therefore , the asymmetry with respect to the filament location , rather than ti of the overlying arcades , seems reasonable and convincing to interpret why the major part of the filament underwent failed eruption . in this study , both successful and failed eruptions occurred in a partially eruptive event , which provides more constraints to the theoretical models of solar eruptions . since the first discovery of coronal loop oscillations during flares @xcite
, such kind of oscillations are found to be ubiquitous and be useful for the diagnostics of coronal magnetic field @xcite . owing to the complex interconnections of the magnetic field lines , blast wave and/or euv wave induced by filament eruption
may disturb the adjacent coronal loops in the same ar or remote loops in another ar , resulting in transverse kink - mode oscillations .
@xcite observed decaying and decayless transverse oscillations of a coronal loop on 2012 may 30 .
the loops experience small - amplitude decayless oscillations , which is driven by an external non - resonant harmonic driver before and after the flare @xcite .
the flare , as an impulsive driver , triggers large - amplitude decaying loop oscillations . in our study , the decayless loop oscillation with moderate amplitude ( @xmath21.6 mm ) occurred during the flare and lasted for only two cycles , which makes it quite difficult to precisely measure the decay timescale if it is decaying indeed .
the loop may cool down and become invisible in 171 while oscillating .
considering that the distance between the flare and ol is @xmath250 mm and the time delay between the flare onset and loop oscillation is @xmath26 minutes , the speed of propagation of the disturbances from the flare to ol is estimated to be @xmath2140 km s@xmath1 , which is close to the local sound speed of the plasmas with temperature of @xmath20.8 mk .
hence , we suppose that the coronal loop oscillation was triggered by the external disturbances as a result of the rising and expanding motions of the filament . flares and cmes play a very important role in the generation of space weather . accurate prediction of space weather is of great significance .
successful eruptions have substantially been observed and deeply investigated .
partial filament eruptions that produce flares and cmes , however , are rarely detected and poorly explored . for the type of partial eruptions in this study ,
i.e. , one part undergoes failed eruption and the other part escapes out of the corona , it would be misleading and confusing to assess and predict the space weather effects based on the information only from the solar surface , since the escaping part may carry or produce solar energetic particles that have potential geoeffectiveness .
complete observations are necessary for accurate predictions .
using the multiwavelength observations from both spaceborne and ground - based telescopes , we studied in detail a partial filament eruption event in ar 11283 on 2011 september 8 .
the main results are summarized as follows : 1 .
a magnetic null point was found above the preexisting positive polarity surrounded by negative polarities in the ar .
a spine passed through the null and intersected with the photosphere to the left . weakly twisted sheared arcade supporting the filament was located under the null point whose height increased slightly by @xmath20.4 mm after the eruption .
2 . the filament rose and expanded , which was probably triggered by the internal tether - cutting reconnection or by continuous magnetic flux emergence and cancellation along the highly complex and fragmented pil , the former of which seems more convincing . during its eruption , it triggered the null - point magnetic reconnection and the m6.7 flare with a single hxr source at different energy bands .
the flare produced a quasi - circular ribbon and a v - shape ribbon where the outer spine intersects with the photosphere .
during the expansion , the filament split into two parts at the eastern leg that is closer to the flare site .
the major part of the filament rose at the speeds of 90@xmath0150 km s@xmath1 before reaching the maximum apparent height of @xmath2115 mm .
afterwards , it returned to the solar surface staggeringly at the speeds of 20@xmath080 km s@xmath1 .
the rising and falling motions of the filament were clearly observed in the uv , euv , and h@xmath3 wavelengths .
the failed eruption of the major part was most probably caused by the asymmetry of the overlying magnetic arcades with respect to the filament location .
4 . the runaway part , however , separated from and rotated around the major part for @xmath21 turn before escaping outward from the corona at the speeds of 125@xmath0255 km s@xmath1 , probably along the large - scale open magnetic field lines as evidenced by the pfss modelling and the type slowromancap3@ radio burst .
the ejected part of the filament led to a faint cme . the angular width and apparent speed of the cme in the fov of c2
are 37@xmath8 and 214 km s@xmath1 .
the propagation directions of the escaping filament observed by sdo / aia and _ sta_/euvi are consistent with those of the cme observed by lasco / c2 and _ sta_/cor1 , respectively .
the partial filament eruption also triggered transverse oscillation of the neighbouring coronal loops in the same ar .
the amplitude and period of the kink - mode oscillation were 1.6 mm and 225 s. we also performed diagnostics of the plasma density and temperature of the oscillating loops .
the authors thank the referee for valuable suggestions and comments to improve the quality of this article .
we gratefully acknowledge y. n. su , p. f. chen , j. zhang , b. kliem , r. liu , s. gibson , h. gilbert , m. d. ding , and h. n. wang for inspiring and constructive discussions .
_ sdo _ is a mission of nasas living with a star program .
aia and hmi data are courtesy of the nasa/_sdo _ science teams .
_ stereo_/secchi data are provided by a consortium of us , uk , germany , belgium , and france .
qmz is supported by youth fund of jiangsu bk20141043 , by 973 program under grant 2011cb811402 , and by nsfc 11303101 , 11333009 , 11173062 , 11473071 , and 11221063
. h. ji is supported by the strategic priority research program@xmath0the emergence of cosmological structures of the chinese academy of sciences , grant no .
yg is supported by nsfc 11203014 .
li feng is supported by the nfsc grant 11473070 , 11233008 and by grant bk2012889 .
li feng also thanks the youth innovation promotion association , cas , for the financial support .
amari , t. , canou , a. , & aly , j .- j .
2014 , , 514 , 465 antiochos , s. k. , devore , c. r. , & klimchuk , j. a. 1999 , , 510 , 485 aschwanden , m. j. , fletcher , l. , schrijver , c. j. , & alexander , d. 1999 , , 520 , 880 aschwanden , m. j. 2011 , living reviews in solar physics , 8 , 5 bi , y. , jiang , y. , yang , j. , et al . 2014 , , 790 , 100 bougeret , j .- l . , kaiser , m. l. , kellogg , p. j. , et al . 1995 , , 71 , 231 bougeret , j. l. , goetz , k. , kaiser , m. l. , et al
2008 , , 136 , 487 brueckner , g. e. , howard , r. a. , koomen , m. j. , et al .
1995 , , 162 , 357 chen , p. f. 2011 , living reviews in solar physics , 8 , 1 chen , p. f. , & shibata , k. 2000 , , 545 , 524 chen , p. f. , innes , d. e. , & solanki , s. k. 2008 , , 484 , 487 chen , h. , zhang , j. , cheng , x. , et al .
2014 , , 797 , ll15 cheng , x. , zhang , j. , saar , s. h. , & ding , m. d. 2012 , , 761 , 62 cheng , x. , ding , m. d. , zhang , j. , et al . 2014a , , 789 , l35 cheng , x. , ding , m. d. , zhang , j. , et al .
2014b , , 789 , 93 dai , y. , ding , m. d. , & guo , y. 2013 , , 773 , l21 fan , y. 2005 , , 630 , 543 feng , l. , wiegelmann , t. , su , y. , et al .
2013 , , 765 , 37 gibson , s. e. , & fan , y. 2006a , , 637 , l65 gibson , s. e. , & fan , y. 2006b , journal of geophysical research ( space physics ) , 111 , 12103 gilbert , h. r. , holzer , t. e. , burkepile , j. t. , & hundhausen , a. j. 2000 , , 537 , 503 gilbert , h. r. , holzer , t. e. , & burkepile , j. t. 2001 , , 549 , 1221 gilbert , h. r. , alexander , d. , & liu , r. 2007 , , 245 , 287 guo , y. , ding , m. d. , schmieder , b. , et al . 2010a , , 725 , l38 guo , y. , schmieder , b. , dmoulin , p. , et al .
2010b , , 714 , 343 guo , y. , dmoulin , p. , schmieder , b. , et al .
2013 , , 555 , a19 guo , y. , erdlyi , r. , srivastava , a. k. , et al .
2015 , , 799 , 151 hood , a. w. , & priest , e. r. 1981 , geophysical and astrophysical fluid dynamics , 17 , 297 howard , r. a. , moses , j. d. , vourlidas , a. , et al .
2008 , , 136 , 67 ji , h. , wang , h. , schmahl , e. j. , moon , y .- j .
, & jiang , y. 2003 , , 595 , l135 jiang , c. , feng , x. , wu , s. t. , & hu , q. 2013 , , 771 , l30 jiang , c. , wu , s. t. , feng , x. , & hu , q. 2014 , , 780 , 55 joshi , n. c. , srivastava , a. k. , filippov , b. , et al .
2013 , , 771 , 65 joshi , n. c. , srivastava , a. k. , filippov , b. , et al .
2014 , , 787 , 11 kaiser , m. l. 2005 , advances in space research , 36 , 1483 keppens , r. , & xia , c. 2014 , , 789 , 22 kliem , b. , titov , v. s. , trk , t. 2004 , , 413 , l23 kliem , b. , trk , t. 2006 , physical review letters , 96 , 255002 kumar , p. , srivastava , a. k. , filippov , b. , & uddin , w. 2010 , , 266 , 39 kumar , p. , srivastava , a. k. , filippov , b. , erdlyi , r. , & uddin , w. 2011 , , 272 , 301 kumar , p. , cho , k .- s . , bong , s .- c . , park , s .- h .
, & kim , y. h. 2012 , , 746 , 67 lemen , j. r. , title , a. m. , akin , d. j. , et al .
2012 , , 275 , 17 li , t. , & zhang , j. 2012 , , 760 , l10 li , y. , ding , m. d. , guo , y. , & dai , y. 2014 , , 793 , 85 lin , r. p. , dennis , b. r. , hurford , g. j. , et al .
2002 , , 210 , 3 lin , j. , & forbes , t. g. 2000 , , 105 , 2375 liu , r. , alexander , d. , & gilbert , h. r. 2007 , , 661 , 1260 liu , y. 2008 , , 679 , l151 liu , r. , gilbert , h. r. , alexander , d. , & su , y. 2008 , , 680 , 1508 liu , y. , su , j. , xu , z. , et al .
2009 , , 696 , l70 liu , r. , kliem , b. , trk , t. , et al .
2012 , , 756 , 59 liu , c. , deng , n. , lee , j. , et al .
2014 , , 795 , 128 mackay , d. h. , karpen , j. t. , ballester , j. l. , schmieder , b. , & aulanier , g. 2010 , , 151 , 333 mandrini , c. h. , schmieder , b. , dmoulin , p. , guo , y. , & cristiani , g. d. 2014 , , 289 , 2041 moore , r. l. , sterling , a. c. , hudson , h. s. , & lemen , j. r. 2001 , , 552 , 833 murawski , k. , solovev , a. , kraskiewicz , j. , & srivastava , a. k. 2014 , arxiv:1411.7465 nakariakov , v. m. , ofman , l. , deluca , e. e. , roberts , b. , & davila , j. m. 1999 , science , 285 , 862 nakariakov , v. m. , & ofman , l. 2001 , , 372 , l53 ning , z. , & cao , w. 2010 , , 264 , 329 nistic , g. , nakariakov , v. m. , & verwichte , e. 2013 , , 552 , aa57 ruan , g. , chen , y. , wang , s. , et al .
2014 , , 784 , 165 schatten , k. h. , wilcox , j. m. , & ness , n. f. 1969 , , 6 , 442 scherrer , p. h. , schou , j. , bush , r. i. , et al .
2012 , , 275 , 207 schrijver , c. j. , & de rosa , m. l. 2003 , , 212 , 165 schuck , p. w. 2005 , , 632 , l53 shen , y. , liu , y. , & su , j. 2012 , , 750 , 12 shen , y. , liu , y. d. , chen , p. f. , & ichimoto , k. 2014 , , 795 , 130 song , h. q. , zhang , j. , cheng , x. , et al .
2014 , , 784 , 48 srivastava , a. k. , zaqarashvili , t. v. , kumar , p. , & khodachenko , m. l. 2010 , , 715 , 292 su , y. , & van ballegooijen , a. 2012 , , 757 , 168 sun , x. , hoeksema , j. t. , liu , y. , et al . 2012a , , 748 , 77 sun , x. , hoeksema , j. t. , liu , y. , chen , q. , & hayashi , k. 2012b , , 757 , 149 tang , j. f. , wu , d. j. , & tan , c. m. 2013 , , 779 , 83 terradas , j. , soler , r. , luna , m. , oliver , r. , & ballester , j. l. 2015 , , 799 , 94 trk , t. , kliem , b. , & titov , v. s. 2004 , , 413 , l27 trk , t. , & kliem , b. 2005 , , 630 , l97 trk , t. , berger , m. a. , & kliem , b. 2010 , , 516 , a49 tripathi , d. , solanki , s. k. , schwenn , r. , et al .
2006 , , 449 , 369 tripathi , d. , solanki , s. k. , mason , h. e. , & webb , d. f. 2007 , , 472 , 633 tripathi , d. , gibson , s. e. , qiu , j. , et al .
2009 , , 498 , 295 tripathi , d. , reeves , k. k. , gibson , s. e. , srivastava , a. , & joshi , n. c. 2013 , , 778 , 142 wheatland , m. s. , sturrock , p. a. , & roumeliotis , g. 2000 , , 540 , 1150 white , r. s. , & verwichte , e. 2012 , , 537 , aa49 wiegelmann , t. 2004 , , 219 , 87 xia , c. , chen , p. f. , & keppens , r. 2012 , , 748 , l26 xia , c. , chen , p. f. , keppens , r. , & van marle , a. j. 2011 , , 737 , 27 xia , c. , keppens , r. , antolin , p. , & porth , o. 2014a , , 792 , l38 xia , c. , keppens , r. , & guo , y. 2014b , , 780 , 130 zhang , j. , cheng , x. , & ding , m .- d . 2012a , nature communications , 3 , 747 zhang , q. m. , chen , p. f. , xia , c. , & keppens , r. 2012b , , 542 , a52 zhang , q. m. , chen , p. f. , guo , y. , fang , c. , & ding , m. d. 2012c , , 746 , 19 zhang , q. m. , chen , p. f. , xia , c. , keppens , r. , & ji , h. s. 2013 , , 554 , a124 zhou , y .- h . ,
chen , p .- f .
, zhang , q .- m . , & fang , c. 2014 , research in astronomy and astrophysics , 14 , 581 | in this paper , we report our multiwavelength observations of a partial filament eruption event in noaa active region 11283 on 2011 september 8 . a magnetic null point and the corresponding spine and separatrix surface
are found in the active region . beneath the null point ,
a sheared arcade supports the filament along the highly complex and fragmented polarity inversion line . after being activated , the sigmoidal filament erupted and split into two parts .
the major part rose at the speeds of 90@xmath0150 km s@xmath1 before reaching the maximum apparent height of @xmath2115 mm .
afterwards , it returned to the solar surface in a bumpy way at the speeds of 20@xmath080 km s@xmath1 .
the rising and falling motions were clearly observed in the extreme - ultravoilet ( euv ) , uv , and h@xmath3 wavelengths .
the failed eruption of the main part was associated with an m6.7 flare with a single hard x - ray source .
the runaway part of the filament , however , separated from and rotated around the major part for @xmath21 turn at the eastern leg before escaping from the corona , probably along large - scale open magnetic field lines .
the ejection of the runaway part resulted in a very faint coronal mass ejection ( cme ) that propagated at an apparent speed of 214 km s@xmath1 in the outer corona .
the filament eruption also triggered transverse kink - mode oscillation of the adjacent coronal loops in the same ar .
the amplitude and period of the oscillation were 1.6 mm and 225 s. our results are important for understanding the mechanisms of partial filament eruptions and provide new constraints to theoretical models .
the multiwavelength observations also shed light on space weather prediction . online - only material : animations , color figures |
our understanding of gamma - ray bursts ( grbs ) has improved tremendously in the last ten years .
observations have firmly established that grbs and their afterglows arise from highly relativistic ( taylor et al 2004 ) , collimated outflows ( frail et al 2001 , panaitescu & kumar 2002 ) . at least a few long - duration grbs are associated with the collapse of massive stars ( galama et al 1998 , della valle et al 2003 , hjorth et al 2003 , malesani et al 2004 ) , but less than a few percent of supernovae of type ib / c give rise to grbs ( soderberg et al 2004 ) .
a fraction of short - duration grbs are associated with old stellar populations ( nakar 2007 ) . among the fundamental questions that remain unanswered
is the generation of @xmath3-rays and the composition of the relativistic jet ( baryonic , e@xmath4 and/or magnetic ) .
the simultaneous monitoring of the optical and @xmath3-ray emissions of grb 080319b can provide insight into the mechanism for @xmath3-ray generation , as described in the next section .
studies of the multiwavelength prompt emission are still incipient because of the rarity of a good temporal coverage of the optical counterpart .
so far , the prompt optical emission was observed @xmath5 to be decoupled from that at @xmath3-rays ( grb 990123 akerlof et al 1999 ) , @xmath6 to contain a fluctuating component that tracks the burst and a smooth one peaking after the burst ( grb 050820a vestrand et al 2006 ) , or @xmath7 to fluctuate in phase with the burst ( grb 041219a vestrand et al 2005 ) .
grb 080319b is of the last type .
grb 080319b was the brightest burst detected by swift .
it was also seen by the konus satellite , which measured its fluence in the 20 kev 7 mev band , @xmath8 erg @xmath9 , corresponding to and isotropic energy release of @xmath10 erg ( golenetskii et al 2008 ) , for the burst redshift @xmath11 ( vreeswijk et al 2008 , cucchiara & fox 2008 ) .
the time - averaged spectrum during the burst peaked at 650 kev .
the konus spectrum was @xmath12 below its peak at 650 kev and @xmath13 above it . the peak optical flux during the burst was @xmath14 mag ( karpov et al 2008 ) , detectable with unaided eye for about 30 s under appropriate conditions . the optical and the gamma - ray light - curves of grb 080319b shown in figure [ prompt ]
are mildly correlated , suggesting that both radiations originated from the same source .
the optical - gamma - ray correlation function for the entire prompt emission has a maximum at lags between 0s and 5s , that being mostly the result of the contemporaneous rise at 0 - 18 s and fall at 43 - 60 s. for the middle part of prompt emission , the optical - gamma - ray correlation function still has a peak at lags between -1s to 3s , but is less prominent
. the time - lag between the ( grb ) emission produced in ( say ) internal shocks and the ( optical ) emission released by the blast - wave caused by the ejecta interaction with the ambient medium can be arbitrarily small , thus the optical and @xmath3-ray fluctuations could be correlated ( as seen for grb 080319b ) even though they arise from different sources .
however , the short duration @xmath15 of the optical pulses ( with @xmath16 ) and the lack of an increasing optical pulse duration throughout the burst ( instead , a decrease is apparent ) , as expected for an increasing source radius , argue against an origin of the optical counterpart in the external shock ( either the reverse - shock crossing the ejecta or the forward - shock propagating in the circumburst medium ) .
this makes it more likely that the two prompt emissions are , in fact , arising from same source .
the extrapolation of the gamma - ray spectrum ( which peaks at 650 kev and has a peak flux of about 7 mjy ) to the optical band under - predicts the observed optical counterpart flux ( which is around 20 jy ) by almost 4 orders of magnitude .
this indicates that , although from same source , the optical and gamma - ray radiation processes are different .
a natural possibility is that optical photons were generated via the synchrotron emission and gamma - rays via inverse - compton scatterings of that synchrotron emission ( i.e. _ synchrotron self - compton _ mechanism )
. the peak energy of the synchrotron spectrum ( @xmath17 ) should be below optical , to account for the optical spectrum of @xmath18 ( woniak et al 2008 ) : measured by raptor after 80 s. we parameterize the synchrotron peak photon energy as @xmath19ev ( @xmath20 ) , above which the optical counterpart spectrum has a spectral index @xmath21 ( i.e. @xmath22 ) .
the measured peak energy of the @xmath3-ray burst spectrum , which is the inverse - compton scattering of synchrotron photons , implies that the radiating electrons have a random lorentz factor @xmath23 , where the factor 5 accounts for the scattering by a population of electrons with a power - law distribution with energy above @xmath24 .
the second inverse - compton scattering component should peak at @xmath25 gev and its fluence should be larger than that of the gamma - rays by a factor equal to the gamma - ray to optical fluence ratio : @xmath26 ( this ratio is the `` compton parameter '' ) . hence , the fluence of the @xmath27 gev emission accompanying grb 080319b should have been @xmath28 erg @xmath9 ; the glast / lat instrument would collect thousands photons at 0.1100 gev during a burst as bright as grb 080319b . for the isotropic - equivalent gamma - ray output of grb 080319b of @xmath29 erg ,
the compton parameter above implies a gev output of @xmath30 erg . the energy release is smaller if the second inverse - compton scattering is in the klein - nishina regime .
the total energy release is reduced further by a few orders of magnitude if the fireball is tightly collimated ( i.e. the explosion energy is released in a narrow jet ) .
these reduction factors are estimated below .
as the peak of the synchrotron spectrum is at @xmath31 hz , where @xmath32 is the source s magnetic field and @xmath33 its bulk lorentz factor , the inferred random lorentz factor @xmath24 of the radiating electrons and the peak frequency of the synchrotron spectrum imply that @xmath34 gauss . for a burst redshift @xmath35 ,
the peak flux of the synchrotron emission is @xmath36 jy , where @xmath37 is the number of radiating electrons , thus observed synchrotron optical peak flux of @xmath38 jy requires that @xmath39 .
the optical thickness of the source to thomson scattering ( @xmath40 ) is the ratio of the inverse - compton peak flux ( at 650 kev ) and the synchrotron peak flux : @xmath41 . using that @xmath42 , where @xmath43 is the electron cross - section for thomson scattering
, it follows that the grb source radius is @xmath44 cm .
this distance is a factor @xmath45 larger than that expected ( eg .
piran 2005 ) for internal shocks .
the very steep decline of gamma - ray and optical flux at the end of the burst suggests that the source turned - off quickly .
the flux decline can not be faster than the limit provided by the photons emitted from source regions outside the @xmath46 opening area moving toward the observer .
this large - angle " emission arising from the fluid moving at an angle larger than @xmath46 yields a flux decay @xmath47 ( fenimore & sumner 1997 , kumar & panaitescu 2000 ) .
since the burst emission has @xmath48 in the 15150 kev bat window ( cummings et al 2008 ) , the large - angle emission should decay as @xmath49 , i.e. much slower than the fall - off displayed by the burst emission at 5055 s ( figure [ aglow ] ) .
the observed decay of the burst tail can be reconciled with the upper limit set by the large - angle emission if the last grb pulse is timed not since the burst trigger ( e.g. fan & wei 2005 ) but since a reference time @xmath50 after it , when the relativistic ejecta producing the last grb pulse must have been released .
this means that the timescale of the grb pulse decay is @xmath51 s. taking into account that @xmath52 , the inferred burst radius implies a source lorentz factor @xmath53 .
then , from @xmath54 g , we find that the magnetic field in the grb source is @xmath55 g. the grb source radius can be used to constrain the location of the synchrotron peak frequency ( @xmath19 ) .
the lack of a progressive lengthening of the burst pulse throughout the burst indicates that the grb source is not decelerating .
the radius at which the ejecta are decelerated by their interaction with the wind expelled by the wolf - rayet progenitor is @xmath56 cm , provided that the kinetic energy of the ejecta is similar to the radiation output .
the condition @xmath57 leads to @xmath58 . at the same time , at the end of the burst , the peak of the synchrotron spectrum can not be well above 1 ev ( i.e. @xmath59 ) because that would bring the synchrotron self - absorption frequency above the optical and yield a constant optical flux from the large - angle emission , which is inconsistent with observations of a decaying optical flux after the burst end .
thus , to a good approximation , @xmath60 , implying that the time averaged optical spectrum during the burst should have been rather flat . for the inferred @xmath24 and @xmath33
, the second inverse - compton scattering occurs just above the klein - nishina limit , where the electron scattering cross - section is @xmath61 .
therefore the twice - scattered photon takes all the electron energy , the peak energy of the 2nd inverse - compton component being at @xmath62 gev and the compton parameter for the 2nd scattering being @xmath63 \sim 10 $ ] ( i.e. 10 times smaller than for the first scattering ) .
thus , the isotropic - equivalent total radiation output of grb 080319b was @xmath64 erg .
from the inferred @xmath32 , @xmath33 , and @xmath65 ( for first and second scattering ) , it can be shown that the cooling electron lorentz factor @xmath66 , thus @xmath67 and the grb electrons are fast - cooling ( sari , piran & narayan 1998 ) .
this has two consequences .
first , since the grb spectrum does not exhibit a @xmath68 spectrum below its peak , as expected for radiatively - cooling electrons , a re - acceleration process must be active during the burst .
second , at the end of burst , when electrons are no longer accelerated , the peak energy of the inverse - compton spectrum falls fast below the bat window and the emission from the @xmath46 region moving toward the observer decreases on a timescale shorter than @xmath15 , allowing the large - angle emission to become dominant ( i.e. initial assumption that the grb tail is the large - angle emission is self - consistent ) .
if the grb ejecta were baryonic with one proton per electron , then the energy stored in protons were larger by that of electrons by a factor of at least @xmath69 , implying that the total grb outflow energy is at least 5 times larger than the radiated energy .
thus the total isotropic - equivalent energy release in this grb explosion must have been @xmath70 erg . although the synchrotron self - compton model is a _
natural _ interpretation of the optical and gamma - ray prompt emission from grb 080319b , it suffers from a few , possibly serious , problems .
one of the problems is the incompatibility between the observed and expected gamma - ray spectrum . in the synchrotron - self compton model , the synchrotron ( optical ) and inverse - compton ( gamma - rays ) spectra
are closely related .
the model parameters derived above lead to an upscattered self - absorbed photon energy of @xmath45 kev , below which the inverse - compton spectrum should be @xmath71 ( panaitescu & mszros 2000 ) .
in contrast , golenetskii et al ( 2008 ) measure for grb 080319b a softer @xmath72 gamma - ray spectrum down to 20 kev .
a similar inconsistency between the self - compton model and gamma - ray observations is encountered for grb 990123 , which we ( panaitescu & kumar 2007 ) suggested that could be explained if the magnetic field decays ( rossi & rees 2003 ) on a length - scale much larger than the plasma skin depth .
furthermore , in the large - angle emission interpretation of the early , steep optical decay ( [ lae ] ) , the optical spectral slope @xmath73 measured by woniak et al ( 2008 ) during that decay is just the slope above the peak of the synchrotron spectrum of the prompt emission .
however , that is inconsistent with the spectral slope of the inverse - compton emission above the grb spectral peak , which golenetskii et al ( 2008 ) report to be much softer , @xmath74 . the possible solution to this problem
is that the early optical afterglow emission of grb 080319b , which we identify with the large - angle emission released during the burst , is dominated by that from a single pulse whose spectrum is harder above its spectral peak than that of the burst - averaged gamma - ray spectrum .
taking onto account that the decay of the large - angle emission from a pulse lasting @xmath15 , peaking at @xmath75 , and of peak flux @xmath76 , is @xmath77^{-2-\beta_o}$ ] , it seems plausible that the large - angle emission is dominated by the pulse with the hardest spectral slope @xmath21 .
another potential problem for the synchrotron self - compton interpretation of grb 080319b s prompt emission is that the gamma - ray light - curve exhibits a higher variability than the optical counterpart , as one would expect the inverse - compton gamma - ray emission to have less fluctuations than the synchrotron seed photon field .
we note that the rise of the optical counterpart emission lagging that of the gamma - rays could be due to the optical band being below the synchrotron self - absorption frequency at early times .
in this case , the adiabatic cooling of ejecta electrons leads to a steep rise ( @xmath78 ) , consistent with that displayed by the optical emission at 815 s. at the other end , the optical counterpart emission turned - off before the gamma - rays .
that the last three grb pulses are not accompanied by optical emission may be due to that their synchrotron spectrum peaked sufficiently below the optical .
after the burst , the optical emission has a spectral slope @xmath80 and displays a steep decay , @xmath81 up to 1 ks , as shown in figure [ early ] , where the contribution of the power - law decaying emission after 1 ks , @xmath82 , was subtracted from the earlier optical flux . assuming that the burst ejecta is in pressure equilibrium with the ambient medium shocked by the forward shock , whose structure is described by the blandford - mckee solution , we find that the steepest decay for the synchrotron emission from the ejecta electrons , after they have been energized by the reverse - shock ( i.e. as they cool adiabatically ) , is obtained for a wind - like medium and an ejecta shell thickness increasing as @xmath83 , with @xmath84 the shock radius and @xmath85 its lorentz factor .
that decay , @xmath86 ( see equation 55 of panaitescu & kumar 2004 ) is too slow to account for the decay measured for grb afterglow 080319b at 1001000 s ( the results of sari & piran 1999 and kobayashi & sari 2000 for the reverse - shock emission give a steeper decay , up to @xmath87 , which is still too slow to account for the observed @xmath81 ) . instead
, the optical light - curve decline is consistent with the large - angle emission model , for which @xmath88 is expected .
therefore , we suggest that the optical afterglow emission between 50 s and 1 ks arises from the same source that produced @xmath3-ray photons . the slower decay of the optical flux observed after 1 ks , @xmath89 ( figure [ aglow ] ) , is most naturally attributed to the emission from the forward - shock driven into the ambient medium ( e.g. paczyski & rhoads 1993 , mszros & rees 1997 ) by the burst ejecta overtaking the large - angle burst emission . the optical flux decay ( @xmath90 ) and spectral slope @xmath91 at 90 ks
( bloom et al 2008 ) satisfy @xmath92 , which is consistent only with a circumburst medium density of @xmath93 radial structure and cooling frequency of the forward - shock synchrotron spectrum above optical ( for which @xmath94 is expected chevalier & li 1999 ) . a homogeneous circumburst medium , for which @xmath95 or -0.5 ( depending on the location of the cooling break ) is inconsistent with the observations at the 4@xmath96 level .
the x - ray light - curve should exhibit the same decay as the optical if the cooling break frequency is not between optical and x - ray or , owing to a wind - like medium , a slower one with decay index @xmath97 if the cooling frequency is between these domains .
the former could be the case for the early ( before 1 ks ) x - ray light - curve , whose decay is compatible with that seen in the optical after 1 ks , but neither expectation is satisfied by the x - ray and optical decays after 1 ks , for which @xmath98 . the same inconsistency with the standard blast - wave model expectations is also illustrated in the lower panel of figure [ aglow ] , which shows that , if the optical and x - ray afterglow emissions were attributed to the same spectral component , then the cooling break would have to exhibit an non - monotonic evolution incompatible with the blast - wave model , in general , and with the @xmath99 evolution expected for wind medium , in particular .
the non - monotonic evolution of the cooling break shown in figure [ aglow ] suggests a departure from the basic assumptions of the standard blast - wave model , such as a non - monotonically evolving magnetic field energy parameter ( on which the cooling frequency has the strongest dependence ) . alternately , the optical and x - ray afterglow emissions may have ( somewhat ) different origins .
for instance , the optical flux , which exhibits a single power - law decay after 1 ks , could be attributed to the synchrotron emission from the forward shock , while the x - ray flux , which displays 5 power - law segments , could arise from reprocessing of the forward - shock emission by scattering off a lagged part of the relativistic outflow ( panaitescu 2008 ) . for grb ejecta collimated into a jet of half - angle @xmath100 ,
the afterglow light - curve should exhibit a steepening to a decay faster than @xmath101 when the jet edge becomes visible to the observer .
using the arrival time of photons emitted at angle @xmath46 and for a circumburst medium of density typical for galactic wolf - rayet stars , the jet - break time is @xmath102 d. since the optical light - curve of grb 080319b afterglows does not exhibit a jet - break during the first 10 days ( figure [ aglow ] ) , the jet opening is @xmath103 deg .
therefore , the collimation - corrected kinetic energy of the grb outflow is @xmath104 erg .
the modest correlation between the optical and gamma - ray light - curves of grb 080319b suggests that these emissions may be from the same source . that the prompt optical flux lies well above the extrapolation of the burst spectrum to optical indicates that the two emissions arose from different radiation processes : optical emission was generated through synchrotron process and @xmath3-rays were due to inverse - compton scattering of the synchrotron photons . as shown by ( kumar & mcmahon 2008 )
, a bright optical emission accompanying @xmath3-rays is a generic prediction of the synchrotron self - compton model for grbs .
however , some features of the gamma - ray data for grb 080319b are not accounted for by that model , as discussed in [ lim ] .
synchrotron self - compton emission may have also been observed during grb 990123 ( panaitescu & kumar 2007 ) . for the bright optical counterpart of grb 080319b
, we find that the spectrum of the synchrotron prompt emission must have peaked in the optical , while for grb 990123 that peak was slightly above optical , which can explain ( at least in part ) why these optical counterparts were so bright . within the synchrotron self - compton model for the prompt emission and without any assumption about the dissipation mechanism , we have shown that the @xmath3-ray and optical measurements of grb 080319b imply a source radius of @xmath0 cm , moving at lorentz factor 500 .
the energy of the 2nd inverse - compton scattering , which should have peaked at @xmath105 gev , must have been 10 times larger than that of the 1st inverse - compton component at sub - mev energies . with the lower limit on the half - angle of the grb outflow set by the lack of an afterglow optical light - curve break until 10 d , and assuming a baryonic , shock - heated @xmath3-ray source
, we estimate that the beaming - corrected kinetic energy of outflow powering grb 080319b was higher than @xmath2 erg ( twice larger for a double - sided jet ) , which provides an important constraint for the collapsar model ( macfadyen & woosley 1999 , woosley & bloom 2006 ) for long bursts .
we find that the optical light - curve decay at 0.11 ks is too fast to be attributed to the external reverse - shock propagating into the grb ejecta . instead
, the early optical decay and spectral slope are consistent with the large - angle emission released during the burst and arriving at observer later , due to the spherical curvature of the emitting surface .
the decay of the late - time ( @xmath106 ks ) optical afterglow flux is consistent with an origin in the forward - shock energizing a wind - like circumburst medium , a homogeneous medium being `` ruled out '' at 4@xmath96 .
however , if the optical and x - ray afterglow fluxes arose from same mechanism , an unusual evolution of the cooling frequency would be required , which may be indicative of evolving blast - wave microphysical parameters .
alternatively , the x - ray afterglow emission may be affected by reprocessing of the optical forward - shock photons through ( bulk and , possibly , inverse - compton ) scattering by a relativistic , pair - rich outflow located behind the forward shock . in this model , the scattering outflow results from continued accretion onto the central black - hole , if such an accretion can be maintained for source - frame timescales comparable to the observer - frame duration of the x - ray afterglow .
the authors are grateful to the referee for his comments .
ap acknowledges the support of the us department of energy through the lanl / ldrd 20080039dr program and of nasa swift gi grant nng06en001 for this work .
bloom j. et al . , 2008 , preprint ( arxiv:0803.3215 ) chevalier r. , li , y. , 1999 , apj , 520 , l29 cucchiara a. , fox d. , 2008 , gcn 7456@xmath107 cummings j. et al . , 2008 ,
gcn 7462@xmath107 della valle m. et al . , 2003 , a&a , 406 , l33 evans p. et al . , 2007 , a&a , 469 , 379 fan y. , wei d. , 2005 , mnras , 364 , l42 fenimore e. , sumner m. , 1997 , in `` all - sky x - ray observations in the next decade '' , eds . m. matsuoka and n. kawai , riken , japan , p.167 ( astro - ph/9705052 ) frail d. et al . , 2001 ,
apj , 562 , l55 galama t. et al . , 1998 , nature , 395 , 670 golenetskii s. et al . , 2008 , gcn 7482@xmath107 hjorth j. et al . , 2003 ,
nature , 423 , 847 karpov s. et al .
, 2008 , gcn 7558@xmath107 kobayashi s. , sari r. , 2000 , apj , 542 , 819 kumar p. , panaitescu a. , 2000 , apj , 541 , l51 kumar p. ,
mcmahon e. , 2008 , mnras , 384 , 33 macfadyen a. , woosley s. , 1999 , apj , 524 , 262 malesani d. et al . , 2004 , apj , 609 , l5 mszros p. , rees m. , 1997 , apj , 476 , 232 nakar e. , 2007 , physics rep . , 442 , 166 paczyski b. , rhoads j. , 1993 , apj , 418 , l5 panaitescu a. , mszros p. , 2000 ,
apj , 544 , l17 panaitescu a. , kumar p. , 2002 ,
apj , 571 , 779 panaitescu a. , kumar p. , 2004 ,
mnras , 353 , 511 panaitescu a. , kumar p. , 2007 ,
mnras , 376 , 1065 panaitescu a. , 2008 , mnras , 383 , 1143 piran t. , 2005 , rev .
phy . , 76 , 1143 racusin j. et al . , 2008 , gcn 7459@xmath107 rossi e. , rees m. , 2003 , mnras , 339 , 881 sari r. , piran t. , narayan r. , 1998 , apj , 497 , l17 sari r. , piran , t. , 1999 , apj , 520 , 641 soderberg a. , frail d. , wieringa m. , 2004 , apj 607 , l13 taylor g. , frail d. , berger e. , kulkarni s. , 2004 , apj , 609 , l1 vreeswijk p. et al . , 2008 ,
gcn 7444@xmath107 vestrand w. et al .
, 2005 , nature , 435 , 178 vestrand w. et al .
, 2006 , nature , 442 , 172 woosley s. , bloom j. , 2006 , ara&a , 44 , 507 woniak p. et al . , 2008 , preprint @xmath107 gcn circulars can be found at + _ http://gcn.gsfc.nasa.gov/gcn/gcn3_archive.html _ | the optical and gamma - ray observations of grb 080319b allow us to provide a broad - brush picture for this remarkable burst .
the data indicate that the prompt optical and gamma - ray photons were possibly produced at the same location but by different radiation processes : synchrotron and synchrotron self - compton , respectively ( but we note that this interpretation of the gamma - ray data faces some difficulties ) .
we find that the burst prompt optical emission was produced at a distance of @xmath0 cm by an ultra - relativistic source moving at lorentz factor of @xmath1 .
a straightforward inference is that about 10 times more energy must have been radiated at tens of gev than that released at 1 mev .
assuming that the grb outflow was baryonic and that the gamma - ray source was shock - heated plasma , the collimation - corrected kinetic energy of the jet powering grb 080319b was larger than @xmath2 erg .
the decay of the early afterglow optical emission ( up to 1 ks ) is too fast to be attributed to the reverse shock crossing the grb ejecta but is consistent with the expectations for the large - angle " emission released during the burst .
the pure power - law decay of the optical afterglow flux from 1 ks to 10 day is most naturally identified with the ( synchrotron ) emission from the shock propagating into a wind - like medium .
however , the x - ray afterglow requires a departure from the standard blast - wave model .
radiation mechanisms : non - thermal - shock waves - gamma - rays : bursts |
in this paper we study large deviations for the zero set @xmath11 of the time dependent planar gaussian analytic function ( gaf ) @xmath12 where @xmath1 are @xmath2 complex valued ornstein - uhlenbeck processes .
specifically , @xmath13 where @xmath14 is a brownian motion in @xmath15 . with probability one , this process defines an analytic function in the entire plane , and at any fixed time @xmath16 the distribution of @xmath11 is translation invariant ( see sodin and tsirelson @xcite for references ) .
sodin and tsirelson @xcite study the large deviation behavior of @xmath17 at a fixed time , and show that the hole probability decays exponentially in the square of the area of the disk .
the over - crowding behavior of @xmath17 at a fixed time has been studied by krishnapur @xcite who shows that the probability a fixed disk contains @xmath10 zeros is asymptotic to @xmath18 .
the study of gaussian analytic functions as dynamic processes was initiated by peres and virg @xcite , who considered the closely related time dependent hyperbolic gaf @xmath19 where @xmath1 are defined as above .
peres and virg showed that @xmath20 can be reconstructed from its zeros , and hence the zero process determined by @xmath21 is a time homogenous markov process . their proof may be easily adapted to show that @xmath22 can be reconstructed from its zero set , and hence @xmath11 is a time homogenous markov process as well . for fixed @xmath16 ,
the zero set @xmath11 exhibits strong repulsive forces between nearby zeros , as one can see visually in figure [ fig1 ] .
the zero set of @xmath23 ( left ) and a poisson point process with the same intensity.,title="fig:",height=168 ] the zero set of @xmath23 ( left ) and a poisson point process with the same intensity.,title="fig:",height=168 ] to appreciate the effect of this repulsion , sodin and tsirelson compare @xmath24 to three toy models .
the first model is a poisson process with the same intensity , @xmath25 .
the second model is a _
perturbed lattice model _ consisting of the points @xmath26 where @xmath27 are independent standard @xmath28 random variables .
the third model is a _ triangular cluster model _ consisting of the points @xmath29 where @xmath27 are as before .
they prove that the fixed time hole probability that @xmath24 contains no zeros in the disk of radius @xmath30 decays as @xmath31 .
this decay rate differentiates @xmath32 from the poisson process for which the hole probability decays as @xmath33 , but not from the perturbed lattice and triangular cluster models . by studying the asymptotic behavior of linear statistics , sodin and tsirelson are able to differentiate @xmath11 from the perturbed lattice model , but not the triangular cluster model .
each of the toy models has a natural extension to a time dependent process which preserves its distribution .
the poisson process can be made into a time dependent process by allowing the points to evolve as independent planar brownian motions .
the perturbed lattice and triangular cluster models may be extended to time dependent processes by replacing the coefficients @xmath27 with independent ornstein - uhlenbeck processes .
we show that certain large deviation probabilities starkly contrast @xmath34 from the other three models .
specifically , let @xmath35 denote the _ hole probability _ that @xmath36 contains no zeros of @xmath23 for all @xmath4 $ ] .
we prove that [ main ] @xmath37 and @xmath38 denote by @xmath39 , where @xmath40 equals @xmath41 for poisson process , @xmath42 for perturbed lattice or @xmath43 for triangular cluster , the event that the time dependent @xmath44 model has no points in @xmath45 for all @xmath4 $ ] .
we have the following results : the zero set of @xmath23 ( left ) and the second toy model ( right ) , conditioned to have a hole of radius five.,title="fig:",height=168 ] the zero set of @xmath23 ( left ) and the second toy model ( right ) , conditioned to have a hole of radius five.,title="fig:",height=168 ] [ toyprop1 ] for the poisson process @xmath46 for the perturbed lattice model and the triangular cluster model ( @xmath47 ) , any @xmath48 and @xmath49 , there exist positive constants @xmath50 and @xmath51 depending only on @xmath52 and @xmath53 so that @xmath54 and @xmath55 we also obtain estimates for the _ overcrowding probability _ , the event that @xmath45 contains at least @xmath10 zeros of @xmath21 for all @xmath4 $ ] .
denote this event by @xmath56 .
[ main2 ] for fixed @xmath3 , we have @xmath57 and @xmath58 for the three toy models , denote by @xmath59 the event that the time dependent @xmath44 model ( @xmath40 equals @xmath41 , @xmath42 or @xmath43 ) has at least @xmath10 points in @xmath45 for all @xmath4 $ ] .
we have the following results : [ toyprop2 ] for the poisson process @xmath60 for the perturbed lattice model and the triangular cluster model ( @xmath61 ) : @xmath62 this paper is organized as follows . in section 2 , we state some well known large and small deviation estimates for ornstein - uhlenbeck processes , and prove the large deviation estimates for the toy models . in section 3 , we prove that @xmath63 can be reconstructed from its zero set , and deduce that @xmath11 is a time homogenous markov process . in section 4 ,
we prove theorem [ main ] , and in section 5 we prove theorem [ main2 ] . in section [ ouprocess ] , we prove the large deviation estimates for ornstein - uhlenbeck processes .
we begin by stating some well known large and small deviation estimates for ornstein - uhlenbeck processes that will be used throughout the paper . for completeness , full proofs of these estimates
are given in section [ ouprocess ] .
these are the only estimates on the coefficients @xmath27 necessary to derive propositions [ toyprop1 ] and [ toyprop2 ] . however , further properties of ornstein - uhlenbeck properties are used ( at least superficially ) to derive theorems [ main ] and [ main2 ] . [ ousm ] let @xmath64 where @xmath65 is a @xmath15-valued brownian motion started from 0 . for all @xmath66 and @xmath49 , there exist constants @xmath67 and @xmath68 depending only on @xmath53 and @xmath52 so that @xmath69)\;<\ ; e^{-c_2t / r^2}.\ ] ] [ oulg ] let @xmath64 where @xmath65 is a @xmath15-valued brownian motion started from 0 . for all @xmath70 and @xmath49 , there exist constants @xmath67 and
@xmath68 depending only on @xmath53 and @xmath52 so that @xmath71 \right)\ ; < \;e^{-c_2 t r^2}.\ ] ] [ oucirc ] let @xmath72 denote the ball of radius @xmath73 centered at @xmath74 . then for fixed @xmath73 and all @xmath75 and @xmath76 there exist constants
@xmath50 and @xmath51 so that @xmath77 ) \leq e^{-c_2 r^2 t}.\ ] ] [ ouhp ] let @xmath78 where @xmath65 is a @xmath15-valued brownian motion . for all @xmath79 and @xmath49
there exists a constant @xmath80 depending only on @xmath53 and @xmath52 so that @xmath81 ) \geq
\exp \left [ -t e^{-cr^2 } \right].\ ] ] we start with the poisson process .
let @xmath82 denote the conditional probability that no points in the poisson process with intensity @xmath83 intersect @xmath45 during the time interval @xmath84 $ ] , given that no points lie in @xmath45 at time 0 . brownian scaling and the fact that the union of two independent poisson processes is another poisson process gives the following : @xmath85 moreover , @xmath86 .
so it suffices to bound @xmath87 . for a complex valued brownian motion @xmath88 , let us denote by @xmath89 the hitting time of @xmath90 .
recall that for @xmath91 @xmath92 it follows that @xmath93 where we write @xmath94 to denote two iterations of the @xmath95 function .
now compute @xmath96 where we have used the reflection principle and lemma 1.3 in @xcite .
it follows that for large @xmath97 we have the estimate @xmath98 .
combining this fact with equation ( [ rad_1_est ] ) we find @xmath99 for @xmath65 starting at radius @xmath100 we can compute the probability that @xmath65 avoids @xmath101 for all @xmath102 $ ] by considering the probability that the brownian motion hits @xmath103 prior to @xmath101 and then use ( [ rad1 ] ) .
this consideration yields @xmath104 similar reasoning for @xmath105 yields the bound @xmath106 now fix @xmath10 , and let @xmath107 for @xmath108 , and @xmath109 for @xmath110 .
write @xmath111 to denote the number of points in @xmath112 at time 0 , and similarly @xmath113 and @xmath114 .
we compute , using equations ( [ smallr ] ) and ( [ bigr ] ) @xmath115^{\ # a_k } \right)\\ & & \e \left(\prod_{k=1}^\infty \left[1 - 4(k-1 ) e^{-(k-1)^2/2}\left(\frac{3 \log_2 t}{\log t}\right ) \right]^{\ # \tilde{a}_k}\right).\end{aligned}\ ] ] now , if @xmath116 is a poisson random variable with mean @xmath117 , then @xmath118 . therefore @xmath119 \\ & & \prod_{k=1}^\infty \left[\exp \left ( ( - \frac{12(k-1)e^{-(k-1)^2/2 } \log_2 t}{\log t})(2 \pi k ) \right)\right]\\ & \geq & e^{-1/n^2 } \exp \left [ \sum_{k=1}^n \left ( ( -\frac{2 \log \frac{n}{k}}{\log t } - \frac{3 \log_2 t}{\log t } ) ( \frac{1}{n})(\frac{2\pi(k+1)}{n})\right ) \right]\\ & & \exp \left [ \sum_{k=1}^\infty(- \frac{12(k-1)e^{-(k-1)^2/2 } \log_2 t}{\log t})(2 \pi k ) \right]\\ & \geq & e^{-1/n^2 } \exp \left[-\frac{c_1
\log_2 t}{\log t } \right ] .\end{aligned}\ ] ] taking @xmath120 , we have @xmath121 . combining this result with ( [ reduction ] )
it follows that as @xmath122 @xmath123 which proves the claim .
we now consider the perturbed lattice model . to
bound @xmath124 from above , note that if @xmath125 occurs then for each point @xmath126 with @xmath127 we have @xmath128 for all @xmath4 $ ] .
applying lemma [ oulg ] , we see that @xmath129 and ( [ toy1ub ] ) follows . for the lower bound , observe that @xmath130 will occur if the following two conditions are satisfied for all lattice points @xmath131 : * if @xmath132 we have @xmath133 for all @xmath4 $ ] . *
if @xmath134 then for all @xmath4 $ ] the process @xmath135 lies in the half plane @xmath136 which is a distance @xmath3 from the origin , is parallel to one of the coordinate axis and maximizes @xmath137 . from lemma [ oulg ]
we see that @xmath138 from lemma [ ouhp ] we estimate : @xmath139^{8k } = e^{-\tilde{c } t}.\ ] ] since the events @xmath140 and @xmath141 are independent we obtain ( [ toy1 lb ] ) for @xmath142 .
the proof of the upper bound for the triangular cluster model is completely analogous .
the proof of the lower bound is also very similar .
note that @xmath143 will occur if the following two conditions are satisfied : * if @xmath144 then @xmath145 for all @xmath4 $ ] . *
if @xmath146 , then @xmath147 lies in the quarter plane @xmath148 such that @xmath149 is a distance @xmath150 from @xmath151 and the vertex of @xmath148 lies on the segment connecting @xmath151 to the origin .
see figure [ toy ] .
construction used for bounding the hole probability for the triangular cluster model . by restricting one of the points in a triad to an appropriate quarter plane
, we can ensure that none of the three points enters @xmath45.,height=96 ] lemma [ ouhp ] implies that @xmath152 moreover , since @xmath153 and @xmath154 are independent and the distribution of @xmath155 is radially symmetric we can apply lemma [ ouhp ] to compute : @xmath156^{c_2 k } = e^{-c_3 t}.\ ] ] we now prove lower bounds for the over - crowding probabilities of the toy models .
we begin with the poisson process .
observe that : @xmath157 \textrm { and } b(\tau+r^2 ) \in d_{r/2 } | b(\tau ) \in \partial d_{r/2 } ) \ ] ] is a constant independent of @xmath3 , and gives a lower bound for the conditional probability @xmath157 \textrm { and } b(\tau+r^2 ) \in d_{r/2 } | b(\tau ) \in d_{r/2 } ) .\ ] ] it follows that @xmath158 from which the result follows . for the perturbed lattice model
, the result follows by using lemma [ oucirc ] to compute the probability that the points corresponding to the @xmath10 closest lattice points lie in @xmath45 for all @xmath4 $ ] .
the computation is analogous for the triangular cluster model , in this case we compute the probability that one point corresponding to each of the @xmath10 closest centers lies in @xmath45 .
this discussion closely parallels the proof given in @xcite that the modulus of a hyperbolic gaussian analytic function ( [ hypergaf ] ) can be reconstructed from its zero set .
the key lemma is the following [ reconstruct ] fix @xmath16 . then with probability one we have @xmath159 where @xmath160 is euler s constant .
assume @xmath23 has no zeros on @xmath90 , then by jensen s formula ( see @xcite , p. 208 ) we have @xmath161 write @xmath162 where @xmath163 and @xmath164
. then @xmath165 define @xmath166 , so that the distribution of @xmath167 is independent of @xmath30 , @xmath16 and @xmath83 , and @xmath168
. set @xmath169 we prove that with probability one @xmath170 as @xmath171 .
the proof is a straightforward application of chebyshev s inequality and the borel - cantelli lemma .
compute : @xmath172 where we have used rotational invariance and absolute integrability to obtain the last expression . by lemma 17 in @xcite
we have the estimate @xmath173 it follows that @xmath174.\end{aligned}\ ] ] for small @xmath74 we have the estimate @xmath175 .
it follows that for @xmath176 close to 1 , @xmath177 .
thus for large @xmath30 we obtain the bound @xmath178.\ ] ] applying chebyshev s inequality to ( [ varest ] ) , it follows from the borel cantelli lemma that @xmath179 a.s .
thus by equation ( [ jenson_reconstruct ] ) : @xmath180 exponentiating ( [ keylemmaeqn ] ) , we obtain ( [ keylemma ] ) . to see that
@xmath22 may be reconstructed from its zero set , note that if @xmath181 , then by computing covariances we see that @xmath182 so , applying lemma [ reconstruct ] to @xmath183 , we can recover @xmath184 with probability one .
iterating this procedure , we can recover @xmath22 from @xmath185 on a dense countable subset , and hence recover @xmath186 everywhere by continuity .
in this section , we compute the probability of the event @xmath35 that @xmath45 contains no zeros of @xmath23 for all @xmath4 $ ] . we claim that a hole of radius @xmath3 will exist for all @xmath4 $ ] if the following three conditions are satisfied for all such @xmath16 : * @xmath187 * @xmath188 for @xmath189 * @xmath190 for @xmath191 .
a similar computation is given in @xcite .
write @xmath192 and compute @xmath193 since we also have @xmath194 ( we used the inequality @xmath195 which follows from sterling s formula ) , it follows that if ( @xmath196 ) , ( @xmath197 ) and ( @xmath198 ) hold then @xmath199 and hence @xmath200 contains no zeros in @xmath45 for @xmath4 $ ] . using lemmas [ ousm ] and [ oulg ] we have : @xmath201 \nonumber \\
\pp(ii ) & \geq & \exp \left [ -t e^{(\frac{1}{2}+o(1))r^2 } \right]^{48r^2 } = \exp \left [ -t e^{(\frac{1}{2}+o(1))r^2 } \right ] \nonumber\\
\pp(iii ) & \geq & \prod_{k>48r^2 } e^{c_1 t/4^k } = e^{-c_2 t}.\end{aligned}\ ] ] events @xmath140 , @xmath141 and @xmath202 are independent , therefore @xmath203\ ] ] as desired .
if a hole of radius @xmath3 exists at time @xmath16 , jensen s formula gives @xmath204 where @xmath117 is the uniform probability measure on @xmath205 .
so for fixed @xmath206 and @xmath207 one of the following three events must occur : * @xmath208 and @xmath209 * @xmath210 * @xmath211 .
write @xmath212 to denote the event that @xmath213 occurs at time @xmath16 , and @xmath214 to denote the event that @xmath215 and @xmath216 to denote the event that @xmath217 .
let @xmath218 denote the event that either @xmath214 or @xmath216 occurs . also , define @xmath219 and @xmath220 .
the method of proof is similar to the proof of the upper bound given for lemma [ oulg ] .
we observe the function @xmath23 at a sequence of times @xmath221 and bound the probability that either condition @xmath213 or condition @xmath222 is satisfied at all @xmath223 .
specifically , define @xmath224 , where @xmath225 is defined as follows : * if @xmath226 is true and @xmath227 fails , @xmath228 * if @xmath229 , then @xmath230 .
if both @xmath226 and @xmath227 fail then we set @xmath231 , i.e. the observation process is halted .
the proof relies on the following [ claim ] for any @xmath232 and @xmath233 satisfying @xmath234 we may choose @xmath235 and @xmath53 sufficiently large so that there exist @xmath236 and @xmath237 such that : @xmath238 and @xmath239 for all @xmath240 .
the proof of this claim is somewhat technical , so we shall first check that it implies the upper bound stated in theorem [ main ] .
let @xmath236 and @xmath237 be chosen as in claim [ claim ] , and consider the the process @xmath241 , where @xmath242 are i.i.d . and have distribution : @xmath243 set @xmath244 .
equations ( [ aineq1 ] ) and ( [ bineq1 ] ) imply that @xmath242 stochastically dominates @xmath225 , so @xmath245 .
using the following lemma [ est2 ] @xmath246 .
it follows that @xmath247,\end{aligned}\ ] ] so the upper bound in theorem [ main ] follows by letting @xmath248 and @xmath249 decrease to zero . to prove lemma [ est2 ] , we compute @xmath250 the following lemmas allow us to bound the conditional probabilities @xmath251 and @xmath252 .
[ stest ] for @xmath253 , and @xmath254 with @xmath255 and @xmath256 we have : @xmath257 define @xmath258 .
observe that for fixed @xmath16 and @xmath259 we may write @xmath260 where @xmath261 and @xmath262 are i.i.d . @xmath28 random variables independent of @xmath263 .
@xmath264 so @xmath265 quoting ( * ? ? ?
* claim 1 ) , this probability is bounded above by @xmath266 provided that @xmath267 . choosing the maximum allowable value for @xmath40 , we obtain @xmath268 if @xmath269 with @xmath270 then [ alemma ] @xmath271 the proof uses several estimates from @xcite , and we use similar notation .
take @xmath272 and @xmath273 where @xmath274 and @xmath275 . by lemma [ stest ]
we see that if @xmath276 then with probability at least @xmath277 we can choose @xmath10 points @xmath278 with @xmath279 such that @xmath280 let @xmath281 be the poisson kernel for the disk @xmath45 with @xmath282 and @xmath283 .
define @xmath284 .
then if @xmath117 is the uniform probability measure on @xmath205 we have @xmath285 for the remainder of the proof we condition on the event that @xmath286 , since otherwise @xmath287 must fail .
so , @xmath288 also , by applying lemma [ stest ] with @xmath3 replaced by @xmath289 and @xmath290 , we know that except on an exceptional set of measure less than @xmath291 we have @xmath292 so that @xmath293 $ ] .
then @xmath294 an easy computation shows that @xmath295 for @xmath296 and @xmath297 , hence @xmath298 combining ( [ logp ] ) and ( [ logpm ] ) we obtain @xmath299 now from ( * ? ? ?
* claim 2 ) we know that @xmath300 combining ( [ star ] ) , ( [ logabs ] ) and ( [ abssum ] ) gives that , except on an exceptional set of probability bounded by @xmath301 : @xmath302 \nonumber \\ & \geq & ( 1/2 - c_6 \delta^{1/4})r^2 + ( 1+c_7 \delta^{1/2 } ) \log \sqrt{1-e^{-\delta t}}.\nonumber \;\;(*)\end{aligned}\ ] ] all we must show is that @xmath303 exceeds @xmath304 for sufficiently small @xmath305 which may be chosen uniformly in @xmath3 .
observe that @xmath303 is increasing in @xmath259 , so it suffices to restrict to @xmath306 .
using the inequalities @xmath307 and @xmath308 for @xmath309 , we compute : @xmath310 - \frac{c_8}{2}(1+c_7 \delta^{1/2 } ) e^{-(1 - 2c-\epsilon)r^2}. \label{final}\end{aligned}\ ] ] it is clear that for fixed @xmath311 and @xmath312 we may choose @xmath305 small enough ( uniformly in @xmath3 ) so that ( [ final ] ) exceeds @xmath304 .
fix @xmath233 small enough so that @xmath313
. then for @xmath314 [ b1lemma1 ] @xmath315^{\frac{\delta t_b(\ell)}{\delta t_a(\epsilon)}+1}.\ ] ] assuming @xmath227 fails , write : @xmath316 where @xmath317 is independent of @xmath318 and @xmath319 . using the inequalities
@xmath320 it follows that if @xmath321 is satisfied , then @xmath322 and therefore , since @xmath323 : @xmath324 now , @xmath325 , so @xmath326^{\frac{\delta t_b(\ell)}{\delta t_a(\epsilon)}+1}\ ] ] the quantity @xmath327 is decreasing for @xmath328 , it follows that @xmath329^{\frac{\delta t_b(\ell)}{\delta t_a(\epsilon)}+1 } \nonumber \\ & = & \left [ \exp \left ( - e^{(6c + 2 \epsilon - 2 + o(1))r^2 } \right)\right]^{\frac{\delta t_b(\ell)}{\delta t_a(\epsilon)}+1}\end{aligned}\ ] ] as desired .
[ b1lemma2]fix @xmath233 so that @xmath234
. then for @xmath314 we have @xmath330^{\frac{\delta t_b(\ell)}{\delta t_a(\epsilon)}+1}.\ ] ] as before , write @xmath331 where @xmath317 is independent of @xmath318 .
assuming @xmath332 occurs , we deduce : @xmath333 so @xmath334 .
now , @xmath335\end{aligned}\ ] ] is decreasing in @xmath336 for all @xmath337 , so we obtain : @xmath338^{\frac{\delta t_b(\ell)}{\delta t_a(\epsilon)}+1}.\ ] ] [ b2lemma1 ] fix @xmath233 so that @xmath234 , then if @xmath339 there exists a constant @xmath340 so that for all @xmath79 : @xmath341^{\frac{\delta t_b(\ell)}{\delta t_a(\epsilon)}+1}\ ] ] write @xmath342 where @xmath343 with @xmath344 i.i.d . and independent of @xmath318 . also , define @xmath345 and observe that : @xmath346 if @xmath347 is satisfied and @xmath227 fails then @xmath348 , and we have ( using the inequalities @xmath349 and @xmath350 ) : @xmath351 from @xcite we have the estimate : @xmath352 provided that @xmath353 , where @xmath354 is decreasing in @xmath83 . since we resticted to @xmath355 we can fix @xmath356 so that ( [ keyest ] ) may be used to estimate the probability of ( [ eq34 ] ) for all @xmath357 . for such constants , we have : @xmath358^{\frac{\delta t_a(\epsilon)}{2\delta t_b(\ell ) } } \nonumber \\ & = & \exp \left [ -\frac{3}{8 } e^{(\tilde{c } + c + \epsilon/2 - 1)r^2 } + ( e^{\ell r^2 } - 1)e^{(\tilde{c } - c - \epsilon/2)r^2 } \right]^{\frac{\delta t_a(\epsilon)}{2\delta t_b(\ell ) } } \nonumber \\ & = & \exp \left[\frac{-\frac{3}{8 } e^{(\tilde{c } + 3c + 3\epsilon/2 - 2)r^2 } + ( e^{\ell r^2 } - 1)e^{(\tilde{c } + c + \epsilon/2 -1)r^2}}{2\delta t_b(\ell)}\right ] \label{dec}.\end{aligned}\ ] ] it is clear that that if @xmath3 is sufficiently large , ( [ dec ] ) is decreasing in @xmath336 for @xmath337 , so evaluating ( [ dec ] ) at @xmath359 we obtain @xmath360^ { \frac{\delta t_b(\ell)}{\delta t_a(\epsilon)}+1}.\ ] ] [ b2lemma2]for @xmath355 and @xmath233 such that @xmath234 then there exists a constant @xmath53 so that for all @xmath79 : @xmath361\right)^{\frac{\delta t_b(\ell)}{\delta t_a(\epsilon)}+1}.\ ] ] equation ( [ decomp1 ] ) holds as before , but now we assume that @xmath227 is satisfied so @xmath362 .
if @xmath347 holds , then @xmath363 by requiring @xmath356 we deduce @xmath364 .
since we have fixed @xmath365 we can fix a constant @xmath53 so that for all @xmath79 equation ( [ keyest ] ) may be used to write @xmath366\right)^{\frac{\delta t_a(\epsilon)}{2\delta t_b(\ell ) } } \nonumber \\ & = & \exp \left [ -\frac{1}{4 \delta t_b(\ell)}e^{(\tilde{c } + \ell + 2c + \epsilon - 3/2)r^2 } \right].\label{dec2}\end{aligned}\ ] ] it is easy to check that ( [ dec2 ] ) is decreasing in @xmath336 for all @xmath328 , so @xmath361\right)^{\frac{\delta t_b(\ell)}{\delta t_a(\epsilon)}+1 } .\ ] ] claim [ claim ] now follows from lemmas [ alemma ] - [ b2lemma2 ] .
in this section we compute the probability of the event @xmath56 that @xmath45 contains at least @xmath10 zeros of @xmath21 for all @xmath4 $ ] . we claim that for sufficiently large @xmath10 , the disk @xmath45 will contain at least @xmath10 zeros for all @xmath4 $ ] if the following three conditions are satisfied for all such @xmath16 : * @xmath368 for all @xmath369 * @xmath370 * @xmath371 for all @xmath372 .
to see that this is the case , observe that condition @xmath140 implies that for @xmath4 $ ] : @xmath373 while @xmath202 implies that for large @xmath10 and @xmath374 $ ] : @xmath375 thus , the claim follows by comparing the functions @xmath376 and @xmath377 on @xmath205 and applying rouch s theorem . from lemmas [ ousm ] , [ oulg ] and [ ouhp ]
we compute : @xmath378 since events @xmath140 , @xmath141 and @xmath202 are independent , this computation yields the lower bound in theorem [ main2 ] .
the style of proof is identical to that given for the upperbound of theorem [ main ] .
jensen s formula states that @xmath379 where @xmath117 is the uniform probability measure on @xmath380 . evaluating ( [ jensen ] ) at @xmath381
, we see that if @xmath382 has at least @xmath10 zeros in @xmath45 and @xmath383 then : @xmath384 thus , if @xmath56 holds , @xmath385 and @xmath386 it follows that at each @xmath4 $ ] one of the following conditions must be satisfied : * @xmath387 * @xmath388 .
write @xmath212 to denote the event that condition @xmath213 is satisfied at time @xmath16 , and @xmath218 to denote the event that @xmath389 .
also , define @xmath219
. we shall observe the function @xmath382 at a sequence of times @xmath390 , and bound the probability that either condition @xmath213 or condition @xmath222 is satisfied at all @xmath391 .
define @xmath224 , where @xmath225 is defined as follows : * if @xmath226 is true and @xmath227 fails then @xmath392 * if @xmath227 is true and @xmath393 , then @xmath394 . if both @xmath226 and @xmath227 fail , then we set @xmath395 , i.e. the observation process is halted .
we assert the following [ claim2 ] for any @xmath396 we may choose @xmath397 sufficiently large so that there exists @xmath236 and @xmath237 such that : @xmath398 and @xmath239 for all @xmath399 .
it then follows exactly as before that @xmath400.\ ] ] the upper bound in theorem [ main2 ] now follows by letting @xmath401 and @xmath402 .
claim [ claim2 ] follows from the following three lemmas : for @xmath403 we have @xmath404 write @xmath405 where @xmath317 .
the probability that @xmath406 holds is maximized when @xmath407 , so we have @xmath408 since @xmath409 it follows that @xmath410 . hence : @xmath411 < 2 e^{-\alpha n \log n}.\end{aligned}\ ] ] if @xmath412 there exists a constant @xmath413 so that for all @xmath414 : @xmath415^{\frac{\delta t_b(\ell)}{\delta t_a(\alpha)}+1}\ ] ] define @xmath416 . following the proof of lemma [ b2lemma1 ] , we write @xmath417 and deduce that if @xmath332 is satisfied but @xmath227 fails then @xmath418 now , using ( [ keyest ] ) we obtain : @xmath419^{\frac{\delta t_a(\alpha)}{2\delta t_b(\ell ) } } \\ & = & \exp \left [ \frac{-(e^{\ell n \log n } - 1)e^{\frac{\alpha}{2 }
n \log n - n/2 } - \frac{3}{8}e^{(\frac{3 \alpha}{2 } - \gamma)n \log n - n/2}}{2\delta t_b(\ell ) } \right]\end{aligned}\ ] ] it is clear that for sufficiently large @xmath10 , the above expression is decreasing in @xmath336 for @xmath337 , so evaluating at @xmath359 we obtain : @xmath420^{\frac{\delta t_b(\ell)}{\delta t_a(\alpha)}+1}.\ ] ] for @xmath421 we have @xmath422^{\frac{\delta t_b(\ell)}{\delta t_a(\alpha)}+1}.\ ] ] equation ( [ decomp ] ) holds as before , but now we assume both @xmath227 and @xmath332 are satisfied so @xmath423
so for @xmath421 we have @xmath424 .
thus , @xmath425 \right)^{\frac{\delta t_a(\alpha)}{2\delta t_b(\ell ) } } \nonumber \\ & \leq & \exp \left [ - \frac{1}{4 \delta t_b(\ell ) } e^{(\alpha - \frac{\gamma}{2 } + \ell)n \log
n - n/2 } \right ] \nonumber \\ & \leq & \exp \left [ - e^{(\alpha - \frac{\gamma}{2 } + o(1 ) ) n \log n } \right].\end{aligned}\ ] ]
in this section we give proofs of the large deviation estimates for ornstein - uhlenbeck processes .
take @xmath426 , and observe that : @xmath427 also , for fixed @xmath16 we may write @xmath428 , where @xmath317 is independent of @xmath429 . thus , @xmath430 since @xmath431 is increasing in @xmath3 .
thus , @xmath432\right ) \leq c_*^{\lfloor t/{\delta t } \rfloor } = e^{-c_2 t / r^2}\ ] ] for the other bound , take @xmath433 and define @xmath434 \textrm { and } |w(t+\delta
t)|<r/2 \nonumber\\ \tilde{q } & = & \textrm { event that } |b(s)|<r \textrm { for } s \in [ 1 , 1 + r^2 ] \textrm { and } |b(1+r^2)|<r/2 \nonumber.\end{aligned}\ ] ] observe that @xmath435 where @xmath436 is independent of @xmath3 by brownian scaling . hence ,
@xmath437 ) \geq \pp(|w(0)|<r/2 ) \tilde{c}^{\lceil t/\log(1+r^2 ) \rceil } \geq e^{-c_1 t / r^2}.\ ] ] for the upper bound , take @xmath438 and define @xmath439 \textrm { and } |w(t+\delta t)|>2r \nonumber\\ \tilde{q } & = & \textrm { event that } |b(s)|>r\sqrt{1 + 1/r^2 } \textrm { for } s \in [ 1 , 1 + 1/r^2]\textrm { and } |b(1 + 1/r^2)|>2r\sqrt{1 + 1/r^2 } \nonumber\\ \tilde{q}^\prime & = & \textrm { event that } |b(s)|>r^2 + 1/2 \textrm { for } s \in [ 1 , 2]\textrm { and } |b(2)|>2r^2 + 1 \nonumber.\end{aligned}\ ] ] now observe that : @xmath440 where @xmath441 is a positive constant and the last inequality follows from brownian scaling and the inequality @xmath442 .
it follows that @xmath443 ) \geq \pp(|b(1)|>2r ) c_*^{\lceil t/\delta t \rceil } \geq e^{-c_1 r^2 t}\ ] ] for the lower bound , we observe @xmath429 at times @xmath444 and then bound the probability that @xmath445 for all @xmath446 . specifically ,
if we observe @xmath429 at time @xmath447 and @xmath448 , then we set @xmath224 where @xmath449 .
if @xmath450 then we halt the observation process and define @xmath231 .
we are interested in bounding @xmath451 from above .
observe that if @xmath369 then we may write : @xmath452 is independent of @xmath453 .
thus , @xmath454 and we see that if @xmath455 then @xmath456 let @xmath457 so that for @xmath458 we have @xmath459 by ( [ est ] ) and ( [ sd2 ] )
. now let @xmath242 be i.i.d .
geometric random variables satisfying @xmath460 for @xmath461 .
define @xmath241 and set @xmath462 .
it follows that @xmath242 stochastically dominates the conditional distribution of @xmath225 given @xmath463 for all @xmath464 , so @xmath465 .
using the following lemma it remains only to prove lemma [ lem1 ] .
let @xmath468 be i.i.d .
bernoulli random variables satisfying @xmath469 .
define @xmath470 and @xmath471 .
observe that we may construct the @xmath468 s so that @xmath472
. then the event @xmath473 corresponds to the event that the @xmath468 process yields more than @xmath474 1 s before 2 consecutive 0 s , and the lemma follows easily . as @xmath484 we have @xmath485 | b(1 )
= r-\rho/2 ) \rightarrow 1\ ] ] so for sufficiently large @xmath3 we obtain @xmath486 where @xmath317 . using the approximation @xmath487
we have @xmath488 thus , @xmath489 is bounded away from zero by a positive constant .
it follows that @xmath490 ) \geq
e^{-c \lfloor \frac{t}{\log(1 + 1/r^2 ) } \rfloor } \geq e^{-c_1 r^2t}.\ ] ] observe that @xmath491 } w_x(s ) < r \textrm { and } w_x(1)<r/2
| w_x(0)=r/2\right ) & \geq & \pp\left ( \sup_{s \in [ 0,e-1 ] } b_x(s ) \leq \left ( \sqrt{e}-1 \right ) \frac{r}{2 } \right ) \\ & = & 1 - 2 \pp \left ( b_x(e-1 ) \geq \left ( \sqrt{e}-1 \right ) \frac{r}{2 } \right ) \\ & \geq & 1 - e^{-c_1 r^2 } \geq \exp \left[-e^{-c_2 r^2 } \right]\end{aligned}\ ] ]
it follows that @xmath492 } w_x(s)<r \right ) & \geq & \pp(w_x(0)<r/2 ) \exp \left[-e^{-c_2 r^2 } \right]^{\lfloor t \rfloor } \\ & \geq & \exp\left [ -t e^{-cr^2 } \right].\end{aligned}\ ] ] | the _ hole probability _ that the zero set of the time dependent planar gaussian analytic function @xmath0 where @xmath1 are @xmath2 complex valued ornstein - uhlenbeck processes , does not intersect a disk of radius @xmath3 for all @xmath4 $ ] decays like @xmath5 .
this result sharply differentiates the zero set of @xmath6 from a number of canonical evolving planar point processes .
for example , the hole probability of the perturbed lattice model @xmath7 where @xmath8 are i.i.d .
ornstein - uhlenbeck processes decays like @xmath9 .
this stark contrast is also present in the _ overcrowding probability _ that a disk of radius @xmath3 contains at least @xmath10 zeros for all @xmath4 $ ] . |
though the mass of the top quark has been measured precisely , its coupling to other particles and production and decay schemes have not been probed in detail so far .
the top s large mass @xmath1 gev , and hence its large coupling to the higgs and longitudinal component of the @xmath2 boson makes it a sensitive probe of the mechanism of electroweak symmerty breaking ( esb ) .
top quark properties are an important input to electroweak precision analysis .
alternative mechanisms of esb and many new physics scenarios incorporate a special role for the top quark .
it is thus important to determine if the couplings of the top follow those of the first two generations as in the standard model ( sm ) or different electroweak and/or strong couplings apply in this case .
an important property of the top is that it decays before hadronization ( with a lifetime @xmath3 which is an order of magnitude smaller than the hadronization lifetime of @xmath4 ) . as a result the spin of the _ bare _ top is not diluted by hadronization , but is imprinted on characteristic angular distributions of its decay products , the @xmath2 boson and @xmath5 quark . for a review see @xcite .
the degree of top polarization depends on its production process and thus can probe sm / bsm scenarios , since any couplings of the top to new particles can alter its degree of polarization . in this talk
, we investigate the effects on top polarization , via decay lepton angular distributions , due to the presence of a high mass vector resonance , like the @xmath6 in littlest higgs models .
the polarization of the top is detected through the angular distribution of its decay products .
mass reconstruction is better for the hadronic decay channel but it suffers from a large qcd background .
the leptonic channel is cleaner but mass reconstruction is difficult due to missing energy . for a @xmath7 final state at a hadron collider
the optimal detection channel is semileptonic with @xmath8 and @xmath9 jets .
top polarization can allow measurements of parameters of the model / new physics and can give more information on the production mechanism than just the cross section .
it requires parity violation and thus measures left - right mixing and can probe @xmath10 violation through dipole couplings . even in parity conserving qcd , though the top is produced unpolarized ( at tree level ) , there are different rates for opposite helicity and same helicity @xmath7 production , i.e , the spin of the @xmath11 is correlated with the @xmath12 .
top spin can be measured by the angular decay distribution of the particle / jet @xmath13 in the rest frame of the top . for hadronic @xmath7 production ,
spin correlations between the decay leptons from the @xmath11 and @xmath12 have been extensively studied @xcite .
these spin correlations measure the asymmetry between the production of like and unlike helicity pairs of @xmath7 which can probe new physics in top pair production
. however , this requires the reconstruction of the @xmath11 and @xmath12 rest frames , which is difficult , if not impossible , at the lhc . here
we investigate if _ single top polarization _ can be a qualitative or quantitative probe of new physics and can provide better statistics for top pair production at the lhc .
new physics can also appear in the @xmath0 decay vertex , apart from that in top production , leading to changed decay width and distributions for the @xmath14 and @xmath15 .
the model independent form factor for the @xmath0 vertex can be parametrised as @xmath16 \label{anomaloustbw}\ ] ] where for the sm @xmath17 and the anomalous couplings @xmath18 .
the simultaneous presence of new physics in top production and decay can complicate the analysis making it difficult to isolate new couplings of the top .
however it has been proved that _
angular distributions _ of charged leptons/@xmath19 quarks from top decay are _ not affected by the anomalous @xmath20 vertex _ , see @xcite and references therein .
this has been shown very generally for a @xmath21 process and assumes the narrow width approximation ( nwa ) for the top and neglects terms quadratic in the anomalous couplings in ( [ anomaloustbw ] ) , assuming new physics couplings to be small .
this implies that charged lepton angular distributions , a _ decoupled observable _ , are more accurate probes of top polarization , and thus to new physics in top production alone .
in contrast , the energy distributions of the @xmath15 or the angular distributions of the @xmath5 and @xmath2 are `` contaminated '' by the anomalous @xmath0 vertex .
invariant mass distribution in gev for top pair production at the lhc for the total ( left ) and polarized ( right ) cross sections with @xmath22 gev and @xmath23.,title="fig : " ] invariant mass distribution in gev for top pair production at the lhc for the total ( left ) and polarized ( right ) cross sections with @xmath22 gev and @xmath23.,title="fig : " ] using the spin density matrix formalism for top production and decay ( which retains the spin information of the decaying quark ) and the nwa , the amplitude squared can be factored into production and decay parts as @xmath24 where @xmath25 and @xmath26 are the @xmath27 production and decay spin density matrices .
after phase space integration of @xmath25 the resulting top production density matrix @xmath28 can be used to define the polarization density matrix @xmath29 where the top longitudinal polarization is @xmath30 .
the transverse polarization in the production plane is @xmath31 while the transverse polarization perpendicular to the production plane is @xmath32 .
the @xmath33 s can be calculated from the decay lepton angular distributions by a suitable combination of polar and azimuthal angular asymmetries , for details see @xcite .
this requires partial or complete reconstruction of the top momentum in the lab frame .
alternatively , the azimuthal distribution of the lepton in the lab frame can be used to probe top polarization .
@xmath35 production at the lhc proceeds therough @xmath36 . in models with an extra heavy gauge boson , @xmath34
, top pair production receives an extra contribution in the @xmath37 channel apart from the @xmath38 and gluon . for concreteness
we choose the littlest higgs model where the left handed couplings to quarks of the resonance @xmath6 can be parametrized as @xmath39 for @xmath40 , where @xmath41 is the electroweak coupling and @xmath42 is a mixing angle in the model @xcite .
the @xmath6 mass and coupling @xmath43 completely specify @xmath35 production and decay dynamics .
1 shows the effects of the @xmath6 resonance on the total cross section and the difference between the cross sections for the left and right handed top s . understandably , the resonace is sharper for the polarized cross section because of the absence of the large qcd background .
distribution for the sm and littlest higgs model for right and left handed top-@xmath44 couplings ( left ) and the azimuthal asymmetry @xmath45 as a function of @xmath46 ( right).,title="fig : " ] distribution for the sm and littlest higgs model for right and left handed top-@xmath44 couplings ( left ) and the azimuthal asymmetry @xmath45 as a function of @xmath46 ( right).,title="fig : " ] we now consider angular distributions of the lepton in the lab frame , since there are easily measured experimentally at the lhc ( considering inclusive decay of the @xmath12 ) .
they are a pure probe of top polarization , since they are independent of any possible new physics in top decay
. however , the polar angle , measuring the forward backward asymmetry with respect to boost direction in the parton cm frame , is only mildly sensitive to the magnitude of top polarization and almost insensitive to the sign of polarization .
this leaves the lepton azimuthal angle @xmath47 measured with respect to the @xmath35 production plane .
2 shows the normalized @xmath47 distribution for the sm and the littlest higgs for @xmath48 gev and @xmath23 .
the peak at @xmath49 is sensitive to the magnitude of top polarization @xmath50 and can distinguish between right and left handed tops ( @xmath51 in this case resulting in the @xmath47 distribution being symmetric around @xmath52 ) .
we can thus construct a normalized azimuthal asymmetry sensitive to top polarization for comparision with the sm case , by taking the difference between cross sections in the ( 1,4 ) and ( 2,3 ) quadrants : @xmath53 . a plot of the azimuthal asymmetry for different @xmath43 is shown in fig .
2 as function of @xmath6 mass . @xmath45
reflects the degree of top polarization , saturating to the sm value for large @xmath46 and equalling it around 600 gev .
we find a large sensitivity for the variable @xmath54 , the difference between the azimuthal asymmetries in the littlest higgs and sm .
work is currently underway to study spin sensitive variables to qualitatively and quantitatively probe top polarization .
we acknowledge support from the indo - french project ifcpar no .
rmg acknowledges dst , india for support under grant no .
sr / s2/jcb-64/2007 and kr thanks the academy of finland for partial support under project no .
115032 . the work of r.k.s .
is supported by the german bmbf under contract 05ht6wwa .
w. bernreuther , j. phys .
g * 35 * , 083001 ( 2008 ) [ arxiv:0805.1333 [ hep - ph ] ] ; m. beneke et al .
2000 [ arxiv : hep - ph/0507207 ] ; s. willenbrock , arxiv : hep - ph/0211067 .
g. mahlon and s.j .
parke , phys .
d * 53 * , 4886 ( 1996 ) [ arxiv : hep - ph/9512264 ] ; phys .
b * 411 * , 173 ( 1997 ) [ arxiv : hep - ph/9706304 ] ; r. m. godbole , s. d. rindani and r. k. singh , jhep * 0612 * , 021 ( 2006 ) [ arxiv : hep - ph/0605100 ] .
t. han , h. e. logan , b. mcelrath and l. t. wang , phys .
d * 67 * , 095004 ( 2003 ) [ arxiv : hep - ph/0301040 ] . | we investigate the effects of new physics scenarios containing a high mass vector resonance on top pair production at the lhc , using the polarization of the produced top .
in particular we use kinematic distributions of the secondary lepton coming from top decay , which depends on top polarization , as it has been shown that the angular distribution of the decay lepton is insensitive to the anomalous @xmath0 vertex and hence is a pure probe of new physics in top quark production .
spin sensitive variables involving the decay lepton are used to probe top polarization .
some sensitivity is found for the new couplings of the top .
address = center for high energy physics , iisc , bangalore , 560 012 , india address = physical research laboratory , navarangpura , ahmedabad 380 009 , india address = helsinki institute of physics , p.o .
box 64 , helsinki 00014 , finland address = institut fr theoretische physik und astronomie , universitt wrzburg , 97074 , germany |
the theoretical understanding of electronic structure of strongly correlated materials have been a long standing problem in condensed matter physics .
local density approximation ( lda ) and generalized gradient approximation ( gga ) based on density functional theory ( dft ) give very reasonable ground state properties including band structures of weakly correlated materials.@xcite however for strongly correlated materials , both approximations fail to predict the experimentally observed insulating ground state of parent materials ( la@xmath0cuo@xmath1 , coo and feo ) and also underestimate the magnitude of magnetic moment.@xcite this is due to that the fact that both approximations account the correlation effects only in limited extent . to include the correlation effects beyond lda / gga
, several research groups have added the on - site repulsion interaction parameter @xmath2 ( so called dft+@xmath2 method).@xcite by adjusting @xmath2 , one can understand the various experimentally observed properties of @xmath6 or @xmath9 electron systems .
however , a major problem in dft+@xmath2 method is the double counting ( dc ) interactions , which is removed by the around mean field ( weakly correlated systems ) and fully localized ( strongly correlated systems ) schemes.@xcite also , because of its naturally static treatment of electronic correlation , dft+@xmath2 approach fails to reproduce the spectral properties .
for example , this method fails to reproduce the experimentally reported quasiparticle peak of 5@xmath9 character of @xmath10-pu near the fermi level.@xcite hence , in order to reproduce the spectral and thermodynamic properties of strongly correlated materials , a more sophisticated methods are needed to account the correlation effects beyond these methods . the most successful method is the dynamical - mean - field theory ( dmft ) , which describe the correlation effects in a periodic lattice by a strongly interacting impurity coupled to a self consistent bath.@xcite merging this with dft method , dft+dmft approach provides more material specific predictions of correlation effects in solids.@xcite using this approach , several groups have explained the electronic structure as well as various physical properties of transition metal , lanthanides and actinides compounds.@xcite despite of the many success of this method , wang @xmath11 @xmath12 . and dang @xmath11 @xmath12 . have shown that the mott insulating state in early and late transition metal oxides are described by tuning of several parameters , including dc and interaction @xmath2.@xcite however , recently haule @xmath11 @xmath12 .
have developed a method , where no fine tuning of parameter is required to predict the mott gaps in early transition metal oxides .
they have shown that for fixed value of @xmath2 , mott gaps of order of experimental data are observed for these compounds.@xcite strongly correlated vanadium spinels , av@xmath0o@xmath1 ( a = zn , cd and mg ) have been thoroughly studied both theoretically and experimentally because of the presence of large variety of physical properties such as electronic , structural and magnetic etc.@xcite . at a room temperature , these compounds shows face - centered - cubic structure .
geometrical frustration is seen in the antiferromagnetically coupled v ions with spin @xmath13=1 reside on the vertices of corner - sharing tetrahedra forming a pyrochlore lattice.@xcite in these system , the interplay of spin , electron , orbital and lattice degrees of freedom leads to two successive phase transitions .
the structural and magnetic transition takes place from cubic to tetragonal and paramagnetic to anti - ferromagnetic , respectively.@xcite the structural ( magnetic ) transition temperature _ @xmath14 _ ( _ @xmath15 _ ) for znv@xmath0o@xmath1 , mgv@xmath0o@xmath1 , and cdv@xmath0o@xmath1 are @xmath750 k ( @xmath740 k ) , @xmath765 k ( @xmath742 k ) and @xmath797 k ( @xmath735 k).@xcite however , their curie - weiss temperature are found to be much higher than the room temperature.@xcite _ @xmath14_@xmath8_@xmath15 _ , indicates that the certain degree of geometrical frustration is still present in these systems .
the degree of geometrical frustration is measured by the frustration index , which is the ratio of curie - weiss temperature and _ @xmath15_. the frustration index is largest for znv@xmath0o@xmath1 ( 21.3 ) and lowest for cdv@xmath0o@xmath1 ( 11.4 ) as compared to mgv@xmath0o@xmath1 ( 14.3).@xcite the experimentally observed magnetic moment per v atom ( band gap ) for znv@xmath0o@xmath1 , mgv@xmath0o@xmath1 and cdv@xmath0o@xmath1 are @xmath70.63 @xmath16@xmath17 ( @xmath70.32 ev , @xmath71.1 ev ) , @xmath70.47 @xmath16@xmath17 ( @xmath70.36 ev , @xmath71.08 ev ) and @xmath71.19 @xmath16@xmath17 ( @xmath73.16 ev ) , respectively.@xcite normally the degree of localization of electrons is measured by the ratio of transfer integral ( @xmath18 ) between neighbouring site and coulomb interaction @xmath2 .
canosa @xmath11 @xmath12 . have shown qualitatively that the order of degree of localization of electrons of v site for cdv@xmath0o@xmath1@xmath8znv@xmath0o@xmath1@xmath8mgv@xmath0o@xmath1.@xcite while we have argued quantitatively in earlier paper that the degree of localization of electrons of v site is largest for cdv@xmath0o@xmath1 and smallest for znv@xmath0o@xmath1 .
in our study , we have calculated the parameter @xmath2 self - consistently , while the parameter @xmath18 is calculated by linear fit of neighbour exchange coupling constant versus inverse of @xmath2.@xcite in order to understand the various physical properties as mentioned above for these compounds , several groups have proposed different mechanisms.@xcite . most of these mechanisms are based on the model calculations , which are fully parameters dependent .
the various parameters used in these calculations are not material specific .
similarly , some @xmath19 @xmath20 electronic structure calculations have also been performed in these systems , where normally @xmath2 and @xmath3 are used as a adjustable parameters .
there is no systemically comparative studied of these compounds available in the literature , where the parameters , @xmath2 and @xmath3 are material specific .
hence , it is important to know the material specific parameters to understand the various comparative study of these spinels . recently , in self - consistent electronic structure calculations , we have computed the material specific parameter @xmath2 for these compounds .
the calculated values of @xmath2 reported in our paper for znv@xmath0o@xmath1 , cdv@xmath0o@xmath1 and mgv@xmath0o@xmath1 are @xmath75.9 ev , @xmath75.7 ev , and @xmath76.2 ev , respectively.@xcite here , in present work we try to understand the comparative electronic behaviour of these spinels by using dft+dmft methods , where the self - consistently computed material specific parameters are used . for znv@xmath0o@xmath1 and mgv@xmath0o@xmath1 ,
the calculated values of @xmath5 are in good agreement with the experimental data .
close to fermi level , the average impurity hybridization function of v site is largest for znv@xmath0o@xmath1 and smallest for cdv@xmath0o@xmath1 .
hence the order of degree of itinerancy is found to be znv@xmath0o@xmath1@xmath8mgv@xmath0o@xmath1@xmath8cdv@xmath0o@xmath1 , which is consistent with our earlier reported data .
the position of lower hubbard band observed in the present study is also consistent with the experimental data for znv@xmath0o@xmath1 .
the paramagnetic electronic structure calculations of vanadium spinels carried out in face - centered cubic ( fcc ) phase have been divided into two parts , dft and dft+dmft .
dft part of the calculations are preformed by using the full - potential linearized augmented plane - wave ( fp - lapw ) method as implemented in wien2k code.@xcite the exchange - correlation functional has been treated within the gga of pbesol.@xcite the experimentally observed crystal structure for these spinels are taken from the literature.@xcite the muffin - tin sphere radii used for zn , cd , mg , v and o atoms are 2.0 , 2.46 , 1.39 , 2.0 and 1.54 bohr , respectively .
k - point grid size of 1000 points in the whole brillouin zone have been used here .
every calculations are converged with charge convergence per cell below 10@xmath21 electronic charge . to see the effect of spin - orbit coupling ( soc ) on zn 3@xmath6 states for znv@xmath0o@xmath1 and cd 4@xmath6 states for cdv@xmath0o@xmath1 , we have included it to the zn and cd atom , respectively . in order to described the itinerant and localized behaviours of strongly correlated electrons ,
we have used dft+dmft method within fp - lapw basis as implemented in wien2k code.@xcite such calculations are carried out in fcc phase of these compounds at room temperature , which are fully self - consistent in the electronic charge density and impurity levels .
continuous time quantum monte carlo impurity solver has been used to solve the auxiliary impurity problem.@xcite the exact dc scheme proposed by haule has been used here.@xcite v t@xmath22 shell are treated within dmft .
the values of on - site coulomb repulsion @xmath2 , hund s coupling @xmath3 and yukawa screening @xmath4 used for v atom in these compounds are shown in the table i. the values of @xmath2 were obtained by self - consistent calculations in our earlier paper.@xcite using these values of @xmath2 , the @xmath4 were uniquely determined through the matrix elements of the yukawa interaction in dmft basis .
similarly , the values of @xmath3 were uniquely determined by @xmath4 through yukawa form of coulomb interaction.@xcite hence , all the parameters used in the present study were material specific .
the convergence of the charge / cell of these systems are set to be less than 10@xmath21 electronic charge .
maximum entropy method to do analytical continuation of the self energy has been used to obtain the spectra on real axis , where we have used k - point grid size of 2000 points in the whole brillouin zone.@xcite .the values of material specific on - site coulomb repulsion @xmath2 , hund s coupling @xmath3 and yukawa screening @xmath4 parameters used for electronic structure calculations within dft+dmft for av@xmath0o@xmath1 ( a = zn , cd and mg ) compounds .
[ cols= " < ,
< , < , < " , ] [ table : the exp ] it is clear from the figure that an incoherent shoulder ( so called lower hubbard band ) appears at @xmath7 - 0.9 ev , @xmath7 - 1.1 ev and @xmath7 - 1.0 ev for znv@xmath0o@xmath1 , cdv@xmath0o@xmath1 and mgv@xmath0o@xmath1 compounds , which are slightly mismatch from experimental data .
however , by including the ib of 0.6 ev to tdos of these compounds , the position of the lower hubbard band are shifted by @xmath70.15 ev towards the lower energy side , which are in good agreement with experimental data for znv@xmath0o@xmath1 compound .
however , it is still @xmath70.2 ev mismatched from the experimental data for cdv@xmath0o@xmath1 compound .
the the behaviour of calculated tdos are in good agreement with the experiment from fermi level to @xmath7 - 4.5 ev for both compounds . however , below @xmath7 - 4.5 ev , the position of the two peaks are not matched with the experimental data , where similar to dft data both peaks are located at @xmath70.6 ev higher energy than experimental results for both compounds .
o@xmath1 , ( b ) cdv@xmath0o@xmath1 and ( c ) mgv@xmath0o@xmath1 compounds .
tdos computed by including instrument broadening ( ib ) are also shown for these compounds .
integrated background subtracted x - ray photoemission spectroscopy measurements data for znv@xmath0o@xmath1 and cdv@xmath0o@xmath1 are taken from the ref .
zero energy corresponds to the fermi level.,title="fig:",scaledwidth=45.0% ] [ ] now in order to match the position of lower hubbard band with the experimental result , we have increased the values of @xmath2 from the self - consistent calculated value of @xmath2 for cdv@xmath0o@xmath1 compound . for @xmath2=8 ev and @xmath3=1.1 ev
, the the position of the lower hubbard band are matched with the experimental data as also shown in the figure .
the slight mismatching of theoretically calculated lower hubbard band with the experiment data corresponding to self - consistent calculated parameters for cdv@xmath0o@xmath1 may be due to some other reasons , where careful study is needed in this direction .
the plot of pdos of zn 3@xmath6 , cd 4@xmath6 , v 3@xmath6 ( @xmath23/@xmath24 and @xmath24 orbitals ) and o 2@xmath25 states of these spinels computed within dft+dmft approach are shown in the fig .
6(a - f ) . for znv@xmath0o@xmath1 ,
the plot of pdos of zn 3@xmath6 , v 3@xmath6 and o 2@xmath25 states are divided into three distinct regions , i ( above fermi level ) , ii ( fermi level to @xmath7 - 7 ev ) and iii ( below @xmath7 - 7 ev ) . similarly , the plot of pdos of cd 4@xmath6 , v 3@xmath6 and o 2@xmath25 states of cdv@xmath0o@xmath1 are also divided into three distinct regions , i ( above fermi level ) , ii ( fermi level to @xmath7 - 8.5 ev ) and iii ( below @xmath7 - 8.5 ev ) . while it is divided into two distinct regions , i ( above fermi level ) and ii ( below fermi level ) for mgv@xmath0o@xmath1 .
now , in order to compare the pdos for these states , we have multiplied the pdos of region ii by a factor of four for these compounds .
it is clear from the figure that the quasiparticle peaks of v 3@xmath6 character crosses the fermi level within dft are disappeared in dft+dmft calculations .
hence , the band gap is created for these vanadates .
we will discuss the band gap for these compounds in the later part of the manuscript in more detail .
in regions i and ii , two peaks of dominant v 3@xmath6 character are sharply peaked around @xmath73 ev ( @xmath72.7 ev ) and @xmath7 - 0.9 ev ( @xmath7 - 1.1 ev ) , respectively for znv@xmath0o@xmath1 ( cdv@xmath0o@xmath1 ) .
similarly , they are sharply peaked around @xmath72.9 ev ( region i ) and @xmath7 - 1 ev ( region ii ) for mgv@xmath0o@xmath1 compound . in region
ii , zn 3@xmath6 , cd 4@xmath6 and o 2@xmath25 states are observed below @xmath7 - 3 ev for these spinels , where they are strongly mixed with v 3@xmath6 states .
the position of the sharp peaks observed here are not shifted appreciably from the dft results for these compounds . in region
iii , below @xmath7 - 7 ev for znv@xmath0o@xmath1 and below @xmath7 - 8.5 ev for cdv@xmath0o@xmath1 , the dominant electron character comes from zn 3@xmath6 and cd 4@xmath6 states , respectively .
the position of these states for both compounds are also not shifted appreciably from the dft calculations .
the plot of pdos of @xmath24 and @xmath23/@xmath26 orbitals are also shown in the fig .
6 , where it is divided into two distinct regions , i ( above fermi level ) and ii ( below fermi level ) . in region
i of cb , the incoherent shoulder of these orbitals are observed around @xmath72 ev for these compounds . in this region , among these orbitals , the large electronic character comes from @xmath24 orbital for znv@xmath0o@xmath1 and cdv@xmath0o@xmath1 , while it comes from @xmath23/@xmath26 orbitals for mgv@xmath0o@xmath1 . in this region ,
the energy of @xmath24 orbital is lower than @xmath23/@xmath26 orbitals for znv@xmath0o@xmath1 and cdv@xmath0o@xmath1 .
however for mgv@xmath0o@xmath1 , the energy of @xmath23/@xmath26 orbitals are found to lower than @xmath24 orbital . in region
ii of vb , all three orbitals are sharply peaked around @xmath7 - 0.9 ev , @xmath7 - 1.1 ev and @xmath7 - 1 ev for znv@xmath0o@xmath1 , , v 3@xmath6 and o 2@xmath25 states and ( b ) @xmath23/@xmath26 and @xmath24 orbitals of v atom for znv@xmath0o@xmath1 , ( c ) cd 4@xmath6 , v 3@xmath6 and o 2@xmath25 states and ( d ) @xmath23/@xmath26 and @xmath24 orbitals of v atom for cdv@xmath0o@xmath1 and ( e ) v 3@xmath6 and o 2@xmath25 states and ( f ) @xmath23/@xmath26 and @xmath24 orbitals of v atom for mgv@xmath0o@xmath1 compounds computed within dft+dmft calculations . for better comparison , pdos of zn 3@xmath6 , cd 4@xmath6 , v 3@xmath6 and o 2@xmath25 states in regions ii
are multiplied by a factor of four for these spinels .
zero energy corresponds to the fermi level.,title="fig:",scaledwidth=45.0% ] [ ] cdv@xmath0o@xmath1 and mgv@xmath0o@xmath1 , respectively . in this region , the opposite behaviour of these orbitals are observed as compared to the region i. the partial occupancy of @xmath24 orbital increases slightly from @xmath70.63 to @xmath70.65 to @xmath70.69 as compound changes from znv@xmath0o@xmath1 to cdv@xmath0o@xmath1 to mgv@xmath0o@xmath1 . while the partial occupancy of @xmath23/@xmath26 orbitals decreases slightly from @xmath70.69 to @xmath70.68 to @xmath70.66 as compound changes from znv@xmath0o@xmath1 to cdv@xmath0o@xmath1 to mgv@xmath0o@xmath1 .
however , the total occupancy of t@xmath22 orbitals is @xmath72.01 for these spinels .
the occupancy of @xmath24 , @xmath23/@xmath26 and t@xmath22 orbitals are about 1.2 , 2 and 1.6 times larger than dft results , respectively .
the different orbital occupancy of these orbitals from compound to compound in both approaches are due to the different v - o and v - v bonds and v - o - v angles in the trigonal distorted vo@xmath27 octahedra for these spinels .
different bonds and angles influence the orbital occupancy are due to the different ionic radii of zn , cd and mg ions in these compounds .
the calculated values of band gap ( @xmath5 ) for znv@xmath0o@xmath1 , cdv@xmath0o@xmath1 and mgv@xmath0o@xmath1 are @xmath70.9 ev , @xmath70.95 ev and @xmath71.15 ev , respectively . for znv@xmath0o@xmath1 and mgv@xmath0o@xmath1 compounds ,
the values of @xmath5 computed in the present study are about three times larger than the experimental data reported by rogers @xmath11 @xmath12 . however , the experimentally observed values of @xmath5 by pardo @xmath11 @xmath12 .
are @xmath71.1 ev , @xmath71.08 ev and @xmath73.16 ev for znv@xmath0o@xmath1 , mgv@xmath0o@xmath1 and cdv@xmath0o@xmath1 , respectively .
the calculated values of @xmath5 are close to the experimental values for znv@xmath0o@xmath1 and mgv@xmath0o@xmath1 compounds .
however , it is about three times smaller than the experimental data for cdv@xmath0o@xmath1 .
as mentioned above for @xmath2=8 ev and @xmath3=1.1 ev , the position of the lower hubbard band match to the experimental result for cdv@xmath0o@xmath1 compound .
corresponding to these parameters , dft+ dmft approach still underestimate the value of @xmath5 ( @xmath71.6 ev ) as compared to experimental result ( @xmath73.16 ev ) for this compound . experimentally , the value of @xmath5 depends on the activation energy for the resistivity , which itself depends whether the spinel sample is made in pure phase or not . however , it is always a challenging task to make the spinel sample in pure phase . in the light of these facts , it is difficult to compare the theoretically computed and experimentally observed values of @xmath5 for these compounds . for these parameters ,
the values of @xmath5 computed in the present study for these compounds are reasonable as corresponding to these parameters , our data is closely matched with the experiment in vb . here , it will be interesting to see whether the inverse photoemission spectroscopy measurements data will be helpful to explain our results in cb for these compounds . in order to compare the degree of itinerancy of v 3@xmath6 electrons in these compounds ,
we have computed the impurity hybridization function , which contains all information about the crystal environment for valance electrons on a given v site .
the plot of the average impurity hybridization function ( average over all t@xmath22 orbitals of v site ) orbitals of v site computed within dft+dmft calculations for av@xmath0o@xmath1 ( a = zn , cd and mg ) compounds .
zero energy corresponds to the fermi level.,title="fig:",scaledwidth=45.0% ] [ ] for these compounds is shown in the fig .
7 . below fermi level
, the average impurity hybridization function is sharply peaked around @xmath7 - 1 ev , @xmath7 - 1.1 ev and @xmath7 - 1.15 ev for znv@xmath0o@xmath1 , mgv@xmath0o@xmath1 and cdv@xmath0o@xmath1 , respectively . around these energies
, it is largest for znv@xmath0o@xmath1 and smallest for cdv@xmath0o@xmath1 as compared to mgv@xmath0o@xmath1 .
large value of the average impurity hybridization function indicates the more itinerant electronic character for the system .
hence , based on this , the degree of localization of v 3@xmath6 electrons decreases from cdv@xmath0o@xmath1 to mgv@xmath0o@xmath1 to znv@xmath0o@xmath1 .
this order is in accordance with our earlier results based on @xmath28 ratio , where the values of @xmath2 and @xmath18 are calculated self - consistently and the linear fit of nearest neighbour exchange coupling constant versus inverse of @xmath2 , respectively.@xcite however , our present and earlier results are not consistent with the results of canosa @xmath11 @xmath29 . based on localized - electron superexchange represented by @xmath3@xmath30@xmath31 ,
they have argued qualitatively that the degree of localization of v 3@xmath6 electrons is largest for cdv@xmath0o@xmath1 and smallest for mgv@xmath0o@xmath1 as compared to znv@xmath0o@xmath1.@xcite
in conclusion , the exact comparative behaviour of various physical properties of av@xmath0o@xmath1 ( a = zn , cd and mg ) compounds have not been found , where material specific parameters were used . in present work
, we have explored the comparative behaviour of electronic structures of these compounds by using dft+dmft approach , where material specific parameters were used .
the calculated order of @xmath5 was found to be mgv@xmath0o@xmath1 ( @xmath71.15 ev)@xmath8cdv@xmath0o@xmath1 ( @xmath70.95 ev)@xmath8znv@xmath0o@xmath1 ( @xmath70.9 ev ) , where the values of @xmath5 were close to experimental data for znv@xmath0o@xmath1 and mgv@xmath0o@xmath1 . among these spinels ,
the average impurity hybridization function of v site ( close to the fermi level ) were found to be largest for znv@xmath0o@xmath1 and smallest for cdv@xmath0o@xmath1 .
hence , the degree of itinerancy was highest for znv@xmath0o@xmath1 and lowest for cdv@xmath0o@xmath1 as compared to mgv@xmath0o@xmath1 compound .
this order of degree of itinerancy were found to be consistent with our earlier results . in the energy range of 0 to @xmath7 - 4.5 ev ,
almost a similar behaviour of calculated total density of states and experimental data were observed for these spinels , where the position of lower hubbard band was in good agreement with the experiment for znv@xmath0o@xmath1 .
the present study shows the importance of material specific parameters to understand the comparative electronic behaviour of these compounds .
s.l . is thankful to ugc , india , for financial support .
g. giovannetti , a. stroppa , s. picozzi , d. baldomir , v. pardo , s. blanco - canosa , f. rivadulla , s. jodlauk , d. niermann , j. rohrkamp , t. lorenz , s. streltsov , d. i. khomskii , and j. hemberger , phys .
b * 83 * , 060402(r ) ( 2011 ) .
p. blaha , k. schwarz , g. k. h. madsen , d. kvasnicka , and j. luitz , wien2k , an augmented plane wave plus local orbitals program for calculating crystal properties ( vienna university of technology , vienna , 2001 ) . | theoretically , various physical properties of av@xmath0o@xmath1 ( a = zn , cd and mg ) spinels have been extensively studied for last 15 years . besides of this
, no systematic comparative study has been done for these compounds , where the material specific parameters are used . here ,
we report the comparative electronic behaviour of these spinels by using a combination of density functional theory ( dft ) and dynamical mean - field theory ( dmft ) , where the self - consistent calculated coulomb interaction @xmath2 and hund s coupling @xmath3 ( determined by yukawa screening @xmath4 ) are used .
the main features , such as insulating band gaps ( @xmath5 ) , degree of itinerancy of v 3@xmath6 electrons and position of lower hubbard band are observed for these parameters in these spinels .
the calculated values of @xmath5 for znv@xmath0o@xmath1 , cdv@xmath0o@xmath1 and mgv@xmath0o@xmath1 are found to be @xmath70.9 ev , @xmath70.95 ev and @xmath71.15 ev , respectively , where the values of @xmath5 are close to experiment for znv@xmath0o@xmath1 and mgv@xmath0o@xmath1 .
the behaviour of calculated total density of states are in good agreement with the experiment in the energy range of 0 to @xmath7 - 4.5 ev for znv@xmath0o@xmath1 and cdv@xmath0o@xmath1 .
the position of lower hubbard band are observed around @xmath7 - 1.05 ev , @xmath7 - 1.25 ev and @xmath7 - 1.15 ev for znv@xmath0o@xmath1 , cdv@xmath0o@xmath1 and mgv@xmath0o@xmath1 , respectively , which are also in good agreement with the experimental data for znv@xmath0o@xmath1 .
close to the fermi level , the order of average impurity hybridization function of v site are found to be znv@xmath0o@xmath1@xmath8mgv@xmath0o@xmath1@xmath8cdv@xmath0o@xmath1 .
hence , the degree of itinerancy ( localization ) of v 3@xmath6 electrons is smallest ( largest ) for cdv@xmath0o@xmath1 and largest ( smallest ) for znv@xmath0o@xmath1 , which is in accordance with our earlier results .
hence , present work shows the importance of material specific parameters to understand the comparative electronic behaviour of these compounds . |
the use of effective hadronic models based on chiral symmetry has been quite successful in describing nuclear matter , finite nuclei and neutron stars ( see e.g. @xcite ) .
while most of the approaches are based on variations of the linear sigma model or its non - linear realization , work on an alternative ansatz starting from the parity - doublet model ( @xcite ) has been quite limited . in this approach
, the scalar sigma meson serves to split the nucleon and its parity partner , while in the chirally restored phase both baryonic states become degenerate but not massless .
this situation is quite analogous to sigma models with hidden gauge symmetry @xcite where the masses of vector and axial vector mesons are split in a similar way to attain degenerate ( but non - zero ) masses in the chirally restored phase similar to the pion and sigma mesons .
it has only quite recently been shown @xcite that with this approach one can obtain saturated nuclear matter with relatively high nuclear compressibilities with a simple potential for the scalar field and a cross - over transition to chiral restoration at large densities . in our investigation
we calculate symmetric nuclear matter and neutron star properties within this framework . in order to get a better understanding of the parameters involved and not just to fix an arbitrary fine - tuned parameter set we vary the parameters over a broad range of values and calculate nuclear matter properties and star masses .
the parameter scan shows that for quite a wide range of values a satisfactory reproduction of nuclear matter binding energy and saturation density is possible .
further demanding that the nuclear matter compressibility lies within a range of reasonable values ( assumed to be between 100 and 400 @xmath0 in the following ) , restricts the parameter space significantly but still allows for a range of values that yield a quantitatively good description of nuclear matter ground state properties in contrast to the results in @xcite . in order to study isospin asymmetric nuclear matter
the rho meson is included in the lagrangian .
its coupling to the nucleons is fixed by the phenomenological value for the asymmetry energy of about 32 mev . by further requiring that the maximum mass of the neutron star , determined by solving the tolman - oppenheimer - volkov ( tov ) equations , should have a mass around @xmath1 @xmath2
, we find additional restrictions for the allowed model parameters .
the article is organized as follows .
the parity doublet model is briefly introduced in the following section .
we then apply the equations to symmetric nuclear matter and to isospin - asymmetric matter as it is present in neutron stars .
the properties of the model are first tested to fit saturation quantities and then to reproduce observational properties such as the maximum mass of the neutron star .
finally , we discuss the results of this investigation and add some remarks on further work to be done .
in the parity doublet model the nucleons and their parity partners belong to the same multiplet and so are true chiral partners @xcite . under @xmath3 transformations @xmath4 and @xmath5 , the two nucleon fields @xmath6 and @xmath7 transform as :
@xmath8 this allows for a chirally invariant mass , @xmath9 : @xmath10 the full chiral lagrangian reads : @xmath11 where @xmath12 , @xmath13 , @xmath14 and @xmath15 are the coupling constants of the mesons fields ( @xmath16 , @xmath17 , @xmath18 and @xmath19 ) to the baryons @xmath6 and @xmath7 . note that we assume the same vector coupling strength for both parity partners , as it would result from minimal coupling of the vector mesons .
the mesonic lagrangian @xmath20 contains the kinetic terms of the different meson species , and potentials for the scalar and vector fields .
the potential for the spin zero fields is the same as in the ordinary su(2 ) linear sigma model .
kinetic and potential terms are added for an isoscalar vector meson , @xmath18 , as in the @xmath16-@xmath18 model of nuclear matter @xcite as well as for the isovector vector meson , @xmath19 : @xmath21 \nonumber\\ & + & \frac 12 \bar{\mu}\,^2 ( \sigma^2+\vec{\pi}^2 ) - \frac \lambda 4 ( \sigma^2+\vec{\pi}^2)^2 \nonumber \\ & + & \epsilon\sigma \ , \end{aligned}\ ] ] where @xmath22 and @xmath23 represent the field strength tensors of the vector fields . as usual , the parameters @xmath24 , @xmath25 and @xmath26 can be related to the sigma and pion masses , and the pion decay constant , in vacuum : @xmath27 with @xmath28 @xmath0 , @xmath29 @xmath0 and the vacuum expectation value of the sigma field @xmath30 . since the mass of the @xmath16 meson in the vacuum can not be fixed precisely by experiment , we will treat it as a free parameter .
the vacuum mass of the @xmath18 field is @xmath31 @xmath0 and the vacuum mass of the @xmath19 field is @xmath32 @xmath0 . to investigate the properties of dense nuclear matter and the chiral phase transition at zero temperature
, we adopt the mean - field approximation @xcite .
the fluctuations around constant vacuum expectation values of the mesonic field operators are neglected , while the nucleons are treated as quantum - mechanical one - particle operators .
only the time - like component of the isoscalar vector meson @xmath33 and the time - like third component of the isovector vector meson @xmath34 survive , assuming homogeneous and isotropic infinite nuclear matter . additionally , parity conservation demands @xmath35 .
the mass eigenstates for the parity doubled nucleons , the @xmath36 and @xmath37 are determined by diagonalizing the mass matrix , eq .
( [ chinvmass ] ) , for @xmath6 and @xmath7 . writing the coupling constants
a and b as a function of @xmath38 , @xmath39 and @xmath40 , the effective masses of the baryons are given by : @xmath41\frac{\sigma^2}{\sigma_0 ^ 2}+m_0 ^ 2}\nonumber\\ & \pm&\frac{m_{n_+}-m_{n_-}}{2}\frac{\sigma}{\sigma_0}\ , \end{aligned}\ ] ] if chiral symmetry is completely restored , i.e. @xmath42 , the two nucleonic parity states become degenerate in mass with @xmath43 . thus
, for large values of @xmath9 , the sigma field mainly serves to split the nucleon and its parity partner nucleon masses and contributes less to the ( dynamical ) mass of the nucleon @xcite .
independent of the value for @xmath9 , in the vacuum , i.e. for @xmath44 , the baryon effective masses reproduce their measured values .
the thermodynamic potential per unit volume is : @xmath45 where @xmath46 denotes the nucleon type , @xmath47 is the fermionic degeneracy , @xmath48 the energy , and @xmath49 the corresponding effective chemical potential .
the single particle energy of each parity partner @xmath50 is given by @xmath51 .
the mean meson fields @xmath52 , @xmath53 and @xmath54 are determined by extremizing the thermodynamic potential @xmath55 : @xmath56 the scalar density @xmath57 and the baryon density @xmath58 for each chiral partner are given by the usual expressions : @xmath59 \ , \nonumber \\
\rho_i&=&\gamma_i \int_0^{k_{f_i } } \frac{d^3k}{(2 \pi)^3}=\frac{\gamma_i k^3_{f_i}}{6 \pi^2 } \ .
\label{densities}\end{aligned}\ ] ] the basic nuclear matter saturation properties we impose can be formulated in the following way . at the point of vanishing pressure we have @xmath60 note that in contrast to former work in @xcite , here we allow for a reasonable range of values for the different observables .
first we apply the model to symmetric nuclear matter and determine the ranges of the parameters that reproduce the saturation properties for binding energy per nucleon and baryon density as given by eq .
[ nucmatprop ] . in principle
the free parameters of the model are the coupling constant of the vector meson to the nucleons @xmath61 , the mass of the chiral partners @xmath39 ( as the identification with a specific experimentally known resonance is not completely unambiguous ) , the bare mass @xmath9 , the coupling constant of the fourth order self interactions of the vectorial fields @xmath62 and the mass of the scalar meson @xmath63 as already described in @xcite .
however , to keep things simple , we will for most of this work restrict our analysis to parameter sets with @xmath64 and @xmath65 @xmath0 . where necessary or interesting
we will discuss the influence of finite values of @xmath62 or smaller values of @xmath39 , as was also done in @xcite .
the other parameters are varied over all possible values . in fig .
( [ gnovsm0 ] ) and fig .
( [ msvsm0 ] ) we show the values of the corresponding parameters vs mass parameter @xmath9 , for which the imposed nuclear matter saturation properties are fulfilled . in both figures ,
the light grey bands represent the allowed regions without any further restriction , while the dark grey bands depict the parameter sets , which in addition yield a reasonable nuclear matter compressibility @xmath66 at saturation between 100 and 400 mev .
value of the nucleon vector coupling constant @xmath61 vs mass parameter @xmath9.,width=340 ] as one can see , successful fits without restricting the compressibility are possible for @xmath67 mev . this region is larger than that found in @xcite because no fine - tuning to fixed value for the nuclear matter saturation properties has been done here .
the corresponding nucleon - vector coupling constant @xmath61 as well as the value of the sigma mass in vacuum @xmath63 decrease with increasing values of @xmath9 , as has also been observed in @xcite .
mass of sigma meson in vacuum vs mass parameter @xmath9.,width=340 ] demanding the nuclear matter compressibility to be in
reasonable ranges restricts the possible parameter values . for our choice of @xmath68 , we see that only the fits with quite large @xmath9 values ( @xmath69 mev ) survive . nuclear compressibility at saturation vs mass parameter @xmath9.,width=340 ] this can be seen in more detail from fig .
( [ kvsm0 ] ) . for smaller values of @xmath9
the nuclear matter compressibility increases very fast beyond any acceptable value . in order to study neutron star properties we calculate the equation of state ( eos ) of charge neutral matter including leptons . due to the different densities of neutrons and protons ,
the expectation value of the vector - isovector meson @xmath19 does not vanish .
the corresponding coupling constant is determined such that an asymmetry energy at saturation of @xmath70 @xmath0 is obtained .
neutron star masses and radii are determined by solving the tolman - oppenheimer - volkov equations @xcite .
the maximum star mass for a given equation of state is shown as a function of @xmath9 in fig .
( [ nstarmass ] ) for the cases that have a compressibility at saturation @xmath71 between @xmath72 and @xmath73 @xmath74 . as can be seen , the maximum mass of the star decreases with increasing @xmath9 .
this results from the fact that for larger bare mass term @xmath9 , medium effects are reduced which leads to a smaller compressibility as can be seen in fig .
[ kvsm0 ] and reduced repulsion at high densities and consequently lower star masses .
thus , demanding the compressibility to be in a reasonable range , especially not to be too big and in addition , to demand to have a maximum star mass above a phenomenologically acceptable value of at least 1.4 to 1.6 solar masses , restricts the value of @xmath9 to be somewhere between 900 and 870 mev .
thus the initially broad band of possible parameter sets is strongly narrowed by these additional observables .
note , that these values of @xmath9 are quite large , which means that the mass generation is very much different from a standard linear @xmath16 model , where the mass is more or less exclusively generated by the dynamical symmetry breaking , while now most of the mass is generated by the parity - mixing mass term .
the results for neutron stars show that keeping a high mass for the chiral partners and a small coupling constant of the fourth order self interactions of the vectorial fields produce more massive stars , being in agreement with phenomenology .
mass of neutron stars vs mass parameter @xmath9.,width=340 ] in figure [ nstarden ] , the central densities of the neutron stars and the threshold densities above which the @xmath37 states are populated , respectively , are shown .
the maximum mass stars have a typical central energy density of around @xmath75 which for the cases at hand with @xmath64 and @xmath76 mev is always smaller than the critical density to create chiral partners and in this case no @xmath37 are present in the star as can be seen in the particle densities shown in fig .
however as has already been observed in @xcite , decreasing the mass of the @xmath37 , reduces the threshold density and thus can allow for the population of @xmath37 states in neutron stars .
this can for example be seen in fig .
[ pop2 ] , for @xmath77 @xmath0 , where the chiral partners become populated already slightly below @xmath78 .
the mass - radius relation for the two cases is shown in fig .
( [ masscrust ] ) . besides the eos for the dense part of the star , the eos for an outer crust , an inner crust and an atmosphere
have been added @xcite .
central densities of neutron stars vs mass parameter @xmath9 and critical density for chiral restoration or density for appearance of n- , respectively.,width=340 ] @xmath79 as can be seen , the maximum mass of the star is higher for @xmath65 @xmath0 ( @xmath80 @xmath81 ) compared with the case for @xmath77 @xmath0 ( @xmath82 @xmath81 ) , as expected .
however , both masses are around @xmath1 @xmath81 , value that is in agreement with the massive stars observed lately with the heaviest one so far having a mass of @xmath83 @xmath81 @xcite .
@xmath79 @xmath79 @xmath79 @xmath79 @xmath79 population of the star vs baryon density for @xmath84 @xmath0 and @xmath65 @xmath0 .
the central density of the respective star would be @xmath85 . ]
population of the star vs baryon density for @xmath86 @xmath0 and @xmath87 @xmath0 .
the central density of the respective star would be @xmath88.,width=302 ] mass radius diagram for different chiral partners masses.,width=302 ]
we applied the parity doublet model to nuclear and neutron star matter .
we were able to achieve a quantitatively good description of saturated nuclear matter as well as to obtain maximum masses of neutron stars that are in agreement with recent observations . in our approach
we explored a wide range of model parameters that reproduce saturation properties of nuclear matter for a window of values for the density ( @xmath89 ) and binding energy per nucleon ( @xmath90 ) . with
that we could study the dependence of results on the free parameters of the model : the coupling constant of the vector meson to the nucleons @xmath61 , the bare mass @xmath9 and the mass of the scalar meson @xmath63 .
initially the mass of the chiral partners is assumed to be @xmath91 , corresponding to one of the likely candidates listed in the particle data book .
no vector meson self - interactions have been taken into account as they generally lead to very soft equations of state and small masses of neutron stars that are in disagreement with observation .
besides that , if we take into account phenomenological values of compressibility ( allowing for a somewhat wider range @xmath92 ) we find that the mixing parameter @xmath9 has to be @xmath93 so that the two different couplings of the nucleon to the scalar field cancel each other to a large degree . in a second step the asymmetry between protons and neutrons has been included demanding a charge neutral state ( including the electron ) and the @xmath19 meson coupling constant is determined to reproduce the phenomenological value for the asymmetry energy at saturation .
if we consider a high mass for the nucleon chiral partners @xmath39 , the critical density above which those particles appear in the star is also high , and in the case of @xmath91 they do not appear in the star .
the value of the maximum mass is @xmath94 .
demanding that the @xmath95 should be at least 1.4 solar masses restricts the value of the explicit mass parameter @xmath9 to be less than 900 mev .
decreasing the mass of the chiral partners by about @xmath96 is enough to decrease the threshold density of the chiral partners for them to be present in the star , which reduces the maximum mass of the star moderately , in the case of @xmath97 to @xmath82 @xmath81 .
an interesting extension of this work would be the generalization to a flavor - su(3 ) approach .
work in this direction is in progress .
we thank g. pagliara and j. schaffner - bielich for valuable discussions .
the numerical work has been done at the computing facilities of the csc frankfurt .
p. papazoglou , d. zschiesche , s. schramm , j. schaffner - bielich , h. stoecker and w. greiner , phys .
c * 59 * ( 1999 ) 411 . ; s. schramm , phys .
b * 560 * ( 2003 ) 164 ; s. schramm , phys .
c * 66 * ( 2002 ) 064310 . | we investigate the properties of isospin - symmetric nuclear matter and neutron stars in a chiral model approach adopting the su(2 ) parity doublet formulation .
this ansatz explicitly incorporates chiral symmetry restoration with the limit of degenerate masses of the nucleons and their parity partners . instead of searching for an optimized parameter set we explore the general parameter dependence of nuclear matter and star properties in the model .
we are able to get a good description of ground state nuclear matter as well as large values of mass for neutron stars in agreement with observation . |
of all ground - based observatory sites that have been tested for submillimeter - wave sky quality , the best results have come from the amundsen - scott south pole station . in the current decade , a year - round observatory has been established at the pole by the center for astrophysical research in antarctica ( cara ) , an nsf science and technology center .
cara operates three major telescope facilities : ast / ro ( the antarctic submillimeter telescope and remote observatory , a 1.7-m telescope@xcite ) , python ( a cosmic microwave background experiment ) , and spirex ( the south pole infrared explorer , a 60-cm telescope ) .
these facilities are conducting site characterization and astronomical investigations from millimeter wavelengths to the near - infrared@xcite .
profiles of temperature , pressure , and water vapor above the pole have been measured at least daily for decades by the south pole meteorology office@xcite .
measured water vapor values are extremely low because the air is dessicated by frigid temperatures ( annual average : -49c , median winter pwv : 0.25 mm ) .
winds at the pole are unusually low ( maximum recorded wind speed over 30 years : 25 m / s ) , and rain is completely absent@xcite .
atmospheric opacity at the pole has been routinely measured at 225 ghz and 492 ghz using skydip techniques@xcite .
the 225 ghz results are comparable to those made at the proposed nrao millimeter array site at atacama , chile .
this , together with the significantly higher atmospheric pressure at the pole , necessarily implies that the pwv and submillimeter - wave opacity at the pole are somewhat better than at atacama . a wide - field
, off - axis 10 m diameter submillimeter - wave telescope has been proposed for the south pole by an international consortium .
the south pole 10 m ( sp 10 m ) telescope will be a powerful and highly flexible instrument , usable for a wide variety of submillimeter observations in observational cosmology , star formation , chemistry and dynamics of the interstellar medium , galactic structure , and solar system studies
. the sp 10 m telescope will be optically superior to existing telescopes with respect to scattered radiation , cross - polarization , and field of view and will be located at a site which is consistently much more transparent in the submillimeter - wave atmospheric windows than any other existing or proposed observatory site@xcite .
it will therefore be orders of magnitude faster than existing instruments and capable of detecting much weaker sources .
it will greatly enhance submillimeter - wave capability in the southern hemisphere and provide a single - dish complement to the smithsonian astrophysical observatory ( sao ) submillimeter array ( sma ) and the nrao millimeter array ( mma ) .
it will make use of large , sensitive bolometer and heterodyne array detectors which are advanced versions of those currently in use on the caltech submillimeter observatory ( cso ) and the james clerk maxwell telescope ( jcmt ) .
the 10 meter telescope will be able to feed 1000 pixels in a detector array .
a workshop was held at harvard university on 28 march 1997 to discuss science goals for the sp 10 m project .
some critical projects that drive the telescope design are : [ [ primary - cosmic - microwave - background - anisotropy - at - arcminute - scales . ] ] primary cosmic microwave background anisotropy at arcminute scales .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + at angular scales near 0.5 , standard cosmological models predict relatively large anisotropy caused by an acoustic bounce " that occurred before recombination .
the spatial frequency at which this peak occurs depends on the fundamental cosmological parameter @xmath0 .
additional smaller peaks are expected at multiples of the peak spatial frequency .
present cmbr observations show a clear maximum in the anisotropy spectrum at a spatial scale somewhere near 1 , but the spectrum lacks detail and the harmonic peaks at smaller scales have not yet been seen .
the sp 10 m will study the damping tail region at arcminute scales , providing a strong test of the acoustic oscillation model and providing measurements of peak positions that will augment the lower spatial frequency information obtained with meter - class telescopes .
data from the sp 10 m will be particularly vital if the universe turns out to be open , and the main spatial frequency peak occurs at scales near 0.1 . to study primary anisotropy at arcminute angular scales , it will be important to simultaneously understand the secondary contribution to anisotropy due to sunyaev - zeldovich ( s - z ) distortion in galaxy clusters between the recombination last scattering surface and the observer . because of the low sky noise available at the pole , the sp 10 m can be used over a wide range of frequencies , including the 200 - 300 ghz range of frequencies required to allow spectral separation of thermal s - z effect from primary anisotropy . since there are @xmath1 to @xmath2 visible clusters of galaxies , both the small beam size and wide spectral range of the sp 10 m will be needed to locate the cluster - free regions of sky that will be observed for primary anisotropy studies . [ [ sunyaev - zeldovich - effect . ] ] sunyaev - zeldovich effect .
+ + + + + + + + + + + + + + + + + + + + + + + + + + as cmb photons travel from the surface of last scattering to the observer , secondary anisotropies can arise due to the interaction of the cmb photons with intervening matter . of particular interest
is the s - z effect , which occurs when cmb photons travel through a cluster of galaxies@xcite .
approximately @xmath3% of the total mass of rich clusters of galaxies is in the form of hot ( @xmath4k ) ionized plasma .
compton scattering of cmb photons by electrons in this intra - cluster plasma can result in an optical depth as high as 0.02 , resulting in a distortion of the cmb spectrum at the mk level .
there are two components of the s - z effect which result from the two distinct velocity components of the scattering electrons .
the thermal component is due to the thermal ( random ) velocities of the scattering electrons .
the kinematic component is due to the bulk velocity of the intra - cluster gas with respect to the rest frame of the cmb .
the two s - z components have distinct spectra which can be separated by observations at millimeter wavelengths
. the thermal component of the s - z effect can be used in combination with x - ray data to provide a measure of the hubble constant ( @xmath5 ) .
in addition , when combined with a measurement of electron temperature , the ratio of the kinematic and thermal component amplitudes provides a direct measurement of the cluster s peculiar velocity relative to the rest frame of the cmb .
the observed surface brightness difference of both the thermal and kinematic components is independent of the cluster redshift , as long as the cluster is resolved . using the sp 10 m
, accurate s - z measurements can be made throughout the universe , all the way back to the epoch of formation of the hot intra - cluster gas .
this will be an important new probe of universal structure formation processes : a sample of peculiar velocities throughout the hubble volume .
[ [ continuum - detection - of - high - redshift - protogalaxies - not - detectable - in - the - visible - or - near - ir . ] ] continuum detection of high redshift protogalaxies not detectable in the visible or near - ir .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the majority of the luminosity resulting from the collapse energy of galaxies and the first generations of stars may not appear at visual or near - infrared wavelengths ( even in the rest frame of the distant galaxy ) , but may instead be reradiated by dust@xcite .
submillimeter - wave point sources with no bright visible counterpart may already have been detected@xcite . in the broadband rest - frame spectrum of normal galaxies
are two roughly equal flux peaks : visible light produced by stars , and far - infrared radiation near 100 produced by dust and the fine - structure cooling lines of the interstellar medium . to understand the nature of protogalaxies
, it will be necessary to observe them at submillimeter , near - infrared , and visible wavelengths .
the energetics of protogalaxy formation will not be understood until there is a deep , high - resolution survey of the submillimeter - wave background@xcite .
[ [ line - detection - of - high - redshift - protogalaxies - not - detectable - in - the - visible - or - near - ir . ] ] line detection of high redshift protogalaxies not detectable in the visible or near - ir .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the @xmath6 line of [ @xmath7 at @xmath8 is the brightest emission line in the spectrum of most galaxies . from the radio to x - ray , the wavelength of highest flux density in a galaxian spectrum is usually the peak of this line ; as much as 0.5% of the total luminosity of a galaxy can be emitted in the single spectral line@xcite . the [ @xmath7 line - to - continuum ratio is high in low metallicity systems , so that line searches for protogalaxies may be more efficient than continuum searches @xcite .
measurement of the [ @xmath7 line yields chemical and kinematic information for galaxies detected by other means . [ [ continuum - survey - of - cloud - cores . ] ] continuum survey of cloud cores .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + an unbiased wide - area continuum survey at @xmath9 of all nearby molecular regions to identify dense cloud cores should be able to locate all such regions within 500 pc of the sun .
this project can best be carried out at a wavelength of @xmath10 350 ; observing at this relatively long wavelength will ensure that even the most heavily obscured ( deeply embedded ) sources will be included .
the sp 10 m will be particularly capable of tracing extended protostellar disks , and connections between these disks and surrounding , placental material .
follow - on studies at 200 wavelength may help discriminate among protostellar models .
[ [ a - spectral - line - survey - of - detected - cloud - cores - in - kinematic - tracers - of - the - star - forming - gas . ] ] a spectral line survey of detected cloud cores in kinematic tracers of the star - forming gas .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + such a survey would enable discrimination among the variety of complex motions found in these regions , and in particular , identification of objects in which infall of material is occurring .
lrrrrrrrrr telescope & @xmath11 & @xmath12 & @xmath13 & & + & @xmath14 & @xmath15 & & & + & & & & @xmath16 & @xmath17 & @xmath18 & @xmath16 & @xmath17 & @xmath18 + + sp 10 m & 79 & 11 & 38 & _ 60 _ & _ 64 _ & _ 74 _ & _ 95 _ & _ 108 _ & _ 144 _ + sp 30 m & 711 & 4 & 4 & _ 7 _ & _ 7 _ & _ 8 _ & _ 11 _ & _ 12 _ & _ 16 _ + ast / ro & 2 & 65 & 92 & _ 2160 _ & _ 2300 _ & _ 2660 _ & @xmath19 & @xmath20 & @xmath21 + jcmt & 177 & 7 & 5 & 80 & 700 & _ 760 _ & @xmath22 & @xmath23 & @xmath24 + cso & 79 & 11 & 11 & _ 150 _ & 2000 & _ 2200 _ & @xmath25 & @xmath26 & @xmath27 + sofia & 5 & 44 & 50 & & _ 200 _ & _ 200 _ & & _ 800 _ & _ 800 _ + first & 7 & 32 & 4.3 & & _ 54 _ & _ 54 _ & & _ 678 _ & _ 583 _ + sma & 227 & 2 & 0.14 & _ 134 _ & _ 1142 _ & _ 1250 _ & @xmath28 & @xmath29 & @xmath30 + mk array@xmath31 & 483 & 0.5 & 0.02 & _ 59 _ & _ 719 _ & _ 790 _ & @xmath32 & @xmath33 & @xmath34 + `` small '' mma & 2010 & 0.2 & 0.08 & _ 7 _ & _ 45 _ & _ 52 _ & @xmath35 & @xmath36 & @xmath37 + `` large '' mma & 7000 & 0.2 & 0.06 & _ 2 _ & _ 12 _ & _ 15 _ & @xmath38 & @xmath39 & @xmath40 notes : telescope area ( @xmath41 ) resolution element ( @xmath42 ) for @xmath43 .
resolution element scales as @xmath44 .
instantaneous sky coverage ( @xmath45 ) for @xmath43 .
instantaneous sky coverage scales as @xmath46 for interferometers , is independent of @xmath44 for most single - dish instruments , and is 4.3 , 5.0 and 2.7 arcmin@xmath47 at @xmath48 , @xmath49 and @xmath50 , respectively , for first ( g. pilbratt , personal communication ) .
noise equivalent flux density , the sensitivity to point sources whose positions are known .
numbers in italics are predicted sensitivities ; numbers not in italics are measured , on - the - telescope values and are subject to downward revision with improved techniques .
predicted sensitivities are optimistic in the sense that in all cases they are near the thermal background limit , a limit that has not yet been achieved in practical submillimeter - wave bolometer systems .
this table is based on the work of hughes and dunlop@xcite .
mauna kea array consisting of the sma , the cso and the jcmt [ [ thermal - balance - of - star - forming - regions - cooling - line - inventory - and - pdr - studies . ] ] thermal balance of star - forming regions : cooling line inventory and pdr studies .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + theoretical studies have established the importance of molecular cooling in the initial phases of star formation ; cooling at densities @xmath51 @xmath52 is by a large number of lines of many different species .
[ [ observations - of - the - ionized - nitrogen - line - near-1.46-thz-205- . ] ] observations of the ionized nitrogen line near 1.46 thz ( 205 ) .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the winter south pole sky becomes 30% transparent at this frequency .
the [ @xmath53 line will be a powerful probe in conjunction with observation of [ @xmath7 from sofia and [ @xmath54 .
these observations can be used to help determine the astrophysically important carbon to oxygen ratio and are relevant for studies of the milky way and external galaxies .
[ [ polarization - studies . ] ] polarization studies .
+ + + + + + + + + + + + + + + + + + + + + the optical design of the 10 m telescope makes it particularly well - suited for measurement of extended polarized submillimeter - wave radiation . during the past decade
, it has been demonstrated that measurement of linear polarization of the thermal emission from interstellar dust grains yields unique information concerning the configuration of magnetic fields in star - forming regions@xcite .
theoretical work suggests that these fields may determine the masses of the stars that form@xcite . in regions with strong ,
ordered fields , the theory predicts that low - mass stars will form .
when the fields are weaker , they are overwhelmed by gravity and higher mass stars result .
the angular resolution and sensitivity of the south pole 10 m telescope will allow polarimetric studies to be extended from high - mass star - forming regions ( where magnetic fields have been studied in detail with airborne telescopes at far - infrared wavelengths and the large submillimeter telescopes on mauna kea ) , to the fainter , more extended regions where low - mass stars are born .
the comparison will provide a crucial test for theories of star formation .
[ [ high - resolution - submillimeter - line - studies - of - star - forming - regions - in - the - magellanic - clouds . ] ] high resolution submillimeter - line studies of star - forming regions in the magellanic clouds .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the sp 10 m provides an unparalleled opportunity to study star formation in a low - metallicity system perhaps an analog of star formation in early galaxies .
these observations would be made in conjunction with the mma .
a 10 meter diameter telescope equipped with modern submillimeter - wave detectors and operating at the south pole in winter is capable of detecting both line and continuum radiation from protogalaxies at @xmath55 in less than an hour . a protogalaxy which will evolve into a large ( @xmath56 ) galaxy produces a continuum flux density of about 5 mjy and a [ @xmath57 line power of about 1 k @xmath58 at the focus of a 10 meter telescope@xcite .
ast / ro has already detected submillimeter spectral lines having 0.5 k @xmath58 power@xcite .
even though this power level corresponds to a significantly larger flux in the beam of a 1.7 meter telescope , detection of these lines with ast / ro demonstrates that an offset 10 meter antenna operating under the winter south pole sky can overcome systematic noise problems at this power level and is capable of making observations of the requisite sensitivity for observing protogalaxies .
table [ table : continuum ] shows the continuum sensitivity and beam size of the sp 10 m and other submillimeter - wave telescopes and illustrates the strength of the sp 10 m for deep continuum surveys .
the nefd values in this table result from different assumptions for the different instruments , assumptions which have varying degrees of certainly and practical verification ; it should be understood that this table is speculative . the high sensitivity per pixel for the sp 10 m results from the combination of a relatively large aperture , large available bandwidth , and fairly good sky ;
the fast survey speed results from high sensitivity and a large field of view ; a relatively small beam size provides freedom from confusion .
this combination of attributes assures that the sp 10 m telescope will make unique and important contributions to astronomy .
a design study was conducted in 1996 by the sao central engineering department using funding contributed for that purpose by the smithsonian institution .
that study has resulted in a preliminary design for the telescope , a static structural analysis of the design , a detailed seven year budget , and a plan for the construction of the telescope using the technology , personnel , and facilities currently being used to construct the six 6-m diameter antennas of the submillimeter array ( sma ) .
avoidance of internal reflections is critical to the design of submillimeter - wave telescopes .
any receiver or detector placed at the focus of the instrument will necessarily emit into the telescope some amount of submillimeter - wave power in the band of interest .
if there is a reflection in the system , for example at the secondary mirror some 6 meters distant , then a resonant cavity is formed whose modes are spaced at roughly 25 mhz intervals , and spurious features at those intervals will appear in the spectrometer whenever there is a change in power level or cavity dimensions .
submillimeter - wave spectrometers cover more than 1 ghz of bandwidth , and a 25 mhz spurious feature can masquerade as a typical astronomical or atmospheric line . also critical to submillimeter - wave telescope design is minimization of variations in antenna thermal emissions as a function of chopper angle .
chopper offsets can more easily be minimized in an off - axies design .
dragone@xcite has shown that if the offset angles in an offset gregorian antenna are chosen correctly , then aberrations and cross - polarization effects in that offset antenna are the same as those in a conventional on - axis antenna with the same diameter and focal length .
the beam efficiency , aperture efficiency , and sidelobe levels in the off - axis antenna are better than those in the on - axis antenna , because in the on - axis design there will be diffraction , reflection , and blockage from the secondary mirror and its supports .
an off - axis gregorian telescope with correctly chosen offset angles will always be optically superior to a similar on - axis configuration .
an offset telescope allows for large prime - focus instrument packages of the type used successfully on python and other cmbr telescopes ; these instruments _ can not _ be used with an on - axis design . .*10 m telescope design parameters . *
[ table : parameters ] [ cols="<,^ , < " , ] in the evaluation of the aberrations for visual wavelength telescopes , it would at this point be noted that the dominant remaining aberration is coma , and the value of @xmath59 would be chosen so that @xmath60 in equation [ eq : coma ] .
this would result in a ritchey - chretien design for positive @xmath61 and an aplanatic gregorian design for negative @xmath61 . in the aplanatic gregorian case
, @xmath59 would be slightly larger than @xmath62 ( @xmath63 ) , resulting in an ellipsoidal primary .
this eliminates third - order coma , but at the expense of a reduced field of view , @xmath64 . for a submillimeter - wave telescope , however , the dominant source of blur at the focal surface is not coma but diffraction .
the diffraction - limited resolution on the sky is @xmath65 , and @xmath66 for all field angles @xmath67 such that @xmath68^{-1 } \approx { { \lambda}\over{d } } [ c + d(k_1 + 1)]^{-1 } f^2 , \ ] ] where @xmath69 is a large number . to put it another way ,
coma becomes significant compared to diffraction only for beams which are @xmath70 $ ] beam diameters away from the center , and @xmath71 .
( for visual wavelength telescopes this is a real limitation , since the diffraction - limited beamsize is @xmath72 , and unless it is corrected , coma becomes significant @xmath73 from the center of the field . ) since diffraction swamps the classical aberrations , the size of the field of view in the submillimeter will tend to be dominated by curvature of the focal surface .
consider , for example , the classical gregorian case @xmath74 in equation [ eq : field ] : @xmath75 .
the size of the usable field of view can be enlarged by setting @xmath59 so as to maximize @xmath64 in equation [ eq : field ] : @xmath76 resulting in a hyperboloidal primary mirror .
substitution of this value into equation [ eq : primary ] shows that the @xmath77 primary deviates from a paraboloid by @xmath78 at the outer edge of the primary surface @xmath79 .
substitution into equation [ eq : conic ] yields @xmath80 , and equation [ eq : secondary ] shows that the secondary is then an oblate spheroid .
substitution in equation [ eq : astigmatism ] shows that this choice worsens the astigmatism , but not to a degree which is significant compared to diffraction .
substitution in equation [ eq : diffraction ] , however , shows that this choice of primary results in coma which is becoming important compared to diffraction . in the fully corrected system , where @xmath81 is chosen so that @xmath82 in equation [ eq : dragone ] , the coma will be reduced by a factor of @xmath83 , so equation [ eq : diffraction ] should be considered an upper bound to the significance of coma .
there is , therefore , an optimal value for @xmath59 such that @xmath84 and the image is maximally flat and minimally aberrated . an intriguing possibility sets
then the secondary is a concave sphere , and the exit pupil of the telescope , which in this design is the position of the tertiary chopper , lies near the center of that sphere .
this choice provides some flattening of the focal surface ( since @xmath86 ) , coma which is reduced compared to the @xmath77 case , and also a high degree of symmetry at the secondary mirror during the beam chop .
this possibility will be investigated using numerical methods .
= 6.0 in + the telescope will be constructed at the smithsonian submillimeter array construction and test facility at haystack observatory during calendar year 2002 . from 1999 through 2001 ,
project personnel will fully specify the instrument , carry out trade studies and design and performance analysis . in the year 2003
, the telescope will be tested using sma test facilities .
the telescope will be disassembled , packed , and then unpacked and re - assembled at haystack observatory .
this will assure that the third assembly at the pole proceeds as smoothly as possible .
the telescope and all associated equipment will be repacked and delivered to the nsf antarctic contractor in port hueneme , california . in 2004 and 2005 , the telescope will be reassembled and commissioned at pole .
the 10 meter telescope will be operated as a user - facility instrument by an international consortium .
observing time will be allocated with consideration for the contributions of the international partners , but will otherwise be freely available on a proposal basis .
the strong survey capabilities of this instrument in the submillimeter and for cmbr argue that a large fraction of the time ( 70% ) should be allocated in large blocks to several key projects .
these key project blocks will be allocated at three year intervals in an open selection process involving a review of solicited proposals .
successful competitors for these key project blocks would likely design and construct detector packages .
the remaining fraction of the observing time ( 30% ) will be allocated to relatively smaller and less complex projects which use the existing detector packages .
these projects will be carried out by remote observing techniques , in collaboration with sp 10 m scientific staff .
we thank p. cheimets , d. caldwell , w. davis , and w. bruckman for their work on the telescope design .
we thank r. w. wilson and a. p. lane for their contributions to the 10 meter proposal .
we are grateful to b. elmegreen , d. fischer , p. goldsmith , a. lane , and g. knapp for their contributions to the science goals for the sp 10 m .
this work was supported in part by the smithsonian institution and in part by the national science foundation under a cooperative agreement with the center for astrophysical research in antarctica ( cara ) , grant number nsf dpp 89 - 20223 .
cara is a national science foundation science and technology center .
a. a. stark , r. a. chamberlin , j. cheng , j. ingalls , and g. wright , `` optical and mechanical design of the antarctic submillimeter telescope and remote observatory , '' _ rev .
* 68 * , p. 2200 , 1997 .
r. a. chamberlin and j. bally , `` 225 ghz atmospheric opacity of the south pole sky derived from continual radiometric measurements of the sky - brightness temperature , '' _ appl .
opt . _ * 33*(6 ) , p. 1095
, 1994 .
r. a. chamberlin and j. bally , `` the observed relationship between the south pole 225-ghz atmospheric opacity and the water vapor column density , '' _ int . j. infrared millimeter waves _ * 16*(5 ) , p. 907
, 1995 .
d. hughes and j. dunlop , `` using new submillimetre surveys to identify the evolutionary status of high - z galaxies , '' in _ observational cosmology with new radio surveys _ , 1997 .
preprint available on www at http://www.roe.ac.uk/research/dhh2.ps.gz .
g. j. stacey , n. geis , r. genzel , j. b. lugten , a. poglitsch , a. sternber , and c. h. townes , `` the 158 micron [ @xmath88 line : a measure of global star formation activey in galaxies , '' _ _ * 373 * , p. 423
, 1991 .
e. l. wright , j. c. mather , c. l. bennett , e. s. cheng , r. a. shafer , d. j. fixsen , r. e. eplee , jr . , r. b. isaacman , s. m. read , n. w. boggess , s. gulkis , m. g. hauser , m. janssen , t. kelsall , p. m. lubin , s. s. meyer , s. h. moseley , t. l. murdock , r. f. silverberg , g. f. smoot , r. weiss , and d. t. wilkinson , `` preliminary spectral obseravations of the galaxy with a @xmath89 beam by the cosmic background explorer ( cobe ) , '' _ _ * 381 * , p. 200
, 1991 .
r. h. hildebrand , j. dotson , c. d. dowell , s. r. platt , d. schleuning , j. a. davidson , and g. novak , `` far - infrared polarimetry , '' in _ airborne astronomy symposium on the galactic ecosystem _ ,
m. r. haas , j. a. davidson , and e. f. erickson , eds .
73 , p. 97 , asp conf .
ser . , 1995 .
a. a. stark , a. d. bolatto , r. a. chamberlin , a. p. lane , t. m. bania , j. m. jackson , and k .- y .
lo , `` first detection of 492 ghz [ @xmath90 emission from the large magellanic cloud , '' _ _ * 480 * , p. l59 , 1997 . | a 10 meter diameter submillimeter - wave telescope has been proposed for installation and scientific use at the nsf amundsen - scott south pole station .
current evidence indicates that the south pole is the best submillimeter - wave telescope site among all existing or proposed ground - based observatories .
proposed scientific programs place stringent requirements on the optical quality of the telescope design .
in particular , reduction of the thermal background and offsets requires an off - axis , unblocked aperture , and the large field of view needed for survey observations requires shaped optics .
this mix of design elements is well - suited for large - scale ( square degree ) mapping of line and continuum radiation from submillimeter - wave sources at moderate spatial resolutions ( 4 to 60 arcsecond beam size ) and high sensitivity ( millijansky flux density levels )
. the telescope will make arcminute angular scale , high frequency cosmic microwave background measurements from the best possible ground - based site , using an aperture which is larger than is currently possible on orbital or airborne platforms .
the telescope design is homologous .
gravitational changes in pointing and focal length will be accommodated by active repositioning of the secondary mirror . the secondary support , consisting of a large , enclosed beam , permits mounting of either a standard set of gregorian optics , or prime focus instrumentation packages for cmbr studies .
a tertiary chopper is located at the exit pupil of the instrument .
an optical design with a hyperboloidal primary mirror and a concave secondary mirror provides a flat focal surface .
the relatively large classical aberrations present in such an optical arrangement can be small compared to diffraction at submillimeter wavelengths .
effective use of this telescope will require development of large ( 1000 element ) arrays of submillimeter detectors which are background - limited when illuminated by antenna temperatures near 50 k. = = = latex = l.42ex -.15emtex # 1= # 1#2=.45 = .45 # 1#2#3#4#5#6#7 to#2 ' '' '' |
in contemporary communication networks , the nodes perform only routing , i.e. , they copy the data on incoming links to the outgoing links . in order to transmit messages generated simultaneously from multiple sources to multiple sinks
the network may need to be used multiple times .
this limits the throughput of the network and increases the time delay too .
network coding is known to circumvent these problems @xcite . in network coding intermediate nodes in a network
are permitted to perform coding operations , i.e. , encode data received on the incoming links and then transmit it on the outgoing links ( each outgoing link can get differently encoded data ) , the throughput of the network increases .
thus , network coding subsumes routing .
for example , consider the butterfly network @xcite of fig .
[ b_fly ] wherein each link can carry one bit per link use , source node @xmath0 generates bits @xmath1 and @xmath2 , and both sink nodes @xmath3 and @xmath4 demand both source bits . with routing only ,
two uses of link @xmath5 are required while with network coding only one .
this is an example of single - source multi - sink linear multicast network coding , wherein there is a single source ( @xmath0 ) , generating a finite number of messages , ( @xmath6 ) , and multiple sinks , each demanding all the source messages and the encoding operations at all nodes are linear . in general , there may be several source nodes , each generating a different number of source messages , and several sink nodes , each demanding only a subset , and not necessarily all , of the source messages .
decoding at sink nodes with such general demands is studied in this paper .
we represent a network by a finite directed acyclic graph @xmath7 , where @xmath8 is the set of vertices or nodes and @xmath9 is the set of directed links or edges between nodes .
all links are assumed to be error - free .
let @xmath10 denote a @xmath11-ary finite field .
the set @xmath12 is denoted by @xmath13 $ ] .
the network has @xmath14 sources , @xmath15 $ ] , and @xmath16 sinks , @xmath17 $ ] .
the source @xmath18 generates @xmath19 messages for all @xmath20 $ ] .
let @xmath21 be the total number of source messages .
the @xmath22-tuple of source messages is denoted by @xmath23}=(x_1,x_2,\ldots , x_\omega)$ ] , where @xmath24 for all @xmath25 $ ] . by @xmath26
we denote the column vector of the source messages .
the demand of the @xmath27 sink node is denoted by @xmath28 $ ] .
given a set @xmath29 $ ] , let @xmath30 , i.e. , @xmath23}$ ] _ restricted _ to @xmath31 .
for disjoint subsets @xmath31 and @xmath14 of @xmath32 $ ] , we do not differentiate between @xmath33 and @xmath34 . for a multi - variable binary - valued function @xmath35 , the subset of @xmath36 whose elements are mapped to @xmath37 by @xmath35
is called its support and is denoted by @xmath38}))$ ] and @xmath39}))$ ] denotes the @xmath40-tuples in the support restricted to @xmath31 .
a source message is denoted by edges without any originating node and terminating at a source node .
data on a link @xmath41 is denoted by @xmath42 .
a network code is a set of coding operations to be performed at each node such that the requisite source messages can be faithfully reproduced at the sink nodes .
it can be specified using either local or global description @xcite .
the former specifies the data on a particular outgoing edge as a function of data on the incoming edges while the latter specifies the data on a particular outgoing edge as a function of source messages . throughout the paper we use global description for our purposes .
[ global description of a network code @xcite ] an @xmath22-dimensional network code on an acyclic network over a field @xmath10 consists of @xmath43 global encoding maps @xmath44 for all @xmath45 , i.e. , @xmath46 .
let @xmath47 , be the incoming edges at the source , then @xmath48 .
when the intermediate nodes perform only linear encoding operations , the resulting network code is said to be a linear network code ( lnc ) .
[ global description of an lnc @xcite ] an @xmath22-dimensional lnc on an acyclic network over a field @xmath10 consists of @xmath43 number of @xmath49 global encoding vectors @xmath50 for all @xmath45 such that @xmath51 .
the global encoding vectors for the incoming edges at the source are standard basis vectors for the vector space @xmath52 .
the global encoding vectors of the lnc for butterfly network is given in fig .
[ b_fly](b ) .
hereafter we assume that the network is feasible , i.e. , demands of all sink nodes can be met using network coding , and the global description of a network code ( linear or nonlinear ) is available at the sink nodes .
if a sink node demands @xmath53 ( @xmath54 @xmath22 ) source messages , it will have at least @xmath53 incoming edges .
the decoding problem is to reproduce the desired source messages from the coded data received at the incoming edges .
thus , decoding amounts to solving a set of at least @xmath53 simultaneous equations ( linear or nonlinear ) in @xmath22 unknowns for a specified set of @xmath53 unknowns .
hence , the global description of a network code is more useful for decoding .
while decoding of nonlinear network codes has not been studied , the common technique used for decoding an lnc for multicast networks is to perform gaussian elimination @xcite , which requires @xmath55 operations , followed by backward substitution , which requires @xmath56 operations @xcite .
this is not recommendable when the number of equations ( incoming coded messages ) and/or variables ( source messages ) is very large . in such cases ,
iterative methods are used .
convergence and initial guess are some issues that arise while using iterative methods @xcite .
we propose to use the sum - product ( sp ) algorithm to perform iterative decoding at the sinks . a similar scheme for decoding multicast network codes using factor graphs @xcite was studied in @xcite in which the authors considered the case of lncs .
the problems associated with the proposed decoding scheme in @xcite are : * to construct the factor graph , full knowledge of network topology is assumed at the sinks which is impractical if the network topology changes . for a particular sink node ( say @xmath57 ) , the factor graph constructed will have @xmath58 variable nodes and @xmath59 factor nodes , where @xmath60 is the set of incoming edges at node @xmath57 . *
complete knowledge of local encoding matrix @xcite of each node is assumed at the sinks which again is impractical since local encoding matrix for different nodes will have different dimensions and hence variable number of overhead bits will be required to communicate to downstream nodes which will incur huge overhead .
we also point out that the motivating examples , _ viz .
_ , examples 1 and 4 , given in @xcite for which the proposed decoding method claims to exploit the network topology admits a simple routing solution and no network coding is required to achieve maximum throughput . solving a system of linear equations in boolean variables
is also studied in ( * ? ? ?
18 ) .
algorithm * + * single - vertex * & @xmath61 & @xmath62 + * all - vertex * & @xmath63 & @xmath64 + * single - vertex with traceback * & not applicable & @xmath65 + [ cols= " < " , ] total number of operations ( ands and ors ) required with traceback is @xmath66 , which is @xmath67 operations , and that without traceback are @xmath68 , which is @xmath69 operations .
thus , running single - vertex sp algorithm followed by traceback step affords computational advantage over the multiple - vertex version.@xmath70
we will now determine the number of semiring operations required to compute the desired marginal functions in an mpf problem using the sp algorithm and the desired supports in an arg - mpf problem using the arg - sp algorithm with and without traceback in the boolean semiring . in this section , by addition and multiplication we mean the boolean or and and operations .
by remark [ rem_supt_or ] , @xmath71 is considered same as addition .
let @xmath72 be an acyclic factor graph with variable nodes @xmath73 and factor node @xmath74 .
the local domain of a node @xmath75 is denoted by @xmath76 , the cardinality of its configuration space @xmath77 by @xmath78 , and its degree by @xmath79 . for an egde @xmath80 between nodes @xmath81 and @xmath82 , @xmath83 and @xmath84 . for every node @xmath85 ,
define @xmath86 if @xmath87 and @xmath88 otherwise .
the message passed from a variable node @xmath89 to a factor node @xmath90 as given in is @xmath91 in the above equation , for each of the @xmath92 values of @xmath93 , product of @xmath94 messages is required which requires @xmath95 multiplications . for each of the @xmath96 values of @xmath97 , @xmath98 additions and @xmath99 multiplications
are required .
thus , the total number of operations required are @xmath100 additions and @xmath101 multiplications .
the messages passed from a factor node @xmath90 to a variable node @xmath89 as given in is @xmath102 this involves product of a local functions with @xmath103 messages for each of the @xmath104 values of @xmath105 .
the total number of operations required for this case is @xmath106 additions and @xmath107 multiplications .
the messages are passed by all nodes except the root node . at the root node @xmath108 ,
the marginal function is the product of @xmath109 messages , requiring @xmath110 multiplications , if it is a variable node and the product of @xmath109 messages with the local function , requiring @xmath111 multiplications , if it is a factor node . in other words ,
computation of marginal function at @xmath108 requires @xmath112 multiplications .
thus , the total number of additions and multiplications required in the single - vertex sp algorithm is @xmath113 and @xmath114 the grand total of the number of additions and multiplications is @xmath115 in the arg - sp algorithm , support of marginal at @xmath108 is computed which requires @xmath116 additions ( by remark [ rem_supt_ops ] ) so that the grand total of operations in this case is @xmath117 in this case , first the single - vertex arg - sp algorithm with @xmath108 as the root is executed on the factor graph
. then the local domain @xmath76 of a neighbor @xmath75 of @xmath108 is partitioned into sets @xmath118 and @xmath119 .
the value @xmath120 is already known from decoding at @xmath108 , and @xmath121 is computed using as follows : @xmath122 where the table of values of the partial marginal @xmath123 was already computed at @xmath75 while passing the message @xmath124 to the root @xmath108 .
we need to look only at the rows for which @xmath125 and output the value of @xmath126 for which @xmath127 .
this requires @xmath128 additions , where @xmath129 and @xmath130 .
the total number of multiplications remains the same as in the single - vertex arg - sp algorithm , which is @xmath131 , but the number of additions is the sum of the number of additions required in single - vertex sp algorithm @xmath132 and the number of additions required at each node , which is at most @xmath133 .
thus , the grand total of operations is at most @xmath134 in the all - vertex sp algorithm , first the messages are passed by all the nodes on the unique path towards the root .
when the root has received messages from all its neighbors , messages are passed on each edge in the reverse direction , i.e. , away from the root and towards the leaves . when all the leaves have received the messages , marginal functions of each node is computed
we use the method suggested in ( * ? ? ?
v ) to compute messages and marginal function .
let a node @xmath75 have degree @xmath135 and has received messages from all but one of its neighbors @xmath136 which is on the unique path from @xmath75 to the root .
for an instance @xmath137 of @xmath76 , let @xmath138 be the values of the known messages , @xmath139 be the value of the message it is yet to receive from @xmath140 , and @xmath141 be the value of its local function , assumed to be @xmath37 if @xmath142 , i.e. , @xmath143 , @xmath144 . the messages @xmath145 involves the product of @xmath141 with all @xmath146s excluding @xmath147 and summing over suitable variables as in and ; there are @xmath135 such messages to be sent , one to each neighbor . . ]
this can be achieved by computing the following products consecutively : @xmath148 , @xmath149 , @xmath150 , @xmath151 , @xmath152 ; this step requires @xmath153 multiplications .
now @xmath75 passes @xmath154 to @xmath140 ( after summing over suitable variables ) and awaits the reception of @xmath139 from @xmath136 .
once @xmath139 is received , the marginal functions is computed , @xmath155 , which requires @xmath37 multiplication .
then the following products are computed consecutively : @xmath156 , @xmath157 , @xmath158 , @xmath150 , @xmath159 ; this step requires @xmath160 multiplications .
subsequently , @xmath145s are computed as follows : @xmath161 , @xmath162 , @xmath150 , @xmath163 , @xmath164 ; this step requires @xmath153 multiplication .
various messages received and passed by node @xmath75 are depicted fig .
[ cplxty ] .
thus , computation of all the messages to be passed by @xmath75 and its marginal function requires @xmath165 multiplications for each of the @xmath78 values in @xmath77 .
this is true for the root node also .
hence , total number of multiplications required is @xmath166 $ ] .
the number of additions required for computing each message remains the same as in the single - vertex sp algorithm , @xmath167 for a variable node passing message to @xmath90 . unlike in the single - vertex case ,
now @xmath89 will pass messages to all its @xmath168 neighbors , thus requiring @xmath169 .
same is true for all the factor nodes also .
hence , the total number of additions required is @xmath170 .
the grand total number of operations is then @xmath171 in the arg - sp algorithm , computation of support of marginal function at a node @xmath75 requires at most @xmath172 additions .
thus , the total number of operations required in all - vertex arg - sp algorithm is @xmath173 the results of sections iv - a , b , and c are tabulated in table [ tab_nops ] .
the operation counts presented in this section apply not only to mpf and arg - mpf problem in boolean semiring , but also to mpf and arg - mpf problem in min - sum , min - product , max - sum , and max - product semiring .
the sp algorithm for decoding a network code is advantageous when the code is either nonlinear or it is linear but the number of messages is very large . for linear network codes with manageable value of @xmath22 , gaussian elimination with backward substitution
is advisable . for a node that demands all the source messages , for example a sink in a multicast network , if application of sp algorithm for decoding network codes leads to computational complexity strictly better than the brute - force decoding complexity , then the code
is called a _ fast sp decodable network code_. the network code for network @xmath174 given in fig .
[ fano ] is fast sp decodable for the sink with vertex label @xmath175 ; decoding complexity is only @xmath176 compared to the brute - force complexity of @xmath177 . as stated above , in order to recover the requisite source messages at a sink we need only run the single - vertex arg - sp algorithm followed by traceback steps . for a given sink node , if the factor graph constructed using the method given in section iii - a is cycle - free and the network code is such that the local domains of all factor nodes have cardinality at most @xmath178 , then the number of operations required for decoding using the sp algorithm is @xmath179 . if the sink demands all the source messages , then the brute - force decoding would require @xmath180 operations . thus , an acyclic factor graph with at most @xmath178 variables per equation is a sufficient condition for fast decodability of the network code at a sink which demands all the source messages .
if the graph is not cycle - free then we remove the cycles by variable stretching .
let @xmath181 be the size of maximum cardinality local domain in the new cycle - free factor graph .
the number of computations required now will be @xmath182 and the code is fast decodable iff @xmath183 .
in a communication network , some nodes may be interested not in the messages generated by some other nodes but in one or more functions of messages generated by other nodes .
for example , in a wireless sensor network that comprises several sensor nodes , each measuring environmental parameters like ambient light , temperature , pressure , humidity , wind velocity etc . for long - term record - keeping and weather forecasting , average , minimum , maximum and variance of these meteorological parameters
are of interest .
environmental monitoring in an industrial unit is another field of application where relevant parameter may include temperature and level of exhaust gases which may assist in preventing fire and poisoning due to toxic gases respectively .
we consider in - network function computation in a finite directed acyclic error - free network , @xmath184 , where codes can perform network coding .
for brevity of expression , we use @xmath185 for @xmath23}$ ] in this section .
the network model is same as given in section i - a for network coding problem with the exception that the sink nodes demand a function of messages rather than a subset of messages , i.e. , a sink node @xmath186 demands the function @xmath187 .
a network code comprises global encoding maps @xmath188 , one for each edge @xmath45 , such that there exist @xmath16 ( decoding ) maps , @xmath189 , for each sink @xmath17 $ ] , such that @xmath190 .
this subsumes the network coding problem of section i as a special case . by @xmath191
we denote the @xmath192-tuple of coded messages received by @xmath186 on its incoming edges .
[ rem_func ] though arguments of a demanded function @xmath193 may only be a subset , say @xmath194 for some @xmath195 $ ] , of messages , we assume it to be a map from @xmath52 to @xmath10 for simplicity rather than from @xmath196 to @xmath10 .
if a sink demands @xmath197 functions , then such a sink may be replaced by @xmath198 sinks each demanding one function but the incoming information to these new sinks is the same ( see fig . [ multi_func ] ) .
the in - network function computation problem is to design network code that maximizes the frequency of target functions computation , called the _ computing capacity _ , per network use . in @xcite
, bounds on rate of computing symmetric functions ( invariant to argument permutations ) , like minimum , maximum , mean , median and mode , of data collected by sensors in a wireless sensor network at a sink node were presented . the notion of min - cut bound for the network coding problem @xcite
was extended to function computation problem in a directed acyclic network with multiple sources and one sink in @xcite .
the case of directed acyclic network with multiple sources , multiple sinks and each sink demanding the sum of source messages was studied in @xcite ; such a network is called a sum - network .
relation between linear solvability of multiple - unicast networks and sum - networks was established .
furthermore , insufficiency of scalar and vector linear network codes to achieve computing capacity for sum - networks was shown .
coding schemes for computation of arbitrary functions in directed acyclic network with multiple sources , multiple sinks and each sink demanding a function of source messages were presented in @xcite . in @xcite , routing capacity , linear coding capacity and nonlinear coding capacity for function computation in a multiple source single sink directed acyclic network were compared and depending upon the demanded functions and alphabet ( field or ring ) , advantage of linear network coding over routing and nonlinear network coding over linear network coding was shown . in order to obtain the value of its desired functions , a sink node may require to perform some operations on the messages it receives on the incoming edges .
though there are many results on bounds on the computing capacity and coding schemes for in - function computation problem , the decoding operation to be performed at the sink nodes to obtain the value of the desired functions has not been studied .
we now formulate computation of the desired functions at sink nodes as an mpf problem over the boolean semiring and use the sp algorithm on a suitably constructed factor graph for each sink to obtain the value of the desired functions .
we consider decoding at the sink node @xmath186 .
it demands the function @xmath199 , where @xmath200 is the set of arguments of @xmath201 for some @xmath202 $ ] . for a realization @xmath203 of the message vector , we are interested in the value @xmath204 .
since a network code only ensures computation of the correct value @xmath205 of the demanded target function given the incoming coded message vector @xmath191 and not the realization @xmath206 of the messages in the argument set , there may be multiple @xmath207-tuples that produce the same values of the incoming coded messages and function value when input to the demanded function , i.e. , the network code is a many - to - one mapping .
we denote one such message vector by @xmath208 .
it need not necessarily be equal to @xmath206 but @xmath209 for all @xmath210 and @xmath211 . using the sp algorithm ,
we will first obtain @xmath208 and then evaluate @xmath212 to obtain @xmath205 .
the arg - mpf formulation for obtaining @xmath208 is given below .
let @xmath213 here @xmath214 is the global product function and @xmath215 are the local functions of the mpf problem at the sink @xmath186 .
the set @xmath216 contains the coordinates indexed by @xmath217 of the message vectors @xmath185 for which @xmath218 , i.e. , the coordinates indexed by @xmath217 of all those message vectors for which @xmath219 , for all @xmath210 .
though @xmath220 may output multiple @xmath207-tuples , we will choose any one as @xmath208 .
the desired function values is then @xmath221 thus , the function computation can be performed by using sp algorithm to solve mpf problem in followed by .
for all @xmath222 obtained using and each @xmath223 $ ] , we have @xmath224 . by remark [ rem_func ] , @xmath225 . a look - up table ( lut )
approach to decoding is to maintain a table with @xmath226 rows and two columns at each sink : first column containing all possible incoming message vectors , @xmath227 , and the second column listing corresponding values of the demanded function , @xmath228 . given an instance of incoming messages , a sink node locates the row containing that @xmath192-tuple in the first column of the lut and then outputs the value in the second column of the row , which is the desired function value . if two rows in the lut have the same entry in the first column ( network code is a many - to - one map ) , the entry in the second column will also be same .
on the contrary , if for two @xmath229 , @xmath230 but @xmath231 for all @xmath210 and some @xmath223 $ ] , then there will be ambiguity at the @xmath232th receiver because there are two distinct possible function values , @xmath233 and @xmath234 , that the decoder may output .
thus , a valid network code that fulfills all receivers demands satisfies @xmath235 for all @xmath210 if @xmath230 for each @xmath223 $ ] and @xmath229 , @xmath236 .
let @xmath203 be a realization of the message vector and @xmath191 the coded message received by @xmath186 on its incoming edges .
the set @xmath237 contains all the message vectors @xmath238 such that @xmath239 including @xmath203 .
thus , @xmath240 for all @xmath241 .
since @xmath242 and @xmath225 , we have that @xmath243 for all @xmath222 .
hence , the sp algorithm for can terminate as soon as a message vector @xmath208 with @xmath244 is found and we need not obtain all possible message vectors which evaluate to the given coded messages on incoming edges of a sink .
for example , let @xmath245 , @xmath246 for all @xmath247 $ ] , and @xmath248 needs to be evaluated using @xmath249 , and @xmath250 . here @xmath251 .
let @xmath252 be a realization of the message vector .
then , @xmath253 , @xmath254 , @xmath255 , and @xmath256 .
from , we have @xmath257}\delta(\tilde{f}_{e_j}(x),y_{e_j } ) \;=\ ; \{000,111\}\end{aligned}\ ] ] any element of @xmath0 can be chosen as @xmath258 and both evaluate to @xmath88 when input to @xmath259 .
this illustrates that @xmath260 .
@xmath70 the factor graph for computation of function @xmath199 at sink @xmath261 $ ] is constructed as follows : 1 .
install @xmath22 variable nodes , one for each source message .
these vertices are labeled by their corresponding source messages , @xmath262 .
2 . install @xmath192 factor nodes and label them @xmath263 .
the associated local domain of each such vertex is the set of source messages that participate in that encoding map and the local kernel is @xmath264 .
a variable node is connected to a factor node iff the source message corresponding to that variable node participates in the encoding map corresponding to the said factor node .
4 . install an additional dummy factor node with local domain @xmath265 , local kernel @xmath37 and label it @xmath201 .
connect this node to variable nodes in the set @xmath265 , i.e. , to the arguments of @xmath201 .
this node corresponds to the demanded function . as before ,
first the cycles in the factor graph are removed , if there are any .
the single - vertex sp algorithm is run on the acyclic factor graph with the dummy factor node as the root using and .
once it has received all the messages , its marginal function ( using ) and subsequently the set @xmath216 are computed as follows : @xmath266 where @xmath267 is the local domain of a neighboring variable node @xmath89 of @xmath201 .
theorem 1 states that obtaining only an element @xmath208 of the set @xmath216 is sufficient to get the desired function value @xmath268 .
in this paper , we proposed to use the sp algorithm for decoding network codes and performing in - network function computation .
we posed the problem of network code decoding at each sink node in a network as an mpf problem over the boolean semiring .
a method for constructing a factor graph for a given sink node using the global encoding maps ( or vectors in case of an lnc ) of the incoming edges and demands of the sink was provided .
the graph so constructed had fewer nodes and led to fewer message being passed lowering the number of operations as compared to the scheme of @xcite .
we discussed the advantages of traceback over multiple - vertex sp algorithm .
the number of semiring operations required to perform the sp algorithm with and without traceback were derived . for the sinks demanding all the source messages
, we introduced the concept of fast decodable network codes and provided a sufficient condition for a network code to be fast decodable .
then we posed the problem of function computation at sink nodes in an in - network function computation problem as an mpf problem and provided a method to construct a factor graph for each sink node on which sp algorithm can be run to solve the mpf problem .
r. appuswamy , m. franceschetti , n. karamchandani , and k. zeger , `` linear codes , target function classes , and network computing capacity , '' _ ieee trans .
inf . theory _ ,
9 , pp . 5741 - 5753 , september 2013 . | while the capacity , feasibility and methods to obtain codes for network coding problems are well studied , the decoding procedure and complexity have not garnered much attention . in this work
, we pose the decoding problem at a sink node in a network as a marginalize a product function ( mpf ) problem over a boolean semiring and use the sum - product ( sp ) algorithm on a suitably constructed factor graph to perform iterative decoding .
we use _ traceback _ to reduce the number of operations required for sp decoding at sink node with general demands and obtain the number of operations required for decoding using sp algorithm with and without traceback . for sinks demanding all messages ,
we define _ fast decodability _ of a network code and identify a sufficient condition for the same .
next , we consider the in - network function computation problem wherein the sink nodes do not demand the source messages , but are only interested in computing a function of the messages .
we present an mpf formulation for function computation at the sink nodes in this setting and use the sp algorithm to obtain the value of the demanded function .
the proposed method can be used for both linear and nonlinear as well as scalar and vector codes for both decoding of messages in a network coding problem and computing linear and nonlinear functions in an in - network function computation problem .
network coding , decoding , sum - product algorithm , traceback , in - network function computation . |
measurements of the cosmic microwave background ( cmb ) radiation have taken a leap forward with the recent announcement of the findings of the wilkinson microwave anisotropy probe ( wmap ) . with these new data it is possible to set important constraints on the shape of the primordial power spectrum and hence to begin to differentiate between the plethora of models for the early universe .
one of the most intriguing results comes from a a combined analysis of the wmap data with the two degree field galaxy redshift survey ( 2dfgrs ) data @xcite , which indicates that the primordial power spectrum might have curvature .
the addition of lyman-@xmath6 forest data on smaller scales to strengthen this conclusion is however contentious @xcite .
the low quadrupole and octopole observed in the cmb temperature power spectrum @xcite has a low probability in standard models , and may be an indication of some feature in the initial power spectrum on very large scales .
both inflationary big - bang @xcite and more speculative cyclic ekpyrotic @xcite models of the early universe predict very nearly gaussian scalar perturbations in the primordial radiation dominated era .
the shape of the perturbation power spectrum depends on the exact model , which typically involves various unknown parameters .
the objective of this paper is to constrain the shape of the initial power spectrum directly from observational data with as few assumptions as possible .
a wealth of cosmological information can be obtained from analysing the shape of the cosmic microwave background ( cmb ) radiation temperature fluctuation power spectrum .
however it has been shown that the effect of changing the cosmological parameters can be exactly mimicked by changes in the shape of the primordial power spectrum @xcite . by including measurements of the cmb polarization and the late time matter power spectrum , as measured for example by galaxy redshift surveys
, the degeneracy can be broken because the cosmological parameters affect these data in different ways . in this paper
we combine temperature and polarization data from the wmap observations and other cmb observations on smaller scales with constraints on the matter power spectrum from the 2dfgrs @xcite .
since the initial power spectrum is an unknown function one is forced to parameterize it .
there are numerous possibilities .
@xcite , @xcite and @xcite use a number of bands in wavenumber .
@xcite also use a model independent approach but using wavelets .
@xcite test a particular inflationary model , in which a phase transition briefly halts the slow roll of the inflaton . inflationary models generically predict a monotonically slowly varying power spectrum , determined by the shape of the inflationary potential .
more complicated inflationary models can give more interesting spectra at the expense of introducing parameters that are fine tuned to give effects in the small range of observable wavenumbers .
in this paper we consider various different power spectrum parameterizations , motivated by theoretical models or features of the observed data .
we also adopt a model independent approach which allows general trends or unexpected features to be detected . in section [ framework ]
we specify the framework used in the rest of the paper .
we summarize the results of the conventional power spectrum parameterization ( ie . a power law slope and running spectral index ) in section [ vanilla ] , and
discuss the possible implications of the small large scale power detected by wmap in section [ cutoff ] .
section [ pinibins ] describes a reconstruction of the primordial power spectrum in wavenumber bands on sub - horizon scales .
we test a double inflation model over a particular range of wavenumbers in section [ barriga ] . in each section
we discuss the implications for estimates of the other cosmological parameters .
the primordial scalar power spectrum @xmath7 is defined by @xmath8 , where @xmath9 is the super - horizon comoving curvature perturbation in the early radiation dominated era .
the commonly assumed power - law power spectrum parameterization is then @xmath10 here @xmath11 is the conventional definition of the scalar spectral index , where @xmath12 corresponds to a scale invariant power spectrum ( we use @xmath13 throughout ) .
the power spectrum amplitude @xmath14 determines the variance of the fluctuations , with @xmath15 to give the observed cmb anisotropy amplitude . in slow roll inflationary models , it is expected that the spectrum varies only very slowly and that @xmath16 is much smaller than unity @xcite . in general
there is a direct relation between the potential of the inflaton field and the spectral index .
as the potential evolves during inflation the spectral index can vary slightly as different modes leave the horizon .
this can be characterized by including a second order term in the logarithmic expansion of the power spectrum @xmath17 , defined by @xmath18 the value of @xmath19 therefore depends on the pivot scale used , for example to convert to a new pivot scale the relation is @xmath20 .
more generally double inflation models or multiple field inflation can lead to breaks and spikes in the primordial power spectrum , see eg .
this motivates a more general parameterization of the primordial power spectrum in terms of amplitudes over discrete bands in wavenumber .
the primordial power spectrum is related to the linear cmb anisotropy power spectrum @xmath21 by a transfer function @xmath22 via @xmath23 where @xmath24 and @xmath25 denote the various temperature and polarization power spectra .
the transfer function for a mode of wavenumber @xmath26 peaks at a multipole of about @xmath27 where @xmath28 is the angular diameter distance . for the concordance model
the relation is roughly @xmath29 .
however the detailed shape of the transfer function is complicated and a range of wavenumbers contribute to each multipole ( see eg .
@xcite ) .
in this paper we vary four cosmological parameters , using flat priors on the baryon density @xmath30 , the cold dark matter density @xmath31 , the hubble constant @xmath32 , and the redshift of reionization @xmath33 ( we assume @xmath34 ) .
the true ionization history may be complicated , so we assume there exists an effective redshift @xmath33 at which a single rapid reionization gives a good approximation to the true cmb anisotropy .
the temperature cmb anisotropy is fairly insensitive to the details of reionization , and the effect on the polarization is only on large scales where it is largely hidden by cosmic variance and the current observational noise , so this assumption should not affect our results significantly .
throughout we assume that the universe is flat with a cosmological constant .
we assume purely gaussian adiabatic scalar perturbations and ignore tensor modes in this paper . since tensor mode perturbations decay on sub - horizon scales they only affect the large scale cmb anisotropy . given our assumptions , if we find excess power on large scales this could , equivalently , be due to a tensor contribution rather than the shape of the scalar initial power spectrum .
we use the latest wmap ] @xcite temperature and temperature - polarization cross - correlation anisotropy data .
we also include almost independent bandpowers on smaller linear scales ( @xmath35 ) from acbar ] @xcite , cbi @xcite and the vsa @xcite .
we constrain the matter power spectrum at low redshift by using the galaxy power spectrum measurements of the 2dfgrs @xcite over the range @xmath36 .
we assume that the galaxy power spectrum measured by the 2dfgrs is a simple unknown multiple of the underlying matter power spectrum ( linear bias ) , so in our analysis the 2dfgrs data serves to constrain the shape but not ( directly ) the amplitude of the matter power spectrum . in principle
the matter power spectrum is directly proportional to the primordial power spectrum however in practice , due to the finite volume observed , the data constrain a smeared version of the underlying matter power spectrum .
this makes it harder to detect any sharp features in the primordial power spectrum , as investigated by @xcite . to evaluate the posterior distributions of parameters given the data we use the markov - chain monte carlo method to generate a list of samples ( coordinates in parameter space ) such that the number density of samples is proportional to the probability density .
we use a modified version of the cosmomc ] code , using camb @xcite ( based on cmbfast @xcite ) to generate the cmb and matter power spectrum transfer functions .
cosmomc uses the metropolis - hasting algorithm to explore the posterior probability distribution in a piece - wise manner , allowing us to exploit the fact that for each transfer function computed many different values of the initial power spectra parameters can be changed at almost no additional computational cost . for further details
see @xcite and references therein .
most of the chains for the analysis here were computed in around a day using spare nodes of the cita beowulf cluster , with each node running one chain parallelized over the two processors .
between 4 and 20 converged chains were generated for each set of parameters , burn in samples were removed , leaving of the order of @xmath37@xmath38 accepted positions in parameter space from which the results in this paper were computed .
in most parameter studies the scalar initial power spectrum is parameterized by a constant spectral index @xmath19 and an amplitude .
even with only two parameters defining the primordial power spectrum there are large degeneracies between these and the other cosmological parameters .
for example larger baryon densities decrease the height of the cmb second acoustic peak thereby mimicking the effect of a red spectral tilt . using the wmap @xcite power spectrum results alone
we find a tight marginalized constraint @xmath39 ( 68% confidence ) .
however the orthogonal direction is very poorly constrained with @xmath40 .
the best fit model to the wmap data @xcite with @xmath41 can be closely approximated by completely different models with @xmath42 , higher reionization redshifts , rapid hubble expansion and high power spectrum amplitude .
on integrating out the value of @xmath43 the constraint on the spectral index weakens to @xmath44 .
similarly the amplitude of the primordial curvature fluctuations on @xmath45 scales is constrained by wmap to be @xmath46 , whereas the parameter combination @xmath47 is much more tightly constrained since it comes directly from the observed temperature anisotropy amplitude . to partially break these degeneracies the wmap team adopt a prior on the reionization optical depth of @xmath48 . by adding the 2dfgrs and additional cmb data at @xmath49 we find the parameter constraints which broadly agree with the analysis reported by the wmap team @xcite , see fig .
[ fig : pinibins_cospars ] below . [ nrun ] by adding a running spectral index we find the marginalized 1-sigma result @xmath50 shown by the solid line in the right hand panel of fig .
[ fig : nrun ] , in rough agreement with the wmap team . for the pivot scale
used ( @xmath51 ) a red tilt is preferred ( left hand panel of fig .
[ fig : nrun ] ) . as expected , the effect of adding @xmath17 as a free parameter is to increase the uncertainties on the cosmological parameters but only within the original uncertainties .
we find that the evidence for running comes predominantly from the very largest scale multipoles .
when we exclude @xmath52 multipoles from the wmap temperature ( tt ) likelihood the running spectral index distribution shifts to becomes highly consistent with @xmath53 , as shown in fig .
[ fig : nrun ] . the constraints on all cosmological parameters except for @xmath54 and @xmath17
are changed very little on removing the lowest multipoles .
the running parameterization is therefore not ideally suited to the data : wmap provides some evidence for low power on the very largest scales , but this is only crudely fit by using a running spectral index .
the wmap analysis relies on lyman-@xmath6 forest at wavenumbers greater than @xmath55 to give evidence for more red tilt on small scales consistent with a running index , but the validity of this analysis is in serious doubt @xcite .
we conclude that the marginal preference for a running spectral index from our cmb + 2dfgrs analysis is primarily driven by the first three cmb multipoles .
the quadrupole and octopole estimators observed by wmap are low compared to the other large scale multipoles . in a given model the low multipole estimators have a wide @xmath56-like distribution ( which has the peak below the ensemble mean ) , so the low values could just be chance .
however any model that predicts small values for the low multipoles would be favoured by the data , by a factor of up to about fourty , so this could be a hint that there is a sharp fall in power on the largest scales .
@xcite find that the quadrupole and octopole appear to be aligned , perhaps indicative of some anisotropic effect on on very large spatial scales which would not be well modeled by a statistically isotropic power spectrum model . here
we assume that the alignment is a coincidence , and proceed to consider whether the shape of the initial power spectrum could help to explain the low large scale signal .
there is only a limited amount of information on the largest scales due to cosmic variance , so we can not hope to fit many extra parameters .
we choose to assume there is a sharp total cut - off in the primordial power on scales larger than @xmath57 ( for a discussion of an exponential cut - off in closed models see @xcite ; see @xcite for possible motivation for a cut - off ) . as discussed in the previous section
a constant spectral index is a good fit on smaller scales , so we parameterize the primordial spectrum as @xmath58 marginalizing over the other parameters we find the constraint on @xmath59 shown in the upper panel of fig .
[ fig : cut ] .
we find a preference for a cut - off at @xmath60 , a scale which can give a significantly lower quadrupole and octopole without significantly affecting the higher multipoles ( lower panels of fig .
[ fig : cut ] ) .
however , according to this model a spectrum with no cut - off is not strongly excluded by the data .
our parameterization can not achieve values for the quadrupole as low as observed because there is a significant integrated - sachs wolfe contribution to the quadrupole from scales smaller than the cut .
the best - fit cut - off model has a very similar probability to the best - fit running model , although the cut - off model is marginally preferred .
the constraints on the cosmological parameters are virtually unaffected by adding this free cut - off scale .
a comparison with @xcite s assessment of the significance of the low cmb power at small multipoles is not straightforward because the statistical tests differ
. a more detailed analysis of this important issue is required , but given the results of fig . 2 it would seem premature to discard simple continuous power law models .
on smaller scales there is now a wealth of data available to constrain the primordial power spectrum in considerable generality .
here we choose to parameterize the spectrum in a number of bands , which has the advantage that it is free from any assumptions about the underlying model .
this generality can in principle reveal unexpected features in the primordial power spectrum on scales comparable to ( or larger than ) the band width .
the disadvantage is that there are a large number of free parameters so care is needed to obtain meaningful results .
the initial power spectrum is only indirectly constrained by the observational data . since we do not model the biasing in the 2dfgrs galaxy power spectrum , the only direct constraint on the amplitude ( as opposed to the shape ) comes from the cmb anisotropy power spectra . on scales smaller than the horizon size at reionization ( @xmath61 )
the observed power in the cmb anisotropy scales as @xmath62 , depending on the optical depth to reionization @xmath63 .
we therefore parameterize the shape of the initial power spectrum by the value of @xmath62 at a set of points , and linearly interpolate ( in @xmath26 ) between the points .
we then refer to the power at each point as the power in a ` band ' at that wavenumber .
our parameters @xmath64 are defined by @xmath65 where the first line applies for @xmath66 .
we set the position of the first band to @xmath67 and that of the second band to @xmath68 , so we only have two bands over the region where the data are effected by the modes which are super - horizon at reionization and hence do not scale simply with @xmath69 .
subsequent band positions on smaller scales are logarithmically spaced with @xmath70 , where the constant is chosen such that @xmath71 .
we assume a flat prior on each of the @xmath64 .
this prior is equivalent to a flat prior on the underlying power spectrum @xmath72 , together with a prior on the optical depth @xmath73 .
constraints on quantities sensitive to @xmath63 depend strongly on the choice prior for large numbers of bands .
this is one reason we reconstruct @xmath64 , which are more directly constrained by the data and have the main dependence on @xmath63 taken out , rather than @xmath74 directly .
our choice of prior gives a posterior constraint on the optical depth similar to ( but somewhat broader than ) that with the simple spectral index parameterization .
[ fig : pinibins_ktoell ] shows the cmb temperature power spectra for top hat primordial power spectra which are zero except for the range @xmath75 ( except for @xmath76 where @xmath77 ) for a concordance model .
all the primordial power spectrum top hats have the same absolute height .
the result illustrates the rough correspondence between the position of the bands we use and the cmb power spectrum peaks , showing that there is one band covering much of the large scales , and several bands over the first acoustic peak , with only one or two bands for each of the subsequent few peaks .
our reconstruction of the shape of the initial power spectrum is shown in fig .
[ fig : pinibins ] . the crosses and error bars
show the peak and 68% upper and lower limits of the marginalized distributions .
the probability distribution of all but the last band is close to gaussian , although some bands are correlated at around the fifty per cent level .
the maximum correlation is 80 per cent , which is a positive correlation between bands around the first acoustic peak of the cmb .
we have checked that our constraints on the sub - horizon @xmath64 agree well with those obtained by fixing the optical depth to @xmath78 rather than marginalizing over it .
the overall shape is consistent with a featureless scale invariant power spectrum , though perhaps an overall red tilt is discernible , bearing in mind the large error bar on the small scale point .
we do not see the feature at @xmath79 suggested by @xcite .
our reconstruction is complementary to that in @xcite since our inter - band separation is one third of theirs .
also , we use linear interpolation instead of wavelets or top - hat bins , which makes the primordial power spectrum relatively smooth . in fig .
[ fig : pinibins_cospars ] we investigate how robust the cosmological parameter estimates are to this general primordial power spectrum .
dashed lines show the cosmological parameter constraints assuming a power law primordial power spectrum , marginalising over @xmath19 .
overlaid , the solid lines show the result after allowing freedom in the amplitudes of the 16 bands .
the error contours are broader but remarkably show that it is still possible to recover interesting constraints on the cosmological parameters even if one allows the large amount of freedom in the primordial power spectrum shape .
it is important to emphasize that this method of power spectrum differs from earlier work such as that by @xcite because the mcmc reconstruction properly accounts for the allowed ranges of all of the parameters defining the model .
in this section we use a parameterization motivated by double inflation , previously explored by @xcite . in this model @xmath80 where @xmath81 and @xmath6 are chosen such that the power spectrum is continuous . as in
@xcite we take flat priors on @xmath82 , @xmath83 and @xmath84 and we limit the parameter space to @xmath85 , @xmath86 and @xmath87 .
in addition we impose a prior @xmath88 so that we explore a transition only in the region probed by the observational data . in fig .
[ fig : bar ] we show the constraints on the model parameters after marginalising over four cosmological parameters .
a conventional scale - invariant spectrum corresponds to @xmath89 and we can see that this possibility is very close to the 1@xmath90 contours .
values of @xmath83 higher than unity are preferred , corresponding to a drop in the initial power spectrum on going from large to small scales .
this is not surprising since the data favour red tilts , and in this parameterization a tilt is obtained by having a long transition between a higher flat spectrum on large scales and a lower flat spectrum on small scales .
the distribution is slightly bimodal , with the first mode roughly corresponding to a wide transition straddling the first cmb acoustic peak and the second corresponding to a drop in power in the dip between the first and second peaks .
the expected strong correlation between @xmath83 and the baryon density is clear .
we find that removing the first three multipoles has no effect on the parameter constraints on this model .
we show in fig .
[ fig : bar_pk ] the power spectra corresponding to 20 samples in the mcmc chain .
there is considerable spread , but the transition , if any , occurs at around @xmath91 .
this corresponds roughly to the scale at which the 2dfgrs data becomes statistically significant .
since the 2dfgrs data may disfavour a transition at higher wavenumber and the wmap data would disfavour a transition at lower wavenumber then neither the amplitude or the scale of the break is likely to be of any physical significance . the effect of adding these additional parameters on the estimates of cosmological parameters is small ; the biggest effect is to widen the allowed range of @xmath30 to include smaller values .
this is due to the fact that a preferred position for any transition is between the first two acoustic cmb peaks , and it is also this ratio that determines the baryon density .
we have explored various parameterizations of the primordial power spectrum and found that in each case a simple scale invariant spectrum is an acceptable fit to the data .
some deviation towards a red tilt is preferred and there is a marginal indication of a cut - off in power on the largest observable scales .
significantly , even if the power spectrum is allowed to vary over a large number of wavenumber bands on sub - horizon scales the reconstructed spectrum is found to be featureless and close to scale invariant . in addition we have shown how the constraints on the other cosmological parameters are affected by the addition of this extra freedom in the primordial power spectrum shape : the error bars on cosmological parameters are significantly amplified but the resulting constraints are still strong and roughly similar to those before the recent wmap results .
the unexpectedly small power on superhorizon scales observed by wmap , if real , provides marginal evidence for a drop in the primordial power spectrum on the largest observable scales .
we thank carlo contaldi and the cambridge leverhulme group for helpful discussions , in particular anze slosar , ofer lahav , anthony lasenby , dan mortlock , carolina odman and mike hobson .
slb acknowledges support from pparc and a selwyn college trevelyan research fellowship .
jw is supported by the leverhulme trust and kings college trapnell fellowship .
the beowulf computer used for this analysis was funded by the canada foundation for innovation and the ontario innovation trust .
we also used the uk national cosmology supercomputer center funded by pparc , hefce and silicon graphics / cray research . | we reconstruct the shape of the primordial power spectrum from the latest cosmic microwave background data , including the new results from the wilkinson microwave anisotropy probe ( wmap ) , and large scale structure data from the two degree field galaxy redshift survey ( 2dfgrs ) .
we tested four parameterizations taking into account the uncertainties in four cosmological parameters .
first we parameterize the initial spectrum by a tilt and a running spectral index , finding marginal evidence for a running spectral index only if the first three wmap multipoles ( @xmath0 ) are included in the analysis . secondly , to investigate further the low cmb large scale power , we modify the conventional power - law spectrum by introducing a scale above which there is no power . we find a preferred position of the cut at @xmath1
although @xmath2 ( no cut ) is not ruled out .
thirdly we use a model independent parameterization , with 16 bands in wavenumber , and find no obvious sign of deviation from a power law spectrum on the scales investigated .
furthermore the values of the other cosmological parameters defining the model remain relatively well constrained despite the freedom in the shape of the initial power spectrum .
finally we investigate a model motivated by double inflation , in which the power spectrum has a break between two characteristic wavenumbers .
we find that if a break is required to be in the range @xmath3 then the ratio of amplitudes across the break is constrained to be @xmath4 .
our results are consistent with a power law spectrum that is featureless and close to scale invariant over the wavenumber range @xmath5 , with a hint of a decrease in power on the largest scales .
[ firstpage ] cosmology : observations cosmology : theory cosmic microwave background large scale structure |
\(175706 ) 1996 fg3 ( hereafter , 1996 fg3 ) is a binary near - earth asteroid ( nea ) belonging to the apollo type , which has a very low @xmath12v value @xmath2 @xmath13 @xcite .
this asteroid was originally chosen as the target for a proposed sample return mission @xcite , called marcopolo - r .
eclipse events have been observed in the binary system 1996 fg3 in optical wavelengths @xcite .
the period of mutual orbit is @xmath14 , and the diameter ratio @xmath15 is @xmath16 @xcite .
the primary is assumed to be an oblate ellipsoid with a major axis ratio @xmath17 , while the secondary is prolate with a major axis ratio about 1.4 .
furthermore , @xcite showed the rotation period of the primary is @xmath18 , and the mass density @xmath19 g @xmath20 , where 1996 fg3 is generally classified as a complex c - type @xcite .
the optical magnitude and phase coefficient are derived as @xmath21 and @xmath22 by @xcite , while @xcite used additional data to derive @xmath23 and @xmath24 .
@xcite obtained an area - equivalent diameter of @xmath25 and a geometric albedo of @xmath26 for 1996 fg3 , based on thermal observations ( 3.6 and @xmath27 ) from the `` warm spitzer '' space telescope .
furthermore , @xcite measured the effective diameter and geometric albedo of the asteroid to be @xmath28 , @xmath29 , @xmath30 for a solar phase angle @xmath31 from an neatm @xcite procedure .
in addition , using neatm method , @xcite also obtained 1996 fg3 s effective diameter and geometric albedo of @xmath32 , @xmath33 , beaming parameter @xmath34 , for a phase angle @xmath35 .
@xcite modeled the formation of the binaries to be a history of rotation acceleration due to the yorp mechanism .
generally , the asymmetric reflection of sunlight and asymmetric thermal emission from an asteroid s surface produces a net force and a net torque .
the net force causes the orbit of the asteroid to drift , i.e. , yarkovsky effect , and the net torque alters its rotation period and direction of its rotation axis , i.e. , yorp effect @xcite .
the asymmetric shape of an asteroid , as well as the existence of its finite rotation period and thermal inertia , plays a major role in affecting yarkovsky and yorp effects .
thermal inertia is an important parameter that controls temperature distribution over the surface and sub - surface of the asteroid , and it is defined by @xmath36 , where @xmath37 is the mass density , @xmath38 the thermal conductivity , and @xmath39 the specific heat capacity . according to the definition , thermal inertia depends mainly on the regolith particle size and depth , degree of compaction , and exposure of solid rocks and boulders within the top few centimeters of sub - surface @xcite . therefore , thermal inertia may act as a vital indicator to infer the presence or absence of loose material on the asteroid s surface .
the thermal inertia of an asteroid may be evaluated by fitting mid - infrared observations by applying a thermal model to reproduce mid - ir emission curves .
@xcite proposed a so - called thermophysical model ( tpm ) , which is adopted to calculate infrared emission fluxes as a function of the asteroid s albedo , thermal inertia , correction factor and so on .
for example , we provide several values of @xmath7 for the targets of complete / future asteroid missions . on the basis of tpm , @xcite measured the thermal inertia of asteroid ( 25143 ) itokawa to be @xmath2 @xmath40 .
again , @xcite showed that the @xmath7 of 1999 ju3 is likely to be in the range of @xmath41 . the average thermal inertia of 1999 rq36 is estimated to be @xmath2 @xmath42 @xcite .
recently , @xcite proposed an advanced thermophysical model ( atpm ) to extensively investigate the thermal nature of the asteroid .
subsequently , @xcite obtained a best - fit thermal inertia @xmath43 for 1996 fg3 with atpm .
the structure of the paper is as follows .
firstly , section 2 gives a brief introduction to the thermophysical model developed by @xcite . in section 3
, we concentrate our study on deriving a new 3d shape model for the primary of 1996 fg3 from the known optical lightcurves @xcite . the rotation period and
spin axis of the primary are again updated in this work but they appear to be slightly different from those of @xcite . subsequently , in section 4 , we independently develop computation codes that duplicates atpm @xcite , and then carry out extensive fittings using three sets of mid - infrared observations of 1996 fg3 @xcite to investigate the primary s surface physical properties , such as the average thermal inertia , geometric albedo and roughness .
the results show that this asteroid may have a very rough surface , in the meanwhile the thermal inertia and geometric albedo seem to be relatively low , indicative of the existence of loose material or regolith spreading over the asteroid s surface . moreover , section 5 deals with the global surface and sub - surface temperature distribution at the aphelion and perihelion , and the minimum depth of regolith layer over the very surface of the primary from the simulations . in final , section 6
presents the primary results of this work and gives a concise discussion .
an asteroid is assumed to be a polyhedron composed of n triangle facets in atpm . for each facet
, the conservation of energy leads to the surface boundary condition @xcite @xmath44 where @xmath45 is the bond albedo , @xmath46 indicates whether facet @xmath47 is illuminated by the sun , @xmath48 is the thermal emissivity , @xmath49 the stefan - boltzmann constant , @xmath38 the thermal conductivity , and @xmath50 the depth below the surface , respectively .
@xmath51 means a function that returns the cosine of the sunlight incident angle .
@xmath52 is the integrated solar flux at the distance of the asteroid , which can be approximated by @xmath53 where @xmath54 is the solar constant , about @xmath55 , and @xmath56 is the heliocentric distance in au . @xmath57 and @xmath58 are the total multiple - scattered and thermal - radiated fluxes incident onto the facet from other facets , respectively .
@xmath59 is the albedo of the surface at thermal - infrared wavelengths .
temperature @xmath60 can be written as a function of time @xmath61 and depth @xmath50 , i.e. , @xmath62 .
it can be described by one - dimension ( 1d ) heat diffusion equation @xmath63 for the asteroid , which may have a regolith layer over its surface , the temperature below the surface regolith is supposed to be constant .
thus an internal boundary condition could be given by @xmath64 theoretically , the temperature distribution can be directly obtained by solving equation ( [ heat ] ) and the boundary conditions of equations ( [ surbc ] ) and ( [ inbc ] ) . however , the surface boundary condition is a nonlinear equation , thus we will attempt to solve them with numerical methods . in order to simplify the solution of this problem , it is useful to introduce a standard transformation @xcite as follows : @xmath65 where @xmath66 is referred to as skin depth . and @xmath67
is called effective temperature .
thus , the 1d heat diffusion equation is rewritten as @xmath68 while the surface boundary condition and internal boundary condition are converted into @xmath69 @xmath70 where @xmath71 and @xmath72 @xmath7 is the thermal inertia , and @xmath73 is the thermal parameter . by numerically solving equation ( [ tf ] ) with boundary condition equations ( [ sbc ] ) and ( [ ibc ] ) , we can acquire the global temperature distribution over the surface and sub - surface of the asteroid . according to the theory ,
the program codes are also developed for investigation . in the following
, we will introduce the results based on numerical calculations . on the basis of surface temperature distribution calculated from atpm
, we may estimate thermal infrared fluxes from the asteroid for a given phase angle @xmath74 and geocentric distance @xmath12 .
first of all , a body - fixed coordinate system is established ( see figure 1 ) , where the origin @xmath75 locates at the center of asteroid , z - axis is parallel to its spin axis , and x - axis is chosen to remain in a plane determined by z - axis and the line of sun - asteroid , directing towards the sun .
the direction vector of earth in the body - fixed system is calculated from the asteroid s orbit , expressed as @xmath76 .
normal vector of each facet @xmath77 is determined from the shape model of the asteroid .
hence , the view factor of each facet relative to earth can be evaluated by @xmath78 where @xmath79 indicates that facet @xmath47 can be seen from earth , otherwise @xmath80 , and @xmath81 is the area of facet @xmath47 . the flux emitted by each facet
is described by the plank function @xmath82 consequently , the total fluxes observed from earth are fully integrated over each facet @xmath83
according to direct images from space missions and radar measurements , the asteroids in small - size have an irregular shape , while those larger objects appear to be relatively regular shape .
for instance , vesta seems to be more circular than eros . considering the asteroid s rotation ,
the observed lightcurves change with a lot of extrema , which provide key information on modeling its shape and morphology .
hence , the substantial shape and spin status of the asteroid can be derived from the observations . @xcite and @xcite developed the lightcurve inverse scheme , and in their model , the inverse problem
is perfectly solved using modern deconvolution method and optimization techniques . with the assistance of the optical data ,
a convex shape of asteroid will be obtained . after repeating the fitting of the convex shape and additional observations , an improved shape of the asteroid
will be better constructed .
the relative chi - square is defined as @xmath84 where @xmath85 and @xmath86 are the modeled and observed brightness from lightcurves , respectively . @xmath87 and @xmath88 are the averaged brightness of the observed data and the model , respectively . at the end of iteration , chi - square will approach a tiny value .
usually , for an asteroid , there are a great many of observations with respect to various solar angles at different epochs .
therefore , such method will yield a reliable result for the shape model .
.14 lightcurves adopted in the shape model of 1996 fg3 . [ cols="^,^,^,<",options="header " , ] in the following section , we will then utilize the above derived parameters to evaluate the surface thermal environment of 1996 fg3 at its aphelion and perihelion respectively .
on the basis of the derived shape model from the observations , the physical parameters in table [ phpa ] , and the thermal inertia determined from the atpm fitting process , we attempt to simulate the temperature variation of 1996 fg3 over a rotation period . in figures [ aph - per ] and [ gtemp ] , we show the equatorial temperature and global surface temperature distribution for the primary at aphelion and perihelion , respectively .
figure [ aph - per ] shows the equatorial temperature distribution of 1996 fg3 at its aphelion and perihelion respectively .
the maximum temperature does not appear at the solar point , but delays about @xmath89 , and the minimum temperature occurs just a little after the local sunrise , delaying about @xmath90 .
such delay effect between absorption and emission is actually caused by the non - zero thermal inertia and the finite rotation speed of the asteroid .
on the other hand , according to figure [ aph - per ] , the equatorial temperature of 1996 fg3 may range from @xmath91 to @xmath92 over a whole orbit period .
figure [ gtemp ] shows the global surface temperature distribution of 1996 fg3 at its aphelion ( left ) and perihelion ( right ) respectively . in this figure
, z - axis represents the asteroid s spin axis , and x - axis points to the sun in the framework of an asteroid center body - fixed coordinate system .
the profile of temperature in figure [ gtemp ] is shown by the index of color - bar , and the red region represents the facets are sunlit , while the blue facets are referred to relatively low temperature . as mentioned previously ,
1996 fg3 is chosen to be a backup target for marcopolo - r sample return mission .
hence , we still show great interest in the surface feature of the asteroid whether there exists a regolith layer on its surface .
apparently , thermal inertia is associated with the surface properties , where it will be helpful to infer the presence or absence of loose material on the surface . as known to all , fine dust has a very low thermal inertia about @xmath93 , lunar regolith owns a relatively low value about @xmath94 , a sandy regolith like eros soil bears a value of @xmath95 , but coarse sand occupies a higher thermal inertia profile @xmath2 @xmath96 ( e.g. , itokawa s muses - sea regio ) . in comparison with the above materials
, bare rock has an extremely higher thermal inertia more than @xmath97 @xcite . in this work
, the thermal inertia of 1996 fg3 is estimated to be @xmath8 .
consequently , it is quite natural for one to suppose that the surface of 1996 fg3 may be covered by loose materials , perhaps a mixture of dust , fragmentary rocky debris and sand . in other word , there may exist a regolith layer on the surface of 1996 fg3 , according to the above derived thermal inertia .
theoretically , the so - called `` skin depth '' : @xmath66 is usually used to characterize the grain size of regolith . according to the definition of thermal inertia @xmath98 .
the profiles of mass density @xmath37 , thermal conductivity @xmath38 , and specific heat capacity @xmath39 may be estimated from the above derived thermal inertia . since the thermal conductivity @xmath38 depends on particle size and temperature much more than @xmath37 and @xmath39 , we attempt to assume a constant value for @xmath37 and @xmath39 , and then estimate @xmath38 from the above derived thermal inertia .
next , the estimation of skin depth will be easily acquired .
@xcite has derived the mass density to be @xmath19 g @xmath20 . since the above derived value @xmath99 is similar to that of the moon despite a little higher
, we assume the specific heat capacity of 1996 fg3 is similar to that of the moon , about @xmath100 .
then the thermal conductivity @xmath38 is estimated to be @xmath101 and @xmath102 is @xmath103 .
such small @xmath38 and @xmath102 obviously support the possible existence of loose material or regolith over the surface of 1996 fg3 .
however , the depth of the regolith layer , covered on the asteroid s surface , attracts our great attention .
we carried out simulations to explore the regolith depth versus temperature distribution of sub - surface for @xmath104 , @xmath105 and @xmath106 , respectively ( see figure [ regolith ] ) .
the left panels show the results are obtained when the asteroid moves at aphelion , while the right panels exhibit those at perihelion . in each panel ,
two profiles , which are respectively , plotted by solid line ( local midday ) and dashed line ( local midnight ) , show that the sub - surface temperature changes with the depth ( the distance from the surface ) . in these panels ,
the temperature goes down as the depth increases at local noon , while it goes up as the depth increases at local midnight until a certain depth , where two curves are overlapped .
this phenomenon results from the internal boundary equation [ inbc ] .
the figure shows the temperature distribution of the very loose regolith layer , the thermal conductivity of which may be in the range of @xmath101 . and
the minimum depth of this layer may be estimated from figure [ regolith ] .
herein the regolith depth of the very surface of 1996 fg3 is reckoned to be @xmath11 .
this minimum value of regolith depth may be considered as a reference for the design of a spacecraft if a lander is equipped . on the other hand ,
the existence of regolith over the surface of the primary of 1996 fg3 actually does good to a sample return mission .
in this work , we have derived a new 3d convex shape model of the primary of 1996 fg3 from the published lightcurves , where the best - fit orientation of its spin axis is determined to be @xmath0 and @xmath1 , with a rotation period of @xmath2 3.5935 h. on the basis of the numerical codes independently developed according to thermophysical model , we apply the shape model and the required input physical parameters to fit three sets of mid - infrared measurements for 1996 fg3 .
herein we summarize the major physical properties obtained for the asteroid as follows : the geometric albedo and effective diameter are , respectively , @xmath107 and @xmath108 ; the diameters of the primary and secondary are calculated to be @xmath109 and @xmath110 , respectively . moreover ,
the thermal inertia @xmath7 is also determined to be a low value of @xmath8 , whereas the roughness fraction @xmath9 is estimated to be @xmath10 . from the simulations
, we find that low thermal inertia ( @xmath111 ) would make a perfect fitting to the visir observations on 19 january , 2011 ( with respect to a low phase angle ) , whereas the observations obtained at higher phase angles ( table 3 and 4 ) are very sensitive to @xmath9 , thereby leading to a best - fit solution with the case of large roughness fraction .
given that we simultaneously perform the computation using the combination data ( tables 3,4 and 5 ) , a broad range of @xmath112 profiles are obtained in the fitting . furthermore , to acquire a significant uncertainty for the outcomes , we finally choose a @xmath113 range of the minimum @xmath112 to determine the best - fit solution for thermal inertia of 1996 fg3 . on the other hand , via the simultaneous fitting with these observations at different solar phase angles , the degeneracy of solutions between thermal inertia and roughness is removed , making it capable to determine the estimation for thermal inertia and roughness separately . the ratio of `` observation / model '' ( see figures [ rphase ] and [ rspectra ] ) is a good indicator that examines how the results from the model match the observations at various phase angles and wavelengths @xcite .
hence , this enables us to conclude that the fitting process is correct and the derived results are reliable . , @xmath114 , and @xmath115 .
] , @xmath114 , and @xmath115 . ]
according to the atpm fitting results shown in table [ outcome ] , the evaluated large roughness fraction @xmath116 for 1996 fg3 may suggest a rough surface on the asteroid .
however , the asteroid s mean surface thermal inertia is estimated to be a relative low value .
then the question arises why this asteroid bears such a low thermal inertia with a very rough surface . as the adopted roughness model in the atpm
is assumed to be an irregularity degree of a surface at scales smaller than the global shape model resolution ( like a facet area ) but larger than the thermal skin depth .
therefore , 1996 fg3 would be likely to have a rough surface , which may or may not include craters of any size , and a porous or dusty material .
the recent work of @xcite suggested the surface of 1996 fg3 may be a compact one , with the existence of regions showing different roughness , similar to that of itokawa .
hence , there would be possible for an asteroid to possess a large roughness fraction and low thermal inertia .
however , the nea binaries may have a high averaged thermal inertia of @xmath117 @xcite , indicating that high thermal inertia could be supportive to the binary formation scenario due to yorp mechanism . in this work ,
we have derived thermal inertia @xmath99 for 1996 fg3 , and it appears below their estimation .
nevertheless , it is still likely for a rubble - pile asteroid to have a low thermal inertia , thereby remaining a rough surface during a long dynamical evolution due to space - weathering and regolith migration . in this sense
, the binary 1996 fg3 may also be produced via the yorp rotation acceleration effect at a very earlier time , but retain a roughness surface from then , resulting in the distribution of a great many meter - size ( or even smaller ) craters on the surface .
thus , these tiny craters might be covered or surrounded by loose materials , making it appear a low thermal inertia .
@xcite showed that the average thermal inertia for neas may be @xmath118 .
figure [ inertia - size ] exhibits the variation of mean thermal inertia with the size of asteroids from the observations . from this figure
, we observe that the value of thermal inertia for 1996 fg3 given in this work ( labeled in red ) deviates a little from the prediction profile .
in addition , our outcome for the binary system is a bit lower than that of @xcite .
this results from that we perform the combination fitting with additional thermal - ir observations ( tables 3 and 4 ) , which may provide new insight for the thermal study of binary asteroids . in conclusion , the surface of 1996 fg3 s
primary may be a very rough surface , on which loose materials such as fine dust , fragmentary rocky debris , sands or most likely a stuff of their mixture are covered , composing a kind of regolith .
the depth of the possible regolith layer is evaluated to be approximately @xmath11 .
such implication may provide substantial information for engineering of the sample return mission , for example , a selection of landing area .
however , we should place special emphasis on that this estimation is simply a roughly minimum value of the regolith layer over the very surface of the asteroid rather than a sort of megaregolith below the layer .
since the thermal conductivity @xmath38 of the megaregolith has a complicated relationship with the depth below the surface @xcite , thereby we can not simulate the temperature distribution accurately from a one - dimension thermophysical model .
on the other hand , the formation mechanism for 1996 fg3 is still a mystery , which the yorp acceleration mechanism may play a role in producing its shape and orbital configuration @xcite . in short ,
the investigations by future space missions will throw new light on the formation scenario for this asteroid .
the authors thank the anonymous referee and s.f .
green for their constructive comments that significantly improve the original contents of this manuscript .
this work is financially supported by the national natural science foundation of china ( grants no .
11273068 , 11203087 , 10973044 ) , the natural science foundation of jiangsu province ( grant no .
bk2009341 ) , the foundation of minor planets of the purple mountain observatory , and the innovative and interdisciplinary program by cas ( grant no .
kjzd - ew - z001 ) . | in this paper , a 3d convex shape model of ( 175706 ) 1996 fg3 , which consists of 2040 triangle facets and 1022 vertices , is derived from the known lightcurves .
the best - fit orientation of the asteroid s spin axis is determined to be @xmath0 and
@xmath1 considering the observation uncertainties , and its rotation period is @xmath2 3.5935 h . using the derived shape model
, we adopt the so - called advanced thermophysical model ( atpm ) to fit three published sets of mid - infrared observations of 1996 fg3 @xcite , so as to evaluate its surface properties . assuming the primary and the secondary bear identical shape , albedo , thermal inertia and surface roughness
, the best - fit parameters are obtained from the observations .
the geometric albedo and effective diameter of the asteroid are reckoned to be @xmath3 , @xmath4 km .
the diameters of the primary and secondary are determined to be @xmath5 km and @xmath6 km , respectively .
the surface thermal inertia @xmath7 is derived to be a low value of @xmath8 with a roughness fraction @xmath9 of @xmath10 .
this indicates that the primary possibly has a regolith layer on its surface , which is likely to be covered by a mixture of dust , fragmentary rocky debris and sand .
the minimum regolith depth is estimated to be @xmath11 from the simulations of subsurface temperature distribution , indicating that 1996 fg3 could be a very suitable target for a sample return mission .
[ firstpage ] radiation mechanisms : thermal minor planets , asteroids : individual : ( 175706)1996fg3 infrared : general |
the relevance of phonons for the low - energy properties of high - t@xmath3 oxides is presently rather controversially discussed .
some experiments suggest that the charge carriers near the fermi surface interact only weakly with phonons .
examples are the magnitude and temperature dependence of the resistivity @xcite and the rather small effects in the phonon spectrum caused in general by superconductivity near or below the transition temperature t@xmath3 @xcite . for instance , the superconductivity - induced softening and width of the zone - center buckling mode in yba@xmath4cu@xmath5o@xmath6 corresponds to a rather small electron - phonon ( ep ) coupling constant @xcite .
the recently measured isotope dependence of angle - resolved photoemission spectra ( arpes ) @xcite do not indicate any isotope dependence of the total electronic band width for binding energies down to 200 mev .
this shows the inapplicability of a simple polaron picture , caused by a strong ep interaction , to cuprates because the associated band narrowing would be sensitive to different isotopes @xcite . on the other hand , there exist many observations which have been taken as evidence for a substantial ep interaction in the cuprates .
the measured isotope effects on the transition temperature @xcite and the the superfluid density @xcite , especially in underdoped samples , are two examples .
another example is the large width and softening observed in bond - stretching phonons in hole - doped cuprates in a very small region in @xmath7-space @xcite .
similarly , the absence of a quasi - particle peak in strongly underdoped ca@xmath8na@xmath9cuo@xmath10cl@xmath4
@xcite as well as the doping dependence of a magnetic transition in the frequency - dependent conductivity were attributed to phonons @xcite . finally , electronic self - energy effects in the phonon energy region have been observed by arpes @xcite and by scanning tunneling microscopy @xcite and interpreted in terms of a coupling of electrons to a bosonic mode . whether this mode is related to phonons
is presently unclear but if this is the case the ep coupling and phonons would certainly be important for the low - energy physics of cuprates .
unfortunately , there exists for virtually every interpretation of experiments in favor or disfavor of phonons alternative explanations .
for instance , the large isotope effects on t@xmath3 observed in the underdoped region do not indicate necessarily a large ep coupling but may be caused by the pseudogap @xcite .
similarly , the bosons which interact with electrons near the fermi surface in the interpretation of arpes data may be not phonons but , for instance , spin fluctuations @xcite . in view of these uncertainties
it seems useful to investigate the properties of the ep coupling independently from any interpretation of experiments , i.e. , from first - principles using the local density approximation ( lda ) . to this end
we extend recent investigations on phonon - induced electronic self - energy effects @xcite and study the momentum , frequency dependence and magnitude of the phonon - induced interaction between electrons in detail .
though our approach is based on the lda and deals only with the stoichiometric case it takes many features of these systems such as the complicated phonon spectrum and screening properties self - consistently and realistically into account .
the retarded , phonon - induced electron - electron interaction , multiplied by -1 for convenience , is given in momentum space by @xcite , @xmath11 @xmath12 denotes the renormalized amplitude for a transition from the electronic state with momentum @xmath7 and band index @xmath13 to the state with momentum @xmath14 and band index @xmath15 creating ( annihilating ) a phonon with branch label @xmath16 and momentum @xmath17 @xmath18 . @xmath19
denote the phonon frequencies and @xmath20 a positive infinitesimal .
the amplitudes @xmath21 are obtained within the lda with the efficient linear - response technique @xcite .
we used a 36x36x4-mesh for the electronic momentum @xmath7 within the brillouin zone of yba@xmath4cu@xmath5o@xmath6 , while phonon frequencies and the self - consistent electron - phonon potential were calculated for transferred momenta @xmath22 on a coarser 12x12x4-mesh .
because even for this reduced mesh an exact calculation was numerically very demanding , we adopted a two - step procedure .
first , phonon related quantities were calculated exactly by linear response on a 4x4x2-mesh .
details can be found in refs . and .
these quantities were then approximated on the 12x12x4-mesh by a fourier - interpolation technique . under the usual assumptions it is sufficient to restrict the electronic momenta @xmath23 and @xmath14 in @xmath24 to the fermi surface .
our momentum meshes allow to put @xmath23 practically right on the fermi surface whereas @xmath14 was chosen as near as possible to the fermi surface .
fermi surface averages were calculated including also a gaussian for the one - particle energies with width @xmath25 .
the fermi surface in yba@xmath4cu@xmath5o@xmath6 consists essentially of three bands @xmath26 where @xmath27,@xmath28 , and @xmath29 denote the antibonding , bonding and chain band , respectively .
the upper panels of fig . [ fig:1 ] show the intraband contribution @xmath30 to the static interaction @xmath31 .
we used @xmath32 and @xmath33 , i.e. , all electronic momenta are restricted to the plane @xmath32 .
the left part in this figure refers to a fixed momentum @xmath7 at the nodal , the right part at the antinodal fermi point .
symbols indicate electronic states whose energies differ from the fermi energy by less than a certain threshold energy .
the large filled circles refer to 0.1 ev , the small filled circles to 0.2 ev , the open circles to 0.3 ev .
lines approximate the fermi surface and are obtained by a linear interpolation between neighboring mesh points . the position in @xmath23-space of the mesh points corresponding to the different circles is shown in the insets together with the fermi line depicted by a solid line .
@xmath24 is presented as a function of the angle @xmath34 at @xmath35 between the vectors @xmath36 and @xmath37 .
varying @xmath34 between @xmath38 and @xmath39 means that @xmath40 , seen from the point @xmath41 , moves around the fermi surface in the anticlockwise sense starting from the antinodal point @xmath42 .
the angle @xmath34 is illustrated in the left inset .
it is rather straightforward to construct a continuous curve for @xmath24 from the discrete points discarding only a very few points which correspond to momenta @xmath40 rather far away from the fermi surface and thus should be omitted .
phonon - induced interaction @xmath43 ( upper part ) and @xmath44 ( lower part ) for a fixed momentum @xmath23 at the nodal ( left diagram ) and antinodal ( right diagram ) fermi point as a function of @xmath40 , encircling the fermi surface around the point @xmath35 in anticlock direction , starting from the antinodal point ( see upper left inset ) .
insets show the considered mesh points near the fermi surface described by the solid line . ,
title="fig:",height=117 ] phonon - induced interaction @xmath43 ( upper part ) and @xmath44 ( lower part ) for a fixed momentum @xmath23 at the nodal ( left diagram ) and antinodal ( right diagram ) fermi point as a function of @xmath40 , encircling the fermi surface around the point @xmath35 in anticlock direction , starting from the antinodal point ( see upper left inset ) .
insets show the considered mesh points near the fermi surface described by the solid line . , title="fig:",height=114 ] phonon - induced interaction @xmath43 ( upper part ) and @xmath44 ( lower part ) for a fixed momentum @xmath23 at the nodal ( left diagram ) and antinodal ( right diagram ) fermi point as a function of @xmath40 , encircling the fermi surface around the point @xmath35 in anticlock direction , starting from the antinodal point ( see upper left inset ) .
insets show the considered mesh points near the fermi surface described by the solid line . , title="fig:",height=117 ] phonon - induced interaction @xmath43 ( upper part ) and @xmath44 ( lower part ) for a fixed momentum @xmath23 at the nodal ( left diagram ) and antinodal ( right diagram ) fermi point as a function of @xmath40 , encircling the fermi surface around the point @xmath35 in anticlock direction , starting from the antinodal point ( see upper left inset ) .
insets show the considered mesh points near the fermi surface described by the solid line . , title="fig:",height=114 ] phonon - induced interaction @xmath45 ( upper part ) and @xmath46 ( lower part ) for a fixed momentum @xmath23 at the nodal ( left diagrams ) and antinodal ( right diagrams ) fermi point as a function of @xmath40 , encircling the fermi surface similar as in fig.1 .
insets show the considered mesh points near the fermi surface described by the solid lines . , title="fig:",height=117 ] phonon - induced interaction @xmath45 ( upper part ) and @xmath46 ( lower part ) for a fixed momentum @xmath23 at the nodal ( left diagrams ) and antinodal ( right diagrams ) fermi point as a function of @xmath40 , encircling the fermi surface similar as in fig.1 .
insets show the considered mesh points near the fermi surface described by the solid lines . ,
title="fig:",height=114 ] phonon - induced interaction @xmath45 ( upper part ) and @xmath46 ( lower part ) for a fixed momentum @xmath23 at the nodal ( left diagrams ) and antinodal ( right diagrams ) fermi point as a function of @xmath40 , encircling the fermi surface similar as in fig.1 .
insets show the considered mesh points near the fermi surface described by the solid lines . , title="fig:",height=117 ] phonon - induced interaction @xmath45 ( upper part ) and @xmath46 ( lower part ) for a fixed momentum @xmath23 at the nodal ( left diagrams ) and antinodal ( right diagrams ) fermi point as a function of @xmath40 , encircling the fermi surface similar as in fig.1 .
insets show the considered mesh points near the fermi surface described by the solid lines . ,
title="fig:",height=114 ] phonon - induced interaction @xmath47 ( upper part ) and @xmath48 ( lower part ) for a fixed momentum @xmath23 at the nodal ( left diagrams ) and antinodal ( right diagrams ) fermi point as a function of @xmath40 , encircling the fermi surface similar as in fig.1 .
insets show the considered mesh points near the fermi surface described by the solid lines . , title="fig:",height=117 ] phonon - induced interaction @xmath47 ( upper part ) and @xmath48 ( lower part ) for a fixed momentum @xmath23 at the nodal ( left diagrams ) and antinodal ( right diagrams ) fermi point as a function of @xmath40 , encircling the fermi surface similar as in fig.1 .
insets show the considered mesh points near the fermi surface described by the solid lines . , title="fig:",height=114 ] phonon - induced interaction @xmath47 ( upper part ) and @xmath48 ( lower part ) for a fixed momentum @xmath23 at the nodal ( left diagrams ) and antinodal ( right diagrams ) fermi point as a function of @xmath40 , encircling the fermi surface similar as in fig.1 .
insets show the considered mesh points near the fermi surface described by the solid lines . , title="fig:",height=117 ] phonon - induced interaction @xmath47 ( upper part ) and @xmath48 ( lower part ) for a fixed momentum @xmath23 at the nodal ( left diagrams ) and antinodal ( right diagrams ) fermi point as a function of @xmath40 , encircling the fermi surface similar as in fig.1 .
insets show the considered mesh points near the fermi surface described by the solid lines .
, title="fig:",height=114 ] the right potential curve in fig .
[ fig:1 ] should be symmetric with respect to @xmath49 due to the orthorhombic symmetry which is roughly fulfilled for our discrete mesh .
if tetragonal symmetry would apply the left potential curve should be symmetric with respect to @xmath50 which holds approximately .
finally , if @xmath24 depends only on the transferred momentum @xmath17 , the left potential curve , shifted rigidly by the angle @xmath51 , would coincide with the right potential curve .
this is qualitatively the case , for instance , the two dominating maxima are close to each other after such a shift .
quantitatively there are , however , differences , for instance , the heights of the maxima differ by about 30% and the ( small ) values at zero momentum transfer by about a factor 3 after the shift .
remarkable is the large variation of @xmath24 by about a factor of 5 between small and large momentum transfers , i.e. , between @xmath52 and @xmath53 in the case of the right potential curve .
interesting for the relation between state and transport relaxation times is that @xmath24 is larger for large than for small momentum transfers . the lower panels in fig .
[ fig:1 ] show potential curves for interband scattering between the antibonding and bonding band .
the curves are rather similar to those of the corresponding upper panels in this figure .
the only qualitative difference occurs for @xmath54 in the right - hand panel where a pronounced forward scattering peak at the point @xmath55 emerges .
the upper and lower parts of fig .
[ fig:2 ] show potential curves in the same way as in fig .
[ fig:1 ] , but for the @xmath56 and @xmath57 contributions , respectively .
the intraband potential of the antibonding band looks similar as for the bonding band if @xmath7 is at the antinodal fermi point but very different if @xmath7 is at the nodal point . in the latter case
it is practically independent of @xmath34 and rather small .
this means that @xmath24 depends in this case strongly on @xmath23 and not only on the transferred momentum @xmath17 which may reflect the strong interaction with the chain band .
the interband contribution between the bonding and antibonding band , shown in the lower part of fig .
[ fig:2 ] , looks qualitatively similar as the interband term between the antibonding and the bonding band of fig .
1 exhibiting well - pronounced maxima at large momentum transfers . at small momentum transfers electrons near the nodal direction
interact only weakly whereas those in the antinodal direction develop a second and rather sharp peak in @xmath24 at @xmath58 which , to a lesser degree , was also present in the intraband contribution of the antibonding band .
the absolute magnitude of @xmath24 is in all considered cases similar which means that interband and intraband terms are comparable in magnitude .
[ fig:3 ] shows potential curves for interband scattering between the band a and b , respectively , and the chain band c. due to the geometry of the chain band the angle @xmath34 is restricted to some region around @xmath59 and @xmath60 .
the potentials peak in general around these to values but the absolute values are rather small compared , for instance , to those in fig .
[ fig:1 ] .
the dimensionless coupling function @xmath61 @xcite can be written in terms of @xmath24 as , @xmath62 fig .
[ fig:4 ] shows @xmath61 for @xmath63 ( red squares joined by straight lines ) , @xmath64 ( black circles ) and @xmath65 ( blue triangles ) for @xmath66 ev as a function of the angle @xmath34 . using @xmath32 and @xmath67 yields two curves denoted by filled and empty symbols , respectively , which are close to each other illustrating the weak dependence of @xmath68 on @xmath69 . both for the bonding and antibonding bands
@xmath68 is approximately symmetric with respect to @xmath70 reflecting the tetragonal symmetry of isolated layers
. moreover , the coupling functions are smaller along the nodal direction by about 20 to 30 % compared to the antinodal direction which characterizes the anisotropy of @xmath68 in the layers .
the fermi surface of the chain band starts in our plot only somewhat below the nodal direction and yields a rather strong and rapidly varying coupling function @xmath71 reflecting the strong perturbation of tetragonal symmetry by the chains .
( color online ) coupling functions @xmath72 ( red squares joined by straight lines ) , @xmath73 ( black circles ) , and @xmath71 ( blue triangles ) for @xmath32 ( filled symbols ) and @xmath67 ( empty symbols ) . angle @xmath34 is the same as in figs . 1 , 2 , and 3 . , width=272
] dimensionless coupling constants @xmath74 with @xmath75 can be defined by @xmath76 with the weight functions @xmath77 , @xmath78 , @xmath79 , @xmath80 , and @xmath81 .
@xmath82 denotes the partial density of band @xmath13 in the symmetry channel @xmath34 .
the second , third , and fourth columns in table i show the calculated values for @xmath74 where we approximated the @xmath25-functions in eq .
( [ average ] ) by gaussians with width @xmath66 ev .
the fifth column contains @xmath83 , the sum of the three band contributions each weighted with the factor @xmath84 where @xmath85 is the total density of electronic states at the fermi surface in the channel @xmath34 .
these partial density factors enter the total coupling constant relevant for superconductivity .
.weighted coupling constants . [ cols="^,^,^,^,^",options="header " , ] the numbers in the columns 2 - 4 of table i show that the coupling constants for the three bands are similar in magnitude for each symmetry component .
the isotropic @xmath2-wave component is in all cases substantially larger than those for the remaining `` non - trivial '' symmetries . according to eq .
( [ lambda ] ) @xmath61 and thus also the above coupling constants contain both intraband ( @xmath86 ) and interband ( @xmath87 ) contributions .
keeping only the intraband parts diminishes substantially the numbers in the table .
for instance , for the isotropic @xmath2-wave channel ( first line ) the numbers for @xmath88 change to 0.090,0.069,0.074 , respectively .
this means that interband transitions contribute much more to the coupling functions than the intraband transitions .
considering @xmath89 ev instead of 0.2 ev does not change much our results : for instance , the first three numbers in the first line become 0.226,0.229,0.230 , respectively .
this suggests that our momentum nets are adequate for calculating average quantities such as @xmath74 .
the most interesting numbers in the table are those in the last column which determine the phonon contribution to superconductivity .
isotropic @xmath2-wave dominates by far but this component is ineffective because of the strong coulomb repulsion . in all other symmetry channels
@xmath90 is very small .
this holds in particular for the @xmath0-wave channel where it is only 0.022 , i.e. , one order of magnitude smaller than the isotropic @xmath2-wave value .
this value is rather stable with respect to @xmath25 .
for instance , it changes from 0.022 to 0.027 if @xmath91 ev is reduced to @xmath92 ev . keeping only intraband contributions one obtains 0.021 , i.e.
, the interband transitions are negligible in this case .
( color online ) coupling functions @xmath93 ( black dashed line ) and @xmath94 ( red solid line ) as a function of frequency using a width of 1 mev for the phonon modes .
, width=234 ] using the kramers - kronig transformation for @xmath24 in eq .
( [ average ] ) @xmath95 can be written as an integral from zero to infinity over a frequency - dependent coupling function @xmath96 .
the same holds for the total functions @xmath83 and @xmath97 .
[ fig:5 ] shows @xmath93 ( black dashed line ) and @xmath94 ( red solid line ) , using a width of 1 mev for the phonons . in the case of @xmath93
the phonons give only positive contributions to @xmath98 which are spread out over the whole range of phonon frequencies .
the spectrum is rather peaky because of the usual occurrence of density peaks .
nevertheless , it is clear that it is not possible to attribute the spectrum to a few distinguished phonons such as the breathing or buckling modes or bond - stretching modes . in the case of @xmath94 phonons
give both positive and negative contributions to the spectrum which leads to large cancellations in the integral for @xmath1 and thus to a small value for @xmath1 .
such cancellations would be trivial in the case of a momentum - independent @xmath24 where they would occur for each phonon frequency separately . in our case @xmath24
depends substantially on its two momenta , yet , there are large cancellations , especially between phonons with different frequencies .
a somewhat curious point is that the high - frequency part of @xmath94 is dominated by two peaks at around 40 and 65 mev corresponding roughly to the buckling and breathing phonon frequencies , respectively .
however , these peaks have different signs and thus cancel each other to a large extent in @xmath1 .
we note that the positive peak in @xmath94 at 27 mev is related to an oxygen buckling - type vibration with odd symmetry with respect to the cuo@xmath4 bilayer . in fig .
[ fig:6 ] we show a comparison between the frequency dependence of the total coupling functions with @xmath2 and @xmath99 ( upper panel ) and with @xmath100 and @xmath101 ( lower panel ) symmetries .
as expected the @xmath2-wave coupling function is much larger throughout the whole frequency interval compared to the other symmetries .
remarkable is that the couplings with @xmath100 and @xmath101 symmetries are mainly negative over the whole frequency region and that they differ from each other reflecting the presence of the chain band and the broken tetragonal symmetry .
( color online ) upper panel : coupling functions @xmath93 ( black dashed line ) and @xmath102 ( red solid line ) as a function of frequency using a width of 1 mev for the phonon modes .
lower panel : coupling functions @xmath103 ( blue dashed line ) and @xmath104 ( green solid line ) as a function of frequency using a width of 1 mev for the phonon modes .
, width=264 ] the resistivity in @xmath105-direction is determined by the transport coupling constant @xmath106 , given by @xcite @xmath107 where @xmath108 denotes the average of @xmath109 over all pieces of the fermi surface .
our calculation yields @xmath110 , @xmath111 , and @xmath112 .
these values are close to @xmath98 illustrating the fact that the pronounced momentum dependence of v in figs .
[ fig:1]-[fig:3 ] not necessarily reflects itself in the above coupling constants .
generalizing @xmath24 in eq .
( [ transport1 ] ) to a finite frequency and taking the imaginary part the right - hand side of this equation is proportional to the function @xmath113 from which the temperature dependence of the resistivity can be calculated using eqs .
( 4 ) and ( 7 ) of ref .
the results are shown in fig .
[ fig:7 ] .
( color online ) temperature dependence of the resistivity @xmath114 along the direction @xmath105 .
, width=234 ] with increasing temperature the curves reach between 120 - 150 k a quasi - linear regime which approaches in a very smooth way the true linear region above the highest phonon frequency at around 1000k ( not shown in fig .
[ fig:7 ] ) .
a typical experimental value is @xmath115 ohm cm @xcite , i.e. , only about 10 % of the experimental scattering is due to phonons .
also the noticeable deviations from a linear law between t@xmath3 = 90k and about 150 k in the theoretical curves seems to be in conflict with the experiment for optimally doped samples .
our approach treats electronic correlations only in an approximate way and its applicability to real yba@xmath4cu@xmath5o@xmath6 may be questioned .
unfortunately , the calculation of corrections to our results due to strong electronic correlations is difficult and presently not very reliable .
however , it seems very improbable that our main results , namely that @xmath1 and @xmath116 are about a factor 50 and 10 too small to account for t@xmath3 and @xmath117 , respectively , will be substantially changed by correlations .
experimental data on the width of some phonons are interesting in this respect .
the anomalous broadening of the buckling phonon at the zone center has been quantitatively explained with lda results for the ep - interaction , once a large anharmonic contribution was subtracted @xcite . on the other hand ,
bond - stretching phonons at low temperatures show anomalies around the wave vector @xmath118 in form of a sharp and localized softening and large widths @xcite which clearly are beyond the lda @xcite . at higher temperatures , where these anomalies are not present , the measured widths are still sizable . assuming that these widths are entirely due to the ep - interaction our lda calculation
could only account for about 20 % of these widths .
however , part of the widths could also be caused by anharmonicity like in ref .
furthermore , it is not clear how much these rather localized anomalies in @xmath7-space contribute to fermi surface averaged quantities like @xmath116 .
a large increase of @xmath116 due to correlation - enhanced coupling to bond - stretching phonons could account for the missing scattering contribution found in the lda calculation of @xmath117 . on the other hand
, its temperature dependence would be in conflict with the observed linear temperature between t@xmath3 and room temperature because the onset of the linear dependence in fig .
[ fig:7 ] would be shifted to higher temperatures due to the high frequency of bond - stretching phonons . | the properties of the phonon - induced interaction between electrons are studied using the local density approximation ( lda ) . restricting the electron momenta to the fermi surface
we find generally that this interaction has a pronounced peak for large momentum transfers and that the interband contributions between bonding and antibonding band are of the same magnitude as the intraband ones .
results are given for various symmetry averages of this interaction over the fermi surface .
in particular , we find that the dimensionless coupling constant in the @xmath0-wave channel @xmath1 , relevant for superconductivity , is only 0.022 , i.e. , even about ten times smaller than the small value of the @xmath2-wave channel
. similarly , the lda contribution to the resistivity is about a factor 10 times smaller than the observed resistivity suggesting that phonons are not the important low - energy excitations in high - t@xmath3 oxides . |
the multitype contact process is a stochastic process that can be seen as a model for the evolution of different biological species competing for the occupation of space .
it was introduced by neuhauser in @xcite as a modification of harris ( single - type ) contact process ( @xcite ) .
let us give the definition of the multitype contact process @xmath3 on @xmath4 with ( at most ) two types .
we will need the parameters : @xmath5 , @xmath6 and @xmath7 , @xmath8 , @xmath9 , @xmath10 .
@xmath3 is then the markov process with state space @xmath11 and generator given by @xmath12 , with @xmath13 + \sum_{\substack{x \in \z^d:\\\xi(x ) = 0}}\;\;\sum_{\substack{y \in \z^d:\\ |x - y|\leq r_i,\\\xi(y ) = i } } \lambda_i \cdot [ f(\xi^{i\to x } ) - f(\xi)],\quad i = 1 , 2,\ ] ] where @xmath14 is a function that depends only on finitely many coordinates , @xmath15 is the @xmath16 norm and @xmath17 we will adopt throughout the paper the following terminology : vertices are called _ sites _ , sites in state 0 , 1 and 2 are respectively said to be _ empty _ or to have a type 1 or type 2 _ occupant _ ( or _ individual _ ) , and elements of @xmath11 are called _ configurations_. additionally , @xmath18 are called _ death rates _ , @xmath19 are _ ranges _ and @xmath20 are _ birth rates _ ( or sometimes _ infection rates _ ) .
let us now explain the dynamics in words .
two kinds of transitions can occur .
first , an individual of type @xmath21 dies with rate @xmath22 , leaving its site empty .
second , given a pair of sites @xmath23 with @xmath24 , @xmath25 ( with @xmath26 or @xmath27 ) and @xmath28 , the occupant of @xmath29 gives birth at @xmath30 with rate @xmath31 , so that a new individual of type @xmath21 is placed at @xmath30 .
note that , under these rules , births only occur at empty sites , so that the state of a site can never change directly from 1 to 2 or from 2 to 1 . in case only one type ( say , type 1 )
is present , this reduces to the contact process introduced by harris in @xcite , to be denoted here by @xmath32 in order to distinguish it from the multitype version .
we refer the reader to @xcite for an exposition of the contact process and the statements about it that we will gather in this introduction and in section [ s : back ] .
let @xmath33 be the ( one - type ) contact process with rates @xmath34 , @xmath35 , @xmath36 and the initial configuration in which only the origin is occupied . denote by @xmath37 the configuration in which every vertex is empty , and note that this is a trap state for the dynamics .
@xmath38 there exists @xmath39 ( depending on the dimension @xmath40 and the range @xmath2 ) such that @xmath41 = 1 \quad \text { if and only if } \lambda \leq \lambda_c.\label{eq : phase_transition}\ ] ] this _ phase transition _ is the most fundamental property of the contact process .
the process is called _ subcritical _ , _ critical _ and _ supercritical _ respectively in the cases @xmath42 , @xmath43 and @xmath44 . in this paper
, we will consider the multitype contact process @xmath45 on @xmath0 with parameters @xmath46 we emphasize that the quantity @xmath47 that appears here is the one associated to the _ one - type process _
, as in . we will be particularly interested in the ` heaviside ' initial configuration , @xmath48 we will denote by @xmath49 the process with rates and initial configuration @xmath50 .
we let @xmath51 the interval delimited by @xmath52 and @xmath53 is called the _ interface _ at time @xmath54 , and @xmath55 is the _ position _ of the interface at time @xmath54 .
the choice of the middle point of the interval as the position of the interface is somewhat arbitrary and will not matter for all the results obtained in this paper . in case @xmath56
, it follows readily from inspecting the generator in that @xmath57 for all @xmath54 .
if @xmath58 , both @xmath57 and @xmath59 are possible ( in the latter case we say that we have a _ positive _ interface , and in the previous case , a _ negative _ interface ) . in @xcite , it is shown that the process @xmath60 , which describes the evolution of the _ size _ of the interface , is stochastically tight : @xcite [ thm : size_inter_tight ] if @xmath61 and @xmath62 , then @xmath63 < \varepsilon \text { for all } t \geq 0.\ ] ] in the present paper , we will continue the study of the interface , but we will focus on its position rather than its size . our main result is if @xmath61 and @xmath62 , then there exists @xmath64 such that @xmath65 where @xmath66 denotes brownian motion with diffusive constant @xmath67 , and convergence holds in the space @xmath68 of cdlg trajectories with the skorohod topology .
our proof of this result follows the usual two steps : verifying convergence of finite - dimensional distributions and tightness of trajectories in @xmath69 ( see section 16 of @xcite ) .
we thus prove the following propositions , both applicable to the case @xmath61 and @xmath62 : there exists @xmath64 such that , for any @xmath70 we have @xmath71 where @xmath72 are independent and @xmath73 .
[ prop : fdd ] [ prop : tight ] for any @xmath74 there exists a compact set @xmath75 such that @xmath76 > 1-\varepsilon.\ ] ] in proving these propositions , we will establish a result of independent interest , which we call _
interface regeneration_. we will explain it here only informally ; the precise result depends on a few definitions and is given in theorem [ thm : interface_regeneration ] .
given @xmath77 , consider the configuration @xmath78 and assume the interface position @xmath79 .
suppose we define a new configuration @xmath80 by putting 1 s in all sites to the left of @xmath81 and 2 s to the right of @xmath81 .
we then show that it is possible to construct , in the same probability space as that of @xmath82 , a multitype contact process started from time @xmath83 , @xmath84 , such that @xmath85 and moreover , _ the interface positions for @xmath45 and for @xmath86 are never too far from each other_. since the evolution of the interface of @xmath87 has the same distribution as that of the original process ( except for a space - time shift ) , this regeneration allows us to argue that , if we consider large time intervals @xmath88 , then the displacement of @xmath81 in each interval follows approximately the same law . in many of our proofs ,
we study the time dual of the multitype contact process .
this dual , called the _ ancestor process _ , was first considered in @xcite and further studied in @xcite . in these references
, it was shown that the ancestor process behaves approximately as a system of coalescing random walks on @xmath0 .
because of this , our proofs of propositions [ prop : fdd ] and [ prop : tight ] are inspired in arguments that apply to coalescing random walks and the voter model , an interacting particle system whose dual is ( exactly ) equal to coalescing random walks . in particular , a key estimate for the proof of proposition [ prop : tight ] ( see lemma [ cla4onevstun ] ) was inspired in an argument by rongfeng sun for coalescing random walks ( @xcite ) .
given a set @xmath89 , we denote by @xmath90 its cardinality and by @xmath91 its indicator function .
we will reserve the letter @xmath92 to denote elements of @xmath93 , as well as the one - type contact process , and the letter @xmath94 for elements of @xmath95 and the multitype process .
we denote by @xmath37 the configuration in which every vertex is in state 0 .
we write @xmath96 ( and similarly for @xmath92 ) . given @xmath97 , `` @xmath98 on @xmath89 '' means that @xmath25 for all @xmath99 ( and similarly for @xmath92 ) . throughout the paper
, we fix the parameters @xmath61 and @xmath100 .
all the processes we will consider will be defined from these two parameters . we will now briefly survey some background material on the ( one - type ) contact process . a _
graphical construction _ or _ harris system _ is a family of independent poisson processes on @xmath101 , @xmath102 we view each of these processes as a random discrete subset of @xmath101 .
an arrival at time @xmath54 of the process @xmath103 is called a _ recovery mark _ at @xmath29 at time @xmath54 , and an arrival at time @xmath54 of the process @xmath104 is called an _ arrow _ or _ transmission _ from @xmath29 to @xmath30 at time @xmath54 .
this terminology is based on the usual interpretation that is given to the contact process , namely : vertices are individuals , individuals in state 1 are _ infected _ and individuals in state 0 are _
although we will focus mostly on the multitype contact process , which we see as a model for competition rather than the spread of an infection , we will still use some infection - related terminology that comes from the study of the classical process .
we will sometimes need to consider restrictions of @xmath105 to time intervals , and also translations of @xmath105 .
we hence introduce the following notation , for @xmath106 and @xmath107 : @xmath108 } = d^x \cap [ 0,t],\quad d^x \circ\theta(z , t ) = \{s - t : s\in d^{x - z},\;s\geq t\},\\[.2 cm ] & d^{x , y}_{[0,t ] } = d^{x , y } \cap
[ 0,t],\quad d^{x , y } \circ\theta(z , t ) = \{s - t : s\in d^{x - z , y - z},\;s\geq t\},\\[.2 cm ] & h_{[0,t ] } = \left((d^x_{[0,t]})_{x\in\z},\;(d^{(x , y)}_{[0,t]})_{\substack{x , y\in\z^d,\;0 < |x - y|\leq r}}\right),\\[.2 cm ] & h \circ \theta(z , t ) = \left((d^x\circ \theta(z , t))_{x\in\z},\;(d^{(x , y)}\circ \theta(z , t))_{\substack{x , y\in\z^d,\;0 < |x - y|\leq r}}\right ) .
\label{eq : harris_sub_interval } \end{split}\ ] ] given a ( deterministic or random ) initial configuration @xmath109 and a harris system @xmath105 , it is possible to construct the contact process @xmath110 started from @xmath109 by applying the following rules to the arrivals of the poisson processes in @xmath105 : @xmath111 where @xmath112 is defined as in . that this can be done in a consistent manner , and that it yields a markov process with the desired infinitesimal generator , is a non - trivial result which ( as the other statements in this section ) the reader can find in @xcite .
given @xmath113 , @xmath114 and a harris system @xmath105 , an _ infection path _ in @xmath105 from @xmath115 to @xmath116 is a path @xmath117\to \mathbb{z}$ ] such that @xmath118\text { and } s \in d^{\upgamma(s-),\upgamma(s ) } \text { whenever } \upgamma(s- ) \neq \upgamma(s)\end{split}\ ] ] in case there is an infection path from @xmath115 to @xmath116 , we write @xmath119 in @xmath105 ( or simply @xmath119 if @xmath105 is clear from the context ) .
given sets @xmath120 , and @xmath121 , we write @xmath122 if @xmath123 for some @xmath124 and @xmath125
. we will also write @xmath126 if @xmath127 for some @xmath128 , and similarly @xmath129 . finally
, we convention to put @xmath130 .
when @xmath131 is constructed with the rules and , we have @xmath132 we will always assume that the contact process is constructed from a harris system ( this will also be the case for the multitype contact process , which , as we will explain shortly , can be constructed from the same @xmath105 as the one given above ) .
additionally , we will often consider more than one process at a given time , and implicitly assume that all the processes are built in the same probability space , using the same @xmath105 .
let us now list a few facts and estimates that we will need . by a simple comparison with a poisson process
, we can show that there exist @xmath133 ( depending on @xmath134 and @xmath2 ) such that @xmath135^c\times \{r+t\}\right ] \leq e^{-ct } \text { for any } x\in \z,\ ; r \geq 0,\ ; t > 0.\ ] ] given @xmath97 and @xmath136 , define @xmath137 we write @xmath138 instead of @xmath139 and , in case @xmath140 , we omit it and write @xmath141 and @xmath142 . by and the assumption that @xmath44 , @xmath143 > 0.\ ] ] in case @xmath144 , we write @xmath145 .
similarly , when @xmath146 , we write @xmath147 . in theorem 2.30 in @xcite , we find that there exists @xmath148 ( depending on @xmath2 and @xmath134 ) such that @xmath149 \leq e^{-ct } \text { for any } t > 0\ ] ] and @xmath150
< e^{-c\#a } \text { for any } a \subset \z,\ ; a\neq \varnothing.\ ] ] in the mentioned theorem , these estimates are obtained for the case @xmath56 , but the method of proof is a comparison with oriented percolation that works equally well for @xmath58 . in contrast , the following coupling result has a straightforward proof for @xmath56 but is much harder for @xmath58 .
[ lem : couple_ones ] there exists @xmath151 such that the following holds . for any @xmath74 there
exists @xmath152 such that , if @xmath153 and @xmath154 satisfy , for some @xmath155 , @xmath156;\\[.2 cm ] & \zeta_0(x ) \neq 0 \text { for all } x \in [ a - s , a ) \cup ( b , b+s];\\[.2 cm ] & \zeta'_0(x ) \neq 0 \text { for all } x \in ( -\infty , a ) \cup ( b , \infty)\end{aligned}\ ] ] and @xmath131 , @xmath157 are contact processes started from @xmath109 and @xmath158 and constructed with the same harris system , then , with probability larger than @xmath159 , @xmath160 \text { and } t \geq 0.\ ] ] the proof follows from proposition 2.7 in @xcite ( see also the treatment of the event @xmath161 in page 11 of that paper ) .
the key idea is an event which the authors called the formation of a _ descendancy barrier _ ; this means that in a space - time set of the form @xmath162 : t \geq 0\}$ ] , every vertex that is reachable by an infection path from @xmath163 is reachable by an infection path from @xmath164 . for the statement of the present lemma , it would suffice to argue that , if @xmath165 is large , with high probability one can find @xmath166 and @xmath167 $ ] so that a descendancy barrier is formed from both @xmath29 and @xmath30 .
the above lemma also holds , with the same proof , for @xmath168 or @xmath169 .
[ lem : desc_bar_sides ] for any @xmath74 and @xmath170 with @xmath171 there exists @xmath172 such that , if @xmath153 and @xmath173 for all @xmath174 $ ] , then with probability larger than @xmath159 , @xmath175 \text { such that } \zeta_t(x ) = 1.\ ] ] by lemma [ lem : couple_ones ] , it suffices to prove that , given @xmath176 , there exists @xmath177 so that , for @xmath153 , @xmath178 $ with } \zeta^\z_t(x ) = 1\right ] > 1-\varepsilon,\ ] ] where @xmath179 denotes the one - type contact process started from full occupancy .
for any @xmath54 we have @xmath180\right]&=\p\left[\z\times \{0\ } \nleftrightarrow ( x , t ) \text { for all } x \in [ -s+\sigma t,\;s + \sigma't]\right]\\[.2 cm ] & = \p\left[(x,0 ) \nleftrightarrow \z\times \{0\ } \text { for all } x \in [ -s + \sigma t,\;s + \sigma't]\right]\\[.2cm]&\leq \p\left[t^{[-s+\sigma t,\;s+\sigma ' t ] } < \infty\right ] \stackrel{\eqref{eq : many_die } } { < } e^{-2cs+ c(\sigma'-\sigma ) t}.\end{aligned}\ ] ] also , using the strong markov property it is easy to verify that , for some @xmath181 that depends on @xmath182 and @xmath2 but not on @xmath54 , @xmath183 > \delta \cdot \p\left[\zeta^\z_s \equiv 0 \text { on } [ -s + \sigma s,\ ; s + \sigma 's ] \text { for some } s \in [ t-1 , t]\right]\ ] ] for all @xmath184 . in conclusion , @xmath185\right ] \\&\leq \sum_{t=1}^\infty \p\left[\text{there exists } s \in
[ t-1 , t ] : \zeta^\z_t \equiv 0 \text { on } [ -s + \sigma s,\;s + \sigma 's ] \right]\leq \frac{1}{\delta } \sum_{t=1}^\infty e^{-2cs + c(\sigma'-\sigma)t},\end{aligned}\ ] ] which is smaller than @xmath186 if @xmath165 is large enough . * graphical construction . * due to our choice of parameters in , it is possible to construct the multitype contact process with the same graphical construction @xmath105 as the one we have given in for the single - type process .
the effects of recovery marks and arrows are : @xmath187 where @xmath188 is defined in .
these rules lead to the correct transition rates , as prescribed in and
. it will often be convenient to construct several processes , one - type or multitype or both , in the same probability space and using a single realization of @xmath105 .
when we do so , the following will be quite useful .
if @xmath82 , @xmath189 are constructed with the same harris system , @xmath190 simply consider the partial order @xmath191 and note that the rules and preserve this order .
we will keep using the infection paths of @xmath105 , as defined in .
we can obtain @xmath192 from and similarly to how is obtained from and .
in particular , @xmath193 this is quite convenient , but there is a drawback : in case @xmath194 , it is not so simple to deduce its value from @xmath195 and the infection paths in @xmath105 .
that is because an infection path is only successful in carrying type @xmath196 if all of the path s arrows land on space - time points that are in state 0 . to deal with this
, we will need some extra definitions . given a realization of @xmath197 and @xmath97 , we define @xmath198 obviously , any infection path of @xmath199 is also an infection path of @xmath105 .
it is also readily seen that the one - type process with initial occupancy in @xmath89 is the same whether it is built using @xmath105 or @xmath199 , that is , @xmath200 additionally , in @xmath199 , the number of infection paths from @xmath201 to any @xmath115 is either 0 or 1 ( corresponding respectively to the cases @xmath202(x ) = [ \zeta^a_t(h)](x ) = 0 $ ] and @xmath202(x)= [ \zeta^a_t(h)](x ) = 1 $ ] ) . in the second case , the unique infection path in @xmath199 from @xmath203 to @xmath115
is denoted @xmath204\to\z$ ] .
we can also characterize @xmath205 as the unique infection path in @xmath105 satisfying @xmath206 now fix @xmath207 and assume @xmath3 is the multitype contact process started from @xmath195 and constructed with a harris system @xmath105 .
we claim that @xmath208 indeed , if the right - hand side is zero , then the indicator function is zero ( as the other term is non - zero by construction ) , so @xmath209 holds by .
if the right - hand side of is non - zero , then the definition of infection paths together with , and imply that @xmath210 for every @xmath211 $ ] , so the equality also follows . these considerations are summarized as follows : let @xmath82 be the multitype contact process started from a fixed @xmath212 and constructed with a harris system .
then , @xmath213\to\z^d$}\\\text{such that } \xi_0(\upgamma(0 ) ) = i,\ ; \upgamma(t ) = x \text { and } \\
\xi_{s-}(\upgamma(s ) ) = 0 \text { whenever } \upgamma(s ) \neq \upgamma(s-)\end{array},\quad i = 1,2.\ ] ] moreover , there exists at most one infection path satisfying the stated properties .
* ancestry process .
* we now define an auxiliary process that is key in making the graphical construction of the multitype contact process more tractable .
again fix a harris system @xmath105 and let @xmath214 .
given @xmath215 , by arguing similarly to how we did in the previous paragraphs , it can be shown that @xmath216 \to \z\text { such that } \uppsi(r ) = x \text { and}\\ ( \uppsi(s-),s)\nleftrightarrow \z\times \{t\ } \text { whenever }
\uppsi(s- ) \neq \uppsi(s).\end{array}\ ] ] in case it exists , we denote this path by @xmath217 , or @xmath218 when we want to make the dependence on the harris system explicit .
note that @xmath217 only depends on @xmath219 $ ] .
we claim that , for @xmath220 , @xmath221.\ ] ] indeed , defining @xmath222 \to \z$ ] by @xmath223;\\[.2cm]\uppsi^*_{y , t , t'}(s ) & \text{if } s \in [ t , t'],\end{array}\right.\ ] ] we have that * if @xmath224 $ ] and @xmath225 , then by the definition of @xmath226 , we have that @xmath227 , so @xmath228 ; * if @xmath229 $ ] and @xmath225 , then by the definition of @xmath230 , we have that @xmath228 , so that , by the uniqueness of @xmath231 , we get @xmath232 , so follows .
we define , for @xmath233 and @xmath234 , @xmath235\triangle & \text{otherwise,}\end{array}\right.\ ] ] where @xmath236 is interpreted as a `` cemetery '' state . the process @xmath237 is called the _ ancestor process _ of @xmath238 . in case @xmath239 , we write @xmath240 instead of @xmath241 , and in case @xmath242 , we omit the superscript and write @xmath243 .
naturally , @xmath244 now can be rewritten as @xmath245 in particular , we get @xmath246 * joint construction of primal and dual processes . * we now explain the relationship between the multitype contact process and the ancestor process . given a harris system @xmath247 and @xmath106 , we recall the notation introduced in and define the _ reversed harris system _ @xmath248}$ ] by @xmath249 } = \left((\hat d^x_{[0,t]})_{x \in \z},\;(\hat d^{x , y}_{[0,t]})_{x , y\in\z,\;0<|x - y|\le r}\right),\text { where } \\[.2 cm ] & \hat d^x_{[0,t ] } = \{t - s : s \in d^x \cap [ 0,t]\ } \text { and } \hat d^{x , y}_{[0,t ] } = \{t - s : s \in d^{y , x } \cap [ 0,t]\}. \end{split}\ ] ] in words , @xmath248}$ ] is the harris system on the time interval @xmath250 $ ] obtained from @xmath251 $ ] by reversing time and reversing the direction of the arrows .
assume we are given @xmath212 and construct @xmath45 started from @xmath195 using the harris system @xmath105 .
fix @xmath106 and assume that we use @xmath248}$ ] to construct the ancestor processes @xmath252 } ) : 0 \leq s \leq t,\ ; x\in \z.\ ] ] one immediate consequence of this joint construction is that @xmath253 more interestingly , by and the definition of the ancestor process , @xmath254 indeed , the @xmath248}$]-infection path @xmath255 , when ran backwards and with arrows reversed , corresponds exactly to the @xmath105-infection path @xmath256 . as a consequence of these considerations
, we have if @xmath257 for all @xmath258 , then , with the convention that @xmath259 , @xmath260 we now recall the renewal structure from which we are able to decompose the ancestor process into pieces that are independent and identically distributed .
this then allows us to find an embedded random walk in @xmath261 and argue that the whole of the trajectory of @xmath261 remains close to this embedded random walk .
most of the results of this subsection are not new ( they appear in @xcite or @xcite or both ) ; in an effort to balance the self - sufficiency of this paper with shortness of exposition , we will include a few key proofs and omit others . [
lem : no_renewals_ab ] there exists @xmath148 such that , for any @xmath262 , we have @xmath263 \right ] < e^{-c(b - a)}.\ ] ] the proof is a simple consequence of ; see proposition 1 , page 474 , of @xcite
. given @xmath97 , we write @xmath264 = \p[\;\cdot\;|\;(y,0)\leftrightarrow \infty \text { for all } y \in a].\ ] ] in case @xmath265 , we write @xmath266 instead of @xmath267 and in case @xmath268 , we omit the superscript .
[ lem_x_goes_to_y_first ] let @xmath269 and @xmath270 for any @xmath271 and events @xmath272 on harris systems , @xmath273 } \in e,\;\eta_\uptau = y\text { and } h \circ
\theta(0,\uptau ) \in f \right ] \\[.2 cm ] & = \p\left[\uptau < \infty,\ ; h_{[0,\uptau ] } \in e,\;\eta_\uptau = y\right ] \cdot \tilde \p^{y } \left[h \in f\right ] .
\end{split}\ ] ] we let @xmath274 and , for @xmath275 , define @xmath276 as follows : @xmath277 \sigma_k&\text{if } \sigma_k < \infty \text { and } \eta_{\sigma_k } = \triangle;\\[.2 cm ] \infty & \text{if } \sigma_k = \infty . \end{cases}\ ] ] @xmath278 is thus an increasing sequence of stopping times with respect to the sigma - algebra of harris systems .
we note that , in case we have @xmath279 , then gives @xmath280 for all @xmath281 .
so we have @xmath282 as a consequence , we obtain @xmath283 using , the left - hand side of becomes @xmath284 } \in e;\;\eta_{\sigma_k } = y \text { and } ( y,\sigma_k ) \leftrightarrow \infty;\ ; h\circ \theta(0,\sigma_k ) \in f \right]\\ & = \tilde \p^{y}\left [ h \in f\right ] \cdot \sum_{k=0}^\infty \p\left[\sigma_k < \infty;\;h_{[0,\sigma_k ] } \in e;\;\eta_{\sigma_k } = y \text { and } ( y,\sigma_k)\leftrightarrow \infty \right]\\ & = \tilde \p^{y}\left [ h \in f\right ] \cdot \p\left [ \uptau < \infty;\ ; h_{[0,\uptau ] } \in e;\;\eta_\uptau = y \right].\end{aligned}\ ] ] given @xmath285 , on the event @xmath286 we define the times @xmath287 we write @xmath288 instead of @xmath289 and @xmath290 instead of @xmath291 .
we now state three simple facts about these random times .
first , it follows from that @xmath292 second , from it is easy to obtain @xmath293 = 1.\ ] ] third , by putting and together , it is easy to show that @xmath294 \leq e^{-ct},\qquad t > 0.\ ] ] our main tool in dealing with the ancestor process is the following result .
[ prop : rmn ] 1 . under @xmath295 , @xmath296 in particular
, @xmath297 is a random walk on @xmath0 with increment distribution @xmath298 = \tilde \p\left[\eta_{\uptau_1 } = w - z\right].\ ] ] 2 .
there exist @xmath299 such that , for any @xmath136 , @xmath300 and @xmath301 , @xmath302}|\eta_s - \eta_t| > x\right ] \leq ce^{-cx^2/r } + cre^{-c|x|}.\ ] ] 3 . under @xmath295 , @xmath303 a proof of part can be found in @xcite , but we give another one here .
let @xmath304 be measurable subsets of @xmath305 $ ] , the space of finite - time trajectories that are right - continuous with left limits .
we evaluate @xmath306\\[.2 cm ] & = \p\left[(0,0)\leftrightarrow \infty\right]^{-1 } \cdot \sum_{z \in \z } \p\left[\begin{array}{l}(0,0 ) \leftrightarrow \infty , \ ; \left(\eta_s\right)_{0 \leq s \leq \uptau_1 } \in e_0,\;\eta_{\uptau_1 } = z,\\[.2cm]\left(\eta_s - \eta_{\uptau_i}\right)_{\uptau_i < s \leq \uptau_{i+1 } } \in e_i \text { for } i = 1 , \ldots , k\end{array}\right]\nonumber\\[.2 cm ] & \stackrel{\eqref{eq : tau_is_fin}}{= } \p\left[(0,0)\leftrightarrow \infty\right]^{-1 } \cdot \sum_{z \in \z } \p\left[\begin{array}{l}\uptau_1 < \infty , \ ; \left(\eta_s\right)_{0 \leq s \leq \uptau_1 } \in e_0,\;\eta_{\uptau_1 } = z,\\[.2cm]\left(\eta_s - \eta_{\uptau_i}\right)_{\uptau_i < s \leq \uptau_{i+1 } } \in e_i \text { for } i = 1 , \ldots , k\end{array}\right ] \label{eq : oaid_uptau}\end{aligned}\ ] ] now , by and , @xmath307 applying these identities and lemma [ lem_x_goes_to_y_first ] , we obtain that is equal to @xmath308^{-1 } \cdot \sum_{z \in \z } \p\left[\uptau_1 < \infty , \ ; \left(\eta_s\right)_{0 \leq s \leq \uptau_1 } \in e_0,\;\eta_{\uptau_1 } = z\right]\\&\qquad\qquad\qquad\qquad\qquad \cdot \tilde \p^z\left [ \left(\eta^{z}_s - \eta^{z}_{\uptau_i}\right)_{\uptau_i <
s \leq \uptau_{i+1 } } \in e_{i+1 } \text { for } i = 0 , \ldots , k- 1\right]\\[.2 cm ] & = \p\left[(0,0)\leftrightarrow \infty\right]^{-1 } \cdot \sum_{z \in \z } \p\left[\uptau_1 < \infty , \ ; \left(\eta_s\right)_{0 \leq s \leq \uptau_1 } \in e_0,\;\eta_{\uptau_1 } = z\right]\\&\qquad\qquad\qquad\qquad\qquad \cdot \tilde \p\left [ \left(\eta_s - \eta_{\uptau_i}\right)_{\uptau_i < s \leq \uptau_{i+1 } } \in e_{i+1 } \text { for } i = 0 , \ldots , k- 1\right]\\[.2 cm ] & = \tilde\p\left[\left(\eta_s\right)_{0 \leq s \leq \uptau_1 } \in e_0 \right ] \cdot \tilde \p\left [ \left(\eta_s - \eta_{\uptau_i}\right)_{\uptau_i < s \leq \uptau_{i+1 } } \in e_{i+1 } \text { for } i = 0 , \ldots , k- 1\right ] .
\end{aligned}\ ] ] we now iterate this computation to obtain . for the remaining statements , we will need a definition . on the event @xmath309 ,
let @xmath310 since we have @xmath311 for all @xmath312 , we obtain @xmath313 we now turn to . the left - hand side is less than @xmath314 + \tilde \p\left[\sup_{t\leq s\leq t+r } |\eta_s-\eta_t| > x\right ] .
\label{eq:12auxrw}\end{aligned}\ ] ] the first term is less than @xmath315 \leq \p\left[\frac{x}{2\kappa } <
t^0 < \infty\right ] + \p\left[\sup_{0\leq s \leq x/(2\kappa ) } |\eta_s| > \frac{x}{2 } \right],\ ] ] where @xmath316 is as in . then , and show that the sum is less than @xmath317 for some @xmath148 . the second term in is less than @xmath318 + \tilde \p \left[\max_{n_t \leq k \leq n_{t+r}}\ ; \sup_{s\in[\uptau_{k } , \uptau_{k+1 } ] } |\eta_{s } - \eta_{\uptau_{k}}| >
\frac{x}{2 } \right]\\[.2 cm ] & \stackrel{\eqref{eq : inc_iid},\eqref{eq : separation}}{\leq } \tilde \p\left[\max_{0 \leq k \leq \lceil r \rceil } |\eta_{\uptau_k}| > \frac{x}{2 } \right ] + \lceil r \rceil \cdot \tilde \p\left[\sup_{s \in [ 0,\uptau_1 ] } |\eta_{s}| > \frac{x}{2 }
\right]\\[.2 cm ] & \hspace{.5cm}\stackrel{\eqref{eq : bht } } { \leq } \tilde \p\left[\max_{0 \leq k
\leq \lceil r \rceil } |\eta_{\uptau_k}| > \frac{x}{2 } \right ] + \lceil r \rceil e^{-cx}.\end{aligned}\ ] ] now , using standard random walk estimates ( see for example proposition 2.1.2 in @xcite ) , we can bound the first term above by @xmath319 .
this completes the proof of .
finally , let us prove .
denote @xmath320 in lemma 2.5 in @xcite , it is shown that @xmath321 \leq e^{-cx } , \qquad t > 0,\ ; x > 0.\ ] ] we write @xmath322 by and the central limit theorem , @xmath323 converges in distribution , as @xmath324 , to @xmath325 with @xmath64 .
using , we have that @xmath326 converges in probability , as @xmath327 , to zero .
hence , will follow if we prove that the remaining term also satisfies @xmath328with this aim , fix @xmath74 . for any @xmath181 we have @xmath329\leq \tilde \p\left[\frac{n_t}{t } - \frac{1}{\mu }
> \delta \right ] \\&+ \tilde \p\left[\frac{\eta_{n_t } - \eta_{\lfloor t/\mu\rfloor}}{\sqrt{t } } > \varepsilon,\ ; n_t \in \left [ \frac{t}{\mu}-\delta t,\ ; \frac{t}{\mu}\right]\right ] + \tilde \p\left[\frac{\eta_{n_t } - \eta_{\lfloor t/\mu\rfloor}}{\sqrt{t } } > \varepsilon,\ ; n_t
\in \left [ \frac{t}{\mu},\ ; \frac{t}{\mu } + \delta t\right]\right]\label{eq : almost_conv}\end{aligned}\ ] ] by the renewal theorem , @xmath330 \to 0 $ ] as @xmath327 .
next , @xmath331\right ] & \leq \tilde \p\left[\max_{\frac{t}{\mu } \leq i \leq \frac{t}{\mu } + \delta t } |\eta_{\uptau_i } - \eta_{\uptau_{\lfloor t/\mu \rfloor}}| > \varepsilon\sqrt{t}\right ] \\[.2cm]&\leq \delta t \frac{\text{var}(\eta_{\uptau_1})}{\varepsilon^2 t } = \delta \frac{\text{var}(\eta_{\uptau_1})}{\varepsilon^2},\end{aligned}\ ] ] where the last inequality is an application of kolmogorov s inequality .
the above can be made arbitrarily small by taking @xmath332 small ( depending on @xmath186 ) .
the other term in is then treated similarly , and the proof of is now complete . in @xcite , results are obtained about the joint behavior of two or more ancestor processes .
the method used to obtain such results involved studying renewal times that are more complicated then the @xmath333 defined above .
we will not present the details here .
rather , let us just mention that , while a single ancestor behaves closely to a random walk ( as outlined above ) , a larger amount of ancestors , when considered jointly , behave closely to a system of coalescing random walks ( that is , a system of random walkers that move independently with the added rule that two walkers that occupy the same position merge into a single walker ) .
taking advantage of this comparison , one can then obtain for ancestor processes several estimates that hold for coalescing random walks .
in particular , in lemma 3.2 in @xcite , it is shown that @xmath334 \leq \frac{c|x - y|}{\sqrt{t}},\ ; x , y \in \z,\ ; t > 0.\ ] ] using this result , it is then possible to show that the density of the set of _ all _ ancestors at time @xmath54 , @xmath335 , goes to zero as @xmath336 ( see proposition 3.5 in @xcite ) , so that @xmath337 \xrightarrow{t
\to \infty } 0.\ ] ] finally , we will need the bound @xmath338 < \frac{c}{\sqrt{t}}. \end{split}\ ] ] for coalescing random walks having symmetric jump distribution with finite third moments , this estimate is given by lemma 2.0.4 in @xcite .
as @xmath339 and @xmath340 are not exactly coalescing random walks , the proof of the mentioned lemma has to be adapted to the present context . given the method of proof of theorem 6.1 in @xcite , this adaption does not involve anything new , so we do not include it here
. given @xmath341 , we write @xmath342 define @xmath343&\#\{x<0:\xi(x ) = 2\ } <
\infty,\;\#\{x>0:\xi(x ) = 1\ } < \infty \end{array}\right\};\ ] ] in particular , @xmath344 and @xmath345 for any @xmath346 . as mentioned in the introduction
, @xmath49 denotes the contact process started from the heaviside configuration , , and @xmath347 the interval delimited by @xmath52 and @xmath53 is the _ interface _ , and @xmath55 is the _ interface position _ , at time @xmath54 . using , it is easy to show that , almost surely , @xmath348 it will be useful to have the following rough bound on the displacement of @xmath52 and @xmath53 .
[ lem : no_faster ] for any @xmath74 and @xmath64 there exists @xmath152 such that , if @xmath153 and @xmath195 satisfies @xmath349 on @xmath350 , then with probability larger than @xmath351 , @xmath352 it is sufficient to prove the result for @xmath353 , where @xmath354 is the constant that appears in lemmas [ lem : couple_ones ] and [ lem : desc_bar_sides ] .
we fix @xmath355 with @xmath356 using the joint construction of the multitype contact process and the ancestor processes ( as described in subsection [ ss : mcp ] and in particular equation ) together with the assumption that @xmath349 on @xmath350 and claim [ eq : attract_multi ] , we have @xmath357 \leq \p\left[\xi^h_t(x ) = 1\right]\leq \p\left[\eta^x_t\neq \triangle,\ ; \eta^x_t \leq 0 \right].\end{aligned}\ ] ] if @xmath301 , the right - hand side is smaller than or equal to @xmath358 = \p\left[\eta_t \neq \triangle , \;|\eta_t| \geq x \right ] \stackrel{\eqref{eq : bondrw}}{\leq } ce^{-cx^2/t } + cte^{-cx}.\ ] ] combining this with a union bound , we get @xmath359 = \p\left[\xi_t(x ) = 1 \text { for some } t
\in \n \text { and } x \geq \frac{s}{3 } + \sigma't\right ] < \frac{\varepsilon}{3}\label{eq : bound_before_intervals}\ ] ] if @xmath165 is large enough
. we then bound @xmath360 \right]\\[.2 cm ] & \leq \p\left[(-\infty,\;s/3 + \sigma ' t ) \times \{t\ } \leftrightarrow
[ 2s/3 + \sigma '' t,\infty ) \times [ t , t+1 ] \right ] <
e^{-c(s + \sigma t/2)}\end{aligned}\ ] ] for some @xmath148 , by a comparison with a poisson random variable ( describing the number of arrivals in a certain space - time region ; we omit the details ) .
together with , this shows that , if @xmath165 is large enough , @xmath361 < \varepsilon/2.\ ] ] by lemma [ lem : desc_bar_sides ] and , increasing @xmath165 if necessary we have @xmath362 \text { for some } t \geq 0 \right ] < \varepsilon/2.\ ] ] to conclude , @xmath363\\&\geq \p\left[\text{for all } t \geq 0,\;r(\xi_t ) < \frac{2s}{3 } + \sigma''t \text { and } \xi_t(x ) \neq 0 \text { for some } x \in \left[\frac{2s}{3}+\sigma''t , s + \sigma t\right ] \right ] > 1-\varepsilon .
\end{aligned}\ ] ] given a harris system @xmath105 and @xmath364 , we define the regenerated interface process @xmath365 as follows : @xmath366 let us explain this definition in words . using the harris system @xmath105 , we construct the contact process started from the heaviside configuration @xmath50 and evolve it up to time @xmath83 , obtaining the configuration @xmath367 with corresponding interface position @xmath81 .
then , we artificially put 1 s on @xmath368 $ ] and 2 s on @xmath369 and , using @xmath370}$ ] , we continue evolving the process ; the resulting interface position at time @xmath54 is @xmath371 . note in particular that @xmath372 in section [ ss : proof_reg ] , we will prove : [ thm : interface_regeneration ] for any @xmath74 there exists @xmath373 such that , for any @xmath364 , @xmath374 > 1-\varepsilon.\ ] ] as a consequence we obtain [ cor : interface_motion_tight ] for any @xmath74 and @xmath300 there exists @xmath373 such that @xmath375 <
\varepsilon \quad \text { for all } s \ge 0.\ ] ] for any @xmath376 , by , @xmath377 } |i_t - i_s| > k \right ] \leq
\p\left[\sup_{t \in [ s,\infty ) } |i^s_t - i_t| > k/2\right ] + \p \left [ \sup_{t \in [ 0,r ] } |i_t| > ( k-1)/2\right].\end{aligned}\ ] ] now , for fixed @xmath378 , the second term vanishes as @xmath379 , and the first term does so as well by theorem [ thm : interface_regeneration ] .
[ lem : interface_nowhere ] for any @xmath74 there exists @xmath269 such that @xmath380 < \varepsilon \text { for any } x \in \frac{1}{2}\z,\ ; t \geq t_0.\ ] ] let @xmath74 . by , we can obtain @xmath381 such that for all @xmath54 , @xmath382 < \varepsilon /2 $ ] . for any @xmath54 and @xmath29 we have @xmath383 \leq \p[|r_t - \ell_t| >
l ] + \p \left [ \text{there exist } z , w \in [ x - l,\;x+l ] : \xi^h_t(z ) = 1,\;\xi^h_t(w ) = 2\right].\ ] ] switching to the dual process , the second probability can be written as @xmath384 : \eta^z_t,\;\eta^w_t \neq \triangle,\ ; \eta^z_t \leq 0,\ ; \eta^w_t > 0\right]\\[.2 cm ] & \hspace{4cm}\leq \sum_{z , w \in [ x - l,\;x+l ] } \ ; \p\left[\eta^z_t , \eta^w_t \neq \triangle,\ ;
\eta^z_t \neq \eta^w_t\right].\end{aligned}\ ] ] by , then @xmath54 is large enough the sum is smaller than @xmath385 for any @xmath29 , so we are done .
[ lem : coinc_eta_inter ] for any @xmath74 there exists @xmath269 such that @xmath386 - \p\left[\xi^h_t(\lfloor x\sqrt t \rfloor ) = 1\;|\ ; \xi^h_t(\lfloor x\sqrt{t}\rfloor ) \neq 0\right]\right| < \varepsilon \text { for any } x \in \r\text { and } t \geq t_0.\ ] ] fix @xmath74 . using
, we can choose @xmath387 such that @xmath388 < \varepsilon\ ] ] using corollary [ cor : interface_motion_tight ] , we then choose @xmath389 such that @xmath390:\;|i_s - i_t| \geq s'\right ] <
\varepsilon \text { for all } t \geq 0.\ ] ] increasing @xmath391 if necessary , by we can also assume that @xmath392 < \varepsilon \text { for all } t \geq 0;\\[.2cm]\ ] ] finally , using lemma [ lem : interface_nowhere ] , we can choose @xmath393 such that @xmath394\right ] < \varepsilon \quad \text { for all } t \geq 0 \text { and } x \in \z.\ ] ] now fix @xmath395 and @xmath396 . denoting by @xmath236 the symmetric difference between sets , we have the following estimates : @xmath397 \\&\hspace{3cm}\leq \p\left[|i_t - x\sqrt{t}| \leq s ' \right ] + \p\left[|r_t - \ell_t| > s'\right ] \stackrel{\eqref{eq : recap_interface},\eqref{eq : recap_nowhere}}{\leq } 2\varepsilon ; \end{split}\\[.2 cm ] \label{eq : rcp_bound_2}\begin{split } & \p\left[\{i_{t - s } >
x\sqrt{t}\}\;\triangle\;\{i_t > x\sqrt{t}\ } \right ] \\&\qquad\leq \p\left[|i_t - x\sqrt{t}| \leq s'\right ] + \p\left[|i_{t - s } - x\sqrt{t}| \leq s'\right ] + \p\left[|i_t - i_{t - s}| > s'\right ]
\stackrel{\eqref{eq : recap_move_fast } , \eqref{eq : recap_nowhere } } { < } 3\varepsilon ; \end{split}\\[.2 cm ] \begin{split}&\p\left[\z\times \{t - s\ } \;\leftrightarrow \ ; ( x\sqrt{t},t),\;\z\times\{0\ } \;\nleftrightarrow\;(x\sqrt{t},t)\right]\\&\hspace{4.5 cm } \leq \p\left[(0,0)\;\leftrightarrow \z\times \{s\},\ ; ( 0,0)\;\nleftrightarrow \;\infty \right ] \stackrel{\eqref{eq : recap_die_fast } } { < } \varepsilon.\end{split}\label{eq : rcp_bound_3}\end{aligned}\ ] ] with these bounds at hand , we are ready to prove the statement of the lemma . in the following computation , the symbol @xmath398 means that the absolute value of the difference between the left - hand side and the right - hand side is at most @xmath399 .
@xmath400 & \stackrel{\eqref{eq : rcp_bound_1}}{\approx } \p\left[\xi^h_t(\lfloor x\sqrt{t}\rfloor ) \neq 0,\ ; i_t > x\sqrt{t}\right]\\[.2 cm ] & \stackrel{\eqref{eq : rcp_bound_2},\eqref{eq : rcp_bound_3}}{\approx } \p\left[\z \times \{t - s\ } \;\leftrightarrow \ ; ( \lfloor x\sqrt{t}\rfloor , t),\ ; i_{t - s } > x \sqrt{t}\right]\\[.2 cm ] & = \p\left[\z \times \{t - s\ } \;\leftrightarrow \ ; ( \lfloor x\sqrt{t } \rfloor , t)\right ] \cdot \p\left[i_{t - s } > x\sqrt{t}\right]\\[.2 cm ] & \stackrel{\eqref{eq : rcp_bound_3}}{\approx } \p\left[\xi^h_t(\lfloor x\sqrt{t}\rfloor ) \neq 0\right ] \cdot \p\left[i_t > x\sqrt{t}\right].\end{aligned}\ ] ] we then have @xmath401 -\p\left [ \xi^h_t(\lfloor x\sqrt{t}\rfloor ) = 1\;|\ ; \xi^h_t(\lfloor x\sqrt{t}\rfloor)\neq 0\right]\right|\\[.2 cm ] & = \p\left[\xi^h_t(\lfloor x\sqrt{t}\rfloor ) \neq 0\right]^{-1}\cdot \left|\p\left [ \xi^h_t(\lfloor x\sqrt{t}\rfloor ) \neq 0\right]\cdot \p\left[i_t > x\sqrt{t}\right ] - \p\left[\xi^h_t(\lfloor x \sqrt{t}\rfloor ) = 1\right ] \right|\\[.2 cm ] & \leq \p[t^0 = \infty]^{-1 } \cdot 15\varepsilon,\end{aligned}\ ] ] that is , at most a universal constant times @xmath186 .
this completes the proof .
[ prop:1dd ] as @xmath336 , @xmath402 converges in distribution to @xmath325 , where @xmath67 is as in .
the statement follows from , lemma [ lem : coinc_eta_inter ] and the fact that @xmath403 = \tilde \p^{\lfloor x \sqrt{t}\rfloor}\left [ \eta^{\lfloor x \sqrt{t } \rfloor}_t \leq 0\right ] = \tilde \p^0\left[\frac{\eta^0_t}{\sqrt{t } } \leq - \frac{\lfloor x \sqrt{t } \rfloor}{\sqrt{t}}\right].\ ] ] _ proposition [ prop : fdd ] .
_ fix @xmath404 .
we have @xmath405 theorem [ thm : interface_regeneration ] implies that the term in converges to zero in probability as @xmath336 .
since the elements of the vector in are independent and satisfy @xmath406 proposition [ prop:1dd ] shows that converges in distribution , as @xmath336 , to the distribution prescribed in proposition [ prop : fdd ] .
most of the effort in this section will go into proving the following uniform bound on the displacement of the interface position . [
lem : inter_not_move ] for any @xmath74 there exists @xmath407 such that , for large enough @xmath54 and any @xmath300 , @xmath408 > 1 - \frac{\varepsilon}{u^2}.\ ] ] the proof will depend on several preliminary results . before turning to them ,
let us first explain how lemma [ lem : inter_not_move ] allows us to conclude .
+ _ proposition [ prop : tight ] .
_ for each @xmath106 , define the process @xmath409 by @xmath410 we want to show that the family of processes @xmath411 is tight in @xmath412 . as explained in section 16 of @xcite , it is sufficient to prove that , for every @xmath413 , @xmath414 } |x^{(t)}_s - x^{(t)}_{k\delta}| > \varepsilon \right ] < \varepsilon \text { for all } t \geq t_0.\end{split}\ ] ] by the identity @xmath415 , it is sufficient to treat @xmath416
. then the above condition becomes @xmath417 } |i_s - i_{k\delta t}| >
\varepsilon \sqrt{t } \right ] < \varepsilon.\end{split}\ ] ] given @xmath74 , using lemma [ lem : inter_not_move ] , we can find @xmath407 and @xmath418 such that , if @xmath419 , @xmath420 < \frac{\varepsilon^3}{u^2}\ ] ] now , set @xmath421 . we then have , for @xmath422 and @xmath423 , @xmath424 } |i_s - i_{k\delta t}| >
\varepsilon \sqrt{t } \right]\leq \frac{1}{\delta } \cdot \sup_{r\geq 0}\;\p\left[\sup_{s \in [ r , r+\delta t]}|i_s - i_r| > \varepsilon\sqrt{t } \right]\\[.2 cm ] & \quad= \frac{1}{\delta } \cdot \sup_{r\geq 0}\;\p\left[\sup_{s \in [ r , r+t']}|i_s - i_r| > \varepsilon\sqrt{\frac{t'}{\delta } } \right ] = \frac{u^2}{\varepsilon^2}\cdot \sup_{r\geq 0}\;\p\left[\sup_{s \in [ r , r+t']}|i_s - i_r| > u\sqrt{t ' } \right ] \stackrel{\eqref{eq : just_bill } } { < } \varepsilon\end{aligned}\ ] ] as required in .
our first step towards the proof of lemma [ lem : inter_not_move ] are the following generalizations of lemmas [ lem : no_renewals_ab ] and [ lem_x_goes_to_y_first ] . since the proofs are line - by - line repetitions of the proofs of these earlier results , we omit them .
[ lem_x_goes_to_y ] 1 .
there exists @xmath148 such that the following holds . for any @xmath425 and @xmath426
, we have @xmath427,\\[.2 cm ] \min\big ( t^{\big(\eta^{(x_1,r_1)}_s , s\big)},\ ; t^{\big(\eta^{(x_2,r_2)}_s , s\big)}\big ) < \infty \text { for all } s \in [ a , b ] \end{array}\right ] < e^{-c(b - a)}.\ ] ] 2 .
let @xmath428 and @xmath67 be a stopping time ( with respect to the sigma - algebra of harris systems ) with @xmath429 almost surely .
let @xmath430 for any @xmath431 and events @xmath272 on harris systems , @xmath432 } \in e,\;\eta^{(x_1,r_1)}_\uptau = y_1,\;\eta^{(x_2,r_2)}_\uptau = y_2\text { and } h \circ
\theta(0,\uptau ) \in f \right ] \\[.2 cm ] & = \p\left[\uptau < \infty,\ ; h_{[0,\uptau ] } \in e,\;\eta^{(x_1,r_1)}_\uptau = y_1,\;\eta^{(x_2,r_2)}_\uptau = y_2\right ] \cdot \tilde \p^{\{y_1,y_2\ } } \left[h \in f\right ] .
\end{split } \label{eq : x_goes_to_y}\ ] ] given @xmath433 and @xmath106 , define @xmath434 and the intervals @xmath435;\\ & j^{u , t}_k = k^{u , t } + \left[(4k-3)u\sqrt{t},\ ; ( 4k-2)u\sqrt{t } \right ] , \qquad 1 \leq k \leq 3;\\ & j^{u , t}_4 = k^{u , t } + [ 13u\sqrt{t},\ ; \infty ) = [ 0 , \infty ) .
\end{aligned}\ ] ] we will often omit the superscripts and write @xmath436 , @xmath437 and @xmath438 .
these definitions , as well as the event treated in the following lemma , are illustrated in figure [ fig4brownie ] .
[ cla4onevstun ] for any @xmath433 , there exists @xmath439 and @xmath440 such that , for any @xmath419 , @xmath441 $ ] , @xmath442 , @xmath443 and @xmath444 , we have @xmath445 \right ] \leq \frac{c}{t^{3/2}}.\ ] ] fix @xmath433 .
choose @xmath54 large enough that @xmath446 fix @xmath447,\ ; x_1 \in i_1,\ ; x_2 \in i_2,\ ; x_3 \in i_3 $ ] .
define @xmath448 the probability in the statement of the lemma is less than @xmath449.\ ] ] we will show that @xmath450 \\[.2cm]&\hspace{3.5cm}\leq ce^{-ct^\gamma } + \frac{c}{\sqrt{t}}\ ; \p\left[\sigma_3 \leq t , \eta^{x_k}_s \in i_k \text { for all } k\leq 2 \text { and } s \leq \sigma_3\right]\end{split}\ ] ] for some @xmath451 that only depend on @xmath452 . to this end , we first define the events @xmath453 : |s - s'| < t^{1/8},\;\eta^{(x , r)}_s\neq \triangle,\ ; \eta^{(x , r)}_{s ' } \neq \triangle,\;|\eta^{(x , r)}_s - \eta^{(x , r)}_{s'}| > t^{1/4 } \right\},\\[.3 cm ] & e_2 = \left\{\begin{array}{l}\exists s , s ' \in [ r , t ] : |s - s'| \geq t^{1/8},\;\eta^{(x , r)}_u\neq \triangle,\;\eta^{x_3}_u \neq \triangle\\[.2 cm ] \text { and } \min\left(t^{(\eta^{(x , r)}_u , u)},\;t^{(\eta^{x_3}_u , u)}\right ) < \infty \text { for all } u \in [ s , s']\end{array}\right\}\end{aligned}\ ] ] and the random time @xmath454 by and , we can find @xmath455 such that @xmath456 \leq ce^{-ct^\upgamma}.\label{eq : first_bound_comp}\ ] ] using , we have @xmath457&\hspace{1cm}\subseteq \left\ { \begin{array}{l } \sigma_3 \leq t - t^{1/8};\ ; \eta^{x_k}_s \in i_k \text { for all } s \leq \sigma_3 \text { and } k \leq 2;\\[.2 cm ] \eta^{(x , r)}_s \leq \sup j_3 \text { and } \eta^{x_3}_s \in i_3 \text { for all } s \in [ \sigma_3,\;\sigma_3 + t^{1/8}];\\[.2 cm ] \eta^{(x , r)}_s > \eta^{x_3}_s + u\sqrt{t } \text { for some } s \in [ \sigma_3 + t^{1/8 } , t ] \end{array}\right\}.\end{aligned}\ ] ] we thus also get @xmath458&\subseteq\left\ { \begin{array}{l}\sigma_3 \leq t - t^{1/8},\;\eta^{x_k}_s \in i_k \text { for all } s \leq \sigma_3 \text { and } k \leq 2,\\[.2cm]\uptau \leq \sigma_3 + t^{1/8},\ ; \eta^{(x , r)}_{\uptau } \leq \sup j_3,\;\eta^{x_3}_{\uptau } \in i_3,\ ; \eta^{(x , r)}_s > \eta^{x_3}_s + u\sqrt{t } \text { for some } s \in[\uptau , t]\end{array } \right\ } \\[.3 cm ] & \subseteq \left\ { \begin{array}{l}\sigma_3 \leq t,\;\eta^{x_k}_s \in i_k \text { for all } s \leq \sigma_3 \text { and } k \leq 2,\\[.2cm]\uptau < \infty,\;\eta^{(x , r)}_{\uptau } \leq \sup j_3,\;\eta^{x_3}_{\uptau } \in i_3,\ ; \eta^{(x , r)}_s > \eta^{x_3}_s + u\sqrt{t } \text { for some } s
\in[\uptau,\uptau + t]\end{array } \right\}\\[.3 cm ] & = \bigcup_{z\leq \sup j_3 } \;\bigcup_{w \in i_3 } \left\ { \begin{array}{l}\sigma_3 \leq t,\;\eta^{x_k}_s \in i_k \text { for all } s\leq \sigma_3 \text { and } k \leq 2,\\[.2cm]\uptau < \infty,\;\eta^{(x , r)}_{\uptau } = z,\;\eta^{x_3}_{\uptau } = w,\ ; \eta^{(z,\uptau)}_s > \eta^{(w,\uptau)}_s + u\sqrt{t } \text { for some } s \in [ \uptau , \uptau + t ] \end{array } \right\}. \end{aligned}\ ] ] using this set inclusion and we obtain @xmath459\\[.3 cm ] & \leq \sum_{z \leq \sup j_3}\ ; \sum_{w \in i_3 } \p\left[\sigma_3 \leq t,\;\eta^{x_k}_s \in i_k \text { for all } s \leq \sigma_3 \text { and } k \leq 2,\ ; \uptau < \infty,\;\eta^{(x , t)}_{\uptau } = z,\ ; \eta^{x_3}_{\uptau}= w \right]\\[.1cm]&\hspace{6cm}\cdot \tilde \p^{\{z , w\}}\left[\eta^z_s >
\eta^w_s + u\sqrt{t } \text { for some } s \in [ 0,t]\right ] \\[.3 cm ] & \leq \p\left[\sigma_3 \leq t,\;\eta^{x_k}_s \in i_k \text { for all } s \leq \sigma_3 \text { and } k \leq 2 \right]\\[.1 cm ] & \hspace{6cm}\cdot \sup_{z \leq \sup j_3,\ ; w \in i_3 } \tilde \p^{\{z , w\}}\left[\eta^z_s > \eta^w_s + u\sqrt{t } \text { for some } s \in [ 0,t]\right ] \\[.3 cm ] & \stackrel{\eqref{eq : key_rw_estimate_rs}}{\leq } \frac{c}{\sqrt{t } } \cdot \p\left[\sigma_3 \leq t,\;\eta^{x_k}_s \in i_k \text { for all } s \leq \sigma_3 \text { and } k \leq 2 \right ] . \end{aligned}\ ] ] the desired result now follows from iterating this computation .
[ lem : still_dual ] for any @xmath74 there exists @xmath407 such that , for large enough @xmath54 , @xmath460 \times \{0,1,\ldots , \lfloor t \rfloor\}:\;\eta^{(x , r)}_t > 0 \right]<\frac{\varepsilon}{u^2}.\ ] ] let @xmath461 be a large integer to be chosen later .
define @xmath462 for @xmath463 , let @xmath464 be the set of @xmath461 points ( or @xmath465 points , depending on parity ) that are closest to the middle point of @xmath466 .
define the events @xmath467 & b_2 = \left\{\text{there exist } k \in \{1,2,3\},\ ; y \in \bar i^{u , t}_k,\ ; s\leq t \text { such that } \eta^y_s
\neq \triangle , \;\eta^y_s \notin i^{u , t}_k\right\},\\[.2 cm ] & b_3 = \left\{\begin{array}{c}\text{there exist } ( x , r ) \in [ -u\sqrt{t}-t^{1/4 } , -u\sqrt{t } ] \times \{0,\ldots , \lfloor t \rfloor \},\\[.15 cm ] y_1 \in \bar i^{u , t}_1,\;y_2 \in \bar i^{u , t}_2,\;y_3 \in \bar i^{u , t}_3 \text { such that } \\[.15 cm ] \eta^{(x , r)}_t > 0,\;\eta^{y_i}_s
\in i^{u , t}_i \text { for each } i = 1 , 2 , 3,\;s\leq t.\end{array}\right\}.\end{aligned}\ ] ] we then have @xmath468\times \{0,1,\ldots , \lfloor t\rfloor\ } : \eta^{(x , r)}_t > 0 \right\ } \subseteq \cup_{k=1}^3 b_k.\ ] ] in what follows , @xmath469 and @xmath470 will denote constants that only depend on @xmath134 and @xmath2 , and @xmath471 will denote constants that also depend on @xmath461 . of course , since @xmath472 , constants that depend on @xmath452 also depend on @xmath461 .
equation implies that , for some @xmath473 ,
@xmath474 \leq 3e^{-cn}.\ ] ] to bound the probability of @xmath475 , fix @xmath463 and @xmath476 .
as long as @xmath54 is large enough that @xmath477 , we have @xmath478 and , by , the probability of this event is less than @xmath479 for some @xmath480 .
we thus get @xmath481 \leq 3n\left(ce^{-c\frac{u^2t}{4 t } } + cte^{-c\frac{u\sqrt{t}}{2}}\right ) = 3n\left(ce^{-c\frac{n^4}{4 } } + cte^{-c\frac{n^2\sqrt{t}}{2}}\right).\ ] ] we now turn to @xmath482 .
note that there are at most @xmath483 candidates for @xmath238 and @xmath484 candidates for @xmath485 .
using lemma [ cla4onevstun ] , there exists @xmath486 such that @xmath487 \leq n^3 \cdot t^{5/4 } \cdot \frac{c_n}{t^{3/2 } } \leq \frac{c_n\cdot n^3}{t^{1/4}}.\ ] ] putting these bounds together and rearranging constants , we get @xmath488\times \{0,1,\ldots,\lfloor t \rfloor \ } : \eta^{(x , r)}_t > 0
\right]\\[.2 cm ] & \qquad\qquad\qquad\qquad\leq cn^4e^{-cn } + cn^5 e^{-cn^4 } + ctn^5 e^{-cn^2\sqrt{t } } + \frac{c_nn^7}{t^\frac{1}{4}}.\label{eq : long_bound}\end{aligned}\ ] ] now , given @xmath74 , we first choose @xmath489 such that the sum of the first two terms on is less than @xmath385 .
next , we choose @xmath490 such that , for @xmath489 and any @xmath491 , the sum of the third and fourth terms in is less than @xmath492 .
this completes the proof of the lemma , with @xmath493 .
[ lem : rell_not_move ] for any @xmath74 there exists @xmath407 such that , for large enough @xmath54 , @xmath494 > 1 - \frac{\varepsilon}{u^2}.\ ] ] given @xmath74 , we will find @xmath407 such that @xmath495 < \frac{\varepsilon}{u^2 } \text { if $ t$ is large enough } \label{eq : u_2u}\ ] ] and @xmath496 \xrightarrow{t \to \infty } 0;\ ] ] the statement of the lemma clearly follows from these statements and symmetry . for , we remark that , using the joint construction of the multitype contact process and the ancestor processes , can be rewritten as @xmath497 \times \{0,1,\ldots , \lfloor t \rfloor\}:\xi^h_r(x ) = 2 \right ] < \frac{\varepsilon}{u^2}.\ ] ] letting @xmath498 be the event that appears in the above probability , we also have @xmath499 \\&\leq \sum_{r = 0}^{\lfloor t \rfloor } \p\left[[-u\sqrt{t},\infty ) \times \{r\ } \;\leftrightarrow \ ; ( -\infty ,- u\sqrt{t } - t^{1/4}]\times [ r , r+1]\right ] \leq te^{-ct^{1/4}}\end{aligned}\ ] ] by a comparison with a poisson random variable . is thus proved . for , note that @xmath500 \\[.2cm]&\leq \p \left[\text{there exists } s \in [ 0,t ] \text { such that } \xi^h_s \equiv 0 \text { on } [ -2u\sqrt{t},\;-u\sqrt{t } - t^{1/4}\right].\end{aligned}\ ] ] the probability on the right - hand side
can be bounded above similarly to how we proceeded in lemma [ lem : desc_bar_sides ] , so that follows .
_ lemma [ lem : inter_not_move ] _ we recall the definition of @xmath501 in . given @xmath74 , using the previous lemma we choose @xmath502 so that @xmath503 > 1 - \frac{\varepsilon}{2u^2}.\ ] ] using theorem [ thm : interface_regeneration ] , we choose @xmath436 so that , for any @xmath300 , @xmath504 > 1 - \frac{\varepsilon}{2u^2}.\ ] ] then , if @xmath54 is large enough that @xmath505 we have , for any @xmath300 , @xmath506 & \leq \p\left[\sup_{r \leq s \leq r+t } |i_s - i^r_s| > \frac{u}{2}\sqrt{t } \right ] + \p\left[\sup_{r \leq s \leq r+t } |i^r_s - i_r| > \frac{u}{2}\sqrt{t } \right]\\[.2 cm ] & \leq \p\left[\sup_{r \leq s \leq r+t } |i_s - i^r_s| > k \right ] + \p\left[\sup_{0 \leq s \leq t } |i_s| > \frac{u}{2}\sqrt{t } \right ] < \frac{\varepsilon}{u^2}. \end{aligned}\ ] ]
in this section we will prove theorem [ thm : interface_regeneration ] .
we will often consider multitype contact processes with different initial configurations simultaneously .
when we do so , we always assume that all these processes are constructed on the same probability space , using a single harris system @xmath105 .
we start defining some classes of subsets of the space of configurations @xmath507 .
recall the definition of @xmath508 in .
define @xmath509 & \xi \equiv 1 \text { on } [ a - s , a ] \text { and } \xi \equiv 2 \text { on } [ b , b+s]\end{array}\right\},\ ; s , l > 0.\end{aligned}\ ] ] the homogeneously and fully occupied intervals @xmath510 $ ] , @xmath511 $ ] that appear in the above definition will be referred to as `` isolation segments '' .
the reason is that we think of them as isolating the interface ( which is contained in @xmath512 ) from the `` outside '' @xmath513^c$ ] , so that , if @xmath165 is large , we can hope that the configuration in the outside never has any effect on the evolution of the interface .
our second class of configurations will depend on a preliminary definition .
given @xmath514 , let @xmath515 } + 2 \cdot \mathds{1}_{(\lfloor i(\xi_0)\rfloor,\infty)}.\ ] ] also let @xmath45 and @xmath516 be contact processes started from @xmath195 and @xmath517 , respectively ( constructed with the same harris system ) .
we now let @xmath518 > 1-\varepsilon \right\}.\label{eq : def_omega_eps_k}\ ] ] we will separately prove the following two propositions : * ( large isolation segments allow for regeneration).*[prop : omega_eps_k ] for any @xmath74 there exists @xmath387 such that the following holds . for any @xmath381 there
exists @xmath519 such that @xmath520 .
_ theorem _ [ thm : interface_regeneration ] .
fix @xmath74 .
choose @xmath522 as in proposition [ prop : omega_eps_k ] , then choose @xmath523 as in proposition [ prop : gamma_sl ] , and finally choose @xmath524 as in proposition [ prop : omega_eps_k ] .
now , for any @xmath136 we have @xmath525 \stackrel{}{\geq } \p \left[\xi^h_t \in \gamma_{s , l}\right ] > 1-\varepsilon.\ ] ] now , for any @xmath364 we have @xmath526 \leq \p\left[\xi^h_s \notin \omega_{\varepsilon , k } \right ] + \p\left[\left.\sup_{t \geq s } |i^s_t - i_t| > k\;\right|\ ; \xi^h_s \in \omega_{\varepsilon , k}\right ] < 2\varepsilon.\end{aligned}\ ] ] [ prop : almost_all_couples_all ] for any @xmath74 and @xmath381 there exists @xmath269 such that the following holds .
if @xmath527 is an interval of length at most @xmath528 and @xmath529 are such that @xmath530 for all @xmath531 , then @xmath532 > 1- \varepsilon.\ ] ] since @xmath533 and @xmath534 are constructed from the same harris system @xmath105 , it suffices to find @xmath535 such that @xmath536 > 1- \varepsilon.\ ] ] for a fixed @xmath269 , consider the system of first ancestor processes @xmath537 constructed from the time - reversed harris system @xmath538}$ ] .
since @xmath539 on @xmath540 , we have @xmath541 & \stackrel{\eqref{eq : duality_equation_multi}}{\geq } \p\left[\hat\xi_0(\eta^x_{t_0 } ) = { \settoheight{\dhatheight}{\ensuremath{\hat{\xi } } } \addtolength{\dhatheight}{-0.35ex } \hat{\vphantom{\rule{1pt}{\dhatheight } } \smash{\hat{\xi}}}}_0(\eta^x_{t_0 } ) \neq 0 \text { for all } x \in \z \text { with } \eta^x_{t_0 } \neq \triangle\right ] \\&\geq \p\left[\eta^x_{t_0 } \notin
i \text { for all } x \in \z\right ] \geq 1-(\#i)\cdot \p\left[0 \in \{\eta^x_{t_0}:x\in\z\}\right].\end{aligned}\ ] ] the result now follows from taking @xmath535 large enough , depending on @xmath186 and @xmath528 , by . [ lem : couple_left ] for any @xmath74 there exists @xmath152 such that the following holds for any @xmath153 .
assume @xmath195 satisfies : @xmath542$};\\ \label{eq : no_2s}&r(\xi_0 ) < 0.\end{aligned}\ ] ] let @xmath543 be the process started from @xmath544 then , with probability larger than @xmath159 we have @xmath545 , \\
\label{eq : xi_coinc_gamma}&\ell(\xi_t ) = \ell(\xi'_t),\ ; r(\xi_t ) = r(\xi'_t)\text { and } \\ & \label{eq : xi_where_2s}\ell(\xi_t),\;r(\xi_t ) < s/2 + \beta t.\end{aligned}\ ] ] given @xmath387 , we write @xmath546 fix @xmath74 . by lemmas
[ lem : couple_ones ] , [ lem : desc_bar_sides ] and [ lem : no_faster ] , if @xmath165 is large enough , then with probability larger than @xmath159 all the following three events occur : @xmath547 = \{\xi'_t = 0\ } \cap ( -\infty , f^{(2)}_t ] \text { for all } t \geq 0\right\},\\[.2 cm ] & e_2 = \left\{\text{there exists } x \in ( f^{(1)}_t,\;f^{(2)}_t):\;\xi'_t(x ) \neq 0 \text { for all } t \geq 0 \right\},\\[.2 cm ] & e_3 = \left\{r(\xi'_t ) <
f^{(1)}_t \;\text{for all } t \geq 0 \right\}.\end{aligned}\ ] ] we will also assume that @xmath548 .
we will now state and prove two auxiliary claims .
+ _ claim 1 . _ on @xmath549 , @xmath550 for all @xmath54 .
+ to see that this holds , first note that @xmath551 so applying we get @xmath552 we now fix @xmath115 with @xmath553 and will show that @xmath554 using , it follows from @xmath553 that there exists an infection path @xmath555\to\z$ ] so that @xmath556 , @xmath557 , the definition of @xmath87 and imply that @xmath558 additionally , the definition of @xmath559 and together give @xmath560 we must have @xmath561,\ ] ] otherwise we would obtain a contradiction as follows .
let @xmath562 be the smallest time for which @xmath563 .
since @xmath564 and each jump of @xmath565 has size at most @xmath2 , we would have @xmath566 .
again using , and , we would get @xmath567 , so @xmath568 , contradicting the assumption that @xmath569 occurs . now , , and another application of give , so the proof of claim 1 is complete .
+ _ claim 2 . _ on @xmath570 , @xmath571 for all @xmath54 .
+ indeed , by the definition of @xmath572 , for any @xmath54 there exists @xmath573 $ ] such that @xmath574 , and by the definition of @xmath569 we have @xmath575 , so that @xmath576 , thus @xmath577 , thus @xmath578 .
+ we are now ready to conclude . from claim 1 and the definition of @xmath559 , we have that @xmath579 from claim 1 and the definition of @xmath569 , @xmath580 from claim 1 and claim 2 , @xmath581 [ cor : all_tog ] for any @xmath74 there exists @xmath152 such that the following holds .
assume @xmath153 and @xmath195 satisfies , for some @xmath582 with @xmath583 : @xmath584;\\
\label{eq_cor_as2}&a < r(\xi_0),\;\ell(\xi_0 ) < b;\\ \label{eq_cor_as3}&\xi_0(x ) \equiv 2 \text { on } [ b , b+s].\end{aligned}\ ] ] let @xmath585 be the process started from @xmath586}(x ) + \mathds{1}_{(a , b)}(x ) \cdot \xi_0(x ) + 2\cdot\mathds{1}_{[b,\infty)}(x).\ ] ] then , with probability larger than @xmath159 , @xmath587 & a -\frac{s}{2 } - \beta t <
r(\xi_t),\ell(\xi_t ) < b + \frac{s}{2 } + \beta t.\end{aligned}\ ] ] we will also need @xmath189 , the process started from @xmath588 given @xmath74 , by lemma [ lem : couple_left ] , @xmath177 can be chosen so that , if and hold , then @xmath589&\text{and } \ell(\xi_t ) , r(\xi_t ) < b + \frac{s}{2 } + \beta t \end{array}\right]>1-\varepsilon/2.\end{aligned}\ ] ] now , note that and the definition of @xmath590 imply @xmath591,\\ & r(\xi'_0 ) > a,\end{aligned}\ ] ] so that we can again use lemma [ lem : couple_left ] ( and symmetry ) to obtain that @xmath592 & \text{and } \ell(\xi'_t ) , r(\xi'_t ) > a - \frac{s}{2 } - \beta t\end{array}\right ] > 1-\varepsilon/2.\label{eq : eff_coup_inf}\end{aligned}\ ] ] putting and together , we obtain the desired result .
_ proposition _ [ prop : omega_eps_k ] . given @xmath74
, we choose @xmath165 large enough corresponding to @xmath593 in corollary [ cor : all_tog ] .
increasing @xmath165 if necessary , by lemma [ lem : no_faster ] , we can also assume the following ( recall that @xmath594 and @xmath595 , where @xmath596 is the process started from the heaviside configuration ) .
@xmath597 \text { for all } t \geq 0\right ] > 1 - \varepsilon.\ ] ] then , given @xmath381 , we choose @xmath535 corresponding to @xmath593 and @xmath528 in lemma [ prop : almost_all_couples_all ]
. now assume @xmath598 .
then , there exist @xmath155 as prescribed in ; note in particular that @xmath599 , so that @xmath600 .
let @xmath601}(x ) + \mathds{1}_{(a , b)}(x ) \cdot \xi_0(x ) + 2 \cdot \mathds{1}_{[b,\infty)}(x),\\[.2 cm ] & \tilde \xi_0(x ) = \mathds{1}_{(-\infty , \lfloor i(\xi_0 ) \rfloor ] } + 2\cdot \mathds{1}_{(\lfloor i(\xi_0)\rfloor,\infty)}\end{aligned}\ ] ] and @xmath533 , @xmath516 be the processes started from these configurations . by our choice of @xmath165 and @xmath535 , with probability larger
than @xmath159 the following three events occur : @xmath602;\\[.2 cm ] & \text{for all } t \ge 0,\;i(\tilde \xi_t ) \in [ \lfloor i(\xi_0 ) \rfloor - s - \bar\beta t,\;\lfloor i(\xi_0 ) \rfloor + s + \bar \beta t ] \subset [ a - s - \bar \beta t,\ ; b+s + \bar \beta t];\\[.2 cm ] & \text{for all } t
\ge t_0,\;i(\hat \xi_t ) = i(\tilde \xi_t).\end{aligned}\ ] ] if these events all occur , we have @xmath603 & |i(\xi_t ) - i(\tilde \xi_t)|=0 \text { if } t > t_0.\end{aligned}\ ] ] the desired result now holds for @xmath604 . if the statement is false , then one can find @xmath181 , @xmath609 and a sequence of times @xmath611 such that @xmath612 > \delta \text { for all } n \in \n . \nonumber\ ] ] by tightness of the size of the interface ( as given by ) , we can then find @xmath613 such that @xmath614 > \delta/2\text { for all } n \in \n .
\label{eq : negation_more}\ ] ] let us denote by @xmath615 the event inside the above probability . note that @xmath616 \subset [ m_{t_n},x^{(k)}_{t_n}]\}.\ ] ] for @xmath617 , define the event @xmath618 , \text { all vertices in } \\[0.2 cm ] [ m_{t_n } - r,\;m_{t_n } ) \cup \{x^{(0)}_{t_n},\ldots , x^{(k)}_{t_n}\ } \cup ( x^{(k)}_{t_n},\;x^{(k)}_{t_n } + r ] \\[.2 cm ] \text{have a death mark and do not originate any arrow . }
\end{array } \right\}\ ] ] since the set of vertices that appears in the definition of @xmath619 contains @xmath620 vertices , we have @xmath621 = e^{(-1 - 2r\lambda)(2r + k ) } , $ ] so that , by , @xmath622 \geq \frac{\delta}{2}\cdot e^{(-1 - 2r\lambda)(2r + k)}.\ ] ] additionally , by , @xmath623 now , and together imply @xmath624 > \frac{\delta}{2}\cdot e^{(-1 - 2r\lambda)(2r + k)}.\ ] ] this contradicts tightness of the interface size , .
we will need one extra subset of @xmath507 , defined for @xmath609 and @xmath625 by @xmath626 \#\{x \in ( m(\xi),\ ; m(\xi ) +
l ] : \xi(x ) = 2\ } \geq k , \\[.2cm]m(\xi ) - m(\xi ) \leq l\end{array}\right\}.\ ] ] fix @xmath74 and @xmath609 . by
, we can choose @xmath628 so that @xmath629 < \varepsilon/2 \text { for all } t \geq 0.\ ] ] let @xmath630
. we now choose @xmath631 corresponding to @xmath632 and @xmath633 in lemma [ lem : order_many ] ; we get @xmath634 : \xi_t(x ) = 2\ } < k\right ] \\[.2cm]&\leq \p\left[\#\{x \in ( m_t , m_t + l ] : \xi_t(x ) \neq 0 \ } < k ' \right ] \\[.2cm]&\leq \p\left[x^{(k')}_t >
m_t + l\right ] \leq \p\left[x^{(k')}_t > l\right ] < \varepsilon/4 . \end{split } \label{eq : llprime}\ ] ] by symmetry we also get @xmath635 < \varepsilon/4.\ ] ] the desired statement now follows from putting together ( with the observation that @xmath636 ) , and .
given @xmath387 and @xmath258 , define the event @xmath640 d^{z , y}_{[0,1 ] } = \varnothing \text { for all } y , z \text { with } |x - y|\leq s,\ ; |x - z| > s \end{array } \right\}.\ ] ] by prescribing the position of a finite number of arrows and the absence of recovery marks and arrows at certain positions , it is easy to show that there exists @xmath641 such that , for any @xmath29 , @xmath642 > \delta_s.\ ] ] we note that @xmath643,\ ; \ell(\xi_{1 } ) > x+s\};\\[.2 cm ] \label{eq : f_good_prop2}&f_s(x ) \cap \{\xi_0(x ) = 1,\;r(\xi_0 ) < x+s\ } \subset \{\xi_{1 } \equiv 2 \text { on } [ x - s , x],\ ; r(\xi_{1 } ) < x+s\}\end{aligned}\ ] ] and that @xmath644 now , given @xmath74 and @xmath387 , we choose @xmath609 so that @xmath645 ( note in particular that @xmath646 ) .
assume that @xmath637 and @xmath647 . by the definition of @xmath648 in
, we can find @xmath649 \text { with } x_{i+1 } >
x_i + 2(s+r ) \text { for all } i;\\[.2 cm ] y_1,\ldots , y_{k ' } \in [ m(\xi_0 ) + 2(s+r ) , m(\xi_0 ) +
l ] \text { with } y_{i+1 } >
y_i + 2(s+r ) \text { for all } i\label{eq : we_can_find_y}\end{aligned}\ ] ] we then have @xmath650 & \stackrel{\eqref{eq : def_gamma},\eqref{eq : f_good_prop1},\eqref{eq : f_good_prop2}}{\geq } \p\left[\left(\cup_{i=1}^k f_s(x_i)\right ) \cap \left(\cup_{i=1}^k f_s(y_i)\right)\right ] \\[.2cm]&\stackrel{\eqref{eq : fcan_happen},\eqref{eq : f_is_indep},\eqref{eq : we_can_find_x},\eqref{eq : we_can_find_y}}{\geq } 1 - 2(1-\delta_s)^k > 1 -\varepsilon.\end{aligned}\ ] ] _ _ p__roposition [ prop : gamma_sl ] . fix @xmath186 and @xmath165 .
we first choose @xmath312 corresponding to @xmath385 and @xmath165 in lemma [ lem : after_one_second ] , and then choose @xmath528 corresponding to @xmath385 and @xmath312 in lemma [ lem : gamma_pi_good ] . we then have , for any @xmath184 , @xmath651 \geq \p\left[\xi^h_t
\in \gamma_{s , l}\;|\;\xi^h_{t-1 } \in \pi_{k , l}\right ] \cdot \p\left[\xi^h_{t-1 } \in \pi_{k , l } \right]\geq ( 1-\varepsilon/2)^2 > 1-\varepsilon.\ ] ] it is then easy to show that we can increase @xmath528 if necessary so that the result also holds for @xmath652 . | we study the interface of the multitype contact process on @xmath0 . in this process , each site of @xmath0 is either empty or occupied by an individual of one of two species . each individual dies with rate 1 and attempts to give birth with rate @xmath1 ; the position for the possible new individual is chosen uniformly at random within distance @xmath2 of the parent , and the birth is suppressed if this position is already occupied .
we consider the process started from the configuration in which all sites to the left of the origin are occupied by one of the species and all sites to the right of the origin by the other species , and study the evolution of the region of interface between the two species .
we prove that , under diffusive scaling , the position of the interface converges to brownian motion . * * |
the circuit @xmath28 and the closed path @xmath48 used to evaluate the circulation of @xmath46 . ] | we discuss several similarities and differences between the concepts of electric and magnetic dipoles .
we then consider the relation between the magnetic dipole and a current loop and show that in the limit of a pointlike circuit , their magnetic fields coincide .
the presentation is accessible to undergraduate students with a knowledge of the basic ideas of classical electromagnetism .
the concept of a magnetic dipole describes the long distance limit of the field produced by a steady current flowing in a small loop of wire.@xcite the word `` dipole '' is borrowed from electrostatics but when used in magnetostatics , this terminology is somewhat deceptive because a magnetic dipole is physically very different from its electric counterpart .
the aim of this paper is to discuss the similarities and differences of these concepts .
recall the definition of an electric dipole .
we start with a configuration in which two charges @xmath0 and @xmath1 ( @xmath2 ) are located at @xmath3 and @xmath4 respectively .
the electric dipole is obtained by taking the limit @xmath5 keeping fixed the quantity @xmath6 which is called the _ electric dipole moment_. the dipole electric field @xmath7 can be obtained from the potential @xcite @xmath8 so that @xmath9 it might be tempting to define a magnetic dipole with moment @xmath10 in a similar way : that is , the object which generates the magnetic field @xmath11 however , eq .
( [ 4 ] ) is inconsistent with the nonexistence of magnetic monopoles , as described by the maxwell equation @xmath12 because @xmath13 in eq .
we used the result@xcite@xmath14 the failure to satisfy eq .
( [ due ] ) is not surprising because @xmath15 in eq .
was constructed as the limit of zero separation between monopole and anti - monopole , which in the magnetic case do not exist . a modification of eq . ( [ 4 ] ) at the origin
@xcite @xmath16 fixes the problem and gives a divergenceless field .
however , the field given by eq .
( [ 7 ] ) is no longer conservative ( irrotational ) , in contrast to its electric counterpart , eq .
( [ 3 ] ) .
the difference between electric and magnetic fields is that , in a stationary situation , the electric field is conservative as a consequence of the faraday equation @xmath17 whereas the magnetic field , which is divergenceless , can not also be irrotational ( unless it is identically zero ) . in a world without monopoles , a magnetic dipole must be defined in terms of current distributions only .
the magnetic effects of a steady current density @xmath18 are described by ampere s equation @xmath19 from eq .
we can calculate the magnetic field @xmath20 provided the condition , @xmath21 which is equivalent to conservation of charge in the steady case , is satisfied .
equation ( [ uno ] ) shows that the non - conservative part of the magnetic field is located at the points at which the current density is nonzero .
therefore in the magnetic dipole case , eq . ( [ 7 ] ) , the only contribution needed to satisfy ampere s equation is the term proportional to @xmath22 because @xmath23= -{\boldsymbol{\mu}}\times { \boldsymbol{\nabla}}\delta^{(3)}({\mathbf{x}}).\ ] ] we shall now show that @xmath15 given by eq .
( [ 7 ] ) is the magnetic field generated by a current loop of infinitesimal size .
we start from the solution of eqs .
( [ due ] ) and ( [ uno ] ) which can be found in textbooks on electromagnetism : @xcite @xmath24 where @xmath25 is the distance between the generic point @xmath26 of the integration region and the observation point @xmath27 . for a coil @xmath28 made of a thin wire , eq .
( [ dieci ] ) becomes@xcite @xmath29 where the line integral , with length element @xmath30 , runs over the wire whose tangent unit vector is denoted by @xmath31 . the circuit @xmath28 in eq .
( [ dieci2 ] ) must be closed because of eq .
( [ uno2 ] ) , and @xmath32 is the ( constant ) current in the circuit .
we assume that the current loop is a plane circuit enclosing an area @xmath33 .
we denote by @xmath34 the unit vector orthogonal to the plane , oriented according to the right - hand rule with respect to @xmath31 .
we also denote by @xmath35 the external normal to the wire ( see fig . [ massimo_testafig01.pdf ] ) .
these unit vectors are related by @xmath36 if we substitute eq .
( [ ventiquattro ] ) into eq .
( [ dieci2 ] ) , we obtain @xmath37 we use the identity @xmath38 and write eq .
( [ venticinque ] ) as @xmath39.\ ] ] if we use green s formula in two dimensions @xmath40 where @xmath41 is the surface element of @xmath33 , and the relation @xmath42 we obtain @xmath43 + { \hat{\mathbf{n}}}\frac{i}{c } \int_s \delta^3({\mathbf{x}}-{\boldsymbol{\xi}})\,d\sigma \\ & \equiv { \mathbf{b}}^{(1)}_\gamma ( { \mathbf{x}})+{\mathbf{b}}^{(2)}_\gamma ( { \mathbf{x}}).\end{aligned}\ ] ] the gradient @xmath44 is irrotational and is nonzero in all of space , in contrast to @xmath45 which is non - zero only inside the plane region @xmath33 delimited by the coil @xmath28 .
it is instructive to show how @xmath46 satisfies ampere s law in its integral form , that is , @xmath47 where @xmath48 is any closed path linked with @xmath28 as shown in fig [ massimo_testafig01.pdf ] .
because @xmath44 is a pure gradient , we have @xmath49 the integral on the right - hand side of eq .
( [ trenta3 ] ) , by virtue of the delta function , has a contribution only from the point of intersection @xmath50 between @xmath48 and @xmath33 , which leads to eq .
( [ trenta21 ] ) .
equation ( [ trenta3 ] ) is surprising because it shows that ampere s law is satisfied only by @xmath45 , which is the part of the magnetic field localized inside @xmath28 . to make contact with the dipole field @xmath15 given by eq .
( [ 7 ] ) , we take the limit as the coil area goes to zero , keeping the product @xmath51 constant .
we have @xmath52 \\ & = \frac{\mu}{4\pi } { \boldsymbol{\nabla}}_x \left[\frac{1}{s } \int_s ( { \hat{\mathbf{n}}}\cdot { \boldsymbol{\nabla}}_x\frac{1}{r})\,d\sigma \right ] \\ & = \frac{\mu}{4\pi } { \boldsymbol{\nabla}}_x \overline{\left [ ( { \hat{\mathbf{n}}}\cdot { \boldsymbol{\nabla}}_x\frac{1}{r } ) \right ] } , \label{last}\end{aligned}\ ] ] where the bar in eq . denotes the mean value in @xmath33 . by the mean value theorem we know that @xmath53}= \left [ ( { \hat{\mathbf{n}}}\cdot { \boldsymbol{\nabla}}_x\frac{1}{r } ) \right]_{\overline{p}}\ ] ] where @xmath54 is a suitable point inside @xmath33 . in the limit of pointlike @xmath33
, we have @xmath55\ ] ] where @xmath56 may be identified with the magnetic moment of the small loop and @xmath57 is the distance between the observation point and the position @xmath58 of the ( pointlike ) circuit . as for @xmath59 , which contains a delta function ,
the pointlike limit must be discussed using generalized functions.@xcite we introduce a test function @xmath60 , which is an infinitely differentiable function vanishing at infinity faster than any inverse power of @xmath61,@xcite , and study the @xmath62 limit of expressions such as @xmath63 . from eq .
( [ trenta ] ) we have @xmath64 when @xmath33 shrinks to the point @xmath58 .
this implies @xmath65 if we compare with eq .
( [ 7 ] ) , we find that an infinitesimal current loop generates a magnetic field identical to the one given by a magnetic dipole of moment @xmath66 .
we thank the reviewers for valuable suggestions on how to improve the presentation of our paper . 5 r. p. feynman , r. b. leighton , and m. sands , _ the feynman lectures on physics _
( addison - wesley , reading , ma , 1999 ) , vol .
2 .
j. d. jackson , _ classical electrodynamics _
( john wiley @xmath67 sons , new york , 1998 ) , 3rd ed .
s. m. blinder , `` delta functions in spherical coordinates and how to avoid losing them : fields of point charges and dipoles , '' am .
j. phys . * 71 * , 816818 ( 2003 ) .
b. d. h. tellegen , `` magnetic - dipole models , '' am .
j. phys .
* 30 * , 650652 ( 1962 ) .
n. d. rao , `` a note on the vector potential of a magnetic dipole , '' am . j. phys .
* 39 * , 12761277 ( 1971 ) .
we use rationalized cgs units .
i. m. gelfand and g. e. shilov , _ generalized functions _ ( academic press , new york , 1964 ) , vol . 1 . j. i. richards and h. k. youn _ the theory of distributions a nontechnical introduction _
( cambridge university press , 1995 ) h. b. g. casimir , _ on the interaction between the atomic nuclei and electrons _ ( w. h. freeman ,
san francisco , 1963 ) .
d. j. griffiths , `` hyperfine splitting in the ground state of hydrogen , '' am .
j. phys .
* 50 * , 698703 ( 1982 ) .
more precisely , we deal with tempered distributions . |
gravitational lensing occurs when light is deflected by gravity .
the first instance of gravitational lensing in a cosmological context was observed in 1979 , where the quasar q0957 + 561 is lensed by by a foreground galaxy , producing two magnified , distorted , images @xcite . since then
many other multiply - imaged lensed quasars have been found ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the properties of the lensed images , for example their number , separation on the sky , relative brightness , and others , are determined mostly by treating the galaxy as a smooth gravitational lens .
galaxies are not smooth , however , being comprised of many compact objects in addition to smoothly distributed matter .
the objects inside the galaxy can exert an influence on the quasar images because the quasar and galaxy are in transverse motion relative to our line of sight .
as the quasar moves over time , this motion shifts the light paths relative to the galaxy components , making the quasar images vary in magnitude over time scales of weeks to years @xcite .
this phenomenon , called gravitational microlensing , was first observed in 1985 @xcite and is believed to occur in most lensed quasars @xcite .
it can be used as a probe of the source quasar because the quasar s size and shape affects the microlensing variability ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
for example , when the source is large , the changes in the light curve are not as prominent , since the larger source smooths out the variability .
microlensing can also be used to probe the mass distribution in the lensing galaxy .
initial studies of quasar microlensing indicated that microlensing was not significantly dependent , within certain constraints , on the mass spectrum within the lens ( e.g. * ? ? ?
* ; * ? ? ?
* ) , in particular how much mass is in compact objects and how much is in smooth matter .
however , during investigations of anomalous flux ratios in the lensed quasar mg 0414 + 0534 @xcite , evidence against this view was found , and numerical modelling presented by @xcite showed that it must be false . beginning with a large number of solar - mass ( 1m@xmath1 ) stars for the lens galaxy , they replaced a substantial fraction of these with smooth matter , keeping the total mass the same , and found that the microlensing variability was enhanced .
this also introduced the `` bi - modal '' mass distribution into computational microlensing , where two very distinct mass components in the lens are modelled ( e.g. * ? ? ? * ) .
@xcite replaced the smooth matter with many compact objects , all of the same mass , and significantly less than 1m@xmath1 .
the objects produced small - scale microlensing variability that was not present when smooth matter was used , as expected .
however , it was found that the variability produced by the small objects could be smoothed out enough so that they could not be distinguished from smooth matter , if the source size was over a threshold .
it was also found that , if source size was taken into account , the amplitude of these variations was an indicator of the mass of the objects in the lens .
recently , @xcite conducted similar modelling , expanding the bi - modal distribution by using a @xcite mass distribution for the stars and a @xcite power - law distribution for the compact objects , considering these as potential dark matter candidates .
they conclude that their small compact objects will add detectable small microlensing events ( `` nanolensing '' events ) to a microlensing light curve of no more than about 0.1 mag over a time scale smaller than a year .
@xcite were less interested in source size and how this diminishes the detectability of nanolensing events , but did confirm that larger sources would smooth out the light curves and make nanolensing harder to detect .
the term `` nanolensing '' , used previously by @xcite and @xcite , is gaining in usage and refers to light deflections on much smaller scales than microlensing , due to planetary - size objects or possible dark matter objects @xcite .
the work of @xcite involved the detection of cosmological planetary masses by the nanolensing of gamma ray bursts @xcite , and nanolensing is also referred to in exoplanet searches @xcite .
we use it here to refer to lensing variability produced by objects in a lens with masses far below that of 1m@xmath1 stars , e.g. the low - mass objects in a bi - modal mass distribution .
this paper expands on the work of @xcite , @xcite and @xcite in investigating nanolensing events due to small masses , and their interaction with source size . using bi - modal mass distributions , we use several mass values down to @xmath3m@xmath1 for the size of the small objects , while increasing their number to over a billion far more than has been modelled in the past .
we use a mock lensed quasar system that has been modelled by @xcite and @xcite , similar to the lensed quasar mg 0414 + 0534 .
mg 0414 + 0534 has not been used in this and past works due to the high magnification of the source quasar , produced by a very large number of objects in the lens galaxy , which are difficult to deal with in numerical models .
simple statistics allow us to use the amplitude of nanolensing events to infer small objects in a lens , based on a range of source sizes .
we then follow this with an investigation into the duration of the nanolensing events and the source sizes needed to resolve them something has not not been done in past studies .
we show that there are two time and source scales involved in these events and indicate how they may be observed .
we discuss how our investigations provide direction to further work that can be conducted in this area .
the structure of the paper is as follows : in section [ method ] we introduce numerical modelling of microlensing , and the lensing model and parameters used for this study , in section [ results ] we present the results of nanolensing of bi - modal mass distributions with different source sizes , including event amplitudes and durations . in section
we [ discussion ] discusses the results and section [ conclusions ] contains our conclusions . throughout this paper , a cosmology with @xmath5 , @xmath6 and @xmath7 is assumed .
the properties of an image in a multiply - imaged quasar are determined mostly by the mass distribution of the lensing galaxy , and the relative locations and distances of the galaxy and quasar @xcite . for image modelling ,
two parameters are used to specify the mass in the lensing galaxy at the image positions : the convergence ( @xmath8 ) , and shear ( @xmath9 ) .
the convergence specifies the effect of mass close to the light path , and the shear is the effect of the overall mass of the galaxy . within the convergence
, a mass spectrum can be chosen for models , we will be using bi - modal distributions as described above , where a small amount of mass is in 1m@xmath1 stars , and the rest in either small objects , or smooth matter .
the distances to the lens and source are subsumed into a distance unit used within lensing models : the einstein radius ( @xmath10 ) . if a point source is perfectly in line with a point lens object , usually chosen to be 1m@xmath1 , the source will appear as a ring around the lens .
projected onto the source plane , the ring radius is given by @xmath11 where @xmath12 is the mass of the lens and @xmath13 refers to the angular diameter distance between @xmath14 and @xmath15 ; the subscripts @xmath16 and @xmath17 representing source , lens , and observer respectively .
microlensing and nanolensing requires the source to change location behind the lens @xcite to produce variation in the source magnification , so a region of the source plane , where the source may lie , is defined . each point in the region
has a magnification value , indicating how a point source will be ( de)magnified at that location .
this is a `` magnification map '' @xcite , examples of which are seen in figure [ maps strip ] .
bright areas in a map indicate locations where the source will be magnified and darker regions are where it will be demagnified , relative to the average for the image being modelled .
the regions of light and dark are delineated by lines called `` caustics '' , where the magnification is formally infinite @xcite .
there will be different maps for each image in a lensed quasar , since the mass distribution producing each image is different .
a magnification map is divided into pixels , which enforces a lower bound on the size of sources that can be studied .
different source sizes and shapes are studied by creating a pixelized source profile and convolving this with the map , producing a map for the microlensing of that source .
light curves can be obtained by taking the magnification along a line across the map , the line corresponding to the path along which a source may travel . for the lensing galaxy , compact masses and smooth matter
are laid down on a lens plane and rays are fired through them to hit the map .
the lens masses are projected onto on a plane because , in a cosmological situation where the distances are great , the lensing galaxy can be approximated as flat @xcite .
rays are fired in the inverse direction , i.e. from observer through the lens to source plane , for reasons of computational ease and efficiency . because microlensing is produced by a large number of objects in the lens galaxy ,
it is difficult to study analytically , and numerical techniques are used ; we use the inverse ray - shooting method of @xcite and @xcite .
numerical approaches have allowed the study of many aspects of quasar microlensing , such as : the size of accretion disks @xcite ; the structure of quasar broad line emission regions @xcite ; the structure of microlensed water masers @xcite ; chromatic effects in microlensing @xcite ; the nature of dark matter in lensing galaxies @xcite ; and the effect of source size on microlensing @xcite , among others .
.parameters for mock lens images [ cols="^,^,^",options="header " , ] parameters for the mock lensed quasar images used in this study .
the convergence ( @xmath8 ) specifies the mass close to a light ray , the shear ( @xmath9 ) specifies the effect of all the mass in the lensing galaxy .
they are the same as the m10 and s10 parameters used in @xcite .
we use a mock lensed quasar system that has been used in other work @xcite . the convergence and shear parameters for each image
are listed in table [ parameters ] ; these values are based on models of the quadruply - imaged quasar mg 0414 + 0534 .
the images in the mock lens system are designated m and s ; image m is positive - parity , s is negative parity , the latter meaning the image is mirror - symmetric to the source .
however , for the purposes of this study , the internal structure of the images is not relevant , only the magnification of the image relative to the source .
images m and s produce different - looking magnification maps and slightly different behaviour in our models , as will be seen later .
recent studies suggest the dark matter fraction in microlensed quasars at the image positions is high .
@xcite preferred 50@xmath18% in mg 0414 + 0534 .
@xcite determined 90% for pg 1115 + 080 , and now @xcite suggests @xmath19% for mg 0414 + 0534 and h1413 + 117 . for our bi - modal mass distribution , we set two per cent of the mass in the lens in 1m@xmath1 objects , the other 98% being in either smooth matter or small compact objects .
the small objects are all of the same mass for each map .
the mass of the small objects can be reduced , which means their number increases to maintain the same total mass in the lens .
the 1m@xmath1 objects stay at the same locations as the other masses change .
following @xcite , we use z = 0.3 for the lens and z = 2 for the source . the einstein radius ( er ) for a 1m@xmath1 object using these distances is 0.0236 pc .
the magnification maps cover a region in the source plane of 20 @xmath20 20 er@xmath21 ( 0.47 @xmath20 0.47 pc@xmath21 ) , at a resolution of @xmath22 pixel@xmath21 , or width 0.002 er ( 4.7@xmath23 pc ) per pixel .
this also means the smallest source size that can feasibly be studied with such maps is 0.002 er . the maps used in this study are shown in figure [ maps strip ] .
the first column has maps for image m , the third column for image s. note how the large scale caustic structure is consistent within each column , but looks different between columns .
this is because of the opposite parity of the quasar images @xcite .
the first map for image m was generated from a lens of 609 1m@xmath1 objects and 14,758,999 objects of @xmath24m@xmath1 , and the second last map for image m represents 1,195,478,868 objects of @xmath3m@xmath1 . the number of compact objects used to generate each map is listed next to the map in column two .
the bottom map was generated from a lens with 609 1m@xmath1 objects and the rest of the mass in smooth matter , being the same mass as all the small objects combined .
the third column is similar to column one , but for image s , reaching a limit of 1,318,703,785 small objects , with the number of objects listed in column four . to ensure the maps are of high quality with good magnification resolution
, we shoot several trillion light rays through the lens , over a million per pixel .
the computational work required to generate and analyse the maps presents considerable challenges , and it is only in recent years that they have been overcome . with the advent of supercomputers it is now possible to use large numbers of objects in the lensing galaxy , and in this work over 1 billion are used .
we also fire rays in parallel , using multiple parallel processes . using the method of @xcite , the map containing the largest number of masses can be generated in @xmath4 14 days on a supercomputer using 16 parallel processes . for the maps with the least number of lens masses ,
the total time is @xmath4 24 hours . about 18% of the compute time
is used to generate the lens objects , and the rest is for firing the rays , with many rays fired in parallel .
the need to place the lens information in large data files , of about 150 gb for 1 billion objects , means ideal speedups are not achievable for either phase of the program , but the situation is improved with the use of 16 gb of computer memory for caching the files ; such memory is only now usable because 64-bit ram addressing has become available .
sources in our simulations have a simple 2-d gaussian profile , with a radius of 3 @xmath25 .
their radii are based on estimated sizes of quasar structure .
for example , accretion disks are of order @xmath26 pc ( * ? ? ?
* estimated in the i - band from h@xmath27 ) , and broad line regions of order @xmath28 pc ( * ? ? ?
* ; * ? ? ?
* using the mgii / ciii@xmath29}$ ] and civ / ciii@xmath29}$ ] lines respectively ) . for our experiments we use
source radii ranging from 0.002 to 1.4 er , or @xmath30 to 0.033 pc .
beginning at a size of 0.002 er , the next size is 0.01 er , then 0.05 er , then increasing by 0.05 er to 0.4 er , then in steps of 0.1 er to 1.0 er , then in steps of 0.2 er to 1.4 er , thus providing more data points at the smaller sizes .
this is necessary as will be seen from the figures presented later on .
we always compare a map generated using stars@xmath31small objects in the lens , to a map using stars@xmath31smooth matter in the lens , and only maps generated for the same image ( m or s ) .
this means that each map in the first column of figure [ maps strip ] will be compared with the bottom one in the column , and the same for the third column .
it can be seen that the maps using small objects become more like the corresponding smooth matter map as the size of the small objects gets smaller and their number increases ; this is expected .
however , it is also necessary to consider source size ; if these maps are convolved with larger sources then they may become similar to the smooth matter case when the small object mass is _ not _ so small .
a simple measure is used to indicate that one map is similar to another .
the maps are converted from magnifications to magnitudes . to avoid edge effects ,
a margin around the map of width equal to the source radius of the largest source ( 1.4 er ) is ignored , and within the margin the maps are normalized to have an average magnitude of 0 . if such a map generated from smooth matter is subtracted from a map generated from compact matter the result is an approximately gaussian distribution of residual magnitude .
the root - mean - squared ( rms ) of the residual values is used as the measure of similarity termed the `` difference measure '' .
it is the average difference between the two maps in units of magnitude .
firstly we demonstrate how two maps , one generated from compact objects and the other from smooth matter , become similar as the source size increases .
figure [ lightcurves ] shows a comparison of such maps ; all were generated from a star field of 609 1m@xmath1 masses at the same fixed locations , plus either smooth matter or small objects of @xmath32 m@xmath1 .
the first column shows maps for a lens containing these objects , the second column shows maps for a lens with smooth matter .
the 1m@xmath1 masses are responsible for the identical large scale caustics in the maps in each column .
the map in the first column shows that between the large caustics there is small - scale structure , produced by the small objects . the smooth matter map ( second column )
does not exhibit this .
the third and fourth columns display a light curve extracted from a horizontal cut across the middle of the maps in the first and second columns respectively .
the light curves make plain the small scale structure in the compact matter map , producing a high degree of variability in the top row that almost masks the peaks in the light curve from the large scale caustics , clearly visible in the smooth matter light curve .
going down the row the source size increases , indicated on the left side of the first map .
the fifth column is the difference measure between the maps , described previously . as the source size increases , the compact matter and smooth matter maps and their light curves become similar , and the difference measure decreases .
both of these things indicate that the increased source size is reducing the small - scale variation in the compact matter map such that it becomes like the smooth matter map , and perhaps indistinguishable from it .
our next step is to execute this same procedure for all the compact matter maps in figure [ maps strip ] .
figure [ graphs ] shows how the difference measure decreases as source size increases , for all the compact matter maps .
figure [ graphs ] ( a ) is for the maps for image m , and ( b ) is for image s. both plots are similar and show a smooth decrease in difference ( vertical axis ) as the source size ( horizontal axis ) is increased .
the difference measure falls faster early on and then flattens out at a value below @xmath33 for all the maps .
the difference starts out smaller for image m , indicating that the compact object maps are slightly more similar to the smooth matter maps in this image .
maps made with larger masses begin at a higher difference measure , indicating larger masses produce higher amplitude variability between the maps .
the discriminating power of the difference measure appears to be highest in the middle of the mass range used .
overall , the behaviour is smooth and consistent within each image , and the trend is the same for each image .
as figure [ lightcurves ] has shown , the difference between the maps is due to nanolensing variability that appears in a light curve as `` wiggles '' that are smaller than microlensing events . by observing this variability , it is possible to infer that small objects are in the lens , and so distinguish these objects from smooth matter .
observing nanolensing requires that the amplitude of nanolensing variability be above the threshold of observing sensitivity , which we define here as above a certain magnitude threshold .
once nanolensing is found it then depends on its amplitude in relation to the observing threshold as to whether it discriminates between masses for the small objects . if the threshold is high , then only high amplitude variations can be seen , and from figure [ graphs ] this means that only large masses can be inferred in the lens .
if the threshold is low , but still only high amplitude variation is seen , it is possible to conclude that smaller masses are _ not _ in the lens because smaller variations are not seen . added to this is the size of the source , which reduces the amplitude of nanolensing ( and microlensing ) variability , so a knowledge of source size is important .
we first look at some properties of the map analysis and then proceed to look at the duration of small - scale changes in light curves , whether these are nanolensing events , and what source sizes are needed to resolve them .
it is evident that the simple measure used for the difference between the compact mass and smooth matter maps exhibits smooth regular behaviour , which we attribute to the physics of the situation , the high quality of the maps , and the very large number of small objects . as
the behaviour is similar in both images , our difference measure is useful for both positive and negative parity images .
the measure is an average of the amplitude of nanolensing variation over the whole map , and hence also an average of light curve deviations , e.g. those seen in figure [ lightcurves ] .
@xcite measured nanolensing from light curve deviations directly , by finding peaks indicating small - scale changes , @xcite used a magnification map analysis similar to ours , based on fractional differences between magnification maps . using figure [ graphs ]
we can determine the mass of the small objects in the lens based on the amplitude of nanolensing variation and an assumption of source size .
for example , because all the curves flatten out to below 0.047 mag at a source size of 1.4 er ( 0.033 pc in our mock system ) , small compact objects will not be distinguishable from each other , or from smooth matter , for a source larger than 1.4 er , unless the observation threshold is smaller than 0.047 mag .
figure [ graphs ] can also be used to correlate nanolensing mass with source size if the amplitude of nanolensing is above a certain threshold .
figure [ diff_versus_mass ] shows the largest source size ( vertical axis ) that will allow events with amplitudes of 0.12 , 0.1 .
0.8 . 0.6 , or 0.04 mag ( separate lines ) to be observed , for the different masses ( horizontal axis ) .
the plots are obtained by drawing horizontal lines through the graphs in figure [ graphs ] .
the plot is log - log , which produces a linear trend , as found by @xcite .
these show that small sources are needed to observe high amplitude variations , while a large source can only produce low magnitude variation .
the source sizes in figure [ diff_versus_mass ] range from about 0.07 er to 1.34 er ( 0.0017 to 0.032 pc ) . from @xcite
we adopt a value of 0.1 mag as a threshold difference measure for observing nanolensing
. then , to infer small objects of mass @xmath34m@xmath1 in the lensing galaxy , the source must be of size 0.09 er ( 0.0021 pc ) or smaller ; for objects of @xmath24m@xmath1 , a source size of 0.94 er ( 0.022 pc ) or smaller will do .
the two plots in figure [ diff_versus_mass ] also show that compact masses are slightly harder to find using image m , because the source sizes needed for this are a bit lower overall .
this may also be seen in light curves shown later on ( section [ obs_by_time ] ) , where the nanolensing variability appears smaller in image s than that in image m , for the same source .
we may use figures [ graphs ] and [ diff_versus_mass ] to determine the lower limit of the masses that could be discriminated , if the source was quasar sub - structure such as an accretion disk .
for example , @xcite estimate the size of the @xmath35 nm emission region in mg 0414 + 0534 to be @xmath36 cm , based on microlensing analysis of image flux ratios .
this size is 0.23 er in the mock lens system ; using 0.1 mag as the threshold for observing nanolensing in figure [ diff_versus_mass ] , variability could be seen for small masses down to about @xmath37m@xmath1 .
to see variability caused by @xmath3m@xmath1 masses , the source needs to be 0.09 er .
the @xmath38 nm emission region in mg 0414 + 0534 is estimated , from h@xmath27 flux , to be @xmath39 cm = @xmath40 pc in mg 0414 + 0534 , or 0.0026 er in the mock system @xcite .
as this is below 0.09 er a source of this size can produce nanolensing that indicates and discriminates between all the small mass values we have used .
in fact from figure [ graphs ] , with a source this small we would observe nanolensing variations of 0.4 mag or better for all the masses .
@xcite and @xcite used different parameters for the lens ( e.g. @xmath8 = 0.2 , @xmath9 = 0.5 ) and a different criterion for differentiating compact vs smooth , but they suggest 0.2 er as the source size below which nanolensing could be distinguished . if we use an observing threshold of 0.1 mag then 0.2 er sits nicely within our range , of 0.09 to 0.95 er , indicating this source size is characteristic of the problem .
the mock lens used here is similar to that of mg0414 + 0534 , but using mg 0414 + 0534 for our analysis would require 10 billion objects in the lens , if other parameters are held the same .
nevertheless we may make comparisons with mg 0414 + 0534 by setting the einstein radius in the mock lens to the einstein radius of mg 0414 + 0534 : 0.0136 pc ( z@xmath41 = 0.9584 , z@xmath42 = 2.639 ; * ? ? ?
* ; * ? ? ?
* ) . in that case , the parsec sizes quoted in the previous discussion scale down by a factor of 0.58 , so that the range in figure [ diff_versus_mass ] becomes 0.0009 to 0.018 pc .
@xcite find a value of 0.007 pc for the i - band emission region in mg 0414 + 0534 , which is within this range , and the einstein radius would be 0.52 er based on 1 er = 0.0136 . for a source of size 0.52 er , and consulting figure [ diff_versus_mass ] , light curve variability due to masses down to @xmath43m@xmath1 in image s could be observed , at a threshold of 0.1 mag .
for an einstein radius of 0.0236 pc in the mock lens , no variability over this threshold would be seen for any mass .
we therefore expect that mg 0414 + 0534 will provide more opportunity for determining small mass size , based on nanolensing amplitude , than the mock lens system .
@xcite used a uniform disk for the source profile .
a uniform disk with radius r has a half light radius of 0.71 r , a disk with a gaussian profile and radius of r = 3@xmath25 has a half - light radius of 0.28 r. therefore our sources are equivalent to a uniform disk that is 0.4 the size of the size stated here .
we have not yet discussed the temporal characteristics of the nanolensing variation whether it is observable , and how to observe it . in this section
we examine these questions , without embarking on a full statistical analysis , which is reserved for future work .
we assume a velocity of the quasar source over the magnification maps of 600 km s@xmath44 , which has been used in the past for the mock lens system @xcite .
this gives a time of 38 years to cross the einstein radius of a solar mass star , and 71 days = 0.00012 er for the einstein radius of our smallest mass object ( @xmath3m@xmath1 ) . in a lens where there are many objects and a complex magnification map , such as we have here , we expect events to be seen on timescales smaller than this .
in what follows we have used horizontal light curves relative to all the magnifications maps .
the variability is at its greatest , i.e. highest frequency and shortest period , in that direction , because the shear operates in the vertical direction on the magnification maps , and the caustics are extended in that direction .
sources moving diagonally across the caustics will show a longer time period for nanolensing variations , compared to a horizontally moving source .
we now turn to an analysis of the nanolensing variations .
figures [ wiggles_by_mass ] and [ wiggles_by_size ] show light curves taken from the image m and image s maps in figure [ maps strip ] in the horizontal direction , over a time period of 100 years .
figure [ wiggles_by_mass ] uses a fixed source size of 0.02 er and varies the mass of the small objects , so that each panel represents a different mass , indicated in the top of each panel .
the top row represents the largest mass , @xmath24m@xmath1 , the next row is @xmath45m@xmath1 , and so on in the same order reading down the page as in figure [ maps strip ] .
the bottom row is the smooth matter map with a source size of 0.02 er .
figure [ wiggles_by_size ] uses the maps that have small objects of @xmath46m@xmath1 , but varies the source size , indicated in the top of each panel .
the top row represents a source of size of 0.05 er , and the next row is 0.01 er , with the bottom row being the smooth matter map at the pixel resolution of the maps 0.002 er . in both figures the left column is image m , the right column is image s. the light curves confirm the results of the map analysis , that the amplitude of the nanolensing variability decreases as the small objects or source size decreases .
however we have the addition of new temporal information : as the mass of the small objects or the source size decrease , the frequency of small - scale variation increases .
it appears , by looking , that the nanolensing variability is easier to distinguish from the large scale ( microlensing ) variability in image s compared to image m , because the microlensing variability is greater in s. to confirm this we find and compare the nanolensing and microlensing variability for image m and s. the nanolensing variability is the difference measure we have been using .
the microlensing variability is the range of magnifications within the smooth matter map .
we present the results for the smallest mass objects , @xmath46m@xmath1 , and a source size of 0.002 er , the map pixel resolution .
the nanolensing variability for these is 31% of the microlensing variability , in image m , in terms of amplitude . for image s ,
the value is 16% ; hence it appears easier to separate the nanolensing from the microlensing in image s , at this scale .
naturally this is partly because we have used a bi - modal mass distribution , and not a continuous mass spectrum .
we will mention this again , below . the source size used in figure [ wiggles_by_mass ] ,
0.02 er , crosses its own radius in 140 days , which is above the crossing time for the einstein radius of a @xmath46m@xmath1 object ( 71 days ) and therefore probably above the time - scale of nanolensing caustics .
so what is producing the events shown in figures [ wiggles_by_mass ] , and in [ wiggles_by_size ] ? to see this we show in figure [ bands](a ) the square region of 100@xmath20100 years around the light curve for the map with @xmath47m@xmath1 objects , image m , and a source size of 0.05 er ( 0.0012 pc ) .
these are mass and source sizes that sit in the middle of our ranges .
the `` wiggles '' in the light curve are caused not by individual nanolensing caustics but by bands of densely packed caustics .
the nanolensing caustics are not resolvable at the sources sizes and time periods of figures [ wiggles_by_size ] and [ wiggles_by_mass ] .
to highlight this point , we generate a map of 0.2 er in width , at the same pixel resolution as the maps in figure [ maps strip ] , and use a source of size 0.0001 er ( @xmath48 pc ) . a square segment of this map , of width 1 year ( @xmath49 pc ) ,
is shown in figure [ bands](b ) . at this resolution
we can see the nanolensing caustics .
figure [ bands ] ( c ) shows a light curve cut across the map in ( b ) .
the highest peak in the light curve has a width of about 8 days ( 5.7@xmath50 er ) , however there is a small peak just to the left of it which has a width of only 2 days .
a small source crossing these will produce true nanolensing _
events_. the events are resolvable for this scale of mass with this scale of source size , due to the fact that the size of 0.0001 er is comparable to the einstein radius of the small objects ( 0.00012 er ) and smaller than the width of the caustics : 8 days = 0.00057 er . the fact that we have so far been looking at caustic bands may explain why the nanolensing variability , compared to the microlensing variability , appears to be smaller in image s compared to m. the formation of the caustics bands may be on a similar scale in image m and s , whereas the microlensing caustics are not they appear to have a larger magnitude range in image s. however , if only true nanolensing and microlensing caustics were compared , we may find that the relationship of nanolensing to microlensing comes out the same for both images .
we leave this for future investigation .
we noted in section [ procedure ] that when a map generated from stars@xmath31smooth matter is subtracted from the corresponding map generated from stars@xmath31small masses , the residual magnitudes have a roughly gaussian distribution .
the residual magnitudes are the nanolensing variability , suggesting that gaussian noise in microlensing observations will be indistinguishable from nanolensing .
however , subtraction of the maps in figure [ maps strip ] resolves only the caustic bands , which produce the variability seen in figures [ wiggles_by_mass ] and [ wiggles_by_size ] .
if variability due to nanolensing caustics could be resolved , we expect the magnitude distribution to assume a shape more typical of microlensing @xcite .
this would make it possible to distinguish nanolensing from instrument noise .
another method of distinguishing noise be may be to examine the power spectrum of the lightcurves in figures [ wiggles_by_mass ] and [ wiggles_by_size ] . by inspection
, the caustics bands produce a periodicity in the lightcurve that should appear as a peak in a power spectrum ; such a spectrum will distinguish nanolensing from white and other `` colours '' of noise .
a detailed investigation of noise issues will be set aside for future work .
as seen in figures [ wiggles_by_mass ] and [ wiggles_by_size ] , we find that events have a time scale on the order of a year or less .
figure [ big_cadence ] shows a 10-year long segment of the light curve for the case of the @xmath3m@xmath1 masses , source size 0.01 er , for image m ( top row ) and image s ( bottom row ) ; that is , corresponding to the middle row in figure [ wiggles_by_size ] .
the light curves have been sampled at regular intervals of 90 days , and the resultant points plotted in crosses on the light curve .
this indicates that an observing cadence of 90 days is enough to see the nanolensing variability produced by the caustic bands . to find out how often observations need to be made to observe true nanolensing caustic crossings
, we use light curves from the map of width 0.2 er , created with small objects of @xmath3m@xmath1 .
we use two source sizes , an accretion disk of 0.002 er , and the smallest source size used previously : 0.0001 er .
light curves are cut horizontally across the map and a representative segment extracted , showing nanolensing events due to nanolensing caustics .
these are in figure [ cadence ] , where the top row is for image m , the bottom for image s ; the left column is the source of 0.002 er , the right column is 0.0001 er . the time span is 1 year in the left column , and 4 weeks in the right column .
the light curves are sampled at 14 days ( left column ) and 2 days ( right column ) which we suggest is the longest sampling time that will accurately reconstruct these light curves .
only the right column shows true ( albeit smoothed ) nanolensing caustic crossings
the left column does not . in the left column , nanolensing caustics
have been smoothed and combined .
therefore the source of 0.002 er is still too large to truly resolve the nanolensing caustics a size of order 0.0001 er is needed . at 814 nm ( 271 nm in the rest frame ) ,
the predicted flux size of the accretion disk in mg 0414 + 0534 is 0.002 er = 2@xmath51 cm @xcite .
we assume that the disk is a shakura - sunyaev thin disk @xcite where the size of the disk ( @xmath52 ) is given by @xmath53 where @xmath54 is the radius at @xmath55 . using this equation and the flux size at 814 nm as a reference
, we determine that an accretion disk of size 0.0001 er in the mock lens system would be emitting at a wavelength of 25 nm , i.e. the extreme uv . at shorter wavelengths such as these
the emitting region is likely to be highly variable , and such intrinsic variability could interfere with observations of nanolensing events . on the other hand ,
the time delay between images in mg 0414 + 0534 is @xmath56 1 day @xcite , making it possible to separate the intrinsic variability from the nanolensing events in that quasar , since intrinsic variability will be correlated between the images , and nanolensing will not .
we also note , however , that 0.0001 er is a very small source for a high redshift quasar . @xcite give a mass of 1.82@xmath57m@xmath1 for the super - massive black hole ( smbh ) , which means the smallest innermost stable circular orbit ( isco ) , using an extremal co - rotating kerr black hole ( @xmath58 ) , is 2.7@xmath51 cm = 8.7@xmath23 pc = 0.003 er for the mock lens .
this is above the source sizes we have used in figure [ cadence ] .
if the einstein radius were altered to be the same as that in mg 0414 + 0534 , the parsec source sizes given previously will scale down by a factor of 0.58 .
the source velocity in mg 0414 + 0534 is 270 km s@xmath44 , and with the different einstein radius means the light curve durations scale up by a factor of 1.27 . in figure
[ cadence ] the durations in the left and right columns become 463 days and 35 days respectively , with the sampling points occurring every 17 days and 2.5 days .
the source size for the left column becomes 0.0011 er and the right column 0.00006 er . the observing cadence has not altered a great deal , but the source needed to resolve nanolensing caustics is now becoming improbably small .
we have shown that high - quality maps generated from very large numbers of compact masses can be analysed with simple statistical measures . using a root - mean - square measure
, we show that if the source size is increased the microlensing by compact objects and microlensing by smooth matter become indistinguishable , but below a threshold it is possible to discriminate between small masses in a lens using the amplitude of nanolensing variability .
it will require small sources , of order 0.1 er , to distinguish very small masses from smooth matter at some reasonable level of detection magnitude , e.g. 0.1 mag . in future work we will increase the number of data points in figure [ diff_versus_mass ] by generating maps with other small mass values ; this is computationally expensive compared to fleshing out the range of sources , because map generation is more expensive compared to source convolution .
however , with new tools @xcite , this may be feasible .
the method also needs to be applied not only to mg0414 + 0534 , which will require billions of lens masses , but to other lensed quasars . rather than using a bi - modal mass function ,
other distributions such as salpeter can be tested , and the percentage of small to large objects @xcite can be varied .
it will also be necessary to obtain a statistical variance of the results by generating maps that are identical to those in figure [ maps strip ] but with different randomly - generated locations for the objects .
these maps can be passed through the analysis and the results compared to those presented here .
a slight difference is expected as the large - scale caustics in the compact matter maps may shift slightly due to a different field of small objects surrounding them
. however , these differences should not be significant .
the analysis of the light curves show that there are three scales or levels of lensing in this study microlensing due to the 1m@xmath1 stars , variability due to the bands of nanolensing caustics , and peaks due to the nanolensing caustics themselves .
further work needs to be done to determine the scale at which the nanolensing caustics resolve out of the caustic bands , and if , at the different scales , and in different images , the amplitude and duration of nanolensing variability remains the same .
it is possible that such analyses will provide tighter constraints on the masses within the lens , rather than the general analysis here .
nanolensing events are very small .
even sources as small as the optical accretion disk of mg 0414 + 0534 will not be enough to resolve individual nanolensing caustics accurately , and smaller regions will be necessary to do this . such regions
are less than the expected isco of the smbh , but if they exist , and are not too intrinsically variable , an observing cadence of 2 days will resolve nanolensing caustics .
an observing proposal could be made on this basis , either using the hst or magellan telescopes , for observations of nanolensing events in a quasar such as mg 0414 + 0534 .
computing facilities were provided by the high performance computing facility , university of sydney , and the nci national facility at the anu .
hugh garsden acknowledges funding support from an australian postgraduate award .
this work is undertaken as part of the commonwealth cosmology initiative ( www.thecci.org ) , and funded by the australian research council discovery project dp0665574 .
we thank the anonymous reviewer whose comments helped us to improve the quality of the paper . | gravitational microlensing of planetary - mass objects ( or `` nanolensing '' , as it has been termed ) can be used to probe the distribution of mass in a galaxy that is acting as a gravitational lens .
microlensing and nanolensing light curve fluctuations are indicative of the mass of the compact objects within the lens , but the size of the source is important , as large sources will smooth out a light curve .
numerical studies have been made in the past that investigate a range of sources sizes and masses in the lens .
we extend that work in two ways by generating high quality maps with over a billion small objects down to a mass of @xmath0m@xmath1 , and by investigating the temporal properties and observability of the nanolensing events .
the system studied is a mock quasar system similar to mg 0414 + 0534 .
we find that if variability of 0.1 mag in amplitude can be observed , a source size of @xmath2 einstein radius ( er ) would be needed to see the effect of @xmath3m@xmath1 masses , and larger , in the microlensing light curve .
our investigation into the temporal properties of nanolensing events finds that there are two scales of nanolensing that can be observed
one due to the crossing of nanolensing caustic bands , the other due to the crossing of nanolensing caustics themselves .
the latter are very small , having crossing times of a few days , and requiring sources of size @xmath4 0.0001 er to resolve . for sources of the size of an accretion disk ,
the nanolensing caustics are slightly smoothed - out , but can be observed on time scales of a few days .
the crossing of caustic bands can be observed on times scales of about 3 months .
[ firstpage ] galaxies : structure gravitational lensing : micro dark matter
quasars : individual : mg0414 + 0534 methods : numerical |
ngc 4051 is a nearby ( @xmath7 ) , low luminosity ( @xmath8 erg s@xmath2 ) agn , optically classified as a narrow - line seyfert 1 ( nls1 ) . in common with other members of the nls1 class , and following the well - known anti - correlation of x - ray variability amplitude with luminosity ( e.g. green , m@xmath9hardy & lehto 1993 ; nandra et al .
1997 ) , ngc 4051 typically shows large amplitude x - ray variability on short time - scales ( e.g. lamer et al .
2003 ; m@xmath9hardy et al .
, in prep . ) . in the medium x - ray band ( 2 - 10 kev ) , the x - ray continuum slope is well - correlated with flux , with photon index ( @xmath10 ) increasing from @xmath11 to 2.4 over a factor @xmath12 increase in observed flux @xcite .
flux variations in the euv and 2 - 10 kev bands are also very well - correlated @xcite , but a recent _ chandra _ high energy transmission grating ( hetg ) observation of ngc 4051 @xcite suggests this result is due to a correlation between the x - ray power - law continuum and a ` soft excess ' which dominates in the soft x - ray band , rather than variations in the x - ray power - law alone .
besides the short - term x - ray variability , ngc 4051 also shows long - time - scale flux changes , as revealed by monitoring with the _ rossi x - ray timing explorer _ ( rxte ) . in particular , on two occasions - in 1998 and
later in 1999 - ngc 4051 appeared to enter a distinct , prolonged low - flux state which lasted 5 and 3 months respectively @xcite , during which the absolute amplitude of x - ray variability was much smaller than the large amplitude variations normally seen .
the x - ray spectrum of ngc 4051 during this ` low state ' , was measured simultaneously by _
bepposax _ and _ rxte _ in may 1998 @xcite and found to be extremely hard ( @xmath13 ) above @xmath14 kev , also showing a prominent ( @xmath15 kev ) iron k@xmath16 line .
below 4 kev , the spectrum was found to be extremely soft ( with a power - law fit giving @xmath17 ) .
flux variability in either the soft or hard band was not detected during these @xmath18day - long observations .
the natural interpretation of the low state was that the central x - ray source had somehow switched off , revealing the hard spectrum and prominent fluorescent iron k@xmath16 line expected from distant ( @xmath19 light - months ) cold material , possibly the molecular torus , which did not have time to respond to the long - term change in flux @xcite .
the soft x - rays might also correspond to some extended component , a view supported by a claim by @xcite that _ rosat _ hri observations of ngc 4051 showed evidence for extended ( @xmath20 arcsec ) soft x - ray emission , of a comparable flux to that observed during the low state .
the hypothesis that the low states in ngc 4051 represent some physically distinct behaviour of the agn , possibly with the central x - ray source switching off , rather than merely a low - flux epoch associated with the same variability processes which produce the short - term variability , is supported by optical monitoring observations . @xcite
have shown that during the 1998 low state , the variable component of the he ii 4686 emission line seems to disappear while h@xmath21 remains strongly variable , implying that the variable component of the ionizing continuum above @xmath22 ev disappears during the low state , possibly due to the transition of the inner accretion disk to a radiatively inefficient advectively dominated accretion flow ( adaf ) . in order to test models for the origin of the low states in ngc 4051
, we were awarded a 50 ks target of opportunity observation ( too ) using the acis - s instrument on _ chandra _ , to measure the continuum spectral shape of the primary x - ray source spectrum over a broad band , and disentangle the soft continuum of the central source from any extended soft emission .
ngc 4051 entered into a low state of @xmath23 months duration in january 2001 , and was observed by _
chandra _ in early february , as described below .
the _ chandra _ too of ngc 4051 was triggered using data from our continuing long - term monitoring observations of ngc 4051 with _
[ monlc ] ) . in late 2000 , ngc 4051 showed a decline in long - term average x - ray flux over several months , before entering a prolonged low - flux state around 2001 jan 10 .
_ chandra _ successfully observed ngc 4051 on 2001 feb 6 06:24:59 - 20:55:58 ut , for a total useful exposure of @xmath24 ksec with the spectroscopic array of the acis instrument ( acis - s ) without any transmission grating in place .
the source was located on the back - illuminated chip ( s3 ) and offset by the standard 20 from the default on - axis location to avoid a node boundary ( see the _ chandra _ proposers observatory guide [ pog ] ) .
we chose a 1/8 subarray mode with 0.43 s frametime to mitigate pile - up effects , resulting in a 1 x 8 image .
the analysis of the data was performed using standard ciao tools ( version 2.2.1 ) .
we regenerated the level 2 event file corrected with the latest gain and aspect data .
we extracted the spectrum of the source from a 12 pixels radius ( 6 ) circular aperture , to yield a total of 57760 counts . assuming the empirical spectral model which we use later ( section [ empmod ] )
, we estimate from web - pimms a pileup fraction of @xmath25% .
the background spectrum was extracted from a 7000 pix@xmath26 rectangular region , away from the target and containing no obvious sources . because the acis qe degradation affects our data
, we used the ` corrarf ' code provided by the _ chandra _ x - ray center to correct all the effective areas used in our spectral analysis . throughout the paper ,
we use xspec v11.0.1 to carry out spectral fits . to detect any possible weak x - ray extended emission , we extracted and smoothed the image of the source and compared it with a simulated chandra psf calculated with chart at the exact location of the source ( assuming our best fitting empirical spectral model described later in section [ empmod ] ) .
the observed and simulated images are shown in fig .
[ psf ] . fitting the
azimuthally averaged source radial profile with the simulated psf radial profile indicates only weak extended emission in the se direction out to 20 .
details of nearby source detections and further imaging analysis will be presented in a forthcoming paper ( fruscione et al . in preparation ) , so we do not carry out any detailed analysis of extended emission here .
however , we note that excluding the readout streak , we find 796 counts in the annular region 12 - 25 pixels ( 6 - 12.5 ) from the central source , which is dominated by the central psf so that any extended emission contributes @xmath27% of the total observed counts , and is intrinsically very weak ( i.e. much weaker than the component claimed by singh 1999 ) .
this result strengthens the finding of @xcite , based on the zeroth order hetg grating image , that there is no significant extended soft x - ray emission in ngc 4051 , and implies that the central source spectrum we extract is dominated by nuclear emission and not contaminated by any extended component .
to demonstrate the broadband shape of the _ chandra _ 2001 low state spectrum relative to different power - law slopes , we first plot an _ unfolded _ spectrum ( fig . [ ufspec ] ) fitted to a simple absorbed power law of fixed photon index @xmath28 and galactic absorbing column , @xmath29 ( from elvis , lockman & wilkes 1989 ) .
the @xmath28 power - law model is plotted as a dashed line and for comparison we plot a steep power law model ( @xmath30 ) as a dotted line
. we also plot the model of a reflection spectrum ( including iron k@xmath16 line ) that provided a good fit to the 1998 may low state spectrum above 4 kev @xcite .
note that although unfolded x - ray spectra are always model - dependent , that dependence is weak for broadband continuum models of the type we plot here , so that in the 0.5 - 10 kev range the unfolded data relative to all three models is almost identical .
therefore the unfolded observed spectrum is a good representation of the true continuum shape .
clearly the 2001 low state spectrum has the same ` hybrid ' form of a very hard component above a few kev , and a soft component at low energies , as seen in the 1998 low state .
however , the normalization of the hard component is a factor @xmath31 greater than in 1998 may , suggesting that the primary continuum itself is intrinsically hard . in fig .
[ chanlcs ] we show the _ chandra _ lightcurves in soft ( 0.3 - 2 kev ) and hard ( 3 - 10 kev ) bands , together with the corresponding hardness ratio
. clearly the soft and hard lightcurves are well correlated ( ccf analysis shows any lag must be less than @xmath32 ks ) , but the hard band is significantly more variable than the soft band ( hard band fractional rms 29% versus 16% in the soft band ) , so that the broadband continuum becomes harder with increasing flux . the large - amplitude continuum variations in the hard band confirm that we are seeing the primary continuum during the 2001 low state , and not simply emission from a distant reflector .
the shape of the broadband spectrum of ngc 4051 in the low state is clearly an unusual hybrid of an extremely hard component above a few kev and a prominent soft excess at lower energies .
we can confirm that this unusual shape is not caused by pileup effects ( which would redistribute low energy photons to higher energies and so harden the spectrum ) by comparing the spectrum with a spectrum extracted from only the wings of the psf ( excluding counts from a circle of 1.5 " diameter centred on the source ) . in fig .
[ plup ] we plot the ratio of the total spectrum and psf wing spectrum to the same simple power - law ( renormalized by 0.223 , the ratio of counts in the two spectra , to fit the psf wing spectrum ) .
the effect of pileup is to harden the soft component of the spectrum somewhat , but the discrepancy is small , consistent with the estimated pileup fraction of 14% , and should not affect the broad interpretation of the spectral shape , although we should be cautious about assigning too much significance to the specific spectral models and parameters which we fit . with this caveat in mind ,
we proceed with more detailed spectral fits to empirically describe the spectral shape and variability . unless otherwise stated , fits to the broadband _ chandra _ spectrum
are carried out to data in the 0.5 - 10 kev range , in order to mitigate effects of pileup and uncertainties in calibration below 0.5 kev
. the 2001 february low state spectrum in the 3 - 10 kev band can be well - fitted ( reduced @xmath33 , @xmath34 for 259 degrees of freedom ) with a hard power law ( @xmath35 ) plus an unresolved gaussian at 6.4 kev , with equivalent width @xmath36 ev ( absolute line flux @xmath37 photon @xmath1 s@xmath2 ) . below 3 kev , the spectrum is complex ( as is apparent in fig .
[ ufspec ] ) , probably due to a combination of emission / absorption features , slight pileup effects on high @xmath38 data and possibly an intrinsically unusual continuum shape .
therefore we can not obtain a good overall spectral fit with any model we have attempted .
however we obtain a simple approximation to the broadband spectral shape ( @xmath39 for 424 degrees of freedom ) by fitting a blackbody plus power law , together with an absorption edge at 0.7 kev ( optical depth @xmath40 ) and an unresolved emission line at 0.89 kev ( these features likely correspond to the 0.74 kev ovii edge and 0.91 kev neix emission line reported by collinge et al .
the best - fitting flux for the 0.89 kev line feature is @xmath41 photon @xmath1 s@xmath2 , consistent with that observed by collinge et al .
the best - fitting blackbody and power law parameters are shown in table [ spectab ] ( together with hard and soft - band fluxes ) and the data , model and data / model ratio are plotted in fig .
[ totfit ] .
note that due to the statistically poor fit , we can not quote meaningful errors on the fit parameters .
the observed fluxes correspond to luminosities of @xmath42 erg s@xmath2 and @xmath3 erg s@xmath2 in the 0.5 - 2 kev and 2 - 10 kev bands respectively ( assuming @xmath43 km s@xmath2 mpc@xmath2 ) .
the power law plus blackbody combination leads to a steeper power law than obtained when fitting only above 3 kev , but residuals above 5 kev suggest that the model underestimates the hard flux , implying that some spectral curvature may be present or some additional spectral complexity at hard energies ( section [ hardspec ] ) . replacing the blackbody with a steep power law does not worsen the fit , ( @xmath44 remains at 1.56 for a best - fitting power - law photon index @xmath45 ) . extending the fitted spectral range to lower energies ( 0.4 kev ) shows an apparent flattening of the continuum around 0.5 kev which can be replicated by the blackbody fit but not the soft power law , although we can not be certain that this flattening is not due to pile - up effects .
however , @xcite also find that the soft excess observed in the ` normal ' state rolls over at low energies and is more consistent with a blackbody than a power law , so we continue with the assumption that a blackbody provides a good approximation to the soft excess in the low state .
all the simple models we have fitted to the broadband spectrum leave residuals above the model at energies @xmath19 kev ( see fig . [ totfit ] ) .
it is likely then that our power - law model for the continuum at harder energies is incorrect .
one possibility is that significant reflection is present in the spectrum , leading to apparent upwards curvature towards hard energies . to test this possibility
, we included a reflection component in the model .
we used the pexrav model in xspec @xcite , assuming solar abundances and fixing inclination at 30@xmath46 and power - law cutoff energy at 100 kev although note that these latter two parameters do not affect substantially the results of the fit .
first , we assumed that the direct power - law continuum included in the observation is the primary continuum to be reflected ( i.e. the reflection responds to the change in continuum flux and @xmath10 in the low state ) .
the fit is improved substantially ( @xmath44=1.42 for 424 d.o.f . ) , with the residuals at high energies being significantly reduced , as expected .
the best - fitting photon index is still quite hard @xmath47 , but the covering factor of the reflector is @xmath48 , and this number can not be substantially reduced without worsening the quality of the fit steradian sky coverage of the reflector ( as seen from the continuum source ) , results in @xmath49 . ] .
such a large covering factor of reflection might be expected if the reflector is distant ( and hence is still illuminated by the normal state continuum ) , for example , if the reflector corresponds to the possible pure reflection spectrum observed in the may 1998 low state , which could originate from cold material light months or light years from the continuum source ( @xcite,@xcite ) .
therefore , we kept the direct primary continuum slope free and fixed the illuminating continuum slope and normalisation at 1 kev to @xmath50 , @xmath51 photon cm@xmath2 s@xmath2 kev@xmath2 respectively , to match those in the reflection model used to fit the 1998 may _ rxte _ pca plus _ bepposax _ mecs spectra @xcite , to see if that model might explain the residuals seen at hard energies in the 2001 feb
_ chandra _ spectrum . with @xmath44=1.52 for 425 d.o.f .
( for observed primary continuum @xmath53 ) , the best fit is worse than the previous fit obtained by forcing the illuminating continuum to be identical to the direct primary continuum .
furthermore , the fit requires a covering factor @xmath54 , compared to @xmath55 which was used to fit the 1998 may spectrum .
since we fixed the normalisation of the illuminating continuum to that fitted to the 1998 may spectrum , a larger covering factor might correspond to a greater time - averaged ( over months ) illuminating continuum flux prior to the 2001 low state .
however in that case we would also expect a comparably larger 6.4 kev narrow line flux which we do not observe .
therefore the high energy residuals from the power law can not be self - consistently explained by the presence of a reflection spectrum from a distant reflector . due to the difficult physical interpretation of any reflection model
, we will continue to use a simple power law in our fits to the broadband _ chandra _ spectrum , as an _ approximate _ representation of the hard spectral component .
we can split the _ chandra _
observation into two halves of exposure @xmath56 ks and @xmath57 ks respectively , corresponding to the low and high - flux epochs seen in the lightcurves in fig .
[ chanlcs ] ( the split is denoted by a dotted line in the figure ) , in order to examine spectral variability in more detail . fitting the simple empirical ( no reflection ) model used to fit the spectrum of the whole observation , we find that only the _ relative _ normalizations of the black body and power - law components change markedly , with blackbody and power law normalization increasing by 28% and 68% respectively between low and high flux epochs , while the black body temperature , power - law slope , edge depth and line fluxes do not change noticably .
the fit is not particularly good however ( @xmath44=1.54 for 643 d.o.f . ) , partly due to the presence of positive residuals at high energies .
the relatively small variation in the flux of the soft spectral component compared to the hard component is as expected considering the lower variability seen in the soft band ( fig .
[ chanlcs ] ) .
we can also use the low and high - flux data obtained during the low state to further test whether a constant reflection component from a distant reflector could be present in the spectrum .
we applied the same fixed reflection model ( i.e. assuming may 1998 parameters ) as used in section [ empmod ] , fixing the covering factors to be the same in both low and high flux spectra ( so the reflected component flux is constant ) .
the best - fitting model again produced a better fit than the empirical power - law model described above ( @xmath44=1.31 for 641 d.o.f . ) , but requires @xmath58 .
furthermore , the model requires the direct power - law continuum to be steeper at low fluxes than at high fluxes ( @xmath59 versus @xmath60 ) , contrary to the result of taylor et al .
( 2003 ) that the varying spectral component hardens as flux decreases in ngc 4051 , due to spectral pivoting .
again , errors can not be well defined as the fit is still not formally acceptable , but if we force the low - flux direct continuum to be flatter or the same as the high - flux direct continuum , we obtain a best fit for for the identical slope case ( for @xmath61 ) , but the fit is significantly worse than when the slopes are left free ( @xmath44=1.42 for 641 d.o.f . ) .
the observed broadband spectral variability is therefore inconsistent with the presence of a constant reflection component from a distant reflector , of a magnitude which can explain the hard residuals .
we stress however that a much weaker reflection component ( comparable to that observed in may 1998 ) may be present , but it does not contribute signficantly to the hard continuum spectral shape .
although a simple power law plus gaussian model provides a good fit to the 3 - 10 kev _ chandra _ spectrum ( section [ empmod ] ) , the fitted power law is extremely hard ( @xmath62 ) . as noted in the preceding sections , combined black body plus power law fits to the 0.5 - 10 kev spectrum
favour a steeper ( and perhaps more plausible ) photon index of @xmath63 , but show significant residuals at energies above 5 kev ( fig . [ totfit ] ) . as shown in the previous sections , these hard residuals in the broadband
fits can not be simply explained by including a constant reflection component as expected from a distant reflector .
it might be argued that the model is incorrect at softer energies , e.g. if the soft component is in fact harder than the assumed black body , then the fitted slope of the hard power law component would steepen to compensate , resulting in an underprediction of the flux at harder energies .
due to the complexity of the spectral shape at soft energies and the expected distortion of the spectrum due to pile - up , we can not easily determine the best model for the soft spectral component .
therefore , to take into account the possibility of biases in the spectral fitting due to an incorrectly modelled soft spectrum , we now investigate the spectral shape in the 3 - 10 kev range , ignoring the data at lower energies .
we first test the constant reflection model fitted to the broadband low and high flux spectra in section [ specvar ] ( direct continuum @xmath10 allowed to be free , we do not include the soft spectral components : black body , edge and soft gaussian in the fit ) .
we find an acceptable fit ( @xmath64 for 329 d.o.f . ) , and hence can determine meaningful confidence limits on the fit parameters .
we confirm the result of the broadband fits that the hard spectral shape can not be largely due to reflection from a distant reflector , finding a 99% confidence upper limit on covering factor @xmath65 and corresponding upper limits on low and high - flux direct continuum slopes of @xmath66 and @xmath67 respectively , i.e. the varying continuum must itself be intrinsically hard .
an alternative possibility to explain the hard spectral shape is that it may be associated with strong disk reflection from close to the central black hole , which is boosted by gravitational light bending effects ( e.g. @xcite ) . in this scenario ,
the shape of at least some part of the hard residuals in the broadband fits above 5 kev may be caused by unmodelled emission in the form of a prominent _ broad _ iron line and the reflection component associated with it . to test this possibility ,
we first investigated the spectrum at higher energies using data from _ rxte _ , which has a greater collecting area and hence greater sensitivity to broad line features than _ chandra_. we combined _ rxte _ pca data from pcu2 ( which is continuously switched on ) , from all observations between 2001 january 10 and 2001 february 23 , using standard reduction techniques and selection criteria ( e.g. lamer et al . 2003 ) . the resulting 3 - 15 kev spectrum shows a prominent feature at @xmath68 kev , which may be a strong broad diskline .
in order to test whether such a broad line feature might be present in the _ chandra _ spectrum ( but is not obvious due to its broadness combined with the low @xmath38 at high energies ) we have attempted to fit both the _ chandra _ 3 - 10 kev spectra and _ rxte _ 3 - 15 kev pca spectra with the same model of a power law , plus narrow gaussian ( adopting the best - fitting line flux from table [ spectab ] ) plus laor diskline ( for a maximally rotating black hole , laor 1991 ) .
we fix the diskline parameters to be the same in both _
chandra _ and _ rxte _ fits , but allow the power law normalization to vary between the fits .
we allow the power law normalization to vary to account for the different fluxes observed during the _ chandra _ and combined _ rxte _ observations .
we caution that since the source flux is particularly low , systematic errors in the _ rxte _ pca spectrum may arise due to uncertainties in the background model ( e.g. see lamer , uttley & m@xmath9hardy 2000 ) .
therefore , the fit we obtain should be taken as merely indicative of whether or not a broad diskline - like feature could plausibly be present in the _ chandra _ spectrum , rather than an attempt to fit a realistic model ( in this context , we further note that a simple power - law is used to approximate what may in fact be a reflection - dominated continuum ) . the data and best - fitting models for both spectra are plotted in fig .
[ xtechan ] . a reasonable fit
is obtained to both spectra ( combined @xmath69 for 282 degrees of freedom ) for a model with power law slope @xmath70 and prominent diskline flux of @xmath71 photon @xmath1 s@xmath2 ( corresponding to line equivalent widths of 1 kev and 1.6 kev in the _ chandra _ and _ rxte _ spectra respectively ) , for line energy @xmath72 kev , emissivity index @xmath73 , a fixed disk inclination angle @xmath74 and inner and outer emitting radii of 1.23 and 33 gravitational radii respectively .
the fit shows that the prominent broad diskline - like feature seen in the _ rxte _ spectrum could be present in the _ chandra _ spectrum , and this feature may be the cause of some of the high - energy residuals observed in the broadband spectral fit .
some positive high energy residuals remain in the _ chandra _ spectrum however , and in the opposite sense to those in the _ rxte _ spectrum .
it is likely that at hard energies , the _ rxte _ pca spectrum may be significantly affected by uncertainties in the background model , so we next only fit the _ chandra _ 3 - 10 kev data with a diskline including an additional reflection component to explain the hard continuum shape in a self - consistent way .
we freeze the diskline parameters at those values obtained in the previous fit to the _ rxte _ and _ chandra _ spectra , and also include a reflection component tied to the direct power - law continuum shape .
the strength of the reflection component is not well constrained ( due to the degeneracy with the hard power law continuum which is required ) .
however , we note that a covering factor @xmath75 ( for @xmath61 ) is formally acceptable ( @xmath76 for 260 degrees of freedom ) and consistent with the large equivalent width of the diskline ( @xmath77 ev ) . the broadband continuum shape of the extended low state spectrum measured by _
chandra _ in 2001 february is remarkably similar to that observed by _
chandra _ during a brief ( duration @xmath6 d ) low flux epoch in 2000 april , while the source was in its normal high flux , highly variable state @xcite . to compare the shapes of 2000 april and 2001 february spectra , we fitted our approximate , blackbody plus power law model to the april 2000 low flux epoch data counts per spectral bin ] used by @xcite .
we allowed only the blackbody and power - law normalizations and the depth of the 0.7 kev edge to vary .
the data / model ratios are shown in fig .
[ combratio ] , together with the ratios for the fitted _ chandra _ spectrum ( section [ empmod ] ) for comparison , and the fit parameters are shown in table [ spectab ] .
we also carry out the same type of fit on the _ bepposax _ lecs and mecs spectra obtained during the 1998 low state ( see fig . [ combratio ] and table [ spectab ] ) .
the fit is not very good , although the broad continuum shapes of both the 1998 low state and 2001 low states seem to match up rather well ( see fig . [ combspec ] ) .
it is perhaps not surprising that significant deviations from the model remain in the _ bepposax _ data , since when the source continuum flux is exceptionally low ( as in 1998 may ) , weak constant - flux components in the spectrum ( e.g. unresolved emission lines ) are likely to be revealed .
therefore we conclude that , although our simple model does not adequately fit the detailed structures in the spectra , the same hard and soft spectral components can approximate quite well the broadband continuum shape of both the 1998 and 2001 low states , as well as the brief low flux epoch observed in april 2000 _ while the source was in a normal state_.
the x - ray spectrum of ngc 4051 during the 2001 low state is unusual , dominated by a very hard component ( @xmath78 ) above a few kev , and a very soft component at lower energies ( @xmath79 , although a @xmath80 kev blackbody may be a more appropriate model ) .
the question naturally arises as to what are the origins of these spectral components . in the 1998 low state spectrum ,
the hard continuum ( and prominent iron k@xmath16 line ) above @xmath14 kev and the much softer continuum at lower energies ( both of which did not vary significantly ) , were interpreted , respectively , as being pure reflection from distant material and extended soft emission @xcite .
we have seen in section [ image ] that the soft emission component in the 2001 low state is not extended in the _ chandra _ image , but originates from the unresolved central source .
this fact , together with a detailed spectral analysis of the _ chandra _ observation , show that the unusual spectrum observed in 2001 can not be explained by the distant - reflection / extended - emission model used to explain the 1998 low state spectrum .
instead , the emitting region must be small , so that a substantial part of the emission can be significantly variable on time - scales of hours , suggesting that both hard and soft components represent mainly the primary continuum emission .
this result does not necessarily invalidate the original interpretation of the 1998 low state spectrum , since this earlier spectrum was obtained when the continuum was significantly fainter than observed in february 2001 , so that reflection from distant material might have been important .
furthermore , a reflection model provided a significantly better fit to the 1998 low state spectrum measured by _
bepposax _ than a power - law model @xcite .
indeed , the continuing presence of a relatively weak , narrow fe k@xmath16 line in the 2001 low state spectrum might imply that the reflection component seen in may 1998 is present but largely masked by the intrinsically hard primary continuum in the 2001 low state .
having established that the 2001 low state spectrum is probably dominated by the primary continuum emission , we must consider how this unusual spectral shape relates to the spectrum observed during the source s normal state .
in fact it appears ( see section [ prevobs ] ) that the 2001 low state spectrum is not that unusual compared to the spectrum observed at comparable flux levels during the normal state . in other words , there may be a simple monotonic relation between continuum spectral shape and x - ray flux which is independent of whether the source is in a prolonged ( i.e. weeks ) low flux state or a much briefer ( hours to days ) excursion to low fluxes .
it is likely that the extreme hardness of the continuum at very low fluxes is a result of the continuation of the correlation between spectral slope and continuum flux which is observed at higher fluxes @xcite .
it is tempting to explain this correlation , which is observed in other agn ( e.g. lamer et al . 2000 , vaughan & edelson 2001 ) in terms of thermal comptonization models , where increases in the seed photon flux cause corresponding increases in the hard comptonized flux but also cool the comptonizing medium so that the emitted spectrum steepens at higher fluxes and the spectrum effectively pivots about a fixed point at high energies @xcite .
in fact , taylor , uttley & m@xmath9hardy ( 2003 ) have shown using a model - independent technique that in medium - energy x - rays ( 2 - 15 kev ) , the spectral variability of ngc 4051 is consistent with a combination of spectral pivoting of a varying power - law together with a constant hard spectral component .
it is possible that in the low state , the constant hard component is more fully revealed , which might partly explain the deviations from a simple power - law observed at hard energies .
we next consider how the spectrum above a few kev can be so hard at low fluxes ( irrespective of the duration of the low flux state ) , and how the hard continuum relates to the soft spectral component .
the extremely hard continuum observed in the low state is difficult to explain using conventional thermal comptonization models ( e.g. haardt , maraschi & ghisellini 1997 ) , so that non - thermal models may be required .
alternatively , if the hard component also contains a substantial disk - reflection component ( as suggested by the _
data , see section [ hardspec ] ) , this would allow the primary continuum to be steeper than the observed continuum , so that standard thermal comptonization models might still be appropriate .
it is unclear how a disk reflection component can be so strong , unless the disk somehow sees a much greater primary continuum luminosity than we observe ( e.g. if the continuum is beamed towards the disk through gravitational light bending effects , fabian & vaughan 2003 ) .
however , we note that evidence for an unusually strong iron diskline at relatively low continuum flux levels has been reported in the seyfert 1 galaxy mcg-6 - 30 - 15 @xcite .
fabian & vaughan ( 2003 ) have shown that the weakly - varying hard component in the spectrum of mcg-6 - 30 - 15 may be attributed to a strong _ disk _ reflection spectrum ( see also taylor et al .
2003 ) , so that at low fluxes the source should start to become reflection dominated . since the normal spectral variability of ngc 4051 also implies the presence of a constant ( or weakly varying ) hard component ( taylor et al . 2003 ) , it is natural to infer that strong disk reflection should be present in the low state of ngc 4051 , which may dominate over any constant reflection from the torus , even at low fluxes .
the problem of explaining the hard continuum component in the low state of ngc 4051 is compounded by the fact that it varies together with , and with greater amplitude than , the soft component .
if the soft component is a source of seed photons for a comptonizing corona ( perhaps from the hot , inner accretion disk ) , one would not expect the hard , multiply scattered x - rays to vary more than the soft photons .
the problem may be resolved if a substantial , constant component , which does not contribute seed photons is present in the soft x - ray band .
the factor @xmath23 reduction in fractional variability from the hard to the soft band can be accounted for if half the observed soft flux arises from a constant component ( perhaps extended emission on scales @xmath81pc so that it is unresolved by _ chandra _ ) .
note that the absence of varying heii emission but presence of variable h@xmath21 during the low state @xcite might be explained if , for whatever reason , the variability amplitude in the low state is smallest in the euv band , but increases again towards the uv .
although we can explain the spectral variability of ngc 4051 within the low state by invoking a constant soft component in addition to a variable soft component , we can not rule out the possibility that the spectral variability is intrinsic to the primary continuum itself .
furthermore , we can not be certain that the blackbody and power - law components that we use to model the low state spectrum really represent physically distinct components ( such as an accretion disk and corona ) .
it is possible that the reason that both hard and soft components vary together is that they have the same origin , perhaps in something like the hybrid thermal / non - thermal plasma which can successfully fit both the soft excess and power - law components in the spectrum of the black hole x - ray binary cyg x-1 in its soft state @xcite .
finally we note that , since there is no apparent difference between the low state spectral shape and that observed during a low flux epoch in the normal state , then the low state may not be as special as originally thought .
this result ties in well with the discovery ( uttley et al .
, in prep . )
that the appearance of low states in ngc 4051 could be related to the rms - flux relation which is intrinsic to the normal variability process in agn and x - ray binaries @xcite .
because of the rms - flux relation , short term variability amplitude is correlated with long - term flux , so that a large decrease in flux on long time - scales is also accompanied by a corresponding large reduction in short time - scale variability amplitude , producing the distinctive ` low states ' .
we therefore consider it likely that the low states in ngc 4051 are simply prolonged low - flux epochs in the normal life of the x - ray source , i.e. they are caused by the same physical process which produces the flux variability on other time - scales and do not correspond to any distinct transitions in the physical properties of the source .
although the low state of ngc 4051 does not appear to be physically distinct from the normal state of the source , it does allow an excellent opportunity for further toos to target one extreme of the flux - correlated spectral variability in ngc 4051 , enabling further progress in understanding the physical origin of the x - ray spectrum and variability in agn .
we have presented results from a _ chandra _ too observation of the narrow line seyfert 1 ngc 4051 , caught in a 6-week duration low state in 2001 february .
we summarize our main results as follows : 1 .
we confirm the report of @xcite that there is no strong extended soft x - ray emission component in ngc 4051 .
the soft x - ray emission during the low state is dominated by the central point source .
the nuclear x - ray spectrum of ngc 4051 in the low state has an unusual hybrid shape , very soft below @xmath4 kev ( @xmath17 for a simple , power - law fit ) and very hard ( @xmath5 ) at higher energies .
both soft and hard x - ray light curves are variable and well - correlated , the hard x - rays being more variable than the soft .
the hard x - ray variability implies that the hard continuum in the 2001 low state is not due to pure reflection from distant cold gas , but must be the primary continuum itself .
we parameterize the spectrum with a simple black body plus power - law model . the black body model for the soft emission is favoured over a steep power - law if one takes into account the spectral flattening which we observe below 0.5 kev ( but do not fit due to systematic uncertainties at those energies ) , and is also reported by @xcite in the normal flux state .
the continuum at hard energies shows residuals above the power - law model , suggesting the presence of reflection .
however the strength of reflection required , together with an analysis of spectral variability , shows that the deviations from a power - law can not be simply explained by reflection from distant material ( e.g. a molecular torus ) .
an alternate possibility is that the hard residuals are caused by the presence of a strong broad feature , possibly an iron diskline and associated reflection at higher energies .
this latter possibility is favored when we consider _ rxte _ data obtained during the 2001 low state .
the 2001 low state spectrum is consistent with being the same shape as the spectrum observed in a brief ( @xmath6 d ) low - flux epoch during the normal active state of the source , in 2000 april @xcite .
this result suggests that the unusual low state spectrum is not a result of a fundamental physical difference in the emission process between the low and normal states , rather it is the continuation to low fluxes of the normal , flux - dependent spectral variability of the source .
the even fainter 1998 may low state spectrum is also consistent with the same hybrid model as the 2001 february spectrum .
however , as the hard component to the spectrum was a factor @xmath31 fainter in 1998 may than in 2001 february , and no significant variability was detected in that earlier observation , the reflection interpretation of the 1998 may spectrum remains valid .
we would like to thank jon miller and the anonymous referee for helpful comments .
p. u. would like to thank the staff of the _ chandra _ x - ray center and the cfa for their hospitality and support while preparing this paper .
this work has been supported by the nasa grant go1 - 2111x .
collinge , m. j. , et al .
2001 , , 557 , 2 elvis , m. , lockman , f. j. , & wilkes , b. j. , 1989 , , 97 , 777 fabian , a. c. , & vaughan , s. , 2003 , , in press ( astro - ph/0301588 ) gierlinski , m. , zdziarski , a. a. , poutanen , j. , coppi , p. s. , ebisawa , k. , & johnson , w. n. 1999 , , 309 , 496 green , a. r. , m@xmath9hardy , i. m. , & lehto , h. j. 1993 , , 265 , 664 guainazzi , m. , mihara , t. , otani , c. , & matsuoka , m. 1996 , , 48 , 781 guainazzi , m. , et al .
1998 , , 301 , l1 haardt , f. , maraschi , l. , ghisellini , g. 1997 , , 476 , 620 lamer , g. , uttley , p. , & m@xmath9hardy , i. m. 2000 , , 319 , 949 lamer , g. , m@xmath9hardy , i. m. , uttley , p. , & jahoda , k. 2003 , , 338 , 323 laor , a. 1991 , , 376 , 90 magdziarz , p. , & zdziarski , a. a. 1995 , , 273 , 837 nandra , k. , george , i. m. , mushotzky , r. f. , turner , t. j. , & yaqoob , t. 1997 , , 476 , 70 peterson , b. m. , et al .
2000 , , 542 , 161 singh , k. p. , 1999
, , 309 , 991 taylor , r. d. , uttley , p. , m@xmath9hardy , i. m. 2003 , , in press ( astro - ph/0304523 ) uttley , p. , m@xmath9hardy , i. m. , papadakis , i. e. , guainazzi , m. , & fruscione , a. 1999 , , 307 , l6 uttley , p. , m@xmath9hardy , i. m. , papadakis , i. e. , cagnoni , i. , & fruscione , a. 2000a , , 312 , 880 uttley , p. 2000b
thesis , university of southampton uttley , p. , & m@xmath9hardy , i. m. , 2001 , , 323 , l26 vaughan , s. , & edelson , r. 2001 , , 548 , 694 wilms , j. , reynolds , c. s. , begelman , m. c. , reeves , j. , molendi , s. , staubert , r. , & kendziorra , e. 2001 , , 328 , l27 zdziarski , a. a. , lubinski , p. , gilfanov , m. , & revnivtsev , m. , 2003 , in press ( astro - ph/0209363 ) lcccccccc 2001 feb & 0.63 & 0.14 & 4.2 & 0.99 & 52.4 & 2.5 & 7.0 & 662/424 + _ chandra _ 2000 apr & 0.77 & 0.14 & 3.6 & 0.99 & 43.3 & 2.2 & 6.0 & 132/86 + _ bepposax _ 1998 may & 0.52 & 0.14 & 1.3 & 0.99 & 9.1 & 0.7 & 1.4 & 108/60 + | we report the results from a _ chandra _ too observation of the low - luminosity narrow line seyfert 1 galaxy ngc 4051 , obtained during a 6-week duration low - flux state in 2001 february . during the _ chandra _ observation , the 2 - 10 kev source flux
was @xmath0 erg @xmath1 s@xmath2 , corresponding to a 2 - 10 kev luminosity of @xmath3 erg s@xmath2 .
we confirm the absence of strong extended soft x - ray emission in ngc 4051 , and that the low state spectrum is dominated by the central point source .
the x - ray spectrum has an unusual , hybrid shape , very soft below @xmath4 kev ( and which we model with a black body of temperature 0.14 kev ) and very hard at higher energies ( power - law slope of @xmath5 ) .
the lightcurves in both soft and hard bands are significantly variable and correlated , implying a connection between hard and soft components and proving that the hard component is dominated by primary continuum emission and is not due to pure reflection from distant cold material .
however , a comparison with _ rxte _ data obtained during the same 2001 low state suggests the presence of unusually prominent disk reflection features , which may help to explain the apparent upwards curvature and extreme hardness of the _ chandra _ spectrum above a few kev .
the shape of the 2001 low state spectrum is consistent with that observed in a brief ( @xmath6 d ) low flux excursion during the source s normal state in april 2000 , suggesting that emission processes during the low state are not significantly different to those in the normal state .
the unusual spectral shape observed in the low state may be a continuation to low fluxes of the normal , flux - dependent spectral variability of the source . |
in the present day context the physical properties and internal structure of the compact stars are topics of active research .
itoh @xcite first suggested that the quark stars may exist in hydrostatic equilibrium .
according to bodmer @xcite the quark matter made of @xmath2 , @xmath3 and @xmath4 quarks is more stable than any ordinary nuclear matter .
cheng et al .
@xcite argued that after a supernova explosion a massive star core collapses to a strange star .
it was also proposed by alford @xcite that in the dense core of a neutron star there is sufficiently high density and low temperature to crush the hadrons into quark matter .
it is quite interesting that the equation of state ( eos ) choosen by the following authors in refs .
@xcite for neutron stars did not explain the properties of the observed compact stars like @xmath5 , @xmath6 , @xmath7 , @xmath8 , @xmath9 , whereas the strange matter eos can explain them
. in the interior of such stars there may exist many exotic phases .
however , in the present work our investigation is mostly restricted to the eos of strange matter only .
rahaman et al .
@xcite using the mit bag model proposed a new model for strange stars where they analysed the physical properties of the stars from 6 km to the surface . here
we choose phenomenological mit bag model @xcite , according to which , the eos of strange matter can be written in a linear form as following @xmath10 where @xmath11 and @xmath12 are the energy density , the radial pressure and the bag constant respectively . in the bag model
the vacuum pressure @xmath12 confirms the quark confinement that equilibrates the pressure of quarks to stabilize the system . in this paper using mit bag model
we have proposed a new stable model for strange stars , which is neutral , static as well as anisotropic fluid spheres valid under the appropriate physical condition . using the observed value of the mass of strange stars we have analysed their physical properties and also predicted respective radius of the stars .
we have organized our study as follows : in sec .
ii basic equations and the solutions for different physical parameters are provided .
we have discussed various physical features of the strange stars in sec .
iii whereas provide data sets for the strange stars along with a comparative study in the sec .
the article is concluded with a short discussion in sec . v.
we consider that the interior of the strange stars is well described by the following space time metric @xmath13 where @xmath14 and @xmath15 are functions of the radial coordinate @xmath16 only .
the energy - momentum tensor for the proposed strange star model is given by @xmath17 @xmath18 @xmath19 where @xmath20 , @xmath21 and @xmath22 in eqs .
( [ eq3 ] ) , ( [ eq4 ] ) and ( [ eq5 ] ) represents the energy density , radial and tangential pressures respectively for the fluid sphere .
the einstein field equations are now can be listed below : @xmath23 @xmath24 @xmath25 let us define the mass function @xmath26 of the star as @xmath27 now assuming that the matter density profile inside the quark star can described as proposed by mak and harko @xcite @xmath28,\label{eq10}\ ] ] where @xmath29 and @xmath30 are the central and surface densities of the star of radius @xmath31 respectively @xmath16 being the radial coordinate .
( [ eq9 ] ) and eq .
( [ eq10 ] ) we find the total mass of the quark star as @xmath32 at the surface the radial pressure must be zero and hence from eq .
( [ eq1 ] ) we have @xmath33 with the choice of eos ( [ eq1 ] ) and using eqs .
( [ eq2 ] ) , ( [ eq6 ] ) - ( [ eq10 ] ) and ( [ eq11a ] ) we obtain the following physical parameters : @xmath34 @xmath35,\notag \\ & \qquad\label{eq13}\end{aligned}\ ] ] as a function of the radial coordinate @xmath16 for the strange star @xmath9 ] as a function of the radial coordinate @xmath16 for the strange star @xmath9 ] @xmath37,\label{eq14}\ ] ] @xmath38},\label{eq15}\ ] ] for the strange star @xmath9 ] for the strange star @xmath9 ] @xmath39{r}^{2}}{36{r}^{2}[{r}^{2 } \left ( \frac{3}{5}\,c_{{1}}{r}^{2 } -\rho_{{c}}{r}^{2 } \right ) \pi + \frac{3}{8}\,{r}^{2}]}},\label{eq16}\ ] ] where @xmath40 , @xmath41 , @xmath42 , @xmath43 , @xmath44 , @xmath45 , @xmath46\\ \hspace*{4cm}-{\frac { 16}{15}}\,{r}^{2 } \left ( \frac{2}{3 } \rho_{{c}}+\rho_{{0 } } \right ) \pi + \frac{2}{3}$ ] , @xmath47 , @xmath48 , @xmath49 , @xmath50{{r}^{2}}$ ] , @xmath51 , @xmath52 .
the behaviour of the above physical parameters are shown in figs . 1 - 5 which are quite satisfactory in their nature .
now let us maximize the anisotropic stress at the surface ( @xmath53 ) without assuming the nature of extremum .
we obtain from eq .
( [ eq16 ] ) as follows @xmath54 . \notag \\ & \qquad \label{eq17}\end{aligned}\ ] ] the above second order differential equation ( [ eq16 ] ) related to anisotropy takes the form on the surface of the system ( i.e. at @xmath53 ) as follows : @xmath55 . \notag\\ & \qquad\hspace{-12 cm } \label{eqa}\end{aligned}\ ] ] now using eqs .
( [ eq11 ] ) , ( [ eq11a ] ) and ( [ eq17 ] ) and taking the value of bag constant as 83 @xmath56 @xcite , we get several solutions for @xmath31 .
however , we examine that among all these solutions only one solution is physically acceptable and also consistent with the buchdahl condition @xcite . for the strange star @xmath9 ]
in this section we shall discuss different physical parameters of the model . according to tolman @xcite , oppenheimer and volkoff @xcite
, the sum of forces should be equal to zero so that the system is subjected to the equilibrium , @xmath57 where @xmath58 , @xmath59 and @xmath60 are the gravitational , hydrostatic and anisotropic forces respectively .
the generalized tov equation @xcite in its explicit form is @xmath61 where @xmath62 is the the effective gravitational mass of the system whcih is defined as @xmath63 where for our system @xmath64\nu_{{r}}$ ] , @xmath65 , and @xmath66 . for the strange star @xmath9 ] we have drawn plot for the above tov equation in fig .
[ fig6 ] . from this
it is clear that our proposed model has achieved equilibrium state under the combined effect of the forces . to satisfy the energy conditions i.e null energy condition ( nec ) , weak energy condition ( wec ) and strong energy condition ( sec ) for an anisotropic fluid sphere composed of strange matter the following inequalities have to be hold simultaneously : @xmath67 @xmath68 @xmath69 the above energy conditions as drawn in fig .
[ fig7 ] shows that our proposed model for the strange stars satisfies all the energy conditions . for the strange star @xmath9 ] using the concept of herrera s cracking ( or overturning ) @xcite we shall examine the stability of our system .
the condition of causality establishes the physical acceptability of a fluid distribution , which demands @xmath70 and @xmath71 . also according to herrera @xcite and andrasson @xcite for the stability of the matter distribution another condition required is @xmath72 and this suggests that ` no cracking '
i.e potentially stable region must be there . from fig .
[ fig8 ] we observe that our system satisfies all of these conditions and thus provides a stable system . for the strange star @xmath9,title="fig : " ] for the strange star @xmath9,title="fig : " ] the maximum allowable mass - radius ratio for the above proposed anisotropic fluid sphere can be calculated following buchdahl @xcite .
he has proposed that the maximum limit of mass - radius ratio for static spherically symmetric perfect fluid star should satisfy the following upper bound @xmath73 .
however , mak and harko @xcite have given the generalized expression for the same mass - radius ratio .
the total mass of the anisotropic compact star which is defined in eq .
( 11 ) can be considered as the maximum mass @xmath74 . from this mass - radius relation
we find out the values of different compact star candidates in table ii and observe that all the values fall within the acceptable range as specified by buchdahl @xcite .
the maximum mass and radius feature is shown in fig . 9 . ] the compactification factor ( @xmath2 ) for the proposed strange star model can be written as @xmath75 so the corresponding surface redshift ( @xmath76 ) as we get from the compactness becomes @xmath77^{- \frac{1}{2}}-1={\frac { 1}{\sqrt { 1 - 8\,\pi \,{r}^{2 } \left ( \frac{2}{15}\rho_{{c } } + \frac{1}{5}\rho_{{0 } } \right ) } } } -1.\label{eq25}\ ] ] the variation of the compactness and redshift with the radial distance are shown in the fig .
[ fig10 ] . for the strange star @xmath9 ]
in this paper we have derived radius of the different strange star candidates using their observational mass and bag constant ( as 83 @xmath56 ) , shown in table [ table1 ] . from figs .
[ fig1 ] - [ fig3 ] it is clear that our solution is free from geometrical and physical singularities .
also it can be observed that the metric potetials @xmath78 and @xmath79 have finite positive values in the range @xmath80 .
the values of the central density , central pressure , buchdahl condition and surface redshift for the different strange star candidates of table [ table1 ] are shown in table [ table2 ] .
it is shown from our model that the surface density of the star @xmath9 is @xmath81 which is very high and consistent with ultra compact stars @xcite .
we also have found out that the redshift of the stars in table [ table2 ] are within the range @xmath82 which is consistent under the constraints @xmath83 and @xmath84 as suggested by the investigators respectively in the following refs .
@xcite and @xcite . from eq .
( [ eq16 ] ) and fig .
[ fig5 ] one can find that the anisotropic force @xmath85 is positive throughout the system , i.e. @xmath86 and hence the direction of the anisotropic force is outward for our system .
the equation @xmath87 suggests that the star has maximum value of radius @xmath88 for @xmath89 . from the equation @xmath90
, the star has maximum value of mass for @xmath91 .
the model yields the values of the maximum mass and maximum central density as @xmath92 and @xmath93 @xmath94 respectively .
.physical parameters of the observed strange stars [ cols= " < , < , < , < , < , < " , ] interestingly , in the present study we find that the anisotropy of the compact star is zero at the center and then it is increasing through out the interior region of the star and reach it s maximum value at the surface . again from eqn .
( [ eqa ] ) it is clear that the @xmath95 is negative , which supports mathematically that the anisotropy is also maximum on the surface of an anisotropic compact star . as a comparative study ,
the maximum anisotropy of few quark stars are given in table [ table3 ] ( the values are calculated from our model ) .
we would also like to mention that the maximum allowable mass to radius ratio is 0.40 as evident from table ii for @xmath9 .
table ii also indicates that all the other values , including this value , lie within the acceptable range as predicted by buchdahl @xcite , i.e. @xmath96 . hence , as a final comment , from the above discussions it can be concluded that the model proposed in this work seems suitable to study the ultra - dense compact strange stars .
sr and fr are thankful to the inter - university centre for astronomy and astrophysics ( iucaa ) , pune , india for providing visiting associateship under which a part of this work was carried out .
sr is also thankful to the authority of the institute of mathematical sciences , chennai , india for providing all types of working facilty and hospitality under the associateship scheme . | we attempt to find a singularity free interior solution for a neutral and static stellar model .
we consider that ( i ) the star is made up of anisotropic fluid and ( ii ) the mit bag model can be used .
the total system is defined by assuming the density profile given by mak and harko @xcite , which satisfies all the physical conditions of a stellar system and is stable by nature .
we find that those stellar systems which obey such a non - linear density function must have maximum anisotropy at the surface .
we also perform several tests for physical features of the proposed model and show that these are mostly acceptable within certain range . as a special mention , from our investigation
we find that the maximum mass and radius of the quark star are @xmath0 and @xmath1 respectively . |
current theories @xcite suggest that leptoquarks would come in three different generations corresponding to the three quark and lepton generations .
leptoquarks would have color , fractional electric charge , both lepton and baryon numbers , and could be scalars or vectors . at the tevatron leptoquarks could appear in pairs through @xmath2 annihilation ( dominates for @xmath3 gev ) or @xmath4 fusion , @xmath5 or through the associated lepton production @xmath6 with the contribution of the last one being small .
the pair production cross section for scalar leptoquarks only depends on the strong coupling constant and on the leptoquark mass .
the vector leptoquark pair production cross section also depends on anomalous @xmath7 couplings @xmath8 and @xmath9 , and the experimental constrains are generally given for three models : `` minimal coupling '' ( @xmath10 , @xmath11 ) , `` yang mills '' ( @xmath12 , @xmath11 ) , and `` minus minus '' ( @xmath13 , @xmath14 )
. leptoquarks decay into a charged lepton and a quark with a branching fraction @xmath15 or into a neutrino and a quark with a branching fraction ( @xmath16 ) .
thus frations of a leptoquark pairs into the @xmath17 , @xmath18 and @xmath19 final states are @xmath20 , @xmath21 and @xmath22 respectively .
this report presents a summary searches for leptoquark pair production in the data sets collected with the d0 detector @xcite during run ii ( started march 2001 ) of the fermilab tevatron collider .
a search explored the final state where both leptoquarks decay into a neutrino and quarks assuming @xmath23 : @xmath24 .
the corresponding detector signature is two acoplnar jets accompanied by missing energy .
no selection specific for jet flavor were applied .
thus the search results applied to all three generations of leptoquarks .
the analysis used 2.5 fb@xmath25 of d0 run ii data .
events were recorded using triggers requiring two acoplanar jets and large missing transverse energy , @xmath26 , or large @xmath27 , the vector sum of the jets transverse energy @xmath28 .
the two leading jets were required to be in the central region @xmath29 of the d0 detector and have transverse momenta greater then 35 gev .
the multijet qcd background was suppressed by requiring @xmath26 to be greater then 75 gev and with cuts on angular correlations between jets and @xmath26 directions : the azimutal angle between @xmath26 and the first jet and the minimal and maximal angles between any jets and @xmath26 . to suppress the dominant standard model ( sm ) background from @xmath30 events a veto on events containing an isolated electron or muon with @xmath31 gev
was applied , and events with an isolated track @xmath32 gev were also rejected .
the two final cuts on @xmath26 and @xmath33 were optimized for different signals by minimizing the expected upper limit on the cross section in the absence of signal table [ tab : lq_allgen ] shows the number of data , background and signal events after all selections for @xmath34 gev and @xmath35 gev leptoquark signals .
no significant excess of data over the predicted backround was found .
figure [ fig : lq_allgen ] shows the observed and expected 95% c.l .
limits on scalar leptoquark pair production cross sections as a function of the lq mass .
the observed and expected lower lq mass limit of 214 gev and 222 gev respectively were obtained at the intersection of the experimental cross section limits with the nominal theoretical production cross section calculated for the factorization and renormalization scale @xmath36 .
.@xmath24 analysis .
number of data , background and signal events after all selections for different lq signals . [ cols="^,^,^,^,^ " , ] [ tab : lq3_nub ] ( points plus solid line ) as a function of @xmath37 for the pair production of third generation leptoquarks .
the expected cross section limits are the same as the observed .
the grey shaded band shows the pdf and the renormalization scale error bounds on the nlo production cross section ( solid line ) calculated for @xmath36 .
the long dashed line shows @xmath38 for the @xmath39 = @xmath40 times the phase suppression factor for the @xmath41 channel.,width=302 ]
searches for pair production of leptoquarks of all three generations were performed in 14 fb@xmath25 d0 data samples . all presented analysis
are in good agreement with the sm predictions .
no leptoquarks signals were observed , and a set of 95% c.l .
limits on lq masses have been obtained improving previous tevatron results .
more details on the presented analyses can be found at @xcite . | we report on d0 searches for leptoquarks ( lq ) predicted in extended gauge theories and composite models to explain the symmetry between quarks and leptons .
data samples obtained with the d0 detector from @xmath0 collisions at a center - of - mass energy of 1.96 tev corresponding to intergrated luminosities of 14 fb@xmath1 were analyzed .
no evidence for the production of such particles were observed and lower limits on leptoquark masses are set . |
pion - nucleon scattering is one of the main sources of information for the baryon spectrum . in addition
, it also plays a fundamental role in the description of nuclear dynamics for which the @xmath0 off - shell amplitude serves as the basic input to most of the existing nuclear calculations at intermediate energies .
knowledge about the off - shell @xmath0 amplitude is also essential in interpreting the experiments performed at the intermediate - energy electron accelerators in order to unravel the internal structure of these hadrons @xcite .
it is hence important to have a sound theoretical description of the @xmath0 interaction .
it is commonly accepted that quantum chromodynamics ( qcd ) is the fundamental theory of the strong interaction . however , due to the confinement problem , it is still practically impossible to derive the @xmath0 interaction directly from qcd . on the other hand ,
models based on meson - exchange ( mex ) pictures have been very successful in describing the @xmath10 scattering . over the past decade , mex approach has also been applied by several groups @xcite to construct models for @xmath0 scattering . in previous works we have constructed several mex @xmath0 models within the three - dimensional reduction scheme of the bethe - salpeter equation @xcite and investigated their sensitivity with respect to various three - dimensional reduction schemes .
the model lagrangian included only @xmath11 and @xmath6 fields .
it was found that all the resulting meson - exchange models can yield similarly good descriptions of @xmath0 scattering data up to @xmath12 mev . the model obtained with the cooper - jennings reduction scheme @xcite
was recently extended up to a c.m .
energy of 2 gev in the @xmath13 channel by including the @xmath7 channel and a set of higher @xmath13 resonances @xcite .
the effects of the other @xmath14 channels like the @xmath15 , @xmath16 , and @xmath17 , instead of including them directly in the coupled - channels calculation , were taken into account by introducing a phenomenological term in the resonance propagators .
an excellent fit to the t - matrix in both @xmath0 and @xmath7 channels was obtained . here
we further extend the model to other higher partial waves up to the @xmath8waves .
the mex @xmath0 model we previously constructed was obtained by using a three - dimensional reduction scheme of the bethe - salpeter equation for a model lagrangian involving @xmath18 and @xmath6 fields .
details can be found in ref .
@xcite . as the energy increases , two - pion channels like @xmath19 as well as a non - resonant continuum of @xmath20 states become increasingly important , and at the same time more and more nucleon resonances appear as intermediate states . in ref .
@xcite the @xmath0 model constructed in @xcite was extended for the @xmath13 partial wave by explicitly coupling the @xmath1 , @xmath2 and @xmath21 channels and including the couplings with higher baryon resonances .
for example ,
in the case when there is only one resonance @xmath22 contributing , the hilbert space was enlarged to include a bare @xmath13 resonance @xmath22 which acquires a width by its coupling with the @xmath0 and @xmath7 channels through the lagrangian @xmath23 where @xmath24 and @xmath2 denote the field operators for the nucleon , bare @xmath22 , pion and eta meson , respectively . then the full @xmath25-matrix can be written as a system of coupled equations , @xmath26 where @xmath27 and @xmath28 denote the @xmath1 , or
@xmath2 channel and @xmath29 is the total center mass energy . in general
, the potential @xmath30 is a sum of non - resonant @xmath31 and bare resonance @xmath32 terms , @xmath33 the non - resonant term @xmath34 for the @xmath0 elastic channel contains contributions from the @xmath35- and @xmath36-channel , pseudovector born terms and @xmath25-channel contributions with @xmath37 , @xmath5 , and @xmath6 exchange .
the parameters in @xmath34 are fixed from the analysis of the pion scattering phase shifts for the @xmath38 and @xmath39waves at low energies ( @xmath40 mev ) @xcite . in channels involving @xmath2 ,
@xmath41 is taken to be zero since the @xmath42 coupling is very small .
the bare resonance contribution @xmath43 where @xmath44 and @xmath45 denote the bare vertex operator for @xmath46 and bare mass of @xmath22 , respectively , arises from the excitation and de - excitation of the resonance @xmath22 .
the matrix elements of the potential @xmath47 can be symbolically expressed in the form @xmath48 where @xmath49 and @xmath50 are the pion ( or eta ) momenta in the initial and final states , and @xmath51 is the resonance vertex couplings . as in @xcite
, we associate with each external line of the particle @xmath52 in a feynman diagram a covariant form factor @xmath53^{n_\alpha}$ ] , where @xmath54 , @xmath55 , and @xmath56 are the four - momentum , mass , and cut - off parameter of particle @xmath52 , respectively , and @xmath57 . as a result , @xmath58 depends on the product of three cut - off parameters . in eq .
( [ v_r1 ] ) we have included a phenomenological term @xmath59 in the resonance propagator to account for the @xmath14 decay channel .
therefore , our `` bare '' resonance propagator already contains some renormalization or `` dressing '' effects due to the coupling with the @xmath14 channel . with this prescription
we assume that any further non - resonant coupling mechanisms with the @xmath14 channel are small .
the form of @xmath59 can be found in @xcite and is characterized by two parameters , a cut - off @xmath60 and the @xmath9 decay width at the resonance @xmath61 .
consequently , one isolated resonance will contain five free parameters , @xmath62 and @xmath63 .
the generalization of the coupled channels model to the case of @xmath3 resonances with the same quantum numbers is then given by @xmath64 with free parameters for the bare masses , @xmath9 widths , coupling constants , and cut - off parameters for each resonance .
after solving the coupled channel equations , the next task is the extraction of the physical ( or `` dressed '' ) masses , partial widths , and branching ratios of the resonances .
it is well - known this procedure is definitely model dependent , because the background and the resonance contributions can not be separated in a unique way . in this work
, we employ the procedure used in ref .
@xcite where , in the case of pion - nucleon elastic scattering with only one resonance contributing , the full t - matrix is written as follows , @xmath65 where @xmath66 the `` background '' @xmath67 includes contributions not only from the background rescattering but also from intermediate resonance excitation .
this is compensated by the fact that the resonance contribution @xmath68 now contains only the terms that start with the bare resonance vertex . in terms of self - energy and
vertex functions , one obtains the result @xcite @xmath69 where @xmath70 @xmath71 describes the dressed vertex of @xmath72 @xcite .
@xmath73 is the self - energy of the dressed @xmath22 arising from one - pion intermediate states and @xmath74 with @xmath75 .
the information about the physical mass and the total width of the resonance @xmath22 are contained in the dressed resonance propagator given in eq .
( [ tr_dmt2 ] ) .
the complex self - energy @xmath76 leads to a shift from the real `` bare '' mass to a complex and energy - dependent value .
we characterize the resonance by energy - independent parameter that is obtained by solving the equation @xmath77 the solution of this equation , @xmath78 , corresponds to the energy at which the dressed propagator in eq .
( [ tr_dmt2 ] ) is purely imaginary . the `` physical '' or `` dressed '' mass and the width of the resonance is then defined by , @xmath79 when there are n resonances contributing in the same channel , eq .
( [ dmt ] ) can be generalized to take the form of @xmath80 the contribution from each resonance @xmath81 can be expressed in terms of the bare @xmath82 and dressed @xmath83 vertex operators as well as the resonance self energy derived from one - pion @xmath84 and two - pion @xmath85 channels , that is @xmath86 where @xmath87 is the bare mass of @xmath81 .
the vertices for resonance excitation are obtained , in analogous to eqs .
( [ dmt]-[sigma_delta ] ) , from the following two equations : @xmath88 where @xmath89 .
the one - pion self - energies corresponding to eq .
( [ sigma_delta ] ) , is @xmath90 .
we wish to emphasize that in the formulation we present above , namely in eqs .
( [ t_pin]-[dmt_bcr ] ) , the n resonances are treated in a completely symmetrical way and the self - energy and the dressing of any resonance receive contributions from all other resonances .
in fig . 1 , we compare our results for the real and imaginary parts of the t - matrix in some selected channels in @xmath91 and @xmath8waves up to 2 gev c.m .
energy with the experimental data as obtained in the said partial wave analysis @xcite .
one sees that we are able to describe the data very well .
the bare and physical resonance masses , and widths extracted according to eq .
( [ mass - width ] ) are presented in table 1 . even though our model describes the data for t - matrix well as seen in fig .
1 , the resonance properties we extract as given in table 1 do show several differences when compared with pdg values @xcite .
the most notable ones are that ( 1 ) .
we require two resonances not listed in pdg : @xmath92 and @xmath93 ; ( 2 ) .
the masses and widths we obtain for the 2nd and 3rd resonances in @xmath94 and @xmath95 deviates substantially from the pdg values .
the pdg values for @xmath96 for these resonances are @xmath97 and @xmath98 ; ( 3 ) .
the width we obtain for @xmath99 is only 58 mev which is much smaller that the pdg value of @xmath100 mev .
.bare @xmath45 and physical @xmath101 resonance masses and total width in mev . [ cols="<,^,^,^,^,^,^,^,^,^",options="header " , ]
we have extended our previously constructed meson - exchange @xmath102 model to energies up to 2 gev by including the @xmath7 channel and all the four stars @xmath0 resonances up to the @xmath8waves .
the effects of other @xmath9 channels are taken into account phenomenologically .
we have treated , in any given channel , all the contributing resonances in a completely symmetrical manner such that every resonance is dressed by the presence of all other resonances .
the extended model gives an excellent fit to both @xmath102 phase shifts and inelasticity parameters in all channels up to the @xmath8waves .
however , a few of the resonance parameters differ substantially from the pdg values .
this @xmath0 model will be applied to evaluate the contribution of the pion cloud to the photopion reactions up to 2 gev c.m .
energy as was done in ref .
@xcite so that the photoexcitation strengths of all resonances below 2 gev can be reliably extracted . | a meson - exchange @xmath0 model , previously constructed using three - dimensional reduction scheme of the bethe - salpeter equation for a model lagrangian involving @xmath1 , @xmath2 , @xmath3 , @xmath4 , @xmath5 , and @xmath6 fields , is extended to energies up to 2 gev by including the @xmath7 channel and all the four stars @xmath0 resonances up to the @xmath8waves .
the effects of other @xmath9 channels are taken into account phenomenologically .
the extended model gives an excellent fit to both @xmath0 phase shifts and inelasticity parameters in all channels up to the @xmath8waves .
however , a few of the extracted resonance parameters differ considerably from the pdg values . |
in the theory of superconductivity , the quantity which characterizes the nature of pairing is the gap function . more precisely , it is the amplitude for pairing @xmath3 for a cooper pair of electrons in states @xmath4 in a singlet spin configuration .
this quantity is closely related to the energy gap in the single - particle excitation spectrum of a superconductor , and has a direct bearing on many of the experimentally measurable quantities . as an example , consider the standard bcs @xmath5-wave superconductor @xcite , which has an isotropic gap function which is _ even _ across the fermi surface ( fs ) , non - zero only in a thin shell of width @xmath6 around @xmath7 .
the energy gap @xmath8 is related to the gap function ( 1 ) through the relation @xmath9 and obeys the celebrated gap equation @xmath10 where @xmath11 is the quasiparticle energy , @xmath12 is the single - particle energy with respect to @xmath7 , and @xmath13 is the effective electron - electron interaction matrix element in the reduced bcs hamiltonian .
the energy gap can be directly observed , for example , through superconductor - normal metal or giaver tunneling . to understand the significance of the gap function ( [ gapfn ] ) ,
let us examine the behaviour of the bcs wavefunction close to the fs : @xmath14 which can be rewritten as @xmath15 where the product is now over only half of the @xmath16-space ( e.g. @xmath17 and all @xmath18 in 2d ) .
here we have @xmath19 as the singlet cooper pair creation operator at points @xmath16 and @xmath20 in @xmath16-space .
it is clear that @xmath21 is the _ probability amplitude _ of finding no singlet pair with momentum @xmath22 , @xmath23 is that of finding one singlet pair ( of charge @xmath24 ) , and @xmath25 that of finding two singlet pairs ( of total charge @xmath26 )
. the bcs state has identical phase relations for various configurations of pair occupancy in @xmath16-space .
that is , when the product in eq .
( [ bcswf ] ) is expanded out , the resulting sum has identical phase for all terms , each term corresponding to different configurations of the @xmath22 occupancy .
superconductivity can thus be thought of as a coherent charge-@xmath24 fluctuating state in @xmath16-space .
since @xmath27 is non - zero only in a thin energy shell around the fs , the coherent @xmath24 charge fluctuation is concentrated around the fs .
( note that this coherence in @xmath16-space results in phase coherence among the cooper pairs in real space also ) .
away from the shell , we either have a completely filled band ( inside the fs ) or a completely empty band ( outside the fs ) and hence no charge fluctuations .
the gap function ( [ gapfn ] ) , for the bcs wavefunction ( [ bcswf ] ) , reduces to @xmath28 and is thus a measure of coherent charge fluctuations in @xmath16-space . in the context of the high-@xmath0 cuprates
, it is generally believed that the gap function is not of simple bcs type ( @xmath5-wave and even across @xmath29 ) .
some of the recent experiments have suggested that the gap is highly anisotropic , possibly with the @xmath2 symmetry .
@xcite yet another proposal is that of odd - gap pairing , which is a favoured mode of pairing in the presence of a strong , purely repulsive interaction .
@xcite a recent proposal in the context of the interlayer tunneling mechanism is that of anisotropic @xmath5-wave pairing , which has deep minima but no actual nodes on the fs .
@xcite however , in spite of intense activity , both on theoretical and experimental sides , the issue of the symmetry of the gap function still remains unresolved .
the importance of this issue can not be overestimated , since the gap function is a quantity which , on the one hand , has direct experimental consequences , and on the other , may offer a unique signature pointing to the right kind of mechanism for high-@xmath0 superconductivity .
the aim of the present work is to investigate the behaviour of @xmath27 using exact diagonalization , especially with reference to its nodal structure .
we would like to find out the most likely pairing symmetries , both angular and radial , for clusters consisting of single or coupled chains and planes .
in particular , we are interested in the @xmath2 symmetry and odd - gap pairing . in
what follows , we first present a brief summary of the various proposed gap functions , especially @xmath2 and odd - gap pairing ( section ii ) .
we then present our numerical procedure in section iii and the results in section iv .
finally , conclusions are presented in section v.
in order to see , on general grounds , why gap functions with zeros are preferable as candidates for pairing in high-@xmath0 superconductors , let us consider the one - band large-@xmath30 hubbard model or the @xmath1 model .
these models are believed to contain the essential low - energy physics of high-@xmath0 superconductivity .
let us begin by noting that in the large-@xmath30 hubbard model , the on - site pairing amplitude @xmath31 is diminished considerably on account of the restriction on double occupancy . in the @xmath1 model , of course , double occupancy is completely projected out .
thus @xmath32 , which reduces , in @xmath16-space , to @xmath33 where we have used the pairing condition @xmath34 .
this is a constraint on the gap function , arising from a strong real - space repulsion , and implies the presence of zeros in the gap function . the constraint , however , is a global constraint , and can be satisfied in several ways .
one possibility is that of odd - gap pairing , in which case the gap function vanishes on the _ entire _ fermi surface , and changes sign across the fs .
yet another possibility is that of a simple @xmath2 superconductor on a 2d square lattice , for which the gap function vanishes along the lines @xmath35 which intersect the fs at exactly four points .
we give below a brief account of these two and some of the other proposed pairing states .
the possibility of odd - gap pairing , in the presence of a dominantly repulsive interaction , was anticipated by cohen @xcite many years ago . the case for odd - gap pairing , in the context of high-@xmath0 superconductivity , was put forward by mila and abrahams @xcite , within the framework of the weak coupling bcs theory . they have shown that an _ even _ solution to the bcs gap equation ( [ gapeq ] ) exists only so long as the effective electron - electron interaction is dominantly attractive , albeit weak .
this is the case with the standard bcs model of conventional superconductivity , where the weak attraction arises due to electron - phonon coupling . in the presence of a repulsive interaction , however , such a solution ceases to exist when the repulsive part of the interaction starts dominating , and an unusual solution is shown to exist which is _ odd _ across the fs .
for example , in the toy models considered in @xcite as well as @xcite , the energy gap is proportional to @xmath36 , which is clearly odd across the fs .
moreover , an odd gap is independent of the _ magnitude _ of the repulsion , provided that it is the repulsion which dominates .
this condition is met in the large-@xmath30 hubbard model , and odd - gap pairing is thus a candidate for pairing in the high-@xmath0 superconductors .
mila and abrahams have further shown that certain key features of the tunneling density of states in the cuprates , such as the temperature - insensitive peak , can be naturally explained on the basis of odd - gap pairing .
this feature is in contrast with @xmath5-wave as well as @xmath2 pairing , in which the peak shifts towards @xmath7 as @xmath37 .
anderson @xcite has also pointed out in the context of the rvb theory that the gap function vanishes on , and changes sign across , the fs .
for example , in the rvb mean field theory at half filling , @xcite the gap function is constant and changes sign across the pseudo fs defined by @xmath38 and satisfies the constraint ( [ gapsum ] ) .
for the doped case , however , the analysis is difficult and anderson did not provide any argument as to why the gap should vanish on the fs away from half - filling .
@xcite in the latter case , it is argued that strong correlations in real space lead to a suppression of coherent charge fluctuations close to the fs , @xcite in the sense of eq .
( [ bcswf ] ) , as follows .
strongly correlated electrons in 1d , described by a large-@xmath30 hubbard model , have certain unique features close to the fs .
it is well known that in the 1d hubbard model there is singular forward scattering between two electrons with opposite spins close to the fs .
this leads to a finite phase shift @xcite at the fs and the consequent failure of the fermi liquid theory , resulting in the vanishing of the discontinuity in @xmath39 at the fs ( luttinger liquid behaviour ) .
it also implies an effective hard - core repulsive pseudopotential between electrons with opposite spins close to the fs .
thus no two electrons close to the fs , with opposite spins , can have the same momentum , thereby making @xmath16-points close to the fs essentially singly occupied .
coherent pair fluctuations are thus unlikely to develop on or very close to the fs , but are not forbidden away from the fs .
it is therefore likely that the line of zeros of @xmath27 implied by the global constraint ( eq .
( [ gapsum ] ) ) will coincide with the fs . whether these arguments carry over to 2d or not is still unclear .
it is believed that if the constraint of no double occupancy in real space leads to singular forward scattering and the consequent failure of fermi liquid theory in 2d ( in the spirit of anderson s tomographic luttinger liquid picture @xcite ) , a gap function which vanishes on the entire fs is a natural candidate for pairing .
odd - gap pairing has so far found very feeble experimental support
. however , a very recent work @xcite shows arpes data in bi2212 cuprates in which the gap _ does not vanish _ along @xmath40 , but vanishes on either side of this line .
this can be interpreted as a possible indication that the nodal lines in the gap coincide with the fs .
the @xmath2 paired state is characterised by a gap function of the form @xmath41 support to @xmath2 pairing comes from numerous sources .
theoretical analysis of motion of holes in an antiferromagnetic background shows that exchange of spin fluctuations can induce singlet pairing of @xmath2 symmetry , for which hole - hole interaction is attractive .
@xcite although antiferromagnetic fluctuations are rather subdued in the doped case in comparison with half - filling , it is reasonable to expect @xmath2 pairing as a possibility . on the experimental side ,
numerous experiments have been performed recently , which seem to support @xmath2 pairing .
london penetration depth data from hardy _ et al . _
@xcite shows a low - temperature linear temperature dependence which is in agreement with a @xmath2 superconductor , and very much in contrast with the bcs @xmath5-wave case ( which shows an exponential behaviour ) .
an experimental result strongly supporting @xmath2 pairing is the @xmath42 variation of cu nmr relaxation rate @xcite .
arpes measurements on bi2212 compunds by shen _
_ @xcite indicate lines of zeros compatible with a @xmath2 state , and incompatible with an extended-@xmath5 state , although these measurements are unable to rule out pairings such as @xmath43 or an anisotropic @xmath5-wave ( next section ) .
certain other experiments , which support @xmath2 pairing , but can not rule out the possibility of other kinds of pairing include squid experiments on ybco @xcite and electronic raman scattering measurements on bi2212 compounds .
@xcite on the numerical side , studies in favour @xmath2 pairing include an exact diagonalization study of possible pairing symmetries by riera and young @xcite ( on which we comment in the next section ) . dagotto and riera @xcite
find bound hole pairs with @xmath2 symmetry near half - filling and on the verge of phase separation , although the value of @xmath44 is rather too large in comparison with the experimentally relevant parameter range @xmath45 .
another notable work @xcite calculates an anomalous green s function for one and two holes using lanczos technique , and finds signals of @xmath2 pairing over a wide range of parameter @xmath46 ( not specified ) . a variational monte carlo study by li _
@xcite supports @xmath43 symmetry rather than a pure @xmath2 .
this calculation compares rvb variational states with extended-@xmath5 , @xmath2 , and mixed states @xmath47 and @xmath43 symmetries .
the authors show that the mixed states are energetically preferred over pure @xmath5 or @xmath2 states , and that the @xmath43 state overcomes finite - size effects at a lattice size smaller than that required by the @xmath2 state .
the authors also argue that the knight shift data in the ybco cuprates is better explained by @xmath43 rather than pure @xmath2 pairing .
wheatley @xcite has also analyzed the stability of various mixed states involving @xmath2 .
chakravarty _ et al .
_ @xcite have recently proposed an anisotropic @xmath5-wave pairing state based on the interlayer tunneling mechanism @xcite .
this state has no actual nodes on the fs although deep minima occur , and the gap is highly anisotropic . since arpes results , such as @xcite , are unable to detect the sign of the gap function ( as well as the presence of a node unambiguously owing to the current energy resolution ) , they are consistent with such an anisotropic @xmath5-wave pairing as well .
interlayer tunneling is also responsible for a temperature - dependent anisotropy in the gap , as shown by muthukumar and sardar @xcite , which is incompatible with the @xmath2 scenario .
such a temperature - dependent anisotropy has been observed in the arpes data by ma et al .
in our exact diagonalization computation , since we have chosen to work in a number - conserving basis , it is not possible to evaluate @xmath27 directly .
@xcite we thus define a correlation function @xmath48 as @xmath49 where the average @xmath50 is the expectation value in the ground - state of a finite cluster , obtained via exact diagonalization .
this quantity is clearly a measure of coherent pair fluctuations between states @xmath22 and @xmath51 in the ground state .
we find the gap function @xmath27 by diagonalizing this two particle reduced density matrix ( [ dm ] ) , which has the eigenfunction decomposition @xmath52 here , @xmath53 and @xmath54 are the @xmath55 th eigenvalue and eigenfunction of the @xmath56 matrix @xmath57 . here
@xmath58 is the number of points in the brillouin zone and the index @xmath55 orders @xmath59 s as @xmath60 .
superconducting odlro is signalled @xcite by a macroscopic separation of the largest eigen value @xmath61 from the next one @xmath62 , i.e. @xmath63 . the required gap function @xmath27 is the eigenvector @xmath64 corresponding the largest eigen value @xmath61 , which is the cooper pair state with least energy for which condensation will take place , whereas the other eigenvectors represent the various excited states of the cooper pair .
for example , for the standard bcs ground state , @xmath65 , @xmath66 and @xmath67 , for @xmath68 .
a similar procedure was adopted by riera and young @xcite in the context of the @xmath1 model with the three - site term .
however , the authors argue that from the expression ( [ dm ] ) translated into real space , @xmath69 ( where @xmath70 creates a spin - singlet pair of electrons at sites @xmath71 ) , one should retain only those terms which have all the four indices @xmath72 distinct , since it is only these terms which correspond to true pairing terms , rather than mere charge and spin correlations .
the authors _ post facto _ justify this procedure on the basis that the @xmath2 symmetry starts dominating very close to where the two - hole binding energy becomes negative , as a function of @xmath73 .
we do not agree with this argument , since there is no clear cut way to justify such a removal _ a priori _ , in our calculation , since we are calculating a quantity in @xmath16-space ; eq .
( [ dmrs ] ) then dictates that all terms be retained in the fourier transform .
indeed , such removal in the @xmath74 case leads to an incorrect symmetry , whereas the gap function is known to be of a simple bcs type , _
i. e. _ an even function across the fs ( with an @xmath5-wave symmetry in 2d ) .
this is illustrated in fig . 1 in which we plot the gap function corresponding to the highest eigenvalue for an 8-site hubbard chain with @xmath75 , with and without following this removal of terms . without the removal of terms ,
we see the expected behaviour , whereas with the removal , we observe a sign change across @xmath76 . on a @xmath77 hubbard plane at @xmath75 , with the removal of terms ,
we have observed a double sign change across the fs , indicating two nodal lines encircling the fs on either side , whereas without such removal , we see the expected bcs - like symmetry .
we shall thus present results _ without _ the removal of these `` on - site '' terms . note that with removal , we indeed obtain similar results as riera and young .
@xcite we would also like to point out that unlike their work , the focus of our work is the nodes in the gap function .
we shall refer to the gap function corresponding to the highest eigenvalue of the density matrix ( [ dm ] ) as the `` topmost '' gap function , and the term eigenspectrum will be used to refer to the eigenvalue spectrum of the density matrix .
the clusters used for these calculations are single - layer clusters of 8 ( hubbard ) and 16 ( @xmath1 ) sites , and clusters with two 8-site @xmath1 layers coupled by an interlayer term of the form @xmath78 where @xmath79 are layer indices (= 1,2 ) , @xmath80 is the site index within a layer and @xmath81 is the spin label .
we first present the gap function for the 1d clusters .
2 displays the topmost gap function for @xmath82 at various fillings .
a clear change of sign is visible across the corresponding fermi point ( @xmath83 at half filling , @xmath84 for 2 and 4 holes ) , indicating the presence of a node near @xmath76 .
similar behaviour is observed for all @xmath85 , and over a range of fillings ( 0 , 2 , 4 and 6 holes ) .
the results for a 16-site @xmath1 chain are presented in fig . 3 and 4 , and are representative of the parameter range @xmath86 .
3 presents eigenspectrum of the density matrix for various cases : half - filling at @xmath87 ( 1 ) , 2 holes at @xmath88 ( 2)-(5 ) .
all these spectra display a prominant , well - separated largest eigenvalue , indicating the possibility of odlro ( as far as a finite size calculation can reveal ) .
for comparison , we have also plotted the eigenspectrum for a antiparallel triplet cooper pair density matrix in fig . 3 ( 6 ) for @xmath89 , 2 holes , which does not show a well - separated eigenvalue .
4 displays the topmost gap function at half - filling and 2 holes , where a clear sign change across @xmath76 indicates a node near @xmath76 . finally , fig .
5 plots the topmost gap function for a _ single _ 8-site chain in a system of two 8-site @xmath1 chains coupled by an interlayer coupling term ( [ ilc ] ) , for 2 holes with @xmath87 and @xmath90 .
we see again a clear node near @xmath91 very similar to the case of a single 16-site @xmath1 chain .
we now turn to the more interesting case of 2d clusters .
these results are obtained for @xmath77 hubbard , @xmath92 @xmath1 and coupled @xmath77 @xmath1 planes .
the results are representative of a range of parameter values and fillings as specified below .
@xmath77 hubbard cluster : all @xmath85 ( 0 , 2 and 4 holes ) ; @xmath92 @xmath1 plane : @xmath93 ( 0 and 2 holes ) ; coupled @xmath77 @xmath1 planes : @xmath94 , @xmath95 ( 0 and 2 holes ) . to bring about the symmetry features of a gap function
, we shall display the numerical values of the coefficients of the corresponding eigenvector ( of the density matrix ( [ dm ] ) ) at all @xmath16-points in the brillouin zone .
the fs ( and the nodal lines of @xmath2 , wherever required ) will be indicated by dashed lines .
we first discuss the characteristics of the eigenspectra . in all the cases
presented , the eigenspectrum displays a separated largest eigenvalue , although not as well - separated as in the 1d case .
( 1)-(3 ) are eigenspectra for 2 holes , respectively for a @xmath77 hubbard cluster with @xmath82 , @xmath92 @xmath1 plane with @xmath87 and coupled @xmath77 @xmath1 planes with @xmath96 .
the most striking feature of these results is the close interplay between the odd - pairing and the @xmath2 gap functions .
the odd - paired state is the topmost for all the three clusters , with the next dominant state as the @xmath2 state on @xmath92 and coupled @xmath77 clusters . for the @xmath77 hubbard cluster with 2 holes , however , the @xmath2 state is quite low down in the eigenspectrum .
these two states , for a @xmath92 plane , are displayed in fig .
the overall picture that emerges out of this study is as follows . in 1d and 2d
, the gap function shows the presence of a node near or at @xmath76 , indicating an odd - paired behaviour . the @xmath2 state
is often seen as the next dominant state .
quite clearly , the numerical results seem to indicate the presence of a nodal `` surface '' ( gns ) in the gap function closely following the fs . recall that in the rvb mean - field theory at half - filling , the gns coincides with the fs in 2d (
also seen in our numerical results ( not shown ) ) .
it may be argued that as doping increases from zero , the fs shrinks , dragging part of the gns with it .
@xcite the non - overlap of the gns with the fs in some parts in the bz is possibly caused due to the enhancement of the interlayer pair tunneling matrix element @xmath97 in those directions in @xmath98-space .
for example , as emphasized by chakravarty _
@xcite , @xmath97 is largest in the @xmath99 and @xmath100 directions which enhances pairing in those regions of the fs by keeping the gns away .
incorporates this idea as well as the eight nodal points of the recent arpes data @xcite .
based on the present numerical work , we believe that it may be possible to construct novel scenarios where such a nodal surface is also consistent with @xmath2 pairing .
this and such other aspects are discussed elsewhere .
@xcite the major handicap of any exact diagonalization calculation is the cluster size restriction .
it is not possible for us to rule out finite size effects , but the features we have observed are consistent within the sizes and geometries considered .
it is indeed possible that finite size effects are more serious when symmetry issues are concerned , in comparison with , for example , the finite size effects in simple spin , charge or pairing correlation functions .
@xcite note that we do not interpret the present results as indicative of the presence of superconducting odlro in purely 2d systems .
however , it does indicate that these pairing correlations show an odd - paired behaviour across the fs , a feature which seems to persist even in the presence of an interlayer coupling term .
to summarise , we have investigated the behaviour of the gap function @xmath27 across the fermi surface in the doped @xmath1 and large-@xmath30 hubbard clusters in one and two dimensions , with special reference to odd - gap and @xmath2 pairing . for this purpose , we diagonalize a reduced two - particle ( cooper pair ) density matrix computed through exact diagonalization . the largest eigenvalue of this density matrix is well - separated from the rest , indicating the possibility of odlro .
the corresponding eigenvector , the cooper pair wavefunction with least energy , reflects the symmetry of the gap function @xmath27 .
the results on 8- and 16-site single or coupled chains and planes clearly indicate a change of sign of @xmath27 across @xmath76 , indicating the presence of a node close to the entire fs , a signature of odd - pairing behaviour .
this state , on single and coupled planes , in addtion , is isotropic , indicating an angular symmetry which is of the @xmath5-wave type .
we also observe a close interplay between the odd - paired and @xmath2 states in the 2d systems .
we would like to thank v. n. muthukumar for many useful discussions .
m. a. thankfully acknowledges r. e. amritkar for his critical comments , the council for scientific and industrial research , new delhi for financial support , and the center for development of advanced computing ( c - dac ) , pune , for computing facility .
partial financial assitance was provided under project no .
sp / s2/m-47/89 by the department of science and technology and dst project sbr 32 of the national superconductivity programme .
f. mila and e. abrahams , phys .
. lett . * 67 * , 2379 ( 1991 ) .
recently some authors have considered a more general type of singlet pairing that is odd in @xmath16 and @xmath101 both ; see , for example , v. l. berezinskii , pisma zh .
. fiz . * 20 * , 628 ( 1974 ) [ jetp lett . * 20 * , 287 ( 1974 ) ] ; e. abrahams , a. v. balatsky , j. r. schrieffer and p. b. allen , phys .
rev . b*47 * , 513 ( 1993 ) .
p. w. anderson , phys .
lett . * 64 * , 1839 ( 1990 ) ; p. w. anderson in the _ princeton rvb book _ on high-@xmath0 superconductors , chapter v ( unpublished ) .
see also f. d. m. haldane , phys .
lett . * 47 * 1840 ( 1981 ) ; h. shiba and m. ogata , prog .
. suppl . * 108 * , 265 ( 1992 ) . k. miyake , s. schmitt - rink and c. m. varma , phys . rev .
b*34 * , 6554 ( 1986 ) ; d. j. scalapino , e. loh and j. e. hirsch , phys .
rev . b*35 * , 6694 ( 1987 ) ; n. e. bickers , d. j. scalapino and s. r. white , phys .
lett . , * 62 * , 961 ( 1989 ) ; p. monthoux , a. balatsky and d. pines , phys .
b*46 * , 14803 ( 1992 ) ; p. monthoux and d. j. scalapino , phys
lett . * 72 * , 1874 ( 1994 ) . more precisely , the largest eigenvalue of a density matrix with odlro scales as the number of particles in the system
. however , for power - law correlations , the scaling will be with a power smaller than one . 1
. topmost gap function for an 8-site hubbard chain with 2 holes at @xmath75 , _ with _ and _ without _ the removal of terms from the gap expression ( [ dmrs ] ) ; @xmath104 .
notice the sign change across the fs _ with _ the removal of terms .
2 . topmost gap functions for an 8-site hubbard chain with @xmath82 , for 0 , 2 and 4 holes ; @xmath105 respectively .
eigenvalue spectra of the density matrix for the 16-site @xmath1 chain .
( 1 ) half - filled , @xmath89 , ( 2)-(5 ) 2 holes , @xmath106 , ( 6 ) _ triplet antiparallel _
eigenspectrum for 2 holes , @xmath89 .
4 . topmost gap functions for a 16-site @xmath1 chain at half filling ( @xmath87 ) and 2 holes ( @xmath86 ) . @xmath107 for 0 and 2 holes respectively .
topmost gap function for a single 8-site @xmath1 chain in a coupled system of two chains with 2 holes , for values of @xmath90 , @xmath87 .
@xmath108 for @xmath109
eigenvalue spectra of the density matrix for the three 2d clusters at half filling and for 2 holes .
( 1 ) @xmath77 hubbard plane , 2 holes , @xmath82 ; ( 2 ) @xmath92 @xmath1 plane , 2 holes , @xmath87 ; ( 3 ) coupled @xmath77 @xmath1 planes , 2 holes , @xmath110 .
the labels `` o '' and `` d '' near a level respectively stand for odd - paired and @xmath2 states .
a numeral near a level indicates the degeneracy .
the topmost two degenerate gap functions for a @xmath92 plane with 2 holes , @xmath89 .
points marked by squares lie within one brilliouin zone , diamonds belong to the neighbouring ones .
the fs and the nodes of the @xmath2 state are indicated by dashed lines .
the numbers above and below a square are the coefficients of the gap function respectively for the topmost ( odd - paired ) and next ( @xmath2 ) states .
a suggested form of the gns with s - symmetry .
the solid line is the fs and the dotted line is the gns .
the + and - symbols indicate relative signs of the gap function across the gns . | one of the most debated issues related to high-@xmath0 superconductivity is the symmetry of the cooper pair or the gap function . in this report , we present numerical results regarding the gap function in strongly correlated electron systems using @xmath1 and hubbard models in one and two dimensions . to this end , we use exact diagonalization to study the ground states of 8- and 16-site clusters consisting of single or coupled layers .
we calculate a reduced two - particle density matrix in momentum space which is a measure of the gap function .
we then analyze the eigenvectors of this density matrix , which display the possible cooper pair symmetries .
the eigenvector corresponding to the largest eigenvalue indicates a vanishing gap on the fermi surface ( which is in favour of odd - gap pairing ) although @xmath2 symmetry is seen to be a very close contestant in many of the cases . |
that the properties of many galaxies have evolved during recent epochs is a long established fact ( butcher & oemler 1978 , bo78 , madau 1998 , cowie 1996 ) .
that the end point of this evolution results in galaxy populations which vary over space has been known for even longer ( hubble 1936 , abell 1965 , dressler 1980 , d80 ) .
many processes may contribute to such evolution , of which the most fundamental are internal to the galaxy .
star formation builds structure , so structure will evolve as stars form . star formation also depletes its own raw material so- unless provided with an unlimited supply of new gas ( q.v .
larson , tinsley , & caldwell 1980 , ltc80)- star formation will necessarily drive its own decline and eventual extinction .
however , the spatial variation in galaxy populations demands that external effects have significant influence on these processes .
it is possible that this influence was exerted during the epoch of formation , setting up those internal properties of the protogalaxy which drove its subsequent evolution ( e.g. d80 ) .
however , the fact that at least some of the environmental differentiation has occurred at recent epochs ( bo78 , dressler 1997 , allington - smith 1993 , mcgee 2011 ) suggests- although it does not require- that much evolution is driven by the continuing interaction of galaxies with their surroundings .
many such processes have been suggested , including shutoff of the external gas supply ( ltc80 ) , stripping of gas by a hot external medium ( gunn & gott 1972 ) , tidal encounters ( richstone & malmuth 1983 , byrd & valtonen 1990 , moore 1996 ) , and mergers with other galaxies ( dressler 1999 , struck 2006 ) .
thus described , understanding galaxy evolution is a choice between galaxy properties set at birth and galaxy properties transformed by later interactions and , if the latter , between a number of distinct processes of transformation .
although the causes of the original galaxy properties may remain mysterious , their latter evolution seems a tractable problem .
however , the success of the @xmath0 cold dark matter model , and rapid progress in high resolution modeling of galaxies in formation has shown that the simple choices outlined above are an inadequate description of the actual histories of galaxies . because the infall of baryons into dark matter halos and the hierarchical merging of these halos is an ongoing process
, there are no clear distinctions between galaxy formation and galaxy evolution , or between galaxies and their surroundings .
recent numerical models of galaxy growth have highlighted the involvement of a number of physical processes , including infall of cold and hot gas , heating and possible expulsion of gas by stellar radiation , stellar mass loss , supernovae , shocks , and agn s , and the transformation of gas content and distribution of stars by mergers .
( the literature on this subject is now quite large ; good recent summaries include elahi 2016 , with and emphasis on early type clustered galaxies , and murante 2014 , emphasizing disk galaxies . )
although these models have had increasing success in reproducing some of the observed properties of galaxies , including mass functions , star formation histories , and even metallicities , it is still difficult to reproduce all of the properties of individual galaxies and galaxy populations .
a very different phenomenological approach to the same questions has become very popular during the last decade , inspired by the discovery of an apparent `` main sequence '' of galaxies in the star formation vs mass plane ( noeske 2007 ) . such a tight correlation between star formation and mass
has suggested to some authors ( e.g. peng 2010 , dutton 2010 , sparre 2015 , rodrguez - puebla 2016 , tacchella 2016 ) that the myriad physical processes mentioned above somehow conspire to produce _ one _ evolutionary track for all galaxies , that track being the main sequence .
the majority of galaxies remain on that track even today , but a significant number have been `` quenched '' ; driven off the main sequence by one or more processes , including perhaps such classic mechanisms as stripping , starvation , etc .
this idea has an obvious appeal ; as is true in stellar astrophysics , the existence of a main sequence might point to a unique set of astrophysics which could bring order to the complicated mess which galaxy astrophysics appears to be .
whatever its appeal , it has yet to be demonstrated that it is correct . in a series of recent papers ( gladders 2013 , hereafter g13 , abramson 2015 , 2016 dressler 2016 ) we have argued for an alternative view . motivated by the demonstration ( oemler 2013 ) that many galaxies reached their star formation peaks at recent epochs
, we have emphasized the _ diversity _ of the star formation histories of galaxies . in g13
this diversity was parameterized by log - normal star formation histories with two independent timescales , corresponding to the epoch of peak star formation , and the width of the peak .
we have argued that such a description of galaxy histories conforms to the observations at least as well as the main sequence evolution model , including explaining the existence of the main sequence itself .
we are faced , then , with two sets of questions : how do the many astrophysical processes enumerated above result in the variation with time and environment of galaxy properties , and do these processes lead to a unique path of evolution or to many diverse paths ? this is a daunting task , but the recent history should be the easiest aspect of galaxy evolution to understand , for several reasons . with some oversimplification , the complex astrophysics of galaxy formation and evolution can be divided into two rather distinct areas of ( 1 ) extrinsic effects which govern the infall of gas into the regions in which stars are forming and ( 2 ) intrinsic effects which determine the fate of the gas once it arrives there .
most of the difficulties in the correct modeling of galaxy formation lie in the intrinsic processes .
it has been clear since the earliest simulations of forming galaxies in a cold dark matter universe ( white & frenk 1991 , navarro & benz 1991 ) that successful models of real galaxies must include massive amounts of thermal and/or mechanical feedback to disrupt infalling cold gas .
models with inadequate feedback inevitably turn too much of the initial baryons into stars , resulting in galaxies which form too fast , are too massive , and are too centrally concentrated .
getting the feedback right has been a difficult task , one that is not yet complete ( see scannapieco 2012 , and elahi 2016 , for comparisons of various recent attempts ) .
this difficulty is usually attributed to the inevitably incomplete resolution of small scale astrophysical processes- such as the interaction of stellar winds and radiation with the very inhomogeneous interstellar medium- which are central to the feedback processes .
the extrinsic regime should be less troublesome , since it is dominated by the dynamics of dark matter and of baryons in a much less non linear regime than that which occurs within galaxies themselves . at low and moderate redshifts
most galaxies seem to be in a quasi - equilibrium state , with ordered internal structure and dynamics , and moderate to low levels of star formation .
however incomplete our understanding may be of the detailed astrophysics of the intrinsic processes of feedback and star formation , the results of these processes can be observed directly in nearby galaxies in such forms as the schmidt - kennicutt law ( schmidt 1959 , kennicutt 1989 ) , which has been shown to govern star formation in a wide variety of galaxies at a range of redshifts ( e.g. daddi 2010 , bouche 2007 ) , and which recent numerical models ( e.g. hopkins 2014 ) are now able to reproduce with some accuracy .
we may be left , then , with only the extrinsic effects as possible variables to explain the recent histories of galaxies .
there is , of course , a very large literature on the systematic properties of nearby galaxy populations .
it is well established that there are strong trends present in these populations , including trends of star formation rates with mass , epoch , structure , and environment , of metallicity with mass and galaxy type , and of structure with mass and environment .
also , much work has gone into attempts to delineate the `` main sequence '' and its evolution with time ( speagle 2014 present a compilation of many of these ) . despite all of this work ,
the subject of local galaxy populations and their implications for galaxy evolution is far from exhausted . in this and subsequent papers we use multiple published surveys of the local population of galaxies to reexamine the systematic behavior of their main observable properties : mass , star formation rate , disk scale , bulge fraction and gas content .
our primary goal in this first paper is to understand the present distribution and recent evolution of gas and star formation in disk galaxies .
a subsequent paper will examine the issue of bulges and disks .
after a careful recalibration of star formation rates , we will show that we can reliably measure star formation rates well below the main sequence , and thus follow galaxies as they evolve from the main sequence towards the passive state .
we will show that the well established bimodality in galaxy colors hides an important intermediate population of not - quite - dead galaxies with significant gas but low star formation rates . delineating the systematic properties of the star forming , dead , and not - quite - dead galaxies will allow us to put useful new constraints of the processes governing recent galaxy evolution .
the layout of this paper is as follows . in section 2
we survey the mass star formation plane , reaching the tentative conclusion that it can be subdivided into three distinct groups : the main sequence , non - star forming objects , and the intermediate group with low but non - zero star formation rates , which comprises about a third of the galaxy population . in section 3
we assemble facts about these three populations , deferring most analysis until section 4 , in which we attempt to understand these properties in light of various theories of galaxy evolution .
section 5 summarizes the results . in the appendix
we will critically examine the star formation rates from brinchmann ( 2004 ) , and recalibrate them using the best available data .
this section is critical for establishing the reliability of our detection of low star formation rates , but can be skimmed by those mostly interested in the results .
brinchmann ( 2004 , b04 ) have derived star formation rates from imaging and spectroscopic observations of 149,660 galaxies from dr2 of the sloan digital sky survey ( york 2000 ) which have spectroscopic observations , r magnitudes in the range @xmath1 and redshifts between 0.005 and 0.22.they also incorporate mass determinations from kauffmann ( 2003 ) . in 2007
they recalibrated their method , and applied it to the dr7 release .
these new data are not published but are available on the web .. we will use this latter data set , which we denote as b04-dr7 , in all of the following analysis .
the mass values we shall use as given ; all determinations of galaxy stellar masses are derived from sed data using stellar population models and we have no reason to think that we can improve of those of kauffmann .
any significant errors are probably systematic to the population model and will only shift the mass scale , which should not affect our analysis .
there are , however , independent methods of determining star formation rates , and in appendix a we use what we believe are the best of these to improve on the b04-dr7 sfr s .
the main changes are to the star formation rates of galaxies with weak or undetected emission lines . unless otherwise noted , all quoted sfr values are based on our recalibration .
first , though , we begin with the original b04 data .
a particularly striking presentation of the star forming main sequence to be found in the b04 data is presented in figure 1 of peng ( 2010 , p10 ) .
we repeat this presentation of the b04 data in figure 1 .
the density of objects is normalized so that the integrated density per interval of log mass is constant , which permits a better appreciation of the behavior of star formation as a function of mass .
this is a quite striking correlation , with a scatter @xmath2 . if , as has often been assumed ( but we show in appendix a to be incorrect ) , the typical errors in b04 star formation rates were as large as 0.2 dex to 0.3 dex , figure 1 would seem to imply that the true scatter in the star forming main sequence is quite small .
unfortunately , this plot does not stand up to close scrutiny .
the first thing to note is that plotting observed sfr versus stellar mass gives a quite misleading impression of the correlation between these quantities , because the stellar mass is simply the time integral of the star formation rate .
more precisely , @xmath3 where @xmath4 is the fraction of the mass that remains locked up in stars , rather than returning to the ism during stellar evolution . unless @xmath5 is completely uncorrelated with @xmath6 which would imply a completely random history of star formation sfr will _ always _ be correlated with mass , whatever the underlying astrophysics or history of star formation .
a less misleading presentation of the b04 data can be obtained by plotting _ specific
_ star formation rate , @xmath7 versus mass .
we can write ssfr as @xmath8 thus , ssfr is an ( incomplete ) measure of the star formation history of a galaxy , in that , while many histories can lead to the same value of ssfr(t ) , different values of ssfr(t ) require different histories . also , we must improve on the data set used in figure 1 .
firstly , we replace the b04 star formation rates with those obtained in the appendix , which are more accurate , particularly for low star formation rates .
secondly , p10 cull their galaxy set in two problematic ways : by excluding galaxies without strong emission lines or with some sign of an agn component , and by excluding galaxies without sufficiently blue @xmath9 colors .
all of these exclusions will narrow the ssfr range .
the result of these corrections is presented in figure 2 where we now plot the ssfr - log(mass ) distribution for all objects with detected star formation ( i.e. @xmath10 ) .
( we now plot individual points rather than a normalized distribution . )
this distribution is obviously much broader than that in figure 1 , but one source of bias remains .
the analyses of b04 and p10 , as well as those presented so far in this paper , have included all sample galaxies ( of particular spectral properties ) at every redshift .
the sdss spectroscopic sample is complete to @xmath11 . to see what this means for mass completeness we must consider the mass - to - light ratios of the galaxies . in figure 3
we plot the distribution of r - band mass to light ratios of the b04-dr7 galaxies versus their ssfr , and fit a polynomial to the median value of m / l(r ) vs log(ssfr ) , over the ssfr range @xmath12 where most of our objects lie .
superimposed on the ssfr - mass distribution in figure 2 are the mass limits derived using that polynomial fit at the first through third quartiles of the redshift distribution of the sample , namely z = 0.054 , 0.083 , 0.119 .
the sample is more than 50% incomplete to the left of each line .
clearly , the data as usually presented are seriously incomplete .
furthermore , because of the slope of the completeness lines , galaxies with lower ssfr s are systematically excluded since they are less luminous at a given mass than those with high ssfr s .
note that this effect operates at every mass , since the higher mass galaxies are drawn preferentially from higher redshifts .
given the results in figures 2 and 3 , we can construct the largest complete sample of b04-dr7 , which we denote the _ maximally complete sample _ , by , at every mass , including only galaxies out to a redshift such that all galaxies with @xmath13 , the maximum found in the sample , will have @xmath14 , and therefore will be included in the complete sdss spectroscopic sample . in order to minimize the mismatch between galaxy size and sdss fiber size
, we also limit the sample to objects with @xmath15 .
note that this sample , although unbiased in star formation properties at each mass , is strongly biased towards high mass , since the sample of higher mass objects is complete over larger volumes .
the results for this sample are presented in figure 4 . the beautifully clean star forming main sequence of figure 1 has almost disappeared .
although some sort of sequence is apparent at low mass , it seems to be entirely obscured at high masses by a broad swath of lower sfr objects . in all following analyses
, the sample will be limited to the maximally complete sample unless explicitly stated otherwise .
figure 4 shows that the ssfr mass distribution is very broad , with a distinct narrow main sequence only apparent at low masses ( @xmath16 ) .
however , closer examination shows that this broad distribution has significant structure .
figure 5 presents the distribution of @xmath17 in three intervals of log(mass ) : 9.39.7 , 9.810.2 , and 10.3 - 10.7 .
three peaks are apparent in all of these : one for those without detectable star formation- which we arbitarily place at @xmath18 , one at @xmath19 , and one at @xmath20 .
the middle peak is broad , and does not shift with mass , but the upper peak is narrow , and moves to lower values of log(ssfr ) as mass increases .
it is reasonable to still call the upper peak the `` main sequence '' .
using all objects with @xmath21 , we obtain the dispersion of the main sequence - @xmath22 , independent of mass , as well as its slope @xmath23 .
it should be noted that the width of the main sequence , though narrow , is real , since the scatter in our best ssfr measure , established in appendix a , is substantially less than 0.3dex at the main sequence .
let us assume for the moment that the main sequence is a distinct component of the ssfr mass distribution , separate from the intermediate ssfr group , with the shape described above .
if we also assume that the main sequence dominates the distribution at values of log(ssfr ) higher than the main sequence peak , and that the distribution is symmetrical about the peak , we can use counts of objects with @xmath24 to obtain its mass function , and using that plus the slope and spread , we can generate the entire main sequence and statistically remove it from the ssfr mass distribution of all galaxies .
the results of this decomposition - ignoring the dead galaxies with no detectable star formation- are presented in figure 6 .
the most striking implication of this decomposition is that the main sequence , rather than dominating the ssfr - mass plane , as implied by , for example , figure 1 , only contains about half of the star forming galaxies .
the other half belong in a broad swath stretching from below the main sequence down the the level at which star formation becomes undetectable .
the existence of a population of galaxies with star formation rates significantly lower than the main sequence is not a new discovery , but it has not received much attention . it is absent from color distributions , which only show the main sequence and no star formation peaks and an almost empty `` green valley '' , because galaxy colors are insensitive to ssfr values below about @xmath25 .
dressler & abramson 2015 ) such a population is apparent in figure 24 of b04 , as well as in figure 22 of salim ( 2007 ) .
schiminovich ( 2010 ) discuss it in relation to star formation efficiency ( to which we will turn in section 3.3 ) , and tojeiro ( 2013 ) in relation to star formation histories .
in most cases its importance has been underestimated because of the sample bias described by figure 2 , as well as by the difficulty in calibrating very low star formation rates .
sdss images of a random subset of these low ssfr galaxies are presented in figure 7 .
most appear to be rather normal early type galaxies .
a set of 51 galaxies in the low
star formation group which are members of the coma cluster and were classified by d80 contains only 4 that were not typed as normal e s or s0 s , including 1 ep , 2 s0/a s , and 1 sb .
the identification of the low ssfr subset of galaxies as a distinct component is only a convenient working hypothesis , although the distributions in figure 5 give some credibility to the idea .
also , the distinction between these objects and those with no detected star formation is presumably an artifact of our inability to measure specific star formation rates below @xmath26 .
nevertheless , for the moment we will divide the galaxy population into three distinct groups : those with vigorous star formation which we will continue to call the _ main sequence _ , the intermediate group , which we will denote as _ quiescent _ , and those with no detected star fomation , which we will call _
passive_. for the following analysis , we will label galaxies as `` main sequence '' or as `` quiescent '' according to their assignment in the statistical decomposition presented in figure 6 . in the overlap region
this is , of course , an arbitrary assignment , but no more arbitrary that doing the cut at some specified value of ssfr .
all objects below the detection limit of @xmath27 are classified as `` passive . '' in coming sections , we will explore how the properties of these populations are related to their gas content , environment , and structure .
in figure 8 we present the incidence of our three star formation classes as a function of environment , in five mass ranges , using two measures of environment .
we take these measures from tempel ( 2012 ) who , as part of a clustering survey of sdss dr8 galaxies , have determined group membership and normalized galaxy densities within 1 , 2 , 4 , 8 , and 16 mpc of a sample of sdss galaxies which includes about 89% of our maximally complete sample ; we will use their group membership , as well as density within 1 mpc , 1 .
errors for points in the left column are typically 0.01 , those in the right column range from 0.005 for small @xmath28 to 0.02 for @xmath29 .
it is clear that star formation is sensitive to environment at every mass and in every environment , though high mass galaxies show a lower sensitivity than do low mass ones .
conversely , in every environment high mass galaxies are more likely to be quiescent or passive than are low mass objects . at first glance , the population of quiescent galaxies appears to be quite stable , with only the ratio of star forming to passive galaxies varying with environment , but this appearance is misleading . in figure 9
we plot , for every data point in figure 8 , the fraction of main sequence galaxies versus the ratio of quiescent to passive galaxies .
the result shows that there is a very strong correlation between these two quantities at every mass and local density .
the lower the fraction of main sequence galaxies , the lower the fraction of galaxies below the main sequence with some star formation .
two hi surveys cover large portions of the volume of the b04-dr7 sample .
the alfalfa survey ( giovanelli 2005 , haynes 2011 ) used the arecibo telescope for a blind survey of a significant fraction of the north galactic cap over the velocity interval @xmath30 .
the survey volume included about @xmath31 b04-dr7 galaxies ( @xmath32 in the maximally complete sample ) , of which about @xmath33 were detected ( @xmath34 in the maximally complete sample ) .
the sensitivity of the survey varies with linewidth , but a typical limit for 50% completeness is @xmath35 , or @xmath36 at the median redshift of the survey . the gass survey ( catinella 2012 ) is a pointed survey of about @xmath37 galaxies within the alfalfa survey volume which were _ not _ detected either by alfalfa or by springob ( 2005 ) .
it includes only objects in the mass range @xmath38 , and the redshift range @xmath39 .
it is about a factor of 5 deeper than alfalfa at the same redshift .
figure 10 summarizes the hi content of the b04-dr7 galaxies derived from these two surveys .
figure 10a plots the median gas to stellar mass ratio as a function of log(ssfr ) , in intervals of galaxy mass , and figure 10b plots the @xmath40 dispersion , calculated from the difference between the 50th and 84th percentiles of the distribution .
above @xmath41 the alfalfa survey detected the majority of the galaxies over some subset of the survey volume , allowing a determination of both quantities . since the survey is approximately flux limited , the detection rate is a function of @xmath42 .
the redshift limit of each @xmath43 and ssfr subset is chosen to sample as large as possible a fraction of the alfalfa volume , while still reaching deep enough into the @xmath44 distribution to allow determination of the median value .
since the detection threshold is rather broad , there will be some detections below the completeness limit and some non detections above the limit .
we handle non detections in two ways ( 1 ) by assigning them a value of @xmath44 equal to the 50% completeness limit or by assigning them values of zero . where these differ ,
both points are plotted .
when the majority of the objects are undetected , the latter value goes to zero , as with the leftmost black point in figure 10a .
the alfalfa results are presented in figure 10 as colored points , divided into the same mass intervals as in figure 8 .
errors in the alfalfa sample range from 0.02 to 0.04 for median values of @xmath45 , and from 0.015 to 0.035 for @xmath46 . below @xmath41 the alfalfa sample in insufficiently deep and we turn to the gass survey . here
things become slightly complicated .
the gass sample is a random subset of objects within the alfalfa survey volume with redshifts @xmath47 and masses @xmath48 , _ but _ excluding objects that were already detected by alfalfa .
thus , the gass sample by itself is biased against high hi content objects . to correct for this bias
, we must add back those objects which would have been included if not detected by alfalfa . since the survey does not provide information on the number of non included objects
, we must estimate it from the alfalfa sample itself , by the following procedure .
the gass sample objects are ( almost ) all contained within our b04 maximally complete sample , which we abbreviate maxsamp .
let @xmath49 be the subset of alfalfa objects in maxsamp and within the gass mass and z ranges .
let @xmath50 be the subset of gass objects within maxsamp .
since the alfalfa detection probability within asamp will depend of galaxy mass , we divide the sample into mass intervals .
then , it is straightforward to show that , in each interval , if @xmath51 is the number of _ undetected _ @xmath49 objects , and @xmath52 is the total number of @xmath50 objects , we can combine the gass objects with the alfalfa detected objects by giving each @xmath49 detection a weight of @xmath53 .
the gass sample is too small to subdivide by mass , so we include all objects
. we deal with gass nondetections in the same manner as with the alfalfa sample .
the results are presented in figure 10 as black points .
typical errors in the gass sample are about 0.08 for the median values of @xmath45 and 0.06 for values of @xmath46 .
the diagonal dashed lines are lines of constant gas to star formation rate , which is , to within a factor of @xmath4 , the inverse of gas consumption timescale , @xmath54 .
the upper line corresponds to @xmath55 , and the lower line to @xmath56 .
several points seem clear .
( 1 ) at a given ssfr , the scatter in gas to stellar mass ratio is rather small , no more than a factor of 2 .
( 2 ) gas consumption timescales seem to be independent of galaxy mass .
( 3 ) the trend of gas content with ssfr is not inconsistent with a simple picture in which _ _ the median gas consumption timescale of main sequence galaxies is 4.7 gyr , independent of ssfr , and the median gas consumption timescale of quiescent galaxies is 21 gyr , independent of ssfr .
another view of the gas consumption issues is illustrated by the gray points . in gladders g13 we constructed models for the star formation histories of galaxies with log
normal shapes of the form@xmath57 fitting the distribution of shape parameters using observations of low and high redshift galaxy populations as constraints . if galaxies were closed boxes , with gas neither entering nor leaving , and if all gas was eventually turned into stars , then a galaxy s gas content at any time would be just that needed to form the mass of stars it will form in its future .
since we know the future evolution of each g13 galaxy , we can determine the gas ssfr distribution of those galaxies .
this is presented as grey points in figure 10 .
the important point is that the median of the grey points lies near or above the median gas content of main sequence galaxies , but well below that of the quiescent objects .
thus , the median gas content is more than sufficient to fuel the future star formation of all of the quiescent galaxies , but not all main sequence objects .
these models represent only one possible description of galaxy formation histories , though they do fit well all of the extant constraints .
however , it is likely that most alternate models , or at least those in which star formation decreases steadily from a peak at early times , will have a qualitatively similar distribution .
therefore , it appears that many ( though far from all ) main sequence galaxies will require a continuing inflow of gas to maintain their star formation , but that quiescent galaxies do not .
we will consider the implications of this fact in section 4 .
we close this section with a brief look at galaxy bulges and disks .
we shall examine this in much greater detail in a subsequent paper , but one important fact will be useful in the following discussion .
it has long been known ( hubble 1935 , baade 1951 ) , that the masses , stellar populations , and structures of galaxies are correlated : high mass , bulge
domination , and low star formation rates tend to be associated with each other , as are low mass , disk domination , and high sfr .
mendel ( 2014 ) have determined the masses of disks and bulges for a large fraction of the b04-dr7 set of galaxies . in figure 11
we compare the bulge mass fractions ( which we will designate ) of the three galactic populations with masses in the range @xmath58 . to avoid the ambiguous overlap region near @xmath59
we only take main sequence galaxies with @xmath60 and quiescent galaxies with @xmath61 .
the import of this figure is clear : quiescent and passive galaxies are structurally almost identical and very different from the star forming population . while star
forming galaxies have an almost uniform distribution of bulge mass fraction , _ there are almost no disk dominated quiescent or passive galaxies _ , those with bulge mass fractions less than 0.2 .
conversely , there are very few bulge dominated ( @xmath62 ) star forming galaxies .
much but not all of this is old news : it is a commonplace that there are very few star forming ellipticals , and the lack of disk dominated s0 s was first noted by sandage & visvanathan ( 1978 ) , who argued from this that s0 s could not be stripped spirals ( more on this later . )
multiple lines of evidence seem to suggest that main sequence and quiescent galaxies form distinct populations .
the evidence includes the ssfr distribution at fixed mass ( figure 5 ) , the variation in the incidence of the populations with environment ( figures 8) , the ssfr dependence of gas content ( figure 10 ) , and the almost total absence of pure disk galaxies among the quiescent and passive populations ( figure 11 ) . that the main sequence is a physically distinct population
has been a common theme of many recent papers , many of which have tried to identify the astrophysical processes responsible for the tight sfr vs mass dependence .
the above list seems to support such a view , but the existence of a well defined population does not necessarily imply a physical process driving the sfr mass correlation . as was pointed out in section 2 , _ any _ star formation histories that are not completely random will produce a correlation of sfr with mass . as equation 2 makes clear , the peak of the resulting ssfr distribution will be determined only by the age of galaxies ( @xmath63 ) and the mean value of @xmath64 .
if the majority of galaxies are of comparable age then only the range of the shapes of the star formation histories will produce a scatter in ssfr .
two very general considerations require that the ssfr distribution fall rather rapidly away from the peak .
equation 2 constrains the high side .
if we assume that very few galaxies have rising star formation rates at the present epoch , equation 2 requires that the ssfr distribution fall fast at high values of ssfr , nearing zero at @xmath65 .
similar considerations constrain the low side . at
early times- shorter than the timescale over which sfr varies- @xmath66 for all galaxies . therefore
, if star formation rates are now falling , and if @xmath67 is zero or negative , the galaxies with lowest values of ssfr today will be those in which sfr has fallen the fastest . if lower ssfr galaxies evolve faster , then galaxies accelerate away from the `` main sequence '' , traversing each interval of ssfr more quickly , resulting in a decreasing density of objects at lower ssfr .
note that these effects depends not at all on any particular assumptions about the physical processes driving the star formation history . as an illustration of this effect
, we construct a random population of galaxies based on the log normal star formation histories , of the form described by equation 3 , developed in g13 . in that paper ,
the distribution of the parameters @xmath68 and @xmath69 were tuned to match observations of the history of the star formation density as well as of the ssfr distribution . here
we take a completely random set of values over the range @xmath70 , @xmath71 , considerably broader than what was needed to fit the observed ssfr distributions .
we evolve the histories until the present age of the universe and calculate the present epoch ssfr of each objects .
values of @xmath72 are assumed to be undetectable and are set to the arbitrary value of -13 .
the results are presented in figure 10 ; from the 16th and 84th percentiles of the distribution of detected objects we calculate @xmath73 , only slightly broader than we have determined for the real `` main sequence '' in section 2 .
thus , a completely random set of star formation histories produces a ssfr distribution almost identical to that observed .
the point is that the existence of a `` main sequence '' signifies little about the history of the star formation in a galaxy population .
we have chosen a log normal form for illustration , but virtually any set of star formation histories satisfying the two constraints that ( 1 ) @xmath67 is zero or negative and ( 2 ) star formation histories are smooth rather than having large random excursions ( which could temporarily push ssfr to arbitrarily high or low values ) will produce a similar distribution . from equation 2
, we expect the peak to be located somewhat lower than @xmath74 , which means that it will rise with increasing redshift , as it is observed to do .
therefore , the attempt to construct physical models to explain the properties of the main sequence is misguided .
although it is possible that some detailed astrophysical considerations could duplicate the behavior which we have seen to occur naturally , simplicity demands that we accept the straightforward explanation described above , until some new information forces us to look for elsewhere .
if the main sequence is a natural consequence of the age of galaxies and a smooth evolution of their star formation rates , the same is not true of the quiescent population .
figure 12 , and the preceding argument , both show that there should be a rapidly falling tail of the ssfr distribution to the low side of the main sequence peak , rather than the large population of galaxies at low ssfr values that are observed . although the existence of a population of intermediate ssfr galaxies is not news , many of the properties of this population are new , and are quite puzzling , so we should first address the possibility that our low ssfr galaxies are spurious , and the result of errors in the star formation rates .
we demonstrated in appendix a , by a comparison with star formation rates derived from ( mostly independent ) uv+ir measures , that our sfr measure is quite accurate , but the possibility remains that _ all _ of the star formation measures are systematically wrong .
it is easy to rule out the possibility that the ssfr values of the quiescent objects are spuriously low , i.e. that they are misclassified main sequence objects .
neither their appearance ( figure 7 ) , dependence on mass and environment ( figure 8) , gas content ( figure 10 ) , or structure ( figure 11 ) look anything like that of the main sequence . at most
, only a very small fraction of the quiescent class could be misclassified main sequence objects .
might they be misclassified passive objects ? many of their properties are closer to passive than to main sequence galaxies .
however , their dependence on mass and environment are quite different ( figure 8) , as is their gas content . in a sample of quiescent and passive galaxies within the alfalfa survey volume , matched in redshift and mass distribution , hi gas
is detected in 10.3% of the quiescent galaxies but only 1.7% of the passive objects . in similarly defined samples from the gass survey ,
the fractions are 57% and 16% .
this is inconsistent with the quiescent objects being drawn from the passive population ( but suggests , as one would expect , that some fraction of the objects in the passive class might be very low ssfr quiescent galaxies ) .
we conclude , therefore , that the great majority of the quiescent galaxies are correctly typed , and the results summarized above can not be dismissed as being due to sfr measurement errors .
that said , we do _ not _ claim that the quiescent and passive objects form separate populations .
the dividing line between the two is set purely by the considerations described in appendix a : below log(ssfr ) = -12.2 we can not reliably detect star formation in our sample galaxies .
it is likely that the two groups form one continuous population differing only quantitatively in their ssfr values .
whether there is a distinct group of galaxies with absolutely no star formation is far beyond our ability to determine . to understand the quiescent galaxies , we need a model of galaxy evolution with more content than the random assemblages of log normals used earlier .
however , we do not need a full - blown numerical simulation ; a toy model of galaxy formation based on the considerations mentioned in section 1 will suffice . as we pointed out earlier
, galaxy evolution divides , roughly , into intrinsic and extrinsic effects .
the intrinsic effects are very complicated , but they result in star forming behavior which at recent times has been directly observed and characterized , i.e. the kennicutt - schmidt law .
extrinsic effects are not , in general , observable and- absent a full numerical model , are not calculable .
we have no such model , and no completely successful model has yet been constructed .
however , if we ignore mergers , the result of these extrinsic processes is to deliver a net amount ( infall - outflow ) of _ cold _ gas to the star forming regions of the galaxy with some unknown time dependence .
here we adapt the simplest possible such history : a constant rate of addition which begins at time @xmath75 years , and continues until a time @xmath76 years .
we assume the the gas flows into the disk with a radial density profile like that of the stars in disks today @xmath77 . from the data of simard ( 2011 ) we obtain mean values @xmath78 , and a dispersion of about a factor of 2.0 . we take the parameters of the kennicutt - schmidt law from kennicutt ( 1998)- more recent compilations ( e.g daddi 2010 ) are still consistent with these values- which can be expressed as @xmath79 , where @xmath80 is the surface density of star formation in solar masses per year per square parsec , and @xmath81 is the surface density of hi gas in solar masses per square parsec .
this model can not deal with bulge formation , but it need not do so . whether bulges are formed very early in galactic history , or present day bulges are the result of the last major merger between previous disk galaxies , the history of the _ present _ disk likely only begins after bulge formation is complete .
after that time the bulge contributes mass but neither gas nor star formation .
since the typical galaxy in our mass range has a bulge mass fraction of about 0.5 , we account for the bulge by shifting both the gas to mass ratio and the specific star formation rate of our model galaxies down by a factor of two .
this model is only a toy , but should not be a bad descriptor of at least the recent evolutionary history of the stellar populations of galaxies .
different infall histories or starbursts may change the short term star
forming behavior , but will have little long term effect . in figure 13
we compare the predictions of this model for ssfr distribution and gas content with the observations reported earlier , for galaxies in the mass range @xmath82 .
the agreement is not good .
the model reproduces , roughly , the ssfr distribution and gas content of the main sequence , but fails completely at lower values of ssfr , producing neither quiescent nor passive galaxies . however , our description of the star formation dependence on gas content is not yet complete .
kennicutt ( 1989 ) , and kennicutt & martin ( 2001 ) have demonstrated that , below a critical density- probably related to the local toomre q disk stability threshold ( toomre 1964 ) - star formation rates fall precipitously . if the q parameter is indeed the governing factor then , for a flat rotation curve , their results imply a critical density @xmath83 . from bell & de jong ( 2001 )
we take a mean relation for disk galaxies @xmath84 , where @xmath85 is the stellar mass in units of @xmath86 , giving us the critical density as a function of galactic radius of @xmath87 .
the star formation rates below the critical density are not well determined .
they are not zero : observations show that parts of galaxies below the critical density typically have small , isolated star forming regions . for lack of a better specification , we shall assume that below the critical density the sfr is 0.1 times the value predicted by the kennicutt
schmidt law .
this more complete model is also shown in figure 13 , and it is is in much better agreement with the observations .
there is now a distinct main sequence , and a large group of quiescent galaxies with ssfr s about an order of magnitude lower than the main sequence .
the gas to stellar mass ratio now flattens out at the location of the quiescent galaxies , with values of the gas to star formation ratio about a factor of 5 higher than on the main sequence .
in fact , even the bump in gas to mass at @xmath88 is reproduced .
the reasons for this different behavior is explained in figure 14 , which presents the sfr histories of a sampling of galaxies of varying mass , @xmath68 , and @xmath89 values , with and without stability quenching .
compared to models without a disk instability threshold , the star formation in models with a threshold drops more rapidly , as less and less of the galaxy is above the density threshold .
once all of the galaxy is below the threshold , the global star formation efficiency stops falling , the gas consumption rate becomes very low , and the star formation rate drops very slowly .
galaxies move slowly in the mass ssfr plane , and pile up near the locus at which they made the transition .
although gas density in the disk drops with radius , so does the critical density , so the entire galaxy makes the transition over a relatively short timescale . _ and note that , again , a wide range of randomly selected star formation histories has produced a narrow main sequence . _
the agreement of model and observations in figure 13 is , in fact , quite remarkable , given how simple our model is .
it must be stressed that no tweaking of parameters has been done ; we have taken star formation behavior straight from kennicutt ( 1998 ) and martin & kennicutt ( 2001 ) , and galaxy structure straight from simard ( 2011 ) .
this strongly suggests that no additional astrophysics , and no additional gas flows are necessary to move galaxies from the main sequence to the quiescent state . by itself , the rather small dispersion in gas to mass ratios seen at a given ssfr in figure 10 implies that the mass of cold gas observed by alfalfa and gass is equal to the mass of star forming gas ; little has been missed and little is beyond the star
forming region .
the agreement of model and observations further implies that this gas mass is neither increased by significant inflows not depleted by outflows , _ at least during the main sequence to quiescent transition _ ( more on this soon . )
the meaning of the quiescent population is now clear : _ quiescent galaxies are those in which the density of cold disk gas is below the gas disk instability threshold . _
all of the evidence presented in this paper is consistent , then , with the hypothesis that the evolution of disk galaxies away from the main sequence is driven by the depletion of gas due to star formation .
the rate of evolution , and the appearance of the galaxies is governed by the schmidt - kennicutt law , as modified by an instability threshold .
we have now solved the problem of the quiescent galaxies , but at the expense of raising another : if galaxy evolution due to star formation almost stops at , say , @xmath90 , how do galaxies make their way to lower values of ssfr , and how do they deplete their gas and enter the passive state ?
two possibilities seem obvious : either ( 1 ) some galaxies , either contemporaneous with those entering the quiescent state or at an earlier epoch , somehow avoided the drop in star formation rate , and managed to turn their remaining gas into stars , or ( 2 ) another process is active which removes the remnant gas by other means .
possibility 1 seems unlikely , at least for disk galaxies . in a following paper ( oemler in preparation )
we will show that disk structure seems almost constant below @xmath91 , so that there is no reason to think that gas in the now passive disks experienced very different conditions than those in quiescent disks today .
ellipticals are , of course , a different matter : it is well known that bulge dominated galaxies are more prevalent among passive galaxies , and our toy model has no relevance to bulge formation .
however , bulge - dominated galaxies comprise at most 30% of the passive population , so different star formation astrophysics in bulges can not be more than part of the answer .
possibility 2 seems more promising , and many mechanisms are known that can expel gas from galaxies .
the strengths of many are proportional to the star formation rate , and are unlikely to be effective at the very low sfrs of the quiescent galaxies , but not all have this limitation .
some type i supernovae occur well after the cessation of star formation , as do winds from agb stars .
agn s may as well , although it is not clear how important agn s are in galaxies with small bulges .
any such mechanism must be able to explain figure 9 , which plotted , over the entire observed range of mass and environment , the ratios of main sequence to other galaxies , and the ratio of quiescent to passive galaxies .
these quantities are independent of each other , but there is a quite tight correlation between the two , at almost all values of mass and density .
thus , if we try to invoke separate explanations for the transformation of main sequence galaxies into quiescent objects and for the transformation of quiescent into passive objects , we must contrive a tight connection between the two processes .
fortunately , there is a natural connection .
comparison of the model and observed ssfr distributions in figure 13 implies that the transformation process we are seeking operates on a fairly long timescale , longer than a few gyrs but shorter than a few 10 s of gyrs . if it were as short as the former , all quiescent galaxies at @xmath92 would have been quickly transformed into quenched objects .
if it were as long as the latter , all would still be at @xmath92 rather than spread all the way down to the passive region . in our model , the evolutionary rate and present state of a galaxy depends mostly on the epoch at which inflow ends . if certain environments and masses skew this epoch towards earlier values , more galaxies will have entered the quiescent state by now , _ and , on average , they will have done so at an earlier time . _
thus , there will have been more time for the slowly working quiescent to passive transformation process to work , and more will have made the transition .
_ kelson 2016 , which come to similar conclusions from a different direction . )
it is our contention that the post
main sequence evolution of disk galaxies is driven by two intrinsic processes : the depletion of hi gas through star formation followed by the expulsion of the remaining gas , probably by the combined effects of stellar winds and supernovae .
however , this intrinsic process only begins after an extrinsic event : the cessation of the net inflow of cold gas from outside .
now , it is possible , since it is the _ net _ inflow that matters , that this is an intrinsic effect , due to an increase in outflows , perhaps driven by an agn , rather than a cessation of inflows . however , unless such outflows were delicately balanced so as to only cancel the continuing inflows , rather than result in the rapid expulsion of all gas , such an event would lead to a much more rapid decline in star formation , and a much more rapid evolution of the galaxy towards the passive state , than is evidenced by the ssfr distribution in figure 10 .
figure 8 confirms what has long been known : dense environments ( and higher masses ) push the galaxy population towards lower star formation rates . given our model , these are , in fact , the trends one might expect .
the environmental dependence on gas inflow was first proposed by ltc80 , who pointed out that the observed star formation rates and gas content of spiral galaxies required continual infall of new gas if star formation was not to fade , and proposed that s0 galaxies were the product of the truncation of gas infall in dense environments .
more recent models of a @xmath0 cold dark universe support such an idea .
for example , dekel & brinboim ( 2006 ) argued how the mass of the dark matter halo might govern the thermal properties of the infalling gas .
lower mass galaxies are fed by streams of cold gas , which can continue to feed star formation for extended epochs . in high mass galaxies
the gas is shock heated to high temperatures at earlier times , after which it is more vulnerable to dispersal by winds generated by star formation and agn s .
feldmann ( 2011 , f11 ) have performed one of the first fine scale numerical simulations of a large enough volume to simultaneously resolve the intrinsic processes of galaxies and also include a large enough volume to study the extrinsic processes driving evolution .
they find that the evolutionary history of disk galaxies in a group are primarily driven by ( a ) mergers , which form the bulge component ( b ) suppression of gas infall within the group environment and ( c ) ram pressure stripping when galaxies pass through the group center .
the stellar and gas content of the galaxies today depend on how long they have been in the group environment : those that entered first have evolved the most towards the passive state . a significant role for ram pressure stripping of the disk gas is _ not _ consistent with our observations , since that would accelerate gas depletion and accelerate the transition of galaxies towards the passive state . however
, the environment in the f11 simulation is denser than that experienced by the average galaxy .
although perhaps half of all galaxies are in groups , meaning that they have at least one neighbor of comparable or larger mass , the typical groups is significantly smaller than that studied by f11 .
cen ( 2014 ) has performed a larger scale simulation of galaxies in their environments , containing 3000 galaxies in a considerably wider range of environments .
he also finds that ram pressure stripping is important , but only for removing the large scale cold gas halos surrounding galaxies which resupply the disk .
the subsequent evolution of the gas content of the disk is driven , as in ltc80 , by star formation ( see also rafieferantsoa , 2015 ) .
we conclude , therefore , that ram pressure stripping of gas disks , as originally proposed by gunn & gott ( 1972 ) will only be important for that small subset of galaxies living in the densest environments . in our picture , the environmental dependence of galaxy stellar populations has two parallel components : the population of quiescent plus passive galaxies is driven by _ how many _ galaxies have had their outer gas halos removed sufficiently long ago for gas content to have fallen below the stability threshold , while the ratio of quiescent to passive galaxies is driven by _ how early _ the halos were removed .
it is reasonable to expect , and consistent with the models of feldmann and cen , that the two trends should go together .
beginning with rates inferred from hydrogen recombination lines , we have recalibrated star formation rates based on a combination of near uv plus mid ir flux , taking account of the fact that even old stellar populations have some flux in each of these bands . using this , we examined the reliability of the largest available set of star formation determinations , that of sdss galaxies by b04 , as later modified and applied to the sdss dr7 data release .
we showed that the b04-dr7 values are generally quite good , but with systematic problems at low star formation rates .
however , combining recalibrated b04-dr7 values with rates derived from the near
uv vs g band colors , we demonstrated that we can reliably measure specific star formation rates down to @xmath93 . after dealing with selection effects ,
which bias optically selected galaxy samples against low ssfr objects , we have constructed a sample of about @xmath94 objects drawn from the b04-dr7 sample , which is biased in mass but unbiased in ssfr . this sample contains a very large number of galaxies with detected specific star formation rates well below that of the main sequence .
the ssfr distribution of the sample suggests that the low ssfr objects might form a distinct set rather than the low tail of the main sequence , and we divide the sample into three subsets : _ main sequence _ , low ssfr objects , which we denote as _ quiescent _ , and _ passive _ objects with no detected star formation .
we suspect that the passive and quiescent objects form a continuum , but since we are unable to resolve it we treat them as members of separate groups .
we use measures of the galaxies environments from tempel ( 2012 ) , determinations of disk and bulge mass from mendel ( 2014 ) , and gas content from the alfalfa and gass surveys , and we find the following trends among these populations : * averaged over all masses and environments , main sequence , quiescent , and passive galaxies each comprise about one third of the total galaxy population . *
disk dominated ( @xmath95 galaxies are found only among main sequence galaxies , but few main sequence galaxies are bulge dominated ( ( @xmath62 ) * the fractions of main sequence , quiescent , and passive galaxies are sensitive to environment and mass , main sequence fractions falling and quiescent and passive fractions rising at higher mass and density * the mass and environment sensitivity of the three ssfr groups seem to depend on only one physical variable : the ratio of quiescent to passive galaxies is a tight function of the fraction of main sequence galaxies .
* there is a tight @xmath96dex correlation of ssfr with gas content ; compared to main sequence galaxies , quiescent galaxies are overabundent in gas .
these trends place significant constraints on the histories of star formation in disk galaxies .
we have shown , however , that one item- the existence of a tight correlation of mass with star formation rate , i.e. the main sequence- does _ not _ provide much constraint , and that any ensemble of star formation histories that satisfy two conditions- varying smoothly with time and decreasing at the present epoch- will display such a `` main sequence '' .
recent work on the evolution of star formation has been dominated by one concept , that of the `` quenching '' of star formation .
this is , perhaps , an unfortunate choice of terminology because `` quench '' is an ambiguous verb , which can be either transitive or intransitive , but is usually thought of as the former .
star formation quenches in a galaxy when star formation ceases , but that does not imply that it is quenched by an outside agency : the expulsion of gas by an agn , the stripping of gas by ram pressure , or some other active force .
the most basic reason why star formation ceases is that a galaxy runs out of gas , and the most basic reason why a galaxy runs out of gas is that it has turned all of it into stars .
unless provided with an endless new supply , that will happen to any galaxy , whether or not it is subjected to any additional gas removal process , as ltc80 pointed out long ago .
multiple pieces of evidence presented above suggest that the evolution of disk galaxies , in almost every environment , is driven by nothing more than gas exhaustion . the lack of disk dominated quiescent and passive galaxies rules out any galaxy histories- stripping for example- in which the cessation of star formation is unconnected to the processes which built the galaxy s structure .
furthermore , the tight correlation of gas content and star formation rate , combined with the existence of a very large population of quiescent galaxies requires that the process behind gas removal acts very slowly , much slower than either ram pressure stripping or agn - driven gas expulsion are likely to work .
since both of these processes work best when the gas density is low ; they should , once started , drive the gas content to zero , rather than to the low but still significant levels seen in the quiescent galaxies .
we have shown , using a simple but not unrealistic model for disk galaxies , that the properties of main sequence and quiescent galaxies are naturally explained by the observed dependence of star formation rate on density of gas in the disk .
the evolution off the main sequence is governed by the schmidt - kennicutt law , and the pile up of galaxies in the quiescent region is the result of the depression of star formation rates when disk gas density falls below a critical level , possibly set by the toomre disk stability criterion .
the only thing needed to start this evolution is for the infall of new gas into the disk cease , or at least drop to such a low level that it is unable to replace the gas that is locked up into new stars . in this picture , the evolution of disk galaxies is driven not by the ( transitive ) quenching of star formation due to some gas removal process , but rather by the ( transitive ) quenching of gas infall into disks , which precedes , by some billions of years , the ( intransitive ) quenching of star formation .
the cutoff in gas infall provides a natural explanation for the environmental dependence of galaxy populations seen in figure 8 , since models show that such a cutoff tends to occur at earlier times in denser environments .
the structure of disk galaxies present other challenges .
if our model is correct , the correlation of bulge to disk ratio with ssfr , and therefore with disk evolutionary state implies a coherence between the bulge building phase of galaxy evolution and the epoch at which gas infall ends .
this may be so , but further consideration is beyond the scope of this paper .
finally , we should comment on the connection between these findings and the log normal model for star formation histories which we have previously described ( g13 , abramson 2015 , 2016 ) .
that model is characterized by star formation rates which peak at intermediate epochs and are characterized by two independent timescales .
the star formation histories in figure 14 share these characteristics , but not a precise log normal shape .
this is not surprising , for several reasons .
the observations which g13 fit were unable to detect the low ssfr levels of the quiescent galaxies and therefore do not include that phase in their evolution . also , the shapes of the star formation histories which we have constructed in this paper obviously depend on our choice of infall histories ; the tophat form which we have used is unlikely to be an accurate description of actual infall
. we are confident that we could more closely duplicate log normal histories by tuning the infall histories , but see little to be gained by such an effort .
whether the star formation histories of galaxies are accurately or only approximately log normal in form remains an open question , but the existence of several timescales for galaxy evolution appears certain .
b.v . acknowledges the support from an australian research council discovery early career researcher award ( pd0028506 )
the largest published set of star formation rates for nearby galaxies comes from brinchmann ( 2004 , b04 ) , and is based on the @xmath97 fiber spectroscopy from the sloan digitial sky survey ( ) .
b04 first used the observed emission line strengths to divide the galaxies into 5 spectral categories based on line strength and the likelihood that the spectrum is contaminated by an agn .
the spectral energy distributions of objects with detectable emission lines and no agn component were fit to a grid of stellar population models with a range of metallicity , ionization properties , and dust content , from which the star formation rate was derived .
the sfr s of all other objects were derived from the strength of the 4000 angstrom break , d4000 , by an empirical relation between break strength and sfr .
since the @xmath97 fibers only sample a fraction of the typical galaxy , b04 apply empirically derived aperture corrections based on the color distributions of the galaxies in sdss imaging .
salim ( 2007 ) derived star formation rates for a very similar sample of sdss galaxies using ultraviolet and optical photometry , and argued that the b04 sfr s of galaxies with no detected emission lines , which were based on the d4000 index , were systematically too high by several orders of magnitude . in response to this ,
brinchmann and collaborators recalculated star formation rates , also basing them on the newer sdss data release 7 .
we will use these values , which we denote as b04-dr7 , in all of the following analysis .
brinchmann have probably done as careful and thorough a derivation of star formation rates as is possible from the sdss data , but the reliability of the results is inevitably limited by the nature of that data : color distributions plus moderate s / n optical spectroscopy over a portion of the galaxy .
as we shall see , the random errors in the b04-dr7 sfr s of vigorously star forming galaxies are small , but the risk of systematic errors remain , particularly at the lower star formation rates which are our particular interest . to calibrate these possible errors we need independent , accurate star formation rates for a representative sample of the b04-dr7 objects .
since aperture corrections are a significant worry , the sample must , among other things , be representative of the redshift distribution of the b04-dr7 sample , most of which lie in the range @xmath98 .
all measures of star formation rates are indirect , but probably the best and least indirect are those based on the hydrogen recombination lines ( kennicutt 1998 , k98 reviews all of the popular methods ) .
sufficiently accurate measures of at least two lines allow one to determine the extinction corrected ionizing flux within hii regions , from which , with a few rather safe assumptions , the mass of young stars can be obtained .
( applying the method to entire galaxies is not free from error : if extinction with a galaxy is large and variable , the observed flux will come preferentially from the least
extincted regions and the average extinction correction will be underestimated . )
the most accurate emission line measurements of a suitable galaxy sample are probably those of moustakis & kennicutt ( 2006 , mk06 ) , which are derived from high signal to noise spectral scans over the entire face of a set of nearby galaxies .
we have derived luminosities from the mk06 and fluxes after correcting for galactic extinction using schlafly & finkbeinder ( 2011 ) .
star formation rates are obtained using equation 2 of k98 , but adjusted for a kroupa ( 2001 ) imf : @xmath99 .
these galaxies are not at all representative of the b04-dr7 sample : they are all at quite low redshift and are all vigorously star forming , however we can use them to calibrate another star formation measure which we can apply to a suitably large fraction of the b04-dr7 sample .
the most suitable such measure is that based on a combination of ultraviolet and mid infrared flux . because they are photometric rather than spectroscopic measures , these fluxes encompass the entire bodies of galaxies rather than the smaller regions usually sampled spectroscopically
. also , because the uv is a measure of that fraction of the ( reprocessed ) ionizing hot star radiation which escapes the galaxies , while the mid ir is a measure of the fraction that is absorbed by and warms the galactic dust , a suitable combination should be ( relatively ) immune to extinction within the galaxy .
there are several ways of combining these two fluxes ; we follow the method applied by kennicutt ( 2009 , k09 ) , and simply assume that @xmath100 , where @xmath101 and @xmath102 are the luminosities in the uv and ir bands , solving for those values of @xmath103 and @xmath104 which minimize the scatter between @xmath105 and @xmath106 .
we use the subset of normal galaxies from mk06 which is complied in table 2 of k09 . for the infrared flux
we take the iras ( rice 1988 , soifer 1989 , moshir 1009 ) 25@xmath107 fluxes quoted by mk06 . for the uv we take fluxes in the near uv band from the galex nearby galaxy atlas ( gil de paz 2007 ) , corrected as before for galactic extinction .
the scatter between and uv - ir derived star formation rates is minimized if we adapt @xmath108 the agreement between and uv - ir star formation rates for the mk06 sample is presented in figure a1a .
the relation appears quite linear , with a scatter @xmath109 . to apply equation a1 to the b04-dr7 sample we take 2 m magnitudes from the wise all sky data release ( wright 2010 , cutri 2012 ) , denoted w4
a comparison of wise magnitudes and iras fluxes for objects observed by both shows that @xmath110 . for k corrections
we use the 24@xmath107 k corrections from rieke ( 2009 ) . near
uv magnitudes are taken from the galex all sky imaging survey data release 5 ( bianchi 2011 ) .
we will also make use of the 2mass ( skrutski 2006 ) ks magnitudes for our galaxies which are included in the wise date release .
we correct k and nuv magnitudes for galactic extinction using schlafly & finkbeinder ( 2011 ) and apply k corrections from chilingarian & zolotukhin ( 2012 ) . before proceeding with our recalibration of the b04-dr7 star formation rates
, we must consider whether we can expect equation a1 to be applicable at all levels of star formation , or more precisely , at all levels of _ specific _ star formation , since that is the quantity that we will be most interested in , and it is at very high or very low values of ssfr that equation a1 is most likely to break down .
figure a1b presents the relation for the mk06 sample ; it is quite linear over the range sampled , but that range only encompasses the `` main sequence '' .
there is reason to expect the relation to fail at low levels of uv and ir flux . as schiminovich ( 2007 )
have pointed out , old stellar populations as well as young populations contain hot stars- post agb stars for example- which will produce uv flux that is not indicative of star formation .
the same is true in the mid infrared , since most dust heating comes from the same hot stars which produce the uv ( wang & heckman 1996 ) .
given the uncertainties in the hot star populations of old stellar populations , it is safest to estimate this effect empirically .
figure a2 plots the distribution of two colors nuv - k and k - w4 , for b04-dr7 galaxies with @xmath111 and well detected w4 .
these two colors compare the uv and ir indicators of hot stars to the cool stellar component which dominates in the k band .
there is a trend in the expected direction , but the distribution is very broad , presumably due to the effect of internal extinction on the nuv flux
. also plotted , as black points , is a set of galaxies which are expected to have minimal star formation : ellipticals in which the atlas3d survey ( young 2011 ) detected no molecular gas .
with a few exceptions- which may be true star
forming galaxies- these points cluster tightly around the region @xmath112 , @xmath113 .
we take these values to represent the uv and ir emission due to old stars , and therefore can produce uv and ir magnitudes corrected for this contamination as : @xmath114 beyond @xmath112 and @xmath113 @xmath115 and @xmath116 are set to arbitrarily large values . from these
we calculate uv and ir luminosities to which we apply equation a1 and we compare , in figure a3 , specific star formation rates so derived with those of b04-dr7 . for reasons we will discuss shortly , we limit the sample to objects with disk inclination angles ( from simard 2011 ) less that @xmath117 .
the solid line has @xmath118 the scatter , @xmath119 , about this line ranges from 0.18 at @xmath120 to 0.56 at @xmath121 .
thus , the b04-dr7 star formation rates are quite good for vigorously star forming galaxies , but deteriorates significantly at lower specific star formation rates .
this is hardly suprising : many of the latter had no detected emission lines , and star formation rates were estimated from d4000 .
this comparison with b04-dr7 is inadequate in one way : the depth of the wise photometry in insufficient to span most of the redshift range of that sample .
the median redshift of objects in figure a3 is only 0.025 , at which the 3sdss fibers only cover an area 3kpc in diameter .
thus , figure a3 provides a stringent test of random errors in the b04dr7 extrapolations , but the limited redshift range is insensitive to systematic errors . to test for such errors we turn to the swire survey ( londsdale 2003 ) , which used spitzer to survey 6 small fields in the two galactic caps- three within the b04-dr7 area- to much greater depth .
intercomparison between the swire 24@xmath107 photometry and the wise and iras photometry show that @xmath122 .
we then apply equations a1 - a4 to the data , and compare the @xmath123 and uv ir specific star formation rates versus redshift in figure a4 .
galaxies with @xmath124 , which have a smaller scatter , are shown as filled circles , others as open circles .
there is no apparent trend with redshift , implying that the b04-dr7 aperture corrections have no systematic errors .
the above presentation confirms that , once a small correction is applied , the b04-dr7 specific star formation rates are remarkably accurate , and free of systematic errors .
the ssfr values of vigorously star forming galaxies are sufficiently accurate for our needs ; however , random errors do increase significantly at lower levels of star formation .
ideally , we would like to keep @xmath119 less than a few tenths at all detectable levels of star formation , but to achieve this we need another accurate and independent measure of star formation .
@xmath125 would be ideal , but adequately deep mid infrared photometry does not exist over most of the b04-dr7 sample volume .
however , galex uv photometry does go deep enough , and we shall examine one ssfr measure , based on nuv - g colors .
we take nuv magnitudes from galex , and g magnitudes from b04 , correcting , as before , for galactic extincton , k corrections , and the uv flux from old populations following equation a2 .
figure a5 presents the relation between @xmath126 and @xmath127 for the same sample used in figure a3 .
the solid line is a best fit for @xmath128 , and is of the form @xmath129 given the sensitivity of uv flux to galactic extinction , it might seem puzzling how tight this correlation is .
the reason , we surmise , is that both galactic star formation rate and dust content are well correlated with gas content .
the trend of observed nuv - g color with ssfr is the sum of two factors : extinction and intrinsic nuv - g color , both of which are well - correlated with ssfr .
we now define our ssfr measure @xmath130 and , in figure a6 , compare this with @xmath125 .
the two clouds of points to the left of @xmath131 include objects for which @xmath132 ; those in the lower cloud also have either @xmath133 or @xmath134 , or both .
it is clear from this plot that we should assume that all values of @xmath135 less than -12.2 should be regarded as upper limits .
next , we must revisit the issue of internal galactic extinction .
b04 attempt to correct for extinction as part of their analysis , we have argued above that our fit of @xmath127 versus @xmath125 automatically accounts for extinction , and the ir+uv method is designed to be free of extinction effects .
thus , we would expect internal extinction to be unimportant , but figure a7 shows that this is not true .
we plot , for the galaxy sample used in previous figures , the median value of ssfr vs disk inclination angle for objects with @xmath136 .
it appears none of the methods properly account for extinction .
b04-dr7 do the best job and nuv - g color does the worst , but even the ir+uv method has serious extinction problems .
it is easy to understand why this is so : the sum of ir and uv flux may , _ in the mean _
account for all hot star ionizing flux , but only in the mean .
the reradiation of absorbed uv flux by dust grains is a form of ( energetic ) scattering : flux is removed from the line of sight and reradiated isotropically .
thus , at high inclination angles more energy is removed from the line of sight in the uv than is returned to the line of sight in the ir , and equation a1 will underestimate the total star formation rate . to correct for this ,
we fit the median of our final ssfr measure : @xmath137 we cut our sample at @xmath138 to minimize the uncertainty from very large corrections and apply this correction to our ssfr values .
the final scatter in @xmath139 at fixed @xmath135 ranges from 0.13 at @xmath120 to 0.16 at @xmath140 , quite satisfactorily small . | we reexamine the systematic properties of local galaxy populations , using published surveys of star formation , structure and gas content .
after recalibrating star formation measures , we are able to reliably measure specific star formation rates well below that of the so called `` main sequence '' of star formation vs mass .
we find an unexpectedly large population of galaxies with star formation rates intermediate between vigorously star forming main sequence galaxies and passive galaxies , and with gas content disproportionately high for their star formation rates .
several lines of evidence suggest that these _ quiescent _ galaxies form a distinct population rather than a low star formation tail of the main sequence .
we demonstrate that a tight main sequence , evolving with epoch as it is observed to do , is a natural outcome of most histories of star formation and has little astrophysical significance , but that the quiescent population requires additional astrophysics to explain its properties . using a simple model for disk evolution based on the observed dependence of star formation on gas content in local galaxies , and assuming simple histories of cold gas inflow ,
we show that the evolution of galaxies away from the main sequence can be attributed to the depletion of gas due to star formation after a cutoff of gas inflow .
the quiescent population is composed of galaxies in which the density of disk gas has fallen below a threshold for disk stability .
the evolution of galaxies beyond the quiescent state to gas exhaustion and a complete end of star formation requires another process , probably wind driven mass loss .
the ssfr distribution of the quiescent and passive galaxies implies that the timescale of this process must be long , greater than a few gyr , but less than a few tens of gyrs .
the environmental dependence of the three galaxy populations is consistent with recent numerical modeling which indicates that cold gas inflows into galaxies are truncated at earlier epochs in denser environments . |
we thank dr e. a. cornell for making his data available to us prior to publication | we present a semiclassical two fluid model for an interacting bose gas confined in an anisotropic harmonic trap and solve it in the experimentally relevant region for a spin polarized gas of @xmath0rb atoms , obtaining the temperature dependence of the internal energy and of the condensate fraction .
our results are in agreement with recent experimental observations by ensher .
# 1#1 bose
einstein condensation ( bec ) has recently been realized in dilute vapours of spin polarized alkali atoms , using advanced techniques for cooling and trapping@xcite .
these condensates consist of several thousands to several million atoms confined in a well which is generated from nonuniform magnetic fields .
the confining potential is accurately harmonic along the three cartesian directions and has cylindrical symmetry in most experimental setups .
the determination of thermodynamic properties such as the condensate fraction and the internal energy as functions of temperature is at present of primary interest in the study of these condensates@xcite .
the nature of bec is fundamentally affected by the presence of the confining potential@xcite and finite size corrections are appreciable , leading for instance to a reduction in the critical temperature@xcite .
interaction effects are very small in the normal phase but become significant with the condensation
induced density increase .
the correction to the transition temperature due to interactions has been recently computed by giorgini @xcite .
the temperature dependence of the condensate fraction was recently measured@xcite for a sample of around 40000 @xmath0rb atoms , the observed lowering in transition temperature being in agreement with theoretical predictions within experimental resolution . in the same work the internal energy was measured during ballistic expansion and found to be significantly higher in the bec phase than predicted by the ideal gas model . while the increase is easily understood as a consequence of the interatomic repulsions ,
a quantitative estimate is still lacking . in this work
we present a two fluid mean field model which is able to explain the above mentioned effects , giving results in agreement with experiment for both the condensate fraction and the internal energy as functions of temperature .
we describe the condensate by means of the gross pitaevskii ( gp ) equation for its wave function @xmath1 , @xmath2 where @xmath3 , @xmath4 being the scattering length , @xmath5 is the confining potential and @xmath6 is the average non condensed particle distribution .
the factor @xmath7 in the third term arises from exchange@xcite and we neglect the term involving the off diagonal density of non condensed particles .
following bagnato @xcite we treat the non condensed particles as non interacting bosons in an effective potential @xmath8 .
thermal averages are computed with a standard semiclassical bose
einstein distribution in chemical equilibrium with the condensate , i.e. at the same chemical potential @xmath9 . in particular , the density @xmath6 is @xmath10 we fix the chemical potential from the total number of particles @xmath11 @xmath12 where @xmath13 and the semiclassical density of states is @xmath14 this completes the self consistent closure of the model
. equation ( [ eqgp ] ) can be solved analytically in the experimentally relevant situation @xmath15 and @xmath16 , where @xmath17 . except for a small region close to the phase transition
the interaction parameter @xmath18 entering the gp equation is large and the kinetic energy can be neglected .
this yields @xmath19 where @xmath20 ( @xmath21 ) for @xmath22 ( @xmath23 ) .
the present strong coupling solution neglects the condensate zero point energy @xmath24 .
as giorgini , pitaevskii and stringari@xcite pointed out finite size effects are thereby excluded . before presenting the complete numerical solution of the self consistent model defined by equations ( [ eqserien1])([eqpsir ] ) and comparing its predictions with existing experimental data@xcite , we display perturbative solutions at zero and first order . an approximate semi
analytical solution can be obtained by treating perturbatively interactions involving the `` dilute gas '' of non condensed particles . to zero order in @xmath25
we have @xmath26 and equation ( [ eqrhodef ] ) gives @xmath27 \label{eqrhoc}\ ] ] for @xmath28 and @xmath29 for @xmath30 .
the self consistent zero
order solution is then completed by equation ( [ eqntot ] ) .
we remark that no assumption of weak interactions _ within the condensate _ has been made .
we now proceed to compute the first order correction to the above zero order solution .
we take @xmath31 and expand equation ( [ eqrhodef ] ) to first order in @xmath32 $ ] .
the choice of the expansion parameter @xmath32 $ ] ensures that the perturbative expansion is regular , since the correction to @xmath33 vanishes where the zero order term vanishes . with the additional approximation @xmath34 in the first order term
we get @xmath35 where @xmath36\ ] ] @xmath37 being the kummer confluent hypergeometric function .
the self
consistent first
order solution is then completed by equation ( [ eqntot ] ) .
we have solved numerically the simplified two fluid model , first treating the parameter @xmath38 to all orders ( equations [ eqserien1][eqpsir ] ) , then to zero order ( equations [ eqntot ] and [ eqn0mu][eqn0sopratc ] ) and finally to first order ( equations [ eqntot ] and [ eqn0approx][eqrho1e ] ) .
each case involves solving the integral equation ( [ eqntot ] ) to obtain @xmath9 as a function of @xmath11 and @xmath39 ; the non perturbative solution also involves the local nonlinear problem posed by equations ( [ eqserien1 ] ) and ( [ eqpsir ] ) . the small differences between our three results justify _ a posteriori _ a perturbative treatment .
experimental parameters are taken from the work of ensher @xcite : @xmath40 , @xmath41 and @xmath42 .
we have verified numerically that our results depend weakly on @xmath11 in the region explored in the experiments and therefore have used a fixed @xmath41 in all our computations .
we use as energy units the semiclassical ideal gas critical temperature @xmath43 , @xmath44 being the riemann zeta function .
figure [ fign0 ] compares the temperature dependence of the condensate fraction @xmath45 with the experimental results of ensher @xcite .
lowering of the transition temperature due to interactions is clearly visible , even if the smoothness of our results around the transition prevents a precise assessment of an interaction induced shift in @xmath46 from the numerical solution . it should be noticed that the strong coupling solution of the gp equation is not valid for @xmath39 close to @xmath46 , since it requires @xmath47 , and that our mean field model does not include critical fluctuations .
both effects being relevant only in a narrow window around @xmath46@xcite , we expect our results to be meaningful in most of the temperature range .
recently giorgini @xcite solved numerically the popov approximation to the finite temperature generalization of the gp equation within a semiclassical wkb approximation .
their results for the temperature dependence of the condensate fraction are in very good agreement with the predictions of our more naive model except for @xmath48 , where they find a sharp change in the slope of @xmath49 .
their result for the interaction induced shift in critical temperature @xmath50 is also in good agreement with our curves .
figure [ figeqst ] reports our results for the temperature dependence of the internal energy .
we remark that the experimentally measured quantity is the sum of the kinetic energy and of the interaction energy , not including the confinement potential energy due to the rapid switching off of the trapping potential@xcite .
the average single particle energy @xmath51 - 1\right\ } $ ] obtained from the semiclassical density of states contains twice the interaction energy , and assuming that on average the kinetic and potential terms are equal is twice the measured quantity .
the kinetic energy of condensed atoms is negligible in our strong coupling limit and their interaction energy per particle is @xmath52 .
the quantity directly comparable to the experimental data is therefore @xmath53 , which we plot in figure [ figeqst ] obtaining good agreement with the measured values .
the calculated internal energy does not contain any sharp feature at transition , paralleling the result discussed above for the condensate fraction .
correspondingly the rapid rise in the specific heat is considerably smoothed with respect to the ideal
gas result ( see figure [ figcv ] ) . apart from this small region around transition , our results on @xmath54 above and below @xmath46 imply a significant reduction of the increase in specific heat across the phase transition . in conclusion , we have presented a mean field , semiclassical two fluid model and discussed its perturbative and non perturbative solution in the experimentally relevant parameter range .
our results on the temperature dependence of the condensate fraction and of the internal energy are in agreement with recent experimental measurements , accounting for the pronounced increase in internal energy with respect to the noninteracting boson case measured below @xmath46 .
we have also verified that our model reproduces the results obtained for the condensate fraction with a more refined theory by giorgini . |
there is widespread hope that the next few years will see the detection of a dark matter particle .
signals of dark matter are searched for obliquely in colliders , directly in underground detectors , and indirectly in cosmic rays .
of these three methods , the last one is _ a priori _ the most challenging because backgrounds from astrophysical processes are difficult to estimate .
nevertheless , indirect detection is a topic of great current interest due to rapid instrumental progress and a constant flow of new and exciting data . during the past year , interest in indirect detection has been further amplified by tantalizing signals from the pamela @xcite , atic @xcite and fermi @xcite experiments , which may or may not be hints of dark matter .
the main search channels for indirect dark matter detection are cosmic ray photons , antiprotons , positrons , and neutrinos .
certain features of neutrinos make them an especially clean detection channel .
neutrinos , like photons , have no electric charge and traverse our galaxy along straight paths . therefore the flux of neutrinos observed at the earth is determined in a straightforward way by the production mechanism and is free of the astrophysical uncertainties which plague the propagation of charged particles .
neutrinos have a further advantage over photons in that astrophysical neutrino backgrounds are far less severe .
detection of a large diffuse flux of astrophysical neutrinos would thus constitute a smoking gun for dark matter . in spite of these attractive features ,
neutrinos are seldom singled out as a primary dark matter discovery channel .
neutrinos are weakly interacting and require large , dedicated experiments to detect .
it is much easier to detect charged particles and photons .
typical models of dark matter encountered in the literature predict a cosmic neutrino flux that is smaller than or comparable to the fluxes in other observable channels .
most often , dark matter dominantly annihilates or decays into other standard model particles@xmath5 bosons , for instance , or tau leptons which produce neutrinos when they subsequently decay .
but then one would expect to first detect that dark matter signal by observing antiprotons from @xmath5 decays or gamma ray photons from @xmath6s in the tau lepton decay chain .
annihilation or decay to primary , hard neutrinos typically appears as only one of several possible final states .
the @xmath7 gauge symmetry of the standard model generically implies that charged particles are produced at comparable rates , and those lead to more accessible signals . for these reasons measurement of the cosmic neutrino flux
sometimes provides useful constraints on dark matter models @xcite , or may serve to break a degeneracy among different models once a signal is detected in another channel , but is not considered the main discovery channel . in this paper
we argue that one can construct dark matter models which predict a cosmic neutrino flux that is more prominent than the antiproton , positron and gamma ray fluxes .
if such a model is realized in nature , the indirect detection signal could first show up in neutrino telescopes .
the neutrino flux from the galactic center can well be larger than the atmospheric neutrino background , leading to a sharp smoking - gun signature .
the current generation of neutrino telescopes ( icecube and antares ) or the future one ( km3net ) may become the discovery machines .
what kind of set - up is required for dark matter to produce a large neutrino flux ?
it is plausible that the dark matter particle is a perfect singlet under the standard model gauge interactions . if such is the case it can couple to the visible sector via gauge singlet operators which can be constructed out of standard model fields .
the lowest dimensional singlet operators are the higgs mass @xmath8 and the hypercharge field strength @xmath9 , both of mass dimension 2 .
dark matter coupling through the higgs portal or via kinetic mixing with hypercharge has been explored in numerous models . however , if the dark matter particle is a _ fermion _ , then the lowest dimensional standard model operator it can couple to is the dimension 5/2 operator h l_ab h ^a
l ^ b = h ^ + e^-_l - h ^0*_l = g^+ e^-_l- ( v+h+ig ^ 0)_l where @xmath10 gev is the standard model higgs vev .
we refer to this operator as _ the neutrino portal_. now , suppose that the fermionic dark matter particle @xmath11 is a part of a larger dark sector that is neutral under the standard model gauge group .
assume that the dark sector couples to the standard model via the neutrino portal as [ e.npc ] _ int = o_dark ( ) ( l h ) where @xmath12 is a fermionic gauge singlet operator constructed from the dark matter particle @xmath11 and other fields in the dark sector .
suppose in addition that the dark sector contains _ bosonic _ states that are lighter than @xmath11 .
if @xmath11 is the lightest _ fermionic _ state in the dark sector , it is stable within the dark sector .
but in the presence of the coupling eq .
( [ e.npc ] ) the dark matter particle is allowed to decay into fermionic states in the standard model .
the leading dark matter decay channel is then + which leads to a neutrino flux as the dominant signal from dark matter decay ! moreover ,
if there is only one dark sector particle in the final state , the neutrino flux is monochromatic . of course , the coupling in eq .
( [ e.npc ] ) also predicts other , subdominant decay channels : @xmath13 , which imply charged particle fluxes . but
these other decays have one more particle in the final state , and are therefore suppressed by phase space .
the branching ratio for these states with on - shell bosons depends on the multiplicity of dark particles in the final state and the mass of the dark matter particle , and can be made as small as @xmath14 .
cosmic neutrinos then become the most prominent signal of dark matter decay .
we present here a simple model with a strongly coupled dark sector which explicitly realizes this idea .
the model contains an @xmath0 dark gauge group plus a massive majorana fermion in the adjoint representation . with this low energy spectrum
the dark gauge group is asymptotically free for any @xmath15 .
therefore the dark color is confined at low energies and the physical states are dark color singlets : glueballs and glueballinos ( bound states of the adjoint fermion and gluons ) .
the lightest , ground state glueballino ( in the following simply called the glueballino ) is our dark matter candidate .
note that the glueballino , being fermionic , can not decay solely into glueballs , even if it is heavier .
we assume that the dark confinement scale is in the ( sub)-gev ballpark which sets the mass scale for the glueballs , while the glueballino mass is set by the majorana mass of the adjoint fermion of order 1 - 5 tev . in the absence of any coupling between the dark sector and the standard model
the glueballino would be perfectly stable .
decay channels open up once the adjoint fermion is coupled to the standard model via the neutrino portal , that is , via the dimension six coupling @xmath16 \sigma_{\mu\nu } ( l h)$ ] . for @xmath17 gev and @xmath18 ( a one - loop factor ) the dark matter lifetime
is much longer than the age of the universe , while at the same time the decay rate is large enough to yield observable signals .
the leading decay of dark matter is the two - body decay into a dark gluon and a neutrino .
this results in a flux of monochromatic neutrinos from the galactic center which can be detected in the current generation of neutrino telescopes , antares and icecube .
other indirect detection signals are suppressed , so that neutrino telescopes are singled out as the most sensitive probes of our scenario .
moreover , the cross - section of our dark matter particle on nuclei is small , giving null predictions for direct detection experiments .
the paper is organized as follows . in section 2
we introduce our example model of the dark sector and discuss the spectrum after the dark gauge group confines . in section 3 we study the thermal relic abundance , including the sommerfeld effect and late annihilation of glueballinos . depending on assumptions about modeling nonperturbative contributions to the annihilation cross - section , the correct relic abundance may be obtained for dark matter masses in the range 1 - 10 tev provided the strong coupling scale is sufficiently small . besides the sizable theoretical uncertainties due to modeling the strongly coupled effects at energies below the confinement scale , the precise value of the mass is also sensitive to the number of degrees of freedom in the hidden sector the larger the number of colors , the smaller the required dark matter mass . in section 4
we compute the decay width and the branching ratio for dark matter decays due to the coupling via the neutrino portal .
we demonstrate that for the dark matter mass close to a tev the neutrino decay channel is by far dominant , while other decay channels have branching ratios at a few percent level . for heavier dark matter ,
close to 5 tev , the decay rate into charged standard model particles becomes comparable to the decay rate into neutrinos . in section 5
we make a slight detour to explain why the glueballs in the dark sector can not be stable .
we enumerate various options for their decay into the standard model final states and indicate options for model building .
glueball decays may do lead to additional indirect detection signals which however are highly model dependent .
for this reason we choose to focus on the more robust signals from the glueballino decays via the neutrino portal . in the later sections of the paper we turn to discussing observational consequences of the general scenario where structure in the dark sector enables dark matter decay to yield a monochromatic neutrino flux as its primary signal . in section 6
we determine the parameter space consistent with indirect detection experiments .
the super - kamiokande bounds on the neutrino flux from the galactic center set a lower bound on the dark matter lifetime , or equivalently on the scale @xmath19 which suppresses the coupling through the neutrino portal .
that scale turns out to lie in the @xmath20 gev ballpark
the typical scale for lepton violating interactions .
we also study the current bounds from the antiproton , positron and gamma ray fluxes .
we conclude that even for the maximum lifetime allowed by super - kamiokande the indirect bounds from other detection channels are less constraining in a large portion of parameter space .
one exception are the bounds from antiprotons , which are produced in subdominant dark matter decay channels with standard model gauge and higgs boson in the intermediate state .
these bounds are comparable to those from super - kamiokande for dark matter masses around 1 tev , and when the dark matter mass is close to 5 tev ( and thus the branching ratio for decays into charged particles becomes large ) the antiproton flux measured by the pamela experiment implies a more severe constraint on the dark matter decay rate than the super - kamiokande bounds .
however , one should note that uncertainties involved in the propagation of antiprotons in our galaxy are very large , and by changing the parameters of the propagation model one can reduce the antiproton flux by a factor of five , thus obliterating the antiproton bounds . we therefore continue to refer to the super - kamiokande bounds , which are comparatively free of astrophysical uncertainties . in section 7
we study the discovery prospects in antares and icecube .
we compute the rate of signal events in both detectors induced by the predicted neutrino flux from the galactic center , and compare it to the background rate from atmospheric neutrinos .
we conclude that neutrino telescopes have a great potential to probe this scenario , and their sensitivity will exceed that of super - kamiokande by 1 - 2 orders of magnitude .
in this section we construct an example of a dark matter model which leads to an enhanced neutrino signal .
the model falls into the _ hidden valley _
class @xcite , where the hidden sector contains fairly light states ( here gev scale ) interacting very weakly with the standard model through higher dimensional operators .
the hidden sector here consists of a @xmath0 gauge group together with a fermion in the adjoint representation , @xmath21 .
incidentally , this is also the field content of pure @xmath221 super - yang - mills theory , but we do not consider supersymmetry here : the fermion has a tev scale mass @xmath23 and the interactions of the hidden sector with the standard model are not supersymmetric .
the dark gauge group is asymptotically free for any @xmath15 so that the theory confines at low energies .
we assume that the confinement scale is in the gev ballpark , well below the adjoint fermion mass .
the low - energy theory below a tev contains only the dark gluons , and after confinement the low - energy degrees of freedom are stable glueballs .
the spectrum of a pure yang - mills theory has been studied on the lattice for the case @xmath24 @xcite .
these studies indicate there are 12 stable glueballs in a pure @xmath25 theory , labeled by their @xmath26 quantum numbers .
the masses of the glueballs are set by the strong coupling scale @xmath27 .
the lightest glueball is the @xmath28 , with @xmath29 where @xmath30 is the `` force radius '' related to the confinement scale @xmath31 ( defined by the running coupling in the ms - bar scheme ) by @xmath32 @xcite .
thus , the lightest glueball mass m_0++ 7 is appreciably larger than the confinement scale .
the spectrum of glueball states for @xmath24 is shown in figure [ fig : glue ] . while lattice results for the glueball spectrum are only available for @xmath33 , the masses of glueballs do not vary appreciably with @xmath15 in the large-@xmath15 limit .
the spectrum of the @xmath33 theory is thus a useful benchmark for more general cases .
gauge theory , after @xcite .
masses are shown in units of the lightest glueball mass @xmath34 .
[ fig : glue],width=377 ] in addition to glueballs , the spectrum of the dark sector contains color singlet bound states of the adjoint fermion and gluons . borrowing from supersymmetric jargon we refer to the lightest of such states as _ the glueballino_.
the glueballino is stable under decays within the dark sector , and is our dark matter candidate .
in the early universe at @xmath35 tev the adjoint fermions and the dark gluons were in thermal equilibrium .
we assume that at even larger temperatures the dark sector was in thermal equilibrium with the standard model , but near a tev the two sectors are already decoupled . below a tev
the adjoint fermions become non - relativistic and drop out of thermal equilibrium , leaving a relic abundance that is in principle determined by the fermion mass , the dark gauge coupling and the number of dark colors . the freezeout calculation of @xmath11 abundance
is closely analogous to the well - studied problem of gluino dark matter @xcite . in the following we estimate the range of fermion masses that are compatible with the observed dark matter abundance .
we first compute the relic abundance ignoring non - perturbative effects , in which case the standard textbook procedure applies . the perturbative matrix element for the annihilation @xmath36
can be found in @xcite . at low energies
the thermally averaged annihilation cross - section is given by _
ann v where @xmath37 is the dark coupling at the scale @xmath23 , given in terms of the confinement scale and the number of dark colors by @xmath38 . integrating
the boltzmann equation relates the annihilation cross section to the relic abundance @xmath39 .
the latter can be translated into the dark matter density fraction today by @xmath40 where @xmath41 , @xmath42 ; wmap fixes @xmath43 to be @xmath44 @xcite .
the final result strongly depends on the number of dark colors ( the number of degrees of freedom in the hidden sector ) .
our results are plotted in fig .
[ fig : contour ] . in the @xmath45@xmath46 plane , for @xmath24 ( red ) , @xmath47 ( green ) , @xmath48 ( cyan ) , @xmath49 ( blue ) .
using the perturbative annihilation cross - section .
[ fig : contour],width=384 ] nonperturbative contributions to the annihilation cross - section are significant and introduce large uncertainties @xcite .
below we estimate the effects of sommerfeld enhancement and of enhanced annihilation after confinement .
the exchange of dark gluons between the nonrelativistic fermions leads to an enhanced annihilation cross section at low velocity @xcite .
the coulomb interaction of two massive particles takes place on a time scale set by the energy of the scattering particles , @xmath50 .
meanwhile , the time scale for a massive particle to interact with a gluon in the thermal bath , and therefore to randomize its color , is set by its mean free path in the thermal bath .
this can be estimated by @xmath51 . since @xmath52
, the two gluinos will remain in a state of definite total color throughout the interaction , and the sign and strength of their interaction will depend on that particular color state .
we can solve the schrdinger equation in the basis of definite total color , and derive the sommerfeld enhancement in each color representation subspace @xmath53 .
two particles of equal mass moving in a non - abelian coulomb potential have a scattering amplitude enhanced from the plane wave result by the factor [ eq : somm ] e(v_r ) = , where @xmath54 is the _ relative _ velocity of the scattering particles , and @xmath55 depends on the quadratic casimir operators of the single - particle representations @xmath56 and @xmath57 and the total representation @xmath58 , = - ( c_2 ( r ) -c_2 ( r_1)-c_2 ( r_2 ) ) . the annihilation cross - section during freeze - out
is then multiplied by the factor [ eq : prescription ] e_ann(v_r ; \ { r } ) = _ i ( r_i ) e_i ( v_r ) , which enhances annihilation in attractive channels and exponentially suppresses annihilation in repulsive channels .
as the gluinos are identical particles , we need to ensure that the total wave function is antisymmetric .
annihilation proceeds dominantly in the @xmath59-wave , where the wave function is spatially symmetric , and therefore the net color and spin wave function must be antisymmetric .
thus for @xmath60 , the total enhancement factor @xmath61 is explicitly e ( v_r ) = ( + 9 - 5 ) incorporating the sommerfeld enhancement has an order @xmath62 effect on the mass of the dark matter particle relative to perturbative freezeout , as demonstrated in figure [ fig : somm ] . as a function of @xmath45 , with ( dashed ) and without ( solid ) sommerfeld enhancement , for @xmath60 .
the solid line indicates @xmath63 .
relic densities are shown for @xmath64 ( red ) , @xmath65 ( green ) , and @xmath66 ( blue ) .
[ fig : somm],width=384 ] more difficult to estimate are the effects of the increased annihilation cross - sections after confinement , when the @xmath67 have hadronized @xcite .
the heavy fermion @xmath11 is localized within a region of order its de broglie wavelength , @xmath68 , while the total radius of the glueballino is of order @xmath69 .
while the _ total _ interaction cross - section between hadronized adjoints should be geometric , @xmath70 , the annihilation cross - section should be significantly smaller .
estimating the annihilation cross - section is a subtle problem .
the core of the question is whether or not two hadronized @xmath11 s in a state with large angular momentum are kinematically able to radiate away that angular momentum and annihilate .
our hidden sector model differs from qcd in that there are no light fundamental quarks , and hence the mass gap in our hidden sector is larger compared to the strong coupling scale than the mass gap in qcd .
we expect that this suppresses the annihilation cross - section in this model , relative to ( e.g. ) gluino freezeout , as the larger mass gap makes it more difficult to radiate away angular momentum .
we therefore model the late - time annihilation cross - section by taking it to saturate the @xmath59-wave unitarity bound , [ eq : np ] _ np .
we expect that this estimate gives an upper bound on the cross - section , and therefore a lower bound on the relic abundance .
this cross - section yields efficient annihilation of hadronized adjoints , lowering the relic density by roughly an order of magnitude relative to the perturbative result .
it is therefore difficult to obtain the relic abundance @xmath71 without going to corners of parameter space , either large @xmath15 or @xmath72 , or small @xmath31 . as we will explain in the following section , the most interesting choice of dark matter masses from the point of view of neutrino signals is the range @xmath73@xmath74
this range is consistent with the perturbative computation of the relic abundance .
however , if the non - perturbative effects enhancing the annihilation cross - section are close to our estimate , this mass range requires either sub - gev @xmath31 or a very large ( @xmath75 ) number of dark colors . building a complete model of the dark sector with @xmath31 significantly less than a gev requires some additional structure in the dark sector to be consistent with all cosmological constraints , as we discuss in section [ sec : poof ] .
the low - energy lagrangian that governs the hidden sector has an accidental @xmath76 symmetry acting as @xmath77 which ensures stability of dark matter .
however , general interaction terms do not have to respect this @xmath76 and may lead to dark matter decay . in this section
we consider the coupling of dark matter to the standard model via the neutrino portal and study possible decay channels . the operator @xmath78 transforms as a fermion under the lorentz group and as a singlet under both the dark and the standard model gauge groups .
therefore it has the right quantum numbers to couple to the standard model through the neutrino portal .
the interaction takes the form [ eq : portal ] _ int = ^a^ g^a _ h l + . here
@xmath79 is a left - handed weak @xmath7 lepton doublet ; for definiteness let as assume that it is the first generation doublet , @xmath80 , while the couplings to the second and third generations are suppressed by a larger scale . from the point of view of neutrino detection it does not matter which generation has the dominant dominant coupling to dark matter : the neutrino oscillations will make sure that all 3 neutrino flavors are equally populated in the neutrino flux from dark matter decay . the interaction breaks the accidental @xmath76 @xmath11-parity and allows the dark matter to decay .
it is a dimension six non - renormalizable operator , but it can be easily obtained from a renormalizable uv theory as follows .
let us introduce right - handed neutrinos @xmath81 which have the usual standard model couplings .
in addition , introduce a pair of fundamentals in the hidden sector : a scalar @xmath82 and a ( dirac ) fermion @xmath83 , which have masses equal to @xmath19 .
the lagrangian = _ kin - ^ a ( q^t ^ a ) -c_i q^n_i - q^q |h|^2 -y_ij^n n_i h l_j + contains the most general gauge - invariant renormalizable operators which can be built from the fields in the model . integrating out the heavy @xmath82 , @xmath83 , and @xmath81 at one loop gives the effective interactions _
int = ^ a^ g ^ a _ h l_l + ^a ^a | h| ^ 2 + g ^ a _
g ^ a | h | ^ 2 + .
the coefficients @xmath84 , @xmath85 , and @xmath86 are of order @xmath87 .
the latter two interactions have negligible effect on phenomenology , but the first is the desired coupling of dark matter to the standard model via the neutrino portal . if the scale @xmath19 is of order @xmath88 gev the typical scale for lepton number violating interactions then the dark matter lifetime is of order @xmath2 seconds .
such a dark matter particle is stable on cosmological time scales and , at the same time , the decay rate may be large enough to allow for detection of dark matter decay products .
the perturbative vertex given by yields two distinct partonic contributions to dark matter decay : first , the @xmath89 process @xmath90 with a ( hidden ) gluon in the final state , and second , the @xmath91 process @xmath92 , with a ( hidden ) gluon in the initial state .
both partonic processes are part of the full gauge invariant amplitude for the decay of the dark matter particle , but , as we will see , the first process is the dominant contribution .
the main contribution to dark matter decay is then the decay of the heavy adjoint @xmath21 via to two - body and three - body final states with a hidden gluon plus standard model neutrinos , gauge and higgs bosons . the hard gluon @xmath93 from the perturbative decay and the remnants of the brown muck that made up the glueballino go on to hadronize into some number of hidden glueballs .
we first consider the two - body decay with one neutrino and one dark gluon in the final state .
the interaction vertex is obtained from by replacing the higgs field by its vacuum expectation value .
the partial decay width is given by [ eq:2bodyg ] ( g ) _
the cp conjugate decay to anti - neutrinos has the same decay width .
next , consider the three - body decay to gluon , neutrino , and on - shell higgs boson .
the partial width for this process is ( g h ) = ( 3 + 44r - 36 r^ 2 - 12 r^3 + r^4 + 12 r ( 2 + 3 r)r ) , where @xmath94 .
again , the cp conjugate decay to anti - neutrinos occurs with the same rate . in the limit @xmath95 ( g h ) = ( g h | ) _ 3b .
in addition , there are three - body decay modes to ( i ) gluon , electron , and @xmath5 boson ; ( ii ) gluon , neutrino and @xmath96 boson . in the limit
@xmath97 these partial widths satisfy the relation ( g z ) = ( g z | ) = 1 2(g w^+ e^- ) = 1 2(g w^- e^+ ) = _ 3b ignoring gauge and higgs masses , the dark matter lifetime is approximately _
, while the ratio of two- and three - body partial widths is [ eq:2to3 ] _ 3b _
2b 0.01 ( m_)^2 varying @xmath98 in the range 1 - 5 tev , the three - to - two - body ratio varies from 1 percent to 30 percent .
therefore the neutrino flux from dark matter decay dominates over charged particle fluxes , especially when the glueballino is not much heavier than 1 tev . apart from the process where the adjoint fermion decays into a dark gluon and standard model fields one can also consider the annihillation of the fermion with a valence gluon .
we argue however that this gives a negligible contribution to glueballino decay .
first , it is readily apparent that the rate for the gluino to completely annihilate with its cloud of gluons is vanishingly small . in order for the glueballino to annihilate into the vacuum ,
all of the momentum distributed among the brown muck must be concentrated in a single partonic gluon which then annihilates with the heavy fermion @xmath11 , or in other words , all of the momentum carried by the gluons must be entirely localized within the compton wavelength of the heavy fermion @xmath11 . the probability for this to occur is negligibly small , of order @xmath99 at most .
moreover , the probability for the operator @xmath100 to connect the initial state glueballino to a single glueball at rest is suppressed .
this process can be thought of as an annihilation of the partonic @xmath21 with a coherent sum of soft gluons inside the hadron .
the residual soft gluons after annihilation must carry enough energy to reform into a stable glueball .
as the glueballs have masses of a factor of a several larger than the strong coupling scale , an annihilation of this form is kinematically forbidden unless the @xmath101 has a virtuality likewise several times larger than the strong coupling scale .
computing the cross - section for @xmath102 and taking @xmath103 , we can estimate ~ .
strictly , as the gluon participating in the interaction is soft we need to integrate this expression over the field configuration of the glueballino .
to obtain an estimate for the annihilation rate we simply estimate the result of this integral as @xmath31 .
this yields an estimate for the rate ~ : this is suppressed relative to the two - body decay rate by a factor of @xmath104 .
however the rate for the full process should be suppressed even further , as the gluino must have emitted a hard gluon before participating in the annihilation in order to be sufficiently off - shell .
this emission suppresses the annihilation process by an additional factor of @xmath37 . in the following
, we will neglect this glueballino decay mode in comparison with the two - and three - body decay modes coming from the decay of the gluino alone .
the hidden sector glueballs can not be stable on cosmological time scales as otherwise they would overclose the universe . in this section
we discuss the thermal relic abundance of glueballs and present several scenarios for their decay .
we will show that there are multiple ways to eliminate the overabundance of glueballs , and will indicate the observable signals associated with each scenario . at the end of the day
it will be clear that the signals and resulting constraints are very model - dependent , in contrast to the sharp predictions from two- and three - body decays via the neutrino portal discussed in section [ s.dmd ] .
therefore , when discussing signals and constraints in the following of this paper we will focus on the more robust signals of the standard model particles coming directly from the neutrino portal operator @xmath105 .
the goal of this section is to point out and characterize a range of options for addressing the glueball problem , and demonstrate that solutions are possible .
readers less interested in model building and more interested in model - independent astrophysical signatures of the neutrino portal are encouraged to jump directly to the next section .
first , we demonstrate why the glueballs must decay . treating hidden sector confinement as adiabatic , the energy and entropy in the thermal gluon bath
are wholly transferred to glueballs after confinement .
if the glueballs do not decay , the present - day energy density in glueballs is then given in terms of the present - day photon temperature @xmath106 by [ eq : gluefoo ] _ glue ( t_0 ) t ^ 3_0 , where @xmath107 is the effective number of relativistic standard model degrees of freedom today , and @xmath108 are the effective number of relativistic degrees of freedom in the dark sector and in the standard model when the two sectors were in thermal equilibrium .
avoiding overclosure then requires the strong coupling scale @xmath31 to be smaller than an ev , ( n ^ 2 - 1 ) 1 .
however , the strong coupling scale @xmath31 is constrained by astrophysical bounds on the dark matter self - interaction cross - section .
analyses of galactic cluster dynamics combine to yield a bound on the total dark matter self - interaction cross - section @xcite , [ eq : xsec ] 0.3 - 1 ^ 2 ^-1 1400 - 4600 ^-3 .
the most stringent such astrophysical bound comes from observations of the bullet cluster , 1e0657 - 56 @xcite . in our model ,
the _ total _ self - interaction cross - section for the glueballino is geometric , @xmath70 . with @xmath109 ,
the bound requires the mass of the glueballino to be very large , @xmath110 , which would be incompatible with thermal relic abundance .
as the glueball masses are intrinsically tied to the confinement scale , there is no way to escape this problem .
the glueballs must therefore decay .
there are then three possibilities : the glueballs can decay within their own sector ; they can decay to the standard model ; or finally , they can decay to yet another sector . adding light flavors to the hidden sector gauge theory
destabilizes the glueballs to hidden sector mesons and opens an interesting range of possibilities for model building , which we discuss below .
another option , which we will pursue here , is to add additional degrees of freedom ( mediators " ) which couple the hidden and visible sector , enabling glueballs to decay to standard model final states .
dark matter decays then yield a model - dependent amount of visible standard model particles from glueball decays in addition to the primary gluino decay , which must be taken into account .
in order to avoid cosmological constraints , it is simplest to ask that the lifetime of at least one species of glueball be less than one second , so that the glueballs can safely decay without a trace before nucleosynthesis . for definiteness
, we will consider two simple models where the mediators couple to the standard model higgs boson , thereby enabling the @xmath111 glueball to decay rapidly through the dimension 6 higgs portal operator to pairs of standard model fermions .
in the first model , we introduce a standard model singlet scalar field @xmath112 transforming as a fundamental under the dark gauge group , with the interaction [ eq : scalar ] _ int = y || ^ 2 |h|^2 . in the second model ,
we introduce a pair of dirac fermions , @xmath113 and @xmath114 , which are fundamentals under the dark gauge group and have the standard model quantum numbers of a left - handed lepton doublet and a right - handed charged lepton , respectively .
the interactions of the new fermions @xmath113 and @xmath114 we take to be @xmath115- and @xmath116-preserving , for simplicity : [ eq : fermion ] l = l_+(y six interaction [ eq : d6 ] l_eff = h^h g^a_g^a . here
@xmath117 is the mass scale of the mediators .
the coefficient @xmath84 is @xmath118 for the scalar mediator model of equation , and @xmath119 for the fermion mediator model of equation .
the @xmath120 glueball decays through the effective interaction with branching fractions dictated by the couplings of the standard model higgs . for the parameter range
we consider , the @xmath111 will decay to pairs of standard model fermions through an off - shell higgs .
the width for this decay is given by [ eq:0width ] _
0^++f|f = ( ) ^2 ^sm_h f|f(m_0 ^
2 ) here @xmath121 denotes the partial width of @xmath122 for a standard model higgs boson with mass @xmath123 , and @xmath124 is the @xmath120 decay constant .
we will use the lattice result for @xmath124 @xcite , f_0^s g^2 0|g^a_g^a|0^++= 6.12 m_0 ^ 3 .
if the @xmath28 is heavy enough to decay to @xmath125 pairs , a lifetime less than one second is readily accomplished with mediator mass scales in the range of tens to hundreds of tev .
lighter glueballs , with masses in the range @xmath126 , are also viable if the mediator mass scales are in the range of a few to tens of tev .
glueballs light enough to decay dominantly to muon pairs are interesting , as this mass range appears best compatible with obtaining the correct relic density for the heavy hadronized adjoint , as well as with possible visible signals .
however , realizing a sufficiently short lifetime for the @xmath127 in this case requires the mediators to be so light that they lie outside the range of validity of our calculations .
models with light standard model - charged mediators are ruled out , but a light scalar mixing with the standard model higgs is entirely possible , and seems like an interesting avenue to pursue . in the case of a scalar mediator , the only interactions induced upon integrating out the heavy mediators are of the form _ eff = g^a_g^a _ k ( h^h)^ k .
the lorentz decomposition of the operator @xmath128 contains only the @xmath28 .
therefore only the @xmath120 can decay directly via the higgs portal to standard model particles .
other glueballs can decay through the @xmath120 channel , e.g. , @xmath129 .
these three - body decay rates are highly suppressed relative to the @xmath120 decay rate due to the smaller available phase space . the only glueballs which can not decay in this manner are the lightest states in the @xmath130 and @xmath115-odd sectors , the @xmath131 and @xmath132 . without explicit @xmath130 or @xmath115 violation , no dimension 6 operator permits the @xmath131 and @xmath132 states to decay into standard model fermions . in the case of bifundamental mediators , loops of the heavy particles
also induce dimension 8 effective interactions coupling the hidden gluons to standard model gauge bosons , [ eq : d8 ] _ eff = c_1 t rf_i^2 t rg^2 + c_2 f_1 t r g^3 , considered in detail in @xcite in the context of hidden valley signals at the lhc . here
@xmath133 @xmath134 is the field strength tensor for standard model ( dark sector ) gauge fields .
the field strength tensors are contracted according to different irreducible representations of the lorentz group which for the sake of brevity are not shown here .
these dimension 8 operators allow all the hidden sector glueballs to decay , either directly to standard model gauge boson pairs ( the @xmath111 , @xmath135 , @xmath131 , @xmath136 ) or radiatively to another glueball via the emission of a photon ( all others ) . for a glueball which can decay either through radiative photon emission or through an off - shell higgs ,
the situation is rather involved .
the branching ratios of these two modes depend critically on the mass scale @xmath137 of the mediators and the nonperturbative transition matrix elements .
the ratio of the partial widths can be written parametrically as ( ) ^4 ( ) ^2 _ d ^2 where @xmath138 and @xmath139 is the fine structure constant .
the dimensionless phase - space function @xmath140 is given by i(a , b)= _ 4b^2^(1-a)^2 dx x ( g(a^2,x ; 1))^l+1/2 ( g(b^2,b^2;x))^3/2 .
where @xmath141 and @xmath79 is the relative orbital angular momentum of @xmath142 and @xmath143 .
given the masses in fig .
[ fig : glue ] , @xmath144 . for @xmath145 , which is a conservative choice
if the mass splittings between the mediator fields are not especially large , the larger suppression of dimension eight operators tends to suppress the radiative decays in favor of the three - body decays .
however , for small enough @xmath137 , the radiative decays can become competitive .
indeed , a simple estimate suggests that for @xmath146 , the dimension 6 operator dominates , while for mediators within the reach of the lhc radiative and three - body rates can be comparable . if the off - shell higgs needs to decay to muon pairs , the small muon yukawa coupling will additionally suppress the three - body rates , making the radiative decays dominate .
to summarize , glueball decays contribute @xmath147 quarks or @xmath148 leptons to the visible signals of dark matter decay , as well as possible photons ( both singly and in pairs ) if the mediators are charged under @xmath149 .
( decays into @xmath5 and @xmath96 bosons are kinematically impossible . )
colored mediators would also allow the glueballs to decay into standard model gluon pairs , but in order to minimize the antiprotons coming from glueball decays we have taken the mediators to be color singlets .
moreover , as most of the mass splittings between glueballs are smaller than @xmath34 , if the @xmath150 is near the threshold to decay to @xmath151 pairs , then kinematics may prevent @xmath151 pairs from being produced in three body decays of heavier glueballs .
a typical final visible state would then be of the form @xmath152 @xmath153 and so on .
a similar story holds for @xmath154 decaying to tau pairs : higher - mass glueballs could yield pion and muon pairs in addition to the tau pairs coming from the decay of the terminal @xmath154 .
glueballs which decay mainly into muon or pion pairs might be obtained with additional model building in the higgs sector . while it is possible to work out glueball branching fractions for a given mediation model with the help of lattice data for @xmath33 , it is considerably more difficult to arrive at the full spectrum of visible standard model particles produced in a single glueballino decay .
the injection spectrum @xmath155 of visible standard model particles depends on the number and spectra of glueballs initially produced in the glueballino decay . in order to make crisp predictions about indirect detection signals coming from dark matter decay , we therefore need to understand the process of fragmentation in pure yang - mills theory .
unfortunately , this is a situation where neither phenomenological examples from qcd nor data from the lattice can be of help . in the absence of light fundamentals ,
the color tube connecting hard partons can not break .
glueball formation in a pure yang - mills theory occurs heuristically through the crossing of the tube onto itself , a process qualitatively different from fragmentation in qcd . one simple approach to estimating the relative abundances of different glueball species is a thermal model .
in such a model , the ratio of the multiplicity of glueballs is given by = ( 2j+1)()^3/2 ^-(m_j - m_0)/t where @xmath156 @xmath157 is the mass ( multiplicity ) of a glueball with spin @xmath158 .
the effective temperature @xmath159 can be taken as the center of mass energy of the colored system immediately following the decay of the heavy adjoint fermion , @xmath160 . with @xmath161 ,
the thermal model predicts glueball democracy in the final state : for @xmath162 and @xmath163 , the heaviest of the stable glueballs , the @xmath164 , is produced two - thirds as frequently as the @xmath111 .
unfortunately , the applicability of the thermal model to pure yang - mills theory is unclear .
in particular , excitations of the flux tube ( such as can easily be produced from gluon showering in the final state ) might enhance the production of the higher - spin states above the thermal estimate .
moreover , the thermal model is not well - suited to address a key question for indirect detection : what is the average _ total _ number of glueballs per decay ?
to arrive at an injection spectrum @xmath155 which can be compared to indirect detection data , we need to specify not only a particular model for glueball decay but also a model for fragmentation in the hidden sector .
while it appears possible to build interesting models which have the potential to avoid overproduction of antiprotons , it is clear that not only is there significant freedom in the detailed choice of model , but also that a reasonable model of pure - glue fragmentation is a necessary ingredient in quantifying the predictions of such models for indirect detection experiments . adding light ( @xmath165 ) fundamental flavors to the hidden sector destabilizes the glueballs to hidden mesons , without destabilizing the adjoint fermion @xmath11 .
as @xmath11 is fermionic , it must decay to an odd number of hidden quarks and anti - quarks .
but such a final state carries non - zero hidden - sector baryon number .
the decay of @xmath11 to light flavor therefore violates baryon number , and is therefore forbidden in the absence of hidden sector baryon number violation .
the meson masses depend on the input light quark masses as well as on the strong coupling scale , and can be tuned independently of the other parameters of the theory .
avoiding overclosure requires that the hidden mesons have sub - ev masses , giving rise to a large but technically natural hierarchy , @xmath166 .
however , the existence of such extremely light mesons leads to concerns that inelastic glueballino scattering @xmath167 may exceed observational bounds on dissipative dark matter interactions . adding such extremely light mesons will also dramatically increase the late - time glueballino annihilation cross - section , likely to a problematic extent .
more attractive model building possibilities are realized by keeping @xmath168 and allowing the hidden sector mesons to decay to the standard model .
this can be realized in multiple ways .
one option is to introduce a @xmath169 coupling to both hidden flavor and visible standard model fermions @xcite .
this allows at least one hidden sector meson to decay to a pair of standard model fermions , with a preference for heavier fermions due to helicity suppression .
the main challenge is then obtaining a short enough lifetime for the shortest - lived hidden sector meson while keeping @xmath170 at the gev scale . for models of the form considered in @xcite , lifetimes of less than a second
can be achieved for @xmath171 , with the mediator mass in the range of a few to tens of tev .
decay to muon pairs can only be realized for a light mediator and @xmath172 close to the @xmath148 threshold .
models with light flavor have one especially nice feature : they allow for more concrete predictions for the visible signals from dark matter decay , as with the addition of light flavor it is possible to adapt models of fragmentation in qcd .
this then allows definite predictions for the spectra and multiplicities of the hidden sector hadrons produced in dark matter decay , which in turn allows detailed comparison to astrophysical data .
while this is appealing , it comes at a cost : the existence of light degrees of freedom in the hidden sector exacerbates the uncertainties in evaluating the annihilation cross - section for glueballinos , and hence in the acceptable mass range for the dark matter .
nonetheless , there are interesting avenues for model building here .
we turn to discussing the constraints on our neutrino portal scenario from various indirect detection experiments . the glueballino decaying via the operator of eq .
( [ e.npc ] ) produces dominantly neutrinos , but also higgs particles , @xmath5 and @xmath96 bosons , and electrons and positrons .
these fluxes of standard model particles produced directly from the neutrino portal operator are relatively model - independent : the relative normalization of the two- and three - body decay modes depends only on the kinematics of the decay .
bounds on these decay modes therefore constrain the general scenario where structure in the dark sector enables dark matter decaying through the neutrino portal to yield a primary flux of monochromatic neutrinos .
we compute the fluxes of neutrinos , antiprotons , positrons , and gamma rays observed at the earth for standard model particles produced in the decay of the heavy dark matter particle , and determine the allowed parameter space of our model . throughout this section
we assume the dark matter distribution in our galaxy is described by the einasto profile [ e.ep ] _
dm(r ) = _ 0 where @xmath58 is the distance from the center of our galaxy , and we use the values @xmath173 and @xmath174 kpc @xcite .
the normalization constant is @xmath175 which gives the local dark matter energy density @xmath176 . here
@xmath177 kpc is the distance of the solar system from the galactic center . because for decaying dark matter the signal depends on @xmath178 ( rather than @xmath179 as for annihilation ) there is limited sensitivity to the choice of the profile .
for example , another standard profile , the nfw profile _
dm(r ) = _ 0 ( ) ^(1 + ( ) ^ ) ^(-)/ with parameter choices @xmath180 , @xmath181 , and @xmath182 @xcite , increases the neutrino flux above that predicted using the einasto profile by about 10 percent .
dark matter decaying in our galaxy produces mostly neutrinos and antineutrinos with energy equal to @xmath183 .
this leads to a flux of cosmic neutrinos concentrated in the direction toward the galactic center .
the interaction cross - sections of dark matter with nuclei in our model are negligible , leading to correspondingly negligible dark matter capture rates in the earth and the sun
. therefore only bounds on neutrino flux from the galactic center are constraining .
to date , the best bounds on such a flux are those from the super - kamiokande collaboration @xcite .
the flux of neutrinos coming from dark matter decay is given by ( m _ ) = ds(s ) , where @xmath184 is the spectrum of neutrinos generated in the dark matter decay , and the integral of the dark matter energy density is taken along a line of sight .
it is conventional to define the dimensionless quantity = , normalizing the line of sight density integral to our local position and density .
the flux in a cone of half - angle @xmath185 around the galactic center is then [ e.gcf ] ( m _ ; ) = r__j ( ) , with j()= ^1 _
d_0 ^ r_0 dx ( ) . notice that eq .
( [ e.gcf ] ) is the total neutrino flux , summed over all three flavors , as oscillation over galactic scales will equally populate all three flavors .
moreover , there is an equal in magnitude anti - neutrino flux .
having fixed the halo profile in eq .
( [ e.ep ] ) , @xmath186 is a function of the half - cone angle which can be computed numerically .
neutrinos produced in a decay travel to the earth without experiencing further interactions .
the normalization of the neutrino flux at the earth depends on the dark matter lifetime and mass via the ratio @xmath187 , and this is the quantity that will be constrained by super - kamiokande .
the published super - kamiokande bounds do not apply directly directly to the neutrino flux but rather to the upward - going muon flux .
these muons are produced by charged current interactions of the cosmic neutrinos in the rock below the detector . in order to translate the neutrino and antineutrino fluxes into a muon flux
we need to compute the conversion functions @xmath188 and @xmath189 , [ e.cfdef ] d _ ^- d e _
= de _ , = de_. which encapsulate the probability that a neutrino will interact in rock to produce a muon of energy @xmath190 which reaches the detector .
the conversion functions depend on the the muon range in rock @xmath191 , the proton and neutron number densities in rock @xmath192 , @xmath193 , and the neutrino inelastic scattering cross section on nucleons @xmath194 : [ e.dupa1 ] d c_^- d e _
= r_(e _ ) , = r_(e _ ) . to compute the neutrino - nucleon scattering cross sections
, we use the tree - level partonic cross - section integrated over the cteq5 parton distribution functions .
we perform our computations using the fiducial material `` standard rock '' , for which the nucleon densities are @xmath195 . the last piece we need to evaluate eq .
( [ e.dupa1 ] ) is the muon range @xmath191 , the distance a muon with the initial energy @xmath190 can travel before its energy is radiated away so as to fall below the super - kamiokande threshold @xmath196 gev .
the muon range can be modeled analytically using the `` continuous slowing down approximation '' , with parameters found in @xcite .
we approximate the muon range @xmath191 as [ eq : range ] r_(e _ ) 10 ^ 5 ( ) , here @xmath197 is the critical energy , where ionization and radiation loss are equal ; we take @xmath197 to be @xmath198 in standard rock @xcite ( we assume that cross sections in standard rock are comparable to cross sections in the rock below super - kamiokande ) .
the continuous slowing down approximation becomes poor at high energies , where stochastic radiative processes dominate the energy loss .
more accurate monte carlo modeling of muon propagation at high energies reduces the flux of upward - going neutrino - induced muons at neutrino detectors for a given incident neutrino flux relative to the flux obtained using the continuous slowing down approximation @xcite . comparing the effective muon range given by to the effective muon range computed in @xcite for muons with energy @xmath199
, we find our muon ranges are overestimated by a factor of less than order unity over most of the energy range .
the bounds we obtain from super - kamiokande s reported upward through - going muon counts are therefore conservative . finally , to obtain the total muon flux in the super - kamiokande detector we have to integrate over all muon energies between @xmath200 and @xmath183 . at the end of the day the muon flux in super - kamiokande is given by [ e.mfsk ] _
^ ( ) = 1 3 _
_ 4 j_ls ( ) _ e_0^m_/2 de _ the factor @xmath201 arises because @xmath202 of the produced neutrinos oscillate into electron and tau neutrinos to which super - kamiokande is far less sensitive . as a function of @xmath23 allowed by the super - kamiokande bounds.,scaledwidth=50.0% ]
the muon flux induced by neutrinos from the galactic center predicted by our model is dominated by the monoenergetic neutrinos from the dominant two - body decay . at higher dark matter masses , @xmath203 ,
the additional neutrinos from the sub - dominant three - body modes begin to make a noticeable contribution to the total muon flux .
we computed the total muon flux predicted from our model and confronted it with the super - kamiokande bounds from @xcite .
the most stringent bounds are obtained for cones of half - angle @xmath204 degrees . in fig .
[ f.superkgamma ] the resulting bound on @xmath187 is plotted as a function of @xmath23 .
the dark matter mass is varied in the range @xmath205 tev , as motivated by the estimate of thermal abundance in the previous section . for @xmath206
tev the bound is @xmath207 , corresponding to the lifetime @xmath208 seconds . for @xmath209
tev the bound is @xmath210 , corresponding to the lifetime @xmath211 seconds .
the sub - dominant three - body decays lead to additional visible signals , in particular , to antiprotons , hard leptons ( which we take here as @xmath212 ) , and photons . in the following subsections we investigate the predicted antiproton , positron and gamma ray fluxes arising from the three - body decay modes to determine whether these subleading decays impose more stringent bounds on the lifetime of our dark matter particle than the two - body decays .
an irreducible source of antiprotons in our model are the decays of the @xmath5 , @xmath96 , and higgs bosons which are produced by the three - body decays of our dark matter particle .
on average , these channels lead to one energetic antiproton per decay .
the branching ratio for three - body decays varies from 4 percent of the total width for @xmath213 tev to as much as 55 percent for @xmath209 tev .
thus we expect that the antiproton bounds on the dark matter lifetime can become quite stringent in some regions of the parameter space .
the production rate of antiprotons per unit energy due to the three - body decays is given by [ eq : qdef ] q_|p(e , r ) = 2 _ 3b m _
_ dm(r ) ( dn_z |p d e + dn_h |p d e + 2 dn_w |p d e ) .
we determined the antiproton spectra @xmath214 by first generating @xmath5 , @xmath96 , and higgs decays at rest using pythia 6.420 , and then boosting the momenta of antiprotons to the galactic rest frame .
the energy distribution of the parent gauge and higgs bosons is given by the differential decay width , which is the same for all three particles in the limit @xmath215 : = e ( m_-2 e ) ( m_-e ) .
throughout , the higgs is assumed to have standard model properties and the mass equal 115 gev . in order to translate the production rate into the flux of antiprotons at the earth
it is necessary to model the propagation of antiprotons in our galaxy @xcite . the antiproton number density per unit energy @xmath216 satisfies a diffusion equation [ e.appe ] k(e)^2 n_|p(e , r ) - _ z ( v_c sgn(z ) n_|p(e , r ) ) - 2 h ( z ) _
ann(e ) n_|p(e , r ) + q_|p(e , r ) = 0 in which the production rate appears as the source term .
energy loss for protons is negligible and the corresponding term in the diffusion equation is dropped .
the diffusion coefficient @xmath217 is parameterized as @xmath218 .
the parameters @xmath219 and @xmath220 are not known from first principles ; instead , they are fitted , along with the magnitude of the convective wind @xmath221 and the size of the diffusion zone , so as to correctly describe the observed flux of cosmic rays .
the third term describes the disappearance of antiprotons due to annihilations with protons in the galactic disk .
the half - height of the galactic disk is set by @xmath222 , and the annihilation width is given by the number densities of hydrogen and helium multiplied by the proton - antiproton annihilation cross - section , @xmath223 .
expressions for @xmath224 and the proton - antiproton annihilation cross - section @xmath225 can be found in @xcite .
the diffusion equation eq .
( [ e.appe ] ) is solved subject to the boundary condition that the antiproton density @xmath216 vanish at the surface of the diffusion region , taken to be a cylinder of radius @xmath226 and half - height @xmath227 . in the following we focus on the med propagation model @xcite ,
in which case the diffusion equation parameters take the following values : k(e ) = 3.5 10 ^ -10 ^2/ ; = 0.7 ; v_c = 1.2 10 ^ 6 / ; l_max = 4 .
once we know @xmath216 , the flux of antiprotons observed at the earth is given by @xmath228 ; in the interesting energy range the antiprotons are highly relativistic and @xmath229
. the propagation equation can be solved semi - analytically @xcite . using that solution , the antiproton flux in our model
is given by = 1 2 _ 3b m _
( dn_z |p d e + dn_h |p d e + 2 dn_w |p d e ) _ dm(r _ ) r(e ) .
[ e.pbarf ] the propagation effects are encoded in the function @xmath230 which can be computed numerically @xcite . , 2 , and 5 tev .
in each case the glueballino decay rate is set to the maximal value allowed by the super - kamiokande bounds.,title="fig:",scaledwidth=30.0% ] , 2 , and 5 tev . in each case
the glueballino decay rate is set to the maximal value allowed by the super - kamiokande bounds.,title="fig:",scaledwidth=30.0% ] , 2 , and 5 tev . in each case
the glueballino decay rate is set to the maximal value allowed by the super - kamiokande bounds.,title="fig:",scaledwidth=30.0% ] with this solution in hand we are ready to estimate the antiproton constraints on the lifetime of our dark matter particle .
we pick three representative values for the dark matter mass : @xmath231 tev . for the sake of comparison with the neutrino bounds , for each mass we choose the maximum @xmath187 allowed by super - kamiokande as plotted in fig .
[ f.superkgamma ] . given @xmath232 and @xmath23 ,
the partial width for three - body decays @xmath233 is completely determined by equation .
the mass fixes the spectra @xmath234 , while the partial width fixes the normalization of the antiproton flux in eq .
( [ e.pbarf ] ) .
our results are plotted in fig .
[ f.pbar3b_flux ] and compared to the antiproton flux measured by pamela .
it is clear that antiprotons , despite being produced by subdominant decay channels , can put important constraints on the dark matter lifetime . for @xmath209 tev ,
the predicted antiproton flux is 5 times larger than observed , which would imply the stringent bound @xmath235 corresponding to the lifetime @xmath236 sec . for @xmath213 tev ,
the predicted antiproton flux is only marginally larger than observed , and is slightly more constraining than the super - kamiokande bound .
one should however keep in mind that the propagation of antiprotons in our galaxy suffers from huge uncertainties , and the predicted antiproton flux changes dramatically with different assumptions about the propagation model .
for example , choosing the min parameters for the propagation model @xcite , k_0 = 5.1 10 ^ -11 ^2/ ; = 0.85 ; v_c = 1.35 10 ^ 6 / ; l_max = 1 , reduces the antiproton flux by a factor of five .
this would make the antiproton bounds comparable to the neutrino ones for @xmath209 tev , and irrelevant for @xmath213 tev .
( it is of course also possible to choose propagation parameters which magnify the antiproton flux . )
thus , the antiproton bounds should be taken with a grain of salt . in the following , we continue using the more robust super - kamiokande constraints as the reference point for estimating the indirect signals of dark matter .
our decaying dark matter particle contributes also to the cosmic flux of positrons and photons .
hard positrons are produced together with @xmath5 bosons in three - body decays and , to a lesser extent , by subsequent decays of @xmath5 , @xmath96 , and higgs .
gamma ray photons are produced via decays of @xmath5 , @xmath96 , and higgs , from final state radiation off charged particles , and through inverse compton scattering of the hard positrons and electrons on starlight .
below we compute the fluxes of hard positrons and photons from the three - body decay modes and argue that the corresponding bounds are far weaker than those from neutrinos and antiprotons .
therefore , the pamela positron measurements and the gamma ray measurements from fermi and hess do not impose new constraints on the lifetime of our dark matter particle .
the positron flux at the earth due to the positrons from three - body decays is given by @xcite = _ 3b m _ 1 2 b(e ) _ dm(r _ ) _ e^m_/2 d e dn_e^+ d e i(e;e ) [ e.positronflux ] here , the energy loss coefficient @xmath237 gives the rate of energy loss due to scattering off of cosmic photons and magnetic fields , and is approximately given by @xmath238 where @xmath239s .
the function @xmath240 , sometimes called the halo function , results from semi - analytically solving the diffusion equation for positrons @xcite .
it is analogous to the function @xmath241 for antiprotons , and similarly depends on the choice of propagation model .
finally , @xmath242 is the positron spectrum from three - body decays , given by = e^ 2 ( m_- e ) . to obtain the positron fraction one has to divide the flux in eq .
( [ e.positronflux ] ) by the total flux of electrons and positrons , for which we take = 1/(^2 ) as measured by fermi @xcite . as an example , in fig .
[ f.eplus3b_flux ] we plot the contribution to the positron fraction from the three - body decays for @xmath243 tev , using the med propagation parameters .
we took the decay rate @xmath244 to be the maximum one allowed by the neutrino bounds from super - kamiokande .
recall that for these parameters antiprotons are overproduced by a factor of 23 . on the other hand ,
the positron fraction for the same propagation parameters is a factor of 24 below that measured by pamela .
we conclude that positrons do not impose any new constraints on the lifetime of our dark matter particle . at the same time
, we can not explain the anomalous positron fraction via the positrons from three - body decays ; another primary source of positrons is needed to explain the pamela positron measurement .
the new source could be of purely astrophysical origin @xcite , but it could be glueball ( or meson ) decays in our model .
tev . left : the positron fraction ( solid red ) compared to the pamela measurement ( dashed ) .
right : the diffuse gamma ray flux @xmath245 , @xmath246 ( solid red ) compared to the fermi measurement ( dashed).,title="fig:",scaledwidth=30.0% ] tev .
left : the positron fraction ( solid red ) compared to the pamela measurement ( dashed ) .
right : the diffuse gamma ray flux @xmath245 , @xmath246 ( solid red ) compared to the fermi measurement ( dashed).,title="fig:",scaledwidth=35.0% ] similarly , the contribution to the gamma ray spectrum from the three - body decays is well below the current observational bounds . as an example
, we calculate the contribution to the diffuse gamma ray spectrum at intermediate galactic latitudes .
recently , fermi published the gamma ray spectrum between 0.1 and 10 gev from between 10 and 20 degrees galactic latitude @xcite .
the contribution of three - body decays to that spectrum is given by = 2 _ 3b m _ d n _ d e i_d(10^,20^ ) i_d(_1,_2 ) = 1 4 _ 0 ^ 2 d _ 0^ dl _ _
dm(r ) where @xmath247 .
using pythia , we determined the photon spectrum @xmath248 from @xmath5 , @xmath96 , and higgs decays , where the parent bosons originate from the three - body decays of dark matter .
the results for @xmath249 tev are plotted in fig .
[ f.eplus3b_flux ] .
it is clear that the contribution of the three - body decays is more than an order of magnitude smaller than the flux measured by fermi .
similarly , the contribution to the gamma ray flux from the galactic center is far below the flux measured by fermi and hess @xcite .
thus , the strongest constraints on the dark matter lifetime in our model are the galactic center neutrino measurement by super - kamiokande and the cosmic antiproton flux measurement by pamela .
our model predicts a monochromatic flux of high - energy ( @xmath250 ) neutrinos and antineutrinos from the galactic center , [ e.mnf ] d _ d e ( ) = d _ | d e ( ) = _ 2b m _
( e - m_/2 ) r _ _ 4 j_ls ( ) , where the line - of - sight integral over the dark matter density profile @xmath251 is given in eq .
( [ e.gcf ] ) .
the normalization of the flux is set by the dark matter lifetime , @xmath232 .
the flux is equally populated by all three species of neutrinos ; the flux of muon neutrinos to which experiments are most sensitive is one third of eq .
( [ e.mnf ] ) .
the super - kamiokande results place constraints on the dark matter lifetime @xmath187 , as shown in figure fig .
[ f.superkgamma ] .
the dark matter lifetime is also constrained by the antiprotons produced in subdominant three - body decay modes .
while the limits on cosmic antiprotons can be comparable to or more stringent than the bounds from super - kamiokande , depending on the choice of model for charged particle propagation in the galaxy , to discuss signals at neutrino telescopes we will continue to use the limits on the dark matter lifetime set by super - kamiokande . in this section
we discuss the sensitivity of the antares and icecube neutrino telescopes to the primary neutrino signal .
both experiments have the potential to discover this signal .
antares is a high - energy neutrino telescope located in the mediterranean sea .
the flux of muon neutrinos from the galactic center produces upward - going muons via charged current interactions in the rock and water below the detector , which antares observes via the cherenkov light produced in sea water by the muons .
the effective detector area @xmath252 for muon neutrinos can be found e.g. in @xcite and at 1 tev is given approximately by @xmath253 .
the energy resolution near 1 tev is expected to be a factor of 23 @xcite , while the angular resolution is better than a degree and limited mostly by the kinematics of the charged current interactions @xcite .
the number of signal counts per second from the cone of half - angle @xmath185 around the galactic center is given by = 2 3 a_eff(m_/2 ) _
_ 4 j_ls ( ) , and the neutrino energy is @xmath254 .
the irreducible background for the dark matter decay signal comes from atmospheric muon neutrinos .
the combined flux of atmospheric muon neutrinos and anti - neutrinos at @xmath255 is , averaged over angle @xcite , .|_1 = 2.710 ^-11 ( ^ 2 ) ^-1 .
the atmospheric neutrino spectrum falls as @xmath256 in the energy range of interest , @xmath257 . in the energy range @xmath258
where we expect hard primary neutrinos from decaying dark matter , neutrino telescopes have poor energy resolution . to differentiate signal from background
we therefore rely on first , absolute rate , and second , angular variation in the signal .
the neutrinos coming from dark matter decay are concentrated at the galactic center , while the spatial variation of the atmospheric neutrino fluxes is well - known and is at most a factor of @xmath259 between zenith and horizon @xcite .
the number of counts per second from atmospheric neutrinos in a given energy bin @xmath260 is given by = _ e_min^e_max de
a_eff(e ) d_atm d e d we approximate the energy resolution of antares by choosing an energy bin @xmath261 , corresponding to a factor of three in energy resolution centered on the energy of the primary neutrino signal . in fig .
[ f.antares ] we plot the number of signal and background counts per year in this bin for cones around the galactic center as a function of the cone half - angle .
we show plots for @xmath262 and @xmath263 tev . to generate these plots , we assume that the detector effective area @xmath264 already incorporates factors accounting for detection efficiency and the different cross sections of neutrinos and anti - neutrinos .
we use the atmospheric @xmath265 + @xmath266 flux @xmath267 computed in @xcite and averaged over the zenith angle . for the signal
, we show expected counts for the maximal decay rate allowed by the super - kamiokande bounds . for that signal rate
, the five - degree cone corresponds to on the order of 5 signal counts and a signal - to - background ratio in the range of 13 ( for 15 tev dark matter ) .
thus , in the best case scenario , a clear signal can be detected already in one year of antares data .
if no signal is detected , antares can significantly improve the bounds on the dark matter lifetime in our scenario .
the theoretical uncertainty in the atmospheric neutrino flux is of order 25 percent at 1 tev @xcite . from
that we estimate that one year of antares data can improve the bounds on @xmath232 by a factor of 4 to 8 , depending on the dark matter mass . around the galactic center , for @xmath2681 , 2 , and 5 tev .
the normalization of the flux corresponds to the maximal flux allowed by super - kamiokande for a given dark matter mass .
, title="fig:",scaledwidth=30.0% ] around the galactic center , for @xmath2681 , 2 , and 5 tev .
the normalization of the flux corresponds to the maximal flux allowed by super - kamiokande for a given dark matter mass .
, title="fig:",scaledwidth=30.0% ] around the galactic center , for @xmath2681 , 2 , and 5 tev .
the normalization of the flux corresponds to the maximal flux allowed by super - kamiokande for a given dark matter mass .
, title="fig:",scaledwidth=30.0% ] while icecube , located at the south pole , has better exposure to the northern hemisphere , the extension deepcore , planned for 2010 , will allow icecube to observe the southern hemisphere sky @xcite .
icecube s new sensitivity to the galactic center opens many exciting possibilities for dark matter studies @xcite .
deepcore adds additional instrumentation in the clearest and best - shielded portion of the icecube detector , and enables study of downward - going neutrinos by using the remainder of the icecube detector to veto cosmic muons .
cosmic downward - going neutrino signals then consist of neutrino - initiated events with initial interaction vertices located within the fiducial volume of deepcore .
muon neutrinos are primarily detected in deepcore as they are in super - kamiokande and antares , as a track from the muon produced in a charged - current interaction of the neutrino with matter .
in addition to these _ track - like events _ , the neutral - current interactions of neutrinos with all flavor , as well as the charged - current interactions of electron and tau neutrinos , deposit energy in a localized region and are reconstructed as _ cascade - like _ events .
cosmic neutrinos , and in particular neutrinos from dark matter decay , populate all three flavors equally . usually at neutrino telescopes , the effective volume probed by muon neutrinos is orders of magnitude larger than the effective volume seen by electron and tau neutrinos , due to the large muon penetration depth .
however , when using deepcore to observe downward - going neutrinos , the volume available for muon neutrino events is equal to the volume available for electron and tau neutrino events , putting observable signal events from all three neutrino flavors on an equal footing .
while angular resolution for cascade events is less than angular resolution for tracks @xcite , the high rate of cascade events predicted by our model suggests another potential handle for discovery .
the fiducial volume of deepcore is estimated to be around 20 megatons @xcite , that is @xmath269 .
detection efficiencies for neutrino - initiated events are estimated to be @xmath270@xmath62 @xcite .
while the angular resolution that can be achieved for deepcore - initiating events is still to be determined , a reasonable goal is 10 degrees for tracks , possibly improving to 5 degrees , and 40 degrees for cascades @xcite .
we will estimate the energy resolution to be @xmath271 in @xmath272 for tracks and @xmath273 in @xmath272 for cascades @xcite .
the number of neutrino - initiated muons in deepcore from a given incident neutrino flux is = de_de _ _ f v_f _
n n_n ( + ) .
the sum runs over nucleons , @xmath274 .
we take ice to have a density @xmath275 , corresponding to the nucleon number densities @xmath276 , @xmath277 .
we include a factor @xmath278 to account for the detection efficiency .
the integral over the neutrino energy is trivial for the signal ; for the background we set the lower integration limits at @xmath279 for tracks and cascades respectively . in fig .
[ f.icecube ] we plot the expected number of deepcore - initiating track - like events originating within a cone of given half - angle around the galactic center . for the maximal signal flux allowed by the super - kamiokande constraints ,
one expects on the order of 25 ( 5 ) counts per year inside a cone of half - angle @xmath280 for a dark matter mass of 1 ( 5 ) tev .
the dominant background from muon atmospheric neutrinos , also plotted in fig .
[ f.icecube ] , gives roughly the same number of track - like events as the signal .
however , it may be possible to reduce as much as 99 percent of the atmospheric neutrino background by vetoing on the associated collinear muon @xcite .
if that proves to be the case , the background could be almost completely suppressed and the discovery prospects for the signal are greatly enhanced .
the sensitivity of icecube would only be limited by statistics , and could reach a lifetime of @xmath4 seconds after several years of operation , two orders of magnitude better than the current bounds . around the galactic center , for @xmath268 1 , 2 , and 5 tev .
the normalization of the flux corresponds to the maximal flux allowed by super - kamiokande for a given dark matter mass .
, title="fig:",scaledwidth=30.0% ] around the galactic center , for @xmath268 1 , 2 , and 5 tev .
the normalization of the flux corresponds to the maximal flux allowed by super - kamiokande for a given dark matter mass . ,
title="fig:",scaledwidth=30.0% ] around the galactic center , for @xmath268 1 , 2 , and 5 tev .
the normalization of the flux corresponds to the maximal flux allowed by super - kamiokande for a given dark matter mass .
, title="fig:",scaledwidth=30.0% ] around the galactic center , for @xmath2681 , 2 , and 5 tev .
the normalization of the flux corresponds to the maximal flux allowed by super - kamiokande for a given dark matter mass .
, title="fig:",scaledwidth=30.0% ] around the galactic center , for @xmath2681 , 2 , and 5 tev .
the normalization of the flux corresponds to the maximal flux allowed by super - kamiokande for a given dark matter mass .
, title="fig:",scaledwidth=30.0% ] around the galactic center , for @xmath2681 , 2 , and 5 tev .
the normalization of the flux corresponds to the maximal flux allowed by super - kamiokande for a given dark matter mass .
, title="fig:",scaledwidth=30.0% ] we also compute the number of cascade - like events initiated by the signal and compare to events from the atmospheric neutrino background as the atmospheric electron neutrino flux is approximately 5 percent of the atmospheric muon neutrino flux in the energy range of interest , the background is dominated by the neutral - current interactions of atmospheric muon neutrinos .
charged - current interactions of atmospheric electron neutrinos contribute approximately an eighth of the total background event rate . for cascade events ,
rate alone can serve to distinguish signal from background , as the signal to background ratio can be much larger than one for decay rates near the super - kamiokande bounds . in fig .
[ f.cascades ] we plot signal and background rates for cascade events for @xmath268 1 , 2 , and 5 tev , for the maximal signal flux allowed by the super - kamiokande bounds .
again , it may be possible to nearly eliminate the background due to atmospheric muon neutrinos , which would reduce the background by nearly 90 percent . thus cascade - like events , even more than track - like events , offer powerful handles on neutrino signals of this nature
. more importantly , observing an excess of both tracks and cascades would be an important confirmation of the galactic origin of the neutrino flux .
we presented a scenario where decay of hidden sector dark matter can give a large flux of neutrinos from the galactic center , dominating over signals in more standard detection channels .
fluxes of the more readily observed @xmath212 , antiprotons , and photons are subleading .
observational bounds on the flux of cosmic antiprotons are stringent , and despite the relatively small production rate of antiprotons in this class of model , these observations can become constraining if the lifetime of the dark matter particle is of order @xmath2 seconds
. however , the large astrophysical uncertainties in the galactic propagation model for antiprotons make it difficult to draw any robust conclusions from the antiproton spectrum .
moreover , such a dark matter particle naturally interacts very weakly with nucleons , giving null predictions for direct detection experiments . in these circumstances ,
neutrino telescopes become the most sensitive probe of dark matter . in the example model we present here ,
the dark matter candidate is a fermion belonging to a hidden sector which is neutral under all standard model symmetries .
such a particle naturally couples to the standard model via the neutrino portal , which is the lowest - dimensional fermionic gauge singlet operator .
the coupling through the neutrino portal enables dark matter to decay to the standard model .
the existence of lighter states in the dark sector allows the decay channel with a single standard model neutrino to dominate the final state .
this decay mode results in a monoenergetic flux of cosmic neutrinos that can be probed by neutrino telescopes .
while we have studied a specific realization of that scenario , where the hidden sector consists of a confining @xmath0 gauge group together with a heavy fermion in the adjoint representation , obviously similar phenomenology can be obtained in a wider class of models .
currently , the coupling via the neutrino portal is most robustly constrained by super - kamiokande , which restricts the dark matter lifetime to be larger than @xmath2 seconds .
the pamela measurement of the antiproton flux yields roughly comparable constraints due to the antiprotons arising from the subleading dark matter decay channels containing gauge and higgs bosons .
these bounds leave a lot of room for discovery at the current generation of neutrino telescopes .
the good angular resolution of both antares and icecube / deepcore ensure that the number of signal counts from the direction of the galactic center may be larger than or comparable to the expected background from atmospheric neutrinos . in the case of antares ,
the sensitivity to the dark matter decay rate exceeds that of super - kamiokande by a factor of several . furthermore , icecube / deepcore has the additional prospect of rejecting the vast majority of atmospheric neutrino events by observing the associated muon ( which is impossible at antares ) .
this implies increased sensitivity by another order of magnitude with respect to antares .
our analysis of the sensitivity of antares and icecube / deepcore should be considered only as suggestive of the experiments ultimate capabilities ; the operational parameters of these experiments are subject to ongoing research by the collaborations .
it is conceivable that designing search algorithms targeted for a monochromatic neutrino signal can further increase the sensitivity to our scenario .
we are grateful to jim braun and tyce deyoung for providing us with many details concerning the icecube experiment .
we would like to thank peter graham , dan hooper , patrick meade , michele papucci , josh ruderman , matt strassler , scott thomas , and tomer volansky for helpful conversations .
this work was supported in part by doe grant de - fg02 - 96er40959 .
99 o. adriani _ et al . _
[ pamela collaboration ] , `` an anomalous positron abundance in cosmic rays with energies 1.5.100 gev , '' nature * 458 * , 607 ( 2009 ) [ arxiv:0810.4995 [ astro - ph ] ] .
j. chang _ et al .
_ , `` an excess of cosmic ray electrons at energies of 300.800 gev , '' nature * 456 * , 362 ( 2008 ) .
a. a. abdo _ et al .
_ [ the fermi lat collaboration ] , `` measurement of the cosmic ray e+ plus e- spectrum from 20 gev to 1 tev with the fermi large area telescope , '' phys
. rev . lett . * 102 * , 181101 ( 2009 ) [ arxiv:0905.0025 [ astro-ph.he ] ] .
p. meade , m. papucci and t. volansky , `` dark matter sees the light , '' arxiv:0901.2925 [ hep - ph ] .
d. spolyar , m. buckley , k. freese , d. hooper and h. murayama , `` high energy neutrinos as a test of leptophilic dark matter , '' arxiv:0905.4764 [ astro-ph.co ] .
m. r. buckley , k. freese , d. hooper , d. spolyar and h. murayama , `` high - energy neutrino signatures of dark matter decaying into leptons , '' arxiv:0907.2385 [ astro-ph.he ] .
j. t. ruderman and t. volansky , `` searching for smoking gun signatures of decaying dark matter , '' arxiv:0907.4373 [ hep - ph ] .
d. e. kaplan , m. a. luty and k. m. zurek , `` asymmetric dark matter , '' phys . rev .
d * 79 * , 115016 ( 2009 ) [ arxiv:0901.4117 [ hep - ph ] ] .
m. j. strassler and k. m. zurek , `` echoes of a hidden valley at hadron colliders , '' phys .
b * 651 * , 374 ( 2007 ) [ arxiv : hep - ph/0604261 ] .
c. j. morningstar and m. j. peardon , `` the glueball spectrum from an anisotropic lattice study , '' phys .
d * 60 * , 034509 ( 1999 ) [ arxiv : hep - lat/9901004 ] .
m. gockeler , r. horsley , a. c. irving , d. pleiter , p. e. l. rakow , g. schierholz and h. stuben , `` a determination of the lambda parameter from full lattice qcd , '' phys .
d * 73 * , 014513 ( 2006 ) [ arxiv : hep - ph/0502212 ] .
h. baer , k. m. cheung and j. f. gunion , `` a heavy gluino as the lightest supersymmetric particle , '' phys .
d * 59 * , 075002 ( 1999 ) [ arxiv : hep - ph/9806361 ] .
a. arvanitaki , c. davis , p. w. graham , a. pierce and j. g. wacker , `` limits on split supersymmetry from gluino cosmology , '' phys .
d * 72 * , 075011 ( 2005 ) [ arxiv : hep - ph/0504210 ]
. h. k. dreiner , h. e. haber and s. p. martin , `` two - component spinor techniques and feynman rules for quantum field theory and supersymmetry , '' arxiv:0812.1594 [ hep - ph ] .
e. komatsu _ et al .
_ [ wmap collaboration ] , `` five - year wilkinson microwave anisotropy probe ( wmap ) observations : cosmological interpretation , '' astrophys .
j. suppl .
* 180 * , 330 ( 2009 ) [ arxiv:0803.0547 [ astro - ph ] ] .
j. hisano , s. matsumoto and m. m. nojiri , `` explosive dark matter annihilation , '' phys .
lett . * 92 * , 031303 ( 2004 ) [ arxiv : hep - ph/0307216 ] .
j. hisano , s. matsumoto , m. nagai , o. saito and m. senami , `` non - perturbative effect on thermal relic abundance of dark matter , '' phys .
b * 646 * , 34 ( 2007 ) [ arxiv : hep - ph/0610249 ] .
a. strumia , `` sommerfeld corrections to type - ii and iii leptogenesis , '' nucl . phys .
b * 809 * , 308 ( 2009 ) [ arxiv:0806.1630 [ hep - ph ] ] .
r. dave , d. n. spergel , p. j. steinhardt and b. d. wandelt , `` halo properties in cosmological simulations of self - interacting cold dark matter , '' astrophys .
j. * 547 * , 574 ( 2001 ) [ arxiv : astro - ph/0006218 ] .
o. y. gnedin and j. p. ostriker , `` limits on collisional dark matter from elliptical galaxies in clusters , '' astrophys . j. * 561 * , 61 ( 2001 ) [ arxiv : astro - ph/0010436 ] .
m. markevitch _ et al .
_ , `` a textbook example of a bow shock in the merging galaxy cluster 1e0657 - 56 , '' astrophys .
j. * 567 * , l27 ( 2002 ) [ arxiv : astro - ph/0110468 ] . s. w. randall , m. markevitch , d. clowe , a. h. gonzalez and m. bradac , `` constraints on the self - interaction cross - section of dark matter from numerical simulations of the merging galaxy cluster 1e 0657 - 5 , '' arxiv:0704.0261 [ astro - ph ] .
y. chen _ et al .
_ , `` glueball spectrum and matrix elements on anisotropic lattices , '' phys . rev .
d * 73 * , 014516 ( 2006 ) [ arxiv : hep - lat/0510074 ] .
j. e. juknevich , d. melnikov and m. j. strassler , `` a pure - glue hidden valley i. states and decays , '' arxiv:0903.0883 [ hep - ph ] . j. f. navarro _ et al .
_ , `` the inner structure of lambdacdm halos iii : universality and asymptotic slopes , '' mon . not .
349 * , 1039 ( 2004 ) [ arxiv : astro - ph/0311231 ] .
j. f. navarro , c. s. frenk and s. d. m. white , `` a universal density profile from hierarchical clustering , '' astrophys .
j. * 490 * , 493 ( 1997 ) [ arxiv : astro - ph/9611107 ] .
s. desai _ et al .
_ [ super - kamiokande collaboration ] , `` search for dark matter wimps using upward through - going muons in super - kamiokande , '' phys . rev .
d * 70 * , 083523 ( 2004 ) [ erratum - ibid .
d * 70 * , 109901 ( 2004 ) ] [ arxiv : hep - ex/0404025 ] .
s. desai _ et al .
_ [ super - kamiokande collaboration ] , `` study of tev neutrinos with upward showering muons in super - kamiokande , '' astropart .
* 29 * , 42 ( 2008 ) [ arxiv:0711.0053 [ hep - ex ] ] .
j. hisano , s. matsumoto , o. saito and m. senami , `` heavy wino - like neutralino dark matter annihilation into antiparticles , '' phys .
d * 73 * , 055004 ( 2006 ) [ arxiv : hep - ph/0511118 ] .
f. donato , n. fornengo , d. maurin and p. salati , `` antiprotons in cosmic rays from neutralino annihilation , '' phys .
d * 69 * , 063501 ( 2004 ) [ arxiv : astro - ph/0306207 ] .
a. barrau , g. boudoul , f. donato , d. maurin , p. salati and r. taillet , `` antiprotons from primordial black holes , '' astron .
astrophys . *
388 * , 676 ( 2002 ) [ arxiv : astro - ph/0112486 ] .
t. delahaye , r. lineros , f. donato , n. fornengo and p. salati , `` positrons from dark matter annihilation in the galactic halo : theoretical uncertainties , '' phys .
d * 77 * , 063527 ( 2008 ) [ arxiv:0712.2312 [ astro - ph ] ] .
d. grasso _ et al .
_ [ fermi - lat collaboration ] , `` on possible interpretations of the high energy electron - positron spectrum measured by the fermi large area telescope , '' arxiv:0905.0636 [ astro-ph.he ] . t. a. porter and f. t.
f. collaboration , `` fermi lat measurements of the diffuse gamma - ray emission at intermediate galactic latitudes , '' arxiv:0907.0294 [ astro-ph.he ] .
f. aharonian _ et al .
_ [ h.e.s.s .
collaboration ] , `` discovery of very - high - energy gamma - rays from the galactic centre ridge , '' nature * 439 * , 695 ( 2006 ) [ arxiv : astro - ph/0603021 ] .
f. aharonian _ et al .
_ [ h.e.s.s .
collaboration ] , `` hess observations of the galactic center region and their possible dark matter interpretation , '' phys .
lett . * 97 * , 221102 ( 2006 ) [ erratum - ibid .
* 97 * , 249901 ( 2006 ) ] [ arxiv : astro - ph/0610509 ] .
a. m. brown and f. t. a. collaboration , `` the antares neutrino telescope : status and first results , '' arxiv:0908.1035 [ astro-ph.he ] .
j. p. ernenwein , `` the antares neutrino telescope , '' _ to appear in the proceedings of 40th rencontres de moriond on electroweak interactions and unified theories , la thuile , aosta valley italy , 5 - 12 mar 2005 .
_ m. honda , t. kajita , k. kasahara , s. midorikawa and t. sanuki , `` calculation of atmospheric neutrino flux using the interaction model calibrated with atmospheric muon data , '' phys .
d * 75 * , 043006 ( 2007 ) [ arxiv : astro - ph/0611418 ] .
e. resconi and f. t. i. collaboration , `` status and prospects of the icecube neutrino telescope , '' nucl .
instrum .
meth . a * 602 * , 7 ( 2009 ) [ arxiv:0807.3891 [ astro - ph ] ] . | we consider a model of dark matter whose most prominent signature is a monochromatic flux of tev neutrinos from the galactic center . as an example of a general scenario
, we consider a specific model where the dark matter is a fermion in the adjoint representation of a hidden @xmath0 gauge group that confines at gev energies .
the absence of light fermionic states in the dark sector ensures stability of dark matter on cosmological time scales .
dark matter couples to the standard model via the neutrino portal , that is , the singlet operator @xmath1 constructed from the higgs and lepton doublets , which is the lowest dimensional fermionic singlet operator in the standard model .
this coupling prompts dark matter decay where the dominant decay channel has one neutrino ( and at least one dark glueball ) in the final state .
other decay channels with charged standard model particles involve more particles in the final state and are therefore suppressed by phase space . in consequence ,
the standard indirect detection signals like gamma - ray photons , antiprotons and positrons are suppressed with respect to the neutrino signal .
this coupling via the neutrino portal is most robustly constrained by super - kamiokande , which restricts the dark matter lifetime to be larger than @xmath2 seconds . in the near future
, the scenario will be probed by the new generation of neutrino telescopes .
antares will be sensitive to a dark matter lifetime of order @xmath3 seconds , while icecube / deepcore can probe a lifetime as large as @xmath4 seconds .
ru - nhetc-09 - 15 + 1.0 in * dark matter through the neutrino portal * + .1 in 0.5 in adam falkowski , jos juknevich , and jessie shelton 0.25 in _ department of physics + rutgers university + piscataway , nj 08854 _ 0.75 in |
in clusters containing high concentrations of massive stars ( b3 and earlier ) , binary and multiple systems provide a virtual cornucopia of information about the massive stars and the nature of their formation process .
in addition to the frequency at which they occur , we may examine the distribution of several quasi - preserved orbital parameters ( e.g. , period , mass ratio , and eccentricity ) and gain insight into many open questions surrounding massive stars , such as whether they form in isolation or as part of a competing menagerie of protostars ( for a review see * ? ? ? * ) .
other issues include whether there exists an evolutionary link between massive binary systems and type ib / c supernovae @xcite , whether there is a correlation between the massive binary fraction and the stellar density of the parent cluster @xcite , whether massive binaries are responsible for the runaway ob stars @xcite , and what role these stars play in short and long @xmath10-ray bursts @xcite . previous and current endeavors to characterize the binary properties of massive stars include the galactic o - star binary study of @xcite , the galactic o - star speckle survey of @xcite , the multi - cluster o - star binary statistical analysis of @xcite , the vlt - flames studies of the large and small magellanic clouds @xcite , the lucky imaging o - star study of @xcite , the adaptive optics o - star studies of @xcite , @xcite , and @xcite , the hst fine guidance sensor survey of cyg ob2 by @xcite , and the individual open cluster binary spectroscopic studies of @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , and @xcite .
these studies have found a broad range of binary fractions ( 063% ; * ? ? ? * ) and widely varying characterizations of the intrinsic orbital parameter distributions , including single power law descriptions with various indices and multiple - component descriptions requiring many free parameters . reasons for the diverse , and seemingly incompatible , results in the literature stem from the varying methodologies and biases inherent in the surveys themselves .
the majority of massive binary surveys in the literature include either a small sample of o stars within a given cluster or larger samples assembled from multiple smaller surveys encompassing different clusters & associations ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the larger surveys have the advantage of better statistics , but the smaller surveys have the advantage of a more homogeneous sample ( same formation environment and similar ages ) . the cyg ob2 radial velocity survey ( @xcite , @xcite , @xcite , @xcite , and @xcite , in prep ; papers i , ii , iii , iv , & vi ) , takes advantage of both a large and homogeneous sample of ob stars ( 114 massive stars ) with radial velocity coverage of up to 12 years to probe the binary characteristics of massive stars .
* kf07 ) analyzed the cyg ob2 radial velocity data from 19992005 and compared the raw velocities with the expectations of monte carlo simulations over a range of mass ratio and orbital separation distributions and at varying binary fractions .
they find that the likely binary fraction of ob stars within cyg ob2 is greater than 70% for an @xcite period distribution ( i.e. , log - flat ) and mass ratio distribution best characterized by a power - law slope of @xmath11 to @xmath12 ( depending on the choice of binary fraction ) .
though kf07 are very thorough in addressing different possible distribution scenarios ( a second `` twin '' population , a @xcite distribution , a @xcite secondary mass distribution , etc ) , they are forced to make certain simplifying assumptions in their code based on the limited binary information available .
a few of the more significant assumptions are circular orbits , treating all higher order systems as binaries rather than triples , quadruples , etc , and finally , that all velocity variation stems from orbital dynamics rather than stellar atmosphere line profile variations , which may be present in massive stars .
assuming circular orbits can cause an underestimation of the binary fraction because fewer circular binaries need to be generated to simulate the seemingly singular eccentric systems that exist in the cluster . with regard to the next assumption , treating ( difficult - to - detect ) triple and quadruple systems as binaries still yields a meaningful estimate of the binary fraction of the cluster , as long as this term is understood to include possible higher order systems .
lastly , the assumption that all velocity variation stems from orbital dynamics can lead to an overestimation of the binary fraction or an overestimation of number of massive companions , but this is an unavoidable limitation to the kf07 analysis .
we present here a follow - up to the initial results of papers i , ii , iii , iv , & vi , and the earlier analysis of kf07 .
we use the results of nearly 12 years of spectroscopic observations obtained on 114 massive stars in the cyg ob2 association to model the observed mass ratio , orbital period , and eccentricity distributions composed from the orbital information of 22 close massive binaries in the cluster .
we make use of all published close binary information for the cluster , both photometric and spectroscopic , including single - lined spectroscopic binaries ( sb1s ) , double - lined spectroscopic binaries ( sb2s ) , and eclipsing binaries .
we also use the radial velocity information of 110 ob stars from our original sample from @xcite to infer the binary fraction of the cluster .
additionally , we examine , in more detail , an apparent excess of binary systems with periods between 3 and 5 days that have been observed in cyg ob2 @xcite and other clusters ( ngc 6231 ; * ? ? ? * ) .
the current work provides an analysis using the largest sample of massive stars and close massive binary systems ever compiled for a group of stars with a shared history .
section 2 reviews the data used in this work .
section 3 summarizes the method employed in this work to model the orbital period , eccentricity , and mass ratio distributions using a monte carlo approach .
section 4 presents the results of the monte carlo simulations and discusses the most probable intrinsic orbital parameter distributions and the probable binary fraction of massive stars within the cluster .
discussion in section 5 explores possible physical origins for the observed orbital distributions and summarizes results that may inform models of close binaries as progenitors to supernova and @xmath10-ray bursts .
finally , section 6 summarizes the survey findings .
we obtained radial velocities for a total of 146 ob stars over 12 years .
these 146 ( given in table 1 of paper i ) were chosen from the _ ubv _ photometric and spectroscopic survey of cyg ob2 by @xcite .
nearly half of the stars were previously identified as an ob - type and the other half had colors consistent with classification of b3 or earlier . in this work ,
we utilize 110 of the 146 @xcite stars , termed the `` unbiased '' sample because they were selected on a `` binary - blind '' basis ( i.e. , no previous evidence of a companion ) . the objects in the `` unbiased '' sample average 14 observations over 1999 - 2011 with a mean velocity uncertainty of 9.7 . stars were removed from the original sample for having fewer than three observations at sufficiently high s / n needed to obtain reliable velocities ( i.e. , s / n@xmath430 where velocity uncertainties are @xmath1340 ) and for having spectral types later than b3 . additionally , we include the published orbital information for four systems not in our original sample , but likely members of cyg ob2 : schulte 5 from @xcite , a36 , a45 , and b17 from @xcite .
this constitutes 114 ob stars and is designated the `` complete '' sample .
the primary components in the `` complete '' sample have spectral types that range from b3v ( e.g. , mt298 ) to o3if ( e.g. , mt457 ) and masses that range from 7.6 to 80 ( initial masses were interpolated from the stellar evolutionary models of * ? ? ?
* ) with a mean mass of @xmath620 and mean temperature class of b0 ( 71 b3b0 stars and 43 o9.5o3 stars ) . there are fewer stars than included in kf07 ( who included stars from table 5 of paper i ) owing to removal of seven stars having temperature classes later than b3 ( mt196 , mt239 , mt271 , mt453 , mt493 , mt576 , and mt641 ) , 13 stars having fewer than three observations with sufficient s / n ( mt238 , mt343 , mt435 , mt477 , mt490 , mt492 , mt493 , mt509 , mt568 , mt621 , mt645 , mt650 , and mt712 ) , and one star that is a newly turned - on be star ( mt213 )
. additionally , ten stars not included in the analysis of kf07 but included here are schulte 3 , schulte 73 , mt005 , mt179 , mt186 , mt222 , mt241 , mt267 , mt421 , and mt426 .
these ten stars were excluded from the earlier work because they had fewer than three observations prior to 2007 . in total , this survey has produced full or partial orbital solutions for 19 binary systems . combined with the other published data ,
there are 24 known ob binary systems in cyg ob2 .
the locations of the first 20 are shown by the open circles in figure 16 of paper iv and listed in table 6 therein .
the remaining 4 are presented in paper vi .
in keeping with prior works , we characterize the orbital parameter distributions as single power - laws .
the mass ratio distribution is represented by @xmath14 ( where @xmath15 ) over a range of @xmath16 to 1 , the period distribution is represented by @xmath17 over a range of periods from @xmath18 to @xmath19 days , and the eccentricity distribution is represented by @xmath20 over a range of @xmath21 to @xmath22 . in the current approach , we constrain the indices , @xmath23 , @xmath24 , @xmath10 , by comparing the 12 years of radial velocity data , in addition to measured mass ratios , orbital periods , and eccentricities for the 22 binaries in cyg ob2 with periods under 30 days , to monte carlo - generated binary populations . for the standard application of the monte carlo ( mc ) code , we adopt @xmath25 ( roughly representing the mass ratio of a b3v @xmath26 m8v system ; @xcite ) , @xmath27 day , @xmath28 days , @xmath29 , and @xmath30 . the period upper limit is chosen based on our observing campaign time span of a few thousand days .
the eccentricity upper limit is chosen based roughly on where an orbital system becomes unbound .
we show later that our results are not sensitive to the exact choices for these upper and lower bounds , the main exception being that the inferred binary fraction does depend on the choice of @xmath19 .
we created synthetic binary populations using monte carlo ( mc ) methods by randomly drawing periods , mass ratios , and eccentricities from power - law distributions with indices ranging from -2 to 2 in increments of 0.1 , with binary fractions ranging from 20%100% in increments of 5% .
the mc code then generates a 4-dimensional parameter space composed of @xmath31 distinct `` families '' that are described by unique combinations of each parameter . during each iteration ,
radial velocities are calculated for a group of 110/114 synthetic stars whose primary masses are determined from the adopted values for cyg ob2 primaries . the number of synthetic velocities for each system is determined by the actual number of observations for each given star in the survey .
the code samples the synthetic velocity curve at intervals corresponding to the actual observing cadence for that star .
if the system is , at random , designated a binary , a mass ratio , eccentricity , and period are drawn from the power - law distributions characterized by the family currently being examined . an inclination is randomly drawn from a probability distribution described by @xmath32 over an interval of 0 to @xmath33 , representing a random orientation on the unit sphere @xcite . the orbital phase and angle of periastron are randomly drawn from uniform distributions over their allowed values ( i.e. , 02@xmath34 ) .
the code adds gaussian noise to each simulated radial velocity , where the magnitude of the noise is set by the observational uncertainty from the actual observations of the selected star .
for example , when a system representing mt720 is simulated , a primary mass of 17.5 is used and 32 radial velocities are calculated using the dates and errors listed in table [ xx.tab ] .
finally , the code produces 300 realizations of each family , designated a `` generation '' .
we compare the average properties of each generation to the cyg ob2 data .
we constrain the binary fraction by comparing the @xmath35 probability density function ( pdf ) for the simulated radial velocities to that of the unbiased sample of cyg ob2 stars .
we use the same method described in paper i , where the probability corresponds to the likelihood that the @xmath35 for each star s radial velocity measurements , as considered around the weighted mean velocity , would be exceeded by chance given @xmath36 degrees of freedom .
we compare the observed @xmath35 pdf to simulated ones for each generation using a 2-sided kolgorov - smirnov ( k s ) test . in our tests of the code
, we were able to recover the binary fraction of a simulated dataset to within @xmath37% . in order to compare the monte carlo distributions of @xmath38 , @xmath39 , and @xmath40 with the observed ones ,
it was necessary to apply a series of filters to the synthetic data to remove all binary systems with orbital solutions undetectable by this survey ( i.e. , the systems are designated `` single '' and the orbital parameters are not included in the final distributions ) .
we first eliminate all synthetic systems have @xmath35 probabilities @xmath135% ; such systems either have large errors , few measurements , and/or small velocity amplitudes that preclude full orbital solutions .
the next filter removes all synthetic systems that have a semi - amplitudes @xmath4115 , the minimum semi - amplitude for which we can derive a complete orbital solution given typical velocity uncertainties . in exceptional cases where the stellar lines are narrow , we are able to do better .
for example , mt174 , with @xmath42 ( paper vi ) , has the smallest velocity semi - amplitude yet measured in the survey .
finally , we retain only systems that have periods within the 1.526.5 day range covered by the 19 systems with orbital solutions where our survey is highly complete ( see section 4.1 ) .
the final mean extracted synthetic orbital parameter distributions for each generation are then compared to the observed ones using 2-sided k s tests .
the final best - fit combination of @xmath23 , @xmath24 , @xmath10 is determined by product of the resultant individual k s probabilities for each of the three parameter distributions . in our testing of the code , we produced a series of simulated datasets ( i.e. , orbital parameter distributions and ephemerides ) , similar to the cyg ob2 dataset , covering a large range of possible @xmath23 , @xmath24 , and @xmath10 combinations and binary fractions .
the power - law indices ranged from @xmath43 to @xmath44 ( and the mc code search grid was increased accordingly ) and binary fractions ranged from 20% ( the absolute minimum binary fraction of the cyg ob2 sample based on current observations ) to 100% .
the simulated datasets were constructed using the observation cadences , velocity uncertainties , and primary masses from the cyg ob2 data .
the code was able to recover the original distribution indices to within @xmath450.3 rms and the original binary fraction to within 5% rms .
the magnitude of the uncertainty on each index is limited by the current sample size of n=22 systems having orbital solutions .
the uncertainty on the binary fraction is limited by the mean number of observations per object and the mean observational uncertainty .
while our current approach explores a larger parameter space than kf07 , incorporates eccentric orbits , and utilizes the information of 22 close massive binary systems in addition to the radial velocity information of 12 years of spectroscopic observations , there are still several limitations stemming from our modeling approach , our dataset , and our current knowledge of cyg ob2 .
* limitations owing to our modeling approach : * as in kf07 , we still consider all velocity variation as stemming from orbital dynamics rather than atmospheric phenomena such as pulsations .
@xcite show that pulsations in evolved early - type stars can significantly affect radial velocity measurements , meaning that this effect may add scatter to radial velocity measurements in roughly one - quarter of our sample .
we also still treat higher order systems as binaries . at present ,
we know of three triple or higher order systems , cyg ob2 no.5 , cyg ob2 no.8 , and mt429 ( see references in table 6 of paper iv ) .
in addition , we limit characterization of each orbital parameter distribution to a single power - law , an assumption that may not hold .
* limitations owing to the dataset : * as mentioned , the cyg ob2 radial velocity survey probes binary systems with periods ranging from one day to a few thousand days , meaning we are not sensitive to extremely close ( and extremely rare ) systems with @xmath461 d. therefore , the mc code only draws from period distributions with periods ranging 1 to 1000 days .
additionally , the vast majority of spectroscopic observations were obtained june through november when cygnus is visible to northern latitudes .
this , coupled with typical velocity uncertainties of 9.7 , reduces our sensitivity to long period , high eccentricity systems ( i.e. , systems that may pass through periastron and exhibit their largest velocity variation during an unobservable period ) , systems with extreme mass ratios , and systems with small inclination angles . even though we have detected and computed solutions for systems with semi - amplitudes as small as 9 ( paper vi ) , we attempt to circumvent these limitations by designating systems in the code with semi - amplitudes smaller than @xmath47 as `` single '' .
another limitation comes in the form of not being able to detect unresolved sb2s . for systems exhibiting highly blended line profiles ( even at quadrature ) , such as long period binaries with mass and luminosity ratios near unity
, radial velocity variations of the blended line profile will be small and possibly undetectable at the resolution of the spectra in this survey , thereby being designated as `` single '' by the code .
@xcite discuss this effect in their analysis of the massive star binary fraction of ngc 6611 and conclude that the effect is minimal ( affecting their modeling of the binary fraction by only a few percent ) . additionally , there is also a negligible bias toward detecting short - period , high - mass - ratio systems that show up as sb2s as it only takes one observation at quadrature to determine that the system warrants follow - up .
these systems , however , have large velocity separations ( i.e. , large semi - amplitudes ) and will get flagged as a `` binary '' by the code with only a few observations . *
limitations owing to incomplete knowledge of cyg ob2 : * a few studies have shown that cyg ob2 may be contaminated by an older population of stars , meaning that evolutionary effects may be present ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
however , @xcite point out that the field star contamination is likely composed of f to late b stars at a radius of 1 degree .
this exceeds the radial extent of the region examined in this study and concerns stars of a later type than are included here .
on the other hand , some doubt remains in the membership of the 114 ob stars included in this work as recent findings by @xcite and @xcite provide conflicting distance estimates to cyg ob2 based on their complete orbital solutions calculated for the eclipsing sb2s , cyg ob2 no.5 ( @xmath61 kpc ) and b17 ( 1.51.8 kpc ) respectively .
we maintain that the contamination is minor , if it exists , owing to the localized region we examine , but even the removal of a few periods or mass ratios could influence the distributions we observe at present .
similar to this is the assumption that the members of cyg ob2 have a common age .
it has been suggested that even within the core region of cyg ob2 , there has been more than one epoch of star formation @xcite .
there are no clear predictions for how a population of close binaries should evolve over the 15 myr lifetime of most massive stars in young clusters , so we have no way to model such effects or even know if they are important .
we use the monte carlo code in conjunction with the `` complete '' sample of cyg ob2 data to infer the underlying distribution of orbital parameters , that is , @xmath23 , @xmath24 , @xmath10 . the orbital parameter information ( but not ephemerides ) for schulte 9 and mt070
are excluded given their much longer periods . in order to address the binary fraction ,
however , we use the data of the `` unbiased '' sample . ultimately , we find that the orbital parameter indices and binary fraction are essentially identical for both the `` unbiased '' and `` complete '' samples . binary surveys that examine the orbital period distribution probe widely varying ranges in period / separation and stellar mass , and they find widely varying characterizations of its functional form . for solar - type stars with periods from 110,000 + days , @xcite
find a log - normal distribution . among massive stars ,
most studies find a uniform distribution in log space ( @xmath48 ) , sometimes known as an @xcite distribution .
for example , @xcite find @xmath49 for a large number of galactic o - star binaries in the northern hemisphere .
@xcite also find an pik s distribution for a survey of visual and spectroscopic binaries in sco ob2 composed of a- and b - type primaries ( for 0.7 days @xmath50 days ) .
@xcite speculate that rather than a simple power - law distribution for o - star binaries , there may be a concentration of systems having periods between 2 and 5 days .
the o - star binaries of ngc 6231 show this trend @xcite .
however , @xcite conclude that a broken pik distribution provides a better description of the period distribution for spectroscopic o - binaries in nearby clusters , where the broken distribution could better explain the higher concentration of systems with periods less than 10 days . the observed orbital period , eccentricity , and mass ratio distributions for the 22 known close massive binaries in cyg ob2 are shown in figure [ disthist ] .
the orbital period and mass ratio distributions from @xcite are also shown and represented by the dotted lines in the upper and lower panels , respectively .
the cyg ob2 and @xcite period distributions show a concentration between 3 and 5 days ( with a peak around 45 days for cyg ob2 ) . a possible explanation for such a concentration might be observational bias owing to the average observing
run lasting a few days to a week .
three systems in cyg ob2 with periods between 3 and 5 days originate from other studies ( mt421 , mt429 , and b17 ) where this may be the case .
however , the remaining five systems in this period range originate from our study which has a time coverage of up to 12 years and individual observing run durations of 13 weeks .
this allows for a sensitivity to orbital periods up to 25 days without appreciable bias .
this abundance of 35 day systems is not limited to the sb2s having the largest mass ratios either .
three of the nine sb1s in cyg ob2 also have periods that are in the 45 days range .
this is illustrated in figure [ peq ] , which shows eccentricity versus orbital period ( upper panel ) and mass - ratio ( lower panel ) versus orbital period .
the sb1s and sb2s are the open and filled circles respectively , and the asterisks represent the orbital periods and eccentricities for the massive star binaries ( primaries of type b3 and earlier ) in the ninth spectroscopic binary catalog @xcite .
one can easily see the concentration of points around 35 days and a peculiar shortage of systems with periods from 714 days in the cyg ob2 data .
an initial hypothesis for this shortage of points might be that completeness plays a role .
however , with multiple studies with varying observing cadences contributing to the catalog of binaries in cyg ob2 , we maintain that a more likely observation would be to see a continually diminishing number of binary detections with increasing period as completeness declines .
given the existence of this excess in periods around 35 days , we caution that a simple power - law may be insufficient to describe the distribution of orbital periods for early - type stars in this cluster . in figure
[ biased1 ] , we show the results from the monte carlo analysis of the 22 close ob binaries in the `` complete '' sample .
the four panels display the @xmath51 ( 68.3% ) , @xmath52 ( 95.4% ) , and @xmath53 ( 99.7% ) contours relative to the maximum for the average probability of a generation as a function of two parameters for fixed best - fit values of the remaining two parameters .
for example , the upper left panel shows the average probability as a function of @xmath23 and @xmath10 for fixed best - fit values of @xmath24 ( 0.2 ) and @xmath54 ( 45% ) .
the upper right panel similarly shows the average probability as a function of @xmath24 and @xmath10 for fixed best - fit values of @xmath23 ( -0.1 ) and @xmath54 ( 45% ) .
the cross in each panel marks the location of the best fit for each parameter ( i.e. , the maximum of the average probability ) .
we find that the period distribution is confined at the 1@xmath55 level to @xmath56 , with a best fit of @xmath57 .
this points to a nearly log - flat distribution , consistent with pik s law @xcite , consistent with the previous o - star findings of @xcite , @xcite , @xcite , and kf07 .
figure [ pdist ] shows the cumulative period distribution of the cyg ob2 data ( diamonds ) and the best fit cumulative period distribution from the mc code ( solid curve ) .
the hump in the distribution at @xmath58 or @xmath59 days greatly influences the fit shown , and it is clear that a power - law of @xmath48 might be a better fit if this hump were not present or less pronounced .
indeed , no single power - law provides a good fit to the cumulative period distribution .
we address the peculiar nature of the period distribution in section 5 .
figure [ completeness_per ] shows the survey completeness as a function of orbital period for the best - fitting power - law indices .
we use the mc code to estimate the survey completeness by tallying the fraction of synthetic systems deemed detectable in our survey given the time sampling and velocity semi - amplitude limit of @xmath60 .
the survey is 88% complete at @xmath61 days and @xmath62%complete at @xmath63 days .
our completeness analysis implies that , given 14 detected binaries with periods @xmath64 days , there are an additional 12 undetected binary systems in this range .
statistically , the monte carlo code shows that all of these would be undetectable by the survey because they have some combination of low inclination , small mass ratio , large eccentricity , and larger radial velocity uncertainties .
if , for the sake of argument , we ignore the excess of short - period systems and assume that the @xmath49 power - law holds , the @xmath615 binaries in the 110 day period range would imply an equivalent number between 10 and 100 days .
there are 8 known systems in this range , leaving an additional 7 putative binaries as undetected .
the mean completeness over this range is @xmath65% , suggesting that @xmath612 of the 15 would be detectable by our survey .
this implies that the ongoing cyg ob2 survey should uncover as many as @xmath65 additional binaries with periods between 10 and 100 days , given sufficient long - term data .
for periods longer than 100 days , where there are presently only two systems with orbital solutions , schulte 9 @xcite and mt070 ( paper vi ) , detection becomes more difficult . in the case of schulte 9 , with a period of @xmath62.355 years and an eccentricity of nearly 0.7 , even an observing campaign such as this one experienced difficulty in uncovering this massive `` twin '' system .
all of the foregoing discussion , however , assumes an pik period distribution , and we have already shown in figure [ pdist ] that no single power - law adequately reproduces the data at the shortest periods , @xmath46 14 days .
observationally , binaries with solar - type primaries and periods longer than a year have a broad distribution of orbital eccentricities with a median value of 0.55 @xcite . for periods shorter than this ,
these systems have a higher concentration near zero eccentricity stemming from circularization of the orbits .
this is also shown to be true for o - star binaries with periods as long as 10,000 days @xcite .
although the method by which binary orbits are circularized and the critical period below which all are circular is unclear from existing data , a standard eccentricity versus period scatter diagram , like figure [ peq ] , illustrates that binaries with short periods do tend toward nearly circular orbits .
the figure shows a decline of mean eccentricity for periods under 10 days .
@xcite found that nearly all ( 92% ) early - type binaries with periods shorter than two days had an eccentricity less than 0.05 .
they also found that the eccentricities of nearly all of the early - type binaries in their survey could be explained by the @xcite tidal theory , which describes the circularization time for an equal mass binary with an initial eccentricity of @xmath66 as being related to the main sequence lifetime of that binary . @xcite
provide a modified version of the @xcite theory and address the evolution of the orbital parameters over the circularization time of the orbit .
the circularization timescales are , on average , a third of those predicted by @xcite and in better agreement than the predictions of @xcite with the massive star binary statistics of @xcite .
the predicted circularization time for an early - type ( pair of b0v stars ) , nearly equal - mass binary at a separation comparable to the cyg ob2 early b - type binaries ( e.g. , @xmath670.16 au for mt720 ) is close to the main - sequence lifetime of an early b star .
however , a study by @xcite found that there are a significant number of early - type binaries with periods up to 10 days with an eccentricity near zero that can not be explained with the @xcite theory and are unlikely to have been primordially circular .
@xcite have proposed a method by which orbits become significantly circularized during the star formation process , thereby diminishing the conflict between circularization time and the theoretical life span of massive stars .
however , the theory assumes the binary is formed concurrently with the formation of the components , and that is a highly debated topic in massive star formation theory @xcite . in figure [ edist ]
we show the mc code best fit to the cyg ob2 eccentricity distribution . as in figure
[ pdist ] , diamonds show the normalized cumulative distribution ( left y - axis ) and the solid curve shows the normalized cumulative distribution for @xmath68 .
figure [ biased1 ] ( upper left and right panels ) shows that the index @xmath10 is confined , at the @xmath51 level , to @xmath69 with a best fit of @xmath2 .
the steep negative power - law is required to reproduce the abundance of systems with low eccentricity .
a k s test shows that the @xmath68 power law is compatible ( i.e. , the observed sample and the synthetic sample are consistent with being drawn from the same parent distribution ) with the data at the @xmath690% level .
the absence of eccentricities with @xmath70 is similar to what is observed for short - period binaries in other ob clusters ( e.g. ngc 6231 ; * ? ? ?
highly eccentric systems are apparently rare in this period regime , either because circularization mechanisms are efficient , or possibly because they are more easily disrupted during encounters with other cluster stars .
finally , we note that the eccentricity distribution within cyg ob2 is also compatible with that of ob binaries in the 9th spectroscopic binary catalog at the @xmath650% level .
we show an estimate of the survey completeness as a function of eccentricity for two different period ranges in figure [ completeness_ecc ] .
the solid and dot - dash curves represent the completeness for @xmath71 days and @xmath72 days respectively .
the plot indicates that the survey is @xmath675% complete at @xmath73 for binaries with @xmath71 days and @xmath688% complete at @xmath74 for binaries with @xmath72 days . for @xmath72 days ,
the survey is likely to have detected the vast majority of systems , even those with large eccentricities , if they exist .
the nature of the mass ratio distribution is one of the more highly debated and unresolved questions pertaining to massive binaries and their formation .
the distribution is difficult to characterize because of the unavoidable bias of observational studies , including this one , against low-@xmath38 systems .
the general consensus is that short - period binaries have a different mass ratio distribution compared to the long - period binaries . @xcite and @xcite
show this with lower - mass binaries , but they also point out that the period range over which this change occurs is still ill - defined .
most studies on the mass ratio distribution are for mid to late - type stars .
a few examples include @xcite , who find @xmath75 ( for @xmath76 and periods 110,000 + days ) for a large survey of solar - type spectroscopic binaries , @xcite , who find @xmath77 to @xmath78 for a large survey of spectroscopic and visual a- and b - type primaries in sco ob2 , @xcite , who find @xmath79 for all the sb1s and @xmath80 for all the sb2s in the solar neighborhood ( for periods 1500 days ) , and @xcite , who find @xmath81 for mass ratios of @xmath82 and @xmath75 for mass ratios of @xmath83 in binaries of b - type and later ( for periods @xmath41 day to 10,000 + days ) . among the massive stars ,
studies such as @xcite conclude from their study of 21 eclipsing massive binaries ( @xmath84 days ) in the small magellanic cloud , that the secondary mass function for massive binaries does not follow a @xcite slope . instead
, the authors find a flat distribution of mass ratios and predict that roughly 55% of all massive systems are part of a `` twin '' system with a mass ratio of @xmath85 .
@xcite have refuted this proposition , but also find a preference for more massive companions among o - stars in clusters .
the distribution of observed mass ratios in cyg ob2 , shown in figure [ disthist ] , appears relatively flat , except for a slight preference for mass ratios between 0.7 and 0.9 .
this may be an artifact of using a standard histogram for small numbers .
not surprisingly , the upper end of the distribution is populated exclusively by sb2s , while the lower end of the distribution is almost entirely populated by sb1s . for five of these single - lined systems , we calculated the expected orbital inclination , @xmath86 , between @xmath87 and 90 , using the inclination lower limits estimated from paper ii , paper iii , and paper vi .
we calculate the most probable mass ratios to be 0.27 , 0.30 , 0.03 , 0.22 , and 0.11 for mt059 , mt145 , mt174 , mt258 , and mt267 , respectively .
these are close to the lower - limit values for @xmath38 because @xmath86 is not far from the upper limit on @xmath88 .
there is a possibility , of course , that for one or more of these systems , the orbital inclination could be less than @xmath86 , and therefore , the system(s ) would have a more massive companion .
in such a case , our present estimates may incorrectly inflate the number of systems with mass ratios between 0.05 and 0.30 , and the true distribution would not be as flat as currently shown in figure [ disthist ] . from the mc code
we find a best fit of @xmath89 for the mass ratio distribution .
figure [ qdist ] shows the normalized cumulative distribution of mass ratios ( diamonds ) , along with the corresponding simulated cumulative distribution for @xmath90 ( thick curve ) .
the 1@xmath55 contours ( top left and bottom left panels in figure [ biased1 ] ) show a range of @xmath91 .
such a shallow positive power - law indicates that mass ratios in the cluster are quasi - uniformly distributed between @xmath92 and @xmath93 with a preference for larger mass ratios .
this type of distribution implies that the secondaries are not drawn from a @xcite stellar imf , in contrast to lower - mass binary populations which exhibit an abundance of low mass ratios .
furthermore , the histogram of mass ratios in figure [ qdist ] is also inconsistent with the `` twin - heavy '' population suggested by @xcite , whereby 45% of systems are predicated to have @xmath85 .
figure [ completeness_mrs ] shows the completeness of the survey as a function of mass ratio . for systems with @xmath72 days ( dot - dash curve ) ,
the survey is at least 90% complete down to @xmath94 , but there is a sharp drop in the fraction of systems detected for @xmath95 .
the average completeness is 1020% lower , overall , for longer period systems with @xmath71 days ( solid curve ) .
our measurement of mt174 ( @xmath96 ; @xmath42 ) below our nominal semi - amplitude detection threshold of 15 provides evidence that such low - amplitude , low-@xmath38 systems exist and are detectable in our survey .
figure [ completeness_mrs ] shows that incompleteness is an issue only at the lowest mass ratios .
any attempt to correct for incompleteness at @xmath97 would add low-@xmath38 systems to the histogram and serve to flatten the distribution .
table [ ostat.tab ] shows a selection of the larger surveys that investigated the o - star binary fraction within individual clusters / associations .
column 1 contains the common cluster identification , column 2 is the number of o stars included in the survey , column 3 is the number of early b stars included in the survey , column 4 is the lower limit on the o - star binary fraction , column 5 is the lower limit on the massive star binary fraction ( o and early - b primaries ) , and column 6 references the original works . the observed o - star binary fractions ( designated @xmath98 rather than @xmath54 here to distinguish from the intrinsic binary fraction ) within the clusters listed range widely from 0
67% ( column 4 ) .
more inclusively , the massive star @xmath98 ( with primaries b3 and earlier ; column 5 ) , varies between 4% ( ngc 330 in the smc ) and 53%(ngc 6231 ) . in paper
i , we find an initial lower limit on the binary fraction of 30% based on the @xmath35 probability density function for radial velocities from the first six years of observations of cyg ob2 , which would place it near the middle of the binary fraction range among massive stars .
kf07 find a more probable binary fraction of @xmath99% based on their monte carlo analysis of the raw radial velocities , which would place it closer to that of ngc 6231 . however , based purely on the binary detections of papers ii iv & vi , the hard lower limit on the binary fraction among massive stars in cyg ob2 is 21% ( i.e. , 24/114 ) .
we show the observed @xmath35 probability density function ( pdf ) for the radial velocities of cyg ob2 stars in figure [ probchi ] .
there are 53 having @xmath1005% and 43 with @xmath1001% .
these systems encompass the 24 known binary systems and additional candidates for radial velocity variables .
the number of variables has increased from paper i despite removing some low s / n spectra that may have caused extraneous velocities .
this is understandable given the additional data acquired on the sample , especially on long period binaries like mt070 ( @xmath101 years ) .
figure [ chisqdist ] shows the mc code best fit to the cumulative @xmath35 pdf for radial velocities in cyg ob2 using the `` unbiased '' sample of 110 stars .
the solid curve represents the cumulative @xmath35 pdf produced by the mc code with best - fit indices of @xmath90 ( @xmath16=0.005 ) , @xmath102 ( @xmath19=1000 days ) , and @xmath68 ( @xmath21=0.0001 , @xmath22=0.9 ) .
the open diamonds represent the normalized cumulative @xmath35 pdf for the unbiased sample of ob primaries .
the best mc code fit results in a binary fraction of @xmath103% if all binary systems are assumed to have @xmath71 days .
simulated binary populations with binary fractions of 4050% in increments of 1% and with fixed @xmath23 , @xmath24 , and @xmath10 , were individually examined for better agreement with the data .
these additional simulations show that a binary fraction of 44% yields a slightly better fit to the data .
figure [ unbiased ] shows the results of the monte carlo analysis for the 18 ob close binaries comprising the `` unbiased '' sample .
similar to figure [ biased1 ] , the four panels provide the @xmath51 , @xmath52 , and @xmath53 probability contours for @xmath23 , @xmath24 , @xmath10 , and binary fraction .
the three panels depicting probabilities for power - law indices ( upper left , upper right , lower left ) are nearly identical to the results from the mc analysis of the `` complete '' sample shown in figure [ biased1 ]
. the best - fit @xmath24 is slightly larger ( 0.4 instead of 0.2 ) owing to the exclusion of several shorter - period binaries from the unbiased sample .
the best - fit @xmath23 and @xmath10 are the same , just with larger uncertainties .
the contours in the lower right panel show that the binary fraction is confined to @xmath104% @xmath105% at the @xmath51 level and @xmath106% @xmath107% at the @xmath53 level for @xmath28 days .
this result is consistent with the range of massive binary fractions for galactic clusters listed in table [ ostat.tab ] . in order to compare our current results to the monte carlo analysis of kf07 , who explored a larger range of periods ( separations of 0.0210,000 a.u . corresponding to @xmath108110@xmath109 days for two 20 stars ) , we also performed a mc run with @xmath110 days . in both analyses
10,000 a.u .
is considered a reasonable upper limit for bound systems . in this case , we find the binary fraction must be @xmath111% in order to replicate our observed @xmath35 pdf .
this is in rough agreement with the conclusions of kf07 , but based on extrapolating the power - law distributions of mass ratios and periods determined from close binaries over many orders of magnitude to characterize loosely bound systems .
is an extrapolation of the power law index @xmath24 to periods as long as 10@xmath112 days warranted when our radial velocity survey is sensitive only to systems with periods less than about 10 years ?
@xcite conducted an adaptive optics and @xmath113 fine guidance sensor survey of 75 cyg ob2 systems and detect highly probable companions around 31 ob primaries at separations @xmath13200 a.u .
( p@xmath13630 years ) . of these ,
nine ( 9/75 = 12% ) have probable periods in the range @xmath114@xmath5 years ( s. caballero - nieves , private communication ) to turn the angular separations into estimated periods . ] .
if we adopt @xmath115 years ( @xmath116 days ) , the mc code provides a best fit binary fraction of @xmath690% , and this extrapolation of the @xmath117 power law predicts approximately 15 binaries with orbital periods in each 1 dex interval ( i.e. , 15 with periods between 10 yr and 100 yr , 15 between 100 yr and 1000 yr , etc . )
given the 110 stars in our unbiased sample .
this translates into about 14% of the binary systems falling within each logarithmic period interval .
hence , the predictions of an extrapolated power - law appear consistent with the emerging high angular resolution observations at larger separations .
we therefore suggest that the binary fraction among massive stars may be as high as @xmath118% if periods as long as 10@xmath119 years are allowed .
in addition to exploring the effect of varying @xmath19 , we also examined the effect of varying @xmath16 and the semi - amplitude detection threshold , @xmath120 , in the mc code .
as @xmath120 is changed from 15 to 20 or larger , a greater fraction of synthetic systems are deemed undetectable by the code , and the detected systems have smaller separations and/or larger companions .
therefore , we expect some combination of increase in @xmath23 ( larger mass ratios ) , decrease in @xmath24 ( shorter periods ) , and increase in binary fraction .
we find insignificant change ( i.e. , @xmath23 , @xmath24 , @xmath10 are the same ) when @xmath121 .
a significant change in the indices is not observed until @xmath120 is increased to much larger values of @xmath122 .
we are also confident our detection threshold is much closer to 15 , as evidenced by the measurement of mt070 @xcite .
we conclude that the minimum semi - amplitude for binary detection is appropriately set at 15 . in the case of increasing @xmath16 , the lowest mass ratio simulated in the code , we primarily expect to see a decrease in @xmath23 and a decrease in the binary fraction .
binaries will have more massive companions and semi - amplitudes will be larger , meaning that a smaller fraction of binary systems will be deemed undetectable in the code .
we observed only a small decrease in binary fraction in runs with @xmath123 ( a few percent and still consistent with our initial results ) .
we also find only a small increase in @xmath23 and no change in @xmath24 or @xmath10 .
however , increases in @xmath23 become larger when @xmath124 .
for example , @xmath23 changes from 0.1 to @xmath125 for the `` complete '' sample when @xmath126 ; however , such a large @xmath16 is probably unphysical given the existence of mt070 ( @xmath127 ) and mt267 ( @xmath128 ) .
these results , for both tests , indicate that our mc results are not sensitive to small changes in the adopted values of the mass ratio lower limit or semi - amplitude threshold .
the presence of an excess of binary systems with periods around 35 days ( with a peak at 45 days ) and a shortage of systems with periods around 714 days is not only observed in cyg ob2 . @xcite
observe a similar occurrence in ngc 6231 , and the excess and shortage of systems occur at roughly the same ranges in period .
systems in cyg ob2 that lie within the excess encompass a range of spectral types , mass ratios , and binary types ( i.e. , eclipsing , sb1 and sb2 ) and are scattered randomly across the cluster , having no preferred location either in dense or sparse regions .
we do not believe this peculiar distribution can be attributed to observational biases given the extended time coverage of this survey and the different observing cadences of the observing runs that have contributed to the cyg ob2 binary catalog . while the existence of an excess of systems with periods of around 35 days seems well - established in the literature and in our present data , the possibility of a shortage in the 714 day range has not been previously noted .
a comparison between the cyg ob2 cumulative period distribution with the data from @xcite reveals a flattening in this 714 day range ( though we should be cautious given the use of a standard histogram and small numbers ) .
their survey ( of 28 systems ) contains three in this 714 day range while ours ( 22 systems ) contains one . while we find that no single power - law can adequately describe the log - period distribution in cyg ob2 , it is unlikely , given any power - law distribution , to find so few systems between 7 and 14 days .
we examined the probability that only a single system would be observed in the 714 day period range given a random drawing of 22 systems with @xmath129 days ( for @xmath130 - 1 to 1 ) .
random samples were drawn 10,000 times and the final percentages were adjusted for a 90% completeness level ( the mean level as measured in figure [ completeness_per ] for the 714 day range ) .
the likelihood of detecting a single system for @xmath48 is @xmath62% .
however , if the power - law describing the log - period distribution is assumed to be closer to @xmath131 , then the probability rises quickly to @xmath613%(with @xmath132% for @xmath133 , and @xmath134% for @xmath1300.5 and 1 ) .
this tells us that if the power - law(s ) that describe the log - period distribution have @xmath135 , the shortage is a real phenomenon at the 2@xmath55 level . in an attempt to replicate the cyg ob2 period distribution with a physically motivated prescription , we made a version of the monte carlo code to take 70% of the systems assigned a period between 7 and 14 days and reassign them assuming the same power - law describing the rest of the distribution to periods between 3 and 5 days .
the 70% reassignment fraction reflects a 90% completeness level and detection of one out of a probable 3 or 4 systems in this range .
this approach approximates what would happen if binary systems with periods 7 and 14 days migrated to shorter periods .
the best fit using this approach is shown in figure [ pdist_hump ] and comes from reassigning systems with a period between 7 and 13.5 days to a period between 4 and 5 days ( corresponding to the peak of the excess in the cyg ob2 distribution ) with only a slight increase in @xmath24 ( 0.5 and within the uncertainty of our results ) .
@xmath23 , @xmath10 , and the binary fraction are same as in section 4 . the normalized cumulative distribution for the cyg ob2 data is represented by the diamonds and the solid curve represents the normalized cumulative period distribution for the best fit using this version of the mc code .
though this fit requires a slightly more positive slope for the initial period distribution , the overall fit to the data is visually better than in figure [ pdist ] , and the multiple changes in the slope are also replicated well . a two - sided k s test yields a compatibility of greater than 99% ( compared to @xmath660% with the simple power - law approach ) .
a plausible explanation for such period migration could be the mechanism proposed by @xcite whereby the circularization of the orbit occurs primarily during the formation process instead . in figure [ pq30 ]
we show the evolution of orbital period / semi - major axis and mass ratio for a binary system with a total mass of 30 m@xmath136 as the primary loses mass to the secondary .
the plotted curves show tracks of constant angular momentum as systems evolve toward shorter periods with mass ratios approaching unity , assuming that mass and angular momentum are conserved , @xmath137 ( e.g. , equation 8 of @xcite ) . here
@xmath138 and @xmath139 are the initial semi - major axis and mass ratio while @xmath140 and @xmath38 are the final semi - major axis and mass ratio after mass transfer .
the two solid curves are tracks for systems with initial mass ratios @xmath141 and initial periods of 7 and 14 days .
these systems may end up having mass ratios near unity , but their final periods are longer than 5 days .
the dotted curves are tracks for systems that begin with mass ratios @xmath142 and periods of 7 and 14 days .
these systems may end up with mass ratios near unity and periods in the range 35 days .
this is precisely where we observe an excess of systems with short periods in cyg ob2 , and we also find a relative abundance of systems with mass ratios @xmath1430.80.9 not exactly `` twins '' , but systems where significant mass transfer may have taken place .
this figure indicates that , if the @xcite mechanism is the operative one producing an abundance of short - period , high-@xmath38 systems , then some fraction of the population of 35 day systems should originate as systems with @xmath144 with 714 day periods . while the evidence for `` missing '' 714 day systems in cyg ob2 is compelling , we can not claim , on the basis of the present data , to see the corresponding lack of systems with @xmath144 .
our survey is sufficiently incomplete in this mass ratio range that observational biases preclude any firm conclusion .
the cyg ob2 radial velocity survey originated as an observational effort to inform theoretical models predicting the frequency of energetic events like supernova and @xmath10-ray bursts that may have close binaries as progenitors . a key ingredient of these modeling efforts is the fraction of massive stars that might interact through mass transfer or undergo a common envelope phase , especially during the post - main - sequence evolution of either star in the system .
red supergiants are the descendants of massive stars , so we adopt the rsg radii deduced by @xcite which range between 200 r@xmath136 and 1500 r@xmath136 .
their figure 3 shows stellar evolutionary tracks indicating that stars in the mass range 15 30 m@xmath136 may reach or exceed 1000 r@xmath136 while stars with initial masses over 10 m@xmath136 can generally be expected to reach 500 r@xmath136 .
we adopt 500 r@xmath136 as a maximum evolved radius for stars in our sample , meaning that systems with semi - major axes smaller than this value will interact strongly , probably through the formation of a common envelope .
other systems may interact via mass transfer after roche lobe overflow .
@xcite provides an analytic description of the roche lobe radius as a function of mass ratio .
his tabulated values show that the roche lobe radii for binary systems with mass ratios @xmath1450.2 1.0 fall in the range 0.38@xmath140 0.55@xmath140 where @xmath140 is the separation of the components .
consequently , we adopt 1000 @xmath146 ( 4.7 a.u . ) as a conservative threshold separation , below which systems are deemed to interact during the evolution of the system . in our monte carlo code , we flag all systems with periastron distances less than 1000 @xmath146 as capable of interaction . assuming @xmath90 , @xmath147 , @xmath68 , and @xmath115 years ,
this predicts that 45% of the binary systems will have sufficiently close companions to interact during the binary s lifetime ( 37% if we adopt the slightly larger @xmath148 from section 5.1 ) .
the luminous contribution from secondary stars , which are unresolved in photometric surveys , even with the @xmath149 or adaptive optics , is not negligible .
consequently , measuring accurate distances to star clusters using spectroscopic parallax , main sequence fitting , or similar standard - candle techniques is systematically in error if close companions are neglected .
as part of our monte carlo code we tallied the mean v - band light contribution from primary and secondary stars by using a zero - age main sequence ( zams ) relation between stellar mass and absolute v magnitude @xcite .
for the best fitting power law indices described above , allowing for periods as long as 10@xmath119 yr , we find that the secondary stars contribute 16% of the total stellar light at v - band .
this is equivalent to a systematic error in the distance modulus of 0.16 mag and a systematic distance error of about 8% , in the sense that distances derived without correcting for luminous secondaries will be too small .
we expect the secondary luminosity fractions in evolved populations to vary stochastically about this 16% zams value as the primary and secondary components throughout the population become giants and supergiants .
the intricacies of mass exchange between components will further limit any simple description of the time evolution in the nominal zams value .
in this , our fifth work of this series , we present the results of a monte carlo analysis using a decade of radial velocity data on 114 massive stars in the cyg ob2 association .
this survey is unique in terms of number of massive stars and binary solutions analyzed .
it improves on kf07 and other works ( e.g. , @xcite , @xcite , @xcite , @xcite ) in that we utilize the radial velocity information obtained over a larger time span and make use of the nearly complete orbital information including eccentricities on 22 of the 24 known massive binaries . the cyg ob2 binaries have been observed by spectroscopic and/or photometeric means and cover a period range of 1.47 days 7.58 years ( with all but a few having @xmath150 days ) , an eccentricity range of 00.53 , and a mass ratio range of 0.030.99 .
on the strength of these data , we compute the binary fraction for a large , young population ( 23 myr ) of massive stars from a common origin . in addition
, we use this information to model the most probable intrinsic distributions for period , mass ratio , and eccentricity for short - period systems having p@xmath426 days . *
the data imply a hard lower limit on the binary fraction of 21% ( i.e. , 24/114 systems ) .
the monte carlo analysis of the unbiased sample of 110 ob stars ( i.e. , not including a36 , a45 , b17 , and schulte 5 ) , indicates a massive star binary fraction of @xmath3% , assuming that the allowed period range is 11000 days ( separations @xmath60.01 few @xmath151 10 au ) . however , if we extrapolate the best fit power - law distributions over several orders of magnitude and allow for periods as long as 10@xmath119 years , the binary fraction may be as large as @xmath118% .
these results are consistent with the probable number of binaries detected in the adaptive optics and _ hst _ survey of cyg ob2 by @xcite and with the previous results of kf07 . *
the best - fitting distributions of mass ratio , period , and eccentricity are described by power - laws with indices @xmath89 , @xmath57 , and @xmath2 , respectively , assuming lower and upper bounds on the power - law distributions of q=0.0051.0 , p=11000 days , and e=0.00010.9 .
these values for @xmath23 and @xmath24 are broadly consistent with the previous results of kf07 , as well as @xcite , @xcite , @xcite , and the analysis of @xcite .
the mass ratio distribution is not consistent with the `` twin - heavy '' population proposed by @xcite , but suggests a weak preference for larger mass ratios ( i.e. , @xmath152 ) .
* the distribution of orbital periods is not consistent with any single power - law , owing to a significant excess of systems with periods around 35 days ( peaking at 45 days ) and a deficit of systems with periods around 714 days .
such a 35 day excess has been observed among massive stars in ngc 6231 @xcite , but the latter phenomenon has not been previously reported .
we propose that it may stem from binary systems with periods initially in the range of @xmath6714 days migrating to shorter periods during a period of pre - main - sequence mass transfer , via the mechanism proposed by @xcite .
* the fraction of massive - star binaries ( or higher order systems ) having companions close enough to interact during pre- or post - main - sequence evolution , defined here as requiring a semi - major axis @xmath44.7 au ( @xmath61000 ) , is @xmath645% .
this fraction is a key ingredient in recipes that predict the cosmic rates of binary - induced phenomena arising from massive stars , such as some classes of supernova and @xmath10-ray bursts .
we present these findings in the anticipation that they , along with the analyses of other clusters ( such as the work with sco ob2 by @xcite , ngc 6231 by @xcite , ngc 2244 by @xcite , and ngc 6611 by @xcite ) , will aid our understanding of numerous issues involving massive stars , including their formation , the predicted rates of supernovae type ib / c arising through binary channels , and whether there is a correlation between cluster density and the binary fraction therein . this ongoing survey will continue to uncover massive spectroscopic binaries in cyg ob2 .
these new sb1s and sb2s will help refine our characterization of the orbital parameter distributions and increase our understanding of the nature of the excess in orbital periods between @xmath635 days .
while the present data has provided solid constraints on the massive binary characteristics in cyg ob2 , it is possible that the binary fraction or distribution of periods / mass ratios / eccentricities varies temporally as a cluster ages or is a function of initial cluster density .
discerning whether these additional variables are important will require more extensive surveys over a range of age and formation environment .
we would like to acknowledge the helpful advice and attention to detail we received from our referee in the process of making this a much better manuscript .
we thank mark krumholz and hans zinnecker for inspiring conversations about phenomena in short - period binaries , chris fryer for helpful discussions regarding interacting binary systems , and saida caballero - nieves for sharing her imaging data of cyg ob2 in advance of publication .
we are very appreciative for the continued advice and encouragement throughout the project from ginny mcswain .
additionally , we are grateful for the continued support from the national science foundation through research experience for undergraduates ( reu ) program grant ast 03 - 53760 , through grant ast 03 - 07778 , ast 09 - 08239 , and the support of the wyoming nasa space grant consortium through grant nng05g165h .
blaauw , a. 1961 , ban , 15 , 265 caballero - nieves , s. m. , gies , d. r. , roberts , l. c. , & turner , n. h. 2011 , bulletin socit des sciences de lige , 80 , 639 clark , j. s. , ritchie , b. w. , & negueruela , i. 2010 , a&a , 514 , 87 comern , f. , pasquali , a. , figueras , f. , torra , j. 2008 , a&a , 486 , 453 comern , f. , et al . 2002 , a&a , 389 , 874 de becker , m. , rauw , g. , manfroid , j. , & eenens , p. 2006 ,
a&a , 456 , 1121 drew , j. e. , greimel , r. , irwin , m. j. , sale , s. e. 2008 , mnras , 386 , 1761 drilling , j. s. , & landolt , a. u. 2000 , in astrophysical quantities , ed .
a. n. cox ( 4th ed .
; new york ; springer ) , 381 duquennoy , a. , & mayor , m. 1991 , a&a , 248 , 485 duchne , g. , simon , t. , eislffel , j. , & bouvier , j. 2001 , a&a , 379 , 147 eggleton , p. p. 1983 , apj , 268 , 368 evans , c. j. , lennon , d. j. , smartt , s. j. , & trundle c. 2006 , a&a , 456 , 623 fisher , j. , schrder , k. p. , & smoth , r. c. 2005 , mnras , 361 , 495 fryer , c. l. , woosley , s. e. , & hartmann , d. h. 1999 , apj , 526 , 152 garcia , b. , & mermilliod , j. c. , 2001 , a&a , 368 , 122 garmany , c. d. , conti , p. s. , & massey , p. 1980 , apj , 242 , 1063 gies , d. r. , 1987 , apjs , 64 , 545 giuricin , g. , mardirossian , r. , & mezzetti , m. 1984 , a&a , 134 , 365 goldberg , d. , mazeh , t. , & latham , d. w. 1991 , apj , 591 , 397 hillwig , todd c. , gies , douglas r. , bagnuolo , william g. , jr . , huang , wenjin , mcswain , m. virginia , wingert , david w. 2006 , apj , 639 , 1069 hogeveen , s. j. 1992 , ap&ss , 196 , 299 khaliullin , kh . f. , & khaliullina , a. i. 2010 , mnras , 401 , 257 kiminki , d. c. , kobulnicky , h. a. , kinemuchi , k. , irwin , j. s. , fryer , c. l. , berrington , r. c. , uzpen , b. , monson , a. j. , pierce , m. j. , & woosely , s. e. 2007 , apj , 664 , 1102 kiminki , d. c. , kobulnicky , h. a. , gilbert , i. , bird , s. , & chunev , g. 2009 , aj , 137 , 4608 kiminki , d. c. , kobulnicky , h. a. , ewing , i. , lundquist , m. , alexander , m. , vargas - alvarez , c. , & choi , h. 2010 , apj , 745 , 1 kiminki , d. c. , mcswain , m. v. , & kobulnicky , h. a. 2008 , apj , 679 , 1478 kobulnicky , h. a. , & fryer , c. l. 2007 , apj , 670 , 747 kobulnicky , h. a. , smullen , r. , kiminki , d. c. , long , g. , & runnoe , j. 2012 , in prep kouwenhoven , m. b. n. , brown , a. g. a. , portegies zwart , s. f. , & kaper , l. 2007 , a&a , 474 , 77 krumholz , m. r. , & thompson , t. a. 2007 , apj , 661 , 1034 kurtz , m. j. , mink , d. j. , wyatt , w. f. , fabricant , d. g. , torres , g. , kriss , g. a. , & tonry , j. l. 1991 , in astronomical data analysis software and systems i , asp conf .
worrall , c. biemesderfer , and j. barnes , p. 432
law , d. r. , steidel , c. c. , erb , d. k. , larkin , j. e. , pettini , m. , shapley , a. e. , & wright , s. a. 2009 , apj , 697 , 2057 lejeune , t. , & schaerer , d. 2001 , a&a , 366 , 538 levesque , e. m. , massey , p. , olsen , k. a. g. , et al .
2005 , apj , 628 , 973 linder , n. , rauw , g. , manfroid , j. , damerdji , y. , de becker , m. , eenens , p. , royer , p. , & vreux , j. m. 2009 , a&a , 495 , 231 mahy , l. , naz , y. , rauw , g. , gosset , e. , m. de becker , sana , h. , & eenens , p. 2009 , a&a , 502 , 937 maz - apellniz , j. , walborn , n. r. , galu , h. .
, & wei , l. h. 2004 , apjs , 151 , 103 maz apellniz , j. 2010 , a&a , 518 , a1 martins , f. , schaerer , d. , & hillier , d. j. 2005 , a&a , 436 , 1049 mason , b. , hartkopf , w. i. , gies , d. r. , henry , t. j. , & helsel , j. w. 2009 , apj , 137 , 3358 massey , p. , & thompson , a. b. 1991 , aj , 101 , 1408 mermilliod , j. c. 1995 , in information & on - line data in astronomy , ed .
d. egret , & m. a. albrecht ( kluwer academic publishers , dordrecht ) , 127 miller , g. e. , & scalo , j. m. 1979 , apjs , 41 , 513 naz et al . 2010 ,
apj , 719 , 634 pik , e. j. 1924 , tartu obs .
, 25 pigulski , a. , & kolaczkowski , z. 1998 , mnras , 298 , 753 pinsonneault , m. h. , & stanek , k. z. 2006 , apj , 639 , 67 podsiadlowski , ph . , joss , p. c. , & hsu , j. j. l. 1992 , apj , 391 , 246 pourbaix , d. , et al .
2004 , a&a , 424 , 727 rauw , g. , & de becker , m. 2004 , a&a , 421 , 693 salpeter , e. e. 1955 , apj , 121 , 161 sana , h. , & evans , c. j. 2011 , iaus , 272 , 474 sana , h. , gosset , e. , & evans , c. j. 2009 , mnras , 400 , 1479 sana , h. , gosset , e. , naz y. , rauw , g. , & linder , n. 2008 , mnras , 386 , 447 sana , h. , momany , y. , gieles , m. , et al .
2010 , a&a , 515 , a26 schulte , d. h. 1958 , aj , 128 , 41 stroud , v. e. , clark , j.s . , negueruela , i. , roche , p. , & norton , a.j .
2009 , a&a , 511 , 84 tassoul , j. 1988 , apj , 324 , 71 turner , n. h. , ten brummelaar , t. a. , roberts , l. c. , et al .
2008 , aj , 136 , 554 witte , m. g. , & savonije , g. j. 2001 , a&a , 366 , 840 wright , n. j. , drake , j. j. , drew , j. e. , & vink , j. s. 2010 , apj , 713 , 871 zahn , j. p. 1977 , a&a , 57 , 383 zinnecker , h. , & yorke , h. w. 2007 , ara&a , 45 , 481 lrrrrr ngc 2004 ( lmc ) & 4 ( 1 ) & 83 ( 21 ) & 25% & 25% & 1 + ngc 330 ( smc ) & 6 ( 0 ) & 76 ( 3 ) & 0% & 4% & 1 + ngc 346 ( smc ) & 19 ( 4 ) & 59 ( 19 ) & 21% & 29% & 1 + trumpler 14 & 7 ( 1 ) & & 14% & & 2,3 + ic 1805 & 10 ( 2 ) & & 20% & & 4,5 + ngc 2244 & 6 ( 12 ) & & 1733% & & 6 + ncc 6231 & 16 ( 10 ) & 33 ( 16 ) & 63% & 53% & 2,7 + ngc 6611 & 9 ( 4 ) & & 4467% & & 2,3,8 + ic 2944 & 16 ( 7 ) & & 44% & & 2,3 + trumpler 16 & 20 ( 7 ) & & 35% & & 2,3 + cr 228 & 21 ( 5 ) & & 24% & & 2,3 + cyg ob2 & 44 ( 12 ) & 69 ( 8) & 27% & 18% & 9,10,11 + | this work provides a statistical analysis of the massive star binary characteristics in the cygnus ob2 association using radial velocity information of 114 b3o3 primary stars and orbital properties for the 24 known binaries .
we compare these data to a series of monte carlo simulations to infer the intrinsic binary fraction and distributions of mass ratios , periods , and eccentricities .
we model the distribution of mass ratio , log - period , and eccentricity as power - laws and find best fitting indices of @xmath0 , @xmath1 , and @xmath2 respectively .
these distributions indicate a preference for massive companions , short periods , and low eccentricities .
our analysis indicates that the binary fraction of the cluster is @xmath3% if all binary systems are ( artificially ) assumed to have p@xmath41000 days ; if the power - law period distribution is extrapolated to @xmath5 years , a plausible upper limit for bound systems , the binary fraction is @xmath690@xmath7% . of these binary ( or higher order ) systems
, @xmath645% will have companions close enough to interact during pre- or post - main - sequence evolution ( semi - major axis @xmath84.7 au ) .
the period distribution for @xmath9 d is not well reproduced by any single power - law owing to an excess of systems with periods around 35 days ( 0.080.31 au ) and a relative shortage of systems with periods around 714 days ( 0.140.62 au ) .
we explore the idea that these longer - period systems evolved to produce the observed excess of short - period systems .
the best fitting binary parameters imply that secondaries generate , on average , @xmath616% of the v - band light in young massive populations .
this means that photometrically based distance measurements for young massive clusters & associations will be systematically low by @xmath68% ( 0.16 mag in the distance modulus ) if the luminous contributions of unresolved secondaries are not taken into account . |
the influence of regions on their immediate surrounding , especially in inducing star formation activities , has drawn a lot of attentions in recent times . @xcite
first proposed that high - mass stars are formed through a `` collect - and - collapse '' sequential process triggered by the influence of an expanding region .
later , @xcite explained why gravitationally unstable fragments are formed in the shocked layers which fragment to produce high - mass stars .
since then many observational studies have targeted triggered star formation ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
although examples of quiescent formation of low mass stars have been reported in regions such as taurus , those for high mass stars are rare , kr140 @xcite being one example . while the influence of expanding regions on their surroundings can not be denied , @xcite argued an isolated case of the star associated with s252a forming spontaneously though located near the shock fronts of the monkey head nebula ( gem ob1 ) .
identifying the formation scenario requires studying the cloud properties and star formation in regions under the influence of regions .
w3main , w3(oh ) and afgl333 are active massive star forming regions located in the high - density layer ( hdl ) on the western edge of the w4 expanding region , or bubble @xcite .
w4 is excited by a group of o stars located at the central part of the region @xcite .
@xcite suggested a three generation system of hierarchical triggered star formation , based on the stellar populations of ic1795 .
the first generation is the 1300 pc loop and the perseus superbubble .
these triggered the formation of ic1795 , and then ic1795 triggered the formation of w3north , w3main , w3oh at the edge of shell .
an x - ray study by @xcite revealed the diversity of the w3 stellar populations .
@xcite analyzed the protoplanetary disks of young stars in the ic1795 region as massive star forming region by a deep spitzer survey .
the wider area including afgl333 and kr140 has been investigated with spitzer and herschel @xcite .
observational results of a sub - mm survey of dense clumps by @xcite , and extensive molecular line observations by @xcite agree well with the collect - and - collapse model .
@xcite also produced the extensive co maps of the w3 region .
star formation in w3main and w3(oh ) are considered to have been triggered by ic1795 . on
the other hand kr140 , which is located in the southwest quadrant of w3 giant molecular cloud ( gmc ) , appears to be formed in quiescently .
recently much of works have been done to reveal the star formation history of the w3 complex .
the stellar contents in w3main @xcite , ic1795 @xcite , and across the complex @xcite were investigated .
the statistical study was done by using herschel datasets @xcite .
these studies showed a complex history of star formation in the w3 gmc , and suggested massive stars and clusters in the active region as w3main and w3(oh ) were formed in the dense material by the progression of star formation process .
although triggering events play an important role to form the active region , it is still unclear in more quiescent region as afgl333 or kr140 .
is a dense molecular cloud located south of w3(oh ) .
it is associated with three known iras sources , namely , iras02252@xmath16120
( sfo05 : * ? ? ? * ) , iras02244@xmath16117 , and iras02245@xmath16115 ( afgl333 ) ( see figure [ fig:01 ] ) . although afgl333 is located in the hdl , its star formation is less active than the other two regions ( w3main and w3(oh ) ) and it has less extreme environmental conditions ( e.g. * ? ? ?
the deep nir photometry by @xcite unraveled the star formation activity of afgl333 region .
the afgl333 region is one of the best sites to study the influence of regions to star formation because of its location ; the dense molecular cloud at the edge of the extended bubble ( heart nebula ) of w4 and the presence of a compact region in iras02245@xmath16115 .
to study the interaction between regions and molecular clouds , investigation of the distribution of dense gas and of the gas temperature is vital .
3 lines prove to be particularly useful for investigating the physical conditions of dense gas clumps of a molecular cloud .
the 3 observations have advantages compared to other lines , such as @xmath3co and c@xmath4o , because 3 is an intermediate - density gas tracer , and gives gas temperature using the @xmath5 line ratio .
the information on gas motion and temperature distribution can indicate signs of interaction .
it also reveals the presence of interstellar shocks through the @xmath6 line ratio , shown by @xcite .
as example of such a study is given in @xcite who surveyed massive young stellar objects and ultra - compact hii regions in ammonia allowing them to derive the basic physical properties of the associated high density gas for their sample .
as such , we made mapping observations of the afgl333 region using 3 ( 1,1 ) , ( 2,2 ) , and ( 3,3 ) lines and estimated the distribution of the gas kinetic temperature over the cloud .
we also simultaneously observed 2o maser emission which is a well known tracer of early star formation activity ( e.g. * ? ? ? * ) . in this paper
, we observationally investigate the dense clumps of the molecular cloud in the afgl333 region that are influenced by the expanding region , and compared our results with archival data / catalogs to reveal the local star formation scenario in this region .
the parallax distance of w3(oh ) was obtained to be 2.0 kpc @xcite . since the line of sight velocities of w4 and w3(oh ) are within of 5 of each other
, we will assume a common distance of 2.0 kpc for the afgl333 region in this paper .
m , green : 5.8 @xmath7 m , blue : 4.5 @xmath7 m ) spitzer - irac image of the afgl333 region .
yellow enclosure shows the mapping area of our 3 observations.,width=566 ] we conducted 3 and 2o maser observations of the afgl333 region using the 45 m telescope at nobeyama radio observatory ( nro ) between january and june 2013 .
3 ( j , k ) = ( 1,1 ) , ( 2,2 ) , and ( 3,3 ) lines , whose frequencies are 23.694495 ghz , 23.722633 ghz , and 23.870129 ghz , respectively , and the 22.235080 ghz 2o maser line were observed simultaneously .
we used the h22 receiver , a cooled hemt receiver , and the sam45 , a digital spectrometer , with eight ifs to observe both polarizations for each line simultaneously . the bandwidth and spectral resolution of each
if were 62 mhz and 30.5 khz , respectively corresponding to @xmath8 and @xmath9 for the observed lines .
the telescope beam size at 23 ghz is 75 , which corresponds to 0.73 pc at 2.0 kpc .
we observed 267 positions with a 37.5 grid along the galactic coordinates using position switching between the target and the reference position comprising of the empty sky .
we investigated the kinematic and density structure of the afgl333 region at clump scales , or 0.5 - 1.0 pc .
the reference position was taken at @xmath10 , where no 3 or 2o maser was detected .
the target mapping area of the afgl333 region was @xmath11 centered at @xmath12 .
figure [ fig:01 ] shows the mapping area of the @xmath13-@xmath14 observations overlaid on the spitzer - irac image .
the pointing accuracy was checked using an 2o maser source cepheus a and was better than 5 .
data reduction was carried out using the newstar software package developed by the nobeyama radio observatory ( nro ) .
emission free channels were used to estimate and subtract a spectral baseline .
we applied only the linear line as the continuum level .
vertical and horizontal polarizations were combined and weighted based on the system noise temperature which was between 100 and 300 k. the antenna temperature in kelvin scale was measured using the chopper wheel method @xcite .
the median @xmath15 rms noise level was 0.04 k. for the 2o maser data .
we converted the antenna temperature to the flux density using the conversion factor of 2.6 jy k@xmath16 . in this paper , the intensities of 3 and 2o maser are presented as the antenna temperature in kelvin and flux density in jansky , respectively .
we used the data cubes of the j=2 - 1 lines of @xmath17co and @xmath3co obtained by the heinrich hertz submillimeter telescope @xcite .
the effective resolution of the maps is 38 fwhm and the sampling velocity of 0.5 .
1420 mhz radio continuum data from the canadian galactic plane survey ( cgps : * ? ? ?
* ) was also used to analyse the ionized gas distribution .
the image was obtained with a synthesized beam and an rms noise of 1.0 @xmath0 0.81 arcmin@xmath18 and 0.063 k. to establish the distributions of young stars and dust , we obtained spitzer space telescope 3.6 , 4.5 , 5.8 and 8.0 @xmath7 m irac image of the afgl333 region from the spitzer heritage archive hosted by the infrared science archive at ipac ( post - basic calibrated data : program i d 30995 ) , which shows the bright iras sources ( clusters ) and a prominent infrared dark filament representing the dense molecular cloud .
we also used the spitzer yso catalog in this region ( catalog 1 by @xcite ) , and the photometric data of all point sources within afgl333 by @xcite .
, shown in figure 3 . , width=302 ] 3 @xmath19 ( 1,1 ) , ( 2,2 ) , and ( 3,3 ) emission for which the integrated intensities exceed the 2@xmath20 noise level ( @xmath21 k ) , were obtained at 165 , 111 , and 71 positions out of the 267 observed positions , respectively .
figure [ fig:02 ] shows the 3 ( 1,1 ) , ( 2,2 ) , and ( 3,3 ) profiles towards the ( 1,1 ) emission peak at @xmath22 .
the 3 line profile comprises five quadruple hyperfine components consisting of a main line and two symmetrical pairs of inner and outer satellite lines .
we could detect the inner and outer satellite lines of the ( 1,1 ) line in 51 positions .
these 51 positions show ( 2,2 ) main lines with @xmath23 , and 36 points show ( 3,3 ) main lines among them .
detections of both main and satellite lines in ( 1,1 ) lines yield an opacity estimation using their intensity ratio ( see subsection 3.3 )
. the satellite lines of ( 2,2 ) and ( 3,3 ) line were not detected at any positions in the observed area . ) and 0.22 k respectively . the integrated velocity range is from @xmath24 to @xmath25 . pluses and
cross denote the positions of two 2o masers , and the ( 1,1 ) emission peak , respectively.,width=453 ] figure [ fig:03 ] shows the @xmath13-@xmath14 map of ( 1,1 ) line , velocity - integrated in the range from @xmath26 to @xmath25 .
no significant ( 1,1 ) emission was detected beyond this velocity range .
( 1,1 ) emission is distributed with an area of @xmath27 @xmath28 @xmath29 ( 5.2 @xmath0 8.7 pc at the distance of 2.0 kpc ) .
the overall morphology shows bow shape with the size of @xmath30 ( @xmath31 pc ) at the level of half of peak intensity .
there is weak and compact emission at north - east of the ridge near @xmath32 corresponding to the position of sfo05 @xcite at the northern tip of the infrared dark filament .
sfo05 is a small bright - rimmed cloud with an optical size of @xmath33 1 .
this bright - rimmed cloud was observed with the green bank telescope in 3 ( 1,1 ) and ( 2,2 ) lines @xcite .
the bow - shaped distribution of the ( 1,1 ) line is similar to those of the @xmath3co and c@xmath4o @xcite , submillimter observations @xcite , herschel far infrared map @xcite , and the filamentary structure of the dark infrared cloud @xcite .
6115 and iras02244@xmath16117.,width=566 ] figure [ fig:04 ] shows the peak velocity and velocity width maps of ( 1,1 ) line .
the peak velocity and the velocity width were obtained from the gaussian fitting to the main line of the ( 1,1 ) spectrum .
we observe a north - south velocity gradient along the ridge from the peak velocity map .
the most blueshifted component of @xmath34 is detected near the position of sfo05 .
this velocity is consistent with that of @xcite .
sfo05 should be on the near side of w4 .
@xcite reported that the afgl333 region shows the most negative velocity emission in the high density layer ( hdl ) .
the velocity structure in the afgl333 region is also consistent with the c@xmath4o results by @xcite , blueshifted at the northside of the nh@xmath35 peak , and redshifted at the southern side .
this may indicates that the gas is kinematically distributed by the large scale motion ( @xmath36 10 pc ) of hdl .
afgl333 is located on the near side of the w4 bubble , and w3main and w3(oh ) are at the edge or far side of the bubble , as suggested by @xcite .
the inverse trend of the velocity structure near the center may reflect the small scale structure within the clumps . ) and 0.03 k , respectively .
the position offset , increasing toward the northwest , is relative to the position of iras02245 + 6115 along the line shown in figure 4.,width=302 ] we found two regions of large velocity width at the eastern side of the ridge , one north and the other south of the dent .
the ( 1,1 ) emission velocity widths of these two regions are 2.8 @xmath37 0.4 , compared to 1.8 @xmath37 0.1 at the other positions .
@xcite reported yso surface density peaks at the center of the cluster associated with iras02245@xmath16115 .
figure [ fig:05 ] shows the position velocity diagram of the ( 1,1 ) emission line along the line in figure [ fig:04 ] .
co j=2 - 1 integrated intensity maps in two velocity intervals , ( a ) @xmath38 @xmath39 , and ( b ) @xmath40 @xmath41 , from the dataset of @xcite .
two iras sources , large 3 line width regions , and an 3 peak position are shown in stars , boxes , and x , respectively .
the lowest contour and contour interval are 2.5 k ( 5@xmath20 ) and 2.5 k , respectively.,width=453 ] we made the @xmath3co j=2 - 1 integrated intensity maps in two velocity intervals , which is shown in figure [ fig:06 ] , by using the data cubes provided by @xcite .
the blue shifted components ( @xmath42 ) mainly consist of the curved ridge at the east of iras02244@xmath16117 , and two clumps at the west and the south of iras02245@xmath16115 .
the northern 3 large velocity width region corresponds to the part of this curved co ridge .
the red shifted components ( @xmath43 ) mainly consist of the circular ridge with the size of @xmath44 around iras02245@xmath16115 , and the southern extension .
the southern 3 large velocity width region corresponds to the part of this circular co ridge . , we confirmed on the 3 maps , are not so conspicuous on the @xmath3co maps .
this may show the difference of the critical density of the molecular species . 3 emission sampled denser main part of the .
we discuss the origin of these 3 large velocity width regions in subsection 4.2 . from the 3 observational results , we derived the physical parameters : optical depth , column density , rotational temperature , local thermodynamical equilibrium mass , and virial mass of
. level in both the main and the satellite lines are plotted .
the error bar shows the @xmath15 noise level .
the estimated intensity ratio is shown in the top left corner of each panel.,width=302 ] the optical depth of 3 can be derived from the intensity ratio between the main and satellite line using the method shown in @xcite . the theoretical intensity ratio of satellite to main lines is @xmath45 in the case of the inner satellite line and outer satellite line , @xmath46 is 0.278 and 0.222 , respectively .
figure [ fig:07 ] shows the correlation between the 3 ( 1,1 ) and the inner / outer satellite lines .
the correlation shows a fairly uniform intensity ratio for all observed positions .
the ratios of the inner and outer satellite lines to main line were obtained to be @xmath47 and @xmath48 , respectively .
these correspond to optical depths of @xmath49 and @xmath50 which are consistent within the error values . from the average of these two values ,
the optical depth of 3 ( 1,1 ) in was derived to be @xmath51 .
the rotational temperature is determined from the the intensity ratio of ( 2,2 ) to ( 1,1 ) lines , assuming that the ( 1,1 ) and ( 2,2 ) lines are emitted from the same gas . if the dynamical state and the excitation conditions are similar , the line shapes of ( 1,1 ) and ( 2,2 ) emission should be similar too . to check the similality
, we made gaussian fitting to the ( 1,1 ) and ( 2,2 ) lines .
the obtained central velocity and the velocity width are @xmath52 and @xmath53 for the ( 1,1 ) line , @xmath54 and @xmath55 for the ( 2,2 ) line .
they are consistent each other . in the case that the optical depth is uniform , the rotational temperature , @xmath56 ,
is directly derived from the the intensity ratio of the ( 2,2 ) to ( 1,1 ) line .
following @xcite , we derived the rotation temperature as : @xmath57 \ } \right ] \right ) \right . \label{eq:2}\end{aligned}\ ] ] the average intensity ratio is obtained to be @xmath58 from the correlation plot of the integrated intensity in the ( 1,1 ) main and ( 2,2 ) main lines .
however the intensity ratio seems to change in the range from 0.2 to 1.0 .
this indicates the rotational temperature is different with the observed positions .
figure [ fig:08 ] shows a histogram of the rotational temperature for the ( 1,1 ) integrated intensity with @xmath59 detection in both the ( 1,1 ) and ( 2,2 ) lines .
the ( 3,3 ) line was used to discuss the interstellar shock through the abundance of the ortho - to - para 3 . however , the average intensity ratio of ( 1,1 ) line to ( 3,3 ) line in this region is not high ( @xmath60 ) as a typical molecular clouds in the galactic disk @xcite , implying that the special condition as the interstellar shock is not necessarily required .
the value of rotational temperature ranges between @xmath61 k. the mean value and standard deviation of the rotational temperature were derived to be 17 k and 3 k , respectively .
scan positions that derived temperatures in the range of @xmath62 k and @xmath63 k account for 94% and 6% of the total integrated intensity , respectively .
@xmath62 k is consistent with the dust temperature estimated from the herschel data of @xcite .
figure [ fig:09 ] shows the rotational temperature map .
the gas with a rotational temperature higher than the dust temperature ( @xmath63 k ) seem to be located at north , east , and south sides of the ridge .
the temperature is enhanced nearly 10 k near the ionization front facing w4 .
such an enhancement is also seen in the excitation temperature of co @xcite .
the column density of 3 can be estimated from the column density of ( 1,1 ) line and the rotational temperature using the assumption of local thermal equilibrium ( lte ) @xcite .
@xmath64 } \right ) \left ( \frac{\delta v_{1/2}}{[{\rm km\ s}^{-1 } ] } \right ) \rm{\ \ [ cm}^{-2 } ] \label{eq:3 } \end{aligned}\ ] ] the column density of the ( 1,1 ) line can be estimated from the product of the optical depth of the ( 1,1 ) line , the rotational temperature , and the velocity width ( fwhm ) of the ( 1,1 ) line .
we used a uniform optical depth of @xmath51 for the whole observed area .
we used the rotational temperature and the velocity width obtained at each observed positions shown in figure [ fig:09 ] and [ fig:04 ] . again , using the lte assumption , the total column density of 3 which is sum of the column density of each rotational levels , was estimated using the column density of ( 1,1 ) line and the rotational temperature at each observed position .
the column density map is shown in figure [ fig:09 ] .
the mean column density was obtained to be @xmath65 @xmath66 .
the error is the standard deviation of whole observed area .
column density contour @xcite overlaid on the 3 column density map.,width=453 ] the derived column density is compared with other observational results .
@xcite made ammonia observations by a petal - shaped scan centered on sub - mm sources in w3 gmc .
their six ammonia clumps , w48-w53 , are located within our observed region excluding w54 , which corresponds to sfo05 .
the column densities of their sources , w48 , w49 , and w50 , which are located near the is @xmath67 @xmath66 .
however , for the sources in the eastern side of ( w51 , w52 , and w53 ) , the column density is @xmath68 @xmath66 .
although their values are nearly a third of ours ( @xmath69 @xmath66 ) , the column density is higher at the eastern side of the .
@xcite showed the h@xmath70 column density map of afgl333 region by the herschel hobys data .
the central column density is @xmath71 @xmath66 .
figure [ fig:10 ] shows their contours of the h@xmath70 column density overlaid on our 3 column density map . as the increase of the 3 column density is mainly around the positions of large velocity widths , one possible cause is the 3 velocity width .
another cause might be due to the actual 3 abundance or excitation conditions within the clumps @xcite .
if we adopt n(h@xmath70 ) derived from dust continuum @xcite , the abundance ratio of 3 relative to h@xmath70 , @xmath72 , is @xmath73 at the eastern edge and @xmath74 at the central ridge .
the lte mass of the cloud , @xmath75 , was estimated from the sum of the lte mass obtained at each observed grid , @xmath76 , using @xmath77 , where @xmath78 is the grid area of @xmath79 pc@xmath18 , @xmath80 is the column density of 3 at each observed grid shown in figure [ fig:09 ] , @xmath81 is the mass of the hydrogen molecule .
the fractional abundance of 3 has typical values of @xmath82 to @xmath83 in dense gas clumps ( e.g. * ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . by assuming @xmath72 of 3@xmath84 as in the case of other massive star forming regions ( e.g. *
* ; * ? ? ?
* ) and infrared dark clouds @xcite , the lte mass of the cloud is @xmath85 .
the other cloud mass estimation is the virial mass , @xmath86 .
this is derived from @xmath87 , where r is the radius of the cloud and @xmath88 is the velocity width @xcite .
the virial mass within the radius of 1.6 pc ( @xmath89 ) , which corresponds to the same area of the lte mass estimation , was estimated to be @xmath90 . in this estimation ,
the mean velocity width of whole observed area of @xmath91 was used .
our results suggest that the is not close to a virialized state .
@xcite identified two cores , a and b , in the from their c@xmath4o observations .
they derived the lte masses and the viral masses of two cores , 2500 and 900 @xmath92 , and 1400 and 500 @xmath92 , for core a and b , respectively .
.,width=302 ] in star forming regions , 2o masers serve as signposts of sites of active star formation .
they may trace shocks , outflows , and other episodic events during the early formation stages of a protostar @xcite . observing 2o maser emission simultaneously to 3lead us to detect 2o masers at two different location in the region .
for identification purposes , we will refer to these masers as the northern - maser and the southern - maser .
the northern - maser is associated with sfo05 and is a known maser which was reported by the maser survey of bright rimmed molecular clouds @xcite .
however , the position around the southern - maser was not searched before and not listed in any maser catalogs ( e.g. * ? ? ?
* ; * ? ? ?
the southern - maser is located near the center of .
figure [ fig:11 ] shows the two 2o maser spectra .
the positions , radial velocities , and peak fluxes with uncertainty of the masers are shown in table [ tab:01 ] .
the error in their determined positions is 30 .
[ tab:01 ] co j=2 - 1 red and blue outflow lobes .
velocities between @xmath93 and @xmath94 are shown as blue contours and velocities between @xmath95 and @xmath96 are shown in red contours .
the blue component at the southwestern part is probably the contamination unrelated to the outflow associated with the young stellar activity .
the lowest contour and contour interval are @xmath97 k ( 2@xmath20 ) and @xmath97 k .
a plus marks the location of the southern - maser .
the dashed box shows the region where the physical parameters of the outflow are calculated . , width=453 ] [ tab:02 ] we found a high velocity @xmath17co wing components , which suggest the activity of young stellar object .
figure [ fig:12 ] shows the distribution of the emission wings integrated over the velocity range @xmath98 @xmath33 @xmath99 ( red component ) , and @xmath100 @xmath33 @xmath101 ( blue component ) around the southern - maser .
the blue components at the southwestern part of the map is probably another velocity ( @xmath102 ) component .
our result suggests the bipolar molecular outflow is associated with the southern - maser . assuming @xmath17co j=2 - 1 line is optically thin , and co abundance relative to h@xmath70 is 10@xmath103 , the lower limit of the outflow masses in the blue and red lobes .
we restricted the area of the outflow within the box area of 150 in latitude and 120 in longitude , shown in the dashed box in figure [ fig:12 ] .
as the temperature of the warm gas in the outflow is uncertain , we choose t@xmath104 k to minimize the column density as @xcite .
the calculated total mass , momentum and energy of the outflow in each lobe are shown in table [ tab:02 ] . there are four class i and one class ii sources @xcite within the positional accuracy of the southern - maser ( table [ tab:03 ] ) .
all five sources were detected by the spitzer irac bands , but have no nir data , and two sources were detected by mips 24 @xmath105 band .
it suggests that the outflow is powered by the source(s ) among them .
there have been extensive studies on w3 gmc , but mainly on the most active star forming regions such as w3main and w3(oh ) .
for example , the deep nir imaging of the w3main , @xcite revealed the disk fraction of the young stellar content .
the age of ic1795 ob association was estimated to be 3 - 5 myr by @xcite and @xcite .
@xcite identified five principal clusters and discussed the history of star formation in w3main and w3(oh ) . across the w3 region ,
extensive survey of the high - mass stellar population by @xcite showed star formation in w3 began spontaneously up to 810 myr ago . @xcite made dust column density maps by using herschel datasets in the whole w3 complex , and created the probability density functions ( pdfs ) .
they analyzed pdf to show the stellar feedback - based constructive process , i.e. stellar feedback is a major player in the cluster formation and the overall characteristics and local evolution of gmc .
they proposed a
convergent constructive feedback " model .
they suggested the combined collection effect in compressing and confining material by the high mass stars create higher column densities by feedback , which lead to the creation of an environment suitable for high mass cluster formation .
around iras02245@xmath16115 , which is shown with white dashed line in ( a ) , are overlaid on the synthesis radio continuum map @xcite .
ysos are shown by the symbols in different size according to their mass estimated based on the color magnitude ( @xmath106 vs @xmath107 ) diagram by @xcite .
sources with @xmath108 , @xmath109 , and @xmath110 are shown in small , medium , and large circles , respectively .
, width=642 ] recently , @xcite made deep jhks photometry of the afgl333 region , complementing with spitzer irac and mips observations .
their new nir photometry is @xmath111 mag deeper in each band compared to the 2mass photometry , and they identified many ysos including low mass populations in this region . to explore the distribution of ysos in our observational field
, we used their candidate sources of class i and ii .
figure [ fig:13]a shows the plot of distribution of the ysos . as their field is not fully covered our mapping area of 3 observations , we supplemented by the spitzer yso catalog ( catalog 1 ) of @xcite . that is why the surface density of class ii sources in the southern region appears to be low .
a rich clustering of class 0/i objects is seen along and in the core region , where the southern - maser source and 3 peak lie .
although in the spitzer catalog 1 clusterings are not found around the positions of two iras sources because of confusion by the bright infrared nebulosities , the deep nir observations by @xcite clearly revealed some peaks of the yso surface density associated with and its surroundings .
the peak around iras02245@xmath16115 was already discussed by @xcite .
@xcite also listed it with the size of 1.7 @xmath0 1.5 in the catalog of infrared clusters and stellar groups .
iras02245 + 6115 is associated with compact region g134.2 + 0.8 @xcite which has an electron density of 450 @xmath112 and size of 0.3 pc .
the peak of the 5 ghz continuum emission agrees well with iras02245@xmath16115 within a positional uncertainty of @xmath113 in position angle @xmath114 ( see figure [ fig:13]b ) .
one of the class i source at @xmath115 , which also corresponds to this peak , have @xmath116 , @xmath117 , and @xmath118 @xcite .
the extinction map made from the co column density @xcite indicates @xmath119 25 mag around g134.2 + 0.8 . on the nir color - magnitude diagram ( figure 8 by @xcite ) , its location is consistent with deeply embedded star with the mass of more than @xmath120 using the 2 myr isochrone by @xcite .
thus , this source should be the exciting star of the compact region as suggested by @xcite .
most ysos are distributed in the south - western of this peak of radio continuum emission .
another iras source , iras02244@xmath16117 , is a bright infrared source birs-104 @xcite and a small arc of infrared emission around it is reported by @xcite .
the peak of yso surface distribution at the south of iras02244@xmath16117 is also evident . as the northern part of
is located between these two iras sources , @xcite suggested the star formation activity of has been induced by these two sources .
the substructure of could be affected by the feedback from the young stellar activities .
sfo05 at the northern end of is the typical case of star formation triggered by radiation - driven implosion .
a pre - existing condensation of about @xmath121 was compressed by the pressure of the ionizing radiation of w4 @xcite . and
the compact region g134.2 + 0.8 lies at the interface between the and w4 .
this implies that the b0.5 star exciting the region g134.2 + 0.8 @xcite could have been formed due to the influence of the expanding bubble of w4 .
the large - scale temperature enhancement in the eastern side of the suggests an interaction between the dense ridge and the w4 bubble .
the interaction with the compact region may be limited to only the periphery or the north - south direction where large velocity widths were observed .
the low ionizing flux can erode the cloud , but would not trigger the formation of stars @xcite .
previous studies ( e.g. * ? ? ?
* ; * ? ? ?
* ) support a triggered or induced mode of star formation to be responsible for the formation of the stars at the interface between the w4 and the dense cloud , .
we suggest the overall structure of has been made by the external feedback from w4 .
although the local stellar feedback was also acting to make small - scale structure of the cloud , the convergent constructive process @xcite did not work effectively in this region .
quiescent mode of star formation can be described as a mode of star formation with minimal external influence .
based on the evidence seen in small velocity widths in the 3 lines around this region , the younger generation of ysos distributed around the center of the peak of the 3 ( 1,1 ) emission at the core of may have formed spontaneously .
@xcite gave a warning in distinguishing triggered population from spontaneously formed ones , and interpreting the observational data of star forming region in terms of triggering .
although our observations show g134.2 + 0.8 and the w4 bubble are interacting with and has affected its structure , it is not evident that most stars in this region were formed by the induced mode . @xcite compared the star formation efficiency and rate with w3 main and the nearby low - mass star forming regions , and showed the star forming activity in afgl333 region is comparable to other low - mass regions . as the mass of the cloud estimated by them is consistent with ours , the star formation efficiency ( sfe @xmath122 ) ) is as low as 3% .
the stellar feedback has not globally enhanced the star forming activity in this region .
our finding of the small velocity width of 3 line in the and the existing literatures as an older population in the western side of the afgl333 region suggest apparent coexistence of induced and quiescent modes of star formation within several parsec scale .
3 clumps associated with active star forming regions exhibit large emission line velocity widths driven by inflow or outflow within the gas , other non - thermal turbulent motion , or numerous unresolved dense clumps ( e.g. * ? ? ?
* ; * ? ? ?
* ) . in the case of gem
ob1 cloud , the dense cloud in contact with an expanding region showed no obvious observable interaction between them @xcite . on the other hand , the observed large velocity widths found in the northern and southern surroundings of region g134.2 + 0.8 ( see figure [ fig:04 ] ) is evident , and these feature suggests g134.2 + 0.8 is interacting with the dense molecular gas of .
no 3 emission was detected in and to the east of g134.2 + 0.8 , but the 3 emission detected to the west of it ( around the center of ) showed no significant velocity width .
the surface of the cloud might be irradiated by uv photons from the g134.2 + 0.8 .
based on the lyman continuum photon number from b0.5 exciting star of region g134.2 + 0.8 at 1 pc from is 8 @xmath0 10@xmath123 @xmath66s@xmath16 . even at the distance of 1620 pc from seven exciting stars of w4 @xcite , the ionizing flux is 4 @xmath0 10@xmath124 @xmath66s@xmath16 at .
these photon flux is approximately one order of magnitude smaller than for the typical star forming clouds at the edge of bubbles described in the collect - and - collapse model ( e.g. * ? ? ?
the activity of young stars embedded in the cloud might be a source of turbulent motion , but there is not so many young objects in these regions .
one explanation for the north - south alignment of this interaction could be an injection of mechanical energy from yso jets or high - velocity outflows located near the center of g134.2 + 0.8 , interacting with the dense cloud around it .
although , such shock activities provide possible explanation for the observed large velocity widths , co observations by @xcite show no sign of a high velocity component expected in the case of such outflows .
the systematic outflow survey of iras sources by @xcite also reported negative results for the afgl333 region .
however , we should notice that co observations by @xcite were not so deep to detect faint high velocity wing , because their work was to study the physical and chemical states of the cloud . and
the central mapping position by @xcite was different from iras02245@xmath16115 .
co j=2 - 1 emission .
two iras sources and an 3 peak position are shown in stars and x , respectively.,width=453 ] figure [ fig:14 ] shows two large velocity width regions of 3 emission on the contours of 1420 mhz radio continuum @xcite and the integrated @xmath3co j=2 - 1 intensity map .
the radio continuum emission of g134.2 + 0.8 fits into the hole of the @xmath3co map . the cavity or shell - like structure of @xmath3co emission with the size of 2 ( 1.2 pc ) hosting the dense ionized gas is reminiscent of the molecular dissociation or dynamical effect of the region on the molecular cloud .
there is the submillimeter source id-298 @xcite at the eastern side of the compact hii region , which corresponds to the co clump .
the synthesized radio continuum map of g134.2 + 0.8 @xcite shows steep decline of intensity toward the eastern direction ( see figure [ fig:13]b ) .
this asymmetric structure of radio continuum map suggests that a b0.5 star has been formed near this co clump , and the compact hii region is expanding to the western direction .
co j=2 - 1 line .
the positions of iras02245@xmath16115 and iras02244@xmath16117 are shown in the dashed lines.,width=566 ] figure [ fig:15 ] shows the position - velocity diagrams of @xmath3co j=2 - 1 emission along the galactic longitude .
the positions of iras02245@xmath16115 and iras02244@xmath16117 are shown by the dashed lines in the panel of @xmath125 and @xmath126 , respectively . from @xmath125 to @xmath127 , we can trace the c - shaped velocity structure ( blue shifted to @xmath128 ) centered around @xmath129 .
such a structure also shows the evidence of the shell with the expanding motion of @xmath130 .
contrary to the north and south of g134.2 + 0.8 , the reason for the 3 velocity width of @xmath131 at the west , nearly as low as the main part of the cloud , might be the geometrical effect .
the most dense part of the 3 ridge is deeply embedded in along the line of sight .
g134.2 + 0.8 is mostly expanding into the west at the foreground of the 3 ridge , and could not interact in the western boundary of the compact region . the interaction is limited in the northern and the southern less dense part of the 3 ridge .
the large velocity width around @xmath132 in figure [ fig:15 ] corresponds to the northern large width 3 region .
although sed of the exciting star candidate ( see subsection 4.2 ) suggests that it is class i source @xcite , there is no other characteristic signatures of young activity .
furthermore , because of its relatively low density , g134.2 + 0.8 is not as young @xcite as the other compact or ultracompact regions , like g133.8 + 1.4 ( w3n ) and w3(oh ) @xcite , associated with infrared sources in the w3/w4 region . for the large line width regions of 3 emission ,
the explanation invoking interaction by the expanding motion of the ionized gas at the periphery of is more preferable at present .
the large line width could be due to turbulent motion supplied through the shock front of g134.2 + 0.8 in the low ambient density .
molecular line observations with higher resolution is needed to clarify the activity of the ysos in g134.2 + 0.8 .
through our 3 mapping observations of afgl333 in the w4 region using the nobeyama 45 m radio telescope and archival co and other data , we have explored the physical properties of this region and studied ysos in two iras sources .
our results and conclusions are summarized as follows : \1 .
we derived a size of @xmath31 pc , the rotational temperature of @xmath61 k , and lte mass of @xmath85 for the molecular cloud .
the viral mass was estimated to be @xmath90 . possibly indicating that the is not close to a virialized state .
we found two regions of large 3 line width at the eastern side of .
the large 3 line width at the north and the south of iras02245@xmath16115 suggests an interaction by the expanding motion of the ionized gas from the compact hii region g134.2 + 0.8 at the periphery of .
@xmath3co data obtained by hht and nro reveal a cavity structure with the size of 1.2 pc in the integrated intensity maps , with an expanding motion of a few in the position velocity maps .
however , the small velocity widths in 3 gas observed west of iras02245@xmath16115 , around the center of the dense molecular gas , suggest that interaction with the compact region is limited .
two 2o masers were detected .
the southern - maser at the core of was a new finding .
a co bipolar outflow , a signpost of active star formation , was found near this maser source and it appears to be associated with some class i / ii sources .
we examined the distribution of ysos in the afgl333 region , especially around iras02245@xmath16115 and iras02244@xmath16117 .
we confirmed that g134.2 + 0.8 associated with iras02245@xmath16115 is excited by a deeply embedded young intermediate - mass star .
although the w4 bubble and have previously interacted , and induced mode of star formation could be responsible for the formation of sfo05 and iras02245@xmath16115 , the small velocity width , low rotational temperature of 3 line , and low sfe suggests that most stars in this region were formed by the quiescent mode in the feedback - driven structure . depending on the local environment , both of the two modes of star formation , induced and quiescent , coexist in scales of several pc or less .
it appears that star formation in this region has taken place mainly without an external trigger , but accompanying stellar feedback environment .
we are grateful to the staff members of the nobeyama radio observatory ( nro ) for observation support .
we also thank john bieging for providing heinrich hertz telescope co data cubes and takeshi sakai for providing his nro maps of the co emission in fits format .
the research presented in this paper has used data from the spitzer space telescope , which is operated by the jet propulsion laboratory , california institute of technology , under contract with nasa , and the canadian galactic plane survey supported by the natural sciences and engineering research council .
ballantyne , d. r. , kerton , c. r. , & martin , p. g. 2000 , , 539 , 283 bica , e. , dutra , c. m. , & barbuy , b. 2003 , , 397 , 177 bieging , j. h. , & peters , w. l. 2011 , , 196,18 bik , a. , stolte , a. , gennaro , m. et al .
2014 , , 561 , a12 bisbas , t. g. , wnsch , r. , whitworth , a. p. et al .
2011 , , 736 , 142 bressan , a. , marigo , p. , girardi , l. et al .
2012 , , 427 , 127 carpenter , j. m. , heyer , m. h. , & snell , r. l. 2000 , , 130 , 381 chibueze , j. o. , imai , h. , tafoya , d. et al .
2012 , , 748 , 146 chibueze , j. o. , imura , k. , omodaka , t. et al .
2013 , , 762 , 17 chibueze , j. o. , omodaka , t. , handa , t. et al .
2014 , , 784 , 114 chira , r. -a . , beuther , h. , linz , h. et al .
2013 , , 552 , a40 dale , j. e. , haworth , t. j. , & bressert , e. 2015 , , 450 , 1199 deharveng , l. , lefloch , b. , zavagno , a. et al .
2003 , , 408 , l25 deharveng , l. , zavagno , a. , & caplan , j. 2005 , , 433 , 565 di francesco , j , johnstone , d. , kirk , h. et al .
2008 , , 175 , 277 dunham , m. k. , rosolowsky , e. , evans , n. j. ii et al . 2010a ,
, 717,1157 dunham , m. k. , evans , n. j. ii , bourke , t. l. et al . 2010b , , 721,995 elmegreen , d. m. 1980 , , 240 , 846 elmegreen , b. g. , & lada , c. j. 1977 , , 214 , 725 feigelson , e. d. , & townsley , l. k. 2008 , , 673 , 354 fukuda , n. , miao , j. , sugitani , k. , et al .
2013 , , 773 , 132 hachisuka , k. , brunthaler , a. , menten , k. m. et al .
2006 , , 645 , 337 hindson , l. , thompson , m. a. , urquhart , j. s. et al .
2010 , , 408 , 1438 ho , p. t. p. , & townes , c. h. 1983 , , 21 , 239 hughes , v. a. , & viner , m. r. 1982 , , 87 , 685 jose , j. , kim , s. k. , herczeg , g. j. et al . 2016 , , 822 , 49 kiminki , m. m. , kim , j. s. , bagley , m. b. et al .
2015 , , 813 , 42 kraemer , k. e. , shipman , r. f. , price , s. d. et al .
2003 , , 126,1423 kutner , m. l. , & ulich , b. l. 1981 , , 250 , 341 lada , c. j. , elmegreen , b. g. , cong , h .- i . , & thaddeus , p. 1978
, , 226 , l39 lefloch , b. , lazareff , b. , & castets , a. 1997 , , 324 , 249 lefloch , b. , & cernicharo , j. 2000 , , 545 , 340 mampaso , a. , gomez , p , , sanchez - magro , c. & selby , m. j. 1984 , , 207 , 465 mangum , j. g. , wootten , a. , & mundy , l. g. 1992 , , 388 , 467 moore , t. j. t. , bretherton , d. e. , fujiyoshi , t. et al .
2007 , , 379 , 663 morgan , l. k. , figura , c. c. , urquhart , j. s. & thompson , m. a. 2010 , , 408 , 157 morgan , l. k. , moore , t. j. t. , eden , d. j. et al .
2014 , , 440 , 1730 nagayama , t. , omodaka , t. , handa , t. et al .
2009 , , 61 , 1023 oey , m. s. , watson , a. m. , kern , k. , & walth , g. l. 2005 , , 129 , 393 polychroni , d. , moore , t. j. t. , & allsopp , j. 2012 , , 422 , 2992 rivera - ingraham , a. , martin , p. g. , polychroni , d. , & moore , t. j. t. 2011 , , 743 , 39 rivera - ingraham , a. , martin , p. g. , polychroni , d. et al .
2013 , , 766 , 85 rivera - ingraham , a. , martin , p. g. , polychroni , d. et al .
2015 , , 809 , 81 roccatagliata , v. , bouwman , j. , henning , t. , et al .
2011 , , 733 , 113 rohlfs , k. , & wilson , t. l. 1996 , tools of radio astronomy ( 2nd edition ; berlin : springer ) rom@xmath133n - z@xmath134iga , c. g. , ybarra , j. e. , meg@xmath135as , g. d. et al .
2015 , , 150 , 80 sakai , t. , oka , t. , & yamamoto , s. 2006 , , 649 , 268 sakai , t. , oka , t. , & yamamoto , s. 2007 , , 662 , 1043 snell , r. l. , dickman , r. l. , & huang , y .- l . 1990 , , 352 , 139 sugitani , k. , fukui , y. , & ogura , k. 1991 , , 77 , 59 sunada , k. , nakazato , t. , ikeda , n. et al .
2007 , , 59 , 1185 taylor , a. r. , gibson , s. j. , peracaula , m. et al .
2003 , , 125 , 3145 tieftrunk , a. r. , megeath , s. t. , wilson , t. l. , & rainer , j. t. 1998 , , 336 , 991 thronson , h. a. , harvey , p. m. , & gatley , i. 1979 , , 229 , l133 thronson , h. a. , lada , c. j. , & hewagama , t. 1985 , , 297 , 662 torrelles , j. m. , patel , n. a. , curiel , s. et al .
2011 , , 410 , 627 ungerechts , h. , walmsley , c. m. , & winnewisser , g. 1980 , , 88 , 259 urquhart , j. s. , morgan , l. k. , figura , c. c. et al .
2011 , , 418 , 1689 urquhart , j. s. , figura , c. c. , moore , t. j. t. et al .
2015 , , 452 , 4029 valdettaro , r. , palla , f. , brand , j. , & cesaroni , r. 2005 , , 443 , 535 valdettaro , r. , migenes , v. , trinidad , m. a. , brand , j. , & palla , f. 2008 , , 675 , 1352 whitworth , a. p. , bhattal , a. s. , chapman , s. j. et al .
1994 , , 268 , 291 xu , y. , reid , m. j. , zheng , x. w. , & menten , k. m. 2006 , science , 311 , 54 | we investigated the star formation activities in the afgl333 region , which is in the vicinity of the w4 expanding bubble , by conducting 3 ( 1,1 ) , ( 2,2 ) , and ( 3,3 ) mapping observations with the 45 m nobeyama radio telescope at an angular resolution of 75 .
the morphology of the 3 ( 1,1 ) map shows a bow - shape structure with the size of 2.0 @xmath0 0.6 pc as seen in the dust continuum . at the interface between the w4 bubble and the dense 3 cloud ,
the compact region g134.2 + 0.8 , associated with iras02245@xmath16115 , is located .
interestingly , just north and south of g134.2 + 0.8 we found 3 emission exhibiting large velocity widths of @xmath2 , compared to 1.8 at the other positions .
as the possibility of mechanical energy injection through the activity of yso(s ) is low , we considered the origin of the large turbulent gas motion as indication of interaction between the compact region and the periphery of the dense molecular cloud .
we also found expanding motion of the co emission associated with g134.2 + 0.8 .
the overall structure of the afgl333-ridge might have been formed by the expanding bubble of w4 .
however , the small velocity widths observed west of iras02245@xmath16115 , around the center of the dense molecular cloud , suggest that interaction with the compact region is limited .
therefore the ysos ( dominantly class 0/i ) in the core of the dense molecular cloud most likely formed in quiescent mode .
as has been previously suggested for the large scale star formation in the w3 giant molecular cloud , our results show an apparent coexistence of induced and quiescent star formation in this region .
it appears that star formation in the afgl333 region has proceeded without significant external triggers , but accompanying stellar feedback environment . |
a variety of phenomena manifest themselves as changes in apparent brightness of astronomical objects .
the amplitudes and time - scales of these changes vary from tens of magnitudes and weeks for supernova explosions to a fraction of a magnitude and minutes for stellar pulsations . with the notable exceptions of light
echoes ( e.g. * ? ? ?
* ) , variable reflecting nebulae ( e.g. * ? ? ?
* ) and the m87 jet ( e.g. * ? ? ?
* ) variable objects are unresolved by single - dish telescopes can not be larger than @xmath1 light seconds , otherwise its variability would be smeared . ] .
variable point - like objects are often embedded in light of a resolved non - variable source ( active nucleus or a supernova in a galaxy ; young stellar object embedded in a nebula ) that complicate measurements of the variable object s brightness .
the variations may be associated with a single catastrophic event ( supernova ) , may be approximately ( dwarf novae ) or strictly periodic ( eclipsing binaries ) or aperiodic ( active galactic nuclei ) in nature .
our understanding of these events depends on the efficient and reliable detection of brightness variations .
photometric measurements are prone to systematic errors that are difficult to characterize .
this makes it challenging to distinguish true low - amplitude variability from the apparent one caused by systematic effects and measurement errors .
imaging artefacts such as cosmetic defects of a ccd , diffraction spikes from bright objects and cosmic ray hits as well as blending between images of nearby objects can corrupt photometry and mimic high - amplitude variability .
three different lines of attack on the problem of variable object detection are described in the literature : direct image comparison , ( `` transient detection '' ) , lightcurve analysis using variability indices and periodicity search .
transient detection techniques seek to identify changes between two sets of sky images taken at different times ( epochs ) .
the changes may be found by subtracting the images pixel - by - pixel after resampling them to a common coordinate grid and accounting for seeing changes ( difference image analysis
dia ; @xcite , , @xcite , @xcite , @xcite ; applications of the method include @xcite ) .
large surveys such as ogle @xcite , ptf @xcite , pan - starrs @xcite and des @xcite implement the image subtraction technique .
alternatively , one may extract astronomical objects ( sources ) from each image independently and compare the resulting source lists ( @xcite , crts
the second - epoch images are often taken in pairs , triplets or even longer series with dithering to eliminate image artefacts that are usually associated with a given position on the image detector , not in the sky .
more sophisticated detection strategies may be applied if measurements are obtained at more than two epochs .
their obvious advantage over the simple two - epoch data comparison is the potential to average - out individual measurement errors and thus detect variability with a lower amplitude .
one class of methods employs various `` variability indices '' characterizing the overall scatter of measurements in a lightcurve and/or degree of correlation between consecutive flux measurements ( some recent examples : * ? ? ?
* ; * ? ? ?
* see the detailed discussion in section [ sec : indices ] ) .
the other class of methods search for significant periodicity in flux variations ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) .
while many types of variable stars show periodic or semi - periodic light variations , flux measurement errors are expected to be aperiodic , or associated with a known periodic process inherent to the observations ( diurnal or seasonal cycle , synodic month , periodic guiding errors , orbital period of a spaceborne telescope , etc . ) .
if a search is aimed at a specific variability type for which a lightcurve shape is generally known in advance ( e.g. exoplanet transits or eclipsing binaries in general , cepheids , rr lyrae stars , novae ) , template fitting ( e.g. @xcite , @xcite , , @xcite ) with various trial periods / flare development time - scales can be performed .
simple cuts on lightcurve parameters as well as advanced machine learning techniques @xcite can be used to select lightcurves of a known shape from a large photometric data set . a pre - selection based on colour
can be used to reduce the number of candidates when searching for variables of a specific type . since period search and template - fitting algorithms
are computationally expensive , a two - step approach can be applied .
candidate variable stars are pre - selected using a fast - to - compute variability index ( and/or colour ) and only the lightcurves that passed this selection are subjected to period search or template fitting ( e.g. * ? ? ?
* ; * ? ? ?
if the total number of observed objects is low , both period search and lightcurve scatter - based selection criteria are applied independently of each other to conduct exhaustive search for both periodic and non - periodic variables @xcite .
selection based on period search may be followed by even more computationally intensive steps like binary system modelling @xcite . used the period along with other variability features as an input for the random forest algorithm to select periodic variable star candidates in the eros-2 database and simultaneously classify them .
the methods described above may efficiently select variable object candidates from a large set of photometric data .
however , the final decision to designate an object as `` variable star '' rather than a `` candidate '' is usually made only after visual inspection of the object s lightcurve by a human expert ( e.g. @xcite , @xcite , @xcite , @xcite ) .
if the number of observations is small , the original images are checked for the presence of obvious problems [ image artefacts , cosmic ray hits , point spread function ( psf ) wings of a bright nearby object ] affecting photometry of a candidate variable ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
while advanced image artifact rejection procedures exist , visual image inspection remains an important data quality control tool as it may uncover unexpected problems .
variable star detection may be considered in the framework of classical hypothesis testing ( e.g. * ? ? ?
* ) : to establish that an object is variable , one needs to rule out the null hypothesis that it is constant given the observations @xcite .
one may compare a value of variability - detection statistic ( section [ sec : indices ] ) derived from the lightcurve to the distribution of this value expected for non - variable objects .
the problem is that objects with corrupted measurements produce long tails in the aforementioned distributions . in the presence of badly measured objects one
is forced to set a low threshold for accepting candidate variables ( section [ sec : cutoffvalue ] ) and rely on additional information not captured by the variability - detection statistic to distinguish true variables from badly measured objects in the distribution tail .
alternatively one may view the search for variable stars as a classification problem that may be approached with machine learning techniques .
the task is to classify a set of objects characterized by their lightcurves , images associated with each lightcurve point and possibly additional pieces of information associated with each brightness measurement ( object s position on the ccd frame , airmass , seeing , temperature , etc . ) .
one needs to distinguish various classes of variable stars from the class of well - measured constant stars and classes of stars affected by various types of measurement errors ( bad pixels , diffraction spikes , blending ) .
objects that do not belong to one of the known classes should also be identified .
while considerable progress has been made in lightcurve - based automated classification of stars already known to be variable , an automated system that could reliably identify variable stars among non - variables remains to be developed . in practice , the following approach to variable star detection is often adopted .
( i)objects affected by blending and image artefacts are flagged at source extraction stage . ( ii)the lightcurves of the detected objects are constructed and may be refined using the available additional information ( section [ sec : filtering ] ) .
( iii)the techniques described in the previous paragraphs are used to select promising variable star candidates based on their lightcurves .
( iv)the list of candidates is examined by a human expert who performs the final classification and removes false variables from the list . in this work
we explore the limits of the traditional approach outlined above and identify the best ways to select candidate variables .
we compare the performance of popular variability detection techniques on various real and simulated photometric data sets .
we refer to any value that quantifies ` how variable ' a given object is as a ` variability index ' .
the discussion is limited to variability indices based on lightcurve scatter ( sections [ sec : chi2][sec : peaktopeak ] ) and correlation ( sections [ sec : lag1autocorr][sec : sb ] ) while the period - search based techniques will be discussed elsewhere .
we attempt to find a general - purpose variability detection technique able to recover a broad range of variability types including currently unknown ones @xcite .
such a technique would also be useful for solving the opposite problem : reliable selection of non - variable objects that can be used as photometric standards ( e.g. * ? ? ? * ) or targets for searches of variations not intrinsic or not typical to the objects such as microlensing events @xcite , occultations of stars by distant solar system bodies @xcite , tidal disruption events in nuclei of non - active galaxies @xcite and failed supernovae @xcite .
publications focused on comparing performance of variability search techniques include who compared planetary - transit detection algorithms , while @xcite and @xcite discussed a number of variability detection tests in the context of active galactic nuclei . compared performance of some multi - band correlation - based variability indices . @xcite and @xcite discussed properties of ` excess variance ' ( section [ sec : sigmaxs ] ) and ` fractional variability amplitude ' , the variability measures often used in x - ray astronomy .
@xcite compared the accuracy and performance of period finding algorithms . @xcite
compare various methods of extracting a characteristic time - scale from an irregular lightcurve .
@xcite provide an extensive list of features useful for lightcurve characterization and classification .
preliminary results of our variability index comparison based solely on photographic lightcurves are presented by @xcite .
this paper is structured as follows .
section [ sec : indices ] defines the variability indices we investigate .
section [ sec : testdata ] describes the test data .
section [ sec : comparison_tech ] presents the technique for comparison of effectiveness of variability indices in selecting variable objects .
section [ sec : resdisc ] discusses the results of the comparison and section [ sec : conclusions ] summarizes our findings .
in this section we define the numerical parameters characterizing the ` degree of variability ' of an object the variability indices , discussed in detail in the following paragraphs .
the scatter - based indices ( sections [ sec : sigma][sec : peaktopeak ] ) consider only the distribution of measured magnitudes ignoring the time information available in a lightcurve .
some also take into account the estimated errors .
the correlation - based indices ( sections [ sec : lag1autocorr][sec : sb ] ) in addition to the measured magnitudes themselves consider the order in which the measurements were taken and some indices also take into account the time difference between measurements .
the use of this additional information makes correlation - based indices more sensitive to low - amplitude variability , but on the downside , correlation - based indices are insensitive to variability on time - scales shorter than the sampling time @xcite .
table [ tab : indexsummary ] summarizes the information used by each index . in the following sections we compare the effectiveness of these variability indices in selecting variable stars .
a @xmath2 test is any statistical hypothesis test in which the sampling distribution of the test statistic is a @xmath2 distribution when the null hypothesis is true
. given @xmath3 magnitude measurements @xmath4 ( assumed to be independent of each other ) and their associated errors @xmath5 ( assumed to be gaussian ) , the null hypothesis , @xmath6 , that an object does not change its brightness can be tested by computing the value @xmath7 where @xmath8 is the weighted mean magnitude .
@xmath2 is compared to the critical value @xmath9 obtained from the @xmath2-distribution with @xmath10 degrees of freedom .
the @xmath11-value indicates the statistical significance level at which @xmath6 can be rejected .
if measurement errors are estimated correctly , the majority of objects should have @xmath2 values consistent with @xmath6 , since the majority of stars are not variable .
a notable exception from this rule is millimagnitude - precision photometric observations such as the ones obtained by most @xcite , corot , kepler @xcite and future photometric space missions ( e.g. * ? ? ?
* ; * ? ? ?
* ) , which are able to detect variability in the majority of field stars , including variability caused by transiting solar system - like planets @xcite . in practice , poor knowledge of @xmath5 limits the applicability of the @xmath2 test for variability detection in ground - based photometry
. in this case
, @xmath2 may still be useful as a measure of scatter in a lightcurve compared to the expected measurement errors , but the cut - off value for discriminating variable objects from non - variable ones should be different from the one suggested by the @xmath2 distribution . in the following we use the reduced @xmath12 ( e.g. * ? ? ?
* ) to compare its value for lightcurves with different @xmath3 .
@xcite note that estimated photometric measurement errors are asymmetric and non - gaussian when converted from flux to magnitude space .
this violates the assumptions behind the critical value @xmath9 calculations .
the @xmath2 test , in its textbook form , should be performed in flux space and only when the contribution from all sources of photometric errors has been properly accounted for .
c c@ c@ c c@ c index & errors & order & time & sec . & ref .
+ + @xmath13 & @xmath14 & & & [ sec : chi2 ] & ( a ) + @xmath15 & & & & [ sec : sigma ] & ( b ) + @xmath16 & @xmath14 & & & [ sec : sigma ] & ( b ) + @xmath17 & & & & [ sec : mad ] & ( c ) + @xmath18 & & & & [ sec : iqr ] & ( d ) + @xmath19 & @xmath14 & & & [ sec : roms ] & ( e ) + @xmath20 & @xmath14 & & & [ sec : sigmaxs ] & ( f ) + @xmath21 & @xmath14 & & & [ sec : peaktopeak ] & ( g ) + + @xmath22 & & @xmath14 & & [ sec : lag1autocorr ] & ( h ) + @xmath23 & @xmath14 & @xmath14 & @xmath14 & [ sec : welch ] & ( i ) + @xmath24 & @xmath14 & @xmath14 & @xmath14 & [ sec : stetson ] & ( j ) + @xmath25 & @xmath14 & @xmath14 & @xmath14 & [ sec : stetsontime ] & ( k ) + @xmath26 & @xmath14 & @xmath14 & @xmath14 & [ sec : stetsonclip ] & ( d ) + @xmath27 & @xmath14 & @xmath14 & @xmath14 & [ sec : stetson ] & ( j ) + @xmath28 & & @xmath14 & & [ sec : cssd ] & ( l ) + @xmath29 & @xmath14 & @xmath14 & @xmath14 & [ sec : excursions ] & ( m ) + @xmath0 & & @xmath14 & & [ sec : vonneumann ] & ( n ) + @xmath30 & & @xmath14 & @xmath14 & [ sec : ea ] & ( o ) + @xmath31 & @xmath14 & @xmath14 & & [ sec : sb ] & ( p ) + a detectable variable star , by definition , should have larger scatter of measurements in its lightcurve compared to non - variable stars that could be measured with the same accuracy .
one way to characterize scatter of measurements is to compute the standard deviation , @xmath32 or alternatively , if the estimated errors are assumed to reflect the relative accuracy of measurements , its weighted version @xmath33 assuming that @xmath4 are drawn from gaussian distributions having variances @xmath34 and the same mean @xmath35 , the choice of weights @xmath36 maximizes the likelihood of obtaining the set of measurements ( @xmath4 ) .
therefore , given a set of measurements ( @xmath4 , @xmath5 ) , equation ( [ eq : weighted_average ] ) is the best estimate of the mean under the above assumptions . we define @xmath15 as a square root from an unbiased estimator of the population variance [ the bessel correction , i.e. @xmath37 instead of @xmath3 in the denominator of equation [ eq : sigma ] ] as this is the definition often adopted in statistical software , notably in the gnu scientific library . for the purpose of variable star search ,
the use of bessel s correction has minimal practical consequences .
standard deviation is relatively sensitive to outlier points . in many cases , lightcurve filtering ( section [ sec : filtering ] )
might be needed before @xmath15 can serve as an efficient variable star selection tool . in the following paragraphs
we describe ways of characterizing lightcurve scatter that are less sensitive to outliers .
the median absolute deviation , mad ( @xcite , @xcite ) , is a measure of scatter of observations @xmath4 defined as @xmath38 for a gaussian distribution @xmath39 where @xmath40 is the inverse of the cumulative distribution function for the gaussian distribution .
the mad statistic is mostly insensitive to outliers @xcite ; its only disadvantage is that it is equally insensitive to real variations that occur only occasionally , like rare eclipses of an algol - type binary that may show virtually constant brightness outside of the eclipses , the use of mad is computationally more expensive than @xmath15 as the sorting needed to compute the median is a relatively slow , @xmath41 , operation compared to calculating the average value , @xmath42 . here @xmath41 ( @xmath42 ) means that there is a constant @xmath43 such that for any number of input measurements , @xmath44 , the computation will be completed in less than @xmath45 ( @xmath46 ) steps . it should be noted that correlation - based indices discussed below in sections [ sec : lag1autocorr][sec : sb ] computationally depend on the order of data points and , therefore , require the input lightcurve to be sorted in time an operation of @xmath41 complexity .
another robust measure of scatter is the interquartile range , iqr , which includes the inner 50% of measurement values ( i.e. excludes 25% of the brightest and 25% of the faintest flux measurements ) . to compute the iqr we first compute the median value that divides the set of flux measurements into upper and lower halves .
the iqr is the difference between the median values computed for the upper and lower halves of the data set . for the normal distribution @xmath47 , where @xmath40 is the inverse of the cumulative distribution function for the gaussian distribution
the iqr may be more appropriate than mad ( section [ sec : mad ] ) for measuring the width of an asymmetric ( skewed ) distribution , such as the distribution of flux measurements of an eclipsing binary . the robust median statistic , roms , was proposed by @xcite and successfully applied for variable star search by @xcite and @xcite .
it is defined as @xmath48 for a non - variable object , the expected value of @xmath19 is around 1 as the majority of the measurements should be within @xmath49 of the median value ( if @xmath15 is estimated correctly ) .
normalized excess variance , @xmath20 , is used in x - ray and optical astronomy to characterize variability amplitude in the presence of changing measurement errors .
it is defined as @xmath50.\ ] ] here we use the symbol @xmath20 for the normalized excess variance as this or similar symbols are widely used in the literature . note that @xmath20 may be negative if the estimated errors @xmath5 are larger than the actual scatter of measurements , @xmath4 .
the fractional root mean square variability amplitude , @xmath51 , another commonly used x - ray variability measure , is simply a square root of the normalized excess variance : @xmath52 @xcite if @xmath20 is positive .
@xcite note that in the presence of red noise , the expected value of @xmath20 depends on the length of a time series .
the value of @xmath20 estimated from a lightcurve is related to the integral of the power spectral density ( psd ) in the frequency range probed by the observations , however this relation is complex @xcite and depends on the psd slope and sampling ( window function ) . the peak - to - peak variation , @xmath21 , can be characterized as @xmath53 where @xmath54 and @xmath55 are the maximum and minimum values of the expressions @xmath56 and @xmath57 over the entire lightcurve .
this variability index , with minor variations in its definition , is widely used in the radio astronomy community .
it is of interest to compare @xmath21 with variability characteristics traditionally used in optical and other bands .
here we use the definition of @xmath21 adopted by @xcite and .
the value of @xmath21 may be negative if the measurement errors , @xmath58 , are overestimated ( c.f .
@xmath20 , section [ sec : sigmaxs ] ) the peak - to - peak variation may be a sensitive variability indicator if we believe that a lightcurve is free from outliers ( thanks to high data quality or successful filtering ) .
while @xmath21 can be computed for a lightcurve consisting of as few as two observations , the expected value of @xmath21 for a non - variable source depends strongly on the number of measurements .
monte carlo simulation is a practical way to estimate expected values of @xmath21 for a non - variable object given a number of observations and their accuracy .
photometric observations are often planned so that the time span between consecutive flux measurements is smaller than the variability time - scale expected for the objects of interest . the simplest way to characterize
the similarity of consecutive flux measurements is to compute the first - order autocorrelation coefficient ( also known as ` serial correlation coefficient ' or ` lag-1 autocorrelation ' ) of a lightcurve ( e.g. * ? ? ?
* ; * ? ? ?
* ) : @xmath59 it has been shown that , assuming that @xmath60 are independent measurements subject to identically distributed measurement errors , @xmath22 follows an asymptotically normal distribution with the expected value of @xmath61 and the variance of @xmath62 , allowing one to assess if the obtained value of @xmath22 is consistent with the expected one under the above assumptions .
this simple method loses efficiency if a lightcurve is unevenly sampled since pairs of data points widely separated in time and weakly correlated or uncorrelated entirely contribute to the value of @xmath22 equally with the pairs of measurements taken close in time that may be well correlated .
@xcite propose a variability index , @xmath23 , characterizing the degree of correlation between @xmath44 quasi - simultaneous pairs of measurements obtained in two filters @xmath63 and @xmath21 : @xmath64 where @xmath65 ( @xmath66 ) are the measured magnitudes , @xmath67 ( @xmath68 ) are the estimated errors and @xmath69 ( @xmath70 ) is the mean magnitude in filter @xmath63 ( @xmath21 ) .
relying on the above assumption that a lightcurve contains pairs of measurements obtained close in time ( compared to the expected variability time - scale ) one can apply @xmath23 to a single - band lightcurve by dividing it into two subsamples that would mimic measurements in two filters .
one obvious way to accomplish this is to sort the lightcurve in time , number measurements ( @xmath71 ) and assign measurements having odd numbers to @xmath21 subsample and even numbers to @xmath63 subsample . in this case
, @xmath72 may be taken to be the mean of all @xmath73 observations , rather than the means of two different samples each of size @xmath44 .
if a single - filter lightcurve does not entirely consist of pairs of closely - spaced points , one would like to avoid forming pairs from measurements taken far apart in time ( c.f .
@xmath22 in section [ sec : lag1autocorr ] ) . in that case
, an additional parameter , @xmath74 , defines the maximum time difference between two observations that are considered to be taken sufficiently close in time for forming a pair .
the performance of the algorithm on a given unevenly sampled data set depends strongly on the choice of @xmath74 .
if @xmath74 is too small , only few lightcurve points will form a pair and contribute to @xmath23 rendering the index unusable .
an optimal value of @xmath74 would be large enough to form many measurement pairs in an unevenly sampled lightcurve but small enough to remain sensitive to a wide range of variability time - scales as @xmath23 is sensitive to variations on time - scales from @xmath74 to the overall duration of the lightcurve .
a histogram of the interval between observations may be useful in selecting an appropriate @xmath74 value for a given data set . in our tests we use @xmath75 for all the test data sets .
isolated data points that can not be paired with others ( for a given choice of @xmath74 ) are omitted from the @xmath23 computation .
[ fig : simlc ] and table [ tab : simlc ] show how an unevenly - sampled single - band lightcurve can be divided into subsamples to calculate @xmath23 or @xmath24 .
a point is assigned to subsample b , v or counted as ` isolated ' depending on the value of @xmath74 and the order in which one considers the lightcurve : from the first point to the last one ( direction indicated by the top arrow in fig .
[ fig : simlc ] , the corresponding samples are named as lower case b and v ) or the reverse direction ( bottom arrow , capital b and v ) . depending on the order , one may compute the ` forward ' and ` reverse ' values of an index that might differ from each other because the points are divided into pairs ( assigned to b and v subsamples ) in a different way ( as illustrated by fig .
[ fig : simlc ] ) . in our implementation of the index ,
the ` forward ' and ` reverse ' values are averaged to have a single value describing a lightcurve . in the case of @xmath23 ( but not @xmath24 ) ,
the ` forward ' and ` reverse ' values are equal if one allows a point to be counted in multiple pairs ( enter two subsamples simultaneously ) .
the @xmath23 and @xmath24 indices are designed to detect variability on time - scales much longer than the typical time difference between observations forming pairs . if , however , the variability time - scale is comparable to the sampling rate of observations , the measurements in pairs may appear anticorrelated ( correlation coefficient @xmath76 , section [ sec : lag1autocorr ] ) rather than uncorrelated ( @xmath77 ) , resulting in near - zero or negative value of @xmath23 ( @xmath24 ) and rendering the index insensitive to the variations .
the actual value of detectable variability time - scale is determined by the data and will be very different for data sets including observations taken minutes apart and data sets that include only observations taken on different nights .
a more robust variability index proposed by @xcite is : @xmath79 where @xmath80 is the sign function . here
the photometric data set is divided into @xmath44 groups each consisting of two or more quasi - simultaneous observations ( in one or more filters ) or a single , isolated measurement .
a single - filter lightcurve can be divided into subsamples to mimic multi - band data in the same way as for the @xmath23 index ( section [ sec : welch ] ) , with the difference that isolated points can be kept in the analysis .
each group consisting of one or more points is assigned a weight @xmath81 .
@xmath82 is defined as @xmath83 the definition of @xmath82 can be generalized for groups containing more than two measurements by multiplying @xmath82 ( for a pair ) by an additional factor of @xmath84 , where @xmath85 are the observations in the third filter or subsample . for simplicity , in the implementation of the stetson indices used throughout this paper , we do not consider groups containing more than two points and do not allow a point to be counted as part of more than one group ( see fig .
[ fig : simlc ] and table [ tab : simlc ] ) . instead of using the weighted arithmetic
mean to derive @xmath70 , @xcite suggests to use an iterative re - weighting procedure to down - weight potential outlier points . after computing @xmath70 as the weighted mean at the first step , weights of all points are multiplied by a factor @xmath86 and @xmath70 is re - computed with these new weights .
the procedure is repeated until it converges .
many types of variable stars show continuous brightness variations ( with notable exceptions , the algol - type eclipsing binaries and flare stars , which remain at about constant brightness most of the time only occasionally showing large variations ) .
@xcite suggests to supplement @xmath24 , which is a measure of the degree of correlation between consecutive brightness measurements , with a robust measure of the kurtosis ( ` peakedness ' ) of the magnitude histogram : @xmath87 for a gaussian magnitude distribution , @xmath88{}\sqrt{2/{\mathrm \pi}}$ ] or will be less if there is an outlier point in the lightcurve making the overall magnitude distribution more ` peaked ' .
the two indices @xmath24 and @xmath78 can be combined to the index @xmath27 @xcite : @xmath89 where @xmath90 is the ratio of the weights of all of the lightcurve points to a total weight that the star would have if it had been successfully measured on all images .
this ratio is designed to reduce the combined variability index @xmath27 value for stars with a small number of measurements for the reasons outlined in section [ sec : filtering ] .
the combined index is designed to maximize chances of detection for well - measured continuously variable stars .
it is less effective for objects that show brightness variations only occasionally ( algol - type binaries , flare stars , transient events ) .
is divided into subsamples to calculate stetson s variability indices ( sections [ sec : welch ] and [ sec : stetson ] ) .
the arrows indicate the order in which lightcurve points are considered : first to last ( subsample names in lower case ) or reverse ( subsample names in upper case ) .
@xmath91 . dashed line is the weighted average.the weights are iteratively scaled by the factor @xmath92 , see equation ( [ eq : weightscale ] ) , section [ sec : stetson ] .
the open circle is an ` outlier ' 1mag brighter than it should be to follow the sine curve.,scaledwidth=48.0% ] c c@ c c@ c c@ c c@ c c@ c c@ c c@ c cl & & & & & & & & & + index & @xmath93 & @xmath94 & @xmath93 & @xmath94 & @xmath93 & @xmath94 & @xmath93 & @xmath94 & @xmath93 & @xmath94 & @xmath93 & @xmath94 & @xmath93 & @xmath94 & sec . & ref .
+ + @xmath13 & 0.157 & 0.928 & 0.185 & 0.936 & 0.871 & 0.991 & 0.566 & 0.990 & 0.571 & 0.991 & 0.684 & 0.993 & 0.253 & 0.981 & [ sec : chi2 ] & ( a ) + @xmath16 & 0.166 & 0.940 & 0.187 & 0.941 & 0.858 & 0.992 & 0.500 & 0.990 & 0.589 & 0.991 & 0.669 & 0.993 & 0.262 & 0.981 & [ sec : sigma ] & ( b ) + @xmath17 & 0.296 & 0.973 & 0.309 & 0.973 & 0.883 & 0.991 & 0.625 & 0.993 & 0.680 & 0.993 & 0.809 & 0.992 & 0.487 & 0.990 & [ sec : mad ] & ( c ) + @xmath18 & 0.269 & 0.968 & 0.279 & 0.969 & 0.880 & 0.992 & 0.611 & 0.993 & 0.675 & 0.992 & 0.801 & 0.992 & 0.464 & 0.991 & [ sec : iqr ] & ( d ) + @xmath19 & 0.215 & 0.957 & 0.237 & 0.954 & 0.882 & 0.991 & 0.654 & 0.993 & 0.626 & 0.992 & 0.768 & 0.991 & 0.437 & 0.993 & [ sec : roms ] & ( e ) + @xmath20 & 0.023 & 0.243 & 0.022 & 0.196 & 0.065 & 0.737 & 0.042 & 0.599 & 0.061 & 0.758 & 0.069 & 0.775 & 0.346 & 0.990 & [ sec : sigmaxs ] & ( f ) + @xmath21 & 0.074 & 0.860 & 0.097 & 0.901 & 0.488 & 0.993 & 0.361 & 0.984 & 0.415 & 0.990 & 0.141 & 0.934 & 0.070 & 0.911 & [ sec : peaktopeak ] & ( g ) + + @xmath22 & 0.668 & 0.993 & 0.530 & 0.993 & 0.157 & 0.993 & 0.250 & 0.998 & 0.332 & 0.986 & 0.032 & 0.994 & 0.025 & 0.980 & [ sec : lag1autocorr ] & ( h ) + @xmath23 & 0.164 & 0.926 & 0.193 & 0.935 & 0.877 & 0.991 & 0.286 & 0.993 & 0.641 & 0.991 & 0.354 & 0.994 & 0.097 & 0.989 & [ sec : welch ] & ( i ) + @xmath24 & 0.231 & 0.956 & 0.249 & 0.958 & 0.890 & 0.991 & 0.320 & 0.996 & 0.672 & 0.991 & 0.777 & 0.991 & 0.329 & 0.988 & [ sec : stetson ] & ( j ) + @xmath25 & 0.240 & 0.959 & 0.249 & 0.958 & 0.891 & 0.991 & 0.345 & 0.996 & 0.685 & 0.992 & 0.279 & 0.996 & 0.081 & 0.992 & [ sec : stetsontime ] & ( k ) + @xmath26 & 0.217 & 0.951 & 0.239 & 0.954 & 0.898 & 0.991 & 0.622 & 0.994 & 0.642 & 0.991 & 0.768 & 0.991 & 0.433 & 0.993 & [ sec : stetsonclip ] & ( d ) + @xmath27 & 0.272 & 0.964 & 0.274 & 0.961 & 0.877 & 0.991 & 0.323 & 0.996 & 0.712 & 0.992 & 0.796 & 0.991 & 0.495 & 0.994 & [ sec : stetson ] & ( j ) + @xmath28 & 0.295 & 0.959 & 0.408 & 0.972 & 0.020 & 0.011 & 0.020 & 0.016 & 0.020 & 0.011 & 0.020 & 0.012 & 0.019 & 0.002 & [ sec : cssd ] & ( l ) + @xmath29 & 0.077 & 0.863 & 0.083 & 0.895 & 0.303 & 0.993 & 0.132 & 0.981 & 0.439 & 0.990 & 0.714 & 0.992 & 0.255 & 0.989 & [ sec : excursions ] & ( m ) + @xmath0 & 0.674 & 0.992 & 0.625 & 0.988 & 0.859 & 0.991 & 0.242 & 0.997 & 0.315 & 0.985 & 0.031 & 0.994 & 0.023 & 0.974 & [ sec : vonneumann ] & ( n ) + @xmath30 & 0.012 & 0.970 & 0.040 & 0.988 & 0.168 & 0.978 & 0.021 & 0.946 & 0.025 & 0.689 & 0.019 & 0.539 & 0.019 & 0.897 & [ sec : ea ] & ( o ) + @xmath31 & 0.138 & 0.893 & 0.141 & 0.888 & 0.826 & 0.990 & 0.292 & 0.984 & 0.469 & 0.989 & 0.492 & 0.991 & 0.136 & 0.986 & [ sec : sb ] & ( p ) + c c@ c c@ c c@ c c@ c c@ c c@ c c@ c cl & & & & & & & & & + index & @xmath93 & @xmath94 & @xmath93 & @xmath94 & @xmath93 & @xmath94 & @xmath93 & @xmath94 & @xmath93 & @xmath94 & @xmath93 & @xmath94 & @xmath93 & @xmath94 & sec . & ref .
+ + @xmath13 & 0.166 & 0.927 & 0.184 & 0.936 & 0.873 & 0.991 & 0.556 & 0.990 & 0.563 & 0.991 & 0.685 & 0.993 & 0.252 & 0.980 & [ sec : chi2 ] & ( a ) + @xmath16 & 0.179 & 0.948 & 0.186 & 0.941 & 0.860 & 0.992 & 0.493 & 0.990 & 0.577 & 0.991 & 0.669 & 0.993 & 0.258 & 0.980 & [ sec : sigma ] & ( b ) + @xmath17 & 0.180 & 0.958 & 0.264 & 0.966 & 0.825 & 0.992 & 0.548 & 0.993 & 0.561 & 0.991 & 0.779 & 0.992 & 0.400 & 0.989 & [ sec : mad ] & ( c ) + @xmath18 & 0.175 & 0.954 & 0.247 & 0.962 & 0.826 & 0.992 & 0.556 & 0.993 & 0.608 & 0.992 & 0.777 & 0.992 & 0.384 & 0.988 & [ sec : iqr ] & ( d ) + @xmath19 & 0.191 & 0.950 & 0.231 & 0.953 & 0.870 & 0.991 & 0.632 & 0.993 & 0.589 & 0.991 & 0.757 & 0.991 & 0.406 & 0.991 & [ sec : roms ] & ( e ) + @xmath20 & 0.023 & 0.240 & 0.022 & 0.192 & 0.065 & 0.737 & 0.042 & 0.599 & 0.061 & 0.759 & 0.068 & 0.776 & 0.348 & 0.990 & [ sec : sigmaxs ] & ( f ) + @xmath21 & 0.096 & 0.898 & 0.107 & 0.911 & 0.668 & 0.993 & 0.400 & 0.989 & 0.471 & 0.991 & 0.165 & 0.957 & 0.096 & 0.952 & [ sec : peaktopeak ] & ( g ) + + @xmath22 & 0.708 & 0.993 & 0.545 & 0.993 & 0.877 & 0.991 & 0.864 & 0.992 & 0.759 & 0.991 & 0.887 & 0.992 & 0.689 & 0.993 & [ sec : lag1autocorr ] & ( h ) + @xmath23 & 0.171 & 0.926 & 0.192 & 0.935 & 0.881 & 0.991 & 0.636 & 0.990 & 0.642 & 0.991 & 0.783 & 0.992 & 0.369 & 0.990 & [ sec : welch ] & ( i ) + @xmath24 & 0.211 & 0.953 & 0.242 & 0.957 & 0.884 & 0.991 & 0.756 & 0.993 & 0.653 & 0.991 & 0.780 & 0.991 & 0.419 & 0.994 & [ sec : stetson ] & ( j ) + @xmath25 & 0.193 & 0.952 & 0.243 & 0.957 & 0.870 & 0.991 & 0.733 & 0.993 & 0.672 & 0.991 & 0.884 & 0.991 & 0.484 & 0.993 & [ sec : stetsontime ] & ( k ) + @xmath26 & 0.198 & 0.949 & 0.232 & 0.953 & 0.892 & 0.991 & 0.704 & 0.992 & 0.613 & 0.992 & 0.768 & 0.991 & 0.403 & 0.993 & [ sec : stetsonclip ] & ( d ) + @xmath27 & 0.211 & 0.956 & 0.260 & 0.959 & 0.858 & 0.991 & 0.732 & 0.993 & 0.682 & 0.992 & 0.795 & 0.991 & 0.551 & 0.994 & [ sec : stetson ] & ( j ) + @xmath28 & 0.197 & 0.950 & 0.324 & 0.974 & 0.018 & 0.009 & 0.019 & 0.014 & 0.019 & 0.007 & 0.019 & 0.010 & 0.019 & 0.001 & [ sec : cssd ] & ( l ) + @xmath29 & 0.509 & 0.988 & 0.413 & 0.989 & 0.854 & 0.992 & 0.700 & 0.993 & 0.622 & 0.992 & 0.784 & 0.992 & 0.430 & 0.994 & [ sec : excursions ] & ( m ) + @xmath0 & 0.715 & 0.992 & 0.631 & 0.988 & 0.876 & 0.991 & 0.873 & 0.992 & 0.756 & 0.991 & 0.887 & 0.992 & 0.700 & 0.994 & [ sec : vonneumann ] & ( n ) + @xmath30 & 0.257 & 0.978 & 0.405 & 0.990 & 0.828 & 0.991 & 0.692 & 0.994 & 0.480 & 0.991 & 0.885 & 0.991 & 0.525 & 0.994 & [ sec : ea ] & ( o ) + @xmath31 & 0.150 & 0.893 & 0.146 & 0.888 & 0.846 & 0.989 & 0.571 & 0.981 & 0.676 & 0.988 & 0.736 & 0.988 & 0.512 & 0.990 & [ sec : sb ] & ( p ) + | photometric measurements are prone to systematic errors presenting a challenge to low - amplitude variability detection . in search for a general - purpose variability detection technique able to recover a broad range of variability types including currently unknown ones
, we test 18 statistical characteristics quantifying scatter and/or correlation between brightness measurements .
we compare their performance in identifying variable objects in seven time series data sets obtained with telescopes ranging in size from a telephoto lens to 1m - class and probing variability on time - scales from minutes to decades .
the test data sets together include lightcurves of 127539 objects , among them 1251 variable stars of various types and represent a range of observing conditions often found in ground - based variability surveys .
the real data are complemented by simulations .
we propose a combination of two indices that together recover a broad range of variability types from photometric data characterized by a wide variety of sampling patterns , photometric accuracies and percentages of outlier measurements .
the first index is the interquartile range ( iqr ) of magnitude measurements , sensitive to variability irrespective of a time - scale and resistant to outliers .
it can be complemented by the ratio of the lightcurve variance to the mean square successive difference , @xmath0 , which is efficient in detecting variability on time - scales longer than the typical time interval between observations .
variable objects have larger @xmath0 and/or iqr values than non - variable objects of similar brightness .
another approach to variability detection is to combine many variability indices using principal component analysis .
we present 124 previously unknown variable stars found in the test data .
[ firstpage ] methods : data analysis , methods : statistical , stars : variables : general |
searching for high - energy tau neutrinos ( @xmath7 gev ) will yield quite useful information about the highest energy phenomenon occurring in the universe @xcite . the same search may also provide evidence for physics beyond the standard model @xcite .
the latter is suggested by the recent measurements of the atmospheric muon neutrino deficit , though yet there is no observation of oscillated atmospheric tau neutrinos at a significant confidence level @xcite .
interestingly , it is also only recently that we have the first evidence of existence of tau neutrinos @xcite .
the high - energy tau neutrinos can be produced in @xmath8 and @xmath9 interactions taking place in cosmos .
these interactions produce unstable hadrons that decay into tau neutrinos . in this paper
, we mainly concentrate on @xmath8 interactions and will only briefly comment on @xmath9 interactions as source interactions for producing high - energy tau neutrinos .
there can be several astrophysical sites where the @xmath8 interactions may occur .
examples of these include the relatively nearby and better known astrophysical sites such as our galaxy and the earth atmosphere , where the basic @xmath8 interactions occur as @xmath10 interactions .
the @xmath8 interactions in these sites form a rather certain background to the extra - galactic high - energy tau neutrino searches .
it is possible that such interactions are the only sources of high - energy tau neutrinos should the search for high - energy tau neutrinos originating from several proposed distant sites such as agns , grbs , as well as groups and clusters of galaxies , turns out to be negative .
therefore , it is rather essential to investigate the high - energy tau neutrino flux expected from our galaxy and the earth atmosphere .
both the galactic and atmospheric tau neutrinos can be categorized into intrinsic and oscillated ones . here
, intrinsic @xmath1 flux refers to the @xmath1 produced directly by an interaction while oscillated @xmath1 refers to the @xmath1 resulted from the @xmath11 oscillation .
presently , there exists no estimate for the intrinsic high - energy @xmath1 flux originating from our galaxy in @xmath8 interactions , although the estimates for @xmath12 and @xmath3 fluxes due to the same interactions are available @xcite . in this work
, we calculate the intrinsic @xmath1 flux from our galaxy by using the perturbative and nonperturbative qcd approaches to model the @xmath8 interactions , and taking into account all major tau neutrino production channels up to @xmath13 gev .
we note that the production of tau neutrinos in the terrestrial context was discussed in ref .
@xcite , which uses a nonperturbative qcd approach for @xmath8 interactions . to calculate the oscillated galactic and atmospheric @xmath1 fluxes , we apply the two - flavor neutrino oscillation analysis @xcite .
it is essential to compare the above galactic tau neutrino flux with the flux of atmospheric tau neutrinos .
the intrinsic atmospheric tau neutrino flux has been calculated for @xmath14 gev @xcite . in this work ,
we extend the calculation up to @xmath13 gev .
such an extension requires the input of cosmic - ray flux spectrum for an energy range beyond that considered in ref .
furthermore , for a greater neutrino energy , the solutions of cascade equations relevant to the neutrino production behave differently .
for the oscillated @xmath1 flux , it is interesting to note that the oscillation length for @xmath15 for the energy range @xmath0 is much greater than the thickness of the earth atmosphere .
hence the oscillated atmospheric @xmath1 flux in this case is highly suppressed .
one may argue that the interaction of the high - energy cosmic - rays with the ubiquitous cosmic microwave background ( cmb ) photons present in cosmos ( @xmath9 rather than @xmath8 interactions ) could also be an important source for high - energy astrophysical tau neutrinos .
the center - of - mass energy ( @xmath16 ) needed to produce a @xmath17 lepton and a @xmath18 is at least @xmath19 1.8 gev . in a collision between a proton with an energy @xmath20 and a cmb photon with an energy @xmath21 ,
the center - of - mass energy squared of the system satisfies @xmath22 .
since the peak of the cmb photon flux spectrum with a temperature @xmath19 2.7 k is at about @xmath23 ev , it requires a very energetic proton with @xmath24 gev in order to produce a @xmath25 pair .
thus , the contribution of the intrinsic tau neutrino flux from the interaction between the cosmic proton and the cmb photon is negligible , unless we are considering extremely high - energy protons , beyond the presently observed highest energy cosmic - rays @xcite . to compute the oscillated @xmath1 flux in this case ,
we consider the non - tau neutrino flux generated by the @xmath26 interaction via the @xmath27 resonance , commonly referred to as the greisen - zatsepin - kuzmin ( gzk ) cutoff interaction , which assumes that the proton travels a cosmological distance @xcite .
a recent calculation of the intrinsic non - tau gzk neutrino flux indicates that this flux peaks typically at @xmath28 gev @xcite , beyond the reach of presently operating high - energy neutrino telescopes such as amanda and baikal experiments @xcite .
this flux decreases for @xmath29 gev .
in fact , it falls below the intrinsic non - tau galactic - plane neutrino flux for @xmath30 gev .
the neutrino flavor oscillations of the intrinsic non - tau gzk neutrinos into tau neutrinos result in a @xmath1 flux comparable to the original non - tau neutrino flux @xcite .
therefore , in the absence ( or smallness ) of tau neutrino flux from other possible extra - galactic astrophysical sites , the only source of high - energy tau neutrinos besides the atmospheric background is from our ( plane of ) galaxy , typically for @xmath31 .
this is an energy range to be explored by the above high - energy neutrino telescopes in the near future .
the organization of the paper is as follows . in section
ii , we discuss the calculation of intrinsic high - energy tau neutrino flux from our galaxy , including the description of the flux formula , the galaxy model , and the various tau neutrino production channels taken into account in the calculation . although this flux will be shown small , we shall go through some details of the calculation since they are also useful for the calculation in the next section . in section iii ,
we present our result on the intrinsic atmospheric @xmath1 flux and compare it with the galactic one . in section iv , we discuss the effects of neutrino flavor mixing , which are used to construct the oscillated @xmath1 fluxes from the galaxy and the earth atmosphere respectively . the total galactic @xmath1 flux ( the sum of intrinsic and oscillated fluxes ) is compared to its atmospheric counterpart , and the dominant energy range for the former flux is identified .
we also mention the currently envisaged prospects for identifying the high - energy tau neutrinos .
we summarize in section v.
we use the following formula for computing the @xmath1 flux : @xmath32 in the above equation , @xmath33 is the tau neutrino energy and the cosmic - ray flux spectrum , @xmath34 , is given by @xcite @xmath35 where @xmath36 gev and @xmath34 is in units of @xmath37 s@xmath38 sr@xmath38 gev@xmath38 .
we assume directional isotropy in @xmath34 for the above energy range .
the function @xmath39 is equal to @xmath40 , where @xmath41 is the @xmath8 interaction length and @xmath42 is a representative distance in the galaxy along the galactic plane .
the target particles are taken to be protons with a constant number density of 1 @xmath43 and @xmath42 is taken to be @xmath44 kpc , where 1 pc @xmath45 @xmath46 cm .
the @xmath47 is the total inelastic @xmath8 cross section .
since the high - energy protons traverse a distance @xmath42 much shorter than the proton interaction length , the proton flux spectrum @xmath48 is assumed to be constant over the distance @xmath42 .
furthermore , we calculate the intrinsic tau neutrino flux along the galactic plane only to obtain the maximal expected tau neutrino flux .
the matter density decreases exponentially in the direction orthogonal to the galactic plane , therefore the amount of intrinsic tau neutrino flux decreases by approximately two orders of magnitude for the energy range of our interest . for further details ,
see ingelman and thunman in ref .
the function @xmath39 in eq .
( [ aa ] ) basically gives the number of inelastic @xmath8 collisions in the distance @xmath42 , while the distribution @xmath49 is the differential cross section normalized by @xmath50 , i.e. , @xmath51 the above distribution gives the fraction of inelastic @xmath8 interactions that goes into @xmath1 s .
we can now simplify eq .
( [ aa ] ) into @xmath52 it is clear that the task of computing @xmath53 relies on the evaluation of the differential cross section @xmath54 in @xmath8 interactions . in this work , we include all major production channels of tau neutrinos , namely , via the @xmath55 meson , @xmath56-hadron , @xmath57 , @xmath58 , and @xmath59 . in general , the symbol @xmath1
shall be used to account for the contribution of @xmath1 and @xmath60 unless otherwise mentioned .
we note that the heavy intermediate states , such as the @xmath61 meson , @xmath56-hadrons and other heavier states , decay ( into @xmath1 ) before they interact with other particles in the galactic plane .
this is due to the rather small matter density of the medium and the large distance between the proton source and the earth . before we proceed ,
let us remark that the intrinsic tau neutrino production by the galactic @xmath9 interactions is suppressed relative to that in the galactic @xmath8 interaction , because @xmath62 for the energy range of interest to us @xcite .
the lightest meson that can decay into a @xmath17-@xmath63 pair is the @xmath55 meson .
it was pointed out that the production and the decay of @xmath55 meson is the major production channel for tau neutrinos in the agn @xcite .
we expect the same will be true for galactic tau neutrinos .
the @xmath55 meson decays into a charged @xmath17 lepton and a @xmath63 .
the charged @xmath17 lepton subsequently decays into the second @xmath63 plus other particles , which can be one prong or three prong .
the kinematics of the @xmath17 lepton decay is treated by the monte carlo technique . for simplicity , we assume that the @xmath17 lepton decays into a @xmath63 and a particle @xmath64 with the mass @xmath65 satisfying @xmath66 . consequently , the second @xmath63 is much more @xmath67 than the first one because the @xmath55 mass is only slightly larger than @xmath68 .
we take the branching ratio @xmath69 as @xmath70 @xcite .
we now turn to the production of @xmath55 mesons . here , we employ two approaches to calculate the production of @xmath55 mesons : ( i ) the perturbative qcd ( pqcd ) and ( ii ) the quark - gluon string model ( qgsm ) . in the pqcd approach ,
we use the leading - order result for @xmath71 : @xmath72 where @xmath73 are the parton distribution functions ( we use the cteqv5 @xcite ) , while the parton subprocesses are @xmath74 .
we use a @xmath75 factor , @xmath76 , to account for the nlo corrections @xcite .
the matrix elements for these subprocesses can be found in appendix a. the @xmath77 or @xmath78 then undergoes fragmentation into the @xmath55 meson , which we model by the peterson fragmentation function @xcite with @xmath79 @xcite ( see appendix a ) .
the probability @xmath80 of a @xmath81 quark fragmenting into a @xmath55 is 0.19 @xcite ( we have added the @xmath82 and @xmath83 ) .
typically , a fraction @xmath84 of the charm - quark energy is transferred to the @xmath55 meson so that the energy spectrum of @xmath55 is softer than that of the charm quark .
the qgsm approach is nonperturbative and is based on the string fragmentation .
it contains a number of parameters determined by experiments @xcite .
the production cross section of the @xmath55 meson is given by the sum of @xmath85-pomeron terms @xmath86 where @xmath87 and @xmath88 .
the functions @xmath89 and @xmath90 are given in appendix b. a comparison of these two approaches for @xmath61 meson is shown in fig .
the dashed line in the figure is the spectrum of the injected proton flux given by eq .
( [ proton ] ) .
the @xmath63 spectra calculated by these two approaches agree well with each other for @xmath4 gev . beyond this energy ,
the qgsm approach gives a relatively harder spectrum .
in fact , this behavior was already seen in the @xmath91 distribution . a comparison between the pqcd and qgsm approach to the energy spectrum of the intrinsic galactic @xmath1 flux coming from the @xmath55 meson .
the thick dashed curve is the injected proton flux spectrum given by eq .
( [ proton]).,width=480 ] nevertheless , in the region where the two approaches differ , the tau neutrino flux is already small . to our knowledge ,
the heaviest meson production that the qgsm has been applied to is the @xmath55 meson production .
the current highest energy collider experiment for @xmath61 meson production is at the fermilab tevatron with @xmath92 gev , which corresponds to an @xmath93 gev , as @xmath94 in our setting .
note that up to this @xmath16 , the agreement between the two approaches is quite good , according to fig .
we have used a factorization scale @xmath95 and the one - loop running strong coupling constant @xmath96 with the value @xmath97 , and the @xmath98 gev .
an important quantity in the neutrino flux calculation is the average fraction of the injected proton energy being transferred to the tau neutrino , i.e. , the ratio @xmath99 .
the average value of @xmath100 is given either by the mean @xmath101 or by the value of @xmath100 at which the distribution @xmath102 attains the peak .
we have found that both averages of @xmath100 are very close to each other .
the @xmath103 ranges from @xmath104 to @xmath105 for @xmath20 from @xmath106 to @xmath107 gev for the production channel @xmath108 , using the pqcd approach . the higher the injected proton energy , the smaller is the fraction of the incident @xmath109 that goes into hadrons .
the production of @xmath112 and @xmath113 in @xmath8 interactions can be calculated quite reliably by the pqcd approach , similar to the calculation of @xmath114 .
the relevant matrix elements , including the ones for @xmath115 and @xmath116 , are listed in appendix a. intrinsic galactic tau neutrino flux calculated via various intermediate states and channels : @xmath55 , @xmath56-hadron , @xmath58 , @xmath59 , and @xmath57 .
the injected proton flux spectrum is also shown .
, width=480 ] the results , @xmath117 , @xmath118 , @xmath42 , @xmath119 , etc .
, we estimate that our calculation of the intrinsic galactic @xmath1 flux is reliable within an order of magnitude.]are shown in fig .
a few observations can be drawn from the figure @xcite .
( i ) the production via @xmath55 mesons dominates for @xmath120 gev , followed by @xmath56-hadrons , @xmath58 , @xmath59 , and @xmath57 respectively .
( ii ) for @xmath121 gev all these production channels become @xmath122 .
( iii ) the intrinsic tau neutrino flux is about @xmath123 orders of magnitude smaller than the injected proton flux .
the earth atmosphere is an interesting extra - terrestrial site where the basic @xmath8 interaction occurs in the form of a @xmath10 collision with @xmath124 the nuclei present in the earth atmosphere .
incidently , it is the only known nearby extra - terrestrial site from where the intrinsic neutrinos are observed as a result of high - energy cosmic - ray interactions .
we have calculated the downward and horizontal intrinsic atmospheric @xmath1 flux for the energy range @xmath125 .
we used the nonperturbative qcd approach mentioned in the last section to model the production of @xmath61 mesons in @xmath10 interactions .
we have used the @xmath34 given by eq .
( [ proton ] ) and the @xmath126-moment description for the calculation of intrinsic tau neutrino flux , which is appropriate for a varying target density medium @xcite .
since the tau neutrino flux is determined by the flux of @xmath55 meson , we briefly discuss the cascade equation for the @xmath55 flux . in general , we have @xcite @xmath127 where @xmath128 is the slant depth , i.e. , the amount of atmosphere ( in @xmath129 ) traversed by the @xmath55 meson ( @xmath130 at the top of the atmosphere ) , @xmath131 and @xmath132 are the fluxes of protons and @xmath55 mesons respectively .
the @xmath133 and @xmath134 are the interaction thickness ( in @xmath129 ) and the decay length of the @xmath55 meson respectively . finally , the @xmath126-moments @xmath135 and @xmath136 describe the effectiveness of generating @xmath55 meson from the higher - energy protons and @xmath55 mesons respectively .
we have @xmath137 where @xmath138 and @xmath139 are defined according to eq .
( [ dnpp ] ) .
we note that eq .
( [ cascade ] ) should be solved together with the cascade equation governing the propagation of high - energy cosmic - ray protons .
in fact , the proton flux equation can be easily solved such that @xmath140 where @xmath141 is the proton attenuation length with @xmath142 the proton interaction thickness ( in @xmath129 ) and the @xmath143 given by @xmath144 an analytic solution of eq .
( [ cascade ] ) can be obtained for either the low or the high energy limit .
such limits are characterized by whether the @xmath61 decays before it interacts with the medium or vice versa .
the critical energy separating the two limits is approximately @xmath145 gev . in the low energy limit
, we disregard the first and third terms in the r.h.s . of eq .
( [ cascade ] ) . on the other hand
, one can drop the second term in the high energy limit . with
@xmath146 determined , the @xmath1 flux can be calculated by considering the decay @xmath147 and the subsequent decay @xmath148 as treated in section ii .
we first obtain two @xmath1 fluxes , valid for low and high energy limits respectively , in terms of @xmath126-moments and then interpolate the two fluxes .
a complete numerical solution without the interpolation is given in a separate work @xcite .
intrinsic horizontal @xmath1 flux via production and decay of the @xmath55 meson in the earth atmosphere . for @xmath149 ,
the results by pr are also shown . in the inset
, we compare our calculated @xmath150 with the one given by rescaling tig s @xmath151 .
the total intrinsic galactic - plane tau neutrino flux is also shown.,width=480 ] in fig .
[ fig2 ] , we show our result for the intrinsic atmospheric @xmath1 flux along the horizontal direction . for comparison , the results by pasquali and reno ( pr ) @xcite , valid for @xmath152 , are also shown .
the @xmath1 flux along all the other direction is small .
for example with a energy @xmath153 gev is degraded in energy to about @xmath154 gev after crossing the earth . ] , the downward @xmath1 flux is about @xmath155 times smaller than the horizontal one for @xmath156 gev .
we remark that the major uncertainty for determining the above @xmath1 flux is the @xmath126-moment @xmath157 . in ref .
@xcite , the authors calculate @xmath157 using two different approaches , which then give rise to different results for the @xmath1 flux . the first approach is based upon next - to - leading order ( nlo ) perturbative qcd @xcite , while the second approach rescales the @xmath151 given by the pythia @xcite calculation of thunman , ingelman , and gondolo ( tig ) @xcite . in the inset of fig .
[ fig2 ] , we also show our calculated @xmath150 in comparison with the one given by rescaling tgi s result for @xmath151
. we do not show the @xmath150 obtained by nlo perturbative qcd since it is not explicitly given in ref .
in the context of two neutrino flavors , @xmath3 and @xmath1 , the total @xmath1 flux , @xmath158 , is given by @xcite @xmath159 here @xmath160 .
the neutrino flavor oscillation length for @xmath11 is @xmath161 . for @xmath162 and with @xmath163 ev@xmath164 , we obtain @xmath165 . we assume maximal flavor mixing between @xmath3 and @xmath1 .
for intrinsic neutrinos produced along the galactic plane , we take @xmath166 given by ingelman and thunman in ref . @xcite by extrapolating it up to @xmath167 gev , whereas for @xmath168 , we use our results obtained in section ii . for galactic - plane neutrinos
, we note that @xmath169 , where @xmath170 kpc is the typical average distance the intrinsic high - energy muon neutrinos traverse after being produced in our galaxy .
( [ osc ] ) then implies that , on the average , half of the muon neutrino flux will be oscillated into tau neutrino flux , reducing its intrinsic level to one half . for downward going neutrinos produced in the earth atmosphere , we take @xmath171 km as an example .
we use the ( prompt ) @xmath166 given in @xcite , whereas for @xmath168 , we use our results obtained in section iii . for
horizontal and upward going atmospheric neutrinos , @xmath172 km .
here , @xmath173 , and so the intrinsic atmospheric tau neutrino flux dominates over the oscillated one for @xmath174 gev , essentially irrespective of the incident direction .
we present these results in fig .
[ fig2 ] , along with the gzk oscillated tau neutrino flux briefly mentioned in section i. for the gzk neutrinos , @xmath175 mpc and @xmath166 is taken from ref . @xcite . from the figure
, we note that the galactic - plane oscillated @xmath1 flux
_ dominates _ over the intrinsic atmospheric @xmath1 flux for @xmath176 gev , whereas the gzk oscillated tau neutrino flux dominates for @xmath177 gev .
a prospective search for high - energy tau neutrinos can be done by appropriately utilizing the characteristic @xmath17 lepton range in deep inelastic ( charged current ) tau - neutrino - nucleon interactions , in addition to the attempt of observing the associated showers . for @xmath33 close to @xmath178 gev
, the ( anti electron ) neutrino - electron resonant scattering is also available to the search for ( secondary ) high - energy tau neutrinos @xcite .
the main advantages of using the latter channel are that the neutrino flavor in the initial state is least affected by neutrino flavor oscillations and that this cross section is free from theoretical uncertainties @xcite .
the appropriate utilization of the characteristic @xmath17 lepton range in both interaction channels not only helps to identify the incident neutrino flavor but also helps to bracket the incident neutrino energy as well , at least in principle .
galactic - plane , horizontal atmospheric and gzk tau neutrino fluxes under the assumption of neutrino flavor oscillations.,width=480 ] for downward going or near horizontal high - energy tau neutrinos , the deep inelastic neutrino - nucleon scattering , occurring near or inside the detector , produces two ( hadronic ) showers @xcite .
the first shower is due to a charged - current neutrino - nucleon deep inelastic scattering , whereas the second shower is due to the ( hadronic ) decay of the associated @xmath17 lepton produced in the first shower .
it might be possible for the proposed large neutrino telescopes such as icecube to constrain the two showers simultaneously for @xmath179 , depending on the achievable shower separation capabilities @xcite ( see , also , @xcite ) . here
, the two showers develop mainly in ice . using the same shower separation criteria
as given in @xcite , we note that the @xmath19 ( 100 m)@xmath180 proposed neutrino detector , commonly called the megaton detector @xcite , may constrain the two showers separated by @xmath181 10 m , typically for @xmath182 .
the two nearly horizontal showers may also possibly be contained in a large surface area detector array such as pierre auger , typically for @xmath183 @xcite .
in contrast to previous situations , here the two showers develop mainly in air .
several different suggestions have recently been made to measure only one shower , which is due to the @xmath17 lepton decay , typically for @xmath184 , while the first shower is considered to be mainly absorbed in the earth @xcite .
the upward going high - energy tau neutrinos with @xmath185 gev , on the other hand , may avoid earth shadowing to a certain extent because of the characteristic @xmath17 lepton range , unlike the upward going electron and muon neutrinos , and may appear as a rather small pile up of @xmath186 ( @xmath187 ) with @xmath188 gev @xcite . however , the empirical determination of incident tau neutrino energy seems rather challenging here .
the above studies indicate that for a rather large range of tau neutrino energy , a prospective search may be carried out .
the event rate in each experimental configuration is directly proportional to the incident @xmath1 flux and the effective area of the detector concerned .
presently , no direct empirical upper bounds ( or observations ) for high - energy tau neutrinos exist .
to have an idea of the event rate , let us consider the downward going high - energy tau neutrinos originating from the galactic - plane due to neutrino flavor oscillations .
the flux of such neutrinos has been given in fig .
4 . following ref .
@xcite , we note that these galactic tau neutrinos give a representative event rate of @xmath189 per year per steradian for two separable and contained showers with @xmath190 gev in a km@xmath180 volume neutrino telescope such as the proposed icecube .
we have calculated the @xmath1 flux due to @xmath8 interactions in our galaxy .
this flux consists of intrinsic tau neutrino flux and that arising from the oscillations of muon neutrinos .
we note that the latter flux is dominant over the former by four to five orders of magnitude for the considered neutrino energy range . from fig . 4 , one can see that the main background for the search of high - energy extra - galactic tau neutrino is due to the muon neutrinos produced in the galactic - plane , which then oscillate into tau neutrinos .
such a flux dominates for @xmath6 .
therefore it is clear that searching for extra - galactic tau neutrinos orthogonal to the galactic - plane is more prospective . in the calculation of galactic tau neutrino flux
, we have used a simplified model of matter distribution along our galactic plane to obtain the maximal intrinsic tau neutrino flux .
we have explicitly calculated the contribution of heavier states such as @xmath110 , @xmath111 as well as @xmath115 and @xmath116 in addition to the more conventional @xmath61 channel to the intrinsic tau neutrino flux .
we have estimated the average fraction of the incident cosmic - ray energy that goes into tau neutrinos and found it to be less than 1@xmath191 .
the contributions from @xmath110 , @xmath111 , @xmath115 and @xmath116 channels are comparable to @xmath55 for @xmath192 gev . for @xmath61 channel , we have used both perturbative and nonperturbative qcd approaches .
we have extended a previous calculation of intrinsic atmospheric @xmath1 flux from @xmath193 gev up to @xmath194 gev . here
, we used the nonperturbative qcd approach to calculate the production of @xmath61 mesons in @xmath10 interactions . in comparison with the intrinsic galactic - plane @xmath1 flux
, it is large .
however , since the distance between the detector and the neutrino source in the galactic plane is sufficiently large , the neutrino flavor oscillations of non - tau neutrinos into tau neutrinos makes the eventual tau neutrino flux along the galactic plane greater than the atmospheric tau neutrino flux for @xmath195 .
however , the intrinsic atmospheric @xmath1 flux dominates over the oscillated galactic @xmath1 flux in the direction orthogonal to the galactic plane .
we have also briefly mentioned the presently envisaged prospects for observations . in summary , we have completed the compilation of all definite sources of tau neutrino flux , i.e. , those from our galaxy and from the earth atmosphere . such a compilation is needed before one conducts the search for tau neutrinos from extra - galactic sources .
h.a . and k.c .
are supported in part by the physics division of national center for theoretical sciences under a grant from the national science council of taiwan .
g.l.l . and j.j.t .
are supported by the national science council of r.o.c . under the grant number nsc90 - 2112-m009 - 023 .
99 for a recent review article , see , for instance , f. halzen , arxiv : astro - ph/0111059 and references therein . y. fukuda _ et al .
_ [ super - kamiokande collaboration ] , phys . rev .
lett . 81 ( 1998 ) 1562
. s. fukuda _ et al .
_ [ super - kamiokande collaboration ] , phys . rev .
85 ( 2000 ) 3999 .
k. kodama _ et al . _
[ donut collaboration ] , phys .
b 504 ( 2001 ) 218 ; j. sielaff [ donut collaboration ] , arxiv : hep - ex/0105042 . f. w. stecker , astrophys .
j. 228 ( 1979 ) 919 ; g. domokos , b. elliott and s. kovesi - domokos , j. phys .
g 19 ( 1993 ) 899 ; v. s. berezinsky , t. k. gaisser , f. halzen and t. stanev , astropart .
phys . 1 ( 1993 ) 281 ; g. ingelman and m. thunman , arxiv : hep - ph/9604286 .
a. de rujula , e. fernandez and j. j. gomez - cadenas , nucl .
b 405 ( 1993 ) 80 ; m. c. gonzalez - garcia and j. j. gomez - cadenas , phys . rev .
d 55 ( 1997 ) 1297 .
l. pasquali and m. h. reno , phys .
d 59 ( 1999 ) 093003 . m. nagano and a. a. watson , rev .
72 ( 2000 ) 689 .
k. greisen , phys .
lett . 16 ( 1966 ) 748
; g. t. zatsepin and v. a. kuzmin , jetp lett . 4 ( 1966 ) 78 [ pisma zh .
. teor .
fiz . 4 ( 1966 )
r. engel , d. seckel and t. stanev , phys .
d 64 ( 2001 ) 093010 .
h. athar , arxiv : hep - ph/0008121 .
t. h. burnett _
[ jacee collaboration ] , astrophys .
j. lett . 349
( 1990 ) l25 . for an estimate of galactic @xmath196 , see , berezinsky _ et al .
, _ in ref .
h. athar , nucl .
phys . proc .
suppl . 76 ( 1999 ) 419 .
d. e. groom _ et al . _
[ particle data group collaboration ] , eur .
j. c 15 ( 2000 ) 1 .
h. l. lai _ et al . _
[ cteq collaboration ] , eur .
phys . j. c 12 ( 2000 ) 375 .
see , for instance , p. nason , s. dawson and r. k. ellis , nucl .
b 327 ( 1989 ) 49 [ erratum - ibid .
b 335 ( 1989 ) 260 ] . c. peterson , d. schlatter , i. schmitt and p. m. zerwas , phys .
d 27 ( 1983 ) 105 .
r. barate _
[ aleph collaboration ] , eur .
j. c 16 ( 2000 ) 597 .
a. b. kaidalov , phys .
b 116 ( 1982 ) 459 ; a. b. kaidalov and o. i. piskunova , sov . j. nucl
. 43 ( 1986 ) 994 [ yad .
43 ( 1986 ) 1545 ] ; g. g. arakelian and p. e. volkovitsky , z. phys . a 353 ( 1995 ) 87
; g. h. arakelian , phys . atom .
. 61 ( 1998 ) 1570 [ yad .
fiz . 61 ( 1998 )
1682 ] ; g. h. arakelian and s. s. eremian , phys . atom .
62 ( 199 ) 1724 [ yad . fiz . 62 ( 1999 ) 1851 ] .
see , also , m. c. gonzalez - garcia and j. j. gomez - cadenas , in ref .
for a more detailed discussion , see k. cheung , in : h. athar , g .-
lin and k .- w .
ng ( eds . ) , proceedings of the first ncts workshop on astroparticle physics , 2001 , kenting ( taiwan ) [ to be published ] . t. k. gaisser , cosmic rays and particle physics , cambridge university press , newyork , 1990 .
tseng , ph.d .
thesis , nctu , 2002 ( in preparation ) .
s. frixione , m. l. mangano , p. nason , and g. ridolfi , nucl .
b 431 ( 1994 ) 453 .
t. sjostrand , comput .
commun . 82 ( 1994 ) 74 .
m. thunman , g. ingelman and p. gondolo , astropart .
phys . 5 ( 1996 ) 309 .
see , for instance , h. athar , m. jezabek and o. yasuda , phys .
d 62 ( 2000 ) 103007 and references therein .
l. v. volkova and g. t. zatsepin , phys .
. 64 ( 2001 ) 266 [ yad .
fiz . 64 ( 2001 )
d. fargion , arxiv : astro - ph/9704205 . h. athar and g .- l .
lin , arxiv : hep - ph/0108204 ; _ ibid _ , arxiv : hep - ph/0201026 . j. g. learned and s. pakvasa , astropart .
phys . 3 ( 1995 ) 267 .
h. athar , astropart .
phys . 14 ( 2000 ) 217 ; h. athar , g. parente and e. zas , phys .
d 62 ( 2000 ) 093010 .
see , for instance , h. chen _ et al . _
[ vlbl study group h2b-1 collaboration ] , arxiv : hep - ph/0104266 .
h. athar , arxiv : hep - ph/0004083 .
d. fargion , arxiv : astro - ph/0002453 ; x. bertou , p. billoir , o. deligny , c. lachaud and a. letessier - selvon , arxiv : astro - ph/0104452 ; j. l. feng , p. fisher , f. wilczek and t. m. yu , arxiv : hep - ph/0105067 ; a. kusenko and t. weiler , arxiv : hep - ph/0106071 . f. halzen and d. saltzberg , phys .
lett . 81 ( 1998 ) 4305 . f. becattini and s. bottai , astropart .
phys . 15 ( 2001 ) 323 ; s. iyer dutta , m. h. reno and i. sarcevic , arxiv : hep - ph/0110245 and references therein ; j. f. beacom , p. crotty and e. w. kolb , arxiv : astro - ph/0111482 .
in this appendix , we list the matrix element squared for the subprocesses of @xmath197 , where @xmath198 , used in the pqcd calculation . @xmath199 \ ; , \nonumber\\ % \frac{\mbox{d } \hat \sigma}{\mbox{d } \cos\theta^ * } ( gg\to q \bar q)&= & \frac{g_s^4 \beta}{768 \pi \hat s } \biggr \ { \frac{4}{(\hat t- m_q^2)^2 } ( -m_q^4 -3m_q^2 \hat t -m_q^2\hat u + \hat
u \hat t ) \nonumber \\ & + & \frac{4}{(\hat u- m_q^2)^2 } ( -m_q^4 -3m_q^2 \hat u -m_q^2 \hat t + \hat u \hat t ) \nonumber \\ & + & \frac{m_q^2}{(\hat u - m_q^2)(\hat t -m_q^2 ) } ( 2m_q^2 + \hat t
+ \hat u ) + 18 \frac{1}{\hat s^2 } ( m_q^4 - m_q^2 ( \hat t+ \hat u ) + \hat t \hat
u ) \nonumber \\ & + & \frac{9}{\hat t -m_q^2 } \frac{1}{\hat s } ( m_q^4 - 2 m_q^2 \hat t + \hat u \hat t ) + \frac{9}{\hat u -m_q^2 } \frac{1}{\hat s } ( m_q^4 - 2 m_q^2 \hat u + \hat u \hat t ) \biggr \},\end{aligned}\ ] ] where @xmath200 , @xmath201 and @xmath202 are the usual mandelstem variables with @xmath203 .
the subprocesses for high - energy tau neutrino production via @xmath115 and @xmath116 are @xmath204 and @xmath205 .
the spin- and color - averaged amplitude squared for these subprocesses are given by @xmath206 \;,\end{aligned}\ ] ] where @xmath207 and @xmath208 is the weak mixing angle , @xmath209 is the third component of the weak isospin and @xmath210 is the electric charge in units of proton charge of the fermion @xmath211 .
the peterson fragmentation function is given by @xmath212^{2}},\ ] ] where @xmath213 is the normalization constant and @xmath214 is given in the text .
in eq . ( [ qgsm ] ) , the functions @xmath89 and @xmath90 are given as follows : where @xmath216 \nonumber\\ & & + \int_x^1 \mbox{d}x_1 \;f_p^{d_{sea}}(x_1,n ) \ ; \left [ g_{d}^{d_s } \left(\frac{x}{x_1}\right ) + g_{\bar{d}}^{d_s } \left ( \frac{x}{x_1 } \right ) \right ] \nonumber \\ & & + \delta_s \int_x^1 \mbox{d}x_1 \;f_p^{s_{sea}}(x_1,n ) \;\left [ g_{s}^{d_s } \left(\frac{x}{x_1}\right ) + g_{\bar{s}}^{d_s } \left ( \frac{x}{x_1}\right ) \right ] \nonumber\\ & & + \delta_c \int_x^1 \mbox{d}x_1 \;f_p^{c_{sea}}(x_1,n ) \ ; \left [ g_{c}^{d_s } \left(\frac{x}{x_1}\right ) + g_{\bar{c}}^{d_s } \left ( \frac{x}{x_1}\right ) \right ] \biggr \}.\end{aligned}\ ] ] in the above , @xmath217 s are the distribution functions describing the @xmath218pomeron distribution functions of quarks or diquarks ( @xmath219 ) with a fraction of energy @xmath220 from the proton , and @xmath221 s are the fragmentation functions of the quark or diquark chain into a hadron @xmath222 which carries a fraction @xmath223 of its energy .
the list of the @xmath226 are given by @xmath227 in the above , the input parameters are as follows : @xmath228 the function @xmath229 is given by the following formulas : @xmath230 where @xmath231 the best fit parameters are as follows : | we compare the tau neutrino flux arising from the galaxy and the earth atmosphere for @xmath0 .
the intrinsic and oscillated tau neutrino fluxes from both sources are calculated .
the intrinsic galactic @xmath1 flux ( @xmath2 gev ) is calculated by considering the interactions of high - energy cosmic - rays with the matter present in our galaxy , whereas the oscillated galactic @xmath1 flux is coming from the oscillation of the galactic @xmath3 flux . for the intrinsic atmospheric @xmath1 flux ,
we extend the validity of a previous calculation from @xmath4 gev up to @xmath5 gev .
the oscillated atmospheric @xmath1 flux is , on the other hand , rather suppressed .
we find that , for @xmath6 , the oscillated @xmath1 flux along the galactic plane dominates over the maximal intrinsic atmospheric @xmath1 flux , i.e. , the flux along the horizontal direction .
we also briefly mention the presently envisaged prospects for observing these high - energy tau neutrinos . |
fig . 1.schematic diagram of the cavity model ; r indicates the radius of curvature of the mirrors . two subcavities of length @xmath33 and @xmath34 are coupled by a bs .
the total cavity is globally stable for @xmath35 .
@xmath36 represents the displacement of the bs with respect to the center of the cavity .
when a ray hits the surface of the bs , which we choose to coincide with the reference plane , it can be either reflected or transmitted with equal probability ; for a @xmath9 beam splitter @xmath8 .
2.sos for ( a ) @xmath25 : the ray dynamics is stable and thus confined on a torus in phase space .
( b ) @xmath37 m , the dynamics becomes unstable and the ray escapes after @xmath27 bounces .
note the ring structure in this plot .
4.(a ) escape time as a function of the initial condition @xmath38 .
( b ) blow up of a small interval along the horizontal axis in ( a ) .
( c ) and ( d ) blow ups of consecutive intervals along the set of impact parameters @xmath38 shown in ( b ) . | we investigate the ray dynamics in an optical cavity when a ray splitting mechanism is present . the cavity is a conventional two - mirror stable resonator and the ray splitting
is achieved by inserting an optical beam splitter perpendicular to the cavity axis .
using hamiltonian optics , we show that such a simple device presents a surprisingly rich chaotic ray dynamics .
= 0.8truecm pacs numbers : 42.60.da , 42.65.sf , 42.15.-i in this letter we present a very simple optical cavity whose ray dynamics is nevertheless fully chaotic .
our starting point is the fact that a two - mirror optical cavity can be _ stable _ or _ unstable _ depending on its geometrical configuration @xcite .
if a light ray is injected inside the cavity it will remain confined indefinitely when the configuration is stable but it will escape after a finite number of bounces when the cavity is unstable .
our interest is in a cavity which has both aspects of stability and instability ( fig [ fig:1 ] ) .
the cavity is modelled as a strip resonator @xcite made of two identical concave mirrors of radius of curvature @xmath0 separated by a distance @xmath1 , where @xmath2 so that the cavity is globally stable .
we then introduce a beam splitter ( bs ) inside the cavity , oriented perpendicular to the optical axis . in this way
the bs defines two planar - concave subcavities : one on the left and one on the right with respect to the bs , with length @xmath3 and @xmath4 , respectively .
the main idea is that depending on the position of the bs the left ( right ) subcavity becomes _ unstable _ for the reflected rays when @xmath3 ( @xmath4 ) is bigger than @xmath0 , while the cavity as a whole remains always _ stable _ ( @xmath5 ) .
consideration of this system raises the nontrivial question whether there will be an `` equilibrium '' between the number of trapped rays and escaping rays .
the trapped rays are those which bounce for infinitely long times due to the global stability of the cavity and the escaping ones are those which stay only for a finite time .
if such equilibrium exists it could eventually lead to transient chaos since it is known in literature that instability ( positive lyapunov exponents ) and mixing ( confinement inside the system ) form the skeleton of chaotic dynamics @xcite . in this letter
we show that under certain conditions such equilibrium can be achieved in our cavity and that chaotic ray dynamics is displayed + in our system the bs plays a crucial role .
it is modelled as a stochastic ray splitting element by assuming the reflection and transmission coefficients as random variables @xcite . within the context of wave optics
this model corresponds to the neglect of all interference phenomena inside the cavity , as required by the ray ( zero - wavelength ) limit .
the stochasticity is implemented by using a monte carlo method to determine whether the ray is transmitted or reflected @xcite . when a ray is incident on the ray splitting surface of the bs , it is either transmitted through it , with probability @xmath6 , or reflected with probability @xmath7 , where we assume @xmath8 for a @xmath9 beam splitter as shown in fig [ fig:1 ] .
we then dynamically evolve a ray and at each reflection we use a random number generator with a uniform distribution to randomly decide whether to reflect or transmit the incident ray .
+ in the context of hamiltonian optics , to characterize the trajectory of a ray we first choose a reference plane perpendicular to the optical axis @xmath10 , coinciding with the surface of the bs .
the intersection of a ray with this plane is specified by two parameters : the height @xmath11 above the optical axis and the angle @xmath12 between the trajectory and the same axis .
we consider the rays as point particles , as in standard billiard theory where the propagation of rays correspond to the trajectories of unit mass point particles moving freely inside the billiard and reflecting elastically at the boundary .
in particular , we study the evolution of the transversal component of the momentum of the ray , i.e. @xmath13 so that we associate a ray of light with the two - dimensional vector @xmath14 .
it is important to stress that we use exact 2d - hamiltonian optics , i.e. we do _ not _ use the paraxial approximation .
+ the evolution of a set of rays injected in the cavity with different initial conditions @xmath15 , is obtained by using a ray tracing algorithm . for each initial condition ,
the actual ray trajectory is determined by a random sequence \{ ...
rrttrrrrt .. } which specifies if the ray is reflected ( r ) or transmitted ( t ) by the bs .
when one evolves the whole set of initial conditions , one can choose between two possibilities , either use the _ same _ random sequence for all rays in the set of initial conditions or use a _ different _ random sequence for each ray . in this letter
we use the _ same _ random sequence for all injected rays in order to uncover the dynamical randomness of the cavity .
+ the three quantities that we have calculated to demonstrate the chaotic ray dynamics inside the cavity are the poincar surface of section ( sos ) , the exit basin diagrams and the escape time function @xcite . in all calculations we have assumed @xmath16 m and the radius of curvature of the mirrors @xmath17 m ; the diameter @xmath18 of the two mirrors was @xmath19 m .
in addition , the displacement @xmath20 of the bs with respect to the center of the cavity was chosen as @xmath21 m ( unless specified otherwise ) , and the time was measured in number of bounces @xmath22 .
+ in fig .
[ fig:2 ] , the successive intersections of a ray with initial transverse coordinates @xmath23 m , @xmath24 are represented by the black points in the soss . for @xmath25
the cavity configuration is symmetric and the dynamics is completely regular ( fig.[fig:2](a ) ) ; the on - axis trajectory represents an elliptic fixed point and nearby stable trajectories lie on continuous tori in phase space . in fig.[fig:2](b ) , the bs is slightly displaced from the center ( @xmath26 m ) , the same initial trajectory becomes unstable and spreads over a finite region of the phase space before escaping after a large number of bounces ( @xmath27 ) . in view of the ring structure of fig.[fig:2](b ) we may qualify the motion as azimuthally ergodic .
the fact that the ray - splitting mechanism introduced by the bs produces ergodicity is a well known result for a closed billard @xcite .
we find here an analogue phenomenon , with the difference that in our case the trajectory does not explore uniformly but only azimuthally the available phase space , as an apparent consequence of the openness of the system .
+ it is well known that chaotic hamiltonian systems with more than one exit channel exhibit irregular escape dynamics which can be displayed , e.g. , by plotting the exit basin diagram @xcite . in our system , this diagram was constructed by defining a fine grid ( @xmath28 ) of initial conditions @xmath15 . each ray is followed until it escapes from the cavity .
when it escapes from above ( @xmath29 ) we plot a black dot in the corresponding initial condition , whereas when it escapes from below ( @xmath30 ) we plot a white dot .
this is shown in fig.[fig:3 ] , the uniform regions in the exit basin diagram correspond to rays which display a regular dynamics , whereas the dusty regions correspond to portions of phase space where there is sensitivity to initial conditions , since two initially nearby points can escape from opposite exits .
moreover , in fig.[fig:3 ] one can see how the boundary between black and white regions becomes less and less smooth as one approaches the center of these diagrams .
it is known that this boundary is actually a fractal set @xcite whose convoluted appearance is a typical feature of chaotic scattering systems @xcite .
+ besides sensitivity to initial conditions , another fundamental ingredient of chaotic dynamics is the presence of infinitely long living orbits which are responsible for the mixing properties of the system .
this set of orbits is usually called repeller @xcite , and is fundamental to generate a truly chaotic scattering system . to verify the existence of this set we have calculated the escape time or time delay function @xcite for a one - dimensional set of initial conditions specified by the initial position @xmath31 ( impact parameter ) taken on the mirror @xmath32 and the initial velocity @xmath24 .
the escape time was calculated in the standard way , as the time ( in number of bounces @xmath22 ) it takes a ray to escape from the cavity .
+ fig.[fig:4](a ) shows the escape time function .
the singularities of this function are a clear signature of the existence of long living orbits and the presence of peaks followed by flat regions are a signature of the exponential sensitivity to initial conditions . in order to verify the presence of an _ infinite _ set of long living orbits , we have zoomed in on the set of impact parameters @xmath31 in three different intervals ( fig .
[ fig:4](b ) , ( c ) and ( d ) ) .
each zoom reveals the existence of new infinitely long living orbits .
infinite delay times correspond to orbits that are asymptotically close to an unstable periodic orbit .
if we would continue to increase the resolution we would find more and more infinitely trapped orbits .
the repeated existence of singular points is a signature of the mixing mechanism of the system due to the global stability of the cavity . + in conclusion , we have demonstrated that our simple optical system displays chaotic ray dynamics .
it is important to stress that a key component for the development of chaos is the inclusion of non - paraxial rays which add the mixing properties to the system @xcite .
in fact , it has been previously shown that paraxial ray dynamics can be unstable but not chaotic , in systems with stochastic perturbations @xcite . in our case , it is the stochastic ray splitting mechanism induced by the bs that destroys the regular motion of rays in the globally stable ( but non - paraxial ) cavity , as shown by the soss . moreover , by calculating the exit basin diagrams we have found that they show fractal boundaries , which is a typical feature of chaotic ray dynamics @xcite .
finally , through the singularities in the escape time function , we have verified the presence of infinitely long living orbits , which in turns reveal the mixing mechanism of our optical cavity .
an experimental confirmation of the fractal properties of the exit basin can be performed , e.g. , in the way suggested in @xcite , by injecting a narrow laser beam into the cavity either in a regular or in a dusty region of phase space . in the former case one
expects the beam to leave the cavity either from above or below , while in the latter case both exits should appear illuminated .
this proposed experiment is fully within the context of geometrical optics ( interference plays no role ) so that our stochastic model of the bs is adequate .
+ this project is part of the program of fom and is also supported by the eu under the ist - atesit contract . we thank s. oemrawsingh for useful contributions to software development .
+ a16 a. e. siegman , _ lasers _ ( university science books , mill valley , ca , 1996 ) .
p. cvitanovi _
et al . , classical and quantum chaos _
( www.nbi.bk/chaosbook/ , 2002 ) .
l. couchman , e. ott , and t. m. antonsen , jr .
, phys .
rev .
a * 46 * , 6193 ( 1992 ) .
e. ott .
_ chaos in dynamical systems _ ( cambridge university press , 2002 ) , 2nd ed .
s. bleher , c. grebogi , e. ott , and r. brown , phys .
rev . a * 38 * , 930 ( 1988 ) .
s. ree and l. e. reichl , phys . rev .
e * 65 * , 055205(r ) ( 2002 ) .
d. sweet , e. ott and j. a. yorke , nature * 399 * , 315 ( 1999 ) .
p. gaspard , _ chaos , scattering and statistical mechanics_(cambridge university press , 1998 ) , 1st ed .
s. bleher and c. grebogi and e. ott , physica d * 46 * , 87 - 121 ( 1990 ) .
a. aiello , m.p .
van exter , and j. p. woerdman , phys .
rev .
e * 68 * , 046208 ( 2003 ) .
s. longhi , phys .
rev .
e * 65 * , 027601 ( 2002 ) .
g. puentes , a. aiello , and j. p. woerdman , submitted to phys .
rev .
e ( 2003 ) . |
one of the important tasks of quantum mechanics is to find exact solutions of the wave equations ( nonrelativistic and relativistic ) for certain type of potentials of physical interest since they contain all the necessary information regarding the quantum system under consideration .
for example , the exact solutions of these wave equations are only possible in a few simple cases such as the coulomb , the harmonic oscillator , pseudoharmonic and mie - type potentials [ 1 - 8 ] . for an arbitrary @xmath0-state
, most quantum systems could be only treated by approximation methods . for the rotating morse potential
some semiclassical and/or numerical solutions have been obtained by using pekeris approximation [ 9 - 13 ] . in recent years , many authors have studied the nonrelativistic and relativistic wave equations with certain potentials for the @xmath4- and @xmath0-waves .
the exact and approximate solutions of these models have been obtained analytically [ 10 - 14 ] .
many exponential - type potentials have been solved like the morse potential [ 12,13,15 ] , the hulthn potential [ 16 - 19 ] , the pschl - teller [ 20 ] , the woods - saxon potential [ 21 - 23 ] , the kratzer - type potentials [ 12,14,24 - 27 ] , the rosen - morse - type potentials [ 28,29 ] , the manning - rosen potential [ 30 - 33 ] , generalized morse potential [ 34 ] and other multiparameter exponential - type potentials [ 35 ] .
various methods are used to obtain the exact solutions of the wave equations for this type of exponential potentials .
these methods include the supersymmetric ( susy ) and shape invariant method [ 19,36 ] , the variational [ 37 ] , the path integral approach [ 31 ] , the standard methods [ 32,33 ] , the asymptotic iteration method ( aim ) [ 38 ] , the exact quantization rule ( eqr ) [ 13,39,40 ] , the hypervirial perturbation [ 41 ] , the shifted @xmath5 expansion ( se ) [ 42 ] and the modified shifted @xmath5 expansion ( mse ) [ 43 ] , series method [ 44 ] , smooth transformation [ 45 ] , the algebraic approach [ 46 ] , the perturbative treatment [ 47,48 ] and the nikiforov - uvarov ( nu ) method [ 16,17,2026,49 - 51 ] and others .
the nu method [ 51 ] is based on solving the second - order linear differential equation by reducing to a generalized equation of hypergeometric type .
it has been used to solve the schrdinger [ 14,16,20,22,48,49 ] , dirac [ 17,28,34,50 ] , klein - gordon [ 21,24,25,50 ] wave equations for such kinds of exponential potentials .
the nu method has shown its power in calculating the exact energy levels of all bound states for some solvable quantum systems .
motivated by the considerable interest in exponential - type potentials [ 12 - 35 ] , we attempt to study the quantum properties of another exponential - type potential proposed by manning and rosen ( mr ) [ 29 - 33]@xmath6where @xmath7 and @xmath8 are two - dimensionless parameters but the screening parameter @xmath9 has dimension of length and corresponds to the potential range [ 33 ] .
this potential is used as a methematical model in the description of diatomic molecular vibrations [ 52,53 ] and it constitutes a convenient model for other physical situations .
figure 1 plots the manning - rosen potential ( 1 ) versus @xmath10 for various screening distances @xmath11 @xmath12 and @xmath13 considering the cases ( a ) @xmath14 and ( b ) @xmath15 it is known that for this potential the schrdinger equation can be solved exactly for @xmath4-wave ( i.e. , @xmath16 ) [ 32 ] .
unfortunately , for an arbitrary @xmath0-states ( @xmath17 in which the schrdinger equation does not admit an exact analytic solution . in such a case , the schrdinger equation is solved numerically [ 54 ] or approximately using approximation schemes [ 18,50,55,56,57 ] .
some authors used the approximation scheme proposed by greene and aldrich [ 18 ] to study analytically the @xmath18 bound states or scattering states of the schrdinger or even relativistic wave equations for mr potential [ 13,21 ] .
we calculate and find its @xmath18 bound state energy spectrum and normalized wave functions [ 29 - 33 ] .
the potential ( 1 ) may be further put in the following simple form@xmath19it is also used in several branches of physics for their bound states and scattering properties .
its spectra have already been calculated via schrdinger formulation [ 30 ] . in our analysis
, we find that the potential ( 1 ) remains invariant by mapping @xmath20 further , it has a relative minimum value @xmath21 at @xmath22 $ ] for @xmath23 which provides @xmath24 as a result of the first derivative @xmath25 for the case @xmath26 we have the criteria imposed on the value of @xmath7 is @xmath27 for example , in @xmath28 the minimum of the potential is @xmath29 the second derivative which determines the force constants at @xmath30 is given by@xmath31 ^{2}}{8b^{4}\alpha ^{3}(\alpha -1)^{3}}.\]]the purpose of this paper is to investigate the @xmath0-state solution of the schrdinger - mr problem within the nikiforov - uvarov method to generate accurate energy spectrum .
the solution is mainly depends on replacing the orbital centrifugal term of singularity @xmath32 [ 17 ] with greene - aldrich approximation scheme .
consisting of the exponential form [ 16 ] .
figure 2 shows the behaviour of the singular term @xmath33 and various approximation schemes recently used in refs .
[ 18,34,55,56 ] .
the paper is organized as follows : in section ii we present the shortcuts of the nu method . in section iii , we derive @xmath18 bound state eigensolutions ( energy spectrum and wave functions ) of the mr potential by means of the nu method . in section iv , we give numerical calculations for various diatomic molecules . section v , is devoted to for two special cases , namely , @xmath16 and the hulthn potential .
the concluding remarks are given in section vi .
the nikiforov - uvarov ( nu ) method is based on solving the hypergeometric type second order differential equation [ 51 ] . employing an appropriate coordinate transformation @xmath34
we may rewrite the schrdinger equation in the following form:@xmath35where @xmath36 and @xmath37 are the polynomials with at most of second - degree , and @xmath38 is a first - degree polynomial .
further , using @xmath39 eq .
( 4 ) reduces into an equation of the following hypergeometric type:@xmath40where @xmath41 ( its derivative must be negative ) and @xmath42 is a constant given in the form@xmath43it is worthwhile to note that @xmath42 or @xmath44 are obtained from a particular solution of the form @xmath45 which is a polynomial of degree @xmath46 further , @xmath47 is the hypergeometric - type function whose polynomial solutions are given by rodrigues relation@xmath48 , \]]where @xmath49 is the normalization constant and the weight function @xmath50 must satisfy the condition [ 51]@xmath51 in order to determine the weight function given in eq .
( 8) , we must obtain the following polynomial:@xmath52 in principle , the expression under the square root sign in eq .
( 9 ) can be arranged as the square of a polynomial .
this is possible only if its discriminant is zero . in this case ,
an equation for @xmath53 is obtained . after solving this equation , the obtained values of @xmath53
are included in the nu method and here there is a relationship between @xmath42 and @xmath53 by @xmath54 after this point an appropriate @xmath55 can be calculated as the solution of the differential equation:@xmath56
to study any quantum physical system characterized by the empirical potential given in eq .
( 1 ) , we solve the original @xmath57 which is given in the well known textbooks [ 1,2 ] @xmath58 where the potential @xmath59 is taken as the mr form in ( 1 ) . using the separation method with the wavefunction @xmath60
we obtain the following radial schrdinger eqauation as@xmath61 -\frac{l(l+1)}{r^{2}}\right\ } r_{nl}(r)=0,\]]since the schrdinger equation with above mr effective potential has no analytical solution for @xmath18 states@xmath62 an approximation to the centrifugal term has to be made . the good approximation for the too singular kinetic energy term @xmath63 in the centrifugal barrier is taken as [ 18,33]@xmath64 in a short potential range . to solve it by the present method
, we need to recast eq .
( 12 ) with eq .
( 13 ) into the form of eq .
( 4 ) by making change of the variables @xmath65 through the mapping function @xmath66 and energy transformation:@xmath67to obtain the following hypergeometric equation:@xmath68@xmath69 ^{2}}\left\ { -\varepsilon ^{2}+\left [ a+2\varepsilon ^{2}-l(l+1)\right ] z-\left [ a+\varepsilon ^{2}+\alpha ( \alpha -1)\right ] z^{2}\right\ } r(z)=0.\]]it is noted that the bound state ( real ) solutions of the last equation demands that@xmath70and thus provide the finite radial wave functions @xmath71 to apply the hypergeometric method ( nu ) , it is necessary to compare eq .
( 15 ) with eq .
subsequently , the following value for the parameters in eq .
( 4 ) are obtained as @xmath72 z^{2}+\left [ a+2\varepsilon ^{2}-l(l+1)\right ] z-\varepsilon ^{2}.\]]if one inserts these values of parameters into eq .
( 9 ) , with @xmath73 the following linear function is achieved@xmath74where @xmath75 , $ ] @xmath76 \right\ } $ ] and @xmath77 according to this method the expression in the square root has to be set equal to zero , that is , @xmath78 thus the constant @xmath53 can be determined as@xmath79 in view of that , we can find four possible functions for @xmath80 as@xmath81we must select@xmath82 in order to obtain the polynomial , @xmath41 having negative derivative as@xmath83we can also write the values of @xmath84 and @xmath85 @xmath86 as@xmath87 , \]]@xmath88respectively . letting @xmath89 and solving the resulting equation for @xmath90 leads to the energy equation@xmath91from which we obtain the discrete energy spectrum formula:@xmath92 ^{2},\text { \ } 0\leq n , l<\infty\]]where @xmath1 denotes the radial quantum number .
it is found that @xmath93 remains invariant by mapping @xmath94 so do the bound state energies @xmath95 an important quantity of interest for the mr potential is the critical coupling constant @xmath96 which is that value of @xmath97 for which the binding energy of the level in question becomes zero .
furthermore , from eq .
( 26 ) , we have ( in atomic units @xmath98@xmath99 next , we turn to the radial wave function calculations .
we use @xmath36 and @xmath80 in eq ( 17 ) and eq .
( 21 ) to obtain@xmath100and weight function@xmath101@xmath102 .\]]the functions @xmath103 , up to a numerical factor , are in the form ofjacobi polynomials , i.e. , @xmath104 and physically holds in the interval @xmath105 @xmath106 @xmath107 [ 58 ] .
therefore , the radial part of the wave functions can be found by substituting eq .
( 28 ) and eq .
( 30 ) into @xmath108 as@xmath109where @xmath90 and @xmath93 are given in eqs .
( 14 ) and ( 19 ) and @xmath110 is a normalization constant .
this equation satisfies the requirements ; @xmath111 as @xmath112 @xmath113 and @xmath114 as @xmath115 @xmath116 therefore , the wave functions , @xmath117 in eq . ( 31 ) is valid physically in the closed interval @xmath118 $ ] or @xmath119 further , the wave functions satisfy the normalization condition:@xmath120where @xmath121 can be determined via@xmath122 ^{2}dz.\]]the jacobi polynomials , @xmath123 can be explicitly written in two different ways [ 59,60]::@xmath124 @xmath125 where @xmath126 after using eqs . ( 34 ) and ( 35 ) , we obtain the explicit expressions for @xmath127@xmath128 @xmath129 @xmath130 inserting eqs .
( 36 ) and ( 37 ) into eq .
( 33 ) , one obtains@xmath131@xmath132where@xmath133using the following integral representation of the hypergeometric function [ 59.60]@xmath134 @xmath135 which gives@xmath136where@xmath137@xmath138for the present case , with the aid of eq .
( 40 ) , when @xmath139 @xmath140 and @xmath141 are substituted into eq .
( 41)@xmath62 we obtain@xmath142finally , we obtain@xmath131@xmath143which gives@xmath144where@xmath145@xmath146
to show the accuracy of our results , we calculate the energy eigenvalues for various @xmath1 and @xmath0 quantum numbers with two different values of the parameters @xmath3 its shown in table 1 , the present approximately numerical results are not in a good agreement when long potential range ( small values of parameter @xmath9 ) .
the energy eigenvalues for short potential range ( large values of parameter @xmath9 ) are in agreement with the other authors .
the energy spectra for various diatomic molecules like @xmath147 and @xmath148 are presented in tables 2 and 3 .
these results are relevant to atomic physics [ 61 - 64 ] , molecular physics [ 65,66 ] and chemical physics [ 67,68 ] , etc .
in this work , we have utilized the hypergeometric method and solved the radial @xmath57 for the m - r model potential with the angular momentum @xmath149 states@xmath150 we have derived the binding energy spectra in eq . ( 26 ) and their corresponding wave functions in eq . ( 31 ) .
let us study special cases .
we have shown that for @xmath151 @xmath152 , the present solution reduces to the one of the hulthn potential [ 16,19,57]:@xmath153where @xmath154 is the potential strength parameter and @xmath155 is the screening parameter and @xmath9 is the range of potential .
we note also that it is possible to recover the yukawa potential by letting @xmath156 and @xmath157 if the potential is used for atoms , the @xmath158 is identified with the atomic number .
this can be achieved by setting @xmath159 hence , the energy for @xmath18 states@xmath160 ^{2}\hbar ^{2}}{8\mu b^{2}(n+l+1)^{2}},\text { \ } 0\leq n , l<\infty .\]]and for @xmath4-wave ( @xmath161 states@xmath162 ^{2}\hbar ^{2}}{8\mu b^{2}(n+1)^{2}},% \text { \ } 0\leq n<\infty\]]essentially , these results coincide with those obtained by the feynman integral method [ 31,56 ] and the standard way [ 32,33 ] , respectively . furthermore , if taking @xmath163 and identifying @xmath164 as @xmath165 we are able to obtain@xmath166 ^{2},\]]which coincides with those of refs . [ 16,19 ] . further
, we have ( in atomic units @xmath167@xmath168 ^{2},\]]which coincides with refs .
[ 16,33 ] .
the corresponding radial wave functions are expressed as@xmath169where@xmath170 , \text { } 0\leq n , l<\infty , \]]which coincides for the ground state with that given in eq .
( 6 ) by gnl _
in addition , for @xmath171 ( i.e. , @xmath172 the hulthn potential turns to become a coulomb potential : @xmath173 with energy levels and wave functions:@xmath174 @xmath175 where @xmath176 @xmath177 and @xmath178 is bohr radius for the hydrogen atom .
the wave functions are@xmath179 r^{l+1}p_{n}^{\left ( \frac{2\mu ze^{2}}{\hbar ^{2}\delta ( n+l+1)},2l+1\right ) } ( 1 + 2\delta r)\]]which coincide with refs .
[ 3,16,22 ] .
in this work approximately analytical bound states for the @xmath0-wave schrdinger equationwith the mr potential have been presented by making a proper approximation to the too singular orbital centrifugal term @xmath180 the normalized radial wave functions of @xmath0-wave bound states associated with the mr potential are obtained .
the approach enables one to find the @xmath0-dependent solutions and the corresponding energy eigenvalues for different screening parameters of the mr potential .
we have shown that for @xmath181 the present solution reduces to the one of the hulthn potential .
we note that it is possible to recover the yukawa potential by letting @xmath156 and @xmath157 the hulthn potential behaves like the coulomb potential near the origin ( i.e. , @xmath182 ) @xmath183 but decreases exponentially in the asymptotic region when @xmath184 so its capacity for bound states is smaller than the coulomb potential [ 16 ] . obviously , the results are in good agreement with those obtained by other methods for short potential range , small @xmath8 and @xmath185 we have also studied two special cases for @xmath186 @xmath149 and hulthn potential .
the results we have ended up show that the nu method constitute a reliable alternative way in solving the exponential potentials .
we have also found that the criteria for the choice of parameter @xmath7 requires that @xmath7 satisfies the inequality @xmath187 this means that for real bound state solutions @xmath7 should be chosen properly in our numerical calculations .
a slight difference in the approximations of the numerical energy spectrum of schrdinger - mr problem is found in refs .
[ 55,56 ] and present work since the approximation schemes are different by a small shift @xmath188 in our recent work [ 17 ] , we have found that the physical quantities like the energy spectrum are critically dependent on the behavior of the system near the singularity ( @xmath189 ) .
that is why , for example , the energy spectrum depends strongly on the angular momentum @xmath0 , which results from the @xmath33 singularity of the orbital term , even for high excited states .
it is found that the @xmath190 orbital term is too singular , then the validity of all such approximations is limited only to very few of the lowest energy states . in this case , to extend accuracy to higher energy states one may attempt to utilize the full advantage of the unique features of schrdinger equation .
therefore , it is more fruitful to perform the analytic approximation of the less singularity @xmath191 rather than the too singular term @xmath33 which makes it possible to extend the validity of the results to higher excitation levels giving better analytic approximation for a wider energy spectrum [ 69 ] .
ikhdair , chin .
46 ( 3 ) ( 2008 ) 291 ; s.m .
ikhdair and r. sever , int .
c 18 ( 10 ) ( 2007 ) 1571 ; int .
c 19 ( 2 ) ( 2008 ) 221 ; s.m .
ikhdair and r. sever , centr .
j. phys . 5 ( 4 ) ( 2007 ) 516 ; s.m .
ikhdair and r. sever , cent .
j. phys . 6 ( 3 ) ( 2008 ) 697 .
ikhdair and r. sever _ _ , _ _ int .
phys . a 24 ( 2009 ) 5341 ; z. phys .
c 56 ( 1992 ) 155 ; z. phys .
c 58 ( 1993 ) 153 ; z. phys .
d 28 ( 1993 ) 1 ; int
a 18 ( 2003 ) 4215 ; int .
a 19 ( 2004 ) 1771 ; int .
phys . a 20 ( 2005 ) 4035 ; int .
phys . a 20 ( 2005 ) 6509 ; int . j. mod
. phys . a 21 ( 2006 ) 2191 ; int .
phys . a 21 ( 2006 ) 3989 ; int . j. mod
. phys . a 21 ( 2006 ) 6699 ; int .
e 17 ( 4 ) ( 2008 ) 669 .
ikhdair and r. sever , j. math .
chem 41 ( 2007 ) 329 ; 41 ( 2007 ) 343 ; s.m .
ikhdair and r. sever , j. molec . struc .
( theochem ) 809 ( 2007 ) 103 ; s.m .
ikhdair and r. sever , int .
phys . a 21 ( 2006 ) 6465 ; c. berkdemir , a. berkdemir and r. sever , j. phys . a : math .
39 ( 2006 ) 13455 .
ikhdair , j. math .
52 ( 2011 ) 052303 ; s.m .
ikhdair and r. sever , j. phys . : math .
44 ( 2011 ) 345301 ; s.m .
ikhdair , c. berkdemir and r. sever , appl .
comput . 217
( 2011 ) 9019 ; s.m .
ikhdair , phys .
83 ( 2011 ) 015010 . | the approximately analytical bound state solutions of the @xmath0-wave schrdinger equation for the manning - rosen ( mr ) potential are carried out by a proper approximation to the centrifugal term .
the energy spectrum formula and normalized wave functions expressed in terms of the jacobi polynomials are both obtained for the application of the nikiforov - uvarov ( nu ) method to the manning - rosen potential . to show the accuracy of our results , we calculate the eigenvalues numerically for arbitrary quantum numbers @xmath1 and @xmath2 with two different values of the potential parameter @xmath3
it is found that our results are in good agreement with the those obtained by other methods for short potential range , small @xmath0 and @xmath3 two special cases are investigated like the @xmath4-wave case and hulthn potential case .
keywords : bound states ; manning - rosen potential ; nu method . |
the dynamics of dark matter ( dm ) halos in the cold dark matter ( cdm ) cosmology can be easily formulated as the classical newtonian @xmath13-body problem . yet
, the understanding of the equilibrium configuration of the dm halos defies a simple analytical approach .
the lack of analytical understanding is often compensated for by numerical simulations which provide empirical knowledge .
the cumulative work in cosmology over the last decade or so has led to a broad consensus about two basic facts that describe the equilibrium structure of dm halos .
one is that the spherically - averaged density profile @xmath14 is well approximated by the so - called nfw profile ( @xcite @xcite , @xcite ) or some close variants of it ( @xcite @xcite ; @xcite @xcite ; @xcite @xcite ) .
the other is the power law behavior of the phase - space density profile , namely @xmath15 , with @xmath16 , where @xmath17 is the velocity dispersion ( @xcite @xcite ) .
two seemingly orthogonal approaches to the problem of the origin of the equilibrium structure in dm halos exist .
one assumes a monolithic collapse of a halo that can be approximated by the spherical infall model @xcite .
the application of the model to the cosmological context , where the shell crossing has to be explicitly accounted for , has resulted in the so - called secondary infall model ( sim ; @xcite @xcite ; @xcite @xcite ; @xcite @xcite ; @xcite @xcite ; @xcite @xcite ; @xcite @xcite ; @xcite @xcite ; @xcite @xcite ) .
the sim has been tested against the @xmath13-body simulations and has proven to faithfully reproduce the density profile of simulated dm halos ( @xcite @xcite ; @xcite @xcite ; @xcite @xcite ; @xcite @xcite ; @xcite ) .
a closely related variant of the sim replaces its dependence on the primordial over - density of the proto - halo by the mass accretion history ( mah ) of the halo ( @xcite @xcite ; @xcite @xcite ; @xcite @xcite ) .
the sim and its mah variant can reproduce also the power law behavior of the ( @xcite @xcite ; @xcite @xcite ) .
however , a close inspection of the @xmath13-body simulations reveals that a dm halo evolves very differently from a monolithic quasi - spherical collapse .
in fact halos are numerically observed to go through a sequence of mergers , some labeled as major mergers in which the two main progenitors are of a similar mass , leading to emergence of the nfw density profile ( @xcite @xcite ; @xcite @xcite ; @xcite @xcite ) .
@xcite ( @xcite ; @xcite , hereafter paper i and ii ) studied the formation and equilibrium configuration of halos by means of controlled @xmath13-body simulations , with the initial conditions set by constrained realizations of gaussian fields .
these simulations were designed to address issues of how the merging history affects the dm halos .
the emerging picture is that of a halo evolving _ via _ a sequence of quiescent phases of a slow mass accretion intermitted by violent episodes of major mergers . in the quiescent phases ,
the density is well fitted by an nfw profile , the inner ( nfw ) scale radius @xmath18 and the mass enclosed within it ( @xmath19 ) remain constant , and the virial radius ( @xmath20 ) grows linearly with the expansion parameter ( @xmath21 ) . in the violent phases , the halos are not in a dynamical equilibrium , but are rather in a transition state , resulting in a discontinuous growth of @xmath18 and @xmath20 .
in such a picture a halo is defined in the context of a merger tree at any given time it is taken as the most massive progenitor along the branch leading to the final halo .
a direct comparison between the sim and numerical simulations has been conducted recently by @xcite .
this comparison is based on selecting the dm halos from a cosmological simulation , tracing them back in time , and recovering their individual initial conditions .
the sim has been applied to the ` primordial ' density profiles , and their virial density profiles have been calculated for different redshifts .
the sim calculated profiles provided a good match to the evolution and structure of the simulated clusters .
this is encouraging .
however , a rigorous fundamental theory that can accommodate both the spherical monolithic collapse and the major merger - driven evolution exhibited by the simulations is still missing . this motivates us to look further into the phenomenology of the , to gain a further insight into this seemingly simple , yet complicated , problem .
the study of the dm halo evolution has been heavily focused on the density profile , while the evolution of the has been largely ignored .
@xcite presented one of the few studies of @xmath0 evolution .
they found that @xmath4 , defined as a global quantity characterizing the halo as a whole , is generally decreasing with time .
specifically , it exhibits an early rapid decrease ( at redshifts @xmath22 ) and a late slow decrease . here
we aim at studying the evolution of the within the framework of the nfw scaling and the dynamical picture formulated in papers i and ii .
in particular , we shall rely on the empirical fact that the dynamics of the dm halos is constrained by a new invariant of motion .
this invariant tags a halo along the main branch of the merging tree .
the nfw scaling and this invariant of motion provide a full description of the cosmological evolution of the for individual halos .
the structure of the paper is as follows .
the analysis of the dm halos in papers i and ii is briefly reviewed in
[ sec : numerics ] .
general considerations of the evolution of dm halos are given in[sec : general ] .
the general evolution of the dm profile is described in [ sec : results ] and self - similarity and scaling relations are given in [ sec : scaling ] .
a general discussion follows in [ sec : disc ] .
in papers i and ii , we investigated the cosmological evolution and structure of five dm halos by setting the initial conditions of the simulations using constrained realizations of gaussian fields .
our basic motivation was to perform controlled numerical experiments designed to study the dependence of the evolution and structure of a given halo on its merging history .
the @xcite algorithm of constrained realizations of gaussian fields has been used to set up the initial conditions .
papers i and ii present the analysis of five different models of a given dm halo , evolving along various merging histories . the models were simulated within the framework of the open cdm ( ocdm ) model with @xmath23 , @xmath24 and @xmath25 , where @xmath26 is the current cosmological matter density parameter and @xmath27 is the variance of the density field convolved with a top - hat window of radius @xmath28 used to normalize the power spectrum .
this model is very close to the ` concordance ' @xmath29cdm model in its dynamical properties .
the models are labeled as ocdma , ocdmb , etc .
, following the notations of papers i and ii . here
we add a new model , run within the flat-@xmath29 cosmology with the parameters of the wmap three years data base @xcite .
the constraints used to set the model are similar to ocdma and the model is labeled as wmap3a , yet it is performed within a @xmath30 computational box and it starts from a different random realization of the initial conditions . a full description of the numerical simulations , the application of the algorithm of constrained realizations and the numerical code are given in papers i and ii .
the wmap3a model is one in a series of runs in the wmap3 cosmology , to be reported in a forthcoming paper .
the nfw fitting algorithm is described in paper ii and the power law profile is fitted in a similar way .
the density profile of dm halos is well approximated by the nfw profile , @xmath31 in which the characteristic density ( @xmath32 ) and scale radius ( @xmath6 ) define the nfw profile . defining the halo as a collection of particles in a spherical ( say ) volume in which the mean density equals some critical over - density ( which is in general redshift dependent ) times the mean cosmological density , the virial mass ( @xmath33 ) and radius ( @xmath34 ) of the halo are determined .
it follows that @xmath20and @xmath18 ( or equivalently @xmath35 and @xmath19 ) are the two independent parameters that define an nfw halo . assuming the nfw parameterization , we can write the profile as @xmath36 where @xmath37 ^ 3},\ ] ] @xmath38 @xmath39 is the mean velocity dispersion within @xmath6 and @xmath40 . to the extent that the dm halos are fitted by the nfw profile , their @xmath41 profile should obey a universal relation .
their cosmological evolution is then determined by the evolution of @xmath5 .
the cosmological evolution of the main halo of the six different models is best presented by fig .
[ fig : evol ] , which shows the evolution of @xmath6 and @xmath5 .
the halo goes through violent episodes of major mergers and quiescent phases of slow accretion . in the ocdm models
the violent events are well separated by the quiescent phases characterized by an nfw structure ( papers i and ii ) .
the wmap3 model goes through an early phase of successive violent mergers , frequent enough so that the halo does not relax to an nfw - like configuration in between .
this early phase is followed by a quiescent phase , which is slightly perturbed by mergers not strong enough to be qualified as major .
this is clearly shown by the behavior @xmath6 and @xmath5 , where @xmath6 ( @xmath5 ) increases ( decreases ) discontinuously in the violent phases and remains constant in the quiescent ones .
one should note that in the violent episodes the halos are not in an equilibrium and therefore the nfw fitting is very unstable and the resulting @xmath6 and @xmath5 parameters are quite erratic .
this is reflected by the spiky behavior of these quantities in the violent phases .
evolution of the dimensionless profile , @xmath41 , is presented in fig .
[ fig : calq ] , where the @xmath42 profiles of the ocdmc and wmap3 models are evaluated at different epochs , covering the time interval from @xmath43 ( @xmath44 ) for the ocdmc ( wma3 ) model to the present epoch .
the evolution of the other models is virtually identical to the ones shown here .
the @xmath41 profiles are very closely approximated by a power law , with a fractional deviation of less than a twenty percent ( lower panel of fig .
[ fig : calq ] ) . fig .
[ fig : calq ] shows not only the power law nature of the profiles but also that indeed the @xmath5 scaling renders the @xmath41 profile to a universal time independent power law .
the evolution of the exponents of the @xmath41 fitted power law of all the models is shown in fig .
[ fig : beta ] .
@xmath45 displays a bumpy behavior , with bumps corresponding to the major mergers , but generally staying within the @xmath46 range .
overall , given the violent character of the halo s evolution , the robustness of the @xmath45 range is remarkable .
a close inspection of the halo evolution reveals that the product @xmath47 remains approximately invariant as the halo evolves along the main branch of its merging tree ( fig .
[ fig : rhosrs ] ) .
the value of @xmath48 fluctuates around its mean value in two different modes .. undergoes fast and correlated fluctuations of a small amplitude ( see next section and paper ii ) . in the violent episodes it exhibits large deviations from the mean value .
this behavior is associated with timings of major mergers when halos are far from the equilibrium .
the nfw fitting fails here and the resulting nfw parameters are ill - defined . ignoring the spikes of the violent phases , and averaging over the jittery fluctuations in the quiescent phases , @xmath48 retains its value along the halo evolution .
_ while the invariance of _ @xmath48 _ in a given quiescent phase is not surprising , its ability to retain the value in and across the major merger events is not obviously expected . _
assuming the empirical finding of @xmath48 invariance , we shall study its ramifications for the evolution of the phase - space density . in the quiescent phases , in which the density follows the nfw profile ,
a halo is expected to be in virial equilibrium , as corrobrated by fig .
13 of paper ii .
in particular the virial ratio , evaluated within @xmath18 , should have a constant value , namely , @xmath49 the virial ratio differs from unity because the inner part of the halo does not constitute an isolated system , and its value depends on the shape of the density profile .
assuming the constancy of the virial ratio within @xmath6 and the invariance of @xmath48 , @xmath50 the evolution of the phase - space density profile in a given halo is described by @xmath51 consequently , for a given halo the following invariance holds : @xmath52 this prediction has been tested against the five models of papers i and ii and the @xmath29cdm halo .
[ fig : qsra ] shows the cosmological evolution of @xmath53 of these models . as with all other quantities that characterize the dm halos ,
the product @xmath54 shows a jittery behavior in the quiescent phases and strong fluctuations in the violent phases , in which the nfw parameters are ill defined .
apart from this , it remains constant throughout the evolution .
only model c exhibits a small deviation from this invariance . in its first quiescent phase @xmath55 is larger by a factor of @xmath56 than its asymptotic value .
the @xmath57 invariance is to be distinguished from the @xmath58 scaling found in its the low - amplitude jitter ( paper ii ) and from @xmath59 of @xcite . in paper
ii we show that @xmath60 in the last quiescent phases of various halos , while at early times @xmath61 .
this stands in good agreement with @xcite s @xmath62 in the `` slow accretion phase '' and @xmath63 in the `` rapid accretion phase . ''
the @xmath60 scaling is obtained by analyzing each quiescent phase , and it reflects the fluctuations of @xmath32 and @xmath18around their mean values within that phase .
this correlation appear to be driven by density fluctuations that originate in the region between the cusp and @xmath6 , where the density slope varies between -1 and -2 .
the mean values of @xmath32 and @xmath18 change from one quiescent phase to another .
the @xmath64 reflects the variation across the different phases .
the association of the @xmath64 and @xmath63 with the early times of the evolution of halos stems from the fact that the violent phases are more abundant at early times .
the combined analysis of paper ii and the present work yields the following picture . in the quiescent phases
the values of @xmath32 and @xmath18 fluctuate around constant values , yet their product remains invariant along the evolution of a halo .
the @xmath48 invariance implies that @xmath65 .
this is very close to the @xmath66 of @xcite and the minor discrepancy results from @xcite not separating explicitly between the quiescent and violent phases .
we , therefore , support and validate one of @xcite s main points , namely the @xmath59 scaling , and set @xmath67 .
thus , we validate also @xcite s relation between the evolution of the concentration parameter and the mah .
moreover , this relation can be easily extended to the mah @xmath5 relation , given the @xmath55 invariance .
a final note concerns the entropy of a system of self - gravitating collisionless particles .
the definition of the entropy ( per particle ) of a monoatomic ideal gas is givein by @xmath68 where @xmath69 is the boltzmann constant ( @xcite @xcite ; @xcite @xcite ) . applying this definition to the dm particles provides one with a formal entropy of the dm .
our findings concerning structure and evolution of the can be easily translated to the language of the entropy of the dm .
it should be noted here that eq .
[ eq : s ] provides a local measure of the entropy .
the long range nature of the gravitational interactions prohibits a simple extension of the entropy to an extensive quantity that characterizes the whole halo . within this framework
we can only refer to the entropy as a local property .
the main elements of the structure and evolution of the dm halos can be summarized as follows .
halos evolve through two phases , quiescent and violent ones , which represent the two extreme cases of smooth accretion and major mergers . in the quiescent phase the halo density distribution is closely approximated by the nfw profile .
the inner ( within @xmath6 ) halo mass surface density , @xmath48 , remains approximately constant throughout its entire evolution .
most importantly , the major mergers that take the halo from one quiescent phase to the other preserve the value of @xmath48 . during the quiescent phases the halo phase - space density profile follows a power law of the form @xmath3 with @xmath16 .
the cosmological evolution of the phase space density is given by @xmath5 .
under the invariance of @xmath48 and the assumption of the virial equilibrium within @xmath6 , the evolution of @xmath5 is dictated by @xmath6 , so that @xmath70 . in the quiescent phases ,
@xmath5 remains constant and it decreases discontinuously in the violent phases . the @xmath48 invariance and the virial theorem ( eq . [ eq : const ] ) imply that the evolution of a halo proceeds while conserving its surface density and its energy per unit volume , or equivalently the pressure , within @xmath6 .
the interesting point is that a typical halo undergoes a few violent events of major merging that destroy its equilibrium .
following each event it regains the nfw structure with a larger @xmath6 , while preserving its pressure and the mean surface density within the new @xmath6 . summarizing ,
a dm halo evolves along a sequence of nfw profiles , with an ever increasing @xmath34 and @xmath6 that grows only discontinuously , in the manner described in paper i , while conserving the mean surface density and the pressure within @xmath6 .
the evolution of the @xmath0 profile of a given dm halo is predicted by eq .
[ eq : q - evol ] .
the controlled numerical experiments of papers i and ii suggest that , for a given halo , @xmath48 remains constant over a few quiescent phases intermitted by violent events .
all the halos analyzed here ( apart from wmap3a ) emerge from the same realization of the initial conditions that has been subjected to different constraints . in many ways these halos can be considered as a single halo that has been manipulated so as to modify its merging history . as such they all have roughly the same value of @xmath48 .
the wmap3a model provides an independent realization of a dm halo in a different cosmology .
the value of @xmath48 is roughly twice as large than in all the other model .
of course the extremely poor statistics of our models can not teach us about the scatter in @xmath48 .
the question arises as to what happens in the general case of a large ensemble of dm halos drawn from a large cosmological simulation .
the scatter in @xmath48 is expected to determine the evolutionary tracks of the phase - spase density profiles .
it is interesting to study the possible environment @xmath71 correlation , and to what extent this affects the evolution of the dm halos .
the @xmath57 invariance implies that the @xcite @xmath72 scaling holds with @xmath67 , and so does the mah - concentration parameter relation .
it follows that this can be easily extended to calculate a similar mah @xmath73 relation .
the interesting ramification of this result is that any statistical algorithm for generating dm halos merging trees and/or mahs can be extended so as to provide the evolution of the full nfw parameters of a halo along its merging tree .
this can be used in semi - analytical modeling of galaxy formation in which realizations of merging trees include the full nfw structure of the halos .
the definition of @xmath0 as a phase - space density is only a poor man s substitute for the real phase - space density in the six - dimensional phase space . yet , pushing the analogy further on , a formal entropy is defined by eq .
[ eq : s ] as a local variable .
this local definition can not be extended to provide a global entropy of the halo .
it follows that statements regarding the evolution of the entropy of dm halos are ill - defined , at least in the current context . nevertheless , in the nfw scaling framework , in which the entropy of dm halos is taken as @xmath5 , it increases with @xmath6 . in this context , the quiescent phases with @xmath74const .
correspond to adiabatic processes which preserve the entropy , and the violent phases to non - adiabatic ones in which the entropy grows .
the method of introducing an entropy by means of eq .
[ eq : s ] can be considered as only formal , and its relation to a standard thermodynamics needs to be questioned .
yet , @xcite have recently shown that the formally defined entropy of the dm is very closely associated with the classical ( ideal gas ) entropy of the intergalactic gas in clusters of galaxies .
@xcite studied the entropy of the gas and dm in high resolution adiabatic sph cosmological simulations ` adiabatic ' here is used in the sense that the entropy of the gas can only grow due to shock waves .
consequently , following the accepted notation , the entropy of the gas can be expressed by means of @xmath75 , and by analogy @xmath76 .
the power law behavior of @xmath0 implies that @xmath77 .
interestingly , outside the entropy core of @xmath78 the gas entropy follows the dm , namely @xmath79 , and the calculated slope agrees well with observations .
faltenbacher amended the definition of @xmath80 , such that the thermal energy is extended to include the kinetic energy of the small scale ( turbulent motion ) and the dm and gas densities are normalized by their mean cosmological values .
the newly defined @xmath80 then coincides with the dm entropy outside the entropy core to a very good approximation .
the analysis of faltenbacher shows that , at least in galaxy clusters , the dm entropy coincides with the classical , ideal gas , entropy of the intracluster medium .
it is not clear a priori why the gas and dm entropies should coincide , but numerical simulations suggest that they do .
the simulations suggest that observational determination of the entropy of the intracluster gas should shed light and constrain the behavior of the dm entropy .
we thank ran rubin for his help with the calculation of the phase space density profile .
fruitful discussions with yuval birnboim , adi nusser , noam soker and saleem zaroubi are gratefully acknowledged .
this research has been supported by isf-143/02 and the sheinborn foundation ( to yh ) , by nasa / ltsa 5 - 13063 , nasa / atp nag5 - 10823 , hst / ar-10284 ( to is ) , and by nsf / ast 02 - 06251 ( to ch and is ) .
, d. n. , bean , r. , dor , o. , nolta , m. r. , bennett , c. l. , dunkley , j. , hinshaw , g. , jarosik , n. , komatsu , e. , page , l. , peiris , h. v. , verde , l. , halpern , m. , hill , r. s. , kogut , a. , limon , m. , meyer , s. s. , odegard , n. , tucker , g. s. , weiland , j. l. , wollack , e. , & wright , e. l. 2006 , arxiv astrophysics e - prints | the evolution of the phase - space density profile in dark matter ( dm ) halos is investigated by means of constrained simulations , designed to control the merging history of a given dm halo .
halos evolve through a series of quiescent phases of a slow accretion intermitted by violent events of major mergers . in the quiescent phases
the density of the halo closely follows the nfw profile and the phase - space density profile , @xmath0 , is given by the taylor & navarro power law , @xmath1 , where @xmath2 and stays remarkably stable over the hubble time . expressing the phase - space density by the nfw parameters , @xmath3 , the evolution of @xmath4
is determined by @xmath5 .
we have found that the effective mass surface density within @xmath6 , @xmath7 , remains constant throughout the evolution of a given dm halo along the main branch of its merging tree .
this invariance entails that @xmath8 and @xmath9 .
it follows that the phase - space density remains constant , in the sense of @xmath10 , in the quiescent phases and it decreases as @xmath11 in the violent ones .
the physical origin of the nfw density profile and the phase - space density power law is still unknown .
yet , the numerical experiments show that halos recover these relations after the violent phases .
the major mergers drive @xmath6 to increase and @xmath5 to decrease discontinuously while keeping @xmath12 .
the virial equilibrium in the quiescent phases implies that a dm halos evolves along a sequence of nfw profiles with constant energy per unit volume ( i.e. , pressure ) within @xmath6 . |
periodic variable stars , in particular eclipsing binary ( eb ) star systems , continue to be of central importance to a variety of applications in stellar astrophysics .
ebs are crucial astrophysical benchmarks for stellar evolution theory as they can provide fundamental physical parameters of stars with very high precision and accuracy .
fundamental issues , from the stellar mass - radius relation using ebs in nearby clusters and the field ( e.g. , * ? ? ?
* ) , to the extragalactic distance scale using ebs in the large magellanic @xcite and small magellanic cloud @xcite , rely on the discovery and analysis of statistically significant samples of ebs and/or of rare but astrophysically important ebs .
indeed , ebs in which at least one component is of very low mass , in a scarce evolutionary phase , or itself a periodic variable are astrophysically very interesting and useful , yet rare and thus only small numbers of these benchmark systems are found in current catalogs .
for example , there is only one known eb system comprising two brown dwarfs @xcite , and only a couple ebs containing a classical cepheid pulsators are currently known ( e.g. , * ? ? ? * ) .
repositories of large time - series photometric survey data , such as the _ kepler _ archive , provide an ideal sample from which to identify periodic variables including these interesting ebs for focused followup investigation .
however , constraints on the rate of human analysis make mining this data using classical techniques prohibitive . an intelligent data pipeline ( idp ) that contains an automated light curve classifier to replace
the human bottleneck is then necessary for the future pace of large survey data rates .
ideally such an idp would be capable of automatically finding ebs in light curve data sets and then accurately classifying these ebs by morphological type ( e.g. , eclipsing contact , eclipsing semi - detached , eclipsing detached ) to enable focused followup analyses .
several algorithms have been developed for this purpose ( e.g. , @xcite these are discussed further in sec .
[ dis ] ) , yet a fully - automated pipeline remains an open task .
an automated classification pipeline with parameters adaptable to multiple time series photometric surveys would be immediately applicable to the _ kepler _ data set @xcite as well as the follow - on k2 mission @xcite and would be well suited to the upcoming transiting exoplanet survey satellite ( tess ) @xcite .
in addition , the development of the large synoptic survey telescope ( lsst ) contains a requirement that data processing enable a fast and efficient response to transient sources ( i.e. , automated identification of variable stars and astrophysically interesting binaries ) with a robust and accurate preliminary classification @xcite . in short
, there is a large need currently and into the near future for methods that can reliably process large quantities of light curve data and automatically identify astrophysical objects of interest such as ebs with minimal human intervention .
so motivated , especially the immediate applicability of an automated pipeline for the study of ebs in k2 @xcite , in this paper we present a fully automated , adaptive , end - to - end computational pipeline the eclipsing binary factory ( ebf )
. the ebf first identifies ebs , and then morphologically classifies them as either eclipsing contact ( ec ) , eclipsing semi - detached ( esd ) , or eclipsing detached ( ed ) .
the classification module of the ebf pipeline was described in the first paper in this series @xcite ; here we add the remaining computational modules for a complete automated pipeline . in this paper
we also then measure the accuracy and completeness of our ebf generated catalog of _ kepler _ ebs against the manually constructed _ kepler _ eb catalog @xcite .
the ebf takes a modular approach to automatically process large volumes of survey data into patterns suitable for recognition by an integrated , fully - trained artificial neural network ( ann ) .
this modular pipeline is designed to accept as input the direct archival data from a time - series photometric survey where each module s parameters are adaptable to the specific characteristics of the data ( e.g. , photometric precision , data collection cadence , flux measurement uncertainty ) and define the output options to produce the ann generated catalog of eb classifications ( i.e. , ec , esd , or ed ) . the computational efficiency of the ann classifier , @xmath2 where @xmath3 is the number of neurons in the neural network , is quite fast of order @xmath4 seconds for classification of @xmath5 light curves using a standard workstation .
this then alleviates the current bottleneck by limiting followup human analysis to just the light curves automatically classified as ebs .
in addition , the portability of this modular software pipeline makes the ebf a solution for the automated probabilistic identification and morphological classification of ebs in other large repositories of time - series , photometric data currently underway or planned for the near future .
the _ kepler _ mission monitored @xmath6 targets brighter than the magnitude limit @xmath7 in the _ kepler _ spacecraft s field of view ( fov ) .
_ kepler _ captured and summed these images as long cadence , 29.4 @xmath8 , data files @xcite that contain one quarter , or @xmath9 90 @xmath10 , of observations across the fov for sixteen quarters until reaction wheel failure placed the instrument into safe mode .
the complete scientific data set , archived in mast at the space telescope science institute , provides simple aperture photometry ( sap ) with the photometric precision necessary to detect planetary transits @xcite and astero - seismological signals @xcite from solar - like stars ; about 20 @xmath11 for @xmath12 magnitude g2v stars for a 6.5 @xmath13 integration @xcite . as such , this data set provides both the appropriate sample size and precision for use in testing the ebf pipeline .
in addition , the _ kepler _ eclipsing binary catalog version 3 ( kebc3 ) of @xmath0 manually classified ebs and a corresponding catalog of manually identified false positives in the _ kepler _ fov @xcite provides a benchmark of the ebf s classification accuracy ( i.e. , false positive rate ) and completeness ( i.e. , false negative rate ) .
moreover , full processing of the kepler dataset with the ebf automated pipeline provides the opportunity to uncover some additional ebs that may have been missed by the human generated kebc3 .
we demonstrate the efficacy of the ebf by automatically identifying and morphologically classifying ebs in the @xmath14 _ kepler _ quarter 3 ( `` q3 '' ) sap light curves . in section
[ met ] of this paper we describe the procedures and methods of each ebf module .
section [ res ] presents the results of the ebf applied to the kepler `` q3 '' light curves .
section [ resp ] benchmarks the pre - classification data processing modules against the metrics of the manually corrected and phased light curves in the kebc3 while section [ resc ] focuses on the ebf pipeline s false positives and false negatives as a measure of classification accuracy and completeness .
finally , in section [ dis ] we discuss the future expansion of the ebf to include the automated identification and classification of other periodic variable stars such as @xmath15 cepheid fundamental mode ( dcep - fu ) and first overtones ( dcep - fo ) , @xmath15 scuti ( dsct ) , rr lyrae @xmath16 ( rrab ) and @xmath17 ( rrc ) , and mira ( mira ) as well the ebf s applicability to other large time - series , photometric surveys .
a set of newly identified candidate ebs that may have been missed by the human generated _ kepler _ eb catalog is provided in appendix [ app ] .
we conclude with a brief summary in section [ con ] .
the ebf pipeline , diagrammed in figure [ figure1 ] , is a compilation of five fully automated , adaptable software modules that process light curves from large astronomical surveys in order to automatically identify ebs and classify the each eb s morphology . in this section
we describe the parameters with which systematics are removed , periodic variable stars are identified , phased light curves are represented as ann recognizable patterns , and constrain how eb candidate classifications are tuned and validated as a function of the survey instrument s photometric precision , the data set duration and cadence , the flux measurement uncertainty , and the desired level of classification confidence , respectively .
the procedure of each ebf module is described below in turn , then we test the overall pipeline performance by the rate of correctly classified systems and the completeness of the ebf classified sample as compared to the kebc3 benchmark sample classified by human analysis .
we use the _ kepler _
q3 " long cadence data files as described in the _ kepler _ data release 4 notes ksci-199044 - 001 @xcite , to tool and test the ebf pipeline .
corrections to the raw photometry collected from large survey instruments are generally necessary to remove systematic artifacts from the data . though most surveys include their own data correction processes , there is no guarantee that the method employed preserves the necessary signal from which the ebf will identify and classify ebs . in the case of the _ kepler _ `` q3 '' light curves corrections to the sap flux
are necessary to remove systematics such as differential velocity aberration , thermal gradients across the spacecraft , and pointing variations @xcite . of course , the _ kepler _ archive data reduction pipeline removes systematics in the `` q3 '' sap data to produce pre - search conditioned data ( pcd ) using the method of cotrending basis vectors ; however , pcd may lack the astrophysical signatures of variables whose harmonic thresholds were not met @xcite . hence , we employ a data correction module ( dcm ) to remove trending artifacts while preserving the necessary signal as well as to normalize the flux in order to present the remainder of the pipeline with a standardized , detrended light curve .
the dcm step is applied to all of the input light curves ( i.e. , no light curves are filtered out at this stage ) .
the dcm accepts as input the time - series flux with the associated uncertainty of each data point and detrends the flux via the sigma - clipping algorithm of @xcite ; where data points outside a standard deviation interval of a least squares _
polynomial fit to order @xmath18 , expressed by the generating function @xmath19,\ ] ] are iteratively discarded until there are no remaining data points outside the interval . for each data point input ,
the dcm returns a point containing a normalized flux value with a new uncertainty .
the ebf s implementation of this method allows the parameters of the module to be set to the specific characteristics of a data set by defining the sigma clipping thresholds and _ legendre _ polynomial order .
additionally , the module allows for the input of the time intervals of known breaks in the light curves due to instrument operations . in this paper
we detrend the _ kepler _ `` q3 '' sap light curves using asymmetric sigma - clipping thresholds of @xmath20 with the @xmath21 order _ legendre _ polynomial @xmath22 and known breaks in the `` q3 '' light curves beginning at bjd 281.0 , 291.0 , and 322.5 .
figure [ figure2 ] shows the uncorrected _ kepler _ `` q3 '' sap light curve for a target and the ebf detrended , normalized light curve where systematics were removed by the dcm and the red circles indicate data gaps given by the aforementioned breaks . the ebf pipeline is currently optimized for identification and classification of ebs , which exhibit periodically varying light curves .
therefore the precursors to morphological classification of ebs via pattern recognition are the identification of a target s periodic variability and a subsequent phase folded light curve .
hence , the detrended normalized light curve is next processed by the harmonics recognition module ( hrm ) where periodic variability is probed by the analysis of variance ( aov ) method @xcite included in the vartools1.202 package @xcite .
the hrm takes as input all of the normalized light curves output by the preceding dcm , but outputs only those light curves that are deemed to exhibit truly periodic behavior ( of any kind , i.e. , the output light curves from the hrm stage include ebs but also any other type of periodic variable ) .
two separate searches for the first three harmonics , each with a coarse sub - sample setting of 0.1 and a fine - tune setting of 0.01 , are conducted over the entire data set .
the first search separates the data into five bins and the second uses fifty bins such that skewed harmonics from bin size related flux variance is minimized .
we then compare the strongest harmonic signal from both periodograms and , in the event of disagreement between the primary periods , assign the period ( along with all other associated aov statistics ) from the periodogram with the smallest false alarm probability ( fap ) .
to eliminate light curves that are unlikely to exhibit truly significant periodic variability , we use the aov calculated signal to noise ratio ( snr ) .
specifically , periodograms with a snr @xmath23 100 are rejected by the module while the remaining light curves , now identified as candidate periodic variables , are accepted .
these accepted candidates light curves are phased by the standard equation @xmath24 where @xmath25 is the phase of an associated time value , @xmath26 , for a given reference time , @xmath27 , and the assigned period , @xmath28 .
for an eb , we want the reference or zero phase @xmath29 , where @xmath30 , to be associated with the primary eclipse .
however , not all of the periodic variables identified by the hrm will be ebs ; there will be other types of periodic variables where the most physically meaningful reference phase is not an eclipse but rather a flux maximum .
therefore , to maintain generality , we associate @xmath29 with either the minimum or maximum flux , dependent upon whether the light curve spends most of its time above or below the median flux .
most true ebs will have flux points in the
high " state and will spend a minority of the time in eclipse . here
the determination is made by the value of the flux ratio ( fr ) as described by @xcite : @xmath31 by definition , the fr has a value on the interval ( 0,1 ) . for reference ,
a perfect sinusoid will have fr = 0.5 .
we assign @xmath29 to the flux minimum , indicative of an eb , for a fr less than a defined threshold , and a flux maximum assigned at @xmath29 otherwise .
the hrm s parameters that are adaptable to the data set are the period search interval , based on the duration of the survey data , and the fr threshold .
the latter is based on the tolerance for ecs , which in general are not perfect sinusoids and whose fr may be greater than 0.5 if they spend more than half the time above the median flux . as we limited the _
input files to `` q3 '' ( @xmath990 @xmath10 ) , the harmonics search was limited to a period range of 0.11 to 20.1 @xmath10 in order to ensure at least three primary eclipses would be present in each light curve .
in addition , we set the fr threshold to 0.7 based on a simple analysis of kebc3 frs in order to include those ecs whose light curves are not perfectly sinusoidal .
figure [ figure3 ] shows a plot of the fr distribution in the kebc3 versus cataloged morphology , and a `` q3 '' target as an example of an imperfect sinusoidal phased ec light curve with a @xmath32 .
an example of a phased esd light curve , and an example of a phased ed light curve , each with a @xmath33 , are shown in figure [ figure4 ] .
as described in section [ ann ] , the ebf s classification of the phased light curves for the candidate periodic variables output from the hrm is dependent upon the ann evaluating these light curves as linearly separable patterns . in its simplest form
linearly separable means two sets of points in a plane may be divided by at least one line such that each data point belonging to one set is on one side of the line and each data point belonging to the second set is on the other side of that line . when generalized to an n - dimensional space , linear separability
is formalized such that every point @xmath34 satisfies @xmath35 and every point @xmath36 satisfies @xmath37 , where @xmath38 and @xmath39 are two sets of points in an @xmath40dimensional _ euclidean _ space and @xmath41 is the @xmath42 component of @xmath43 for @xmath44 real numbers @xmath45 , @xmath46 ... , @xmath47 , @xmath48 .
the pattern generation module ( pgm ) therefore represents the phased light curves via piecewise smooth polynomial chains @xmath43 to pass to the ann as linearly separable patterns .
since ebf efficiency goes as @xmath49 where @xmath3 is the number of elements in @xmath43 , it becomes computationally expensive to process light curves possessing on the order of @xmath50 data points .
therefore , we condense the light curve by phase binning down to the order @xmath51 data points over the phase interval @xmath52 , with the primary eclipse centered at @xmath29 .
in addition to meeting the efficiency requirements , the phase binning process also minimizes @xmath29 error by smoothing over uncorrected systematics that may present a stray minimum flux below the true primary eclipse depth .
the pgm therefore fits the phase binned curves to piecewise smooth @xmath53 order polynomial chains using the ` polyfit ` algorithm of @xcite .
however , the ann s use of the sigmoid activation function ( see sec .
[ ann ] ) leads to errors in processing large polynomial coefficients that may arise during polynomial fitting .
hence , we modify the ` polyfit ` algorithm to fit polynomial chains @xmath43 with components @xmath41 described by three points in a @xmath54dimensional _ euclidean _ space ( phase , flux ) .
these three data points represent the chain s two end - points , or knots , and the midpoint . here
each chain shares one knot with the next , thus a single chain is effectively described by two points in a plane .
we add the additional modification of allowing either a four - polynomial chain @xmath55 or two - polynomial chain @xmath56 fit based on a @xmath57 goodness of fit criterion . with these modifications the pgm processes phased light curves initially containing on order @xmath58 data points into at most four linearly separable patterns @xmath43 where @xmath59 for each @xmath60 .
that is , we reduce the recognizable pattern of the phase folded light curve which may contain a very large number of data points to a maximum of sixteen points . bin size selection is an adjustable parameter .
this option allows for a condensed light curve based on the survey cadence and/or duration in order to manage the point density per bin .
of course , more bins lead to longer run times for the pgm and fewer bins lead to sparse light curves where the narrow eclipses of wide ebs might be smoothed over or missed .
the minimum number of points per bin may also be adjusted in an attempt to capture narrow eclipses , however from trial and error we find that no fewer than 50 phase bins and no fewer than three bins per chain is optimal . to fit the _ kepler _ `` q3 '' phased light curves we bin each light curve s @xmath95,000 long cadence data points into 200 ( size @xmath61 ) phase bins with a polynomial chain that contains no fewer than three data points .
the ebf generated binned light curves and linearly separable patterns , plotted as the knots and midpoints of the module s fitted flux , as examples of _ kepler _ ebs with the morphology ec , esd , ed are shown in figure [ figure5 ] , figure [ figure6 ] , and figure [ figure7 ] , respectively . the ebf s trained , validated , and tested ann is the single hidden layer feed forward perceptron network diagrammed in fig .
[ figure8 ] .
this module takes as input the pgm output set of data points as a simple representation of the target light curve .
these patterned inputs are recognized by the ann and classified into the eb categories that it has been trained to recognize .
the outputs of the ann are therefore the classification of each light curve with an associated confidence level in the classification .
a detailed description of the ann , including its training , validation , and output classifications , is described in @xcite . here
we provide a summary of its specifications and functions .
the ann functions as a two - stage pattern recognition classifier based on nonlinear statistical models . the input layer , nodes @xmath62 to @xmath63 , are populated with the pgm fitted polynomial chains @xmath43 as an input vector of @xmath64 components that is then propagated via weighted connections @xmath65 to the hidden layer .
the activation function @xmath66 is in the form of the sigmoid @xmath67 where @xmath68 is the variable used by the ridge function to derive the features @xmath69 to @xmath70 as linear combinations of the inputs .
the signals are again weighted by @xmath71 and propagated to the output layer where there are @xmath72 target classifications coded as a 0 to 1 categorical target variable @xmath68 for the @xmath73class classification using the same sigmoid activation function to model the target nodes @xmath74 to @xmath75 as linear combinations of @xmath69 to @xmath70 ; where @xmath76 , @xmath77 , and @xmath78 .
in other words , the ann accepts @xmath79 input parameters representing the 16 polynomial chain parameters from the pgm as well as the logarithm of the period in days , the total @xmath57 as a measure of fit , and the maximal amplitude of the light curve s normalized flux and outputs a probability associated with each of the @xmath80 known light curve classifications .
the 10 known classifications are the three eb categories ( ec , esd , ed ) , the other variable types ( dcep - fu , dcep - fo , dsct , rrab , rrc , mira ) , and a miscellaneous classification ( misc ) .
training of the ann was conducted by back - propagation , and gradient descent was used to minimize the error in the set of all weights @xmath81 , or @xmath82 , where the error function is given by @xmath83 for @xmath84 and @xmath85 .
@xmath86 is described by the softmax function @xmath87 as @xmath88 , and was used to approximate the nonlinear input functions . as described in @xcite , training and validation of the ann was done using 32,278 light curves from the automated all - sky survey ( asas ) @xcite .
these tests showed the ann is capable of an accuracy of 90.6% , 76.5% , and 90.3% , and a completeness of 95.8% , 63.7% , and 87.8% for ecs , esds , and eds , respectively . in summary
, for each input patterned light curve from the pgm , the ann outputs a most likely classification and an associated confidence in that classification . the output classifications for ebs can be either ec , esd , or ed .
because the ann was originally developed on the asas sample @xcite , there is also a classification of misc for light curves that are not clearly identified as one of the three eb classes .
in addition , the ann outputs up to 9 additional possible classifications ( and associated confidence levels ) for each light curve . as we describe below ,
this permits us to identify possible ebs at lower confidence that , e.g. , have a primary classification of misc but secondary classification of ec , esd , or ed .
once ann classifications are complete , candidate ec , esd , and ed light curves are filtered for false positives by the candidate validation module ( cvm ) .
we position the cvm as a post classification process to provide the option to examine some or all of the filtered candidates .
the cvm filters candidates by two distinct metrics .
the first is the ` polyfit ` algorithm s total goodness of fit , @xmath89 , such that @xmath90 where @xmath91 is the normalized , detrended flux of phase bin @xmath18 and @xmath92 is the @xmath42 component of the @xmath93 polynomial chain @xmath43 for @xmath94 that satisfies the requirement of linear separability from section [ pgm ] .
the @xmath89 metric describes how well the pgm represents the candidate light curves and is a useful measure of whether the ann s recognition of the ebf generated pattern translates to an accurate classification as an eb .
the second metric is the normalized , detrended light curve s scaled sum of all second derivatives for the portion of the curve between each and every data point , which we refer to as @xmath95 and defined as @xmath96 where @xmath97 and @xmath98 are the phase and flux for the @xmath42 of @xmath64 data points . as such , @xmath95 is a measure of the dispersion in the detrended light curve and when compared to the average detrended flux measurement uncertainty calculated in the dcm provides a criterion for filtering a pseudo - periodic signal that may arise from random noise .
the cvm adapts the @xmath89 and @xmath95 filter metrics by using @xmath99 known ebs contained in the survey .
the module then calculates the upper limits on both the @xmath89 and @xmath95 from this benchmark subset .
a cvm analysis of the kebc3 ebs contained in the _ kepler _ `` q3 '' data set and recovered by the ann provides a @xmath89 upper limit of 100 and a @xmath95 upper limit of 1,200 as shown by the histograms in figure [ figure9 ] .
figure [ figure10 ] shows the light curves of two example _ kepler _ targets where the pgm fits to some quasi - systematic artifact that the ann consequently mis - classifies as an esd and where the pgm fits to a pseudo - sinusoidal signal generated by random noise that the ann consequently mis - classifies as an ec .
these examples demonstrate the cvm determined criteria of @xmath100 and @xmath101 filtering false positives from the automated eb catalog .
notice that a cvm filter with the criteria of only @xmath100 would have passed the target , with @xmath89 of only 0.01 , yet a @xmath95 of 4,752 to the ebf catalog .
similarly , a cvm criteria of only @xmath101 would have passed the target , with a @xmath95 of 1,165 yet a , @xmath89 of 3,378 to the ebf catalog .
thus , we set the cvm cut - off parameters to @xmath102 and @xmath103 in order to minimize eb classification false positives .
we use the 2,612 ebs cataloged in the kebc3 to benchmark the ebf pipeline outputs as an initial assessment of pipeline capabilities on standard morphology ecs , esds , and eds .
thus , we first remove from the benchmark sample those kebc3 ebs flagged as eclipse timing variations ( etv ) induced by third body candidates @xcite , heart beat ( hb ) , and those classified in the kebc3 as uncertain ( unc ) .
we also exclude those kebc3 ebs with cataloged periods outside the _ kepler _ `` q3 '' duration parameters of @xmath104 period @xmath105 and of course those cataloged ebs that do not appear in the _ kepler _ `` q3 '' data set .
thus , we benchmark the ebf pipeline against an adjusted kebc3 sample of 1,198 cataloged ebs .
we compare the metrics of the ebf pipeline s pre - classification output to the same metrics of the manually determined kebc3 .
note that the kebc3 light curves contain @xmath5 data points from more than ten quarters of observations while the `` q3 '' observations include only @xmath106 .
we evaluate the dcm s performance by comparing the 200 phase binned , normalized flux values of each common target .
we find the dcm s performance to be in excellent agreement , mean @xmath107 , with the kebc3 benchmark .
this dcm light curve agreement breaks down to a mean @xmath57 of 0.00373 , 0.00264 , 0.00052 for cataloged ecs , esds , and eds respectively .
the hrm assigned period agrees to within @xmath108 for 99.67% of the ebf identified periodic variable candidates found in the kebc3 , though the hrm failed to identify statistically significant periodicity in 5.59% of the benchmark sample , and therefore were not considered for follow on classification by the ann .
a further comparison of the hrm assigned periods to those manually determined in the kebc3 reveals a period alias rate of 21.65% , ( i.e. , the percentage of ebf determined periods that are an alias of the kebc3 period ) .
however , aliasing does not negatively impact ann pattern recognition and the true period is easily determined by examining the candidate light curve .
figure [ figure11 ] shows examples of both the ebf generated and the manually determined kebc3 light curves for targets where the ebf assigns a half - period alias .
it was also discovered that phase disagreement was common for targets with a @xmath32 .
though there is no adverse effect on pattern recognition for the example targeted ec in figure [ figure12 ] , the ambiguity in phasing a light curve with an fr of @xmath109 on the minimum flux led to diminished pipeline accuracy as other variable types mistaken by the pipeline for ebs ( e.i . , @xmath110cepheids incorrectly phased on the flux minimum ) contributed to the false positive rate .
the pgm fitted the ebf phased , detrended light curves with either four- or two - polynomial chains and a goodness of fit described by a median @xmath89 of 5.39 as compared to the unaltered ` polyfit ` algorithm used to fit the kebc3 samples with a median @xmath89 of 3.74 .
this goodness of fit includes cases where the pgm patterned fit is in excellent agreement with the manually determined kebc3 light curves , such as the target shown in figure [ figure13 ] , as well as instances where the patterned fit fails to capture to full depth of the eclipses as circled in figure [ figure14 ] .
the subtle deviation between a phased light curve and an ebf generated pattern may present a challenge to an automated solution estimator , yet the pgm s representation of the light curves shape allows the ann to successfully recover and classify these systems .
this and other deviations , such as the small aberrant @xmath111 chain circled in red on figure [ figure11 ] , are the result of the _ three data points per chain minimum _
parameter we adopted in the pgm in order to capture narrow eclipses . though @xmath910% of the non - recovered kebc3 benchmark sample is comprised of wide eds with narrow
eclipses that contain less than three binned data points , the three data point rule remains a hard minimum to ensure pattern generation integrity .
it is important to note that the pgm failed to pattern 3.33% of the benchmark sample and that these light curves were not considered for follow on recognition by the ann .
the performance of the ann for cataloged binaries in the _ kepler _ `` q3 '' data set is benchmarked by kebc3 ebs with period @xmath28 where @xmath104 @xmath28 @xmath112 while etvs , hbs , and uncs were excluded .
the ann reports each of its classifications with a posterior probability @xmath113 interpreted as a level of confidence ( loc ) for the morphological class of ec , esd , ed , or any other periodic variable class contained in the asas training set @xcite . before discussing the performance of the ebf pipeline as a whole
, we consider the ann performance alone .
this is evaluated only for those light curves passed forward by the dcm , hrm , and pgm such that the ann performance metrics are not hindered by the exclusion of light curves by the preceding modules .
the ann s classification accuracy metric is calculated using the kebc3 targets flagged in the kebc3 database as incat : false ( icf ) , a false positive eb ( e.g. , blends , planets , pulsators , rotators ) .
thus we only select those targets designated icf that lack a cataloged morphology ( e.g. , pulsators , rotators ) as an inaccurate classification .
that is , the ann s accuracy is given as the percentage of ann ec , esd , ed classifications that are _ kepler _ targets currently cataloged in the kebc3 as an eb as opposed to those currently cataloged as a pulsating or rotating variable .
the ann s completeness metric is simply the percentage of the adjusted kebc3 benchmark sample that the ann successfully recovered as an eb . using the _ kepler _
`` q3 '' data set , the ann accuracy is found to be 93.68% , 84.27% , and 83.03% with a completeness of 94.17% , 93.62% , and 93.63% for ec , esd , and ed respectively .
the cvm s performance was evaluated by a manual inspection of the 91 high - confidence ( @xmath114 ) eb candidate light curves generated by the ebf . here we remove from these candidates 21 _ kepler _ `` q3 '' targets that suffer from the phasing ambiguity discussed in section [ resp ] .
this manual inspection of the remaining 68 candidates found that the high - confidence cvm validated ebs are likely reliable to @xmath115% accuracy , commensurate with the integrated ebf pipeline s accuracy as discussed below . to evaluate the performance of the integrated ebf pipeline
, we describe the ebf generated catalog for candidate ecs , esds , and eds with two distinct queries in order to evaluate ebf accuracy and completeness in the _ kepler _ fov both at high - confidence ( @xmath114 ) and with no regard to confidence ( @xmath116 ) . here , _ no regard to confidence _
means either the primary ann classification is ec , esd , ed , or the primary classification is the placeholder misc with a secondary classification of ec , esd , or ed with any posterior probability on the interval ( 0,1 ) .
the accuracy and completeness for each condition are listed in table [ table1 ] .
lcrrrrrr query & & & ec & esd & ed & + @xmath114 & for & @xmath117 & & & & + & & accuracy ( % ) & 96.12 & 95.53 & 91.93 & + & & completeness ( % ) & 63.93 & 45.80 & 32.40 & + @xmath116 & for & @xmath117 & & & & + & & accuracy ( % ) & 93.68 & 84.27 & 83.03 & + & & completeness ( % ) & 86.07 & 73.95 & 61.59 & + these two queries serve to exemplify the range of utility in the ebf s automated output of candidate _
for example , a high - confidence ( @xmath114 ) query for ecs in the interest of analyzing the characteristics of a homogeneous sample of _ kepler _ ecs , with a high certainty of morphological classification , recovers @xmath118 of those available in the `` q3 '' data set with only @xmath119 false positives .
conversely , a query with no regard to confidence ( @xmath116 ) in the interest of a statistical analysis using a large number of _ kepler _ ecs , without regard to the certainty of their morphology ( @xmath116 ) , recovers @xmath120 of those in the `` q3 '' data with @xmath121 false positives .
in addition , for the high - confidence condition we find 68 cvm validated eb candidates that do not appear in the kebc3 catalog .
the set of these newly identified candidates are provided in appendix [ app ] . in general , these newly identified candidates are in many cases visually more complex and less obviously identified as ebs via manual visual inspection of the light curves .
this is not surprising , as the ebs previously cataloged in the kebc3 were by definition those ebs that could be readily identified by visual human analysis , and therefore these new candidates are only those that would have been most likely to be missed by human visual inspection .
nonetheless , visual inspection of these new candidates identified in an automated manner by the ebf indicates that most if not all of these are bona fide ebs . in any case
, these new candidates will be examined further in a follow - on paper .
previous works have developed aspects of an automated eb classifier , although to date a fully automated end - to - end pipeline has remained an open challenge .
for example , the eb identification algorithm of @xcite automatically identified candidates in the optical gravitational lensing experiment ( ogle ) , but then relied on visual inspection of all candidate light curves for manual classification . similarly , the semi - automated method of @xcite relied on visual inspection to manually remove artifacts from the ogle - iii candidate light curves . though the pipeline of @xcite automatically filtered @xmath122 of the trans - atlantic exoplanet survey ( tres ) data as non - periodic , a manual removal of @xmath123 of the remaining light curves was necessary to finalize the catalog .
more advanced algorithms described by @xcite , such as the eclipsing binaries via artificial intelligence ( ebai ) method @xcite , used sophisticated anns to estimate the characteristic parameters of detached ebs in the _ kepler _ field ; however , the classification of eb morphological sub - classes remained largely a manual task .
in contrast to the algorithm of @xcite , where the authors are only able to estimate that their eb catalog is complete to @xmath124 with an efficiency of @xmath125 , and where the classification of eb sub - classes is achieved only through visual inspection , the ebf s completeness and accuracy rates are precisely benchmarked by the kebc3 and based upon the ann s fully - automated classification of sub - classes . additionally , the ebf s single - layer ann represents a much simpler representation of the classifier algorithm with a more robust training set ( i.e. , @xmath126 input nodes and trained on @xmath127 distinct asas exemplars ) .
the @xcite method requires the conversion of ogle photometric light curves into representational pixel images that are then analyzed by a @xmath128 input node three - layer ann trained on variations of only @xmath129 ogle eb exemplars .
the ogle - iii analysis of @xcite is based on a @xmath130 step algorithm that , when complete , reports a false positive rate of @xmath131 and thus requires a manual inspection of each and every eb candidate light curve .
in contrast , the ebf s automated filter reduces false positives to less than @xmath132 .
the semi - automatic method of @xcite also requires manual classification of the eb sub - classes and is further constrained to the analysis of ed and esd tres light curves while the ebf extends the result to include the ec class .
the ebf pipeline again represents an advance in full automation and adds the ability to automatically classify the ec sub - class when compared to more advanced solution estimators , such as the ebai method @xcite .
the purpose of the ebf pipeline is to alleviate the human bottleneck in discovering astrophysically interesting eclipsing binary systems by automatically identifying and classifying these systems from large volumes of survey data as a foundation for an idp designed for the future of astroinformatics .
the performance results using the _ kepler _ `` q3 '' data release demonstrate the ebf s accuracy for those ebs flagged at high confidence , however the completeness metric leaves room for improvement .
an examination of the pipeline s unclassified kebc3 ebs reflects the 5.59% exclusion rate of the aov - snr output filter , as well as the 3.33% failure rate of the pgm .
here we find that 8.92% of the benchmark sample never reaches the ann .
in addition , we discover that a simple evaluation of the flux ratio in relation to an arbitrary threshold of @xmath133 is insufficient to discriminate between ecs and other variable types .
thus , the necessity for the pipeline to first reliably identify harmonics as well as correctly phase and fit the input data without manual adjustment is obviously an area in need of refinement .
these period and subsequent phasing errors can potentially be mitigated in the future by layering the hrm with multiple harmonic search methods for varied period length .
for example , the computationally inexpensive fast chi squared ( f@xmath57 ) method of @xcite and the boxed least squares ( bls ) transit search algorithm @xcite may supplement the aov method .
this multi - layered harmonics search could be combined with a revised pgm where a post - fitting @xmath57 analysis may be used to detect errors and validate assigned periods . phasing
may then be subjected to other constraints ( e.g. , the symmetry vs asymmetry of the primary eclipse ingress and egress for those targets with fr @xmath133 ) .
training deficits such as the lack of a sufficient number of wide eb light curves with narrow transits in the asas exemplars as well as an inherent morphological ambiguity in binary systems can potentially be rectified with further training , testing , and validation using expanded survey data sets .
the ebf s adaptability , facilitated by the modular parameters described in section [ met ] , and the automated catalog s query options , demonstrates the pipeline s utility for a variety of applications . applying the ebf to other current time - domain photometric surveys , such as k2 and
the upcoming tess mission should provide similar results .
in addition , the ebf s ann is currently tooled and trained to recognize and classify the other types of periodic variables discussed in @xcite ( i.e. , the asas periodic variable classifications dcep - fu , dcep - fo , dsct , rrab , rrc , and mira ) .
only a benchmark of variables in the _ kepler _ fov , parallel to the kebc3 , is needed to tool and test the ebf s expansion into automatically cataloging these and other types of periodic variables .
the eclipsing binary factory ( ebf ) generated a fully automated catalog of eclipsing binary systems in the _ kepler _ field of view from the _ kepler _ `` q3 '' data release .
this catalog , benchmarked by the manually generated _ kepler _ eclipsing binary catalog v3 , identified and classified to 90% certainty eclipsing contact ( ec ) , eclipsing semi - detached ( esd ) , and eclipsing detached ( ed ) systems ( @xmath117 ) with an accuracy of 96.12% , 95.53% , and 91.93% , while complete to 63.93% , 45.80% , and 32.40% respectively . when classification certainty is not considered , the ebf catalog identified and classified ecs , esds , and eds ( @xmath117 ) with an accuracy of 93.68% , 84.27% , and 91.93% , while complete to 86.07% , 73.95% , and 61.59% respectively . in addition , the ebf identified 68 new candidate ebs that may have been missed by the human generated kebc3 ( see appendix [ app ] ) .
this demonstrates the ebf pipeline as an effective alternative to the constraints of human analysis when probing large time - domain , photometric data sets . as such
, the ebf computational pipeline is a viable framework for the development of a more complex intelligent data pipeline containing a follow on automated solution estimator that would enable the automatic determination of physical eb parameters .
future work on the ebf will include enhancing the harmonics identification capabilities with multiple search methods and phasing constraints beyond the flux ratio , a revised pattern generator to constrain polynomial chain fits to smaller errors , and more robust training for the ann on wide binary systems .
in addition , an expansion of ebf searches in the _ kepler _ data , and in the upcoming k2 data , to include the periodic variables of type @xmath110cepheid ( fundamental and first overtones ) , @xmath110scuti , rr - lyrae ( @xmath16 and @xmath17 ) , and mira will be published in follow - up papers . over the long term
, the ebf general classification approach can be adopted by future surveys such as tess and lsst .
we are grateful to a. pra , k. conroy , and all those at villanova university who contribute to the _ kepler _ eclipsing binary catalog , to j. pepper for his insights and experience with large surveys , and to r. siverd for his expert assistance with harmonics .
kepler was selected as the @xmath21 mission of the discovery program .
funding for this mission is provided by nasa , science mission directorate .
christiansen , j. l. , jenkins , j. m. , caldwell , d. a. , et al .
2013 , _ kepler data characteristics handbook _
( ksci-19040 - 004 ) nasa ames research center , moett field , ca , http://archive.stsci.edu/kepler/manuals conroy , k. , pra , a. , stassun , k. g. , et al .
2013 , , 147 , 45 coughlin , j. , lpez - morales , m. , harrison , t. e. , ule , n. , and hoffman , d. i. 2011 , , 141 , 78 devinney , e. j. , pra , a. , guinan , e. f. , et al .
2012 , from interacting binaries to exoplanets : essential modeling tools .
proceedings of the international astronomical union , symposium no .
doi:10.1017/s1743921311027712 devor , j. 2008 , , 135 , 850 devor , j. 2005 , , 628 , 411 gilliland , r. l. , brown , t. m. , christensen - dalsgaard , j. , et al .
2010 , , 122 , 131 graczyk , d. , pietrzyski , g. , thompson , i. b. , et al .
2014 , , 780 , 59 graczyk , d. , soszyski , r. , poleski , g. , et al .
2011 , , 61 , 103 hartman , j. d. , gaudi , b. s. , holman , m. j. , et al .
2008 , , 675 , 1233 howell , s. b. , sobeck , c. , haas , m. , et al .
2014 , , 126 , 398 ivezi , z. , tyson , j. a. , acosta , e. , et al .
2011 , arxiv:0805.2366v2 jenkins , j. , caldwell , d. a. , chandrasekaran , h. , et al .
2010 , , 713 , l160 kinemuchi , k. , barclay , t. , fanelli , m. , et al .
2012 , , 124 , 919 koch , d. , borucki , w. j. , basri , g. , et al .
2010 , , 713 , l79 kovcs , g. , zucker , s. , and mazeh , t. 2002 , , 391 , 369 lacy , c. h. 1977 , , 218 , 444 mcquillan , a. , aigrain , s. , and roberts , s. 2012 , , 539 , a137 paczyski , b. , szczygiel , d. m. , pilecki , b. , and pojmaski , g. 2006 , , 368 , 1311 paegert m. , stassun , k. g. , and burger , d. m. 2014 , , 148 , 31 palmer , d. 2009 , , 695 , 496 pepper j. , stanek , k. z. , pogge , r. w. , et al .
2008 , , 135 , 907 pietrzyski , g. , thompson , i. b. , gieren , w. , et al . 2010 , , 468 , 542 pietrzyski , g. , graczyk , d. , gieren , w. , et al .
2013 , , 492 , 76 pojmaski g. 1997 , , 47 , 467 pra , a. , guinan , e. f. , devinney , e. j. , et al .
2008 , , 687 , 542 pra a. , batalha , n. , slawson , r. w. , et al .
2011 , , 141 , 83 , http://keplerebs.villanova.edu ricker g.r . ,
winn , j. n. , vanderspek , r. , et al .
2014 , arxiv:1406.0151v1 schwarzenberg - czerny , a. 1989 , , 241 , 153 slawson r. , pra , a. , welsh , w. f. , et al .
2011 , , 142 , 160 stassun , k. g. , mathieu , r. d. , & valenti , j. a. 2007 , , 664 , 1154 stassun , k. g. , mathieu , r. d. , & valenti , j. a. 2006 , , 440 , 311 van cleve , j. , and caldwell , d. 2009 , _ kepler instrument handbook ( ksci-10933 - 001 ) _ , nasa ames research center , moett field , ca , http://archive.stsci.edu/kepler/manuals van cleve , j. 2010 , kepler data release 4 notes ( ksci-10944 - 001 ) , nasa ames research center , moett field , ca , http://archive.stsci.edu/kepler/release_notes/release_notes4/ welsh , w. , bloemen , s. , conroy , k. , et al . 2013 , arxiv:1309.1176 wyrzykowski , l. , udalski , a. , kubiak , m. , et al . 2003 , , 53 , 1
in table [ table2 ] , we provide the kic , ra , dec , period , and _ kepler _ magnitude for 68 high - confidence ( @xmath134 ) , astrophysically interesting eb candidates newly identified by the ebf pipeline from the kepler `` q3 '' light curves that are not currently included in the human generated _ kepler _ eclipsing binary catalog v3 .
figures [ figure15][figure17 ] represent a sample of `` q3 '' candidate light curves .
the list of candidates also includes three targets with a current disposition of `` candidate '' in the _ kepler _ objects of interest ( koi ) .
figure [ figure18 ] shows the light curves for koi k00188.01 , k00830.01 , and k00183.01 .
1872192 & 292.29273 & 37.32021 & 0.669979 & 13.74 + 2164791 & 292.427369 & 37.556019 & 6.703367 & 15.22 + 2285420 & 286.776414 & 37.63955 & 5.347988 & 13.5 + 2696217 & 286.90421 & 37.90544 & 12.807157 & 13.45 + 3441414 & 290.84053 & 38.54963 & 8.114992 & 11.51 + 3761175 & 295.0293 & 38.89788 & 5.162696 & 12.83 + 3831911 & 285.05514 & 38.98501 & 7.339436 & 15.02 + 3836276 & 286.88712 & 38.90174 & 8.176842 & 15.32 + 3847822 & 290.738025 & 38.971723 & 7.256877 & 11.85 + 4164363 & 293.210621 & 39.2407 & 5.791975 & 11.37 + 4175707 & 295.698821 & 39.26685 & 3.746223 & 15.02 + 4271063 & 293.72196 & 39.38955 & 0.989814 & 14.11 + 4451525 & 287.60961 & 39.59948 & 16.264671 & 13.88 + 4926962 & 292.68917 & 40.04848 & 4.463288 & 13.13 + 5000179 & 288.35829 & 40.17466 & 3.632168 & 13.76 + 5083330 & 286.319321 & 40.23009 & 7.814972 & 15.73 + 5083911 & 286.585518 & 40.228043 & 12.133361 & 12.96 + 5357901 & 290.35802 & 40.56774 & 3.797303 & 14.74 + 5358624 & 290.58173 & 40.5774 & 3.525784 & 15.22 + 5391520 & 298.830296 & 40.507433 & 3.984854 & 11.93 + 5857714 & 284.69325 & 41.16556 & 0.163569 & 11.21 + 6207355 & 292.59219 & 41.53943 & 5.210091 & 13.46 + 6425891 & 285.25158 & 41.8777 & 0.951153 & 14.77 + 6507888 & 286.7031 & 41.99742 & 7.453959 & 15.27 + 6613627 & 293.97651 & 42.04829 & 0.226224 & 12.55 + 6616211 & 294.67262 & 42.01478 & 3.85509 & 11.79 + 6924320 & 282.394541 & 42.44624 & 16.505877 & 15.49 + 6953069 & 293.20455 & 42.48449 & 0.294107 & 15.74 + 7053456 & 297.03561 & 42.53181 & 11.211459 & 12.0 + 7219251 & 296.87916 & 42.70233 & 3.190739 & 13.84 + 7363684 & 292.06889 & 42.93308 & 0.729569 & 13.72 + 7431838 & 287.45459 & 43.04541 & 3.036295 & 12.09 + 7434110 & 288.35027 & 43.04639 & 7.223437 & 15.99 + 7523733 & 290.58686 & 43.11656 & 3.121057 & 11.81 + 7549265 & 297.31412 & 43.14078 & 1.962315 & 13.8 + 7748449 & 290.22456 & 43.49293 & 15.705378 & 13.43 + 7905209 & 296.36599 & 43.62922 & 0.209777 & 13.27 + 8029556 & 292.09914 & 43.88178 & 18.040183 & 15.35 + 8098181 & 292.01466 & 43.97829 & 0.986971 & 12.86 + 8432040 & 292.30794 & 44.48024 & 7.991428 & 11.95 + 8520065 & 299.48401 & 44.50633 & 9.365808 & 13.99 + 8694772 & 293.97896 & 44.85794 & 15.187454 & 14.83 + 8707639 & 297.887694 & 44.868264 & 7.785182 & 12.71 + 8774912 & 298.697591 & 44.90004 & 15.137412 & 12.99 + 8881943 & 290.48391 & 45.18651 & 7.777886 & 14.79 + 8972908 & 297.87374 & 45.24537 & 4.195616 & 12.58 + 9011963 & 287.87358 & 45.31096 & 8.400634 & 15.5 + 9283572 & 292.93673 & 45.73978 & 1.217898 & 14.36 + 9511937 & 284.28423 & 46.11628 & 0.277782 & 15.54 + 9651668 & 292.85574 & 46.39118 & 2.68435 & 14.29 + 9700181 & 286.87755 & 46.43377 & 2.846435 & 12.74 + 9726020 & 297.60386 & 46.48611 & 10.642508 & 12.94 + 10389596 & 284.504783 & 47.550999 & 2.608079 & 13.36 + 10669516 & 293.44307 & 47.98749 & 10.485432 & 13.03 + 10817620 & 298.71137 & 48.13449 & 2.946091 & 13.96 + 10982373 & 294.90293 & 48.47721 & 10.01776 & 11.7 + 10990083 & 297.79189 & 48.49364 & 10.715948 & 13.24 + 11087095 & 293.33433 & 48.69283 & 0.77833 & 15.57 + 11152786 & 298.17105 & 48.7121 & 8.229868 & 15.41 + 11236035 & 287.51454 & 48.93461 & 2.318711 & 14.65 + 11284185 & 284.144111 & 49.036 & 15.705099 & 13.66 + 11653122 & 286.610501 & 49.79567 & 2.650525 & 14.76 + 11922402 & 295.85559 & 50.28799 & 5.917946 & 15.93 + 12102573 & 286.24793 & 50.61843 & 10.942348 & 11.81 + 12105278 & 288.17649 & 50.67654 & 7.110381 & 12.32 + 12314646 & 295.343081 & 51.07426 & 13.579375 &
14.14 + 12458797 & 290.53101 & 51.30857 & 0.241319 & 13.89 + 12602567 & 290.586697 & 51.602235 & 10.241997 & 13.02 + | large repositories of high precision light curve data , such as the _ kepler _ data set , provide the opportunity to identify astrophysically important eclipsing binary ( eb ) systems in large quantities
. however , the rate of classical by eye " human analysis restricts complete and efficient mining of ebs from these data using classical techniques . to prepare for mining ebs from the upcoming k2 mission as well as other current missions
, we developed an automated end - to - end computational pipeline the eclipsing binary factory ( ebf ) that automatically identifies ebs and classifies them into morphological types .
the ebf has been previously tested on ground - based light curves . to assess the performance of the ebf in the context of space - based data , we apply the ebf to the full set of light curves in the _ kepler _ `` q3 '' data release .
we compare the ebs identified from this automated approach against the human generated _ kepler _ eb catalog of @xmath0 ebs . when we require eb classification with @xmath1 90% confidence , we find that the ebf correctly identifies and classifies eclipsing contact ( ec ) , eclipsing semi - detached ( esd ) , and eclipsing detached ( ed ) systems with a false positive rate of only 4% , 4% , and 8% , while complete to 64% , 46% , and 32% respectively .
when classification confidence is relaxed , the ebf identifies and classifies ecs , esds , and eds with a slightly higher false positive rate of 6% , 16% , and 8% , while much more complete to 86% , 74% , and 62% respectively . through our processing of the entire _ kepler
_ `` q3 '' dataset , we also identify 68 new candidate ebs that may have been missed by the human generated _ kepler _ eb catalog .
we discuss the ebf s potential application to light curve classification for periodic variable stars more generally for current and upcoming surveys like k2 and the transiting exoplanet survey satellite . |
the stars of the bulge of the milky way have a broad metallicity distribution , with [ fe / h ] between @xmath3 and @xmath4 dex .
spectroscopic surveys of the bulge , both high and medium resolution , have reported a metallicity gradient in latitude .
this was first measured to be about @xmath5 dex kpc@xmath6 along the minor axis of the bulge by minniti et al ( 1995 ) , and then @xmath7 dex kpc@xmath6 by @xcite from observations of about 900 k giants across @xmath8 to @xmath9 on the minor axis .
@xcite later found a similar minor axis gradient of @xmath10 dex kpc@xmath6 from spectroscopy of about 2000 red clump giants with @xmath11 kpc , across @xmath12 to @xmath13 .
numerous studies ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) have reported similar mdfs for the bulge .
figure [ fig : thomas ] , from @xcite compares the mdfs from a number of different studies in various bulge fields .
in addition to the vertical metallicity gradient , there is also a weaker gradient in longitude , at @xmath14 .
this is seen in the photometric metallicity distribution by @xcite .
the vertical abundance gradient was initially interpreted as inconsistent with the formation of the bulge from the disk via dynamical instabilities , which results in a boxy / peanut morphology similar to that already observed in the milky way @xcite .
this large abundance gradient was instead interpreted as evidence for a bulge formed via mergers or dissipational collapse @xcite .
subsequently , however , bulge metallicity gradients in @xmath15 like those measured for the milky way have been shown to be a natural consequence of bulge formation via instability of the disk , due to the preferential redistribution of stars according to their initial phase space energies in the disk @xcite .
furthermore , the overall large negative gradient in the mean of the mdf is now believed simply to reflect the changing contribution of populations which have different spatial distributions in the innermost regions .
the more metal - rich population of stars is concentrated to the plane , while the more metal - poor populations are present in greater number at larger heights from the plane .
although the various observers find very similar mdfs ( see figure [ fig : thomas ] ) , there is no consensus yet in the interpretation of this mdf ( see section [ sec : pops ] ) .
s window from @xcite c ) 220 red clump stars in baade@xmath16s window from @xcite d ) 1813 red giant stars from the argos survey fields at ( l , b ) = ( 0@xmath17 , 5@xmath17 ) , ( 5@xmath17 , 5@xmath17 ) , ( 5@xmath17,5@xmath17 ) from @xcite ; e ) 363 red giants at ( l , b ) = ( 0 , 10@xmath17 ) from @xcite ; the curves are generalised histograms and the dotted lines mark the peaks of the mdf components proposed by @xcite . ] the vertical and ( lesser ) radial metallicity gradients reported for the mdf in the bulge by numerous studies have been found to flatten for latitudes @xmath18 .
this was first reported from studies using small samples of stars near the minor axis @xcite . from the apogee survey
, it has been revealed that the flattening of the gradient near the plane is seen right across the bulge in longitude , out to @xmath19 and not just on the minor axis , again for latitudes @xmath20 .
this is shown in figure [ fig : apogee1 ] which includes 8500 stars from the apogee survey and 8,000 stars from the argos survey ( those at positive longitudes where apogee has observed ) .
these stars are at distances from @xmath21 to @xmath22 kpc from the sun , thus showing stars in the bulge region ( and stars in the disk away from the sun to illustrate the smooth transition between the bulge and the disk ) .
the mean of these stars is shown in figure [ fig : apogee1 ] ( a ) and figure [ fig : apogee1 ] ( b ) shows the mean .
the argos fields , which each have about 600 stars , are shown in the larger boxes with black outlines for figure ( a ) only .
there are typically @xmath23 30 stars in each of the bins which show the mean measurement for the apogee sample . from figure
[ fig : apogee1 ] ( a ) it is clear that in the inner region with ( l , b ) @xmath24 , @xmath25 the metallicity is nearly constant at constant latitude .
the most metal - rich and correspondingly alpha - poor stars are concentrated to the layer @xmath26 , and these metal - rich stars show a smooth transition out into the disk . in the plane , the stars reach the highest metallicity along the thin long bar reported by wegg et al .
( 2015 ) , the profile of which is shown by the thin long box in figure [ fig : apogee1 ] , extending out to @xmath27 .
the outline of the cobe boxy / bulge profile is shown in this figure as the larger box , going out to longitudes of @xmath28 .
the metallicity measurements for @xmath14 from apogee and argos are in good agreement with the photometric metallicities of @xcite 0.5 0.5
several studies ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) have interpreted the mdf of the bulge across a broad region in @xmath15 as being made up of multiple populations . @xcite
interpret the mdf as having 5 populations which are shown in figure [ fig : ness ] : 3 dominant populations ( a c ) with [ fe / h ] @xmath29 and two minor metal - poor populations ( d & e ) . from the argos bulge survey described in @xcite ,
the three populations a - c with @xmath29 have peaks at metallicities of about + 0.15 , -0.25 , -0.7 dex , respectively and provide about 95% of stars in the bulge .
they associate these populations with the stars of the boxy / peanut bulge ( a and b ) , the thick disk ( c ) , the metal weak thick disk ( d ) and stellar halo ( e ) .
they find that these populations are present in different proportions throughout the inner region of the milky way .
0.3 0.3 0.3 using red clump stars , @xcite and @xcite showed that only the more metal rich stars in populations a and b ( i.e. stars with @xmath30 ) are part of the split density distribution of stars in the innermost region that reflects the x - shaped morphology of the bulge .
this implies that stars with @xmath2 that are present in the inner region are not part of the boxy / peanut bulge morphology and are not on x - profile - supporting x1 orbit families .
the most metal - rich stars show the largest minima between peaks in the k - magnitude distribution of stars and are the most strongly involved in the x - shape @xcite .
population a was found by @xcite to be concentrated toward the plane and to be the thinner part of the boxy / peanut bulge .
population b corresponds to the vertically thicker stars in the bulge , with a similar contribution fraction across @xmath31 to @xmath32 .
population c is not significantly involved in the split density distribution or boxy / peanut structure but transitions smoothly from the bulge to disk with longitude ; it is identified with the inner thick disk .
the two populations of stars ( a and b ) associated with the boxy / peanut structure have similar peaked velocity dispersion profiles with longitude .
the velocity dispersion profiles for the more metal - poor stars have a different shape ( e.g. * ? ? ?
* ; * ? ? ?
* ) ( see below ) .
the 5% of stars in the bulge with metallicities
@xmath33 are chemically similar to stars of the metal - weak thick disk and halo near the solar neighbourhood ( e.g. * ? ? ?
* ; * ? ? ?
* ) and were associated by @xcite with these populations .
the rotation and dispersion profiles reported by @xcite and shown in figure [ fig : kinematics ] as a function of support the differentiation of the populations .
the kinematics of stars in populations a & b are related and are different from the other components .
the stars in populations a and b , which are part of the split clump and the boxy / peanut bulge , show the same characteristic peaked pattern of velocity dispersion in the two left hand panels of figure [ fig : kinematics ] , with population a being a colder replica of population b. population c , associated with the thick disk in the inner galaxy , is rotating as rapidly as the more metal - rich populations ( all show the latitude - independent rotation profiles that are commonly seen in boxy bulges ) but its velocity dispersion profile has a different shape .
the most metal - poor stars with [ fe / h ] @xmath34 have the slow rotation profile and high dispersion that might be expected of a population that had no ancestral association with a disk , such as a spheroidal population in the bulge or the stars of the inner halo or an underlying merger component .
other studies @xcite have proposed that the mdf of the bulge comprises two populations , a metal - rich population that is part of the boxy / peanut bulge and a metal poor population that is an old spheroid ( i.e. with a possible formation history that is distinct from the disk and halo components of the milky way ) .
although note that @xcite concludes that whether their two mdf components have different formation histories is yet to be determined .
the decomposition of stars into two roughly equal components from @xcite is shown in figure [ fig : gonzalez ] , from the gibs survey @xcite , where these two populations peak at [ fe / h ] of about + 0.25 dex and -0.3 dex .
the 2-component decomposition places fewer stars within the boxy / peanut or x - shaped structure , attributing approximately 50% to an old spheroid @xcite .
the 5-population model of @xcite attributes 95% of stars to be disk stars , with the population c ( at = 0.7 ) being disk stars but simply not part of the x - shaped morphology ( possibly because they were originally part of a hotter thick disk which was dynamically less responsive to the instability ) .
these two interpretations therefore have different implications for the stars in the inner region in terms of their origin from the disk and the contribution of any additional population that is distinct from any other milky way population and specifically from the bulge , such as an old spheroid formed via mergers at high redshift . whereas a large fraction , up to 50% of stars in the two - component decomposition ,
are associated with an old - spheroid component , only @xmath35 5% of the population in the 5 population decomposition is associated with any old spheroid component , and this component may simply be metal - poor halo stars in the inner region and not a unique population of the bulge .
the alpha - enhancement of stars in the bulge shows similar trends to that of the thin and thick disk in the solar neighbourhood , as reported by alves brito et al .
( 2013 ) , bensby et al .
( 2013 ) and gonzalez et al.(2015 ) .
the apogee survey includes observations across the milky way s disk and bulge and allows the trends to be homogeneously mapped from the outermost to the innermost region of the galaxy ( see e.g. * ? ? ?
* ; * ? ? ?
comparing the trends and the individual abundance patterns ( [ x / fe ] ) for bulge stars with stars at other galactic radii is important for understanding the detailed formation of the bulge and its relationship with the disk .
a large fraction of the bulge stars is likely to have come from the disk , and therefore the broad expectation would be that disk and bulge abundance trends will be similar . however , there are a number of perturbing processes that may show abundance signatures .
these include the star formation rates in the inner and outer disk and the growth of the disk over time , gas inflow driving star formation , and radial migration .
the alpha and abundance trends of the rare metal - poor stars in the inner milky way are particularly interesting due to the prediction that the oldest stars of the galaxy will now be concentrated to the bulge ( e.g * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) figure [ fig : apogee_alpha ] shows the [ fe / h][@xmath36/fe ] maps for the apogee stars for different galactic zones in @xmath37 and @xmath1 .
the stellar parameters have been obtained with the cannon @xcite and demonstrate that the inner galaxy has the narrowest alpha sequence , which is offset to higher metallicity than in the outer galaxy .
the dotted line in figure [ fig : apogee_alpha ] marks the approximate trend of the high - alpha sequence that is seen in the innermost region .
the density distribution shows that , in the zones further from the galactic plane at a given radius , more stars are found in the low metallicity / high alpha sequence than in the high metallicity / low alpha sequence .
figure [ fig : apogee_alpha ] shows that there is a smooth transition from the outer to inner region at a given height from the plane , with an increasing mean metallicity .
the low - alpha sequence dominates in the outer region and a narrower sequence extending to lower and higher dominates in the inner region . at intermediate radii , near the sun , both low and high alpha sequences are present .
the narrower and high - alpha sequence seen in the inner galaxy implies that the star formation and chemical evolution rate was high at the early epoch in the disk at which the bulge stars were forming .
the star formation continued over a relatively brief period , and this single sequence is a snapshot in time of the rapid chemical evolution of the early disk from which the bulge was formed .
@xcite showed that bulges form early on after the disk , and over a short period of 1 - 2 gyr .
this rapid formation at early times is consistent with the single continuous alpha sequence seen in the inner 3kpc of the milky way that is observed in apogee data .
the simulations of @xcite show that stars are mapped into the bulge from the disk according to their initial phase space location .
stars now in the bulge that reside nearest to the plane originate from the innermost region of the disk .
similarly , the stars in the bulge whose orbits take them to higher @xmath1 came from larger radii in the disk and preferentially at larger heights above the plane .
simulations that include star formation ( e.g. * ? ? ?
* ; * ? ? ?
* ) combined with new observational data e.g. from apogee which covers from the bulge and into the disk may offer insight into the observed chemodynamical distribution of stars across @xmath15 . at small galactic radii in the bulge ,
the low and high alpha sequences are both present near the sun , and only the low - alpha sequence is present at the largest radii in the disk . [
fig : apogee_alpha ]
about 5% of the stars in the bulge have metallicities @xmath0 1.0 .
the argos survey showed that the fraction of these stars , which they associated with the galactic halo population , increases with height above the plane .
this population also rotates more slowly , at about 50% of the speed of more metal rich stars @xcite .
it also has a high and relatively latitude - independent velocity dispersion as seen in fig .
have also suggested that the halo population in the bulge is not insignificant , even within 1 kpc of the galactic center .
@xcite measured the individual abundances of microlensed dwarf bulge stars and found that the stars with @xmath38 1.0 have very similar chemical abundance trends to the nearby thick disk stars . the bulge rr lyrae population , whose mdf peaks around @xmath39 @xcite , has been used to trace the distribution of the old metal - poor population of stars in the bulge .
the rr lyrae stars are ideal tracers of structure because their distances can be accurately estimated .
although @xcite reported the spatial distribution of the rr lyrae to be spheroidal with only a slight elongation , the survey of @xcite showed that the metal - poor rr lyrae very closely trace the barred structure of intermediate - age red clump giants .
@xcite show that the spatial distribution of the rr lyrae population is barred but that these stars are not part of the x - shape structure .
they also report two sequences of rr lyrae with marginally different metallicities , proposing evidence for the role of mergers in the initial formation of the bulge .
although they may follow the elongated distribution of the bar , the rr lyrae stars may still be part of the halo but which has been pulled into a bar shape by the potential of the other stars ( e.g. * ? ? ?
the high - resolution detailed chemical abundance analysis of a handful of the most metal - poor bulge stars with @xmath40 by @xcite and @xcite reported similar individual element enhancement and alpha - enhancement trends to halo stars of the same metallicity ( although @xcite noted an abundance scatter that may be larger ) . from the analysis of the dynamics and detailed chemical abundances of three bulge stars with @xmath41 , @xcite report that these stars follow the abundance trends identified previously for metal - poor halo stars , except for scandium .
simulations ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) predict that the oldest stars in the galaxy will be concentrated to the bulge .
these very metal poor stars found by @xcite may well be members of the oldest populations of stars in the milky way , formed at redshift @xmath42 .
the abundance patterns of this rare stellar population which is concentrated to the bulge could provide critical insight into the composition of the universe at the highest redshift and the conditions at the beginning of the milky way s formation .
numerous studies , including small programs and large surveys observing the bulge across a broad spatial extent ( e.g. argos , gibs , vvv ) have characterised the overall metallicity distribution of the bulge . until recently , most spectroscopic studies have looked off the plane , at @xmath43 . although a small number of fields nearer to the plane have been observed ( e.g. * ? ? ?
* ) , with the near - ir apogee survey @xcite , the mdf can be examined for the bulge and for regions extending out into the disk , at lower latitudes and in the plane across the entire bulge and into the disk . detailed analysis from spectroscopic surveys
has also shown the trends of alpha and individual element abundances for stars of the bulge , at all metallicities , to be comparable with those of disk and halo stars of similar metallicity in the local neighbourhood .
overall the stars of the bulge show similarities to corresponding stars in the nearby disk and halo populations , but there are some critical differences
. the [ @xmath36/fe ] behaviour of bulge stars shown in fig .
6 illustrates the rapid history of chemical evolution that occurred in the inner galaxy , probably before the bulge structure was in place .
the anomalous scandium abundance of the most metal - poor bulge stars ( @xmath44 ) may indicate that these very metal - poor stars in the bulge may indeed be among the first stars to have form in the very early galaxy .
coupling the stellar metallicity distribution data with the kinematics and morphological structure @xcite is key to understanding the relationship between the bulge and the other populations of the milky way that were present at the time that the bulge was formed .
specifically this is important for understanding what fraction of the bulge was formed from pre - existing disk material , and what is the nature of the bulge stars that are not associated with the boxy bulge .
are they simply stars of the disk and inner halo , or is there a unique bulge population that is not related to the disk ? a lower limit of @xmath38 25% on the mass fraction in the x - structure has been recently determined by @xcite . constraining
the total fractional mass of stars in the x - shape is of great interest , as a check on the consistency of the fractional contributions of the different mdf populations with their likely origins . ,
t. , adn , d. , melndez , j. , gould , a. , feltzing , s. , asplund , m. , johnson , j. a. , lucatello , s. , yee , j. c. , ramrez , i. , cohen , j. g. , thompson , i. , bond , i. a. , gal - yam , a. , han , c. , sumi , t. , suzuki , d. , wada , k. , miyake , n. , furusawa , k. , ohmori , k. , saito , t. , tristram , p. , and bennett , d. ( 2011 ) . .
, * 533 * , a134 . ,
t. , yee , j. c. , feltzing , s. , johnson , j. a. , gould , a. , cohen , j. g. , asplund , m. , melndez , j. , lucatello , s. , han , c. , thompson , i. , gal - yam , a. , udalski , a. , bennett , d. p. , bond , i. a. , kohei , w. , sumi , t. , suzuki , d. , suzuki , k. , takino , s. , tristram , p. , yamai , n. , and yonehara , a. ( 2013 ) . . ,
* 549 * , a147 . , f. , gibson , b. k. , michel - dansac , l. , stinson , g. s. , cignoni , m. , dotter , a. , pilkington , k. , house , e. l. , brook , c. b. , few , c. g. , bailin , j. , couchman , h. m. p. , and wadsley , j. ( 2012 ) .
, * 427 * , 14011417 . , k. c. ( 2012 ) . . in w. aoki , m. ishigaki , t. suda , t. tsujimoto , and n. arimoto , editors , _ galactic archaeology : near - field cosmology and the formation of the milky way _ ,
volume 458 of _ astronomical society of the pacific conference series _
, page 393 .
, a. e. , cunha , k. , shetrone , m. , majewski , s. r. , johnson , j. a. , smith , v. v. , schiavon , r. p. , holtzman , j. , nidever , d. , zasowski , g. , allende prieto , c. , beers , t. c. , bizyaev , d. , ebelke , g. , eisenstein , d. j. , frinchaboy , p. m. , girardi , l. , hearty , f. r. , malanushenko , e. , malanushenko , v. , meszaros , s. , oconnell , r. w. , oravetz , d. , pan , k. , robin , a. c. , schneider , d. p. , schultheis , m. , skrutskie , m. f. , simmonsand , a. , and wilson , j. c. ( 2013 ) . .
, * 767 * , l9 . , l. m. , asplund , m. , casey , a. r. , keller , s. c. , yong , d. , gilmore , g. , lind , k. , worley , c. , bessell , m. s. , casagrande , l. , marino , a. f. , nataf , d. m. , owen , c. i. , da costa , g. s. , schmidt , b. p. , tisserand , p. , randich , s. , feltzing , s. , vallenari , a. , allende prieto , c. , bensby , t. , flaccomio , e. , korn , a. j. , pancino , e. , recio - blanco , a. , smiljanic , r. , bergemann , m. , costado , m. t. , damiani , f. , heiter , u. , hill , v. , hourihane , a. , jofr , p. , lardo , c. , de laverny , p. , magrini , l. , maiorca , e. , masseron , t. , morbidelli , l. , sacco , g. g. , minniti , d. , and zoccali , m. ( 2014 ) . . , * 445 * , 42414246 . , d. l. , bovy , j. , bird , j. c. , andrews , b. h. , hayden , m. , holtzman , j. , majewski , s. r. , smith , v. , robin , a. c. , garca prez , a. e. , cunha , k. , allende prieto , c. , zasowski , g. , schiavon , r. p. , johnson , j. a. , weinberg , d. h. , feuillet , d. , schneider , d. p. , shetrone , m. , sobeck , j. , garca - hernndez , d. a. , zamora , o. , rix , h .- w . , beers , t. c. , wilson , j. c. , oconnell , r. w. , minchev , i. , chiappini , c. , anders , f. , bizyaev , d. , brewington , h. , ebelke , g. , frinchaboy , p. m. , ge , j. , kinemuchi , k. , malanushenko , e. , malanushenko , v. , marchante , m. , mszros , s. , oravetz , d. , pan , k. , simmons , a. , and skrutskie , m. f. ( 2014 ) . .
, * 796 * , 38 . , a. , recio - blanco , a. , hill , v. , de laverny , p. , schultheis , m. , babusiaux , c. , zoccali , m. , minniti , d. , gonzalez , o. a. , feltzing , s. , gilmore , g. , randich , s. , vallenari , a. , alfaro , e. j. , bensby , t. , bragaglia , a. , flaccomio , e. , lanzafame , a. c. , pancino , e. , smiljanic , r. , bergemann , m. , costado , m. t. , damiani , f. , hourihane , a. , jofr , p. , lardo , c. , magrini , l. , maiorca , e. , morbidelli , l. , sbordone , l. , worley , c. c. , zaggia , s. , and wyse , r. ( 2014 ) . . , * 569 * , a103 . ,
m. , gonzalez , o. a. , vasquez , s. , hill , v. , rejkuba , m. , valenti , e. , renzini , a. , rojas - arriagada , a. , martinez - valpuesta , i. , babusiaux , c. , brown , t. , minniti , d. , and mcwilliam , a. ( 2014 ) . .
, * 562 * , a66 . | the galactic bulge of the milky way is made up of stars with a broad range of metallicity , 3.0 @xmath0 [ fe / h ] @xmath0 1 dex .
the mean of the metallicity distribution function ( mdf ) decreases as a function of height @xmath1 from the plane and , more weakly , with galactic radius .
the most metal rich stars in the inner galaxy are concentrated to the plane and the more metal poor stars are found predominantly further from the plane , with an overall vertical gradient in the mean of the mdf of about 0.45 dex / kpc .
this vertical gradient is believed to reflect the changing contribution with height of different populations in the inner - most region of the galaxy .
the more metal rich stars of the bulge are part of the boxy / peanut structure and comprise stars in orbits which trace out the underlying x - shape .
there is still a lack of consensus on the origin of the metal poor stars ( [ fe / h ] @xmath2 ) in the region of the bulge .
some studies attribute the more metal poor stars of the bulge to the thick disk and stellar halo that are present in the inner region , and other studies propose that the metal poor stars are a distinct ` old spheroid ' bulge population .
understanding the origin of the populations that make up the mdf of the bulge , and identifying if there is a unique bulge population which has formed separately from the disk and halo , has important consequences for identifying the relevant processes in the the formation and evolution of the milky way .
the galaxy bulge abundances formation evolution |
recently a lot of attention has been paid on the properties of s / n / s and n / n / n microbridge structures , and noise measurement ( shot noise @xcite as well as thermal noise @xcite ) was used as an additional technique besides transport measurements . but to measure the noise at cryogenic temperature is difficult since the sample noise usually is much smaller than the thermal noise of the components in the test circuit .
noise thermometry for electrons at low temperature was previously performed by current noise measurement with squid @xcite , which is very sensitive and has very low noise level .
however it is limited for small resistance and not for voltage noise measurement , also it can not be used when magnetic field is applied .
cross - correlation technique @xcite provides an alternative method to measure the sample noise at low temperature .
one bottleneck for this technique is that it requires a lot of time to do cross - correlation to converge and achieve the required sensitivity , and the trade off between sensitivity and the time needed to converge makes it difficult to use for low level noise experiments . in this article
we present an improved cross - correlation algorithm as well as the test instrument set - up for measurement of thermal noise .
this algorithm does a vector average over both time and specific frequency range .
it is worth noting that for commercial spectrum analyzer usually only average over time is used , and it is impossible to realize this algorithm within the instrument because of limitations like memory and computation speed . as shown in the following sections , much faster convergence can be achieved with the new algorithm .
it is well known 4-probe resistance measurement eliminates contact resistance by measuring the voltage signal across the sample that was stimulated by the current though the sample .
it can be also considered as measuring the `` in phase '' signal between current and voltage across the sample .
cross - correlation is similar in a sense it eliminates the channel s noise by measuring the `` similarity '' or `` in phase '' signal between two different voltage channels . to better illustrate it ,
let consider the voltage signals from two channels .
the fourier components at particular frequency @xmath1 are : @xmath2 here a is the amplitude of the noise signal from our sample at frequency @xmath1 and @xmath3 , @xmath4 are the amplitude of unwanted noise at @xmath1 generated in those two channels , for example , the thermal noise generated by the 20@xmath5 lead resistor as shown in fig .
[ fig1 ] .
schematic of the set - up .
the sample resistor is 10 @xmath5 and 1.5@xmath5 .
the thermal noise is 20 @xmath5 resistor along the channel is much bigger than the sample thermal noise . ] to do cross - correlation we calculate the product of the two vectors : @xmath6 for the last three terms in eq.[eqn2 ] , since all those phases @xmath7 , @xmath8 are random for white noise , and @xmath9 and @xmath4 does not change over time , when we do the average of @xmath10 over enough long time , the random phase terms will cancel each other and will give negligible contribution to the total amplitude .
so the real part of the @xmath10 will converge at @xmath11 , and the imaginary part will converge at @xmath12 .
ideally if we do the average over infinite time the amplitude will converge at @xmath11 .
but in practical situation , the measurement time should be limited to some reasonable extent . to estimate the time needed to approach convergence limit
, we need to find when the deviation of the random phase term is much smaller than the @xmath11 term .
first in order to identify the signals @xmath13 at frequency @xmath1 , the sampling time @xmath14 should be much longer than @xmath15 to get an accurate fourier component . then @xmath13 at different times are acquired to do the vector average to eliminate the random phase terms .
if @xmath16 , which happens when the sample is at low temperature and the sample signal amplitude @xmath17 is very small , it will take very long time to approach the convergence limit , which requires large n in the following equation : @xmath18 from standard textbook@xcite we know for iid ( independent identical distributed ) random sequence , n times average gives n times smaller variance .
assuming our channel noises are iid , we can expect that the variance of the averaged random phase term @xmath19 , similarly for the averaged correlation the variance @xmath20 , since our goal is to find noise magnitude @xmath17 , we need the variance of root square of real part @xmath21 , @xmath22 which is proportional to @xmath23 , and finally what need standard deviation to compare with @xmath17 . the standard deviation @xmath24 should be proportional to @xmath25 .
the @xmath26 exponent is indeed observed in our experiment as shown in sec.[exponent ] .
this @xmath26 exponent might also be used as a criteria to decide if the channel noise in different time steps can be fitted as iid random sequence , i.e. whether there is some correlation in time domain .
as indicated above by the @xmath26 exponent , it is not very effective to eliminate the channel noise by increasing the measurement time steps . to accelerate the convergence process , a new algorithm is described below .
consider cross - correlation results at two different frequencies @xmath27 and @xmath28 : @xmath29 here the last vector term @xmath30 represents the vector sum of the last three terms in eq.[eqn2 ] .
for `` white '' noise , the amplitude @xmath17 is the same for different frequencies but the phases @xmath31 , @xmath32 are random .
this means a vector average over frequency domain is equivalent to the vector average over time domain . since in practice
we usually acquire a series of data points in one sampling time @xmath14 and then do a fft transform , we could compute all the points in the frequency domain and use them to do vector average .
for example the commercial spectrum analyzer usually takes 1024 scaler voltage points in @xmath14 ( ca nt take more because of limited memory ) and give 512 vector points in the frequency domain . if we do vector average of the 512 points , according to the above statement it is similar to the result of average over 512 time steps for one particular frequency , which means the convergence at @xmath17 can be achieved 512 times faster . to test this we built some simple experimental set - up as described in the following section
a schematic plot of the test set up is shown in fig . [ fig1 ] .
since people usually use resistive stainless steel coax cable to connect the sample in the low temperature stage , here a 20@xmath5 resistor is used along each channel to simulate the channel resistor .
the sample resistors used here are 10@xmath5 and 1.5@xmath5 to simulate the low noise level from real sample . in this case
the amplitude of the sample noise @xmath17 is much smaller than that of channel noise @xmath33 , so it can only be retrieved by cross - correlation technique .
the output of the two channels feeds separately to two par116 preamplifier ( transformer mode ) and par124 lock - in amplifier(only as an additional cascade amplifier , the monitor output is used ) .
the transformer is used to match the impedance .
two transformers need to be similar because otherwise if may change the phase and amplitude and affect the convergence .
for example if there is a fixed phase difference @xmath34 between two transformer , the amplitude @xmath17 will be reduced to @xmath35 .
after amplification the signals are fed to left / right channel input of a standard pci sound card installed in a pii pc . the sound card then digitize the signal with 16 bit resolution and 44.1khz sampling rates .
after that the data is acquired by a program written in labview to the computer memory . then the program calculates fourier spectra and and performs correlation and vector average etc . , and shows all results on the screen in real time .
a pc is much better than a commercial analyzer when considering the memory size and computation speed .
in fig . [ fig2 ] , for the 10@xmath5 sample resistor three curves are shown to demonstrate the result of conventional cross - correlation algorithm and to compare it with the new algorithm .
curve a shows that the standard deviation of the average over time , @xmath36 decreases as the time elapse@xcite . as shown in sec.[cross ] it is proportional to @xmath25 , which can be find easily from the log - log plot .
curve b shows the average over time approaches slowly to @xmath17 , which is around 0.3nv/@xmath0 , after more than 100 times average .
curve c shows that when using the new algorithm , the average over frequency _ and _ time almost converge at @xmath17 from the first point !
in fact , curve a is proportional to the difference between curve b and curve c. the measured amplitude of @xmath17 is close to expected amplitude of @xmath17 , which is @xmath37 0.4nv/@xmath0 for 10@xmath5 sample resistor .
this result is not bad when considering there may be affects from non ideal phase and amplitude properties of transformers and amplifiers , and uncertain pre - factors that came in from the data processing like the use of windows when doing fft . for a 10@xmath5 sample resistor with 20@xmath5 channel resistor ,
the cross - correlation results are shown .
curve a shows standard deviation of the average over time .
the slope is proportional to @xmath25 .
curve b shows vector average over time .
curve c shows vector average over both frequency _ and _ time domain . with this new algorithm
the convergence is achieved almost from the first point , much faster than the conventional cross - correlation result shown by curve b. ] fft spectrum after 100 times average is shown in fig .
curve b shows the result of the conventional cross - correlation vector average over 100 time steps . for comparison ,
curves c , d shows separately the noise spectrum of left / right channel measured in @xmath14 .
the vector average over both frequency and time domain is just a number , so it can not be shown in this frequency spectrum figure . from curve
b , despite of those pick up noise peaks , we can still `` see '' the real noise level that is around 0.3nv/@xmath0 , which was also found by the program and shown as the last point of curve b in fig .
the program actually average _ scalarly _ the spectrum amplitude from 1khz to 2khz where the spectrum is almost flat and the affect of power line noise peaks is smaller . and those points close to power line noise peaks were abandoned .
it is worth noting this _ scalarly _ average of amplitude over frequency is different with the vector average over frequency .
fft spectrum after 100 times average .
curve b shows the cross - correlation vector average over 100 time steps , curve c , d shows the noise spectrum of left / right channel in @xmath14 .
the noise floor level of curve b is close to 0.3nv/@xmath0 as shown in fig .
the decrease of amplitude below 60hz and the dip near 8khz is due to the amplitude and phase properties of transformers .
the cut - off near 22khz is due to the shannon limit , i.e. , half of 44.1khz sampling frequency . ] with the presence of huge noise peaks as shown in fig .
[ fig3 ] , to observe the sample noise level it is required that the spectrum leakage and sidelobe background of the unwanted power line noise peaks should nt mask the real white noise floor .
this is usually achieved by using special window function when doing fft and by increasing the frequency resolution@xcite . since hann window
has a fast decreasing sidelobe magnitude , it is preferred in this situation than uniform window which is conventionally used for flat noise spectrum measurement . and by increasing the frequency resolution the peaks mainlobe can be narrowed and their sidelobes can be attenuated . in our case
the sampling rate is 44.1khz , sampling number is chosen to be 32768 ( @xmath38 ) points for each step , so the sampling time @xmath14 is @xmath39 second for each step , and frequency resolution @xmath40 is @xmath41hz .
it is possible to increase the sampling number to increase @xmath14 and decrease @xmath40 .
this will require only larger pc memory and higher speed cpu which is inexpensive .
a simple algorithm is used here to eliminate 3 points from both sides of those power line peaks frequency when doing the average .
this is already good enough to find the real noise floor of curve b in fig .
more complex ways using adaptive filter program to remove the noise peak and extract the floor level is also possible . as shown in sec .
[ cross ] , the number of points used for vector average over frequency domain decides how much times faster of this new algorithm compared to conventional algorithm . here
since we used the range from 1khz to 2khz , with resolution 1.364hz , we get 733 points . after subtracting the number of those points that are too close to noise peaks
, we have around 600 points .
so in principle we should get 600 times faster . to test this we measured room temperature noise for 1.5@xmath5 sample resistor with same 20@xmath5 channel resistors .
the result is shown in fig .
the start point of curve a and b are mostly determined by the 20@xmath5 resistor .
there is a ratio about 3 between those two curves around the start point . to detect the noise level from 1.5@xmath5 sample resistor
, we can assume the required standard deviation @xmath42 to be 5 times smaller than the convergence limit @xmath17 , which is @xmath43 , we would need @xmath42 decrease to magnitude @xmath44 . the time needed for conventional cross - correlation methods
can be estimated by : @xmath45 for the new algorithm , we expect @xmath46 steps . two cross - correlation test results for 1.5 @xmath5 sample with 20@xmath5 channel resistor
, one stopped after 100 time steps , the other stopped after 1000 time steps .
curve b shows that using conventional average over time methods , it approaches convergence limit after 1000 averages .
curve c shows that using vector average over both frequency and time , the convergence was approached within a few time steps . ] as shown in fig .
[ fig4 ] the convergence limit is about 0.126 nv/@xmath0 .
it is close to the estimated value of thermal noise level of a 1.5@xmath5 resistor at room temperature , which is 0.158 nv/@xmath0 . and
@xmath47 is consistent with the 10@xmath5 case . at 1000th time step , the last point of curve a in fig .
[ fig4 ] has the value 25.6 pv/@xmath0 .
the ratio between convergence limit and standard deviation is @xmath48 , which is close to our estimation that is 5 for 1372 steps . as for curve c , it approaches the convergence limit from the first point in the 100 time step case , in the 1000 time steps case , despite of some fluctuations that may caused by some broad band noise or data processing , it also approaches the convergence limit from the first a few points .
so it is proved that with the algorithm of vector average over frequency and time , convergence limit can be approached hundreds of times faster than conventional cross - correlation algorithm with this simple setup .
if there are less noise peaks and if large frequency band is available , this algorithm could give even faster result .
for white noise measurement , an improved cross - correlation algorithm using vector average over both frequency and time domain is presented . with consideration of low temperature noise measurement , a simple
test set - up using pc and sound card is built and tested .
it is proved that this algorithm can achieve convergence hundreds of times faster than the conventional cross - correlation algorithm .
even with much bigger channel noise and huge pick up noises , 100 pv/@xmath0 noise level and 25pv/@xmath0 sensitivity can be achieved in seconds . with a broader frequency band width , better a / d card and larger pc memory the convergence can be reached even faster . in principle
this algorithm could be used for other type of noises as long as the shape of the spectrum is known and phase in frequency domain is random(for example 1/f noise ) .
dimitris g. manolakis , vinay k. ingle and stephen m. kogon , _ statistical and adaptive signal processing : spectral estimation , signal modeling , adaptive filtering , and array processing_(mcgraw - hill , boston , 2000 ) .
curve a and b is smooth because in fact it is result of _ scalarly _ averaged of amplitude over a frequency range for every time step , and then _ vector _ average over time is conducted .
this is further explained by fig . | white noise measurement can provide very useful information in addition to normal transport measurements . for example
thermal noise measurement can be used at sub kelvin temperature to determine the absolute electron temperature without applying any heating current . and shot noise measurements helped to understand the properties of nano and mesoscopic normal metal / superconductor structures .
but at low temperature and for relatively small resistance it is difficult to measure the sample s noise magnitude because the background thermal noise can be much larger and usually there are other pick - up noises .
cross correlation technique is one way to solve this problem .
this article describes an improved cross correlation algorithm that averages in both frequency and time domain , and the realization of a simple instrument set - up with pc and sound card . with this
set - up it is shown even with much larger background noise and pickup noises , 100pv/@xmath0 white noise level can be easily measured in seconds .
compared to the normally used cross - correlation methods , it is several orders of magnitude faster . |
transport in superconducting systems of reduced dimensions ( thin films and wires ) is known to be strongly affected by fluctuations in the order parameter .
a prominent manifestation of the role of fluctuations is the finite electrical resistance of narrow superconducting ( sc ) wires at any finite temperature @xmath1 below the bulk critical temperature @xmath5 , established when the wire thickness @xmath6 is reduced below the superconducting coherence length @xcite @xmath7 .
the finite voltage drop along the wires is generated by phase slips : these are processes whereby the superconducting phase slips by @xmath8 at points on the wire where superconductivity is temporarily destroyed .
the resulting voltage is related to the rate of phase slips via the josephson relation .
the pioneering theoretical studies of this phenomenon @xcite accounted for thermally activated phase slips across barriers separating metastable phase configuration states , corresponding to local minima of the landau ginzburg free energy .
this theory turns out to be supported by experimental studies @xcite , in particular providing a good fit of the resistivity @xmath0 slightly below @xmath5 .
as @xmath1 is lowered further , @xmath0 is exponentially suppressed and becomes practically undetectable .
the advance of nanostructure fabrication techniques during the 1980 s opened up the possibility to study narrow wires of smaller diameter , down to a few 100 .
the consequent weakening of superconductivity leads to enhancement of the rate of phase slips , and results in a measurable finite resistance even at @xmath9 . in this low @xmath1 regime , however , thermal activation is considerably suppressed , and the dominant mechanism for phase slips becomes quantum tunneling .
the first experimental indication of quantum phase slips ( qps ) has been seen in thin strips of indium by giordano @xcite .
the curves of @xmath0 vs. @xmath1 exhibit a ` kink ' at some @xmath10 smaller than @xmath5 , below which the decrease of @xmath0 upon lowering @xmath1 becomes more moderate . for @xmath11 , the fit to the theory of ref .
[ ] fails .
motivated by the assumption that quantum tunneling dominates over thermal activation in this regime @xcite , giordano fitted the data to a phenomenological expression in which the temperature @xmath12 in the thermal activation rate @xmath13 is replaced by a `` characteristic frequency '' given by @xmath14 , where @xmath15 is the relaxation time in the time
dependent landau ginzburg theory .
while this phenomenological expression fitted the data quite successfully , it is not justified by a rigorous theoretical derivation . in particular , the landau ginzburg theory ( which is employed in the calculation of the barrier hight @xmath16 ) is based on an expansion in the close vicinity of @xmath5 , and is not expected to hold far below @xmath5 .
subsequent theoretical studies of the qps contribution to resistivity below @xmath5 @xcite have addressed the quantum dynamics of phase fluctuations in the low temperature limit ( @xmath9 ) , yielding a radically different behavior of @xmath0 vs. @xmath1 . at low @xmath1 ,
the magnitude of the sc order parameter field is approximately constant and its dynamics is dominated by phase fluctuations .
the corresponding @xmath17 effective model is a ( 1 + 1)dimensional xy model ,
in which the topological excitations are qps and anti qps ( vortices and anti vortices in space
time ) @xcite . upon tuning a stiffness parameter ( which in particular is proportional to the wire diameter @xmath6 ) below a critical value , the system undergoes a quantum kosterlitz thouless ( kt)@xcite transition from a sc phase to a metallic phase driven by the unbinding of qps anti qps pairs .
the latter are responsible for a finite resistivity at any finite @xmath1 even in the sc phase , where @xmath18 with @xmath19 .
this power
law dependence was predicted both for a granular wire composed of weakly coupled sc grains separated by a tunnel barrier ( a josephson junction array ) @xcite and for the case of a homogeneous wire in the dirty limit @xcite ; however , the exponent @xmath4 and the prefactor @xmath20 are different ( the latter , in particular , strongly depends on the microscopic details ) . the quantitative estimate of @xmath20 and the consequent size of the resistivity in realistic systems have been discussed in detail in the theoretical literature , and led to a debate among the authors @xcite regarding the relevance of the homogeneous wire limit to the experimental system of ref .
[ ] . however , we are not aware of a direct attempt to test the validity of the power law scaling of @xmath0 as an alternative to giordano s phenomenological formula . more recently ,
several experimental groups @xcite have reported qps induced resistivity in a variety of sc nanowire samples with @xmath6 of order 100 and below .
the data in certain samples also exhibit a transition to a normal ( weakly insulating ) state below a critical wire diameter @xmath21 , consistent with the theoretical expectations . in the sc samples ,
the data was fitted by an effective circuit which accounts for both quantum and thermally activated phase slips contributions to @xmath0 . here as well , giordano s phenomenological expression was implemented to describe the qps term in the fitting formula , rather than a power
law @xmath22 ( see , however , the unpublished notes in ref . [ ] and [ ] ) .
the comparison between theory and experiment is therefore not entirely settled . in the present paper
, we introduce a derivation of qps induced resistivity which enables a systematic account of all possible scenarios in a realistic experimental setup , and carry out a comparison with the experimental data .
we focus our attention on dirty sc wires , and account for non uniformity of arbitrary size in the wire parameters , in particular the diameter @xmath6 .
our approach implements the memory function formalism , which directly relates the mechanism responsible for generating finite resistivity to the violation of conservation laws of the unperturbed hamiltonian @xmath23 ( see below ) , describing non singular fluctuations .
the formalism also underlines the interplay of phase
slip processes and disorder .
we consider two distinct types of disorder : the first corresponds to impurities in the underlying normal electronic state , characterized by a mean free path @xmath24 , and the second type is associated with the random spatial fluctuations in the wire diameter on a length scale of order @xmath7 , whose size is characterized by a granularity parameter " @xmath2 .
we find that @xmath0 exhibits a smooth crossover from a homogeneous to a `` granular '' behavior upon increase of @xmath1 .
the latter is expected to dominate in most of the measurable range of @xmath1 due to an exponential enhancement of the prefactor @xmath20 in eq .
( [ powerlaw ] ) .
in addition , we examine the dependence of the exponent @xmath4 on an external magnetic field applied parallel to the wire .
the model and the main steps of our calculation are described in sec .
[ sec : model ] ; details of our derivation of the resistivity within the memory approach are given in appendix a. in sec . [ sec : exp ] we show that the power law eq .
( [ powerlaw ] ) is consistent with the experimental data obtained in a variety of samples , provided the wire is sufficiently long .
the values of @xmath4 extracted from the data , and the corresponding field dependence , are consistent with known parameters of the corresponding samples .
our conclusions are summarized in sec .
[ sec : conclude ] .
we consider a long and narrow superconducting ( sc ) wire of length @xmath25 and cross section @xmath26 , such that @xmath27 and @xmath28 where @xmath7 is the superconducting coherence length and @xmath29 the london penetration depth .
fluctuations in the sc order parameter are therefore effectively one dimensional . as a first stage
we assume that the wire is homogeneous , and the sc material is in the dirty limit where the mean free path in the underlying electronic system obeys @xmath30 .
an effective model in terms of the sc order parameter field is obtained as a result of integrating over the electron fields@xcite . at low temperatures @xmath9 ( where @xmath5 is the bulk sc transition temperature ) , we assume that fluctuations in the magnitude of the order parameter are suppressed , while the quantum dynamics of phase fluctuations @xmath31 is described by the sine - gordon hamiltonian@xcite @xmath32 where @xmath33 \ ; , \\ h_n^{ps}&=&\frac{-2 y^nv}{a^2}\int_{-l/2}^{l/2 } dx \cos(2n\theta)\label{hps}\ ; .\end{aligned}\ ] ] here and throughout the rest of the paper we use units where @xmath34 ; @xmath35 is the field conjugate to @xmath31 , satisfying @xmath36=i\delta(x^\prime - x)$ ] , which physically represent fluctuations in the number of cooper pairs and is related to the field @xmath37 via @xmath38 .
@xmath39 is the effective capacitance per unit length , @xmath40 is the ( three - dimensional ) superfluid density , and @xmath41 , @xmath42 are the electron charge and mass , respectively . in @xmath43
, @xmath44 is the fugacity of quantum phase slip ( qps ) , where @xmath45 is the action associated with the creation of a single qps ( of winding number 1 ) due to the suppression of the sc order parameter in its core ; the characteristic velocity @xmath46 is given by @xmath47 @xmath48 is the winding number ( `` charge '' ) of a qps , and @xmath49 is a short distance cutoff @xmath50 . note that the hamiltonian eq .
( [ hsg ] ) is the same effective model as in ref .
[ ] expressed in a dual representation@xcite .
the hamiltonian @xmath23 describes the non singular phase fluctuations , which are characterized by a free mode ( the mooij
schn mode@xcite ) propagating at a velocity @xmath46 .
it can be recast in the familiar luttinger form @xmath51 where @xmath46 is defined in eq .
( [ v - def ] ) and @xmath52 note that we also assume the wire to be sufficiently long so that @xmath25 is large compared to @xmath53 , in which case it can be practically taken to be infinite .
the nature of the @xmath54 fixed point of @xmath55 [ eq . ( [ hsg ] ) ] depends crucially on the luttinger parameter @xmath56 .
in particular , as noted by ref .
[ ] , the system undergoes a quantum kosterlitz - thouless ( kt)@xcite transition from a sc phase at @xmath57 ( @xmath58 ) to a metallic phase at @xmath59 . for given material parameters ,
the transition can be tuned by varying the wire cross section @xmath60 below a critical value @xmath61 , related to @xmath62 through eq .
( [ k - def ] ) .
we hereon focus on the sc phase corresponding to @xmath63 , in which all the terms @xmath43 [ eq . ( [ hps ] ) ] are irrelevant and the @xmath54 fixed point hamiltonian is @xmath23 .
indeed , it describes a true sc state , where the resistivity @xmath0 vanishes in the limit @xmath17 .
a key feature of the sc state is that the charge current , dominated by the superconducting component @xmath64 is an _ almost conserved _ quantity .
formally , this is manifested by the vanishing of its commutator with the low energy hamiltonian @xmath23 , which implies that in the absence of the phase slip terms @xmath43 , the current can not degrade ( @xmath65 ) and hence the resistivity vanishes . the leading contribution to @xmath0
can therefore be obtained perturbatively in @xmath43 .
as we show in detail in appendix a , the calculation of @xmath0 can be viewed as a particularly simple application of the more general _ memory matrix _ approach@xcite , which directly implements this insight .
in essence , this approach incorporates a recasting of the standard kubo formula for the conductivity matrix ( a highly singular entity in the case of an almost perfectly conducting system ) in terms of an object named a memory matrix " .
the latter corresponds to a matrix of decay - rates of the slowest modes in the system , and is perturbative in the irrelevant terms in the hamiltonian , in particular all processes responsible for degrading the currents and hence generating a finite resistivity .
the separation , in this approach , of the slow modes generated by the irrelevant operators around @xmath23 from the fast modes allows a controlled approximation as the temperature is lowered and provides a lower bound on the conductivity @xcite . our derivation of the d.c .
electric resistivity ( see appendix a for details ) yields the following expression : @xmath66 where @xmath67 is the static susceptibility @xmath68 and @xmath69 ( the memory function ) can be expressed , to leading order in the perturbations @xmath70 , in terms of correlators of the ` force ' operators @xmath71\ ; , \end{aligned}\ ] ] which dictate the relaxation rate of the current @xmath72 via @xmath73 this yields @xmath74 in which @xmath75 is the retarded correlation function , the expectation value being evaluated with respect to @xmath76 . to leading order in perturbation theory , we evaluate the expectation value in eq .
( [ chi - ee ] ) as well with respect to the low energy hamiltonian @xmath23 .
this yields @xmath77 using eqs .
( [ je : main ] ) , ( [ hps ] ) and ( [ fps - ee ] ) we obtain an expression for the force operator @xmath78 inserting in eq .
( [ mmee ] ) , we find @xmath79 where @xmath80 in which the green s function @xmath81 at finite @xmath1 is given by @xcite @xmath82 \\ \ , & + & \frac{1}{4k}\ln \left[\frac{\pi at / v } { \sinh\{\pi t(x+v t - ia)/v\}}\right]\ ; . \nonumber\end{aligned}\ ] ] to find the leading @xmath1dependence of @xmath0 for small @xmath83 and @xmath84 , we neglect the contributions of @xmath85 . substituting the resulting @xmath69 combined with @xmath67 from eq .
( [ chi ] ) in eq .
( [ rho - ee ] ) , we obtain @xmath86^{2/k-3}\end{aligned}\ ] ] where @xmath87 is the gamma function .
this essentially recovers ( up to a numerical prefactor ) the result of ref .
[ ] , which indeed corresponds to the homogeneous wire limit .
note that the resulting @xmath0 exhibits a sc behavior as long as @xmath88 , in accord with the renormalization group analysis of ref .
we next turn our attention to the more realistic situation , where inhomogeneities along the sc wire are allowed .
random fluctuations are possible in all the wire parameters , and in particular the diameter may varry in space leading to a local cross section @xmath89 , which can be assumed to be a random function of @xmath90 .
the most prominent modification of the hamiltonian in the presence of such spatial fluctuations is manifested in the fugacity @xmath83 , which depends exponentially on the wire diameter via the core action @xmath45 .
consequently , it becomes space dependent , i.e. @xmath91 where @xmath92 where @xmath93 is a random correction to the uniform core action @xmath94 . the leading ( @xmath95 ) phase slip contribution to the hamiltonian [ eq .
( [ hps ] ) ] now becomes @xmath96 where @xmath97 .
assuming a gaussian distribution of the random function @xmath93 @xmath98={\cal n}\exp \left\{- \frac{1}{2ad}\int dx \lambda^2(x ) \right\}\ ] ] yields the disorder averages @xmath99 , @xmath100 and @xmath101 e^{\lambda(x ) } e^{\lambda(x^\prime ) } \nonumber\\ & = & e^{d } + a\delta(x - x^\prime ) \left [ e^{2d } - e^{d } \right]\ ; .\end{aligned}\ ] ] the last result is derived from the discrete version of the above functional integral , which leads to @xmath102 [ here @xmath103 .
note that the parameter @xmath2 characterizes the degree of granularity in the sc wire , with @xmath104 proportional to the typical amplitude of spatial fluctuations in the wire cross section .
the inhomogeneous phase
slip term eq .
( [ hdis ] ) modifies the expression for the force operator @xmath105 substituting in eq .
( [ mmee ] ) and performing the disorder averaging using the correlation function ( [ lam_corr ] ) we find @xmath106 \int_{0}^{\infty } dt \,t\,{\rm im}\{c_{ps,1}^{ee}(0,t)\ } \nonumber\end{aligned}\ ] ] and @xmath107 is given by eq . ( [ cee ] ) for @xmath95 .
this yields the leading @xmath1dependence of the resistivity for arbitrary granularity @xmath2 in the sc wire : @xmath108^{2/k-3 } \\ & \times & \left[a_{h}e^{d}+a_{g}\left ( e^{2d } - e^{d } \right ) \left(\frac{2\pi at}{v}\right)\right]\ ; , \nonumber \\ a_{h}&=&\frac{\pi^2\gamma^4(1/2k)}{\gamma^2(1/k)}\ ; , \quad a_{g}=\frac{\pi^2\gamma^2(1/k)}{4\gamma(2/k ) } \nonumber \ , .\end{aligned}\ ] ] for extremely weak granularity where @xmath109 , the first term in ( [ rho - final ] ) dominates , and the homogeneous result eq .
( [ rho - hom - final ] ) is recovered . however , for @xmath2 of order 1 or more , the second term is _ exponentially enhanced _ by the factor @xmath110 , yielding @xmath111^{2/k-2 } \ , .\end{aligned}\ ] ] this approximation is consistent with earlier predictions for granular sc wires@xcite .
indeed , it indicates that the phase slips dominating the resistivity occur at narrow constrictions in the inhomogeneous wire .
our more general expression eq .
( [ rho - final ] ) implies that for a fixed @xmath2 , the resistivity @xmath0 vs. @xmath1 exhibits a crossover from a `` homogeneous '' power law behavior [ eq .
( [ powerlaw ] ) with @xmath112 to a `` granular '' limit with a higher exponent @xmath113 . in most realistic systems
we expect this crossover to occur at very low temperatures : this is in view of the exponential dependence on the granularity parameter @xmath2 .
the typical core action @xmath94 has been estimated to be of order @xmath114 or larger@xcite ; this implies that even small irregularities ( @xmath115 ) in the wire diameter corresponding to @xmath116 are sufficient to enhance the prefactor of the second term in ( [ rho - final ] ) by at least an order of magnitude .
as shown in the previous section , the resistivity of a sc wire with a moderately non uniform diameter [ eq . ( [ rho - gran ] ) ] far enough below @xmath5 is expected to be well approximated by a power
law @xmath1dependence of the form ( [ powerlaw ] ) .
in particular , the exponential enhancement by the granularity parameter @xmath2 partially compensates for the exponential suppression by the fugacity @xmath117 , and consequently the prefactor [ @xmath20 in eq .
( [ powerlaw ] ) ] becomes comparable to the observed resistivity in a number of experiments . in this section
we test the relevance to available experimental data by a direct attempt to fit a power law . in the cases where the fit appears to be reasonably good
, we can extract the luttinger parameter @xmath56 from the exponent @xmath113 and compare it to an independent estimate based on the sample parameters . according to eq .
( [ k - def ] ) , @xmath56 depends in particular on the wire diameter @xmath118 , the capacitance per unit length @xmath39 and the superfluid density @xmath40 , which can be expressed in terms of the london penetration depth @xmath29 : @xmath119 here @xmath120 is a numerical factor which depends on the geometry of the wire cross section ( e.g. , @xmath121 in ref . [ ] and
@xmath122 in ref .
[ ] ) , and @xmath39 is typically of order unity@xcite and depends very weakly ( i.e. logarithmically ) on @xmath6 ; we hereon regard it as a fitting parameter .
the dependence on @xmath29 implies that a smooth tuning of @xmath56 is possible by application of a magnetic field parallel to the wire .
indeed , this was done by altomare _
et al . _ ( ref . [ ] ) , as will be discussed in more detail below .
we have considered data obtained in several different experimental setups : indium strips studied by giordano ( ref . [ ] ) , moge wires deposited on carbon nanotubes studied by the harvard group ( ref .
[ ] ) , more recent studies of tin ( sn ) nanowires ( tian _ et al . _ , ref . [ ] ) and long aluminum wires ( altomare _ et al .
_ , ref .
all of these indicate substantial deviation from the thermal activation theory @xcite , attributed to qps .
in addition , we note that in most of the sc samples involved in those experiments the normal resistivity is too low to be considered in the strictly granular limit ( where distinct grains are weakly coupled ) , although the wire diameter is likely to be non uniform with varying degree of nonuniformity .
the samples of ref .
[ ] exhibit a rather complicated behavior , following from a number of reasons .
first , most of the wires fabricated in this particular methods are relatively short , hence finite size effects interfere with the @xmath1dependence , and the dynamics of phase slips is crucially affected by their backscattering from the boundaries @xcite . in addition , some of the nanotube substrates are not insulating , and their ( unknown ) resistance complicates the fitting by additional parameters . hence , although some of these samples can be fitted reasonably well by eq .
( [ powerlaw ] ) , a more detailed quantitative analysis has focused on the other experimental papers . presenting the resistance vs. @xmath1 , re
plotted on a log log scale ; the solid lines correspond to fitting functions of the form @xmath123 .
( a ) in the low@xmath1 ( qps ) regime : blue circles ( red squares ) correspond to the samples of diameter @xmath124 ( @xmath125 ) ; the corresponding exponents are @xmath126 and @xmath127 , respectively . ( b ) the @xmath124 and @xmath125 diameter samples in the high@xmath1 regime , and all data points of the @xmath128 diameter sample ( green diamonds ) ; @xmath129 , @xmath130 and @xmath131 , respectively.,title="fig:",width=302 ] presenting the resistance vs. @xmath1 , re plotted on a log log scale ; the solid lines correspond to fitting functions of the form @xmath123 .
( a ) in the low@xmath1 ( qps ) regime : blue circles ( red squares ) correspond to the samples of diameter @xmath124 ( @xmath125 ) ; the corresponding exponents are @xmath126 and @xmath127 , respectively . ( b ) the @xmath124 and @xmath125 diameter samples in the high@xmath1 regime , and all data points of the @xmath128 diameter sample ( green diamonds ) ; @xmath129 , @xmath130 and @xmath131 , respectively.,title="fig:",width=302 ] we first consider the data of ref .
[ ] . fig .
( [ fig : gio ] ) presents the resistance of two samples as a function of @xmath1 on a log log scale .
note that the length of the shortest wire studied is 80 @xmath132 m . in comparison ,
the effective length set by the temperature scale @xmath133 for the relevant @xmath134 is of order 10 @xmath132 m ; this follows from the estimate @xmath135 , where @xmath136 is the velocity of light and the measured @xmath137 obtained for indium films of comparable thickness @xcite .
the wires therefore fulfill the long wire condition @xmath138 , and can be considered good candidates for testing the scaling of @xmath0 with @xmath1 . in the low @xmath1 regime [ fig .
( [ fig : gio](a ) ) ] , the fit to a power
law is very good@xcite .
in addition , the values of @xmath4 corresponding to the two samples of diameters @xmath124 and @xmath125 yield the luttinger parameters @xmath139 and @xmath140 , respectively . these are consistent with the values obtained by inserting the relevant @xmath6 and @xmath29 in eq .
( [ k - dlambda ] ) , provided @xmath141 . in principle , one may argue that a fit by a power law with a high exponent ( especially within a limited range of @xmath1 ) is not easily distinguishable from an exponential function .
however , to put the above analysis to a further test , we tried to fit the data obtained in the high @xmath1 regime ( in the close vicinity of @xmath5 ) to a power law as well . in this regime , there is no justification for such a functional dependence on @xmath1 , as the thermal activation theory for phase
slips@xcite is expected to work much better .
indeed , the result depicted in fig .
( [ fig : gio](b ) ) indicates an obvious failure of a @xmath123 trial function : the best fit to a straight line on a log log plot yields an anomalously large exponent ( @xmath4 ranging from @xmath142 to @xmath143 ) , and a systematic deviation from the fitting function .
this is in sharp contrast with the situation in the low@xmath1 regime , and therefore strengthen our confidence that in the latter case , the analysis is valid and provides a meaningful confirmation of the theoretical prediction . presenting the resistance vs. @xmath1 , re plotted on a log log scale : sample diameters are 20 nm ( blue circles ) , 40 nm ( red squares ) , and 60 nm ( green diamonds ) .
the solid lines correspond to fitting functions of the form @xmath123 , with @xmath144 , @xmath145 and @xmath146 , respectively.,width=302 ] we next consider the data of tian _ et al _ ( ref . [ ] ) .
the results obtained from resistivity measurements in three different tin ( sn ) nanowires of diameters ranging between 20 nm and 60 nm are depicted in fig .
( [ fig : tian ] ) . similarly to giordano s data , in all three samples a kink " is observed at some temperature @xmath147 , below which the decrease of @xmath148 as @xmath1 is lowered becomes more moderate . in the two thinner samples ,
the low@xmath1 section of the data is quite noisy , and @xmath149 seems to saturate .
this behavior is possibly due to serial contribution from the contacts . otherwise , however , the power
law dependence appears to be a good fit for @xmath11 . to test the validity of eq .
( [ k - dlambda ] ) in this system , one requires an independent estimate of @xmath29 .
a direct measurement is not available , however , an indirect estimate of the ratio @xmath150 based on a measurement of the effective critical magnetic field@xcite yields @xmath151 for @xmath152 nm , which is close in diameter to the thickest sample of fig .
( [ fig : tian ] ) .
this is consistent with eq .
( [ k - dlambda ] ) for @xmath153 , a quite reasonable estimate for the capacitance .
unfortunately , reliable data on @xmath150 in the thinner samples is not available . a naive extrapolation of the @xmath6dependence of @xmath29 to lower values of @xmath6 based on the landau ginzburg theory@xcite ( @xmath154 ) yields @xmath155 for the @xmath156 nm sample , a reasonable approximation to the value of @xmath56 obtained from the fit ( @xmath157 ) , however the expected scaling with @xmath6 fails in the case of the @xmath158 nm sample .
indeed , as shown explicitly in ref .
[ ] , the naive scaling @xmath159 works well above @xmath160 nm , but breaks down for the @xmath161 nm wire ( no data is given for even thinner samples ) .
we finally focus our attention on the most recent experimental work of altomare _
_ ( ref . [ ] ) .
this group has studied long and thin aluminum wires , which appear to be ideal candidates for comparison with the qps theory .
in addition , the application of a magnetic field @xmath55 enables a continuous tuning of the superfluid density in a single sample of fixed @xmath6 .
following eq .
( [ k - def ] ) , we therefore expect the @xmath55dependence of @xmath56 to be related to @xmath40 via @xmath162 .
the functional dependence of @xmath163 on @xmath55 can be derived from a simple calculation using the landau
ginzburg theory at finite magnetic field@xcite .
this yields @xmath164 where @xmath165 is the bulk critical field , and @xmath166 a numerical factor which depends on the geometry : e.g. , @xmath167 in a thin film of thickness @xmath6 , and @xmath168 in a cylindrical wire of diameter @xmath6 .
@xmath169 marks an estimated critical field where the inverse luttinger parameter is expected to vanish according to @xmath170 ) , plotted in terms of the inverse luttinger parameter ( @xmath171 ) as a function of magnetic field @xmath55 .
all data points correspond to the lowest temperature measurements ( @xmath172 k ) .
the solid curve corresponds to the formula eq .
( [ kvsh ] ) , with @xmath173.,width=302 ] the unpublished notes included in ref .
[ ] enable a direct comparison of the experimental data to the above prediction .
the experimental results include the ohmic resistance as well as current voltage characteristics which include a non linear part . in the notes
, it is shown that the latter can be fitted to a power
law @xmath174 .
this is suggested by the authors as a plausible alternative to the phenomenological exponential expression used as a fitting function in the published letter .
such power law in the @xmath175 characteristic is consistent with a @xmath1dependence of the ohmic resistance of the form @xmath176 , with @xmath177 .
the exponent @xmath178 is then extracted from the data for different values of the magnetic field , and plotted as a function of @xmath55 . using the linear relation of @xmath178 to @xmath171 , we plot the data in the form depicted in fig .
( [ fig : altomare ] ) , where it is fitted to a trial function of the form ( [ kvsh ] ) .
the fit is reasonably successful , and is better than the @xmath55dependence derived from giordano s formula@xcite .
it yields a critical field @xmath173 . to compare with an independent estimate of @xmath169 based on the experimental parameters , we use the dirty limit expression for the bulk critical field@xcite @xmath179 ( with @xmath7 the sc coherence length in the dirty limit ) . employing eq .
( [ ns - h ] ) we observe that , up to the numerical constant @xmath166 , @xmath169 is essentially determined by @xmath6 and @xmath7 only .
the experimental parameters mentioned in ref . [ ] imply @xmath180 , @xmath181 ( the latter , however , relies on the text book expression for @xmath7 in a dirty aluminum and the measured mean free path ) .
the wires are actually thin strips with a rectangular cross section , hence @xmath166 can not be determined accurately , but is expected to be intermediate between a thin film and a wire , yielding @xmath169 between @xmath182 and @xmath183 .
the order of magnitude is consistent with the value extracted from fig .
( [ fig : altomare ] ) .
considering the fact that the above estimate of @xmath169 does not rely on any fitting parameters , this is a reasonably good agreement .
we have studied the low @xmath1 resistivity of a dirty sc wire using a memory formalism approach .
this method allows a perturbative treatment of corrections to the low energy effective theory , describing the dynamics of phase fluctuations in the sc order parameter . in the limits of either ideally homogeneous wire or strongly inhomogeneous ( granular ) wire , our results for @xmath0
recover the power law behavior @xmath184 derived in earlier theoretical literature using the instanton technique .
we show that more generally , the expression for @xmath0 interpolates between the two limits , where the relative weights of each are smoothly tuned by a `` granularity parameter '' @xmath2 describing the typical fluctuations in the diameter of the wire . in particular , the inhomogeneity of the wire leads to an exponential enhancement of the phase slips induced resistivity .
following these calculations , we infer that the low energy theory is likely to be relevant to experimental measurements of @xmath0 in the low @xmath1 , qps regime .
we directly test the validity of the power
law ansatz as a fit to experimental data obtained in a variety of different samples , and found a very good agreement in sufficiently long wires .
the exponent @xmath4 extracted from the @xmath0 data is found to be consistent with the values obtained from an independent estimate , based on known parameters of the corresponding samples ( as much as the information was available ) .
in addition , its dependence on a parallel magnetic field is consistent with the theoretical expectation .
we conclude that the low@xmath1 theory for dirty sc wires provides a plausible interpretation of data presented in the experimental literature , which is better justified than previously used phenomenological effective models .
our conclusions will hopefully encourage a further , more systematic investigation of the comparison between theory and experiment .
we wish to thank konstantin arutyunov , sergei khlebnikov , ying liu , yuval oreg , gil rephael , dan shahar , nayana shah and especially achim rosch for numerous useful discussions .
in addition , we thank konstantin arutyunov , ying liu and m. tian for showing us unpublished data . this work was partially supported by a grant from gif , the german israeli foundation for scientific research and development .
in this appendix we review the general aspects of the memory matrix approach@xcite , which turns out to serve as a useful tool for evaluating transport properties in systems where approximate conservation laws can be easily identified . as we show below , the calculation of d.c .
electric resistivity of a sc wire provides a particularly simple example for the application of this approach , which amounts to a straightforward perturbative treatment of any processes responsible for relaxing the transport current . at finite but low @xmath1 ,
the fixed point hamiltonian @xmath23 [ eq . ( [ h0 ] ) ] provides the leading contribution to thermodynamic properties of the sc wire .
however , by itself it does not give access to transport properties : being a translationally invariant integrable model , it possesses an infinite number of conserved currents , including in particular the electric charge transport current . since the current can not degrade , the d.c .
conductivity is infinite even for @xmath185 . to get the leading non trivial contribution to transport , it is therefore necessary to add the irrelevant corrections , leading to a slow but finite relaxation rate of the currents . in our case ,
the prominent corrections are the phase slips terms @xmath43 [ eq . ( [ hps ] ) ] : note that other irrelevant terms , e.g. , of the form @xmath186 , @xmath187 with @xmath188 as well as pair breaking terms , which are already neglected in eq .
( [ hsg ] ) , do not contribute significantly to transport@xcite .
we then evaluate the resistivity @xmath0 employing the memory matrix approach , which takes advantage of the fact that the memory matrix a matrix of decay rates of the slowest modes in the system is perturbative in the irrelevant operators .
this is in contrast with the conductivity matrix , which is a highly singular function of these perturbations .
a crucial step in the derivation of transport coefficients in the memory matrix approach is the identification of primary slowly decaying currents of the system .
these are conserved `` charges '' of the fixed point hamiltonian @xmath23 , whose conservation is slightly violated by certain irrelevant perturbations . in the case of the luttinger model @xcite , these include the charge current @xmath72 [ eq .
( [ je : main ] ) ] and , in addition , the total translation operator @xmath189 in the above , we approximate the currents by contributions from the collective degrees of freedom ( phase fluctuations ) only . non superconducting contributions associated with unpaired electrons are exponentially suppressed as @xmath190 ( with @xmath191 ) for @xmath9 .
the correlators of @xmath72 , @xmath192 determine the conductivity matrix @xmath193 at frequency @xmath194 and temperature @xmath1 via the kubo formula : @xmath195 where following ref .
[ ] we have introduced the scalar product ( of any two operators @xmath120 and @xmath196 ) @xmath197 the d.c .
electrical resistivity is then given by @xmath198 , where @xmath199 is the @xmath200 limit of @xmath201 .
however , a direct application of eq .
( [ sigma - def ] ) at low @xmath202 is rather subtle : since @xmath203=0 $ ] ( for @xmath204 ) , the relaxation rate of the currents @xmath205 $ ] is dictated by the irrelevant corrections , hence tends to vanish in the limit @xmath17 .
this leads to divergences in the conductivities , since the currents do not decay in this limit . to enable a controlled perturbative expansion in the relaxation rates @xmath206
, we therefore recast the conductivity matrix in terms of a memory matrix @xmath207 : @xmath208 in which @xmath209 and @xmath210 is the matrix of static susceptibilities @xmath211 here @xmath212 is the liouville operator defined by @xmath213 $ ] , and @xmath214 is the projection operator on the space perpendicular to the slowly varying variables @xmath215 , @xmath216 note that similarly to ref . [ ] , we choose a convenient definition of the memory matrix [ eq .
( [ m ] ) ] , which is slightly different than the standard literature .
the perturbative nature of @xmath207 is transparently reflected by this expression : in particular , the operators @xmath206 are already linear in the irrelevant corrections to @xmath23 .
this enables a systematic perturbative expansion of the correlator in eq .
( [ m ] ) in the small parameter characterizing the relative size of the corrections in our case , the exponential factors @xmath217 [ see eq .
( [ hps ] ) ] .
we now obtain an expression for the d.c .
electric resistivity by setting @xmath218 in eq .
( [ sigma - t ] ) .
this yields @xmath219 where @xmath207 is evaluated from eq .
( [ m ] ) in the limit @xmath200 .
( note that here we have used the fact that both matrices @xmath220 and @xmath221 are symmetric . )
the relaxation rate operators @xmath206 are given by @xmath222 where we have defined the ` force ' operators @xmath223\ ; .\end{aligned}\ ] ] to leading order in the perturbations @xmath43 , eq .
( [ m ] ) for @xmath207 is greatly simplified : one can set @xmath224 with @xmath225 $ ] and @xmath226 .
this yields @xmath227 where the elements of @xmath228 are given by @xmath229 in which @xmath230 is the retarded correlation function evaluated with respect to @xmath76 . substituting eqs .
( [ jt ] ) and ( [ hps ] ) in eq .
( [ fps - def ] ) and taking the @xmath231 limit , it is easy to see that @xmath232 .
indeed , this follows from the translational invariance of @xmath43 . as a result , @xmath233 and
@xmath234 identically vanish .
this implies that _ if _ both @xmath235 and @xmath69 do not vanish , eq .
( [ rho - m ] ) yields @xmath236 _ even at finite @xmath1 _ ; namely , the creation of free phase slips appears to be insufficient to generate a finite dissipation .
such a result can not be reconciled with our understanding , that dissipation should occur at the normal cores during a phase slips event , where the wire behaves temporarily as a normal dirty metal . indeed , we should recall that the translational invariance of @xmath43 is not of microscopic origin ; rather , these terms in the effective hamiltonian are obtained after averaging over disorder in the electron system@xcite .
their form reflects the total absorption of the finite momentum @xmath237 generated in a phase slip process by the core electrons . in a clean sc wire ,
such momentum transfer is not effective , leading to an exponential suppression of @xmath233 , @xmath69 and yielding a finite but exponentially small resistivity @xmath238 ( see ref . [ ] ) . stronger disorder in the underlying electronic system
should only enhance the resistivity , and not vice versa !
this apparent paradox is resolved by the remarkable observation@xcite that , following the commensurate relation between the total momentum @xmath239 and density per unit length , @xmath235 actually _ vanishes identically _ , even at finite @xmath1 .
setting @xmath240 in eq .
( [ rho - m ] ) , we then find that the expression for the resistivity @xmath0 reduces to the simplified form eq .
( [ rho - ee ] ) .
the above arguments are accurate as long as indeed the currents @xmath72 , @xmath192 are given by eqs .
( [ je : main ] ) , ( [ jt ] ) , i.e. , when normal electron contributions to the currents are neglected . however , if such normal contributions are not negligible , they would at the same time modify the translation invariance of the hamiltonian , and all entries in eq .
( [ rho - m ] ) would be finite .
similarly , translational invariance is explicitly broken once we account for random spatial fluctuations in the sc wire cross section , by introducing the inhomogeneous phase
slip term eq .
( [ hdis ] ) .
this modifies the force operator @xmath241 [ now given by eq .
( [ fe - dis ] ) ] , and induces a non trivial contribution to @xmath242 : @xmath243 however , the resulting matrix @xmath220 obtained after substitution in the approximate form eq .
( [ mmm ] ) is still diagonal : to see this , we note that the off - diagonal element is given ( after disorder averaging ) by @xmath244 where @xmath245 in which @xmath81 is given by eq .
( [ green - t0 ] ) .
it is apparent from eqs .
( [ cet ] ) and ( [ green - t0 ] ) that @xmath246 is an antisymmetric function of @xmath90 , and hence the integral over @xmath90 in eq .
( [ met ] ) vanishes , yielding @xmath247 .
the static susceptibility matrix [ eq .
( [ chi - def ] ) ] evaluated to leading order in the perturbation @xmath248 is also approximately diagonal , with @xmath67 given by eq .
( [ chi ] ) .
we therefore find that the leading contribution to eq .
( [ rho - m ] ) for the resistivity again practically reduces to eq .
( [ rho - ee ] ) .
it should be pointed out that the crossover temperature @xmath249 is not too far from the value of @xmath5 as quoted by giordano in ref .
[ ] , which corresponds to the thin film critical temperature , assumed to apply for the wire as well . as a consequence
, one might expect a @xmath1dependence of the exponent @xmath4 associated mainly with the @xmath1dependence of @xmath29 near @xmath5 .
such deviation from a pure power
law is , however , not supported by the experimental data for @xmath250 . a possible resolution for that might be that the actual @xmath5 in a wire and film are actually quite different .
indeed , @xmath5 depends crucially on dimensionality : for example , it is significantly enhanced in a thin film compared to the bulk ( three dimensional ) material . in practice
, @xmath5 is often treated as a fitting parameter ( see , e.g. , ref . | we study the contribution of quantum phase fluctuations in the superconducting order parameter to the low temperature resistivity @xmath0 of a dirty and inhomogeneous superconducting wire . in particular
, we account for random spatial fluctuations of arbitrary size in the wire thickness . for a typical wire thickness above the critical value for superconductor insulator transition ,
phase slips processes can be treated perturbatively .
we use a memory formalism approach , which underlines the role played by weak violation of conservation laws in the mechanism for generating finite resistivity .
our calculations yield an expression for @xmath0 which exhibits a smooth crossover from a homogeneous to a `` granular '' limit upon increase of @xmath1 , controlled by a `` granularity parameter '' @xmath2 characterizing the size of thickness fluctuations .
for extremely small @xmath2 , we recover the power
law dependence @xmath3 obtained by unbinding of quantum phase slips .
however in the strongly inhomogeneous limit , the exponent @xmath4 is modified and the prefactor is _ exponentially enhanced_. we examine the dependence of the exponent @xmath4 on an external magnetic field applied parallel to the wire . finally , we show that the power
law dependence at low @xmath1 is consistent with a series of experimental data obtained in a variety of long and narrow samples .
the values of @xmath4 extracted from the data , and the corresponding field dependence , are consistent with known parameters of the corresponding samples . |
as widely discussed in many papers , the spectrum of the cosmic microwave background ( cmb ) carries unique informations on physical processes occurring during early cosmic epochs ( see e.g. danese & burigana 1993 and references therein ) .
the comparison between models of cmb spectral distortions and cmb absolute temperature measures can constrain the physical parameters of the considered dissipation processes . by jointly considering distortions generated in a wide range of early or intermediate cosmic epochs and at late cosmic epochs , we recently discussed the implications of the current cmb spectrum data ( salvaterra & burigana 2002 ) as well as the great improvements in the knowledge of dissipation processes , in particular on those possibly occurred at early and intermediate epochs , achievable with a significant progress in the experiments at long wavelength , @xmath6 cm , ( burigana & salvaterra 2003 ) , such that expected from a _ dimes_-like project ( kogut 1996 , 2003 ) .
fixsen & mather ( 2002 ) recently proposed a new generation of millimeter and submillimeter spectrum experiment , firas ii , aimed to improve by a factor @xmath11 the accuracy of the cobe / firas measures . while in the last decade many efforts have been ( see e.g. bennett et al .
1996 , 2003 and references therein ) and are currently ( see e.g. mandolesi et al .
1998 , puget et al .
1998 , tauber 2000 and references therein ) dedicated to greatly improve our knowledge of the whole sky cmb anisotropies since their cobe / dmr discovery ( smoot et al . 1992 ) , the combination of high precision and large frequency coverage jointly forseen for _ dimes _ and firas ii opens a completely new perspective also in cmb spectrum studies , as we will discuss in this paper . in section 2
we briefly summarize the general properties of the cmb spectral distortions and the main physical informations that can be derived from the comparison with the observations . in section 3
we briefly report on the performances forseen for the _ dimes _ and firas ii experiments and describe the generation of the simulated observations used in this work .
the implications of such experiments for our knowledge of the thermal history of the universe are presented in section 4 : models including an undistorted spectrum and distorted spectra in the case of a single kind of spectral distortion or by jointly considering three kinds of spectral distortions are compared with data simulated according to the sensitivities quoted for the above experiments ( the joint analysis of two spectral distortions is reported in appendix a for completeness ) .
section 5 is dedicated to a comparison between some different classes of astrophysical and particle processes recently discussed in the literature occurring at relatively late epochs , able to distort the spectrum at different wavelength regions , and , possibly , to significantly contribute to the reionization of the universe .
finally , we discuss the results and draw our main conclusions in section 6 .
the cmb spectrum emerges from the thermalization redshift , @xmath12 , with a shape very close to a planckian one , owing to the strict coupling between radiation and matter through compton scattering and photon production / absorption processes , radiative compton and bremsstrahlung , which were extremely efficient at early times and able to re - establish a blackbody ( bb ) spectrum from a perturbed one on timescales much shorter than the expansion time ( see e.g. danese & de zotti 1977 ) .
the value of @xmath13 ( burigana , danese & de zotti 1991a ) depends on the baryon density ( in units of the critical density ) , @xmath14 , and the hubble constant , @xmath15 , through the product @xmath16 ( @xmath15 expressed in km / s / mpc ) . on the other hand ,
physical processes occurring at redshifts @xmath17 may lead imprints on the cmb spectrum .
the timescale for the achievement of kinetic equilibrium between radiation and matter ( i.e. the relaxation time for the photon spectrum ) , @xmath18 , is @xmath19 where @xmath20 is the photon
electron collision time , @xmath21 , @xmath22 being the electron temperature and @xmath23 ; @xmath24 is the mean fractional change of photon energy in a scattering of cool photons off hot electrons , i.e. @xmath25 ; @xmath26 is the present radiation temperature related to the present radiation energy density by @xmath27 ; a primordial helium abundance of 25% by mass is here assumed .
it is useful to introduce the dimensionless time variable @xmath28 defined by @xmath29 where @xmath30 is the present time and @xmath31 is the expansion time given by @xmath32^{-1/2 } \sec \ , , \ ] ] @xmath33 being the redshift of equal non relativistic matter and photon energy densities ( @xmath34 is the density of non relativistic matter in units of critical density ) ; @xmath35 , @xmath36 being the number of relativistic , 2component , neutrino species ( for 3 species of massless neutrinos , @xmath37 ) , takes into account the contribution of relativistic neutrinos to the dynamics of the universe , is itself a function of the amount of energy dissipated .
the effect , however , is never very important and is negligible for very small distortions . ] .
burigana , de zotti & danese 1991b have reported on numerical solutions of the kompaneets equation ( kompaneets 1956 ) for a wide range of values of the relevant parameters and accurate analytical representations of these numerical solutions , suggested in part by the general properties of the kompaneets equation and by its well known asymptotic solutions , have been found ( burigana , de zotti & danese 1995 ) .
the cmb distorted spectra depend on at least three main parameters : the fractional amount of energy exchanged between matter and radiation , @xmath38 , @xmath39 being the radiation energy density before the energy injection , the redshift @xmath40 at which the heating occurs , and the baryon density @xmath41 .
the photon occupation number can be then expressed in the form @xmath42 where @xmath43 is the dimensionless frequency @xmath44 ( @xmath45 being the present frequency ) , and @xmath46 characterizes the epoch when the energy dissipation occurred , @xmath40 being the corresponding redshift ( we will refer to @xmath46 computed assuming @xmath47 , so that the epoch considered for the energy dissipation does not depend on the amount of released energy ) .
the continuous behaviour of the distorted spectral shape with @xmath48 can be in principle used also to search for constraints on the epoch of the energy exchange . of course , by combining the approximations describing the distorted spectrum at early ( sunyaev & zeldovich 1970 ) and intermediate epochs with the comptonization distortion expression ( zeldovich & sunyaev 1969 ; zeldovich , illarionov & sunyaev 1972 ) describing late distortions , it is possible to jointly treat two heating processes ( see burigana et al .
1995 and salvaterra & burigana 2002 and references therein for a more exhaustive discussion ) . in this work simulated measures of the cmb absolute temperature are compared with the above models of distorted spectra by using a standard @xmath49 analysis .
we determine the limits on the amount of energy possibly injected in the cosmic background at arbitrary primordial epochs corresponding to a redshift @xmath40 ( or equivalently to @xmath48 ) . this topic has been discussed in several works ( see e.g. burigana et al .
1991b , nordberg & smoot 1998 , salvaterra & burigana 2002 ) .
we apply here the method of comparison with the theoretical models described in salvaterra & burigana 2002 and burigana & salvaterra 2003 , by investigating the possibility of properly combining data simulated according to the _ dimes _ and firas ii performances respectively at long and short wavelengths .
we will consider the recent improvement in the calibration of the firas data , that sets the cmb scale temperature at @xmath50 k at 95 per cent confidence level ( cl ) ( mather et al .
then , we study the combined effect of two different heating processes that may have distorted the cmb spectrum at different epochs .
this hypothesis has been also taken into account by nordberg & smoot 1998 , who fit the observed data with a spectrum distorted by a single heating at @xmath51 , a second one at @xmath52 and by free - free emission , obtaining limits on the parameters that describe these processes .
as in salvaterra & burigana 2002 , we extend their analysis by considering the full range of epochs for the early and intermediate energy injection process , by taking advantage of the analytical representation of spectral distortions at intermediate redshifts ( burigana et al .
long wavelength measures are particularly sensitive to the free - free distortion due to its well known dependence on the frequency . at moderately long wavelengths , where quite accurate observations can be carried out , @xmath53 . here
@xmath54 is the ratio between the electron and radiation temperature before the beginning of the dissipation process , @xmath5 is the so called free - free distortion parameter ( burigana et al .
1995 ) , and @xmath55 is the ( frequency dependent ) equivalent thermodynamic temperature . at very long wavelengths
the equivalent thermodynamic temperature approaches the electron temperature because of the very high bremsstrahlung efficiency .
the relationship between free - free distortion and comptonization distortion is highly model dependent , being related to the details of the thermal history at late epochs ( see e.g. danese & burigana 1993 , burigana et al .
1995 ) , and can not be simply represented by integral parameters . in the combined analysis of the implications of future cmb spectrum experiments for energy dissipations at different cosmic times and free - free distortions we avoid to physically link comptonization and free - free distortions .
( 4 ) generalizes then to @xmath56 comptonization and free - free emission could be also produced in several astrophysical scenarios .
for example , ionized halos in the early stages of galaxy formation ( oh 1999 ) can produce a relevant free - free distortions and coupled comptonization and free - free distortions can be produced in the reionization epoch , indicated by the _ wmap _ data ( kogut et al .
1993 ) .
it is also possible to extend the limits on @xmath3 for heatings occurred at @xmath57 , where @xmath58 is the redshift corresponding to @xmath59 , when the compton scattering was able to restore the kinetic equilibrium between matter and radiation on timescales much shorter than the expansion time and the evolution on the cmb spectrum can be easily studied by replacing the full kompaneets equation with the differential equations for the evolution of the electron temperature and the chemical potential .
this study can be performed by using the simple analytical expressions by burigana et al .
1991b instead of numerical solutions . to evaluate the scientific impact represented by the future experiment improvements , we create different data sets simulating the observation of a distorted and non distorted spectra from a space experiments like _ dimes _ and firas ii through the method described in section 3.1 .
each data set will be then compared to models of distorted spectra by using the method described in salvaterra & burigana 2002 ( see also burigana & salvaterra 2000 for the details of the code ) to recover the values of the parameters appearing in eq . ( 5 ) . for simplicity ,
we restrict to the case of a baryon density @xmath60 our analysis of the implications for the thermal history of the universe , but the method can be simply applied to different values of @xmath41 .
the cmb spectrum experiments considered here are dedicated to improve our knowledge both at the same and at longer wavelengths with respect to the firas frequency coverage ( @xmath61 cm @xmath62 cm ) . as representative cases , and without the ambition to cover the whole set of planned experiments , we briefly refer here to the _ dimes _ experiment from space ( kogut 1996 ) designed to reach an accuracy close to that of firas up to @xmath63 cm and to the firas ii experiment ( fixsen & mather 2002 ) which will allow a sensitivity improvement by a factor @xmath64 with respect to firas .
_ dimes _ ( diffuse microwave emission survey ) is a space mission submitted to the nasa in 1995 , designed to measure very accurately the cmb spectrum at wavelengths in the range @xmath65 cm ( kogut 1996 ) .
_ dimes _ will compare the spectrum of each 10 degree pixel on the sky to a precisely known blackbody to precision of @xmath66 mk , close to that of firas ( @xmath67 mk ) .
the set of receivers is given from cryogenic radiometers with instrument emission cooled to 2.7 k operating at six frequency bands about 2 , 4 , 6 , 10 , 30 and 90 ghz using a single external blackbody calibration target common to all channels to minimize the calibration uncertainty .
the _ dimes _ design is driven by the need to reduce or eliminate systematic errors from instrumental artifacts .
the _ dimes _ sensitivity represents an improvement by a factor better than 300 with respect to previous measurements at cm wavelengths .
fixsen & mather ( 2002 ) described the fundamental guidelines to significantly improve cmb spectrum measures at @xmath7 cm .
a great reduction of the residual noise of cosmic rays , dominating the noise of the firas instrument , can be obtained by eliminating the data on - board co - add process or applying deglitching before co - adding , by reducing the size of the detectors and by using `` spiderweb '' bolometers or antenna - coupled microbolometers .
they are expected to show a very low noise when cooled below 1 k while ruo sensors can reduce to 0.1 mk the read noise of thermometers .
the lagrangian point l2 of the earth - sun system is , of course , the favourite `` site '' for firas ii .
also , the calibration can be improved by order or magnitudes with respect to that firas by reducing the contribution to the calibrator reflectance of light from the diffraction at the junction between the calibrator and the horn .
a complete symmetrical construction of the instrument is recommended .
this allows the cross - check between calibrators and between calibrators and the sky and to realize `` an end - to - end calibration and performance test before lunch '' .
according to the authors , firas ii can be designed to have a frequency coverage from 60 to 3600 ghz ( i.e. from 5 mm to 83 @xmath68 m ) with a spectral resolution @xmath69 and sensitivity in each channel about 100 times better than firas .
we collect different data sets , simulating measurements of different cmb spectra , distorted or not , at the frequency ranges of the considered experiments . for the cases of distorted spectra
we calculate the theoretical temperature of the cmb spectrum as discussed in the previous section . of course , the equivalent thermodynamic temperature held obviously constant at all the frequencies for the case of a non distorted spectrum . the theoretical temperatures are then fouled to simulate real measurements affected by instrumental noise .
the simulated temperature @xmath70 at the frequency @xmath45 is given by @xmath71 where @xmath72 is the theoretical temperature at the frequency @xmath45 and err(@xmath45 ) is the expected rms error ( at 1 @xmath73 ) of the experiment at this frequency .
the numbers @xmath74 are a set of random numbers generated according to a gaussian distribution with null mean value and unit variance with the routine gasdev by press et al .
1992 ( 7 ) .
the considered frequency channels are those of _
dimes _ and those achievable with the spectral resolution forseen for firas ii . for sake of simplicity and
differently from the analysis carried out in salvaterra & burigana 2002 , we do not consider here the wavelength region at @xmath75 m , accessible to firas ii , and neglect the problem of separating the signatures of cmb spectral distortions , dominant at @xmath6 mm , from the contribution from far - ir galaxies , dominant at @xmath76 m .
we generate a set ( s - bb ) of simulated data in the case of a blackbody spectrum at a temperature of 2.725 k in order to evaluate the joint capability of two experiments like _ dimes _ and firas ii to improve the constraints on the amount of the energy injected in the cosmic radiation field .
the analysis of this case is in fact directly comparable with the results obtained from the fit to the currently available data . of course
, it is extremely important to evaluate the capability of so accurate experiments to detect and , possibly , precisely measure the spectral distortion parameters , the fractional energy exchanges and the dissipation epochs . to this aim
, we generate other three sets of simulated data ( respectively , s - e5c , s - e1.5c , s - e0.1c ) assuming an instantaneous ( @xmath77 ) earlier energy dissipation with @xmath78 ( @xmath79 , where @xmath68 is the chemical potential of the bose - einstein ( be ) formula describing the high frequency tail of the spectrum for distortions at sufficiently high redshifts ) occurring at @xmath59 or at @xmath80 or at @xmath81 plus a later comptonization distortion associated to an energy dissipation @xmath82 ( here @xmath83 is the `` energy '' comptonization parameter ) . for simplicity , we assume here @xmath84 ( see sect . 5 for a discussion of combined comptonization and free - free distortions on the basis of some specific models ) .
all these distorted spectra are computed by setting @xmath85 and @xmath86 km / s / mpc .
these above simulated data set will be compared with models including spectral distortions generated by a dissipation process occurred at very different cosmic epochs , on a grid represented by the dimensionless time @xmath87 5 , 4 , 3 , 2 , 1 , 0.5 , 0.25 , 0.1 , 0.05 , 0.025 , 0.01 , and @xmath88 , plus / or a comptonization distortion , plus / or a free - free distortion , as described in sect . 2 , in order to analyse the quality of the recovery of the input parameters achievable with the considered experiments .
we fit the simulated data set s - bb with a spectrum distorted by a single energy injection at different values of @xmath48 or by a free - free distortion in order to recover the value of @xmath3 or of @xmath5 , expected to be null , and the limits ( at 95 per cent of cl ) on them .
the results are reported in table 1 for a representative set of the considered dissipation epochs .
the table shows also the results obtained by using the firas data alone combined with the long wavelength data accumulated in the last two decades ( see also salvaterra & burigana 2002 ) , the results achievable by combining the firas data with the subset of the data set s - bb achievable by _ dimes _ alone ( see also burigana & salvaterra 2003 ) , and , finally , the results achievable by exploiting the subset of data set s - bb achievable by firas ii alone . the sensitivities reported in table 1 for a given distortion parameter apply in the case in which quite strong priors can be assumed on the other two distortion parameters . in any case , firas ii alone or _
dimes _ alone either is able to improve by more than one order of magnitude the current limits on the free - free distortions and on the energy exchanges .
it is evident from the table that in this case firas ii alone is able to provide the best constraints on the energy exchanges ( with a sensitivity improvement by a factor @xmath11 ) while _ dimes _ alone is able to provide the best constraints on the free - free distortions ( with a sensitivity improvement by a factor @xmath89 ) , without any significant improvement from the combination of the two experiments .
lccccc + data set & & @xmath90 + + & heating at & heating at & heating at & heating at & free - free dist . at + & @xmath51 & @xmath91 & @xmath92 & @xmath52 & @xmath52 + +
f & 20@xmath93533 & 25@xmath93453 & 25@xmath93359 & 28@xmath93233 & 272@xmath93920 + f + r & 64@xmath93532 & 39@xmath93453 & 28@xmath93359 & 28@xmath93233 & -455@xmath93424 + + + d + f & [email protected] & [email protected] & [email protected] & 3.08@xmath93210 & [email protected] + + + f - ii & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + d + f - ii & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + + we discuss here the possibility to significantly improve the constraints on the spectral distortion parameters in the more general case of a joint analysis of an early / intermediate dissipation process occurred at a given @xmath48 , of a late dissipation process ( @xmath94 ) , and of a free - free distortion .
we exploit the same data sets of the previous subsection but without any prior on the distortion parameters .
the results are reported in table 2 ( 95 per cent cl ) for a representative set of the earlier dissipation epoch @xmath48 and displayed in fig . 1 for the whole range of @xmath48 . clearly , the sensitivities on each distortion parameter are worse than the `` analogous '' one in table 1 , but the relative improvements with respect to the current observational status are similar .
although firas ii ( _ dimes _ ) still plays the most relevant role for the energy exchanges ( for the free - free distortions ) , it is interesting to note that in this case the spectral distortion parameter recovery appreciably improves by combining the two experiments .
this is because such levels of accuracy on a so wide frequency range allow to remove the approximate degeneracy both between free - free and be - like distortions and between comptonization and be - like distortions that remain in presence of future significant improvements only at @xmath6 cm or at @xmath7 cm , respectively .
this point will be discussed again in the next subsections , where the analysis of the capability of such experiments to measure spectral distortions possibly present in the cmb spectrum and the corresponding dissipation epochs is presented .
lccccc + data set & & @xmath90 + + & heating at & heating at & heating at & heating at & free - free dist . at + & @xmath51 & @xmath91 & @xmath92 & @xmath52 & @xmath52 + + f & 414@xmath931561 & & & -85@xmath93546 & 635@xmath931531 + f & & 353@xmath931334 & & -88@xmath93560 & 614@xmath931473 + f & & & 286@xmath931091 & -99@xmath93601 & 583@xmath931392 + f + r & -465@xmath931056 & & & 163@xmath93436 & -540@xmath93469 + f + r & & -418@xmath93922 & & 176@xmath93452 & -542@xmath93459 + f + r & & & -343@xmath93789 & 189@xmath93492 & -521@xmath93453 + + + d + f & [email protected] & & & 68.51@xmath93231 & [email protected] + d + f & & [email protected] & & 71.6@xmath93240 & [email protected] + d + f & & & [email protected] & 63.5@xmath93275 & [email protected] + + + f - ii & [email protected] & & & [email protected] & [email protected] + f - ii & & [email protected] & & [email protected] & [email protected] + f - ii & & & [email protected] & [email protected] & [email protected] + d + f
- ii & [email protected] & & & [email protected] & [email protected] + d + f - ii & & [email protected] & & [email protected] & [email protected] + d + f - ii & & & [email protected] & [email protected] & [email protected] + + in this subsection we exploit the simulated data s - e5c , s - e1.5c , and s - e0.1c ( see sect . 3.2 ) to evaluate the quality of spectral distortion parameter recovery achievable with the considered experiments .
we first consider the information achievable by combining _
dimes _ and firas ii . in the case of the data set s - e5c , a fit in terms of two energy exchanges at the epochs assumed in the input model ( @xmath59 and @xmath94 ) shows that the values of @xmath95 can be accurately recovered : for example , in the considered test , we find @xmath96 and @xmath97 ( errors at 95 per cent cl ) with a @xmath49/d.o.f . @xmath98 .
this results are not significantly affected by allowing also for a free - free distortion .
we find in this case : @xmath99 , @xmath100 , and the recovered @xmath101 is clearly in agreement with the null input value ( @xmath49/d.o.f .
@xmath102 ) .
we verified also that a ( wrong ) interpretation of these simulated data in terms of a single kind of spectral distortion is clearly ruled out ( and the recovery of the spectral distortions parameters is , of course , wrong ) . even in the most favourite case of a comptonization distortion
we find a @xmath49/d.o.f .
@xmath103 ( the @xmath49 goes from @xmath104 of previous fits to @xmath105 ) .
analogously , an interpretation in terms of a single energy dissipation plus a free - free distortion is ruled out ( we find @xmath49/d.o.f .
@xmath106 in the most favourite case of a comptonization distortion coupled to a free - free distortion ) .
we then tried to estimate the epoch of the earlier energy injection . by assuming @xmath107 we find a @xmath108 but a degraded spectral distortion parameter recovery .
in particular , the @xmath95 value of the later process is very weakly affected , while that of the earlier process is underestimated of @xmath109 % and a ( wrong ) small negative free - free detection , @xmath110 , appears , although clearly compatible with the input value @xmath84 at @xmath111 .
if we try a fit assuming @xmath112 the results , substantially unchanged in terms of @xmath49 , degrades further from the physical point of view : we find @xmath113 , i.e. a sign no longer compatible ( at @xmath114 ) with that of the , again only slightly affected , recovered comptonization distortion . to verify the physical trouble of such possible sign differences we computed the ( negative ) comptonization distortion parameter produced by simple thermal histories ( namely , with @xmath115 and fully ionized medium ) at pre - recombination epochs or after the recombination able to generate free - free distortions in the range on the negative values found above .
for example , we find @xmath116 by assuming a process in the redshift range @xmath117 and @xmath118 for a process in the redshift range @xmath119 , while a process in the redshift range @xmath120 can produce such negative free - free distortions only for unphysical matter cooling processes involving @xmath121 . since also a cooling process with significant ionization at @xmath122 seems to be excluded by _
wmap _ , the only possible scenarios producing different signs of @xmath5 and @xmath83 should involve or a pre - recombination cooling process able to produce negative comptonization and free - free distortions followed by a late heating process with a very small free - free distortion and a significant comptonization distortion able to properly compensate the previous one , or , in general , particularly ad hoc thermal histories . in general , these possible scenarios should involve accurate fine tunings and appear difficult and untenable and should be discarded in favour of an early energy injection at relevant @xmath48 followed by a late dissipation mechanism . in spite of the insignificant variations of the @xmath49
, the disagreement between the signs of free - free and comptonization distortions due to the wrong assumption on the earlier process epoch , tends to `` physically '' support the ( right ) interpretation of the s - e5c data set involving an energy dissipation at @xmath48 certainly larger than 1 and , probably , larger than 2 or 3 .
we then conclude that reliable estimates of the epochs of the possible dissipation processes are achievable by combining _ dimes_-like and firas ii - like experiments , even for very small energy exchanges . by exploiting the data set s - e1.5c
, we find that a fit in terms of two energy exchanges at the epochs assumed in the input model ( @xmath80 and @xmath94 ) allows to recover the values of @xmath95 with a sensitivity similar to that quoted above for the data set s - e5c . again , allowing for a free - free distortion does not represent a problem and models involving a single kind of spectral distortion or a single energy dissipation plus a free - free distortion can be easily ruled out by simple @xmath49 arguments . concerning the determination of the epoch of the earlier energy injection , in this case the comparison between the recovered signs of comptonization and free - free distortions does not play a relevant role and the direct analysis of the @xmath49 rules out values of @xmath48 smaller than @xmath123 . finally , we analysed the data set s - e0.1c .
we note that the differences between a spectrum resulting from a dissipation at @xmath124 and that produced by a late ( @xmath94 ) process are significantly less evident that those between a spectrum resulting from a dissipation at relevant @xmath48 ( @xmath125 ) and that produced by a late ( @xmath94 ) process .
consequently , we may expect a degradation of the fit result quality .
in fact , we find an appreciable underestimation of the earlier energy dissipation and an analogous overestimation of the later energy dissipation , and error bars significantly larger that in the previous cases : @xmath126 and @xmath127 ( errors at 95 per cent cl ) .
note that , even in this unfavourite case , the recovered values of @xmath95 are in agreement with the input ones at @xmath128 level . also in this case , allowing for a free - free distortion does not degrade further the fit energy exchange recovery and models involving a single kind of spectral distortion or a single energy dissipation plus a free - free distortion can be directly ruled out by simple @xmath49 arguments .
a firm estimate of the epoch of the earlier energy injection can not be achieved in this case : fits involving models with an earlier energy injection at @xmath125 do not show a significant increase of the @xmath49 . on the other hand , from the analysis of the previous cases
we know that the epoch of a possible energy injection at relevant @xmath48 can be quite well estimated from such high quality data
. therefore , in the situation in which a fit to the data in terms of two energy injections at different epochs ( plus , possibly a free - free distortion ) does not allow to estimate the epoch of the earlier process , it is quite reasonable to argue that the earlier process has been occurred at an epoch not particularly early ( by looking at the results found for the data set s - e1.5c , we argue in this case an upper limit @xmath129 ) . we conclude that the combined information contained in future _ dimes _ and firas ii experiments will allow to clearly identify the presence of spectral distortions also in the quite general case of a thermal histories involving early / intermediate and late processes , to measure the corresponding parameters with a good precision even for small amounts ( @xmath130 ) of dissipated energy , and to quite well constrain the epochs at which the processes have been possibly occurred . in the previous subsection we have considered the information achievable by combining _
dimes _ and firas ii .
burigana & salvaterra 2003 discussed the scientific capabilities of accurate long wavelength measures coupled to the available firas data .
we then report here the results obtained by repeating the same analyses carried out above , but by considering the subsets of the data sets s - e5c , s - e1.5c , and s - e0.1c achievable by using firas ii alone . in the case of the data set s - e5c , we find that the two values of @xmath95 can be recovered with an accuracy only slightly degraded with respect to that of the previous section only by performing the fit in terms of two energy exchanges at the epochs assumed in the input model ( the resulting @xmath49/d.o.f .
@xmath131 in this case ) while allowing for a free - free distortion implies a degradation of a factor @xmath8 in the @xmath95 recovery sensitivity .
models with a single spectral distortions can be again ruled out ; the same holds for models involving a single energy dissipation plus a free - free distortion , although with a @xmath49 increase less significant than that found in the previous subsection ( for example , for a fit in terms of a comptonization distortion coupled to a free - free distortion we find @xmath49/d.o.f .
@xmath132 , to be compared with the value of 1.774 found in the previous subsection ) .
results similar to those described above are found by considering the data set s - e1.5c .
the main difference concerns the impact of free - free distortion : by allowing for it in the fit we find in the test a significant underestimation ( @xmath133 % , no longer compatible at @xmath134 with the input one ) on the recovery of the earlier energy injection ( @xmath135 ) and a corresponding ( false ) detection of a negative ( at @xmath136 ) free - free distortion ( @xmath137 ) [ while the effect on comptonization distortions is small ; we find @xmath138 , errors at 95 per cent cl ] .
clearly , by using only high frequency data , the two errors tends to compensate each other , because of their signatures in the final spectrum . the exploitation of the data set s - e0.1c shows analogous results .
the above problem related to the impact of free - free distortion is further enhanced and the parameter recovery sensitivity is significantly degrades : we find @xmath139 , @xmath140 , @xmath141 ( errors at 95 per cent cl ) .
in addition , while models with a single spectral distortions can be again directly ruled out , and the same holds for models without a late energy injection , models involving a single late ( @xmath94 ) energy dissipation plus a free - free distortion can not be longer ruled out on the basis on a direct @xmath49 analysis but on the requirement of having comptonization and free - free distortions with the same sign ( we recover in this case the values @xmath142 , i.e. about corresponding to the sum the energy dissipated by the two processes in order to fit the higher frequency tail of the spectrum , and a negative @xmath5 ( @xmath143 ) that tends to mimic the effect of an earlier energy dissipation ) .
finally , we find that for all the three data sets only a poor information on the epoch of the earlier energy dissipation can be achieved . from the comparison between the results reported in the two previous subsections
we can draw the following conclusions .
a firas ii - like experiment shows a sensitivity to the values of the two energy exchanges at different cosmic times similar to that achieved by combining a firas ii - like experiment and a _ dimes_-like experiment only in presence of a strong prior on the free - free distortion . in the realistic case in which
also @xmath5 should be recovered from the cmb spectrum data , the sensitivity of the spectral distortion parameter recovery will result significantly affected with respect to the case in which both firas ii and a _ dimes _ data are available , with a sensitivity degradation factor of about @xmath144 , related to the epoch of the earlier dissipation process , and possible significant deviations of the recovered values from the input ones .
in addition , the rejection of `` wrong '' ( i.e. different from those assumed in the input model ) models becomes less reliable in the case of earlier dissipations at @xmath145 , being based no longer on direct @xmath49 analyses but on physical requirements about the signs of the recovered free - free and comptonization distortions .
finally , quite accurate estimates of the epochs of the earlier processes are no longer possible . by combining these results with those reported in sect . 4.1 and with those presented in burigana & salvaterra 2003
, we conclude that , although a significant progress of cmb spectrum measures only at @xmath6 cm or at @xmath7 cm certainly implies a great improvement of the current knowledge of thermal history of the universe , only the combination of very precise measures both at @xmath6 cm and at @xmath7 cm will allow to recover the spectral distortion parameters and to map the thermal history of the universe under quite general conditions .
we extend here at @xmath146 ( i.e. @xmath147 ) the constraints on @xmath3 that would be possible to derive at @xmath148 ( @xmath59 ) with the combination of a _
dimes_-like and a firas ii - like experiment .
we remember that at @xmath149 the compton scattering is able to restore , after an energy injection , the kinetic equilibrium between matter and radiation , yielding a be spectrum , and the combined effect of compton scattering and photon production processes tends to reduce the magnitude of spectral distortions , possibly leading to a blackbody spectrum .
the contribution to the thermalization process by a possible cyclotron emission associated to cosmic magnetic fields by including also stimulated emission and absorption ( afshordi 2002 and references therein ) has been recently evaluated for realistic shapes of the distorted spectrum at early epochs ( zizzo 2003 , zizzo & burigana 2003 ) : the strong decreasing of the frequency dependent chemical potential , @xmath150 , of the be - like spectrum ( sunyaev & zeldovich 1970 ) at the very long wavelengths relevant for the cyclotron process implies a large suppression of the cyclotron efficiency .
therefore , only for values of the cosmic magnetic field , @xmath151 , larger than @xmath152 g , well above current observational evidencies and theoretical predictions , this process is found to play a significant role , comparable to ( or larger than ) the combined effect of radiative compton and bremsstrahlung that drives the thermalization process for reasonable values of @xmath151 .
we consider here for simplicity only the case of the simulated observation of a not distorted spectrum , that represents a good example of the possible improvements represented by such kinds of experiments , directly comparable to the results based on the current data .
the comparison is shown in fig . 2 , where the results obtained by considering only simulated data from a _ dimes_-like experiment together with the real firas data ( burigana & salvaterra 2003 ) are also reported .
note how the constraints on @xmath3 can be improved by a factor @xmath153 for processes possibly occurred in a wide range of cosmic epochs , corresponding to about a decade in redshift at @xmath154 about @xmath155 .
it is also evident the improvement of the constraints on @xmath3 at @xmath156 , possible by adding firas ii data to _ dimes _ data . finally , note that the results based on simulated firas ii and _ dimes _ data are obtained by relaxing any assumption on @xmath5 , assumed to be null in the other two cases
( compare also the different cases reported in tables 1 and 2 ) .
of course , large energy injections are still possible at very early epochs close to the thermalization redshift , when primordial nucleosynthesis set the ultimately constraints on energy injections in the cosmic radiation field .
in the previous sections the constraints on free - free and comptonization distortions from the new generation of cmb spectrum experiments have been considered from a general point of view , without assuming detailed physical models for their genesis .
such kinds of late distortions are expected in many astrophysical scenarios involving some energy dissipation mechanisms during various stages of the star and galaxy evolution or some particle or scalar field decay model ( see e.g. de zotti & burigana 1992 , tegmark & silk 1995 ) .
the relevance of their detection ( kogut 2003 ) has been partially renewed by the _ wmap _ satellite discovery of a reionization phase at relevant redshifts , mainly supported by _
wmap _ detection of an excess in the cmb te cross - power spectrum on large angular scale ( multipoles @xmath157 ) , indicating an optical depth to the cmb last scattering surface of @xmath158 ( kogut et al .
2003 ) . without the ambition
to exhaustively cover a so wide topic , we compare here the coupled free - free and comptonization distortions for some of the models recently discussed in the literature and discuss the possibility to reveal them with the new generation of cmb spectrum experiments .
table 3 summarizes the chances of observing the comptonization and free - free distortions for the four classes of processes discussed in the following subsections on the basis of the sensitivities reported in table 2 ( of course , a more optimistic conclusion for the possibility to measure the comptonization distortion , i.e. a sensitivity improvement by a factor @xmath159 , could be derived by assuming a quite strong prior on the earlier energy exchange ; compare the last row of table a1 with the last three rows of table 2 ) .
oh ( 1999 ) predicted the level of free - free distortion can be produced by ionized halos in the early stage of galaxy formation by computing the moments of the intensity distribution from a press - schechter based model .
the predicted free - free distortion results to be @xmath160 k at @xmath103 ghz corresponding to a free - free distortion parameter @xmath161 , well within the sensitivity of _ dimes_. the associated comptonization distortion predicted in this work is below the current firas upper limit by about one order of magnitude , and is then likely measurable by a firas ii - like experiment .
moreover , this work assumed that free - free emitters are the only radio - bright sources in the sky .
the existence of radio - loud galaxies and agns is likely to increase the spectral distortion due to radio sources . the integrated comptonization distortion from quasar driven blastwaves has been recently computed by platania et al .
2002 through a quite general `` energetic '' approach , not particularly related to a detailed physical assumption .
they found @xmath162 , clearly measurable with a firas ii - like experiment , with approximately the same fractional contribution from low-@xmath154 ( 02 ) and high - z ( 24 ) objects .
the evaluation of the corresponding integrated free - free distortion is less model independent , being significantly sensitive to the density profile inside the blastwave because of the @xmath163 dependence of the free - free emission . by assuming the model described in sect . 2 of platania et al .
2002 , we have computed the ratio between the comptonization parameter @xmath83 and the free - free parameter @xmath5 integrated over the line - of - sight of the blastwave centre between the galaxy outer cut - off radius , @xmath164 , approximately separating the virialized and infall regions , and an inner radius , @xmath165 ( @xmath166 , where the time required for the shock front to reach the outer radius of the galaxy equals the free - free cooling timescale ( it is in fact reasonable to assume that the physical conditions at @xmath167 are more related to the central engine than to the blastwave evolution ) . by assuming an electron temperature @xmath168 k ( @xmath169 k ) and a wide range of line - of - sight velocity dispersions , @xmath170 km / s , we find @xmath171 ( @xmath172 ) for objects both at @xmath173 and at @xmath174 , while only for low values of @xmath22 ( @xmath175 k ) and high values of @xmath73 ( @xmath176 ) @xmath177 approaches the unity .
a reasonable upper limit to @xmath5 in this model is then @xmath178 , a value significantly smaller than the free - free distortion predicted in the model by oh 1999 and below or comparable to the sensitivity of a _
dimes_-like experiment .
the reionization of the universe indicated by the _ wmap _
data must have begun at relatively high redshift .
ciardi , ferrara & white ( 2003 ) have studied the reionization process using supercomputer simulations of a large and representative region of a universe which has cosmological parameters consistent with the _ wmap _ results .
the simulation follows both the radiative transfer of ionizing photons and the formation and evolution of the galaxy population which produces them .
they have shown that the _ wmap _ measured optical depth to electron scattering is easily reproduced by a model in which reionization is caused by the first stars in galaxies with total masses of a few @xmath179 . moreover , the first stars are `` normal objects '' , i.e. their mass is in the range 150 @xmath180 , but metal - free . among
the different model explored , the `` best '' _ wmap _ value for @xmath181 is matched assuming a moderately top - heavy initial mass function ( imf ) ( larson imf with @xmath182 ) and an escape fraction of 20% .
a salpeter imf with the same escape fraction gives @xmath183 , which is still all the suggested 68% confidence ranges .
decreasing @xmath184 to 5% gives @xmath185 , which disagrees with _ wmap _ only at @xmath186 level , the exact level depending on the detailed cosmological scenario adopted to fit the wmap ( and auxiliary ) data .
using the redshift evolution of the density of free electrons obtained by these simulations ( kindly provided us by a. ferrara ) , we compute the expected cmb spectral distortions . we obtain @xmath187distortions of the order of @xmath188 ( @xmath189 ) for a constant electron temperature of @xmath190 k ( @xmath191 k ) corresponding to an energy injection of @xmath192 ( @xmath193 ) well below the current firas upper limits , but easily detectable by an experiment with a sensitivity such that proposed for firas ii .
we compute also the corresponding free - free distortion that results to be so small ( @xmath194 ) that can not be detected by any future experiment . in the best - fit model of ciardi
et al . ( 2003 ) reionization is essentially complete by @xmath195 .
this is difficult to reconcile with observations of the gunn - peterson effect in @xmath196 quasars ( becker et al .
2001 ; fan et al .
this implies a volume - average neutral fraction above @xmath197 and a mass - averaged neutral fraction @xmath198% at @xmath199 .
a fascinating ( although speculative ) possibility is that the universe was reionized twice ( cen 2003 ; whyte & loeb 2002 ) with a relatively short redshift interval in which the igm became neutral again .
an alternative scenario has been proposed by hansen & haiman ( 2003 ) .
they considered the case of a cosmologically significant particle with the property that it decays around a redshift @xmath200 .
the decay products could have energies high enough to reionize the light elements .
if this hypothetical particle is sufficient abundant , then it may explain the high value of the optical depth , without the need for stars to reionize the universe at high redshift , but normal stellar populations could form at @xmath201 with the usual efficiencies and account for the completion of reionization epoch that appears to be occurring at @xmath202 .
hansen & haiman ( 2003 ) proposed that this particle could be a sterile neutrino with a mass @xmath203 mev and decay time @xmath204 s. the decaying of this neutrino can account for the electron scattering optical depth @xmath205 ( kogut et al . 2003 ) measured by _
wmap _ without violating existing astrophysical limits on the cmb and gamma ray background .
in particular , they have shown that the expected energy injection in the cmb of this process is @xmath206 , depending on the fraction of the mean electron energy that goes into heating the igm .
these values are well below the current firas upper limits , but in principle can be detected by future experiments such as firas ii ( fixsen & mather 2002 ) .
we calculate the associated free - free distortion of the cmb spectrum generated during a redshift interval corresponding to a time interval between @xmath207 and @xmath208 for different simple thermal histories ( @xmath209 or @xmath210 ) producing the above values of @xmath3 .
the expected @xmath5 is in the range @xmath211 depending mainly on the exact amount of energy that goes into the igm and only slightly on the considered thermal history of the process . particularly for cases in which a large amount of the electron energy goes into heating the igm , this distortion can be in principle observable by the future _ dimes _ experiment .
lccccc + process & comptonization distortion & free - free distortion + & ( mainly through firas ii ) & ( mainly through _ dimes _ ) + ffe & measurable & measurable + qsob & measurable & @xmath212 at limit + fsg & measurable & unobservable + pd & measurable & observable +
we have studied the implications of the new generation of cmb spectrum space experiments for our knowledge of the thermal history of the universe .
the combination of two experiments with the sensitivity and the frequency coverage jointly forseen for _ dimes _ and firas ii will be able to greatly change our vision of the capability of the cmb spectrum information to constrain physical processes at different cosmic ages .
the limits on the energy dissipations at the all cosmic times accessible to cmb spectrum investigations ( @xmath0 ) could be improved by about two order of magnitudes and even dissipation processes with @xmath1 could be detected and possibly accurately studied .
we tested also the possibility to have an independent cross - check on the baryon density accurately measured by cmb anisotropy experiments in presence of possible small early distortions , by using such accurate measures at @xmath7 cm to determine the chemical potential and the longer wavelengths to estimate @xmath41 through the decreasing of the equivalent thermodynamic temperature at increasing wavelengths , without finding a significant improvement with respect to the results previously described in burigana & salvaterra 2003 : only high accuracy measures at very long wavelengths ( @xmath213 cm ) about the minimum of the equivalent thermodynamic temperature are able to provide a firm and accurate independent cross - check on @xmath41 . with a joint analysis of two dissipation processes occurring at different epochs , we demonstrated that the sensitivity and the wide frequency range jointly covered by _
dimes _ and firas ii would allow to accurately recover the two amounts of energy exchanged in the primeval plasma and to constrain quite well also the epochs of the two processes even when possible , a priori unknown , imprints from free - free distortions are taken into account in the data analysis .
all the three distortion parameters could be in fact accurately reconstructed in this perspective : the sensitivity at 95 per cent cl is @xmath2 for the two values of @xmath3 and of @xmath4 for @xmath5 .
these results are possible because such levels of accuracy on a so wide frequency range allow to remove the approximate degeneracy both between free - free and be - like distortions and between comptonization and be - like distortions that remain in presence of future significant improvements only at @xmath6 cm or at @xmath7 cm , respectively .
the sensitivity on @xmath3 , mainly determined by a firas ii - like experiment , improves by a factor @xmath8 by adding the information from a _ dimes_-like experiment , while the sensitivity on @xmath5 , mainly determined by a _
dimes_-like experiment , improves by a factor @xmath9 , by adding the information from a firas ii - like experiment . of course , not only a very good sensitivity , but also an extreme control of the all systematic effects and , in particular , of the frequency calibration is crucial to reach these goals .
finally , we discussed the different signatures imprinted on the cmb spectrum by some late astrophysical and particle decay models recently proposed in the literature and possibly related to the reionization of the universe indicated by _
wmap _ , and compared them with the sensitivity of such classes of cmb space spectrum experiments .
different processes produce different levels of coupled comptonization and free - free distortions and the combination of very accurate cmb spectrum measures at long and short wavelengths can in principle probe or constrain these models .
we warmly thank l. danese , g. de zotti , a. ferrara and p. platania for numberless conversations on theoretical aspects of cmb spectral distortions and on the ionization history .
it is a pleasure to thank m. bersanelli and n. mandolesi for useful discussions on cmb spectrum observations .
some of the calculations presented in sect .
5 have been carried out on an alpha digital unix machine at the ifp / cnr in milano by using some nag integration codes .
for sake of completeness we report in table a1 the constraints that can be in principle derived from a firas ii - like experiment and from the combination of a _
dimes_-like experiment with a firas ii - like experiment in the joint analysis of two kinds of spectral distortions .
these sensitivities on a couple of distortion parameters apply in the case in which a quite strong prior is assumed on the third distortion parameter ( in practice , assuming it ranging within an interval much smaller than the corresponding sensitivity quoted in table 2 ) .
compare table a1 with table 2 .
although the best sensitivity on @xmath5 is reached when the earlier energy exchange is assumed known , the sensitivity on @xmath5 does not depend much on the considered case and not much degrades by considering two energy exchanges . finally , the sensitivity on one of the two energy exchanges is better by assuming known the other one and unknown the free - free distortion than by assuming known the free - free distortion and unknown the other energy exchange .
lccccc + data set & & @xmath90 + + & heating at & heating at & heating at & heating at & free - free dist . at + & @xmath51 & @xmath91 & @xmath92 & @xmath52 & @xmath52 + + f - ii & [email protected] & & & [email protected] & + f - ii & & [email protected] & & [email protected] & + f - ii & & & [email protected] & [email protected] & + f - ii & [email protected] & & & & [email protected] + f - ii & & [email protected] & & & [email protected] + f - ii & & & [email protected] & & [email protected] + f - ii & & & & [email protected] & [email protected] + d + f - ii & [email protected] & & & [email protected] & + d + f - ii & & [email protected] & & [email protected] & + d + f - ii & & & [email protected] & [email protected] & + d + f - ii & [email protected] & & & & [email protected] + d + f - ii & & [email protected] & & & [email protected] + d + f - ii & & & [email protected] & & [email protected] + d + f - ii & & & & [email protected] & [email protected] + + afshordi n. , 2002 , prl , submitted , astro - ph/0202082 becker r.h .
et al . , 2001 ,
aj , 122 , 2850 bennett c.l .
et al . , 1996 ,
soc . meet .
, 88.05 bennett c.l .
et al . , 2003 ,
apj , in press , astro - ph/0302207 burigana c. , danese l. , de zotti g. , 1991a , a&a , 246 , 59 burigana c. , de zotti g. , danese l. , 1991b , apj , 379 , 1 burigana c. , de zotti g. , danese l. , 1995 , a&a , 303 , 323 burigana c. & salvaterra r. , 2000 , int .
itesre / cnr 291/2000 , august burigana c. & salvaterra r. , 2003 , mnras , 342 , 543 cen r. , 2003 , apj , 591 , 12 ciardi b. , ferrara a. & white s.m.d . , 2003 , mnras , submitted , astro - ph/0302451 danese l. & burigana c. , 1993 , _ theoretical aspects of the cmb spectrum _ , in `` present and future of the cosmic microwave background '' , lecture in physics , vol .
429 , eds .
sanz , e. martinez - gonzales , l. cayon , springer verlag , heidelberg , frg , p. 28
danese l. & de zotti g. , 1977 , riv .
nuovo cimento , 7 , 277 de zotti g. & burigana c. , 1992 , _ the cosmic microwave background spectrum : theoretical framework _
, in `` highlights of astronomy '' , vol .
9 , ed . j. bergeron , kluwer academic publishers , london , uk , p. 265
fan x. et al . , 2002 ,
aj , 123 , 1247 fixsen d.j .
et al . , 1994 , apj , 420 , 457 fixsen d.j .
et al . , 1996 , apj , 473 , 576 fixsen d.j . & mather j.c . , 2002 , apj , 581 , 817 hansen s.h . &
haiman z. , 2003 , apjl , submitted , astro - ph/0305126 kogut a. , 1996 , _ diffuse microwave emission survey _
, contribution to the xvi moriond astrophysics meeting `` microwave background anisotropies '' , march 16 - 23 , les arcs , france , astro - ph/9607100 kogut a. , 2003 , _ reionization and structure formation with arcade _ , to be published in the proceedings of `` the cosmic microwave background and its polarization '' , new astronomy reviews , eds . s. hanany & k.a .
olive , astro - ph/0306044 kogut a. et al .
, 2003 , apj , in press , astro - ph/0302213 kompaneets a.s . , 1956 , zh .
fiz . , 31 , 876 [ sov . phys .
jept , 4 , 730 , ( 1957 ) ] mandolesi n. et al . , 1998 , planck lfi , a proposal submitted to the esa mather j.c .
, fixsen d.j .
, shafer r.a . ,
mosier c. , wilkinson d.t . , 1999 , apj , 512 , 511 nordberg h.p .
& smoot g.f . , 1998 ,
astro - ph/9805123 oh s.p . , 1999 ,
apj , 527 , 16 platania p. , burigana c. , de zotti g. , lazzaro e. , bersanelli m. 2002 , mnras , 337 , 242 press w.h . ,
teukolsky s.a .
, vetterling w.t . ,
flannery b.p . , 1992 , `` numerical recipes in fortran '' , second edition , cambridge university press , usa puget , j.l .
et al . , 1998 , hfi for the planck mission , a proposal submitted to the esa salvaterra r. & burigana c. , 2000 , int .
itesre / cnr 270/2000 , march , astro - ph/0206350 salvaterra r. & burigana c. , 2002 , mnras , 336 , 592 smoot g.f .
et al . , 1992 ,
apj , 396 , l1 sunyaev r.a . &
zeldovich ya.b . , 1970 ,
ap&ss , 7 , 20 tauber j.a . , 2000 , _ the _ planck _ mission _ , in `` the extragalactic infrared background and its cosmological implications '' , proceedings of the iau symposium , vol . 204 , eds .
m. harwit and m. hauser tegmark m. & silk j. , 1995 , apj , 441 , 458 whyte s. & loeb a. , 2003 , apjl , in press , astro - ph/0302297 zeldovich ya.b . &
sunyaev r.a . , 1969 , ap&ss , 4 , 301 zeldovich ya.b . , illarionov a.f . ,
sunyaev r.a .
, 1972 , zh .
fiz . , 62 , 1216 [ sov .
jept , 35 , 643 ] zizzo a. , 2003 , degree thesis , bologna university zizzo a. & burigana c. , 2003 , in preparation [ lastpage ] | we have studied the implications of the new generation of cmb spectrum space experiments for our knowledge of the thermal history of the universe .
the combination of two experiments with the sensitivity and the frequency coverage jointly forseen for _ dimes _ and firas ii will be able to greatly change our vision of the capability of the cmb spectrum information to constrain physical processes at different cosmic ages .
the limits on the energy dissipations at the all cosmic times accessible to cmb spectrum investigations ( @xmath0 ) could be improved by about two order of magnitudes and even dissipation processes with @xmath1 could be detected and possibly accurately studied . with a joint analysis of two
dissipation processes occurring at different epochs , we demonstrated that the joint performances of _ dimes _ and firas ii would allow to accurately recover the two amounts of energy exchanged in the primeval plasma and to constrain quite well also the epochs of the two processes even when possible imprints from free - free distortions are taken into account .
all the three distortion parameters could be accurately reconstructed in this perspective : the sensitivity at 95 per cent cl is @xmath2 for the two values of @xmath3 and of @xmath4 for the free - free distortion parameter @xmath5 .
these results are possible because such levels of accuracy on a so wide frequency range allow to remove the approximate degeneracy both between free - free and bose - einstein ( be ) like distortions and between comptonization and be - like distortions that remain in presence of future significant improvements only at @xmath6 cm or at @xmath7 cm , respectively .
the sensitivity on @xmath3 , mainly determined by a firas ii - like experiment , improves by a factor @xmath8 by adding the information from a _ dimes_-like experiment , while the sensitivity on @xmath5 , mainly determined by a _
dimes_-like experiment , improves by a factor @xmath9 , by adding the information from a firas ii - like experiment .
finally , we discussed the different signatures imprinted on the cmb spectrum by some late astrophysical and particle decay models recently proposed in the literature and possibly related to the reionization of the universe indicated by _
wmap _ , and compared them with the sensitivity of such classes of cmb space spectrum experiments . # 1 # 1 2truept2truept 2truept2truept @xmath10 # 1#2 = cmbx12 et al.et al . [ firstpage ] cosmology : cosmic microwave background
cosmology : theory |
understanding the metal - insulator transition ( mit ) in transition metal oxides is pivotal in explaining the rich phenomenology that these compounds display @xcite . recently
, strong spin - orbit coupling has been shown to lead to novel mits and associated electronic and magnetic states in oxides containing a @xmath8 element @xcite . while the strongly correlated @xmath9 oxides are described as mott insulators , a range of mechanisms
have been proposed to account for the insulating ground states of the @xmath8 oxides @xcite . in particular , it remains unclear how the metal - insulator transition evolves in the correlated , strong spin - orbit coupling regime relevant to these materials . in the ruddlesden - popper series of perovskite iridates sr@xmath10ir@xmath11o@xmath12 ( where @xmath13 is the number of ir - o layers , separated by sr - o layers ) , the large
octahedral crystal - field splitting and spin - orbit coupling produce a @xmath14 band for ir@xmath15 ( @xmath16 ) . for sr@xmath1iro@xmath17 ( @xmath18 )
, it was proposed that the narrow @xmath14 band is split by the modest electronic correlations to create a spin - orbit mott insulator @xcite .
the ideal realization of the @xmath14 state carries a magnetic moment of 1@xmath19 , however , experimental values of 0.5@xmath19 @xcite suggest significant degrees of itineracy .
the perovskite iridates are therefore located in the intermediate - coupling regime , and the insulating state has been proposed to exhibit both slater and mott characteristics @xcite .
the bilayer compound sr@xmath0ir@xmath1o@xmath2 ( @xmath20 ) marginally retains the insulating @xmath14 state @xcite . the effective change in dimensionality in the bilayer structure places sr@xmath0ir@xmath1o@xmath2 in close proximity to a metal - insulator transition , as sriro@xmath0 ( @xmath21 )
is found to be metallic @xcite .
the charge gap of @xmath5130 mev in sr@xmath0ir@xmath1o@xmath2 @xcite can be closed by introducing mobile carriers through substituting la@xmath22 on sr@xmath23 sites @xcite .
this weakens electron correlations such that a conventional fermi liquid state is recovered . in this context ,
application of pressure is an attractive control parameter , as it allows to continuously and isoelectronically tune the salient energy scales .
et al . _ reported a second - order structural transition in sr@xmath0ir@xmath1o@xmath2 at 14 gpa , accompanied by a reduction in resistivity @xcite .
however , high - pressure transport measurements by zocco
_ @xcite have seen no evidence for this transition , and instead report a gradual reduction of the electronic gap up to 104 gpa .
while the authors do not report definitive evidence for a transition to a metallic state , they concede that pressure gradients could affect their data and that an insulator - metal transition may occur at lower pressures . here , we present a comprehensive study of the pressure dependence of the structure of sr@xmath0ir@xmath1o@xmath2 up to 61 gpa , using powder x - ray diffraction ( xrd ) at ambient temperature and at 20 k. we identify a first - order structural transition to a monoclinic symmetry accompanied by a volume collapse of 4% at 54 gpa at room temperature .
rietveld refinement suggests that the stacking sequence of the two perovskite bilayers is altered in the high - pressure phase , resulting in a novel modification of the ruddlesden - popper structure .
in contrast to the study by zhao _
@xcite , no evidence for a second - order structural transition around 14 gpa was found either at room temperature or at 20 k. as the oxygen atoms scatter only weakly compared to the electron - heavy sr and ir atoms , a reliable refinement of the oxygen positions was not possible from the diffraction data .
we employed density functional theory [ local - density approximation ( lda)+@xmath3+spin - orbit ( so ) ] calculations to refine the structure of sr@xmath0ir@xmath1o@xmath2 at high pressure and to examine the evolution of the electronic and magnetic states .
the calculations suggest that the in - plane rotation of iro@xmath4 octahedra increases with pressure in the tetragonal phase .
the decrease of the ir - o bond distance , however , widens the @xmath6 bands , such that an insulator - metal transition occurs at @xmath520 gpa . for the monoclinic high - pressure phase , optimization of the oxygen positions leads to complex rotations and tilting of octahedra , which considerably lowers the total energy with respect to the structure proposed by experiment .
in this high - pressure structure , we find strongly overlapping @xmath6 and @xmath7 bands . due to the wide @xmath6 bands ,
both the spin - orbit coupling and electronic correlations become ineffectual in opening an electronic gap and we obtain a metallic state for the high - pressure phase .
we conducted high - pressure x - ray diffraction experiments at beamlines id09a ( european synchrotron radiation facility ) and i15 ( diamond light source ) .
single crystals of sr@xmath0ir@xmath1o@xmath2 were flux grown as described in ref . and characterized by squid magnetometry .
we used symmetrical diamond - anvil cells , fitted with boehler - almax diamonds with a culet diameter of 250 @xmath24 m .
the gasket material was rhenium , indented to 30 @xmath24 m thickness , into which a 100-@xmath24m - diameter hole was laser - drilled .
the sr@xmath0ir@xmath1o@xmath2 single crystals were ground into powders and loaded with helium as the pressure transmitting medium .
pressure was measured _ in situ _ using the srb@xmath17o@xmath2:sm@xmath23 fluorescence pressure scale @xcite .
the diffraction patterns were recorded on mar345 and mar555 detectors and integrated using the fit2d software @xcite .
full profile rietveld refinements were performed using the fullprof software @xcite .
ir@xmath1o@xmath2 at high pressure at ambient temperature .
( a ) sr@xmath0ir@xmath1o@xmath2 in its monoclinic phase ( space group @xmath25 ) at 61 gpa ; ( d ) the corresponding structure .
( b ) tetragonal phase of sr@xmath0ir@xmath1o@xmath2 ( space group @xmath26 ) at 17 gpa ; ( c ) the corresponding structure . in ( a ) and ( b ) the red circles are background - subtracted data points
, the black solid line presents the rietveld refinement , blue tick marks are the calculated bragg reflections , and the residuals of the refinement are shown by the black line beneath .
reflections from srb@xmath17o@xmath2:sm@xmath23 are marked by an asterisk .
the inset in ( a ) shows the pressure evolution of small - angle reflections . in ( c ) and
( d ) the iridium atoms are shown in bronze , the oxygen atoms in red and the strontium atoms in purple .
the structures in ( c ) and ( d ) were visualised via vesta @xcite . ]
diffraction patterns of sr@xmath0ir@xmath1o@xmath2 at high pressure are shown in fig .
[ xrd ] . at ambient temperature in a helium pressure - transmitting medium
, sr@xmath0ir@xmath1o@xmath2 remained in its tetragonal phase ( space group @xmath26 ) up to pressures of 53 gpa .
it has been argued that the rotations of the oxygen octahedra are correlated , resulting in a 90@xmath27 twinned structure of pseudo - orthorhombic space groups @xmath28 and @xmath29 @xcite .
as our measurements were performed on powder samples , the tetragonal @xmath26 structural model effectively averages over the two space groups .
rietveld refinements of our data confirm the @xmath26 structural model to be adequate .
crucially , no second - order structural transition was observed near 14 gpa .
we used helium as the pressure - transmitting medium ( at room temperature and 20 k ) , and not neon as in ref .
( at 25 k ) . the discrepancy between our studies may therefore arise from nonhydrostatic pressure conditions in the previous study . in our experiments ,
the pressure was determined by srb@xmath17o@xmath2:sm@xmath23 fluorescence .
no significant broadening of the srb@xmath17o@xmath2:sm@xmath23 fluorescence line shapes was observed , indicating that hydrostatic pressure conditions were maintained . at pressures above 54 gpa
, new reflections started to emerge .
further increase in pressure resulted in the complete disappearance of the @xmath26 reflections .
the new reflections were indexed in a @xmath30-centered monoclinic system , space group @xmath31 , @xmath25 , or @xmath32 .
indexing the reflections in a higher symmetry proved not to be possible .
this high - pressure phase remained stable up to 61 gpa , the maximum pressure reached in this study . upon decompression , the monoclinic signature disappeared below 49 gpa , and sr@xmath0ir@xmath1o@xmath2 reverted to its tetragonal phase .
the high - pressure phase of sr@xmath0ir@xmath1o@xmath2 departs strongly from the ruddlesden - popper structure .
most notably , the reflection with the largest @xmath33 spacing , ( 002 ) at @xmath34 , disappears [ see inset fig . [ xrd](a ) ] .
all new reflections possess smaller @xmath33 spacings , indicating that the long @xmath35 axis of tetragonal sr@xmath0ir@xmath1o@xmath2 is not retained .
figure [ eos ] shows the refined lattice parameters as a function of pressure . in the tetragonal phase ,
the compression of sr@xmath0ir@xmath1o@xmath2 is anisotropic ; the @xmath35 axis exhibits a lower compressibility than the @xmath36=@xmath37 axes .
the @xmath38 ratio thus increases with pressure by 3% to a critical value , where the system undergoes a structural phase transition .
we indexed the new phase in a monoclinic symmetry , spacegroup @xmath31 , @xmath25 , or @xmath32 , with lattice parameters @xmath39 , @xmath40 , @xmath41 , and @xmath42 at 61 gpa .
this corresponds to a volume collapse of 4% at the phase transition .
a discontinuity of 2 in the long @xmath35 axis of the unit cell ( labeled the @xmath37 axis in the high - pressure phase ) occurs at the structural transition .
the lower compressibility along the @xmath35 axis in the tetragonal phase appears to be alleviated by the phase transition to monoclinic symmetry .
a vinet equation of state was fitted to the pressure - volume data of the tetragonal phase , yielding a bulk modulus of @xmath43 gpa [ @xmath44 , in agreement with the literature @xcite .
( a ) and ( b ) lattice parameters as a function of pressure at 298 k. the inset in ( a ) shows the normalized lattice constants as a function of pressure at 20 k. ( c ) equation of state of sr@xmath0ir@xmath1o@xmath2 at 298 k. the solid line is a vinet - type fit to the observed equation of state of sr@xmath0ir@xmath1o@xmath2 .
the inset in ( c ) shows the @xmath38 ratio as a function of pressure at 298 k. ] to understand the transition at an atomic level , we attempted structural refinement of the high - pressure phase through rietveld analysis . in general ,
the calculated intensities reproduce adequately the experimental intensities when one of the perovskite bilayers of sr@xmath45ir@xmath46o@xmath47 is translated in plane , by half a unit cell [ see fig . [ xrd](d ) ] .
this may explain the discontinuity in the lattice parameter of the long axis of the unit cell .
while in its tetragonal phase the compressibility along the @xmath35 axis appears to be limited , this is alleviated in the monoclinic phase by an alteration in the stacking sequence of the perovskite bilayers .
furthermore , the sr@xmath23 atoms are arranged within a single plane , parallel to the @xmath36 axis in the unit cell . to accommodate this configuration , the lattice parameter @xmath36 increases ,
such that the in - plane lattice constants ( @xmath48 ) of the monoclinic structure differ by 10% , and the angle spanned between @xmath36 and @xmath35 is 91.8@xmath49 .
this high - pressure structure could be described in the space group @xmath31 .
the presence of a mirror plane perpendicular to the @xmath37 axis minimizes the number of free parameters to two @xmath50 coordinates for ir and sr ( the second sr atom is fixed at @xmath51 ) .
however , the fit of the calculated diffraction pattern was deemed unsatisfactory .
we therefore decided to remove the mirror plane and adopt @xmath25 symmetry .
this allowed to move atoms independently along the @xmath50 axis , through two adjustable @xmath50 coordinates for both ir and sr atoms ( here , the third sr atom remained at @xmath51 ) .
this provided the best fit to the data .
the weighted profile rietveld @xmath52 factor @xmath53 of the monoclinic structure at 61 gpa then became comparable to the one obtained from the rietveld refinement of the tetragonal phase at 17 gpa .
thus our proposed structural model for the high - pressure phase provides as good a description of the data as the @xmath26 structure at lower pressures .
we note that the visually worse fit of the calculated intensities at 61 gpa at least partly originates from the difficulty of accurately describing the background at very high pressures .
as the oxygen atoms scatter only weakly compared to the electron - heavy ir and sr atoms , a precise refinement of the oxygen positions was not possible for the high - pressure phase .
small deviations from the proposed positions [ as shown in fig . [ xrd](d ) ] through rotations or tilting can not be excluded or confirmed within the quality of our data .
to investigate the pressure dependence of the electronic and magnetic properties of sr@xmath45ir@xmath46o@xmath47 we performed first - principles calculations with two complementary approaches : the full - potential linearized - augmented - plane wave ( fp - lapw ) method , as realized in the elk package @xcite ; and the plane wave pseudopotential method , as implemented in the quantum - espresso simulation package @xcite . in these calculations
we used the lda , and its popular refinement taking into account the on - site coulomb interaction and spin - orbit coupling ( lda+@xmath3+so ) @xcite . in pseudopotential calculations , we set an energy cutoff of 400 ev in the plane - wave basis construction and the energy convergence criteria of 10@xmath54 ev . for the brillouin - zone integration , a 6 @xmath55
6 @xmath55 6 monkhorst - pack mesh was used .
ir@xmath1o@xmath2 in the ambient , tetragonal phase ( space group @xmath26 ) with oxygen disorder .
( middle ) @xmath28 structure of sr@xmath0ir@xmath1o@xmath2 used in first - principles calculations .
the difference between blue and gray iro@xmath4 octahedra is the octahedral rotation around the @xmath56 axis .
the red and purple spheres denote oxygen and strontium atoms , respectively .
( right ) @xmath57-plane projection of the @xmath28 structure .
crystallographic plots were made with the vesta software @xcite.,scaledwidth=50.0% ] we first discuss the lda and lda+so results for the ambient pressure phase of sr@xmath0ir@xmath1o@xmath2 . in these calculations , we started from the @xmath26 crystal structure with oxygen disorder @xcite and then doubled the planar basis lattice vectors and deleted specific planar oxygen atoms in the supercell , to achieve the @xmath28 space group ( fig .
[ structuresbbcb ] ) .
ir@xmath1o@xmath2 in the @xmath28 structure at ambient pressure using the lda method.,scaledwidth=45.0% ] from the calculated partial densities of states presented in fig .
[ pdoslda_ap ] , the splitting between @xmath6 and @xmath58 states can be estimated to be about 3.3 ev , in good agreement with the experimental value of 3.5 ev @xcite .
we find that the @xmath8 ir states of @xmath57 character are filled and the @xmath59 ( @xmath60 ) orbitals are partially occupied .
the @xmath58 band is fully located in the unoccupied part of the electronic spectrum due to much stronger hybridization with oxygen states .
the obtained level picture agrees with symmetry analysis of the sr@xmath0ir@xmath1o@xmath2 system @xcite .
as we show below , the degeneracy of @xmath60 and @xmath59 states plays an important role for the orbital magnetism . the comparison of the lda and lda+so band structures ( fig .
[ bandlda_ap ] , left ) shows that the spin - orbit coupling is responsible for the formation of four doubly degenerate nearly separated narrow bands at the fermi level .
the corresponding bandwidth can be estimated as 0.5 ev .
given the large crystal field , an _ effective _ total angular momentum approach can be used to construct a minimal microscopic model of the insulating ground state of sr@xmath0ir@xmath1o@xmath2 , as proposed in ref .
however , in contrast to the ideal @xmath14 state described in ref . , we have a more complicated situation .
the splitting of the @xmath6 manifold results in a mixing of @xmath14 and @xmath61 bands , which should be taken into account within a microscopic analysis @xcite .
we employed the lda+@xmath3+so method to first reproduce the insulating ground state and magnetic properties of sr@xmath45ir@xmath46o@xmath47 at ambient pressure .
the obtained band structure is illustrated in fig .
[ bandlda_ap ] . in our simulations ,
the on - site coulomb interaction @xmath3 and intra - atomic exchange parameter @xmath62 were chosen to be 1.8 and 0.3 ev , respectively , as discussed in ref . .
the electronic gap obtained from the calculations is 110 mev , in reasonable agreement with the experimental estimate of 130 mev @xcite .
the calculated spin and orbital magnetic moments are m@xmath63 and m@xmath64 , respectively , in good agreement with the total magnetic moment of 0.52(8)@xmath65 measured by neutron diffraction @xcite .
it is interesting to note that the calculated orbital moment of sr@xmath45ir@xmath46o@xmath47 is comparable to the one calculated for sr@xmath1iro@xmath17 @xcite , while the spin moment is two times larger .
we have simulated different @xmath35-oriented magnetic configurations of sr@xmath0ir@xmath1o@xmath2 : ferromagnetic , and antiferromagnetic of @xmath66 , @xmath30 , and @xmath67 types .
we find that the @xmath35-axis antiferromagnetic configuration of @xmath66-type minimizes the total energy of the system , in agreement with experiments @xcite .
the calculations for other types of the magnetic order revealed strong changes in electronic and magnetic properties of sr@xmath0ir@xmath1o@xmath2 .
in general , the magnetic moments and energy gap values are suppressed in comparison with those obtained for @xmath66-type structure . in turn
, the in - plane antiferromagnetic solution with weak ferromagnetism also corresponds to the excited states of the sr@xmath0ir@xmath1o@xmath2 system .
the corresponding energy difference between in - plane and out - of - plane @xmath66-type solutions is 353 mev ( per unit cell containing four ir atoms ) .
the calculated orbital and spin magnetic moments of ir atom in the in - plane magnetic configuration are 0.186@xmath65 and 0.063@xmath65 , respectively .
the canting angles of the orbital and spin moments are equal to 15.2@xmath27 and 18.4@xmath27 , respectively . as in the case of sr@xmath1iro@xmath17 @xcite
we found a strong contribution from the oxygen atoms to the resulting net magnetic moment .
to understand the evolution of the electronic and magnetic properties of sr@xmath0ir@xmath1o@xmath2 as a function of pressure , we have performed lda+@xmath3+so calculations for the @xmath28 structure , using the experimental lattice parameters ( fig .
[ eos ] ) . for each pressure point , the crystal structure was optimized .
the resulting structural , electronic , and magnetic properties of sr@xmath0ir@xmath1o@xmath2 as a function of pressure are shown in table [ pressuredata ] .
our calculations reveal that the compression of the unit cell is accompanied by an increase of the in - plane rotation angle of the iro@xmath4 octahedra , while the ir - o distance decreases .
intuitively , this can be understood from the experimentally observed anisotropic compression of sr@xmath0ir@xmath1o@xmath2 , where the compressibility along the @xmath36 and @xmath37 axes is higher than that along the @xmath35 axis . at ambient pressure ,
the rotation angle is about 11.8@xmath27 , while the high - pressure solution at 50 gpa reveals an octahedral rotation of 16.3@xmath27 . for sr@xmath1iro@xmath68 and ba@xmath46iro@xmath68
, it was demonstrated that the in - plane rotation of the iro@xmath4 octahedra with fixed ir - o distance leads to a more localized electronic structure @xcite .
therefore the width of the @xmath6 band was found to be narrower for sr@xmath46iro@xmath17 than for ba@xmath1iro@xmath17 .
for sr@xmath0ir@xmath1o@xmath2 , applying pressure also decreases the ir - o distance , which leads to a strong increase of the lda bandwidth for @xmath6 states .
figure [ dos_amb_high ] shows that the spectral function for degenerate @xmath60 and @xmath59 states retains its structure at high pressure ; however , the width of the spectral function notably increases at 50 gpa .
.pressure dependence of the structural , electronic , and magnetic properties of sr@xmath0ir@xmath1o@xmath2 , using the @xmath28 structure .
the results were obtained with lda+@xmath3+so calculations , as described in the text .
@xmath69 is the angle of the in - plane ir - o - ir bond . e@xmath70
is the value of the energy gap .
the lattice parameters corresponding to the pressure of 60 gpa were obtained by extrapolation of the experimental data . [ cols="^,^,^,^,^,^,^",options="header " , ] it is important to analyze the difference in the electronic structure of the high - pressure structure of sr@xmath0ir@xmath1o@xmath2 .
figure [ dos_60gpa_comp ] compares the total densities of states of the compressed @xmath28 and optimized @xmath25 structures , obtained with lda .
the plane - dependent rotation of the oxygen octahedra and compression of the unit cell in the @xmath25 structure lead to a complete delocalization of the @xmath8 ir states in sr@xmath0ir@xmath1o@xmath2 .
no electronic gap between the @xmath6 and @xmath58 bands remains .
our lda+so and lda+u+so calculations for an optimized @xmath25 structure with a complex rotation of the oxygen octahedra have demonstrated that neither spin - orbit coupling nor on - site coulomb interaction can significantly affect the electronic structure .
the corresponding densities of states are presented in fig .
[ comp_c2dist ] .
the system remains metallic , which still follows from the increased @xmath6 bandwidth .
ir@xmath1o@xmath2 in the compressed @xmath28 structure ( blue dotted line ) and optimized @xmath25 structure ( red solid line ) at @xmath71 gpa .
the fermi level corresponds to zero energy . ]
structure of sr@xmath0ir@xmath1o@xmath2 obtained from lda ( black solid line ) , lda+so ( blue dotted line ) and lda+@xmath3+so ( red dashed line ) calculations .
the fermi level corresponds to zero energy . ]
we have studied the structural evolution of sr@xmath0ir@xmath1o@xmath2 with powder x - ray diffraction to unprecedented pressures .
we observe a novel first - order structural transition at 54 gpa , preceded by a strongly anisotropic compression , characterized by an increasing @xmath38 ratio with pressure .
the high - pressure phase could be indexed in a monoclinic symmetry ( space group @xmath31 , @xmath25 , or @xmath32 ) .
rietveld refinement of the diffraction data suggests that the stacking sequence of the perovskite bilayers is altered in the high - pressure structure .
a precise refinement of the oxygen positions was , however , not possible due to the relatively weak x - ray scattering signal compared to ir and sr atoms .
we used first - principles lda+@xmath3+so calculations to both optimize the structure and examine the electronic properties of sr@xmath0ir@xmath1o@xmath2 as a function of pressure .
below the high - pressure structural transition , we find that the anisotropic compression observed by experiment is accompanied by an increase in the in - plane iro@xmath4 octahedral rotation .
the simultaneous decrease of the ir - o bond length however increases the @xmath6 bandwidth and gradually closes the band gap . in our calculations
, we find an insulator - metal transition at @xmath520 gpa .
this value is comparable to the transition to a nearly metallic state at 13.2 gpa observed by li _
@xcite , but contrasts with measurements by zocco _
@xcite , which show an electronic gap remaining up to at least 30 gpa . for the high - pressure phase ,
structural optimization of the oxygen position resulted in complex in - plane and out - of - plane rotations of the octahedra .
this considerably lowered the total energy with respect to the structure obtained from experiment . in the optimized high - pressure structure
, the @xmath6 and @xmath7 bands strongly overlap .
the shortened ir - o bond length does not allow an electronic band gap to open , even in the presence of strong spin - orbit coupling and on - site coulomb interactions .
this suggests a robust metallic state for the high - pressure monoclinic phase of sr@xmath0ir@xmath1o@xmath2 .
+ this work is supported by the uk engineering and physical sciences research council , grant no .
ep / j016713/1 .
h.m.r . acknowledges support from the swiss national science foundation and its sinergia network mott physics beyond the heisenberg model . the work of i.v.s . and v.v.m .
is supported by the grant program of the russian science foundation , 14 - 12 - 00306 .
we would like to thank d. daisenberger , a. kleppe and h. wilhelm from beamline i15 , diamond light source for technical support .
33ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1103/revmodphys.70.1039 [ * * , ( ) ] link:\doibase 10.1146/annurev - conmatphys-020911 - 125138 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.101.076402 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.101.226402 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.108.086403 [ * * , ( ) ] link:\doibase 10.1126/science.1167106 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.102.017205 [ * * , ( ) ] link:\doibase 10.1103/physrevb.66.214412 [ * * , ( ) ] link:\doibase 10.1103/physrevb.86.100401 [ * * , ( ) ] link:\doibase 10.1103/physrevb.86.035128 [ * * , ( ) ] link:\doibase 10.1103/physrevb.89.165115 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.109.037204 [ * * , ( ) ] link:\doibase 10.1103/physrevb.85.184432 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.111.246402 [ * * , ( ) ] http://dx.doi.org/10.1038/nmat3653 [ * * , ( ) ] link:\doibase 10.1103/physrevb.87.235127 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.113.256402 [ * * , ( ) ] http://dx.doi.org/10.1038/nmat4273 [ * * , ( ) ] http://stacks.iop.org/0953-8984/26/i=21/a=215402 [ * * , ( ) ] http://stacks.iop.org/0953-8984/26/i=25/a=255603 [ * * , ( ) ] \doibase http://dx.doi.org/10.1063/1.365025 [ * * , ( ) ] link:\doibase 10.1080/08957959608201408 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/0921-4526(93)90108-i [ * * , ( ) ] link:\doibase 10.1107/s0021889808012016 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/j.jssc.2004.06.056 [ * * , ( ) ] @noop _ _ ( , ) http://stacks.iop.org/0953-8984/21/i=39/a=395502 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.80.5758 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/0025-5408(94)90120-1 [ * * , ( ) ] link:\doibase 10.1103/physrevb.90.085126 [ * * , ( ) ] link:\doibase 10.1103/physrevb.89.121101 [ * * , ( ) ] @noop * * , ( ) link:\doibase 10.1103/physrevb.92.235109 [ * * , ( ) ] | we study the structural evolution of sr@xmath0ir@xmath1o@xmath2 as a function of pressure using x - ray diffraction . at a pressure of 54 gpa at room temperature ,
we observe a first - order structural phase transition , associated with a change from tetragonal to monoclinic symmetry , and accompanied by a 4% volume collapse .
rietveld refinement of the high - pressure phase reveals a novel modification of the ruddlesden - popper structure , which adopts an altered stacking sequence of the perovskite bilayers .
as the positions of the oxygen atoms could not be reliably refined from the data , we use density functional theory ( local - density approximation+@xmath3+spin orbit ) to optimize the crystal structure , and to elucidate the electronic and magnetic properties of sr@xmath0ir@xmath1o@xmath2 at high pressure . in the low - pressure tetragonal phase , we find that the in - plane rotation of the iro@xmath4 octahedra increases with pressure .
the calculations further indicate that a bandwidth - driven insulator - metal transition occurs at @xmath520 gpa , along with a quenching of the magnetic moment . in the high - pressure monoclinic phase ,
structural optimization resulted in complex tilting and rotation of the oxygen octahedra , and strongly overlapping @xmath6 and @xmath7 bands .
the @xmath6 bandwidth renders both the spin - orbit coupling and electronic correlations ineffectual in opening an electronic gap , resulting in a robust metallic state for the high - pressure phase of sr@xmath0ir@xmath1o@xmath2 . |
careful control of the beam quality is essential for linear accelerators designed to deliver very high brightness electron beams for short wavelength free electron lasers ( fels ) .
there are many beam properties which have to be observed and manipulated such as suppression of microbunching instability , creation of needed peak current , preservation of slice and projected emittances , etc .
in this paper we are interested in some aspects of the control of transverse projected emittances .
among the sources of the growth of transverse projected emittances are the incoherent and coherent synchrotron radiation ( csr ) withing magnetic bunch compressors as well as the other wake fields along the accelerator .
a number of approaches which could help to reduce emittance growth due to csr wake during bunch compression were developed during last decades including different optics tricks , preparation of the initial beam current profile at the bunch compressor entrance and etc .
( see , for example , @xcite and references therein ) . still ,
because the suggested schemes provide reduction but not complete cancellation of the emittance growth , the beam considered downstream of the compression system ( or at the linac exit ) could have nonzero transverse to longitudinal coupling terms in the beam matrix and therefore projected emittances could be further reduced if these correlations will be removed . in general , in order to make complete transverse to longitudinal decoupling
, it is necessary to have the possibility to act on particles depending on their longitudinal position within the bunch ( for example , one may involve transverse deflecting cavities for this purpose ) , which means that the system designed for the complete decoupling could be too complicated and somewhat difficult to operate in comparison with the benefit coming from the achievable reduction of the transverse projected emittances .
in this paper we consider a more simple and more practical question : what one can do having at hand a magnetostatic correction system ? because the transfer matrix of a magnetostatic system could couple transverse and longitudinal particle coordinates only when the dispersions of the underlying magnetic structure are nonzero , the reduction of transverse projected emittances ( if any possible ) will always be accompanied by the creation of a potential source of beam transverse jitter due to the beam energy jitter , and one has to look for an appropriate balance of both .
we show , in the framework of linear particle dynamics and with the self field effects neglected , that if in the beam matrix there are nonzero correlation terms between energy of particles and their transverse positions and momenta ( beam dispersions ) , then the transverse projected emittances can be reduced by letting the beam pass through magnetostatic system ( correction system ) with specially chosen nonzero lattice dispersions .
the maximum possible reduction of the transverse projected emittances occurs when all beam dispersions are zeroed , and the values of the lattice dispersions required for that are completely determined by the values of the beam dispersions and the beam rms energy spread and are independent from any other second - order central beam moments . besides that
, we prove that , alternatively , one can also use the lattice dispersions to remove linear correlations between longitudinal positions of particles and their transverse coordinates ( beam tilts ) , but in this situation solution for the lattice dispersions is nonunique and the reduction of the transverse projected emittances is not guaranteed . note
that this paper is an extended version of the unpublished note @xcite , which was written during discussion of the influence of different dispersive effects on the performance of the flash facility @xcite , and recently , when we get acquainted with the paper @xcite , our interest to this problem was renewed
. both papers , @xcite and @xcite , employ unclosed lattice dispersions as a tuning knob for the control of the transverse projected emittances , but have somewhat different points of view on the practical realization of this idea and therefore the direct comparison of their results and recommendations is not very straightforward . as far as we are mostly discussing possibilities for correction which can be made downstream of the emittance growth and coupling source either by means of a dedicated correction system or even simply by special ( dispersive ) beam steering , the paper @xcite suggests to create dispersion nonclosure already in the bunch compressor , where the csr effect is strongest and can not be neglected .
we consider the linear beam dynamics in an electromagnetic system which conserves the reference beam energy and take the path length along the reference orbit @xmath0 to be the independent variable .
we use a complete set of symplectic variables @xmath1 as particle coordinates @xcite . here
@xmath2 measure the transverse ( horizontal and vertical ) displacements from the ideal orbit and @xmath3 are the corresponding canonical monenta scaled with the constant kinetic momentum of the reference particle @xmath4 .
the variables @xmath5 and @xmath6 which describe the longitudinal dynamics are @xmath7 where @xmath8 and @xmath9 are the energy of the reference particle , its velocity in terms of the speed of light @xmath10 and its arrival time at a certain position @xmath0 , respectively .
let @xmath11 be an @xmath12 square matrix .
then @xmath13 denote the determinant of @xmath11 .
let @xmath14 be a nonempty subset of @xmath15 with its elements listed in increasing order .
then @xmath16 denote the principal submatrix of @xmath11 whose entries are in the intersection of those rows and columns of @xmath11 specified by @xmath14 .
if @xmath11 is a symmetric matrix , we denote by @xmath17 the associated with this matrix quadratic form in @xmath18-variables @xmath19 @xmath20 besides that , we denote by @xmath21 the @xmath12 identity matrix and by @xmath22 the @xmath23 symplectic unit matrix . as usual
, we describe the properties of a collection of points ( a particle beam ) in the three degrees of freedom ( 3d ) phase space by a @xmath24 symmetric matrix ( beam matrix ) of the second - order central beam moments @xmath25 where the brackets @xmath26 denote an average over a distribution of the particles in the beam .
let @xmath27 be the nondegenerated @xmath24 matrix which propagates particle coordinates from the state @xmath28 to the state @xmath29 , i.e let @xmath30 then from ( [ intr_2 ] ) and ( [ intr_4 ] ) it follows that the matrix @xmath31 evolves between these two states according to the congruence @xmath32 in the following we assume that the beam transport matrix @xmath27 is symplectic , which is equivalent to say that it satisfies the relation @xmath33 by definition , the beam matrix @xmath31 is symmetric positive semidefinite and we restrict our considerations to the situation when this matrix is nondegenerated and therefore positive definite . for simplification of notations
we also assume that the beam is proper centered and therefore has vanishing first - order moments @xmath34 . with this assumption
the beam matrix takes on the form @xmath35 where the elements @xmath36 and the elements @xmath37 we call beam dispersions and beam tilts , respectively .
the matrix @xmath31 has twenty - one different entries which can be varied independently within the positive definiteness conditions .
of course , not all of them ( or their combinations ) are equally interesting for any particular accelerator physics application . in this paper
we concentrate on the study of the evolution of 1d horizontal , vertical and longitudinal projected emittances @xmath38 @xmath39 @xmath40 and 2d transverse projected emittance @xmath41 under the transformation rule ( [ form07 ] ) with the additional assumption that the matrix @xmath27 is the transport matrix of a magnetostatic system . besides that
, we also pay attention to the changes in the beam energy chirp @xmath42 ( linear energy slope along the bunch length ) and in the rms bunch length squared @xmath43 .
note that due to hadamar s determinantal inequality @xmath44 and the equality in ( [ eindet ] ) holds if and only if the transverse degrees of freedom in the beam matrix are decoupled from each other @xcite , i.e. if and only if @xmath45
the most general form of the transport matrix of a magnetostatic system is @xmath46 where the elements @xmath47 are ( transverse and longitudinal ) lattice dispersions .
the special form ( [ form01 ] ) of the matrix @xmath27 allows to rewrite the symplecticity condition ( [ form07symplcond ] ) in the form of a system of two equations @xmath48 and @xmath49 and using condition ( [ form022 ] ) one can show that every matrix @xmath27 of the form ( [ form01 ] ) can be represented as a product @xmath50 where @xmath51 is the dispersion - free part of the matrix @xmath27 and @xmath52 is its dispersive part . substituting the decomposition ( [ form03 ] ) into the beam matrix propagation equation ( [ form07 ] ) one obtains @xmath53 this formula is a two step transformation . at first the incoming beam matrix @xmath54 is transported using the matrix @xmath55 and then this intermediate result
is transformed using the matrix @xmath56 .
because the action of the matrix @xmath56 does not alter longitudinal beam parameters , does not couple transverse and longitudinal projected emittances , and propagates the vector of beam dispersions and the vector of beam tilts simply as transverse coordinates of the particle trajectories ( i.e. without possibilities to create or to remove vectors of beam dispersions and beam tilts , and even without possibility simply to mix the vector of the beam dispersions with the vector of the beam tilts ) , the second step in the transport of the beam matrix can be omitted without loss of generality for any result of this paper .
so , in the rest of this paper , we consider the changes in properties of the incoming beam matrix @xmath54 which are of interest for us under the action of the matrix @xmath55 only . because it is impossible to associate with this action some certain position in the beam line
, we write it symbolically as follows @xmath57 and call this transformation as the beam passage through the dispersive part of the correction system .
note that the formulas obtained below for the simplified propagation rule ( [ s01 ] ) can be translated into the formulas for the complete transport equation ( [ form07 ] ) with the help of the decomposition of the matrix @xmath27 in the form of a product @xmath58 where @xmath59 and the matrix @xmath56 remains the same as given in ( [ form04 ] ) . to make such a translation in the selected formula one has to make the following changes in its right hand side : substitute the lattice dispersions @xmath60 , @xmath61 , @xmath62 , and @xmath63 instead of the lattice dispersions @xmath64 , @xmath65 , @xmath66 , and @xmath67 according to the rule @xmath68 and substitute the elements of the matrix @xmath69 instead of the corresponding elements of the matrix @xmath54 .
note that the decompositions ( [ form03 ] ) and ( [ form033 ] ) are still valid if one simply shifts the element @xmath70 from the matrices @xmath55 and @xmath71 to the corresponding position in the matrix @xmath56 .
it gives an additional possibility to simplify calculations if one cares about transport of the projected emittances only , but because we are also interested in the behavior of the rms bunch length and the beam energy chirp , we prefer to keep the @xmath70 coefficient in the matrices @xmath55 and @xmath71 . in order to obtain convenient representation for the emittance transport problem ,
let us introduce a @xmath72 symmetric matrix @xmath73 because leading principal minors of this matrix can be expressed through the principal minors of the positive definite matrix @xmath31 as follows @xmath74 @xmath75 @xmath76 @xmath77 all leading principal minors of the matrix @xmath78 are positive , which means that the matrix @xmath78 is positive definite according to the sylvester criterion @xcite .
note that the elements of this matrix ( similar to the elements of the matrix @xmath79 given below in ( [ matrixb ] ) ) do not depend on the second - order beam moments involving the longitudinal variable @xmath5 . with the help of the positive definite quadratic form @xmath80 associated with the matrix @xmath78
, the evolution of the 1d projected emittances through the dispersive part of the correction system can be expressed as follows : @xmath81 @xmath82 where @xmath83 @xmath84 @xmath85 where @xmath86 @xmath87 @xmath88 @xmath89 where @xmath90 one sees from the propagation rules obtained that while the beam tilts and the beam energy chirp can influence the evolution of the longitudinal projected emittance @xmath91 through the solution of the equation ( [ optsolutionz ] ) , they do not enter the formulas for the evolution of the transverse projected emittances @xmath92 and @xmath93 at all
. in the previous subsection the evolution of all three 1d projected emittances was expressed using the single quadratic form @xmath80 .
unfortunately , to describe the evolution of the 2d transverse projected emittance another , different from @xmath80 , quadratic form is needed .
we denote this form @xmath94 and associated it with the positive definite symmetric matrix @xmath95 @xmath96^{-1}\left\{1,2,3,4\right\}. \label{matrixb}\end{aligned}\ ] ] with the help of this new quadratic form the evolution of the 2d transverse projected emittance can be expressed as follows : @xmath97 @xmath98 @xmath99 where @xmath100 , @xmath101 , @xmath102 , and @xmath103 are the same as given by the formulas ( [ optsolutionxx ] ) and ( [ optsolutionyy ] ) . note
that though one may accept without additional questions the fact that the right hand sides of the formulas ( [ evolutionex ] ) and ( [ evolutioney ] ) are the second order polynomials with respect to the lattice dispersions , the same property of the right hand side of the formula ( [ evolutionet ] ) might be somewhat more surprising .
for example , let us assume that the beam matrix is transversely uncoupled at the exit of the dispersive part of the correction system .
then the right hand side of the formula ( [ evolutionet ] ) must coincide with the product of the right hand sides of the formulas ( [ evolutionex ] ) and ( [ evolutioney ] ) and therefore should contain a polynomial of the fourth order with respect to the variables @xmath64 , @xmath65 , @xmath66 , and @xmath67 .
because the formula ( [ evolutionet ] ) does not provide such a possibility , our assumption must be wrong and , during the passage of the dispersive part of the correction system , the coupling between transverse degrees of freedom in the beam matrix must be created .
this coupling is described by the following propagation rules @xmath104 @xmath105 @xmath106 @xmath107 and , as it can be shown by direct calculations , it really does not allow to the terms of the order higher than two with respect to the variables @xmath64 , @xmath65 , @xmath66 , and @xmath67 to appear in the right hand side of the formula ( [ evolutionet ] ) .
transformation of the transverse to longitudinal coupling terms in accordance with the transport rule ( [ s01 ] ) produces the following changes in the beam dispersions @xmath108 and the following changes in the beam tilts @xmath109 @xmath110 where the parameter @xmath111 is defined by the expression @xmath112 the transformation of the beam energy chirp @xmath113 is given by the above introduced parameter @xmath111 @xmath114 and for the description of the change in the rms bunch length squared @xmath43 the new quadratic form is needed again .
this time it must be quadratic form not in four but in five variables , because as far as the evolution of the projected emittances does not depend from the @xmath70 matrix coefficient , the evolution of the bunch length certainly does .
so , let us introduce quadratic form @xmath115 associated with the positive definite matrix @xmath116 and represent the evolution of @xmath43 in the form @xmath117 @xmath118 @xmath119 where @xmath120 note that , if for some reasons the variation of the @xmath70 coefficient is not allowed and it can be treated as a given parameter , then one can return to the usage of quadratic form in four variables and rearrange the formulas ( [ evolutionbl11 ] ) and ( [ optsolutionbl11 ] ) as follows : @xmath121 @xmath122 @xmath123 where now @xmath124 @xmath125
with the formulas developed in the previous section for the emittance transport the problem of optimization of transverse projected emittances by an appropriate choice of the lattice dispersions becomes ( at least from the theoretical point of view ) fairly simple and straightforward .
for example , the formula ( [ evolutionex ] ) tell us that the change in the horizontal projected emittance @xmath92 after the system passage is the same for all lattice dispersions @xmath64 and @xmath65 belonging to the same level set @xmath126 because the function @xmath80 is a positive definite quadratic form its level sets for @xmath127 are ellipses all centered at the same point @xmath128 and contracting to this point as @xmath129 .
the level set @xmath130 plays a special role .
it separates the lattice dispersions which lead to the emittance increase from the lattice dispersions which provide emittance reduction or preservation . the level surface ( [ ellipseex ] )
is an ellipse if at least one horizontal beam dispersion is nonzero at the correction system entrance and it is a point coinciding with the common center ( [ ellipseexc ] ) of all ellipses ( [ ellipseex01 ] ) otherwise . in any case there
exists unique optimal choice ( optimal solution ) for the horizontal lattice dispersions which is given by the equation ( [ ellipseexc ] ) and which provides the largest possible reduction of the horizontal projected emittance @xmath92 ( the largest possible reduction is zero if both horizontal beam dispersions are equal to zero ) . by analogy
, the optimal solution for the transport of the vertical projected emittance @xmath93 is reached in the point @xmath131 and the optimal solution for the transport of the complete 2d transverse projected emittance @xmath132 is given by the union of the solution for the horizontal motion ( [ ellipseexc ] ) and the solution for the vertical motion ( [ ellipseeyc ] ) , which is a very pleasant fact ( in general , if the chosen lattice dispersions are different from the optimal solution , then the reduction of both @xmath92 and @xmath93 does not guarantee the reduction of @xmath132 , and vice versa ) .
one sees that the values of the lattice dispersions required for the simultaneous minimization of all transverse projected emittances are completely determined by the values of the beam dispersions and the beam rms energy spread , but , even if these quantities are unknown and there is no appropriate diagnostics to measure them , the projected emittances still can be optimized if there is a possibility to measure the horizontal and vertical projected emittances downstream of the correction system .
in this situation , minimization of emittances can be done iteratively ( and independently for horizontal and vertical degrees of freedom ) employing one of the many available effective algorithms for minimizing a convex quadratic objective function of two variables .
the optimal solution for all four lattice dispersions @xmath64 , @xmath65 , @xmath66 , and @xmath67 can be written in the form @xmath133 and therefore satisfies the orthogonality condition @xmath134 with the optimal choice of the lattice dispersions ( [ optsol01 ] ) the beam dispersions are zeroed at the correction system exit and the tilts are transformed according to the rule @xmath135 @xmath136 where , as one sees , the dependence from the @xmath70 matrix coefficient presented in the formula ( [ evolutionbt ] ) disappeared , though no assumptions were made about this coefficient and the optimal lattice dispersions ( [ optsol01 ] ) also do not depend on it .
the transport of the transverse coupling terms is given now by the formulas @xmath137 @xmath138 @xmath139 @xmath140 and one sees that if both , horizontal and vertical , beam dispersion vectors are nonzero at the entrance , then the interplay between them during the passage of the dispersive part of the correction system becomes a source of the transverse coupling .
this coupling could be removed by adding to the correction system an appropriate set of the skew quadrupoles and , therefore , the 1d transverse projected emittances can be reduced even further . in order to find better expressions for the reduction of the transverse projected emittances than the expressions which one can obtain by the direct substitution of the optimal solution ( [ optsol01 ] ) into the emittance propagation formulas ,
let us introduce positive definite quadratic forms @xmath141 and @xmath142 associated with the positive definite matrices @xmath143 and @xmath144 respectively .
the advantage of these quadratic forms over the quadratic forms @xmath80 and @xmath94 is that the elements of their matrices @xmath145 and @xmath146 are functions of the transverse beam moments only and do not depend on the beam moments involving the longitudinal variable @xmath6 as do the elements of the matrices @xmath78 and @xmath79 . besides that @xmath147 @xmath148 where
@xmath149 @xmath150 are the familiar ( but nonnormalized ) horizontal and vertical courant - snyder quadratic forms . with the help of the quadratic forms @xmath141 and
@xmath142 the evolution of the transverse projected emittances for the optimal choice of the lattice dispersions can be expressed as follows : @xmath151 @xmath152 @xmath153 @xmath154 @xmath155 and for the longitudinal projected emittance one obtains : @xmath156 @xmath157 , \label{evolutionezopt}\end{aligned}\ ] ] where @xmath158 one sees that the influence of the energy chirp @xmath42 on the propagation of the longitudinal projected emittance , which was presented in the formula ( [ evolutionez ] ) through the solution of the equation ( [ optsolutionz ] ) , is now canceled . as concerning the transport of the energy chirp itself ,
it is simplified owing to the orthogonality condition ( [ optsol02 ] ) to the form @xmath159 and the propagation formula for the rms bunch length squared @xmath43 , if needed , can be obtained using the equations ( [ evolutionezopt ] ) and ( [ energychirpopt ] ) , and the relation @xmath160 the formulas ( [ evolutionetopt ] ) and ( [ evolutionezopt ] ) for the transport of the 2d transverse projected emittance @xmath132 and the longitudinal projected emittance @xmath91 can be further simplified if one assumes that the conditions ( [ nocouplxy678 ] ) hold and the transverse degrees of freedom in the beam matrix @xmath31 are decoupled from each other at the correction system entrance . with this assumption
matrices @xmath145 and @xmath146 become block diagonal , quadratic forms @xmath141 and @xmath142 get representations @xmath161 @xmath162 and , as the result , one obtains @xmath163 @xmath164 , \label{evolutionetopt2}\end{aligned}\ ] ] @xmath165 @xmath166 @xmath167 note that the formula ( [ optlongemmnocoupl ] ) also can be obtained from equations ( [ coplingxyopt])-([coplingpxpyopt ] ) , ( [ evolutionexopt])-([evolutioneyopt ] ) and conditions ( [ nocouplxy678 ] ) using conservation of the lysenko invariant @xcite @xmath168 @xmath169 during symplectic transport of the beam matrix .
as the formulas ( [ evolutionez ] ) and ( [ optsolutionz ] ) show , the problem of optimization of the longitudinal projected emittance by the proper choice of the lattice dispersions ( when considered alone ) has the same geometry as the corresponding problems for the transverse projected emittances .
the resulting @xmath91 increase or reduction depends on the positioning of the chosen lattice dispersions @xmath64 ,
@xmath65 , @xmath66 , and @xmath67 with respect to the four dimensional ellipsoid but , in contrast to the transport of the transverse projected emittances , the longitudinal projected emittance can be reduced even when all beam dispersions are equal to zero at the correction system entrance , because ( according to the equation ( [ optsolutionz ] ) ) the optimal solution ( [ evolutionezls2 ] ) depends also on the values of the beam tilts . it is outside of the purpose of this paper to make a detailed study of the influence of the solution ( [ evolutionezls2 ] ) on the propagation of the other beam parameters and let us only note that it makes the vector of the beam dispersions and the vector of the beam tilts linearly dependent ( parallel ) at the exit of the dispersive part of the correction system .
it comes from the fact that the choice of the lattice dispersions according to the equations ( [ evolutionezls2 ] ) gives us the following transport rule for the beam dispersions thus , what is important for our further considerations , both solutions ( [ optsol01 ] ) and ( [ evolutionezls2 ] ) zero the last two terms in the lysenko invariant ( [ lysenkoinvariant ] ) at the exit of the correction system .
[ [ conditions - for - simultaneous - optimization - of - transverse - and - longitudinal - projected - emittances ] ] conditions for simultaneous optimization of transverse and longitudinal projected emittances ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ with our approach , the problem of the simultaneous optimization of the selected projected emittances by a proper choice of the lattice dispersions becomes a geometrical problem .
for example , emittances @xmath92 and @xmath91 can be decreased simultaneously if and only if the surfaces ( [ evolutionezls ] ) and ( [ ellipseex ] ) are non - degenerate ( i.e. they are not points but real ellipsoids ) and the projection of the inner points of the ellipsoid ( [ evolutionezls ] ) onto the plane @xmath177 has nonempty intersection with the set of the inner points of the ellipse ( [ ellipseex ] ) . the optimal solutions ( [ optsol01 ] ) and ( [ evolutionezls2 ] ) will be equal to each other and therefore the maximal possible reductions will be achieved for all , horizontal and vertical , projected emittances simultaneously ,
if and only if the following relations between the elements of the beam matrix @xmath31 hold we will discuss these relations in more detail in the following subsections and now let us only point out that the requirement for solution ( [ optsol01 ] ) to coincide with the first four components of the vector which corresponds to the complete chirp removal at the correction system exit .
note that the choice of the lattice dispersions ( including setting of the @xmath70 coefficient ) to be equal to the values ( [ vecs2opt ] ) minimizes the rms bunch length squared @xmath43 after the correction system passage .
according to the relations ( [ evolutionbt ] ) and ( [ lambdad ] ) the beam tilts can be zeroed at the correction system exit by an appropriate choice of the correction lattice dispersions if and only if the system of equations has at least one real solution with respect to the variables @xmath64 , @xmath65 , @xmath66 , @xmath67 , and @xmath70 . in general , it is a nonlinear system .
nevertheless , as we prove below , it always has at least one real solution for every fixed real value of the @xmath70 coefficient , which therefore can be treated as a parameter . to show this ,
let us assume first that @xmath111 in the system ( [ zbt01 ] ) is not simply a notation introduced for brevity , but is an additional real - valued variable , and let us consider an extended system consisting of equations ( [ zbt01 ] ) and ( [ zbt02 ] ) .
now we want to apply the method of successive elimination of variables to the system obtained and with this purpose in mind let us observe that is invertible .
it means that for every real value of @xmath111 equations ( [ zbt01 ] ) can be solved with respect to the variables @xmath64 , @xmath65 , @xmath66 , @xmath67 and the solution is unique . substituting this solution into equation ( [ zbt02 ] ) and multiplying both sides of the result by we obtain the polynomial equation of the fifth degree with respect to the single variable @xmath111 ( consistency equation ) and , because the order of this equation is odd
, it always must have at least one real root .
so we proved that the zeroing of the beam tilts by an appropriate choice of the correction lattice dispersions is always possible .
at least one solution can be found for all real values of the @xmath70 matrix coefficient and , for the fixed @xmath70 value , the number of solutions can vary from one to five .
the effect of the zeroing of the beam tilts in the matrix ( [ numm01 ] ) on projected emittances is presented at figures 1 and 2 , where the resulting emittances are shown for all possible real solutions of the equation ( [ numm03 ] ) .
one has to compare these emittances with the emittances @xmath197 , @xmath198 of the original beam matrix ( [ numm01 ] ) and with the emittances @xmath199 , @xmath200 which can be obtained after removal of the beam dispersions .
the example considered in the previous subsection tells us that zeroing of the beam tilts does not necessarily implies reduction of the transverse projected emittances .
the situation , of course , will be different if zeroing of the beam tilts will simultaneously remove the beam dispersions , i.e. if the longitudinal and transverse degrees of freedom in the beam matrix @xmath31 will be decoupled from each other at the correction system exit .
the necessary and sufficient conditions for the complete transverse to longitudinal decoupling can be obtained by the requirement that the solution for the lattice dispersions ( [ optsol01 ] ) which removes the beam dispersions also zeros the beam tilts , and substituting ( [ optsol01 ] ) into the equations ( [ zbt01 ] ) and ( [ zbt02 ] ) we obtain ( without big surprise ) again the equations ( [ decouplcond1 ] ) , which were derived as conditions for the simultaneous minimization of all projected emittances .
the most difficult question , for which we do not have any good answers yet , is the question of the physical interpretation of the conditions ( [ decouplcond1 ] ) .
it is clear , for example , that if the distortions to the initially uncoupled beam matrix @xmath31 were produced by a magnetostatic system , then the decoupling also can be done by a magnetostatic system , but how such beam matrices can be described more intuitively and what are the other possibilities ? currently , as more physical example in the comparison with the conditions ( [ decouplcond1 ] ) description , we only can state that all beam matrices with equal eigenemittances ( definition and properties of eigenemittances can be found in @xcite ) always can be decoupled by a magnetostatic system .
it follows from the observation that the conditions ( [ decouplcond1 ] ) are equivalent to the property of the matrix @xmath201 to have zeros in the positions we have seen that if in the beam matrix @xmath31 there are nonzero correlations between energy of particles and their transverse positions and momenta , then the values of the transverse projected emittances can be reduced , but how these reduced emittances are related to the emittances of the particle beam before it was damaged by the csr wake ( or by some other effects ) remains , of course , completely unclear .
so , let us consider an example which would give at least some insights into this problem .
our choice of the matrix @xmath205 as a source of the growth of the projected emittances and also as a source of the transverse to longitudinal coupling is motivated by the following reasons : the matrix @xmath205 , from one side , is symplectic and therefore all changes which it introduces are reversible , but it is not the matrix of a magnetostatic system and it is interesting to see up to what extend the original projected emittances of the matrix @xmath206 can be recovered afterwards by a magnetostatic correction system . from the other side ,
this matrix , similar to the wake field action , provides transverse kick and energy loss to the particle depending on its longitudinal position within the bunch .
note that if parameters @xmath207 and @xmath208 in this matrix are related to each other in some special way , then the matrix @xmath205 becomes equal to the matrix of the thick - lens horizontally deflecting cavity when it is sandwiched between two drifts of equal negative lengths ( see , for example , @xcite ) .
the passage of the beam matrix ( [ form1960 ] ) through the system described by the matrix @xmath205 gives equal increase of horizontal and longitudinal projected emittances ( the vertical degree of freedom remains decoupled from the others and is ignored in the following considerations ) among the elements of the matrix @xmath31 .
it means that both projected emittances can be recovered by a magnetostatic correction ( and also the beam matrix can be decoupled ) if and only if horizontal and longitudinal projected emittances were equal in the beginning before the passage through the system described by the matrix @xmath205 . but
let us see what can be done if they were not .
so , as the next step , let the beam pass through the dispersive part of the downstream correction system and , because we would like to express the final results using the elements of the original matrix ( [ form1960 ] ) , let us consider the transformation the formulas ( [ evolutionex ] ) and ( [ evolutionez ] ) for the transport of the horizontal and longitudinal projected emittances , when adapted to the transport equation ( [ exam009 ] ) , can be rewritten as follows : note that in the above formulas @xmath230 is a positive definite quadratic form in two variables obtained from the quadratic form @xmath80 , and the exact expression for the @xmath231 matrix associated with the quadratic form @xmath230 is unimportant for the further consideration .
so , as one sees , if @xmath234 , then only the larger of the two can be repaired and even can be further reduced , but only on expense of the increase of the other
. nevertheless , even if the horizontal ( or longitudinal ) projected emittance can not be recovered to its original value , the distorted value ( [ example02 ] ) ( or ( [ example03 ] ) ) always can be reduced , as follows from the theory developed in this paper .
let us now consider three extreme cases : solution for the lattice dispersions which minimizes @xmath92 , solution which minimizes @xmath91 and solution which zeros beam tilts . even before making any calculations
, one can state that in all these three cases the sum will be conserved , which follows from the preservation of the lysenko invariant ( [ lysenkoinvariant ] ) and the fact that all these solutions make the vector of the beam dispersions and the vector of the beam tilts linearly dependent at the correction system exit .
note also that due to the conservation of the sum ( [ geom03 ] ) and due to the extremum properties of the solutions which minimize projected emittances , any solution which zeros beam tilts will give the final value of the transverse projected emittance which must lie between the values given by the solution which zeros beam dispersions and the solution which minimizes longitudinal projected emittance .
unfortunately , it is practically impossible to find the general solutions for the lattice dispersions which are required for the zeroing of the beam tilts in the analytical form , and we will give it only for the partial case when @xmath244 . with this assumption the solution for the zeroing of the beam tilts is unique and is given by the following formulas one can check that while for @xmath250 the result of ( [ example11 ] ) is always smaller than the result of ( [ example02 ] ) ( as expected ) , the values produced by the formulas ( [ lonem02 ] ) and ( [ exr8 ] ) for @xmath234 can reduce the distorted emittance ( [ example02 ] ) only under specific conditions . v.balandin , r.brinkmann , w.decking , and n.golubeva , `` twiss parameters of coupled particle beams with equal eigenemittances '' , proceedings of ipac 2012 , new orleans , louisiana , usa , may 20 - 25 , 2012 , tuppc008 . | we show that if in the particle beam there are linear correlations between energy of particles and their transverse positions and momenta ( linear beam dispersions ) , then the transverse projected emittances always can be reduced by letting the beam to pass through magnetostatic system with specially chosen nonzero lattice dispersions .
the maximum possible reduction of the transverse projected emittances occurs when all beam dispersions are zeroed , and the values of the lattice dispersions required for that are completely defined by the values of the beam dispersions and the beam rms energy spread and are independent from any other second - order central beam moments . besides that
, we prove that , alternatively , one can also use the lattice dispersions to remove linear correlations between longitudinal positions of particles and their transverse coordinates ( linear beam tilts ) , but in this situation solution for the lattice dispersions is nonunique and the reduction of the transverse projected emittances is not guaranteed . |
the experimental spectra from @xcite were fitted to the toyozawa formula instead of the absorption sepctrum for low driving strength , from here on denoted as rabi spectrum , cf .
( 1 ) and ( 7 ) from the main text , respectively .
both functions may be easily fitted to the experiments . to compare them theoretically
, we will use the following parameters : the apparent resonance as the maximum of the distribution , the fwhm value , and the peak area .
the maximum of the distribution is given for a certain detuning between laser and actual exciton resonance @xmath69 due to the asymmetry .
hence , for @xmath70 , @xmath71 should become 0 as well for either model function . with this in mind
, we can determine the maximum unambigously to be @xmath72\ ] ] for the toyozawa formula and @xmath73\ ] ] for the rabi spectrum . in case of @xmath46
, we find a different relation between @xmath16 and @xmath15 , but without a theory for @xmath15 this can easily be included in the fit parameters .
moreover , in the region where @xmath74 and for @xmath75 , the maxima for both models become @xmath76 the fwhm value follows , after some lengthy algebra as @xmath77 for the toyozawa formula .
the rabi spectrum fwhm value requires the solution of a fourth - order polynomial .
more interestingly , when neglecting the asymmetry , we find for the rabi spectral width @xmath78 power broadening arises for sufficiently large @xmath11 , whereas the overall width is scaled down by a factor of around 1.55 .
crucially , the @xmath11-dependence of the linewidth for @xmath79 is not changed as the scaling is constant . for a similar small change of the width for @xmath80 for both model functions
, we again can not detect differences between the models for sufficiently low ratios of @xmath81 .
finally , the integrated peak area @xmath82 reads as @xmath83 for the toyozawa formula and @xmath84 for the rabi model . inserting the known dependencies of the different parameters we find that both scale only with @xmath13 for low pumping strengths , while for larger ones , the rabi area would scale with @xmath85 .
overall , we thus deduce that for low pumping strengths no detectable difference between these models can be found by freely fitting the model parameters to the experiments .
exemplarily , we have fitted the resonance @xmath86 in fig .
[ fig.comp ] with both models . applying the master equation @xcite for the three - level system , eq .
( 8) from the main text , we can derive the equations of motion for the relevant occupations @xmath87 and coherences @xmath88 , @xmath89 , respectively , which read as @xmath90 here , @xmath91 for @xmath92 .
equation ( [ eq.ajj ] ) and the first line of eq .
( [ eq.a0j ] ) represent the terms known from an isolated exciton resonance based on eq .
( 5 ) from the main text .
the only difference is due to the adjusted completeness relation @xmath93 however , the second line in eq .
( [ eq.a0j ] ) stems from the laser driven resonance @xmath94 and couples the two excited states .
it is based on the common ground state of both resonances allowing an effective quadrupolar coupling via the intermediate state @xmath28 .
the equation of motion for such a coupling @xmath95 then reads as @xmath96\langle\hat a_{kj}\rangle\nonumber\\ & -i\omega_j\langle[\hat a_{kj},\hat a_{j0}]\rangle - i\omega_k\langle[\hat a_{kj},\hat a_{0k}]\rangle\\ = & -[i(\delta_j-\delta_k)+\tfrac{\gamma_j+\gamma_k}{2}]\langle\hat a_{kj}\rangle\nonumber\\ & -i\omega_j\langle\hat a_{k0}\rangle+i\omega_k\langle\hat a_{0j}\rangle.\end{aligned}\ ] ] from the above , two effects couple the resonances @xmath11 and @xmath55 , the completeness relation and the nonvanishing value of @xmath97 .
as we do not yet have a full theory for the peak asymmetry @xmath15 , we can use the following argument . according to toyozawa @xcite ,
the asymmetry comprises couplings to all other excited states via a phonon background . for large @xmath11
, one may assume that this coupling is rather insensitive as the main couplings to low-@xmath11 states is almost the same for similarly large values of @xmath11 .
in this case the @xmath15 in the rabi model would become constant , @xmath98 for @xmath99 . with this in mind
, we can set the toyozawa and the rabi spectrum equal and determine the apparent @xmath15 fitted in the former .
this yields @xmath100 ^ 2}q\\ & + \frac{\gamma_n}{4\delta_n}\left[1-g_n^2\omega_n^2\frac{(\tfrac{\gamma_n}{2})^2+\delta_n^2}{[(\tfrac{\tilde\gamma_n}{2})^2+\delta_n^2]^2}\right ] \end{split}\ ] ] here , @xmath101 in order to include the different fitted widths from the two models .
unsurprisingly , @xmath15 is now a function of @xmath9 , meaning that we can only evaluate it at certain laser frequencies .
we thus evaluate the arithmetic average between the values at the two half width values @xmath102 . including again the different @xmath11-dependencies
, we obtain after some algebra @xmath103 with @xmath104 and @xmath105 positive constants .
for @xmath106 we reobtain a constant behaviour of the apperent asymmetries , whereas for large @xmath11 , even though @xmath107 is constant @xmath15 goes down to zero .
this behaviour seems to be roughly confirmed by the experimental fitting to the toyozawa spectrum , see fig .
[ fig.qn ] .
f. schne , s .- o .
krger , p. grnwald , j. thewes , m. amann , j. hecktter , d. frhlich , m. bayer , h. stolz , and s. scheel , _ deviations of the exciton level spectrum in cu@xmath0o from the hydrogen series _ , phys .
b * 93 * , 075203 ( 2016 ) .
f. schne , s .- o .
krger , p. grnwald , m. amann , j. hecktter , j. thewes , h. stolz , d. frhlich , m. bayer , and s. scheel , _ coupled valence band dispersions and the quantum defect of excitons in cu@xmath0o _ , j. phys .
b : at . mol .
opt . phys . * 49 * , 134003 ( 2016 ) .
j. p. reithmaier , g. sk , a. lffler , c. hofmann , s. kuhn , s. reitzenstein , l. v. keldysh , v. d. kulakovskii , t. l. reinecke , and a. forchel , _ strong coupling in a single quantum dot - semiconductor microcavity system _ , nature * 432 * , 197 ( 2004 ) .
t. yoshie , a. scherer , j. hendrickson , g. khitrova , h. m. gibbs , g. rupper , c. ell , o. b. shchekin , and d. g. deppe , _ vacuum rabi splitting with a single quantum dot in a photonic crystal nanocavity _ ,
nature * 432 * , 200 ( 2004 ) .
a. muller , e. b. flagg , p. bianucci , x. y. wang , d. g. deppe , w. ma , j. zhang , g. j. salamo , m. xiao , and c. k. shih , _ resonance fluorescence from a coherently driven semiconductor quantum dot in a cavity _ , phys .
lett . * 99 * , 187402 ( 2007 ) . | coherent optical control of individual particles has been demonstrated both for atoms and semiconductor quantum dots . here
we demonstrate the emergence of quantum coherent effects in semiconductor rydberg excitons in bulk cu@xmath0o . due to the spectral proximity between two adjacent rydberg exciton states , a single - frequency laser may pump both resonances with little dissipation from the detuning . as a consequence ,
additional resonances appear in the absorption spectrum that correspond to dressed states consisting of two rydberg exciton levels coupled to the excitonic vacuum , forming a v - type three - level system , but driven only by one laser light source .
we show that the level of pure dephasing in this system is extremely low .
these observations are a crucial step towards coherently controlled quantum technologies in a bulk semiconductor .
[ [ introduction . ] ] introduction .
+ + + + + + + + + + + + + rydberg atoms are understood to be those whose valence electron has been promoted to a quantum state with principal quantum number @xmath1 .
these exotic quantum states feature large coupling strengths to other rydberg states and extremely long life times ( scaling with @xmath2 and @xmath3 , respectively ) , making them an excellent choice for the study of quantum aspects of light - matter interaction @xcite .
rydberg atoms have been used in the first direct observation of quantum collapses @xcite as well as quantum non - demolition experiments @xcite . in recent years , they have matured into a promising platform for quantum information research @xcite . rydberg physics is not confined to the realm of atomic physics . recently ,
rydberg states with principal quantum numbers up to @xmath4 were observed in the semiconductor @xmath5 @xcite . here ,
mott - wannier excitons , bound states of electrons and holes , substitute for real atoms .
a number of concepts developed within the framework of rydberg atoms are displayed by these exotic states , such as typical scaling laws for the natural linewidth or the oscillator strength ( @xmath6 @xcite ) .
additionally , a quantum defect associated with an effective quantum number @xmath7 @xcite was observed and shown to be mainly due to the nonparabolicity of the valence bands @xcite .
a hitherto unexplained feature is the appearance of additional resonances in the absorption spectra of rydberg excitons for principal quantum numbers larger than @xmath8 ( see fig .
[ fig.expadd ] ) .
they occur halfway between each pair of excitonic resonances after excitation with a narrow - bandwidth laser , but vanish for large laser powers , where rydberg blockade sets in .
these resonances constitute a significant deviation from simple rydberg models , as well as established methods of many - body physics @xcite .
we will demonstrate here that they represent coherent ( quantum ) interactions between exciton states of different principal quantum numbers .
in the following , we consider quantum coherence in terms of the off - diagonal elements of the density matrix in fock basis .
an effect is based on quantum coherence if these off - diagonal elements are a necessary constituent of said effect . for open quantum systems , and in particular for the emission of light from a given source ,
this is achieved using a dominant coherent drive of the source of the emission .
such a dominance is usually diminished by dissipative effects such as energy and phase relaxation ( dephasing ) .
semiconductor systems are typically prone to strong phase relaxations due to phonon coupling and other many - body effects .
nevertheless , in low - dimensional systems such as quantum dots in microcavities , quantum coherent processes have been observed @xcite .
however , so far no such coherent effects could be demonstrated in bulk systems such as those used for rydberg - exciton generation .
the quantitative description of absorption and emission processes in semiconductors generally requires both the medium properties as well as a quantum - optical description of the emitters , here the excitons .
for example , the fluorescent emission of a coherently pumped quantum well required the combination of both the absorption ( not to be confused with the absorption spectrum ) as well as the spectral intensity distribution of the excitons @xcite . in that case , the necessity to combine both processes follows from kirchhoff s law @xcite and from the quantum - optical input - output relations @xcite . more recently , such an approach has also been applied to highly doped quantum wells in thermal pumping @xcite . using a similar approach ,
it becomes possible to infer that quantum coherence has been achieved in rydberg exciton systems directly from the experimental absorption spectra .
[ [ absorption - spectra . ] ] absorption spectra .
+ + + + + + + + + + + + + + + + + + + the absorption spectrum of a semiconductor is commonly measured via a frequency- insensitive detector such as a photodiode placed behind the sample .
this device registers the spectrally integrated transmission of light , which is then compared to the full laser intensity impinging on the crystal . a spectral distribution as shown in fig .
[ fig.expadd ] is obtained as a function of the laser frequency @xmath9 . the lost intensity is often assumed to be equal to the absorption , thereby neglecting secondary emission and losses due to scattering .
the commonly applied model for the absorption spectrum @xmath10 of an exciton with principal quantum number @xmath11 follows the toyozawa formula @xcite @xmath12 where @xmath13 is proportional to the oscillator strength and @xmath14 and @xmath15 are the resonance frequency of the exciton and the asymmetry parameter of the line shape , respectively .
the natural linewidth @xmath16 should scale with @xmath17 , as known from atomic systems .
according to elliott @xcite , in @xmath5 one also observes @xmath18 .
however , eq . ( [ eq.toyo ] ) only describes the probability density of absorbing a photon at frequency @xmath19 by the crystal , whereas the measurement setup detects the transmitted light as well as contributions from resonance fluorescence , which includes light scattered or reemitted along the forward direction .
therefore , in the event that a photon is indeed absorbed , @xmath10 is scaled by the resonance fluorescence contributions from the excitons @xmath20
@xcite . that scaled spectrum impinges on the frequency - insensitive detector yielding a spectrum @xmath21 the determined `` absorption '' spectrum of the @xmath5 crystal thus also contains components from coherent and incoherent emissions of the sample .
here the coherent contributions correspond to phase - preserving elastic scattering processes , which replicate the excitation field .
meanwhile , the incoherent contributions include inelastically scattered fields , which can result , for example , in the appearance of the mollow triplet for large rabi frequencies @xcite
. @xmath20 is determined by the spectral intensity distribution of the excitons @xmath22 via @xmath23 with @xmath24 being proportional to the dipole transition moment @xmath25 between the excitonic and vacuum states . in turn
, @xmath22 is directly connected to the quantum optical dynamics of the exciton states driven by the laser beam .
modelling adequately the quantum states involved in excitation and relaxation processes hence suffices to determine @xmath26 and compare with the experiments .
[ [ master - equation - approach . ] ] master equation approach .
+ + + + + + + + + + + + + + + + + + + + + + + + + in order to obtain the correct exciton dynamics , we employ the following description .
a single exciton resonance is modeled by only two states , the excited state @xmath27 and the excitonic vacuum @xmath28 as the ground state .
both states are dipole - coupled ( dipole moment @xmath29 ) by the pump laser with rabi frequency @xmath30 .
further , the excited state undergoes radiative , markovian damping with energy relaxation rate @xmath16 . according to standard wigner
weikopf theory for the damping , we find that @xmath31 the hamiltonian and the corresponding master equation for the density operator follow a bloch theory : @xmath32,\\ \frac{d\hat\varrho_n}{dt}=&\frac{1}{i\hbar}[\hat h_n,\hat\varrho_n ] + \frac{\gamma_n}{2}\mathcal{l}_{\hat a_{0,n}}[\hat\varrho_n],\\ \mathcal{l}_{\hat x}[\hat\varrho]=&2\hat x\hat\varrho\hat x^\dagger -\hat x^\dagger\hat x\hat\varrho-\hat\varrho\hat x^\dagger\hat x , \end{split}\ ] ] with @xmath33 being the excitonic operators and @xmath34 . here , @xmath35 describes the lindblad term for operator @xmath36 , in this case the exciton annihilation operator @xmath37 .
it is instructive to discuss the limit of weak light - matter coupling first , as that is usually expected in these experiments . for low excitation , @xmath38
, the incoherent spectral components from the excitons can be neglected as well as excitation - induced dephasing via exciton - exciton scattering .
thus , the spectral distribution of the excitons reduces to a rayleigh component @xcite @xmath39 , where @xmath40 is the coherent part of the exciton intensity .
thus , we find in this limit @xmath41 meaning that fitting the observed spectra in ref .
@xcite to eq .
( [ eq.toyo ] ) is sufficiently accurate if @xmath42 can be regarded as insensitive to variations of the laser frequency and the principal quantum number @xmath11 . for higher laser powers with respect to the light - matter couplings ,
this condition becomes invalid and @xmath40 must be investigated in more detail . in the steady - state solution of eq .
( [ eq.ham_n ] ) , @xmath43 , and the measured absorption spectrum becomes @xmath44 ^ 2}. \label{eq.cohfull}\ ] ] comparing eq .
( [ eq.cohfull ] ) with eq .
( [ eq.toyo ] ) , a few results should be noted .
first , the oscillator strength is calculated by integrating an absorption peak over all laser frequencies @xmath45 . in the linear regime ,
where @xmath46 , all @xmath11-dependencies apart from @xmath13 cancel out in the integrated transmission .
thus , the expected @xmath17-dependence of the oscillator strength is precisely what is observed in the experiment .
second , the corrected line shape of the spectrum differs slightly from the toyozawa formula , with the denominator being squared .
this yields deviations to the fit parameters @xmath15 and @xmath16 in a least - squares fit .
the line shape of isolated resonances can be fitted well to both models , eqs .
( [ eq.toyo ] ) and ( [ eq.cohfull ] ) , making a differentiation impossible ( see som ) .
this means that our corrected model can fully explain the standard toyozawa line shape observed in earlier experiments .
overall we can conclude that , in the linear regime and considering only isolated resonances , we find no indication of additional resonances .
third , and notwithstanding the previous statements the approximation of using solely toyozawa s formula breaks down , however , when reaching the nonlinear regime by increasing either the laser intensity or the principal quantum number @xmath11 of the exciton resonance . for @xmath47 , power broadening sets in , yielding a perceived larger line width and a steeper slope for the fitted oscillator strength .
given the scaling @xmath48 from eq .
( [ eq.deps ] ) , in the strong - coupling limit ( @xmath49 ) the area under the absorption peak scales as @xmath50 compared to the @xmath17-scaling in the weak - coupling regime .
[ [ additional - resonances - in - the - absorption - spectrum . ] ] additional resonances in the absorption spectrum . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + at intermediate excitation strengths @xmath47 , the incoherent part of the exciton spectrum becomes relevant , i.e. @xmath22 is no longer proportional to a @xmath51-distribution but instead broadens .
the incoherent spectrum of a single two - level system and its quantum features are well documented in the literature @xcite .
one may assume that the additional resonances result from strong - coupling effects , representing light - matter dressed states .
however , at the given laser intensities , we are not yet in the strong - coupling regime , and thus , only a single peak occurs in the exciton spectrum for one isolated rydberg resonance .
additional peaks resulting from strong coupling effects would show different spectral behaviour , such as the mollow - triplet .
hence , an isolated exciton state @xmath11 is insufficient to explain the additional resonances .
however , excitons with large @xmath11 can no longer be considered spectrally isolated , see fig .
[ fig.expadd ] .
these resonances are spectrally close , e.g. , @xmath52 hence , tuning the laser frequency around a high-@xmath11 resonance , different states @xmath53 of the exciton can be populated with limited influence by the detuning .
it is therefore appropriate to consider a system of two exciton states @xmath27 and @xmath54 as the possible excited states , and the excitonic vacuum @xmath28 as their common ground state .
the resonances do not couple directly to one another but they are dipole - coupled to the vacuum state and driven by the same laser light , yielding a v - type three - level system with a single pump laser of frequency @xmath9 ( see fig .
[ fig.3ls ] ) .
note that we do not consider two excitons , but the presence of a single exciton in a system with different possible excited quantum states . and
@xmath55 , coupled to the excitonic vacuum state @xmath28.,width=264 ] comparing with eq .
( [ eq.ham_n ] ) , we now take into account the sum of the two hamiltonians , @xmath56 and @xmath57 , with the corresponding lindblad terms and relaxation rates @xmath16 and @xmath58 : @xmath59\\ & + \frac{\gamma_n}{2}\mathcal{l}_{\hat
a_{0,n}}[\hat\varrho_\text{3ls}]+ \frac{\gamma_{n+1}}{2}\mathcal{l}_{\hat a_{0,n+1}}[\hat\varrho_\text{3ls } ] .
\end{split}\ ] ] the common ground state @xmath28 permits a transfer of an excitation from one exciton state to the other , forming an effective quadrupolar coupling , i.e. @xmath60 , see som for details .
hence , the coherent drive induces an effective transition between the exciton states , whose transition rate is directly linked to the rabi frequencies @xmath30 and @xmath61 . the incoherent part of the three - level exciton intensity spectrum @xmath22 is displayed in fig .
[ fig.3ls_x_spec ] for the laser frequency @xmath62 . with increasing rabi frequency ( curves from bottom to top )
, an internal resonance builds up .
this resonance corresponds to a dressed state of the two excitons , formed by the effective quadrupole coupling .
the measured absorption spectrum @xmath26 , following eq .
( [ eq.ameas ] ) , now includes @xmath22 from the three - level hamiltonian and the toyozawa formula ( [ eq.toyo ] ) .
it features prominently the additional resonances from the experiments ( see bottom panel in fig . [ fig.rabivar ] ) . for two resonances with @xmath63 , coupled to the excitonic vacuum with the parameters taken from ref .
@xcite .
the laser is tuned to the center between the resonances and the rabi frequencies are @xmath64 . ] to account for the discrepancies between system parameters determined via fitting toyozawa s formula and the complex three - level model , we apply the following rules : the decay rates @xmath16 follow the basic @xmath17 law , fitted to the lower @xmath11- resonances .
deviations from this behaviour are thus thought to be included in the full description .
the @xmath15 are taken from the fit parameters . for a simple model discussion on their behaviour see the som .
the rabi frequencies were not fitted in @xcite , so we will use values that reproduce the experiments
. further increase of the laser intensity results in two effects ( see top panel in fig . [ fig.rabivar ] ) .
first , the additional resonances develop an asymmetry towards the lower - frequency side of the spectrum .
second , for even higher field intensities , they start to disappear , while the absorption in general decreases .
the former effect follows directly from our model hamiltonian , where the combination of symmetric exciton intensity spectrum and the asymmetric toyozawa formula tilts the additional resonances ( see lower panel in fig .
[ fig.rabivar ] ) .
the regime of strong pumping is signified by the latter effect .
the dressed states are shifted out of the spectral region by the large rabi frequency .
furthermore , for strong driving , the dipole- and quadrupole moments decrease because of the saturated level occupation . in that case
, the effective system reverts back to the isolated resonance scenario . also ,
at this point the rydberg blockade sets in @xcite .
if one exciton is excited to a rydberg state , the excitation of another one becomes suppressed .
note that rydberg blockade depends strongly on parameters like sample thickness and the exact geometry used for excitation and has thus not been included in the theoretical results shown in the lower panel of fig .
[ fig.rabivar ] . and
different laser intensities as shown in the legend .
bottom : absorption spectrum for the two resonances @xmath63 for different rabi frequencies with @xmath64.,title="fig : " ] and different laser intensities as shown in the legend .
bottom : absorption spectrum for the two resonances @xmath63 for different rabi frequencies with @xmath64.,title="fig : " ] so far , we have limited our discussion to energy relaxation , neglecting the influence of pure dephasing .
if the additional resonances are indeed based on quantum coherent coupling of the different resonances , increasing the pure dephasing rate ( by inclusion of another lindblad term @xmath65 $ ] into the master equation ) should influence their magnitude . indeed , as shown in fig .
[ fig.deph ] , already for dephasing rates close to the pumping rate , the resonances disappear .
two conclusions can be drawn from this .
first , radiationless dephasing is surprisingly small in rydberg excitons in @xmath5 and second , the additional resonances appearing between the exciton resonances are clear signatures of quantum coherence between adjacent exciton states .
remarkably , these coherences between exciton states with consecutive quantum numbers are driven by the incoherent part of the exciton spectrum instead of the coherent part , which drives the coherences between the ground state and the exciton states .
we also note that the curves for higher dephasing rates are very similar to the experimental curves for higher pumping , see top panel in fig .
[ fig.rabivar ] .
it seems reasonable that , when increasing the laser intensity beyond saturation , phonons become dominantly excited and induce stronger dephasing rates .
this can only be better understood by a thorough analysis of the decay rates and rabi frequencies present in the experiments . for different dephasing rates
@xmath66 and @xmath67 @xmath68ev and @xmath64 . ]
[ [ summary - and - conclusions . ] ] summary and conclusions .
+ + + + + + + + + + + + + + + + + + + + + + + + we have analyzed the quantum coherent properties of the rydberg exciton spectrum in @xmath5 .
the observed absorption in different experimental situations includes , besides the absorption - induced attenuation , also the intensity spectrum of the rydberg excitons . in the linear regime of a coherently driven system
, this additional effect simply scales the absorption spectrum . for stronger excitation ,
the incoherent part of the exciton spectrum induces additional resonances , which were also observed in the experiments .
their appearance indicates the onset of measurable quantum coherence between exciton states of consecutive principal quantum number , as well as very low levels of pure dephasing .
our model for the absorption spectrum of rydberg excitons accounts for a number of hitherto unexplained features , which should also be applicable to a variety of similar semiconductor experiments .
rydberg excitons in @xmath5 have now reached a stage at which coherent quantum effects become visible , and the precise nature of the excitonic hamiltonian becomes detectable .
this allows for controlled quantum manipulation and state generation for excitons and opens up the possibility for developing semiconductor quantum technologies .
[ [ acknowledgements . ] ] acknowledgements .
+ + + + + + + + + + + + + + + + + we gratefully acknowledge support by the collaborative research centre sfb 652/3 strong correlations in the radiation field and the international collaborative research centre trr 160 coherent manipulation of interacting spin excitations in tailored semiconductors , both funded by the deutsche forschungsgemeinschaft . |
random walk metropolis ( rwm ) algorithms are widely used generic markov chain monte carlo ( mcmc ) algorithms .
the ease with which rwm algorithms can be constructed has no doubt played a pivotal role in their popularity .
the efficiency of a rwm algorithm depends fundamentally upon the scaling of the proposal density .
choose the variance of the proposal to be too small and the rwm will converge slowly since all its increments are small . conversely , choose the variance of the proposal to be too large and too high a proportion of proposed moves will be rejected . of particular interest is how the scaling of the proposal variance depends upon the dimensionality of the target distribution .
the target distribution is the distribution of interest and the mcmc algorithm is constructed such that the stationary distribution of the markov chain is the target distribution .
the introduction is structured as follows .
we outline known results for continuous independent and identically distributed product densities from @xcite and subsequent work .
we highlight the scope and limitations of the results before introducing the discontinuous target densities to be studied in this paper . while the statements of the key results ( theorem [ main ] ) in this paper are similar to those given for continuous target densities , the proofs are markedly different .
a discussion of why a new method of proof is required for discontinuous target densities is given .
finally , we give an outline of the remainder of the paper .
the results of this paper have quite general consequences for the implementation of metropolis algorithms on discontinuous densities ( as are commonly applied in many bayesian statistics problems ) , namely : full- ( high- ) dimensional update rules can be an order of magnitude slower than strategies involving smaller dimensional updates .
( see theorem [ thmprop ] below . ) for target densities with bounded support , metropolis algorithms can be an order of magnitude slower than algorithms which first transform the target support to @xmath5 for some @xmath0 . in @xcite ,
a sequence of target densities of the form @xmath6 were considered as @xmath7 , where @xmath8 is twice differentiable and satisfies certain mild moment conditions ; see @xcite , ( a1 ) and ( a2 ) .
the following random walk metropolis algorithm was used to obtain a sample @xmath9 from @xmath10 .
draw @xmath11 from @xmath12 . for @xmath13 and @xmath14 let @xmath15 be independent and identically distributed ( i.i.d . ) according to @xmath16 and @xmath17 . at time @xmath18 , propose @xmath19 where @xmath20 is the proposal standard deviation to be discussed shortly .
set @xmath21 with probability @xmath22 otherwise set @xmath23 .
it is straightforward to check that @xmath24 has stationary distribution @xmath12 , and hence , for all @xmath13 , @xmath25 .
the key question addressed in @xcite was : starting from the stationary distribution , how should @xmath20 be chosen to optimize the rate at which the rwm algorithm explores the stationary distribution ?
since the components of @xmath26 are i.i.d . , it suffices to study the marginal behavior of the first component , @xmath27 . in @xcite
, it was shown that if @xmath28 @xmath29 and @xmath30,1}^d$ ] @xmath31 , then @xmath32 where @xmath33 satisfies the langevin sde @xmath34 with @xmath35 and @xmath36 with @xmath37 being the standard normal c.d.f . and @xmath38 $ ] .
note that the `` speed measure '' of the diffusion @xmath39 only depends upon @xmath40 through @xmath41 .
the diffusion limit for @xmath42 is unsurprising in that for a time interval of length @xmath43 , @xmath44 moves are made each of size @xmath45 .
therefore the movements in the first component ( appropriately normalized ) converge to those of a langevin diffusion with the `` most efficient '' asymptotic diffusion having the largest speed measure @xmath46 .
since the diffusion limit involves speeding up time by a factor of @xmath0 , we say that the mixing of the algorithm is @xmath44 .
the optimal value of @xmath47 is @xmath48 , which leads to an average optimal acceptance rate ( aoar ) of 0.234 .
this has major practical implications for practitioners , in that , to monitor the ( asymptotic ) efficiency of the rwm algorithm it is sufficient to study the proportion of proposed moves accepted .
there are three key assumptions made in @xcite .
first , @xmath49 , that is , the algorithm starts in the stationary distribution and @xmath20 is chosen to optimize exploration of the stationary distribution .
this assumption has been made in virtually all subsequent optimal scaling work ; see , for example , @xcite and @xcite .
the one exception is @xcite , where @xmath11 is started from the mode of @xmath12 with explicit calculations given for a standard multivariate normal distribution . in @xcite , it is shown that @xmath50 is optimal for maximizing the rate of convergence to the stationary distribution .
since convergence is shown to occur within @xmath51 iterations , the time taken to explore the stationary distribution dominates the time taken to converge to the stationary distribution , and thus overall it is optimal to choose @xmath52 .
it is difficult to prove generic results for @xmath53 .
however , the findings of @xcite suggest that even when @xmath54 , it is best to scale the proposal distribution based upon @xmath55 .
it is worth noting that in @xcite it was found that for the metropolis adjusted langevin algorithm ( mala ) , the optimal scaling of @xmath20 for @xmath11 started at the mode of a multivariate normal is @xmath56 compared to @xmath57 for @xmath55 .
second , @xmath12 is an i.i.d .
product density .
this assumption has been relaxed by a number of authors with @xmath50 and an aoar of 0.234 still being the case , for example , independent , scaled product densities ( @xcite and @xcite ) , gibbs random fields @xcite , exchangeable normals @xcite and elliptical densities @xcite .
thus the simple rule of thumb of tuning @xmath20 such that one in four proposed moves are accepted holds quite generally . in @xcite and @xcite , examples where the aoar is strictly less than 0.234
are given .
these correspond to different orders of magnitude being appropriate for the scaling of the proposed moves in different components .
third , the results are asymptotic as @xmath7 .
however , simulations have shown that for i.i.d .
product densities an acceptance rate of 0.234 is close to optimal for @xmath58 ; see , for example , @xcite .
departures from the i.i.d .
product density require larger @xmath0 for the asymptotic results to be optimal , but @xmath59 is often seen in practical mcmc problems . in @xcite and @xcite , optimal acceptance rates are obtained for finite @xmath0 for some special cases . with the exceptions of @xcite and @xcite , in the above works @xmath60
is assumed to have a continuous ( and suitably differentiable ) probability density function ( p.d.f . ) .
the aim of the current work is to investigate the situation where the target distribution has a discontinuous p.d.f . , and specifically , target distributions confined to the @xmath0-dimensional hypercube @xmath61^d$ ] .
that is , we consider target distributions of the form @xmath62 where @xmath63 and @xmath64 is twice differentiable upon @xmath61 $ ] with @xmath65 we then use the following random walk metropolis algorithm to obtain a sample @xmath66 from @xmath12 .
draw @xmath11 from @xmath12 .
for @xmath13 and @xmath14 let @xmath67 be independent and identically distributed ( i.i.d . ) according to @xmath68 $ ] and @xmath69 . at time @xmath18 , propose @xmath70 set @xmath21 with probability @xmath71 otherwise set @xmath23 . in @xcite and @xcite ,
spherical and elliptical densities are considered which have very different geometry to the hypercube restricted densities . therefore different approaches are taken in these papers with results akin to those obtained for continuous target densities .
densities of the form ( [ eq1b ] ) have previously been studied in @xcite , where the expected square jumping distance ( esjd ) has been computed .
the esjd is @xmath72 = d { \mathbb{e}}_{\pi_d } [ ( x_{1,1}^d - x_{0,1}^d)^2 ] , \ ] ] the mean squared distance between @xmath11 and @xmath73 , where @xmath55 . in @xcite , appendix b , it is shown that for @xmath74 @xmath75 and @xmath76 , @xmath77 \rightarrow\frac{l^2}{3 } \exp\biggl ( - \frac{l}{2 }
\biggr ) \qquad\mbox{as } { d \rightarrow\infty}.\ ] ] thus asymptotically ( as @xmath7 ) the esjd is maximized by taking @xmath78 which corresponds to an aoar of @xmath79 ( @xmath800.1353 ) . in this paper , we show that @xmath74 and an aoar of @xmath79 holds more generally for target distributions of the form given by ( [ eq1a ] ) and ( [ eq1b ] ) .
moreover , we prove a much stronger result than that given in @xcite , in that , we prove that @xmath81,1}^d$ ] converges weakly to an appropriate langevin diffusion @xmath82 with speed measure @xmath83 as @xmath7 , where @xmath84 .
this gives a clear indication of how the markov chain explores the stationary distribution .
by contrast the esjd only gives a measure of average behavior and does not take account of the possibility of the markov chain becoming `` stuck . ''
if @xmath85 $ ] is very low , the markov chain started @xmath86 is likely to spend a large number of iterations at @xmath87 before accepting a move away from @xmath87 . note that since @xmath88 involves speeding up time by a factor of @xmath89 , we say that the mixing of the algorithm is @xmath90 .
the esjd is easy to compute and asymptotically , as @xmath7 , the esjd ( appropriately scaled ) converges to @xmath39 .
thus in discussing possible extensions of the langevin diffusion limit proved in theorem [ main ] for i.i.d .
product densities of the form given in ( [ eq1a ] ) and ( [ eq1b ] ) , we make considerable use of the esjd . however , we highlight the limitations of the esjd in discussing extensions of theorem [ main ] . in most previous work on optimal
scaling , the components of @xmath91 are taken to be independent and identically distributed @xmath16 random variables .
the reason for choosing @xmath68 $ ] for discontinuous target densities is mathematical convenience .
the results proved in this paper hold with gaussian rather than uniform proposal distributions , but some elements of the proof are less straightforward .
for discussion of the esjd for densities ( [ eq1a ] ) for general @xmath92 subject to @xmath93 < \infty$ ] , see @xcite , appendix b. while the key result , a langevin diffusion limit for the movement in the first component , is the same as @xcite , the proof is markedly different .
note that , for finite @xmath0 , @xmath42 and @xmath94 are not markov chains since whether or not a proposed move is accepted depends upon all the components in @xmath12 . in @xcite , it is shown that there exists @xmath95 such that @xmath96 } \{\mathbf{x}_t^d \notin f_d \ } ) \rightarrow0 $ ] as @xmath7 and @xmath97 - 2\phi\biggl ( -\frac{l \sqrt{i}}{2 } \biggr ) \biggr| \leq\varepsilon_d,\ ] ] where @xmath98 as @xmath7 . while ( [ eqrevb7 ] ) is not explicitly stated in @xcite ,
it is the essence of the requirements of the sets @xmath99 , stating that for large @xmath0 , with high probability over the first @xmath100 iterations the acceptance probability of the markov chain is approximately constant , being within @xmath101 of @xmath102 .
( note @xmath103 rather than @xmath0 is used for dimensionality in @xcite . )
thus in the limit as @xmath7 the effect of the other components on movements in the first component converges to a deterministic acceptance probability @xmath102 .
the situation is more complex for @xmath12 of the form given by ( [ eq1a ] ) and ( [ eq1b ] ) as the acceptance rate in the limit as @xmath7 is inherently stochastic .
for example , suppose @xmath12 is the uniform distribution on the @xmath0-dimensional hypercube so that @xmath104^d \}}$ ] .
letting @xmath105 and @xmath106 , this gives @xmath107 = \prod_{i \in r_d^l } \biggl ( \frac{1}{2 } + \frac{x_i}{2 \sigma_d } \biggr ) \times\prod_{i \in r_d^u } \biggl ( \frac{1}{2 } + \frac{1-x_i}{2 \sigma_d } \biggr).\ ] ] thus the acceptance probability is totally determined by the components at the boundary ( within @xmath20 of 0 or 1 ) .
the total number of components in @xmath108 is @xmath109 which converges in distribution to @xmath110 as @xmath7 .
thus the number of components close to the boundary is inherently stochastic .
moreover , the location of the components within @xmath111 plays a crucial role in the acceptance probability ; see ( [ eqrevb8 ] ) .
therefore there is no hope of replicating directly the method of proof applied in @xcite and subsequently , in @xcite and @xcite .
we need a homogenization argument which involves looking at @xmath112 over @xmath113 $ ] steps ; cf .
in particular , we show that the acceptance probability converges very rapidly to its stationary measure , so that over @xmath113 $ ] iterations approximately @xmath114 $ ] proposed moves are accepted . by comparison , @xmath115,1}^d - x_{0,1}^d | \leq[d^\delta ] \sigma_d$ ] ; thus the value of an individual component only makes small changes over @xmath113 $ ] iterations .
that is , we show that there exists @xmath116 such that , for any @xmath117 , @xmath118 } \{\mathbf{x}_t^d \notin\tilde{f}_d \ } ) \rightarrow0 $ ] as @xmath7 and for @xmath119 , @xmath120 } \sum_{t=0}^ { [ d^\delta]-1 } { \mathbb{e } } [ \alpha(\mathbf{x}^d_t , \mathbf{x}^d_t + \sigma_d \mathbf{z}_t^d ) | \mathbf{x}_0^d = \mathbf{x}^d ] - \exp \biggl ( - \frac{lf^\ast}{2 } \biggr ) \biggr| \leq\varepsilon_d\hspace*{-25pt}\ ] ] for some @xmath98 as @xmath7 . for large @xmath0 , with high probability over the first @xmath121 $ ] iterations the markov chain stays in @xmath122 , where the average number of accepted proposed moves in the following @xmath113 $ ] iterations is @xmath123 .
the arguments are considerably more involved than in @xcite , where spherically constrained target distributions were studied , due to the very different geometry of the hypercube and spherical constraints applied in this paper and @xcite , respectively .
in particular , in @xcite , @xmath28 with an aoar of @xmath124 . by exploiting the homogenization argument it is possible to prove that @xmath94 converges weakly to an appropriate langevin diffusion @xmath125 , given in theorem [ main ] . in section
[ secalg ] , theorem [ main ] is presented along with an outline of the proof .
also in section [ secalg ] , a description of the pseudo - rwm algorithm is given .
the pseudo - rwm algorithm plays a key role in the proof of theorem [ main ] .
the pseudo - rwm process moves at each iteration and the moves in the pseudo - rwm process are identical to those of the rwm process , conditioned upon a proposed move in the rwm process being accepted .
the proof of theorem [ main ] is long and technical with the details split into three key sections which are given in the ; see section [ secalg ] for more details . in section [ secextsim ] , two interesting extensions of theorem [ main ] are given .
in particular , theorem [ thmprop ] has major practical implications for the implementation of rwm algorithms by highlighting the detrimental effect of choosing rwm algorithms over metropolis - within - gibbs algorithms .
the target densities for which theoretical results can be proved are limited , so discussion of possible extensions of theorem [ main ] are given .
in particular , we discuss general @xmath60 restricted to the hypercube , general discontinuities in @xmath40 and @xmath53 .
we begin by defining the pseudo - random walk metropolis ( pseudo - rwm ) process . we will then be in position to formally state the main theorem , theorem [ main ] .
an outline of the proof of theorem [ main ] is given , with the details , which are long and technical , placed in the .
for @xmath126 , let @xmath127 let @xmath128 denote the probability of accepting a move in the rwm process given the current state of the process is @xmath87 .
then @xmath129 let @xmath130 , the total number of components of @xmath87 in @xmath131 . by taylor
s theorem for all @xmath132 and @xmath133 , @xmath134 with @xmath135 defined in ( [ eq1c ] ) .
hence , for all @xmath136^d$ ] , @xmath137^d \ } } \,d \mathbf{z}^d \nonumber\\ & \geq&\int h_d ( \mathbf{z}^d ) \ { 1 \wedge\exp(- dg^\ast \sigma_d ) \ } 1_{\ { \mathbf{x}^d + \sigma_d \mathbf { z}^d \in [ 0,1]^d \ } } \,d \mathbf{z}^d \nonumber\\[-8pt]\\[-8pt ] & = & \exp(-l g^\ast ) \int h_d ( \mathbf{z}^d ) 1_{\ { \mathbf { x}^d + \sigma_d \mathbf{z}^d \in[0,1]^d \ } } \,d \mathbf{z}^d \nonumber\\ & \geq & \exp(-l g^\ast ) \biggl ( \frac{1}{2 } \biggr)^{b_d^l ( \mathbf{x}^d)}.\nonumber\end{aligned}\ ] ] this lower bound for @xmath138 will be used repeatedly .
the pseudo - rwm process moves at each iteration , which is the key difference to the rwm process .
furthermore , the moves in the pseudo - rwm process are identical to those of the rwm process , conditioned upon a move in the rwm process being accepted , that is , its jump chain . for @xmath139 ,
let @xmath140 denote the successive states of the pseudo - rwm process , where @xmath141 .
the pseudo - rwm process is a markov process , where for @xmath13 , @xmath142 and given that @xmath143 , @xmath144 has p.d.f .
@xmath145 note that @xmath146 for @xmath147 . since @xmath148
, we can couple the two processes to have the same starting value @xmath11 .
a continued coupling of the two processes is outlined below .
suppose that @xmath149 .
then for any @xmath150 , @xmath151 that is , the number of iterations the rwm algorithm stays at @xmath87 before moving follows a geometric distribution with `` success '' probability @xmath128 .
therefore for @xmath152 , let @xmath153 denote independent geometric random variables , where for @xmath154 , @xmath155 denotes a geometric random variable with `` success '' probability @xmath156 . for @xmath157 ,
let @xmath158 and for @xmath159 , let @xmath160 where the sum is zero if vacuous . for @xmath161 , attach @xmath162 to @xmath163 .
thus @xmath164 denotes the total number of iterations the rwm process spends at @xmath163 before moving to @xmath165 .
hence , the rwm process can be constructed from @xmath166 by setting @xmath167 and for all @xmath150 , @xmath168 .
obviously the above process can be reversed by setting @xmath169 equal to the @xmath18th accepted move in the rwm process . for each @xmath126 ,
the components of @xmath11 are independent and identically distributed
. therefore we focus attention on the first component as this is indicative of the behavior of the whole process . for @xmath126 and @xmath13 , let @xmath170,1}^d$ ] and @xmath171,1}^d$ ] .
[ main ] fix @xmath172 .
for all @xmath126 , let @xmath173 .
then , as @xmath7 , @xmath174 in the skorokhod topology on @xmath175 , where @xmath176 satisfies the ( reflected ) langevin sde on @xmath61 $ ] @xmath177 with @xmath178 .
note that @xmath179 is standard brownian motion , @xmath180 and @xmath181 . here
@xmath182 denotes the local time of @xmath125 at @xmath183 @xmath184 and the sde ( [ eq2a1 ] ) corresponds to standard reflection at the boundaries @xmath185 and @xmath186 ( see , e.g. , chapter of @xcite ) .
as noted in section [ secint ] , the acceptance probability of the rwm process is inherently random and therefore it is necessary to consider the behavior of the rwm process averaged over @xmath113 $ ] iterations , for @xmath119 .
fix @xmath187 and let @xmath188 be a sequence of positive integers satisfying @xmath189 \leq k_d \leq[d^\delta]$ ] . for @xmath161
, let @xmath190}^d$ ] and for @xmath191 , let @xmath192,1}^d$ ] . for all @xmath13 , @xmath193 and @xmath194 - [ d^\delta ]
\times[d^2 t/[d^\delta]]| \leq[d^\delta]$ ] .
hence , for all @xmath195 , @xmath196 \sigma_d.\ ] ] therefore by @xcite , theorem 4.1 , @xmath197 as @xmath7 , if @xmath198 as @xmath7 .
hence we proceed by showing that @xmath199 let @xmath200 be the ( discrete - time ) generator of @xmath201 and let @xmath202 be an arbitrary test function of the first component only .
thus @xmath203 } { \mathbb{e } } [ h ( \tilde{\mathbf{x}}_1^d ) - h ( \tilde{\mathbf{x}}_0^d ) |
\tilde{\mathbf{x}}_0^d = \mathbf{x}^d].\ ] ] the generator @xmath204 of the ( limiting ) one - dimensional diffusion @xmath125 for an arbitrary test function @xmath202 is given by @xmath205 for all @xmath206 $ ] at least for all @xmath207 , where @xmath208 is defined in ( [ eqmain1a ] ) below .
first note that the diffusion defined by ( [ eqmain1 ] ) is regular ; see @xcite , page 366 .
therefore by @xcite , chapter 8 , corollary 1.2 , it is sufficient to restrict attention to functions @xmath209 ) \cap c^2 ( ( 0,1 ) ) \cap \mathcal{d}^\ast , gh \in\hat{c } ( [ 0,1 ] ) \},\ ] ] where @xmath210 is the set of twice differentiable functions upon @xmath211 , @xmath212 $ ] is the set of bounded continuous functions upon @xmath61 $ ] and @xmath213 is obtained by setting @xmath214 @xmath215 in @xcite , page 367 , ( 1.11 ) and is given by @xmath216 let @xmath217 and @xmath218 .
then @xmath219 combined with @xmath220 implies that @xmath221 .
it then follows from @xmath222 being bounded on @xmath61 $ ] and @xmath223)$ ] that @xmath224 .
these observations will play a key role in appendix [ secgen ] .
now ( [ eqn112 ] ) is proved using @xcite , chapter 4 , corollary 8.7 , by showing that there exists a sequence of sets @xmath225 such that for any @xmath117 , @xmath226 } \ { \mathbf{x}_j^d \notin\tilde{f}_d \ }
\biggr ) \rightarrow0 \qquad\mbox{as } { d \rightarrow\infty}\ ] ] and @xmath227 let the sets @xmath228 and @xmath229 be such that @xmath230 and @xmath231 } \ { \hat{\mathbf{x}}_j^d \notin f_d \ } | \hat{\mathbf{x}}_0^d = \mathbf{x}^d \biggr ) \leq d^{-3 } \biggr\},\ ] ] where @xmath232 , @xmath233 , @xmath234 and @xmath235 are defined below .
recall that @xmath236 , the total number of components of @xmath87 in @xmath237 .
we term @xmath238 the rejection region , in that , for any component in @xmath238 , there is positive probability of proposing a move outside the hypercube with such moves automatically being rejected .
let @xmath239}^{[d^\delta ] } \bigl\ { \mathbf{x}^d ; |b_d^{k^{3/4 } } ( \mathbf{x}^d ) - { \mathbb{e } } [ b_d^{k^{3/4 } } ( \mathbf{x}_0^d ) ] | \leq\sqrt{k } \bigr\ } , \\ \label{eqn11f3 } f_d^3 & = & \bigl\ { \mathbf{x}^d ; \sup_{[d^\beta]\leq k_d \leq[d^\delta ] } \sup_{0 \leq
r \leq l } | \lambda_d ( \mathbf{x}^d;r;k_d ) -
\lambda(r)| \leq d^{-\gamma } \bigr\ } , \\
\label{eqn11f4 } f_d^4 & = & \biggl\ { \mathbf{x}^d ; \biggl| \frac{1}{d } \sum_{j=1}^d g^\prime(x_j)^2 - { \mathbb{e}}_f [ g^\prime(x_1)^2 ]
\biggr| < d^{-{1}/{8 } } \biggr\},\end{aligned}\ ] ] where @xmath240 $ ] and @xmath241 . in appendix
[ secsets ] , we prove ( [ eqn114 ] ) for the sets @xmath229 given in ( [ emainx1 ] ) .
note that ( [ eqn114 ] ) follows immediately from theorem [ lem321 ] , ( [ eqss46 ] ) since @xmath173 .
an outline of the roles played by each @xmath242 @xmath243 is given below .
for @xmath244 ( @xmath245 ) the total number of components in ( _ close to _ ) the rejection region are controlled .
for @xmath246 after @xmath247 iterations the total number and position of the points @xmath248 in @xmath238 are approximately from the stationary distribution of @xmath249 .
finally , for @xmath250 , @xmath251 $ ] ; this is the key requirement for the sets @xmath99 given in @xcite , cf .
@xcite , page 114 , @xmath252 .
the proof of ( [ eqn115 ] ) splits into two parts and exploits the pseudo - rwm process .
let @xmath253 ; \frac{1}{[d^\delta ] } \sum_{j=0}^{k-1 } m_j ( j_d ( \hat{\mathbf{x}}_j^d ) ) \leq1 \biggr\ } \big/[d^\delta],\hspace*{-25pt}\ ] ] the proportion of accepted moves in the first @xmath113 $ ] iterations , where the sum is set equal to zero if vacuous . then @xmath254}^d = \hat{\mathbf{x}}_{[p_d d^\delta]}^d$ ] and @xmath255 } { \mathbb{e}}\bigl [ h \bigl(\hat{\mathbf{x}}_{[p_d d^\delta]}^d\bigr ) - h ( \hat{\mathbf{x}}_0^d )
| \hat{\mathbf{x}}_0^d = \mathbf{x}^d\bigr].\ ] ] in appendix [ secpd ] , we show that for all @xmath256 , @xmath257 as @xmath7 .
consequently , it is useful to introduce @xmath258 @xmath259 which is defined for fixed @xmath260 as @xmath261 } { \mathbb{e}}\bigl [ \bigl(h\bigl(\hat{\mathbf{x}}^d_{[\pi d^\delta]}\bigr ) - h ( \hat{\mathbf{x}}^d_0)\bigr ) |
\hat{\mathbf{x}}^d_0 = \mathbf{x}^d \bigr ] \nonumber\\ & = & \frac{d^2}{[d^\delta ] } \sum_{j=0}^{[\pi d^\delta-1 ] } { \mathbb{e } } [ h ( { \hat{\mathbf{x}}}_{j+1}^d ) - h ( { \hat{\mathbf{x}}}_j^d ) | { \hat{\mathbf{x}}}_0^d = \mathbf{x}^d ] \\ & = & \frac{1}{[d^\delta ] } \sum_{j=0}^{[\pi d^\delta-1 ] } { \mathbb{e } } [ \hat{g}_d h ( { \hat{\mathbf{x}}}_j^d ) | { \hat{\mathbf{x}}}_0^d = \mathbf{x}^d ] , \nonumber\end{aligned}\ ] ] where @xmath262.\ ] ] finally in appendix [ secgen ] , we prove in lemma [ lem45 ] that @xmath263 the triangle inequality is then utilized to prove ( [ eqn115 ] ) in lemma [ lem45 ] using ( [ eqn117 ] ) and @xmath264 as @xmath7 .
it should be noted that in appendix [ secgen ] , we assume that @xmath265 > 0 $ ] , in particular in lemma [ lem40 ] . in appendixes [ secsets ] and [ secpd ]
we make no such assumption .
however , @xmath265=0 $ ] corresponds to @xmath266 ( uniform distribution ) , and proving lemma [ lem45 ] in this case follows similar but simpler arguments to those given in appendix [ secgen ] . a key difference between the diffusion limits for continuous and discontinuous i.i.d .
product densities is the dependence of the speed measure @xmath39 upon @xmath40 . for continuous ( suitably differentiable ) @xmath40
, @xmath39 depends upon @xmath38 $ ] , which is a measure of the `` roughness '' of @xmath40 . for discontinuous densities of the form ( [ eq1b ] ) , @xmath267 depends upon @xmath268 , the ( mean of the ) limit of the density at the boundaries ( discontinuities ) .
discussion of the role of the density @xmath40 in the behavior of the rwm algorithm is given in section [ secextsim ] .
the most important consequence of theorem [ main ] is the following result .
[ mc1 ] let @xmath269 .
then @xmath270 \rightarrow a(l ) \qquad\mbox{as } { d \rightarrow\infty}.\ ] ] @xmath39 is maximized by @xmath271 with @xmath272 clearly , if @xmath273 is known , @xmath274 can be calculated explicitly . however , where mcmc is used , @xmath273 will often only be known up to the constant of proportionality .
this is where corollary [ mc1 ] has major practical implications , in that , to maximize the speed of the limiting diffusion , and hence , the efficiency of the rwm algorithm , it is sufficient to monitor the average acceptance rate , and to choose @xmath47 such that the average acceptance rate is approximately @xmath3 . therefore there is no need to explicitly calculate or estimate the constant of proportionality .
in this section , we discuss the extent to which the conclusions of theorem [ main ] extend beyond @xmath60 being an i.i.d . product density upon the @xmath0-dimensional hypercube and @xmath275 .
first we present two extensions of theorem [ main ] .
the second extension , theorem [ thmprop ] , is an important practical result concerning lower - dimensional updating schema .
suppose that @xmath273 is nonzero on the positive half - line .
that is , @xmath276 and @xmath277 otherwise .
[ thmhalf ] fix @xmath172 . for all @xmath126 , let @xmath173 , given by ( [ eqtri1 ] ) , with latexmath:[$\sup_{x \geq0 } @xmath7 , @xmath174 in the skorokhod topology on @xmath175 , where @xmath176 satisfies the ( reflected ) langevin sde on @xmath279 @xmath280 with @xmath178 , @xmath281 and @xmath282 .
the proof of the theorem is virtually identical to the proof of theorem [ main ] , and so , the details are omitted .
note that we have assumed that @xmath283 is bounded on @xmath284 .
this assumption is almost certainly stronger than necessary with @xmath283 being lipschitz and/or satisfying certain moment conditions probably being sufficient ; cf . @xcite .
theorem [ thmhalf ] is unsurprising with the speed of the diffusion depending upon the number of components close to the discontinuity at 0 .
[ mc2 ] let @xmath285 where @xmath40 satisfies ( [ eqtri1 ] ) .
then @xmath286 \rightarrow\exp ( -f^\star l/4 ) \equiv a(l ) \qquad\mbox{as } { d \rightarrow\infty}.\ ] ] @xmath39 is maximized by @xmath287 with @xmath272 therefore the conclusions are identical to corollary [ mc1 ] that in order to maximize the speed of the limiting diffusion it is sufficient to choose @xmath47 such that the average acceptance rate is @xmath3 . the second and more important extension of theorem [ main ] follows on from @xcite . in @xcite ,
the metropolis - within - gibbs algorithm was considered , where only a proportion @xmath288 @xmath289 of the components are updated at each iteration . for given @xmath126 , at each iteration @xmath290 of the components are chosen uniformly at random and new values for these components are proposed using random walk metropolis with proposal variance @xmath291 .
the remaining @xmath292 components remain fixed at their current values .
finally , it is assumed that @xmath293 as @xmath7 .
the following result assumes that @xmath273 is nonzero on @xmath211 only .
the extension to the positive half - line is trivial .
[ thmprop ] fix @xmath294 and @xmath172 . for all @xmath126 , let @xmath295 be such that all of its components are distributed according to @xmath296 .
then , as @xmath7 , @xmath174 in the skorokhod topology , where @xmath297 and @xmath125 satisfies the ( reflected ) langevin sde on @xmath298 $ ] @xmath299 where @xmath179 is standard brownian motion , @xmath300 and @xmath301 .
let @xmath302 denote the average acceptance rate of the rwm algorithm in @xmath0 dimensions where a proportion @xmath303 of the components are updated at each iteration .
let @xmath304 we then have the following result which mirrors corollaries
[ mc1 ] and [ mc2 ] . [ mc3 ] let @xmath293 as @xmath7 .
then @xmath305 for fixed @xmath294 , @xmath306 is maximized by @xmath307 and @xmath308 also @xmath309 corollary [ mc3 ] is of fundamental importance from a practical point of view , in that it shows that the optimal speed of the limiting diffusion is inversely proportional to @xmath288 .
therefore the optimal action is to choose @xmath288 as close to 0 as possible .
furthermore , we have shown that not only is full - dimensional rwm bad for discontinuous target densities but it is the worst algorithm of all the metropolis - within - gibbs rwm algorithms .
we now go beyond i.i.d .
product densities with a discontinuity at the boundary and @xmath55 .
we consider general densities on the unit hypercube , discontinuities not at the boundary and @xmath53 . as mentioned in section [ secint ] , for i.i.d .
product densities , the speed measure of the limiting one - dimensional diffusion , @xmath39 , is equal to the limit , as @xmath7 , of the esjd times @xmath0 .
therefore we consider the esjd for the above - mentioned extensions as being indicative of the behavior of the limiting langevin diffusion .
we also highlight an extra criterion which is likely to be required in moving from an esjd to a langevin diffusion limit . using the proof of theorem [ main ] , it is straightforward to show that @xmath310 \lim_{{d \rightarrow\infty } } { \mathbb{e}}\bigl [ 1 _ { \ { \mathbf{x}_0^d + \sigma_d \mathbf{z}_1^d \in[0,1]^d \ } } \bigr ] \\ & = & \frac{l^2}{3 } \lim_{{d \rightarrow\infty } } { \mathbb{e}}\biggl [ \biggl ( \frac{3}{4 } \biggr)^{b_d^l ( \mathbf{x}_0^d ) } \biggr].\nonumber\end{aligned}\ ] ] the first equality in ( [ eqextd1 ] ) can be proved using lemma [ lem33 ] , ( [ eq33b ] ) , where for @xmath311 , @xmath312 = 1/3 $ ] .
the second equality in ( [ eqextd1 ] ) comes from the fact that for @xmath313 , @xmath314 and for a component @xmath315 uniformly distributed on @xmath316 or @xmath317 , @xmath318 ) = 3/4 $ ] .
that is , the acceptance probability of a proposed move is dominated by whether or not the proposed move lies inside the @xmath0-dimensional unit hypercube . proposed moves inside the hypercube
are accepted with probability @xmath319 for any @xmath320 ; see lemma [ lem35 ] .
thus it is the number and behavior of the components at the boundary of the hypercube ( the discontinuity ) which determine the behavior of the rwm algorithm .
this is also seen in theorems [ thmhalf ] and [ thmprop ] . first , we consider discontinuities not at the boundary .
suppose that@xmath285 , where @xmath321 \ } } \exp(g ( x ) ) \qquad(x \in\mathbb{r})\ ] ] for some @xmath322 .
further suppose that @xmath323 is continuous ( twice differentiable ) upon @xmath324 $ ] except at a countable number of points , @xmath325 , say , on @xmath326 .
set @xmath327 and @xmath328 , with @xmath74 . for @xmath329 ,
let @xmath330 and @xmath331 , with @xmath332 and @xmath333 , where @xmath334 . then following @xcite , ( 4.23 )
, we can show that @xmath0 times the esjd @xmath335 \rightarrow\frac{l^2}{3 } { \mathbb{e}}\biggl [ 1 \wedge \prod_{j=0}^{k+1 } \biggl ( \frac{f_j^-}{f_j^+ } \biggr)^{y_j^+ - y_j^- } \biggr ] \qquad\mbox{as } { d \rightarrow\infty}.\hspace*{-25pt}\ ] ] thus the optimal scaling of @xmath20 is again of the form @xmath336 and the acceptance or rejection of a proposed move is determined by the components close to the discontinuities .
furthermore , it is straightforward to show that for each @xmath337 , @xmath338 as @xmath339 , implying that the optimal choice of @xmath47 lies in @xmath340 .
proving a langevin diffusion for the ( normalized ) first component of the rwm algorithm should be possible with appropriate local time terms at the discontinuities in @xmath40 . while ( [ eqem1 ] ) holds regardless of @xmath341 and @xmath342 for a diffusion limit we require that @xmath343 , that is
, the density is strictly positive on @xmath326 .
( if this is not the case , the rwm algorithm is reducible in the limit as @xmath7 . )
extensions to the case where either @xmath344 and/or @xmath345 are straightforward .
second , we consider general densities which are zero outside the @xmath0-dimensional hypercube , @xmath346^d \ } } \exp(\mu_d ( \mathbf{x}^d))$ ] , where @xmath347 is assumed to be continuous and twice differentiable .
let @xmath348 and assuming that @xmath349 we have that @xmath0 times the esjd satisfies @xmath350 \rightarrow\frac{l^2}{3 } \lim_{{d \rightarrow\infty } } { \mathbb{e}}\biggl [ \biggl ( \frac{3}{4 } \biggr)^{b_d^l ( \mathbf{x}_0^d ) } \biggr ] \qquad\mbox{as } { d \rightarrow\infty}.\ ] ] note that ( [ eqextd2 ] ) is a weak condition and should be straightforward to check using a taylor series expansion of @xmath351 . for i.i.d .
product densities , @xmath352 as @xmath7 .
more generally , the limiting distribution of @xmath353 will determine the limit of the right - hand side of ( [ eqextd3 ] ) .
in particular , so long as there exist @xmath119 and @xmath354 such that @xmath355 , the right - hand side of ( [ eqextd3 ] ) will be nonzero for @xmath172 .
it is informative to consider what conditions upon @xmath60 are likely to be necessary for a diffusion limit , whether it be one - dimensional or infinite - dimensional as in @xcite .
suppose that as @xmath7 .
for a diffusion limit we will require moment conditions on @xmath356 , probably requiring that there exists @xmath357 such that @xmath358 < \infty$ ] .
this will be required to control the probability of the rwm algorithm getting `` stuck '' in the corners of the hypercube .
this highlights a key difference between studying the esjd and a diffusion limit . for the esjd
, we want a positive probability that the total number of components at the boundary of the hypercube is finite in the limit as @xmath7 . for the diffusion limit ,
as seen with the construction of @xmath359 in theorem [ main ] , we want that the probability of there being a large number of components @xmath360 at the boundary is very small @xmath361 .
third , suppose that @xmath53 .
there are very bad starting points in the `` corners '' of the hypercube .
for example , if @xmath362 , @xmath363 which even for @xmath59 is less than @xmath364 .
thus the rwm process is likely to be `` stuck '' at its starting point for a very long period of time .
this is rather pathological and a more interesting question is the situation when @xmath365 , where the components of @xmath366 are i.i.d . in particular , suppose that @xmath367 $ ] , so that @xmath11 is chosen uniformly at random over the hypercube .
note that , if @xmath366 is the uniform distribution , @xmath368 with the right - hand side maximized by taking @xmath369 compared with @xmath370 for @xmath275 .
we expect to see similar behavior to @xcite , in that the optimal @xmath20 ( in terms of the esjd ) will vary as the algorithm converges to the stationary distribution but will be of the form @xmath74 throughout .
the rwm algorithm is unlikely to get `` stuck '' with it conjectured that for any @xmath117 and @xmath371 , @xmath372 } \
{ b_d^l ( \mathbf{x}_t^d ) \geq \gamma\log d \ } \qquad\mbox{as } { d \rightarrow\infty}.\ ] ] simulations with @xmath373 and @xmath374 and @xmath375 suggest that convergence occurs in @xmath90 . for convergence ,
we monitor the mean of @xmath26 for @xmath376 and the variance of @xmath26 for @xmath377 .
the sets @xmath378 consist of the intersection of four sets @xmath379 @xmath380 . for @xmath381
, we will define @xmath379 and discuss the role that it plays in the proof of theorem [ main ] , one at a time .
furthermore , we show that in stationarity it is highly unlikely that @xmath26 does not belong to @xmath378 .
since we rely upon a homogenization argument , it is necessary to go further than the sets @xmath378 to the sets @xmath382 . in particular , if @xmath383 , then it is highly unlikely that any of @xmath384}^d$ ] do not belong to @xmath378 .
the above statement is made precise in theorem [ lem321 ] below , where the constructions of @xmath228 and @xmath116 are drawn together .
it is possible that all @xmath0 components of @xmath11 are in @xmath238 .
however , this is highly unlikely and we show in lemma [ lem31 ] that with high probability , there are at most @xmath385 components in the rejection region .
let @xmath386 .
[ lem31 ] for any @xmath387 , @xmath388 fix @xmath389 .
note that @xmath390 if and only if @xmath391 .
however , @xmath392 with @xmath393 fix @xmath394 . by markov s inequality and using independence of the components of @xmath11 , @xmath395 & & \qquad \leq d^\kappa{\mathbb{e } } [ \exp(\rho b_d^l ( \mathbf{x}_0^d))]/ \exp ( \rho\gamma\log d ) \nonumber\\[-1pt ] & & \qquad= d^\kappa{\mathbb{e}}\bigl [ \exp\bigl(\rho1_{\ { x_{0,1}^d \in r_d^l \}}\bigr)\bigr]^d / d^{\rho\gamma } \\[-2pt ] & & \qquad = d^\kappa\biggl ( 1 + ( e^\rho-1 ) \int_0^{l / d } \ { f(x ) + f(1-x)\ } \,dx \biggr)^d \big/ d^{\rho\gamma } \nonumber\\[-2pt ] & & \qquad \leq
d^{\kappa- \rho\gamma } \exp\biggl ( ( e^\rho-1)d \int_0^{l / d } \ { f(x ) + f ( 1-x)\ } \,dx \biggr).\nonumber\end{aligned}\ ] ] the lemma follows since ( [ eq31a ] ) implies that the right - hand side of ( [ eq31b ] ) converges to 0 as @xmath7 . for @xmath244
, it follows from ( [ eq21ay ] ) that @xmath396 this is a useful lower bound for the acceptance probability and as a result the random walk metropolis algorithm does not get `` stuck '' at values of @xmath397 . to assist with the homogenizing arguments , we define @xmath398 by @xmath399 } \hat{\mathbf{x}}_j^d \notin f_d^1 | \hat{\mathbf{x}}_0^d = \mathbf{x}^d \biggr ) \leq d^{-3 } \biggr\}.\ ] ]
that is , by starting in @xmath400 it is highly unlikely that the pseudo - rwm algorithm leaves @xmath232 in @xmath113 $ ] iterations . to study @xmath400 and
later @xmath122 we require the following lemmas .
[ lem31a ] for a random variable @xmath401 , suppose that there exist @xmath402 such that @xmath403 and for all @xmath404 , @xmath405
. then @xmath406 first note that @xmath407\\[-8pt ] & = & { \mathbb{p}}(x \in a | x \in d^c , x \in b ) { \mathbb{p}}(x \notin d ( [ eq31e ] ) and using ( [ eq31c ] ) and @xmath408 . [ lem34 ] suppose that a sequence of sets @xmath409 is such that there exists @xmath387 such that @xmath410 fix @xmath411 and let @xmath412 } \ { \hat{\mathbf{x}}_i^d \notin f_d^\star\cap f_d^1 \ } | \hat{\mathbf{x}}_0^d = \mathbf{x}^d \biggr ) \leq d^{-\varepsilon } \biggr\}.\ ] ] then @xmath413 since @xmath414 , @xmath415 } \ { \mathbf{x}_i^d \notin f_d^\star\cap f_d^1 \ } \biggr ) \leq d^{2 + \delta+ \gamma } { \mathbb{p}}(\mathbf{x}_0^d \notin f_d^\ast\cap f_d^1).\ ] ] therefore for all sufficiently large @xmath0 , @xmath416 by bayes s theorem , @xmath417 . therefore taking @xmath418 } \
{ \mathbf{x}_i^d \notin f_d^\ast\cap f_d^1 \}$ ] and @xmath419 , it follows from ( [ eq34e ] ) and ( [ eq34ea ] ) that @xmath420 } \!\
{ \mathbf{x}_i^d \notin f_d^\star\cap f_d^1\ } | \mathbf{x}_0^d \in f_d^\star\cap f_d^1 \!\biggr ) \leq\frac{d^{2 + \delta+ \gamma } { \mathbb{p}}(\mathbf{x}_0^d
\notin f_d^\star\cap f_d^1 ) } { 1/2}.\hspace*{-40pt}\ ] ] let @xmath421 } \
{ \mathbf{x}_i^d \notin f_d^\star\cap f_d^1 \ } | \mathbf{x}_0^d = \mathbf{x}^d \biggr ) \leq d^{-\varepsilon } \biggr\}.\ ] ] it follows from lemmas [ lem31 ] and [ lem31a ] that @xmath422 since @xmath423 as @xmath7 , it follows from ( [ eq34 g ] ) that @xmath424 for @xmath126 and @xmath425 let @xmath426 be independent and identically distributed bernoulli random variables with @xmath427 where @xmath428 .
it is straightforward using hoeffding s inequality to show that @xmath429 } \theta_i^d < d^\delta\biggr ) \rightarrow0 \qquad\mbox{as } { d \rightarrow\infty}.\ ] ] now @xmath430 and @xmath431 can be constructed upon a common probability space such that if @xmath432 and @xmath433 , @xmath434 . for @xmath435 , consider @xmath436 , if @xmath437 and @xmath438 , a coupling exists such that there exists @xmath439 such that @xmath440 . exploiting the above coupling , @xmath441 } \
{ \mathbf{x}_j^d \in f_d^\star\cap f_d^1 \}$ ] and @xmath442 } \theta_i^d \geq d^\delta$ ] together imply that @xmath443 } \ { \hat{\mathbf{x}}_j^d \in f_d^\star\cap f_d^1\}$ ] .
thus @xmath444 } \theta_i^d < d^\delta\biggr),\ ] ] and ( [ eq34c ] ) follows from ( [ eq34h ] ) , ( [ eq34j ] ) and ( [ eq34k ] ) .
as noted in section [ secalg ] , we follow @xcite by considering the behavior of the random walk metropolis algorithm over steps of size @xmath113 $ ] iterations .
we find that a single component moves only a small distance in @xmath113 $ ] iterations , while over @xmath113 $ ] iterations the acceptance probability , which is dominated by the number and position of components in @xmath238 , `` forgets '' its starting value .
moreover , we show that approximately @xmath445 $ ] of the proposed moves are accepted . however , we need to control the number of components which are _ close to _ the rejection region ( @xmath233 ) and the distribution of the position of the components in the rejection region after @xmath189 $ ] iterations ( @xmath234 ) , where @xmath446 . for any @xmath447 ,
let @xmath448 | \leq\sqrt{k } \bigr\}\ ] ] and let @xmath449}^{[d^\delta ] } \hat{f}_d^2 ( k).\ ] ] before studying @xmath233 , we state a simple , useful result concerning the central moments of a sequence of binomial random variables . [ lemz32 ] let @xmath450 .
suppose that @xmath451 and @xmath452 as @xmath7 ; then for any @xmath453 , @xmath454)^{2m}\bigr]/(dp_d)^m \rightarrow\prod_{j=1}^m ( 2j-1 ) \qquad\mbox{as } { d \rightarrow\infty}.\ ] ] [ lem32 ] for any @xmath387 and sequence of positive integers @xmath188 satisfying @xmath189 \leq k_d \leq [ d^\delta]$ ] , @xmath455 consequently , for any @xmath387 , @xmath456 as @xmath7 . fix @xmath387 . by stationarity and markov s inequality , for all , @xmath457 | \geq\sqrt{k_d}\bigr )
\nonumber\\[-8pt]\\[-8pt ] & \leq & \frac{d^\kappa}{k_d^m } { \mathbb{e}}\bigl [ \bigl ( b_d^{k_d^{3/4 } } ( \mathbf{x}^d_0 ) - { \mathbb{e } } [ b_d^{k_d^{3/4 } } ( \mathbf{x}^d_0)]\bigr)^{2 m } \bigr].\nonumber\end{aligned}\ ] ] however , @xmath458 , so by lemma [ lemz32 ] for any @xmath453 , for all sufficiently large @xmath0 , @xmath459\bigr)^{2 m } \bigr ]
\leq k_m k_d^{3m/4},\ ] ] where @xmath460 . since @xmath461 $ ] , the right - hand side of ( [ eq32a ] ) converges to 0 as @xmath7 by taking @xmath462 , proving ( [ eq32c ] ) .
note that @xmath463}^{[d^\delta ] } { \mathbb{p}}\bigl(\mathbf{x}_0^d \notin \hat{f}_d^2 ( k)\bigr).\ ] ] the right - hand side of ( [ eq32b ] ) converges to 0 as @xmath7 since ( [ eq32c ] ) holds with @xmath464 replaced by @xmath465 . before considering @xmath234 , the distribution of the position of the components in the rejection region after @xmath189 $
] iterations , we introduce a simple random walk on the hypercube ( rwh ) . the biggest problem in analyzing the rwm or
pseudo - rwm algorithm is the dependence between the components .
however , the dependence is weak and whether or not a proposed move is accepted is dominated by whether or not the proposed moves lies inside or outside the hypercube .
therefore we couple the rwm algorithm to the simpler rwh algorithm . for @xmath126 ,
define the rwh algorithm as follows .
let @xmath466 denote the position of the rwh algorithm after @xmath467 iterations .
then @xmath468^d$ , \cr \mathbf{w}_k^d , & \quad otherwise.}\ ] ] that is , the rwh algorithm simply accepts all proposed moves which remain inside the hypercube and rejects all proposed moves outside the hypercube .
define the pseudo - rwh algorithm in the obvious fashion with @xmath469 denoting the position of the pseudo - rwh algorithm at iteration @xmath467 .
then for @xmath470 , @xmath471 , where @xmath472 $ ] .
for our purposes it will suffice to consider the coupling of the pseudo - rwm and pseudo - rwh algorithms over @xmath113 $ ] iterations and study how the pseudo - rwh algorithm evolves over @xmath113 $ ] iterations .
note that the rwh algorithm coincides with the rwm algorithm with a uniform target density over the @xmath0-dimensional cube , so in this case the coupling is exact .
the components of the pseudo - rwh algorithm behave independently . for @xmath473 ,
let @xmath474 and for @xmath475 , let @xmath476 .
then @xmath477 is the probability that a proposed move from @xmath87 is accepted in the rwh algorithm .
[ lem33 ] for any @xmath478 and @xmath136^d$ ] , there exists a coupling such that @xmath479 let @xmath480 $ ] ; then we can couple @xmath73 and @xmath481 using @xmath482 and @xmath483 as follows .
let @xmath484^d$ , \cr \mathbf{x}^d , & \quad otherwise , } \\
\mathbf{x}_1^d & = & \cases { \mathbf{x}^d + \sigma_d \mathbf{z}_1^d , & \quad if $ \mathbf{x}^d + \sigma_d \mathbf{z}_1^d \in[0,1]^d$\vspace*{2pt}\cr & \quad and $ \displaystyle u \leq1 \wedge \exp\biggl(\sum_{j=1}^d \{g ( x_j + \sigma_d z_{1,j } ) - g(x_j ) \ } \biggr)$ , \vspace*{2pt}\cr \mathbf{x}^d , & \quad otherwise.}\end{aligned}\ ] ] therefore , @xmath485 if @xmath486^d$ ] and @xmath487 . thus @xmath488^d,\nonumber\\ & & \hspace*{59pt }
u > 1 \wedge\exp\biggl(\sum_{j=1}^d \{g ( x_j + \sigma_d z_{1,j } ) - g(x_j ) \ }
\biggr ) \biggr ) \nonumber\\[-8pt]\\[-8pt ] & & \qquad= d^\alpha{\mathbb{e}}\biggl [ \prod_{j=1}^d 1 _ { \ { 0
< x_j + \sigma_d z_{1,j } < 1 \ } } \nonumber\\ & & \hspace*{73pt}{}\times\biggl\{1 - 1 \wedge\exp\biggl ( \sum_{j=1}^d \{g ( x_j + \sigma_d z_{1,j } ) - g(x_j ) \ } \biggr ) \biggr\ } \biggr ] \nonumber\\ & & \qquad\leq d^\alpha{\mathbb{e}}\biggl [ \biggl| \sum_{j=1}^d \{g ( x_j + \sigma_d z_{1,j } ) - g(x_j ) \ } \biggr| \biggr],\nonumber\end{aligned}\ ] ] since for all @xmath489 , @xmath490 . by taylor s theorem , for @xmath491
, there exists @xmath492 lying between 0 and @xmath493 such that @xmath494 since @xmath323 is continuously twice differentiable on @xmath211 , there exists @xmath495 such that @xmath496 since the components of @xmath482 are independent , by jensen s inequality , ( [ eq33d ] ) and @xmath497 \leq2 { \mathbb{e}}[x^2 ] + 2 c^2 $ ] , for any random variable @xmath401 and constant @xmath288 , we have that @xmath498 and the lemma is proved .
[ lem35 ] fix @xmath499 . for any @xmath136^d$
] , there exists a coupling such that @xmath500 } \{\mathbf{x}_j^d \neq\mathbf{w}_j^d \}| \mathbf{x}_0^d \equiv\mathbf{w}_0^d = \mathbf{x}^d \biggr ) \rightarrow0 \qquad\mbox{as } { d \rightarrow\infty}.\ ] ] moreover , if @xmath501 and @xmath502 , there exists a coupling such that @xmath503 } \{\hat{\mathbf{x}}_j^d \neq \hat{\mathbf{w}}_j^d \ }
| \mathbf{x}_0^d \equiv\mathbf{w}_0^d = \mathbf{x}^d
\biggr ) \rightarrow0 \qquad\mbox{as } { d \rightarrow\infty}.\ ] ] for @xmath504 and @xmath505 let @xmath506 let @xmath507 $ ] and let @xmath508 .
note that the movement of the components of the pseudo - rwh algorithm are independent .
the next stage in the proof is to show that , if @xmath509 is started in @xmath234 , then after @xmath247 iterations of the pseudo - rwm algorithm has forgotten its starting value in terms of the total number and position of the components in @xmath238 ( the rejection region ) .
moreover , the total number and position of the components in @xmath238 after @xmath247 iterations of the pseudo - rwm algorithm are approximately from the stationary distribution of @xmath510 . before defining and studying @xmath511 , we require the following lemma and associated corollary concerning the distribution of the components in the rejection region after @xmath247 steps .
[ lem38a ] let @xmath512 be any sequence of positive integers satisfying @xmath189 \leq k_d \leq[d^\delta]$ ] . for any sequence of @xmath513 such that @xmath514 , @xmath515 also for all @xmath516 , @xmath517 fix @xmath518 and set @xmath519 . to prove ( [ eq38aa ] ) and ( [ eq38ab ] ) we couple the components of @xmath520 to a simple reflected random walk process @xmath521 .
set @xmath522 for some @xmath523 .
let @xmath524 be i.i.d . according to @xmath525 $ ] . for @xmath526 ,
set @xmath527 with reflection at the boundaries 0 and 1 so that @xmath528 .
for @xmath473 , let @xmath529 .
consider @xmath530 with identical arguments applying for the other components of @xmath520 . since @xmath531 as @xmath7 , we assume that @xmath0 is such that @xmath532
. then @xmath533 for @xmath534 we can couple @xmath535 and @xmath530 such that @xmath536 for @xmath537 , if @xmath538 , then set @xmath539 .
now if @xmath540 @xmath541 , @xmath542 and @xmath543 can be coupled such that , if @xmath538 , then @xmath544 @xmath545 .
furthermore , for @xmath546 @xmath547 , the above coupling can be extended to give , if @xmath548 and @xmath549 , then @xmath550 @xmath551 . since in @xmath247 iterations either process
can move at most a distance @xmath552 , ( [ eq38ad ] ) follows from the above coupling . without loss of generality , we assume that @xmath553 [ symmetry arguments apply for @xmath554 . by the reflection principle , @xmath555\\[-8pt ] &
= & { \mathbb{p}}\biggl ( - 1 < \frac{x}{\sigma_d } + \sum_{i=1}^{k_d } \tilde{z}_i < 1 \biggr).\nonumber\end{aligned}\ ] ] by the berry essen theorem , there exists a positive constant , @xmath556 say , such that for all @xmath557 , @xmath558 where @xmath559 denotes the c.d.f .
of a standard normal .
therefore it follows from ( [ eq38ae ] ) and ( [ eq38af ] ) that there exists a positive constant , @xmath560 say , such that for all @xmath561 , @xmath562 by hoeffding s inequality , for any @xmath563 , @xmath564\\[-8pt ] & = & 2 \exp\bigl(- \varepsilon^2 \sqrt{k_d}/2\bigr).\nonumber\end{aligned}\ ] ] hence for @xmath565 , by taking @xmath566 in ( [ eq38ah ] ) , we have that @xmath567 furthermore , note that for @xmath568 , @xmath569 .
then ( [ eq38ab ] ) follows immediately from ( [ eq38ad ] ) and the above bounds for @xmath570 since @xmath571 as @xmath7 .
finally , for @xmath518 , it follows from ( [ eq38ad ] ) , ( [ eq38af ] ) and ( [ eq38ag ] ) that there exists @xmath572 such that @xmath573\\[-8pt ] & \leq & d^{2 \gamma } \biggl\ { k_3 k_d^{3/4 } \biggl ( \frac{k_2}{\sqrt{k_d } } \biggr)^2 + 2 d \exp\biggl(- \frac{\sqrt{k_d}}{8 l^2 } \biggr ) \biggr\}\nonumber\end{aligned}\ ] ] with the right - hand side of ( [ eq38j ] ) converging to 0 as @xmath7 . [ lem37 ] for any @xmath574 , any sequence @xmath575 satisfying and any sequence of positive integers @xmath576 satisfying @xmath189
\leq k_d \leq[d^\delta]$ ] , there exists @xmath495 , such that for all @xmath126 , @xmath577 \leq k d^{- ( 1 + \beta m/8)}.\ ] ] fix @xmath574 .
note that @xmath578 & \leq & { \mathbb{e } } [ q^d ( x_{0,1}^d;l ; k_d)^m ] \nonumber\\ & = & \int_0 ^ 1 q^d ( x ; l ; k_d)^m f ( x ) \,dx \nonumber\\[-8pt]\\[-8pt ] & = & \int_{r_d^{k_d^{3/4 } } } q^d ( x ; l ; k_d)^m f ( x ) \,dx\nonumber\\ & & { } + \int_{(r_d^{k_d^{3/4 } } ) ^c } q^d ( x ; l ; k_d)^m f ( x ) \,dx.\nonumber\end{aligned}\ ] ] the two terms on the right - hand side of ( [ eq37aa ] ) are bounded using ( [ eq38ag ] ) and ( [ eq38ai ] ) , respectively .
thus it follows from the proof of lemma [ lem38a ] that there exist constants @xmath579 such that , for all @xmath126 , @xmath580 & \leq & \int_{r_d^{k_d^{3/4 } } } \biggl ( \frac{k_1}{\sqrt{k_d } } \biggr)^m f ( x ) \,dx\nonumber\\ & & { } + \int_{(r_d^{k_d^{3/4 } } ) ^c } \biggl\ { 2 \exp\biggl ( - \frac{\sqrt{k_d}}{8l^2 } \biggr ) \biggr\}^m f ( x ) \,dx \nonumber\\ & \leq & { \mathbb{p } } ( x_{0,1}^d \in r_d^{k_d^{3/4 } } ) \biggl ( \frac{k_1}{\sqrt{k_d } } \biggr)^m\\ & & { } + { \mathbb{p } } ( x_{0,1}^d \notin r_d^{k_d^{3/4 } } ) \times2 \exp\biggl ( - \frac{\sqrt{k_d}}{8l^2 } \biggr ) \nonumber\\ & \leq & k_2 \frac{k_d^{3/4}}{d } k_d^{-m/2 } + 2 \exp\biggl ( - \frac{\sqrt{k_d}}{8l^2 } \biggr).\nonumber\end{aligned}\ ] ] the corollary follows from ( [ eq37b ] ) since @xmath574 and @xmath581 $ ] .
we are now in position to define @xmath582 . for any @xmath583 and @xmath584 , let @xmath585 where @xmath586
let @xmath587 \leq k_d \leq[d^\delta ] } \sup_{0 \leq r \leq l } | \lambda_d ( \mathbf{x}^d ; r ; k_d ) - \lambda ( r ) |
< d^{-\gamma } \bigr\}.\ ] ] we study @xmath588 as a prelude to analyzing @xmath582 where @xmath589 and @xmath247 are defined in lemma [ lem36 ] below . [ lem36 ] for any sequence @xmath590 satisfying @xmath591 , any sequence of positive integers @xmath188 satisfying @xmath189 \leq k_d \leq[d^\delta]$ ] and @xmath387 , @xmath592 by the triangle inequality ,
@xmath593 | > d^{-\gamma}/16\bigr ) \\ & & \qquad\quad { } + d^\kappa{\mathbb{p}}\bigl ( | { \mathbb{e } } [ \lambda_d ( \mathbf{x}_0^d ; r_d ; k_d ) ] - \lambda(r_d)| > d^{-\gamma}/16\bigr).\nonumber\end{aligned}\ ] ] in turn we show that the two terms on the right - hand side of ( [ eq36b ] ) converge to 0 as @xmath7 . by markov s inequality
, we have that for any @xmath453 , @xmath594
| > d^{-\gamma}/16\bigr ) \nonumber\\ & & \qquad\leq16^{m } d^{\kappa+m \gamma } { \mathbb{e}}\biggl [ \biggl ( \sum_{j = 1}^d \ { q^d ( x_{0,j};r_d ; k_d ) - { \mathbb{e } } [ q^d ( x_{0,j};r_d ; k_d ) ] \ } \biggr)^m \biggr ] \nonumber\\[-8pt]\\[-8pt ] & & \qquad= 16^m d^{\kappa+m \gamma } \sum_{i_1 = 1}^d \cdots\sum_{i_m = 1}^d { \mathbb{e}}\biggl [ \prod_{j=1}^m \ { q^d ( x_{0,i_j};r_d ; k_d)\nonumber\\ & & \hspace*{167pt } { } - { \mathbb{e } } [ q^d ( x_{0,i_j};r_d ; k_d ) ] \ } \biggr].\nonumber\end{aligned}\ ] ] since the components of @xmath11 are independent and identically distributed , we have for any @xmath595 , there exists @xmath596 and @xmath597 with @xmath598 such that @xmath599 \ } \biggr]\nonumber\\[-8pt]\\[-8pt ] & & \qquad = \prod_{j=1}^j { \mathbb{e}}\bigl [ \ { q^d ( x_{0,1};r_d ; k_d ) - { \mathbb{e } } [ q^d ( x_{0,1};r_d ; k_d ) ] \}^{l_j } \bigr].\nonumber\end{aligned}\ ] ] note that if any @xmath600 , then the right - hand side of ( [ eq36d ] ) is equal to 0 . by corollary [ lem37 ] ,
if @xmath601 , there exists @xmath556 such that the right - hand side of ( [ eq36d ] ) is less than or equal to @xmath602
. furthermore , there exists @xmath560 such that for any and @xmath603 , there are at most @xmath604 configurations of @xmath605 such that for @xmath606 , @xmath607 of the components are the same .
therefore there exists @xmath495 such that @xmath608 \ } \biggr]\nonumber\\[-8pt]\\[-8pt ] & & \qquad \leq k d^{-m \beta/8}.\nonumber\end{aligned}\ ] ] taking @xmath609 , it follows from ( [ eq36e ] ) that the right - hand side of ( [ eq36c ] ) converges to 0 as @xmath7 .
the lemma follows by showing that for all sufficiently large @xmath0 , @xmath610 - \lambda(r_d)| \leq d^{-\gamma}/16.\ ] ] note that @xmath611 & = & d { \mathbb{e } } [ q^d ( x_{0,1};r;k_d ) ] \nonumber\\ & = & d \int_0^{k_d^{3/4}/d } q^d ( x;r_d ; k_d ) f ( x ) \,dx\nonumber\\[-8pt]\\[-8pt ] & & { } + d \int_{k_d^{3/4}/d}^{1-k_d^{3/4}/d } q^d ( x;r_d ; k_d ) f ( x ) \,dx\nonumber\\ & & { } + d \int_{1-k_d^{3/4}/d}^1 q^d ( x;r_d ; k_d ) f ( x ) \,dx.\nonumber\end{aligned}\ ] ] by ( [ eq38ai ] ) , the second integral on the right - hand side of ( [ eq36 g ] ) is bounded above by @xmath612 as @xmath7 .
let @xmath613 . then by taylor s theorem , for @xmath614 , @xmath615 thus @xmath616\\[-8pt ] & & \qquad\leq d \times f_\star\frac{k_d^{3/4}}{d } \times \int_0^{k_d^{3/4}/d } q^d ( x;r_d ; k_d ) \,dx.\nonumber\end{aligned}\ ] ] similarly , we have that @xmath617\\[-8pt ] & & \qquad\leq d \times f_\star\frac{k_d^{3/4}}{d } \times \int_{1-k_d^{3/4}/d}^1 q^d ( x;r_d ; k_d ) \,dx.\nonumber\end{aligned}\ ] ] by symmetry , @xmath618 , so @xmath619 - 2 f^\ast d \int_0 ^ 1 q^d ( x;r_d;k_d ) \,dx
\biggr| \rightarrow0 \qquad\mbox{as } { d \rightarrow\infty}.\hspace*{-35pt}\ ] ] since @xmath620 , using lemma [ lem38a ] , ( [ eq38ab ] ) , we have that , for all sufficiently large @xmath0 , @xmath621\\[-8pt ] & & \qquad\quad { } + d \int_{\sigma_d}^{1-\sigma_d } \biggl\ { \frac{1 } { \int_0 ^ 1 \omega_d ( y ) \,dy } - 1 \biggr\ } q^d ( x;r_d;k_d ) \,dx \nonumber\\ & & \qquad\leq4 d^{1 + \gamma } \sigma_d d^{-2 \gamma } + d^{1 + \gamma } \int_0 ^ 1 \frac{2 \sigma_d}{\int_0 ^ 1 \omega_d ( y ) \,dy } q^d ( x;r_d;k_d ) \,dx.\nonumber\end{aligned}\ ] ] let @xmath622 be defined as in lemma [ lem38a ] . note that @xmath623 $ ] is the stationary distribution of a reflected random walk on @xmath211 .
therefore for any @xmath447 , @xmath624 therefore , it follows from lemma [ lem38a ] , ( [ eq38ad ] ) , that @xmath625 hence the right - hand side of ( [ eq36k ] ) converges to 0 as @xmath7 .
note that the stationary distribution of a single component of the pseudo - rwh algorithm has p.d.f .
@xmath626 @xmath627 .
therefore @xmath628\\[-8pt ] & = & \frac{r_d}{2 } \biggl(1 + \frac{r_d}{2l } \biggr ) \bigg/ \biggl(1 - \frac{l}{2d } \biggr).\nonumber\end{aligned}\ ] ] finally , combining ( [ eq36j ] ) , ( [ eq36k ] ) and ( [ eq36n ] ) , we have that ( [ eq36f ] ) holds and the lemma is proved .
[ lema311 ] for any @xmath387 , @xmath629 fix @xmath387 .
fix a sequence of positive integers @xmath576 such that @xmath189 \leq k_d \leq[d^\delta]$ ] . fix @xmath630 and let @xmath631 d^{- \theta},l \}$ ] .
thus the elements of @xmath632 are separated by a distance of at most @xmath633 . for any @xmath634 and @xmath126
, there exist @xmath635 such that @xmath636 with @xmath637 . by the triangle inequality , @xmath638\\[-8pt ] & & \qquad\quad { } + \lambda(\hat{r}_d)- \lambda(\tilde{r}_d )
\nonumber\\ & & \qquad\leq| \lambda_d ( \mathbf{x}_0^d ; \hat{r}_d ; k_d ) - { \mathbb{e } } [ \lambda_d ( \mathbf{x}_0^d ; \hat{r}_d ; k_d ) ] |\nonumber\\ & & \qquad\quad { } + 2 |\lambda_d ( \mathbf{x}_0^d ; \tilde{r}_d ; k_d ) - { \mathbb{e } } [ \lambda_d ( \mathbf{x}_0^d ; \tilde{r}_d ; k_d ) ] | \nonumber\\ & & \qquad\quad { } + 2 |\lambda(\hat{r}_d)- \lambda(\tilde{r}_d)|.\nonumber\end{aligned}\ ] ] by lemma [ lem36 ] , for any sequence @xmath639 satisfying @xmath640 , @xmath641 hence @xmath642 for all sufficiently large @xmath0 , @xmath643 therefore it follows from ( [ eqss4 ] ) , ( [ eqss5 ] ) and ( [ eqss6 ] ) that @xmath644 since ( [ eqss7 ] ) holds for any sequence @xmath188 satisfying @xmath189 \leq k_d \leq[d^\delta]$ ] , the lemma follows since @xmath645 \leq k \leq[d^\delta ] } \sup_{0 \leq
r \leq l } | \lambda_d ( \mathbf{x}_0^d;r ; k ) - \lambda(r ) | > d^{-\gamma }
\bigr ) \nonumber\\[-8pt]\\[-8pt ] & & \qquad\leq d^ { \kappa } \sum_{k=[d^\beta]}^{[d^\delta ] } { \mathbb{p}}\bigl ( \sup_{0 \leq
r \leq l } | \lambda_d ( \mathbf{x}_0^d;r ; k ) - \lambda(r ) | > d^{-\gamma } \bigr).\nonumber\end{aligned}\ ] ] finally , we consider @xmath646 \biggr| < d^{-{1}/{8 } } \biggr\}.\ ] ] the sets @xmath647 mirror the sets @xmath648 in @xcite and are used when considering @xmath649 and @xmath258 but play no role in analyzing @xmath650 .
[ lem320 ] for any @xmath387 , @xmath651 let @xmath652 and fix @xmath387 . then by hoeffding s inequality , @xmath653 \biggr| > d^{7/8 }
\biggr ) \nonumber\\[-8pt]\\[-8pt ] & \leq & d^\kappa\times2 \exp\biggl ( - \frac{2 d^{7/4}}{d ( g^\ast)^4 } \biggr ) \rightarrow0 \qquad\mbox{as } { d \rightarrow\infty}.\nonumber\end{aligned}\ ] ] finally we are in position to consider @xmath228 and @xmath654 .
recall that , for @xmath126 , @xmath655 and @xmath656 } \hat{\mathbf{x}}_j^d \notin f_d | \hat{\mathbf{x}}_0^d = \mathbf{x}^d \biggr ) \leq d^{-3 } \biggr\}.\ ] ] combining lemmas [ lem31 ] , [ lem32 ] , [ lema311 ] and [ lem320 ] , we have the following theorem .
[ lem321 ] for any @xmath387 , @xmath657 hence , by lemma [ lem34 ] , for any @xmath387 , @xmath658 also using the couplings outlined above , we have that @xmath659 } \ { \hat{\mathbf{w}}_j^d \notin f_d \ } |\hat{\mathbf{w}}_0^d \in \tilde{f}_d \biggr ) \rightarrow0 \qquad\mbox{as } { d \rightarrow\infty}.\ ] ]
we show that for any sequence @xmath513 such that @xmath660 , @xmath661 the key result is lemma [ lem311 ] which states that after @xmath247 iterations , the configuration of the components in the rejection region @xmath238 resemble the configuration of the points of a poisson point process with rate @xmath586 on the interval @xmath662 $ ] . for any @xmath663 and @xmath664 , let @xmath665 with @xmath666 let @xmath667 where the components of @xmath668 are independent poisson random variables with @xmath669 and @xmath670 [ lem311 ] for any @xmath663 , any sequence of positive integers @xmath188 satisfying @xmath189 \leq k_d
\leq[d^\delta]$ ] and @xmath671 , @xmath672 fix @xmath663 and @xmath671 . let @xmath673 where for @xmath664 , @xmath674 are independent poisson random variables with means @xmath675 the lemma is proved by showing that @xmath676 by @xcite , theorem 1 , @xmath677 by lemma [ lem38a ] , ( [ eq38aa ] ) the right - hand side of ( [ eq311c ] ) converges to 0 as @xmath7 .
for the second term on the right - hand side of ( [ eq311b ] ) , it suffices to show that @xmath678 ( for discrete random variables convergence in distribution and convergence in total variation distance are equivalent ; see @xcite , page 254 . )
the components of @xmath679 and @xmath668 are independent , and therefore it is sufficient to show that , for all @xmath664 , @xmath680 for all @xmath664 , ( [ eq311d ] ) holds , if @xmath681 therefore the lemma follows from ( [ eq311e ] ) since @xmath189 \leq k_d \leq[d^\delta]$ ] and @xmath682 .
[ see ( [ eqn11f3 ] ) for the construction of @xmath582 . ]
lemma [ lem311 ] is the key result stating that if the pseudo - rwh process is started from the set @xmath378 , then after @xmath189 $ ] iterations the distribution of the components in the rejection region are approximately given by @xmath668 .
we show that studying the pseudo - rwh algorithm over @xmath113 $ ] iterations suffices in analyzing @xmath683 } \sum_{j=0}^{[\pi d^\delta-1 ] } m_j ( j_d ( \hat{\mathbf{x}}_j^d))$ ] .
note that @xmath650 satisfies @xmath684).\ ] ] let @xmath685 } \sum_{j=0}^{[\pi d^\delta -1 ] } m_j ( \omega_d ( \hat{\mathbf{w}}_j^d))$ ] . before establishing a coupling between @xmath686 and @xmath687 ,
we give a simple coupling for geometric random variables .
[ lem310 ] suppose that @xmath688 and that @xmath401 and @xmath689 are independent geometric random variables with success probabilities @xmath156 and @xmath690 , respectively , that is , @xmath691 and @xmath692 .
let @xmath693 be a bernoulli random variable with @xmath694 and @xmath695 .
then if @xmath693 , @xmath401 , @xmath689 and @xmath92 are mutually independent , @xmath696 therefore there exists a coupling of @xmath401 and @xmath689 such that @xmath697 [ lem312 ] for any @xmath698 and @xmath256 , there exists a coupling of @xmath686 and @xmath699 such that @xmath700 for @xmath660 , by corollary [ lem35 ] , we have that @xmath701 } \ { \hat{\mathbf{x}}_j^d \neq \hat{\mathbf{w}}_j^d \ } | \hat{\mathbf{x}}_0^d \equiv \hat{\mathbf{w}}_0^d = \mathbf{x}^d \biggr ) \rightarrow0 \qquad\mbox{as } { d \rightarrow\infty}.\ ] ] suppose that for @xmath702 $ ] , @xmath703
. then using lemma [ lem310 ] , ( [ eq310b ] ) , @xmath704 and @xmath705 can be coupled such that @xmath706 since @xmath707 , @xmath708 , the right - hand side of ( [ eq312d ] ) is less than @xmath709 .
note that @xmath710 so by lemma [ lem33 ] for any @xmath711 , @xmath712 times the right - hand side of ( [ eq312d ] ) converges to 0 as @xmath7 .
taking @xmath713 such that @xmath714 , @xmath715 } { \mathbb{p}}\bigl(m_j ( j_d ( \hat{\mathbf{x}}_j^d ) ) \neq m_j ( \omega_d ( \hat{\mathbf{w}}_j^d))| \hat{\mathbf{w}}_j^d = \hat{\mathbf{x}}_j^d \in f_d^1\bigr)\nonumber\\[-8pt]\\[-8pt ] & & \qquad\rightarrow0 \qquad \mbox{as } { d \rightarrow\infty}.\nonumber\end{aligned}\ ] ] the lemma then follows from ( [ eq312b ] ) and ( [ eq312e ] ) .
we show that it suffices to study @xmath716 } \sum_{j=0}^{[\pi d^\delta-1 ] } \omega_d ( \hat{\mathbf{w}}_j^d)^{-1}$ ] . in other words , replace the mean of the geometric random variables @xmath717 , @xmath718}^d ) ) \}$ ] by the mean of the means of the geometric random variables .
[ lem313 ] for any @xmath698 and for any sequence of @xmath719 such that , @xmath720 if @xmath721 as @xmath7 . let @xmath722 } \{\hat{\mathbf{w}}_j^d \notin f_d \}$ ]
. then for any @xmath723 , @xmath724 as @xmath7
. for any @xmath725 with @xmath726 , the characteristic function of @xmath687 conditional upon @xmath727 and @xmath519 is given by @xmath728 } { \mathbb{e}}\biggl [ \exp\biggl ( \frac{i \tau}{[d^\delta ] } m_j ( \omega_d ( \hat{\mathbf{w}}_j^d ) ) \biggr ) \big| a_d^c , \
{ \hat{\mathbf{w}}^d \ } \biggr ] \big| a_d^c , \hat{\mathbf{w}}_0^d = \mathbf{x}^d \biggr ] \\ & & \qquad= { \mathbb{e}}\biggl [ \prod_{j=0}^{[\pi d^\delta-1 ] } \frac{\exp(i \tau/[d^\delta ] ) \omega_d ( \hat{\mathbf{w}}_j^d)}{1 - ( 1 - \omega_d ( \hat{\mathbf{w}}_j^d ) ) \exp(i \tau/[d^\delta ] ) } \big| a_d^c , \hat{\mathbf{w}}_0^d = \mathbf{x}^d \biggr].\nonumber\end{aligned}\ ] ] conditional upon @xmath727 , @xmath729 .
hence , for all , @xmath730 ) \omega_d ( \hat{\mathbf{w}}_j^d)}{1 - ( 1 - \omega_d ( \hat{\mathbf{w}}_j^d ) ) \exp(i \tau/[d^\delta ] ) } = 1 + \frac{i \tau}{[d^\delta ] } \omega_d ( \hat{\mathbf{w}}_j^d)^{-1 } + o \biggl ( \frac{1}{[d^\delta ] } \biggr).\ ] ] thus @xmath731 $ ] has the same limit as @xmath7 ( should one exist ) as @xmath732 } \biggl ( 1 + \frac{i \tau}{[d^\delta ] } \omega_d ( \hat{\mathbf{w}}_j^d)^{-1 } \biggr ) \big| a_d^c , \hat{\mathbf{w}}_0^d = \mathbf{x}^d \biggr],\ ] ] which in turn has the same limit as @xmath7 as @xmath733 } \exp \biggl ( \frac{i \tau}{[d^\delta ] } \omega_d ( \hat{\mathbf{w}}_j^d)^{-1 } \biggr ) \big| a_d^c , \hat{\mathbf{w}}_0^d = \mathbf{x}^d \biggr]\nonumber\\[-8pt]\\[-8pt ] & & \qquad = { \mathbb{e}}[\exp(i \tau \tilde{t}_d ( \pi ) ) | a_d^c , \hat{\mathbf{w}}_0^d = \mathbf{x}^d].\nonumber\end{aligned}\ ] ] the lemma follows since @xmath734 as @xmath7 .
we shall show that @xmath735 as @xmath7 using chebyshev s inequality in lemma [ lem318 ] .
we require preliminary results concerning@xmath736 , @xmath737 with the key results given in lemma [ lem317 ] .
first , however , we introduce useful upper and lower bounds for @xmath738 which allow us to exploit lemma [ lem311 ] and prove uniform integrability @xmath739 . for @xmath663 , @xmath664 and @xmath740 ,
let @xmath741 with @xmath742 .
for @xmath663 and @xmath743 , let @xmath744 then for all @xmath475 , @xmath745 [ lem314 ] for any @xmath453 , any sequence of @xmath513 such that @xmath660 and any sequence of positive integers @xmath188 satisfying @xmath189 \leq k_d \leq[d^\delta]$ ] , @xmath746 \rightarrow\exp\bigl ( ( 2^m -1 ) \lambda(l ) \bigr ) \qquad\mbox{as } { d \rightarrow\infty}.\hspace*{-20pt}\ ] ] note that @xmath747 .
then since the @xmath748 are independent bernoulli random variables , @xmath749 & = & \prod_{j=1}^d { \mathbb{e}}\bigl [ ( 2^m)^{\chi_j ( x_j;l;k_d ) } | \hat{\mathbf{w}}_0^d = \mathbf{x}^d \bigr ] \nonumber\\[-8pt]\\[-8pt ] & = & \prod_{j=1}^d \bigl\ { \bigl(1 - q^d ( x_j;l;k_d)\bigr ) + 2^m q^d ( x_j;l;k_d ) \bigr\}.\nonumber\end{aligned}\ ] ] by lemma [ lem38a ] , ( [ eq38aa ] ) , for @xmath660 , @xmath750 as @xmath7 , so the right - hand side of ( [ eq314b ] ) has the same limit as @xmath7 as @xmath751 the lemma follows since for any @xmath660 , @xmath752 as @xmath7 .
[ lem315]fix @xmath753 .
for any sequence @xmath719 such that @xmath671 , and any sequence of positive integers @xmath188 satisfying @xmath189 \leq k_d \leq [ d^\delta]$ ] , we have that @xmath754 & \rightarrow&{\mathbb{e } } [ \check{\nu}_n ( \mathbf{s}_n)^m ] \qquad\mbox{as } { d \rightarrow\infty } , \\ { \mathbb{e } } [ \hat{\nu}_n ( \tilde{\mathbf{s}}_n^d ( \mathbf{x}^d ; k_d))^m ] & \rightarrow&{\mathbb{e } } [ \hat{\nu}_n^m ( \mathbf{s}_n)^m ] \qquad\mbox{as } { d \rightarrow\infty}.\end{aligned}\ ] ] by @xcite , theorem 29.2 , and lemma [ lem311 ] @xmath755 the lemma follows since ( [ eqnu3 ] ) and lemma [ lem314 ] ensure the uniform integrability of the left - hand sides of ( [ eq315a ] ) and ( [ eq315b ] )
. [ lem316 ] for any sequence @xmath756 such that @xmath671 and sequence of positive integers @xmath512 satisfying @xmath189 \leq k_d \leq[d^\delta]$ ] , @xmath757 \rightarrow\exp(f^\ast l/2 ) \qquad\mbox{as } { d \rightarrow\infty}.\ ] ] for any @xmath660 and sequences of positive integers @xmath758 and @xmath512 satisfying @xmath189 \leq k_d \leq[d^\delta]$ ] and @xmath759 $ ] , @xmath760 { \stackrel{p}{\longrightarrow}}\exp(f^\ast l/2 ) \qquad\mbox{as } { d \rightarrow\infty}.\ ] ] an immediate consequence of lemma [ lem315 ] is that @xmath761 , \lim_{{d \rightarrow\infty } } { \mathbb{e } } [ \hat{\nu}_n(\tilde{\mathbf{s}}_n^d ( \mathbf{x}^d ; k_d ) ) ] \rightarrow\exp(f^\ast l/2 ) \qquad\mbox{as } { n \rightarrow\infty},\ ] ] from which ( [ eq316a ] ) follows by ( [ eqnu3 ] ) . by theorem [ lem321 ] ,
( [ eqss47 ] ) , @xmath762 as @xmath7 , so ( [ eq316b ] ) follows from ( [ eq316a ] ) .
[ lem317 ] for any sequence @xmath756 such that @xmath660 and any sequences of positive integers @xmath763 and @xmath512 satisfying @xmath189 \leq i_d , k_d \leq[d^\delta]$ ] , @xmath764 and @xmath765 using ( [ eqnu3 ] ) , lemma [ lem314 ] and markov s inequality , it is straightforward to show that for any @xmath766 , there exists @xmath495 such that @xmath767 therefore it follows from lemma [ lem316 ] that , for any sequence @xmath513 such that @xmath660 , @xmath768 \nonumber\\[-2pt ] & & \hspace*{47pt}\qquad{}- { \mathbb{e } } [ \omega_d ( \hat{\mathbf{w}}_{j_d + k_d}^d)^{-1 } | \hat{\mathbf{w}}_0^d = \mathbf{x}^d ] \ } | \hat{\mathbf{w}}_0^d = \mathbf{x}^d \\[-2pt ] & & \qquad { \stackrel{p}{\longrightarrow}}0\qquad \mbox{as } { d \rightarrow\infty}.\nonumber\end{aligned}\ ] ] the uniform integrability of the left - hand side of ( [ eq317d ] ) follows from ( [ eqnu3 ] ) and lemma [ lem314 ] .
hence ( [ eq317a ] ) follows .
it is straightforward to show that @xmath769 , { \mathbb{e } } [ \hat{\nu}_n ( \mathbf{s}_n)^2 ] \rightarrow \exp(f^\ast l ( 4 \log2 - 3/2))$ ] as @xmath770 .
therefore from ( [ eqnu3 ] ) and lemma [ lem314 ] , we have that @xmath771 \rightarrow\exp\bigl(f^\ast l ( 4 \log2 - 3/2)\bigr ) \qquad\mbox{as } { d \rightarrow\infty}.\hspace*{-35pt}\ ] ] then ( [ eq317b ] ) follows immediately .
we are now in position to prove lemma [ lem318 ] , which is the final step in proving that for any sequence @xmath756 such that @xmath660 , @xmath772 as @xmath7 .
[ lem318 ] for any @xmath698 and any sequence @xmath719 such that , @xmath773 fix a sequence @xmath756 . let @xmath774 } \sum_{j=0}^{[d^\beta-1 ] } \omega_d ( \hat{\mathbf{w}}_j^d)^{-1}$ ] and let @xmath775 } \sum_{j= [ d^\beta]}^{[\pi d^\delta-1 ] } \omega_d ( \hat{\mathbf{w}}_j^d)^{-1}$ ] .
thus @xmath776 . let @xmath777 } \
{ \hat{\mathbf{w}}_j^d \notin f_d^1 \}$ ] . by theorem [ lem321 ] , ( [ eqss47 ] ) , @xmath778 as @xmath7 and conditional upon @xmath727 , @xmath779 d^\gamma}{[d^\delta]}$ ] .
hence @xmath780 as @xmath7 . by lemma [ lem316 ] , ( [ eq316a ] ) , @xmath781 & = & \frac{1}{[d^\delta ] } \sum_{j=[d^\beta]}^{[\pi d^\delta -1 ] } { \mathbb{e } } [ \omega_d ( \hat{\mathbf{w}}_j^d)^{-1 } | \hat{\mathbf{w}}_0^d = \mathbf{x}^d ] \nonumber\\[-9pt]\\[-9pt ] & \rightarrow & \pi\exp ( f^\ast l /2).\nonumber\vadjust{\goodbreak}\end{aligned}\ ] ] by chebyshev s inequality , for any @xmath563
, @xmath782 \bigr| > \varepsilon| \hat{\mathbf{w}}_0^d = \mathbf{x}^d \bigr ) \nonumber\\[-8pt]\\[-8pt ] & & \qquad\leq\frac{1}{\varepsilon^2 [ d^\delta]^2 } \sum_{j=[d^\beta ] } ^{[\pi d^\delta-1 ] } \sum_{l=[d^\beta]}^{[\pi d^\delta-1 ] } \operatorname{cov } \bigl ( \omega_d ( \hat{\mathbf{w}}_j^d)^{-1 } , \omega_d ( \hat{\mathbf{w}}_l^d)^{-1 } | \hat{\mathbf{w}}_0^d = \mathbf{x}^d \bigr).\nonumber\end{aligned}\ ] ] since for all @xmath783 , @xmath784\\[-8pt ] & & \qquad\leq \operatorname{var } \bigl ( \omega_d ( \hat{\mathbf{w}}_j^d)^{-1 } | \hat{\mathbf{w}}_0^d = \mathbf{x}^d \bigr)^{{1}/{2 } } \operatorname{var } \bigl ( \omega_d ( \hat{\mathbf{w}}_l^d)^{-1 } | \hat{\mathbf{w}}_0^d = \mathbf{x}^d \bigr)^{{1}/{2}},\nonumber\end{aligned}\ ] ] it is straightforward to show , using lemma [ lem317 ] , that the right - hand side of ( [ eq318c ] ) converges to 0 as @xmath7 .
thus @xmath785 as @xmath7 and the lemma follows immediately . [ lem319 ] for any sequence @xmath756 such that @xmath660 , @xmath786 for any @xmath698 , by lemmas [ lem312 ] , [ lem313 ] and [ lem318 ] , @xmath787 since @xmath650 satisfies @xmath788)$ ] , for any @xmath563 , @xmath789 for all sufficiently large @xmath0 .
the lemma follows , since ( [ eq319b ] ) ensures that the right - hand side of ( [ eq319c ] ) converges to 0 as @xmath7 .
from appendix [ secpd ] , we have that for any sequence @xmath513 , such that @xmath660 , latexmath:[$p_d as @xmath7 . therefore we proceed by showing that , for any @xmath260 , @xmath791 where @xmath792 } { \mathbb{e } } [ ( h(\hat{\mathbf{x}}^d_{[\pi d^\delta ] } ) - h ( \hat{\mathbf{x}}^d_0 ) ) @xmath793 equation ( [ eqn115 ] ) will then be proved using the triangle inequality .
we analyze @xmath794 ) , before using ( [ eqmain6 ] ) to study @xmath795 . however , first we require some definitions and preliminary results . throughout
we will utilize the following key facts noted in section [ secalg ] : @xmath796 and that @xmath797 , where @xmath798 and @xmath799 .
we follow @xcite and @xcite in noting that , for any function @xmath800 which is a twice differentiable function on @xmath801 , the function @xmath802 is also twice differentiable , except at a countable number of points , with first derivative given lebesgue almost everywhere by the function @xmath803 the second derivative can similarly be obtained but will not be explicitly required for our calculations . for @xmath804 ,
let @xmath805 denote the probability of accepting a move in the rwm algorithm given that @xmath806 and let @xmath807\\[-8pt ] & & \hspace*{10.3pt}{}\times 1_{\{\sum_{j=2}^d ( g ( x_j + \sigma_d z_{1,j } ) - g ( x_j ) ) < 0 \ } } \prod_{j=2}^d 1 _ { \ { 0
< x_j + \sigma_d z_{1,j } < 1 \ } } \biggr].\nonumber\end{aligned}\ ] ] then for all @xmath87 , using taylor s theorem , @xmath808 therefore for @xmath809 , @xmath810 [ lem40 ] @xmath811 let @xmath812 $ ] and let @xmath813 $ ] , the probability a proposed move stays inside the unit cube given that the first component does not move . the proof of ( [ eq33b ] ) can be adapted to show that , for any @xmath478 , @xmath814 as @xmath7 .
therefore since for @xmath671 , @xmath815 , ( [ eqlowerj ] ) , we have that @xmath816 let @xmath817 and let @xmath818 .
since @xmath819 , we have that @xmath820 then using a taylor series expansion , there exists @xmath495 such that , for all @xmath671 , @xmath821 since @xmath822 are independent , and whether or not a proposed move from @xmath87 stays inside the hypercube depends only upon @xmath823 , @xmath824\\[-8pt ] & & \qquad \leq\tilde{\omega}_d^0 ( \mathbf{x}^d ) \leq\omega_d^0 ( \mathbf{x}^d ) { \mathbb{p}}\bigl(i_d ( \mathbf{x}^d ) <
k \log d /d\bigr).\nonumber\end{aligned}\ ] ] for all @xmath671 , @xmath825 $ ] , so @xmath826)\qquad \mbox{as $ { d \rightarrow\infty}$}.\ ] ] therefore it follows that @xmath827 with the lemma following from ( [ eq40b ] ) and ( [ eq40e ] ) by the triangle inequality .
[ lem41 ] for @xmath809 and @xmath671 , @xmath828 where @xmath98 as @xmath7 .
for @xmath829 , @xmath830 for @xmath126 , fix @xmath671 and suppose that @xmath831
. then @xmath832 \nonumber\\[-4pt]\\[-12pt ] & = & \frac{d^2}{j_d ( \mathbf{x}^d ) } { \mathbb{e}}\biggl [ \bigl(h ( \mathbf{x}^d + \sigma_d \mathbf{z}^d ) - h ( \mathbf{x}^d)\bigr ) \biggl\ { 1 \wedge \frac{\pi_d ( \mathbf{x}^d + \sigma_d \mathbf{z}^d)}{\pi_d ( \mathbf{x}^d ) } \biggr\ } \biggr].\nonumber\end{aligned}\ ] ] the right - hand side of ( [ eq41ex ] ) is familiar in that it is the generator of the rwm - algorithm divided by the acceptance probability ; see , for example , @xcite , page 113 .
first , note that @xmath833 using ( [ eqn4b ] ) , ( [ eqn4c ] ) and noting that @xmath834 , we have that @xmath835 \nonumber\\ & = & \frac{d^2 j_d^0 ( \mathbf{x}^d)}{j_d^0 ( \mathbf{x}^d ) + o ( \sigma_d^2 ) } \sigma_d { \mathbb{e}}[z_1 ] h^\prime(x_1)\nonumber\\ & & { } + \frac{d^2 j_d^0 ( \mathbf{x}^d)}{j_d^0 ( \mathbf{x}^d ) + o ( \sigma_d^2 ) } \frac{\sigma_d^2}{2 } { \mathbb{e}}[z_1 ^ 2 ] h^ { \prime\prime } ( x_1 ) \nonumber\\ & & { } + \frac{d^2 j_d^0 ( \mathbf{x}^d)}{j_d^0 ( \mathbf{x}^d ) + o ( \sigma_d^2 ) } \frac{\sigma_d^2}{2 } { \mathbb{e}}[z_1 ^ 2 \{h^ { \prime\prime } ( x_1 + \psi_1^d ) - h''(x_1 ) \ } ] \\ & & { } + \frac{d^2 \tilde{j}_d^0 ( \mathbf{x}^d)}{j_d^0 ( \mathbf{x}^d ) + o ( \sigma_d^2 ) } \sigma_d^2 g^\prime(x_1 ) h^\prime(x_1 ) { \mathbb{e}}[z_1 ^ 2]\nonumber\\ & & { } + \frac{d^2 } { j_d^0 ( \mathbf{x}^d ) + o ( \sigma_d^2 ) } o ( \sigma_d^3 ) .
\nonumber\end{aligned}\ ] ] the first term on the right - hand side of ( [ eq41eb ] ) is 0 . since @xmath224 , by the continuous mapping theorem , @xmath836 as @xmath7 and then since @xmath837 is bounded the third term on the right - hand side of ( [ eq41eb ] ) converges to 0 as @xmath7 . for @xmath671 , @xmath838 , and
so , the right - hand side of ( [ eq41eb ] ) equals @xmath839 where @xmath98 as @xmath7 .
thus ( [ eq41fx ] ) is proved .
the proof of ( [ eq41fa ] ) follows straightforwardly using taylor series expansions since @xmath840 . since @xmath841 , an immediate consequence of lemma [ lem41 ] is that , there exists @xmath842 such that @xmath843 [ lem42 ] for any sequence of positive integers @xmath188 satisfying @xmath189 \leq k_d \leq[d^\delta]$ ] , @xmath844 - \hat{g } h ( x_1 ) \bigr|
\rightarrow0 \qquad\mbox{as } { d \rightarrow\infty}.\vspace*{-2pt}\ ] ] fix @xmath576 and note that @xmath845 \nonumber\\ & & \qquad= { \mathbb{p}}({\hat{\mathbf{x}}}_{k_d}^d \in f_d
| \hat{\mathbf{x}}_0^d = \mathbf{x}^d ) { \mathbb{e } } [ \hat{g}_d h ( { \hat{\mathbf{x}}}_{k_d}^d ) | \hat{\mathbf{x}}_0^d = \mathbf{x}^d , { \hat{\mathbf{x}}}_{k_d}^d \in f_d ] \\ & & \qquad\quad { } + { \mathbb{p}}({\hat{\mathbf{x}}}_{k_d}^d \notin f_d | \hat{\mathbf{x}}_0^d = \mathbf{x}^d ) { \mathbb{e } } [ \hat{g}_d h ( { \hat{\mathbf{x}}}_{k_d}^d ) | \hat{\mathbf{x}}_0^d = \mathbf{x}^d , { \hat{\mathbf{x}}}_{k_d}^d \notin f_d ] .\nonumber\end{aligned}\ ] ] since @xmath846 , @xmath847 . therefore , for all @xmath848^d$ ] , @xmath849 . by ( [ emainx1 ] ) , @xmath850 as @xmath7 .
thus the latter term on the right - hand side of ( [ eq42ab ] ) converges to 0 as @xmath7 .
now @xmath851 \nonumber\\ & & \qquad= { \mathbb{p}}(\hat{x}_{k_d,1}^d \notin r_d^l | \hat{\mathbf{x}}_0^d = \mathbf{x}^d , { \hat{\mathbf{x}}}_{k_d}^d \in f_d ) \nonumber\\ & & \qquad\quad{}\times{\mathbb{e } } [ \hat{g}_d h ( { \hat{\mathbf{x}}}_{k_d}^d ) | \hat{\mathbf{x}}_0^d = \mathbf{x}^d , { \hat{\mathbf{x}}}_{k_d}^d \in f_d , \hat{x}_{k_d,1}^d \notin r_d^l ] \\ & & \qquad\quad { } + { \mathbb{p}}(\hat{x}_{k_d,1}^d \in r_d^l | \hat{\mathbf { x}}_0^d = \mathbf{x}^d , { \hat{\mathbf{x}}}_{k_d}^d \in f_d )
\nonumber\\ & & \qquad\quad\hspace*{11pt}{}\times{\mathbb{e } } [ \hat{g}_d h ( { \hat{\mathbf{x}}}_{k_d}^d ) |
\hat{\mathbf{x}}_0^d = \mathbf{x}^d , { \hat{\mathbf{x}}}_{k_d}^d \in f_d , \hat{x}_{k_d,1}^d \in r_d^l ] .\nonumber\end{aligned}\ ] ] consider first the latter term on the right - hand side of ( [ eq42ac ] ) . by lemma [ lem41 ] , ( [ eq41fa ] ) , @xmath852 \leq\tfrac{3}{2 } l^2 h^\ast_2.\ ] ] note that @xmath853\\[-8pt ] & \leq & \frac{{\mathbb{p}}(\hat{x}_{k_d,1}^d \in r_d^l | \hat{\mathbf{x}}_0^d = \mathbf{x}^d)}{{\mathbb{p}}({\hat{\mathbf{x}}}_{k_d}^d \in f_d |\hat{\mathbf{x}}_0^d = \mathbf{x}^d)}.\nonumber\end{aligned}\ ] ] by ( [ emainx1 ] ) , for @xmath660 , @xmath854 as @xmath7 .
use corollary [ lem35 ] and lemma [ lem38a ] to show that @xmath855 as @xmath7 .
hence , the right - hand side of ( [ eq42ad ] ) converges to 0 as @xmath7 and consequently the latter term on the right - hand side of ( [ eq42ac ] ) converges to 0 as @xmath7 .
it follows from the above arguments that @xmath856 also it follows from ( [ eqn4d ] ) that there exists @xmath495 such that @xmath857 \leq k.\ ] ] therefore , it is straightforward using ( [ eq42ab ] ) , ( [ eq42ac ] ) and the triangle inequality to show that @xmath858 \nonumber\\ & & \qquad\quad\hspace*{-8.6pt}{}- { \mathbb{e } } [ \hat{g}_d h ( { \hat{\mathbf{x}}}_{k_d}^d ) | \hat{\mathbf{x}}_0^d = \mathbf{x}^d , { \hat{\mathbf{x}}}_{k_d}^d \in f_d , \hat{x}_{k_d,1}^d \notin r_d^l ] \bigr|\\ & & \qquad \rightarrow0 \qquad\mbox{as } { d \rightarrow\infty}.\nonumber\end{aligned}\ ] ] by lemma [ lem41 ] , ( [ eq41fx ] ) , there exists @xmath859 as @xmath7 , such that @xmath860 \bigr| \nonumber\\[-1pt ] & & \qquad\leq\frac{l^2}{3 } \sup_{0 \leq y \leq1 } | g^\prime(y ) h^\prime(y)| \nonumber\\[-8.5pt]\\[-8.5pt ] & & \qquad\quad{}\times\sup_{\mathbf{x}^d \in\tilde{f}_d } { \mathbb{e}}\biggl [ \biggl| \frac{\tilde{j}_d^0 ( { \hat{\mathbf{x}}}_{k_d}^d)}{j_d^0 ( { \hat{\mathbf{x}}}_{k_d}^d ) } - \frac{1}{2 } \biggr| \big| \hat{\mathbf{x}}_0^d = \mathbf{x}^d , { \hat{\mathbf{x}}}_{k_d}^d \in f_d , \hat{x}_{k_d,1}^d \notin r_d^l \biggr ] + \varepsilon_d^1 \nonumber\\[-1pt ] & & \qquad\leq\frac{l^2}{3 } g^\ast h^\ast_1 \sup_{\mathbf{y}^d \in f_d } \biggl| \frac{\tilde{j}_d^0 ( \mathbf{y}^d)}{j_d^0 ( \mathbf{y}^d ) } - \frac{1}{2 } \biggr| + \varepsilon_d^1.\nonumber\end{aligned}\ ] ] by lemma [ lem40 ] , the right - hand side of ( [ eq42ah ] ) converges to 0 as @xmath7 . using the triangle inequality , the lemma follows by showing that @xmath861 - \hat{g } h ( x_1 ) \bigr|\nonumber\\[-8.5pt]\\[-8.5pt ] & & \qquad\rightarrow0 \qquad\mbox{as } { d \rightarrow\infty}.\nonumber\end{aligned}\ ] ] note that @xmath862 , and so , ( [ eq42ai ] ) follows since @xmath863 is continuous .
we are in position to prove ( [ eq41e ] ) .
[ lem43 ] for any @xmath260 , @xmath864 since ( [ eq43a ] ) trivially holds for @xmath865 , we assume that @xmath866 . for all sufficiently large @xmath0 , by the triangle inequality , @xmath867 & & \qquad= \biggl| \frac{1}{[d^\delta ] } \sum_{j=0}^{[\pi d^\delta-1 ] } { \mathbb{e } }
[ \hat{g}_d h ( { \hat{\mathbf{x}}}_j^d ) |
\hat{\mathbf{x}}_0^d = \mathbf{x}^d ] - \pi\hat{g } h ( x_1 ) \biggr| \nonumber\\[-1pt ] & & \qquad\leq\biggl| \frac{1}{[d^\delta ] } \sum_{j=0}^{[d^\beta ] -1 } { \mathbb{e } } [ \hat{g}_d h ( \hat{\mathbf{x}}_j^d ) |
\hat{\mathbf{x}}_0^d = \mathbf{x}^d ] \biggr| \\[-1pt ] & & \qquad\quad { } + \frac{1}{[d^\delta ] } \sum_{j=[d^\beta]}^{[\pi d^\delta-1 ] } \bigl| { \mathbb{e } } [ \hat{g}_d h ( \hat{\mathbf{x}}_j^d ) |
\hat{\mathbf{x}}_0^d = \mathbf{x}^d ] - \hat{g } h ( x_1 ) \bigr|\nonumber\\[-1pt ] & & \qquad\quad { } + \biggl ( \pi- \frac{[\pi d^\delta ] - [ d^\beta]}{[d^\delta ] } \biggr ) \hat{g } h ( x_1).\nonumber\end{aligned}\ ] ] since @xmath868 \nonumber\\ & & \qquad = { \mathbb{e } } [ \hat{g}_d h ( \hat{\mathbf{x}}_j^d ) |
\hat{\mathbf{x}}_0^d = \mathbf{x}^d , { \hat{\mathbf{x}}}_j^d \in f_d ] { \mathbb{p}}({\hat{\mathbf{x}}}_j^d \in f_d | \hat{\mathbf{x}}_0^d = \mathbf{x}^d ) \\ & & \qquad\quad { } + { \mathbb{e } } [ \hat{g}_d h ( \hat{\mathbf{x}}_j^d ) |
\hat{\mathbf{x}}_0^d = \mathbf{x}^d , { \hat{\mathbf{x}}}_j^d \notin f_d ] { \mathbb{p}}({\hat{\mathbf{x}}}_j^d \notin f_d | \hat{\mathbf{x}}_0^d = \mathbf{x}^d),\nonumber\end{aligned}\ ] ] it is straightforward , following a similar argument to the proof of lemma [ lem42 ] , ( [ eq42af ] ) , to show that there exists @xmath869 such that , for all @xmath870 $ ] , @xmath871 \bigr| \leq\tilde{k}.\ ] ] therefore the first term on the right - hand side of ( [ eq43c ] ) is bounded by @xmath189 \tilde{k}/[d^\delta]$ ] . by lemma [ lem42 ] the supremum over @xmath660 of the second term on the right - hand side of ( [ eq43b ] ) converges to 0 as @xmath7 and the lemma follows .
[ lem44 ] @xmath872 fix @xmath563 and let @xmath873 \varepsilon , 1\}$ ] .
it follows from lemma [ lem43 ] that , for all sufficiently large @xmath0 , @xmath874 consider any @xmath260 .
there exists @xmath875 such that @xmath876 . by the triangle inequality ,
@xmath877 again by the triangle inequality , @xmath878 } \sum_{j = [ \tilde{\pi } d^\delta]}^{[\pi d^\delta-1 ] } \sup_{\mathbf{x}^d \in\tilde{f}_d } \bigl|{\mathbb{e } } [ \hat{g}_d h ( \hat{\mathbf{x}}_j^d ) | \hat{\mathbf{x}}_0^d = \mathbf{x}^d ] \bigr|.\nonumber\end{aligned}\ ] ] since for all sufficiently large @xmath0 , @xmath879 - [ \tilde{\pi } d^\delta])/[d^\delta ] \leq2 \varepsilon$ ] , it follows from ( [ eq43d ] ) that the right - hand side of ( [ eq44d ] ) is bounded by @xmath880 , where @xmath881 is defined in lemma [ lem43 ] .
let @xmath882 .
note that since @xmath883 , we have that @xmath884
. therefore it follows from ( [ eq44c ] ) that for all sufficiently large @xmath0 , latexmath:[\[\label{eq44e } \sup_{\mathbf{x}^d \in\tilde{f}_d } @xmath260 and @xmath563 , the lemma follows .
finally we are in position to prove ( [ eqn115 ] ) , and hence complete the proof of theorem [ main ] .
[ lem45 ] @xmath886 note that @xmath649 is given by ( [ eqn113 ] ) and @xmath887 . therefore by the triangle inequality , @xmath888 } { \mathbb{e}}\bigl [ h \bigl(\hat{\mathbf{x}}_{[p_d d^\delta]}^d\bigr ) - h ( \hat{\mathbf{x}}_0^d ) | \hat{\mathbf{x}}_0^d = \mathbf{x}^d \bigr ] - \exp(- l f^\ast/2 ) \hat{g } h ( x_1 ) \biggr| \nonumber\\ & & \qquad\leq\sup_{\mathbf{x}^d \in\tilde{f}_d } \biggl| { \mathbb{e}}\biggl [ \frac{d^2}{[d^\delta ] } \bigl ( h \bigl(\hat{\mathbf{x}}_{[p_d d^\delta]}^d\bigr ) - h ( \hat{\mathbf{x}}_0^d ) \bigr ) - p_d \hat{g } h ( x_1 ) | \hat{\mathbf{x}}_0^d = \mathbf{x}^d \biggr ] \biggr| \nonumber\\ & & \qquad\quad { } + \sup_{\mathbf{x}^d \in\tilde{f}_d } \bigl| { \mathbb{e } } [ p_d \hat{g } h ( x_1 ) | \hat{\mathbf{x}}_0^d = \mathbf{x}^d ] - \exp ( - l f^\ast/2 ) \hat{g } h ( x_1 ) \bigr| \nonumber\\[-8pt]\\[-8pt ] & & \qquad\leq\sup_{0 \leq\pi\leq1 } \sup_{\mathbf{x}^d \in \tilde{f}_d }
\biggl| { \mathbb{e}}\biggl [ \frac{d^2}{[d^\delta ] } \bigl ( h \bigl(\hat{\mathbf{x}}_{[\pi d^\delta]}^d\bigr ) - h ( \hat{\mathbf{x}}_0^d ) \bigr ) - \pi\hat{g } h ( x_1 ) | \hat{\mathbf{x}}_0^d = \mathbf{x}^d \biggr ] \biggr| \nonumber\\ & & \qquad\quad{}+ \sup_{\mathbf{x}^d \in\tilde{f}_d } \bigl| { \mathbb{e } } [ p_d | \hat{\mathbf{x}}_0^d = \mathbf{x}^d ] - \exp(- l f^\ast/2 ) \bigr| \sup_{0 \leq
y \leq1 } |\hat{g } h ( y)| \nonumber\\ & & \qquad\leq\sup_{0 \leq\pi\leq1 } \sup_{\mathbf{x}^d \in \tilde{f}_d } | \hat{g}_d^{\delta , \pi } h ( \mathbf{x}^d ) - \pi \hat{g } h ( x_1 ) | \nonumber\\ & & \qquad\quad { } + \sup _ { \mathbf{x}^d \in\tilde{f}_d } \bigl| { \mathbb{e } } [ p_d | \hat{\mathbf{x}}_0^d = \mathbf{x}^d ] - \exp(- l f^\ast/2 ) \bigr|
\sup_{0 \leq y \leq1 } |\hat{g } h ( y)|.\nonumber\end{aligned}\ ] ] by corollary [ lem44 ] , the first term on the right - hand side of ( [ eq45ba ] ) converges to 0 as @xmath7 . by theorem [ lem319 ] , for any sequence @xmath756 such that @xmath660 , latexmath:[$p_d as @xmath7 .
hence the latter term on the right - hand side of ( [ eq45ba ] ) converges to 0 as @xmath7 , since @xmath883 implies that @xmath889 .
we thank the anonymous referees for their helpful comments which have improved the presentation of the paper . | we consider the optimal scaling problem for high - dimensional random walk metropolis ( rwm ) algorithms where the target distribution has a discontinuous probability density function .
almost all previous analysis has focused upon continuous target densities .
the main result is a weak convergence result as the dimensionality @xmath0 of the target densities converges to @xmath1 . in particular ,
when the proposal variance is scaled by @xmath2 , the sequence of stochastic processes formed by the first component of each markov chain converges to an appropriate langevin diffusion process .
therefore optimizing the efficiency of the rwm algorithm is equivalent to maximizing the speed of the limiting diffusion .
this leads to an asymptotic optimal acceptance rate of @xmath3 @xmath4 under quite general conditions .
the results have major practical implications for the implementation of rwm algorithms by highlighting the detrimental effect of choosing rwm algorithms over metropolis - within - gibbs algorithms . , . |
one of the challenging properties of qcd is the rapid increase with energy of the cross sections of the hard processes . initially the increase has been predicted within the leading order dglap approximation @xcite .
the rapid increase of the structure functions of the proton with the energy has been observed in ep scattering , for the review and proper references see ref .
the deep - inelastic structure functions of a proton , calculated in the perturbative qcd ( pqcd ) within the leading twist ( lt ) approximation , can be fitted at small @xmath0 as @xmath2 , where @xmath3 at @xmath4 gev@xmath5 .
the generalization of the qcd factorization theorems to the amplitudes of hard diffractive processes shows that these cross sections ( that are higher - twist effects ) should increase with energy even faster than the structure functions , calculated in the leading - twist approximation @xcite .
predicted by the perturbative qcd increase @xmath6 of cross sections of hard diffractive processes , has been recently observed at hera ( desy ) in the diffractive photoproduction of the heavy flavor mesons @xmath7 , in the high @xmath8 diffractive electroproduction of @xmath9 mesons , for the review and references see ref .
the rapid increase of amplitudes of hard processes with energy predicted in perturbative qcd within the dglap approximation can not continue forever since it violates at sufficiently small @xmath0 the strict inequality : @xmath10 . here
@xmath11 is the full , and @xmath12 is the differential cross - section , and the cross - section @xmath13 is calculated within the lt approximation , see ref .
@xcite . to quantify the theoretical challenge it is useful to introduce an auxiliary theoretical object - the amplitude describing the scattering of colorless and spatially small dipole of the transverse size @xmath14 of a hadron target or of another small size dipole .
it follows from the probability conservation that properly normalized such amplitude for the scattering at the fixed impact parameter @xmath15 should be restricted from above : @xmath16 , see refs . @xcite and references therein .
this inequality is violated within the dglap approximation beyond the value of bjorken @xmath0 equal to @xmath17 .
probability conservation restricts region of applicability of the dglap approximation but it does not precludes the fast increase of the dipole cross section @xmath18 corresponding to black disk limit , cf . ref .
the structure functions should increase with energy even faster : @xmath19 resulting from the increase with energy of essential impact parameters and from the ultraviolet divergence of hadronic contribution into renormalization of the electric charge , see ref .
@xcite .
the applicability of the competing leading order ( lo ) @xmath20 approximation requires rather large energies , because the convergency of the perturbative qcd series at small @xmath0 is rather slow , ( c.f .
( 2.5 ) below ) .
the evaluation of the next - to - leading order effects found huge correction and significant reduction of the increase of the amplitudes with energy obtained within lo approximation @xcite due to the necessity to restore the energy - momentum conservation @xcite .
the further reduction of the energy dependence of the amplitude has been found in ref .
@xcite where the separation of scales characteristic for hard and soft qcd processes has been performed and the running of the coupling constant and the energy - momentum conservation were taken into account .
the practical conclusion is that almost in the whole range of lhc energies the lo+nlo dglap and bfkl approximations predict rather similar energy dependence of the structure functions , cf .
it seems now that the challenge of violation of probability conservation becomes important at lesser energies than the difference between the different pqcd calculations reveals itself at small x. small @xmath0 physics problem for the two scale processes has been discussed recently in @xcite within the color glass condensate ( cgc ) approach .
the unitarity of scattering matrix has not been achieved within the made approximations @xcite .
( see however the refs .
@xcite whose approach is similar to that used in ref . @xcite but different from the mechanism discussed in this paper . )
the perturbative qcd calculation of the scattering of a color - neutral two - gluon dipole of the size @xmath14 off a proton target violates probability conservation for @xmath21 gev@xmath22 for the scattering at central impact parameters at @xmath23@xmath24 , i.e. in the kinematics typical for the large hadron collider ( lhc ) @xcite .
the puzzle arises in the kinematics when few gluons are produced only .
it reveals itself in the gluon structure function of a proton as the consequence of the large gluon density in the initial condition for qcd evolution and the rapid increase with energy of the perturbative qcd amplitudes @xcite .
the violation of the leading twist approximation should become important at the values of @xmath0 where the perturbative qcd calculations in the leading twist approximation are still more or less unambiguous .
the scale of the hard processes @xmath25 gev@xmath5 is chosen to guarantee the smallness of the running invariant charge .
note that the existence of the discussed above puzzle in the amplitude for the scattering of two spatially small dipoles investigated in this paper is unrelated to the poorly understood physics of the quark confinement and the spontaneous broken chiral symmetry . with the increase of the collision energy
the higher - twist effects blow up : higher is the twist of the term more rapidly it increases with the energy . at sufficiently small @xmath0
the perturbative qcd series have the form : @xmath26 where coefficients @xmath27 are rapidly increasing with @xmath28 .
the rapid increase of @xmath27 with @xmath28 is known for a long time from the study of the topology of the iterations of the ladder diagrams @xcite .
the challenging problem is to develop the unambiguous method of summing the divergent series .
one of the aims of this paper is the reconstruction of the nonperturbative terms related to this divergence , the terms that can not be decomposed into powers of @xmath1 . at sufficiently large energies ,
the hadron scattering at central impact parameters becomes completely absorptive , as the consequence of the compositeness of the projectile and the increase of the interaction with the energy , see ref .
@xcite and references therein .
the fnal data on the differential cross sections of elastic @xmath29 collisions shows that the @xmath29 scattering at central impact parameters is black ( cf .
discussion in refs .
this fact can be considered as an experimental hint of expected limiting behavior of hard processes also at small @xmath0 where the interactions become strong .
another distinctive property of small x hard processes is the rapid increase with energy of the longitudinal distances important in the scattering process .
the coherence length @xmath30 in the total cross section of the deep inelastic scattering in the kinematics near the black disk limit is increasing with energy as @xmath31 .
this increase is somewhat less rapid than the one familiar from the analysis of the leading twist approximation , see ref . @xcite and references therein .
nevertheless , at current and future accelerators @xmath30 significantly exceeds the static radius of a hadron or nucleus .
in fact , @xmath30 becomes comparable with the electromagnetic radius of the hydrogen atom in the kinematic region to be studied at the large hadron collider . as a result
, a variety of observable new coherent phenomena are expected to appear in the small-@xmath0 processes @xcite . in addition
it has been shown that in the coordinate space the correlators of the currents evaluated within both the dglap and bfkl approximations increase rapidly with the distance @xcite .
such increase is a necessary condition for the onset of the critical phenomena @xcite .
the perturbative qcd produces branch points in the angular momentum plane located at @xmath32 i.e. in the region forbidden by the causality and conservation of probability .
thus the theory has a tachyon . in this respect
, certain similarity may exist between the theory of small @xmath0 phenomena and the theory of strings in 26 dimensions where tachyon is also present in the perturbative ( nonphysical ) vacuum , cf .
discussion in ref .
@xcite . in these theories
the account of the nonperturbative phenomenona is necessary to find a true ground state .
the well understood property of qcd is the strong dependence on bjorken @xmath0 of the piece of the quark - gluon component of the wave function of the virtual photon that dominates the structure function of a hadron target . at moderately small @xmath0 , both soft components in the wave function of the virtual photon and the quark - gluon configurations , where the constituents have large relative transverse momenta , give comparable contributions , in the spirit of the aligned jet model @xcite . with the increase of the collision energy
this conspiracy disappears : the hard configurations that occupy most of the phase volume in the photon wave function begin to dominate due to the rapid increase of the hard cross sections with the energy .
this corresponds to the serious change of the @xmath8 dependence of the structure functions to the regime where @xmath33 , for the review and references see ref .
in the new regime hard pqcd phenomena should dominate in the structure functions .
the cross section of the diffractive production of quark - gluon system with large mass determines the triple ladder vertex which enters our calculations . in order to simplify our task
we restrict ourselves to the kinematics where the wave function of the longitudinally polarized photon @xmath34 is dominated by the configurations of the constituents with large transverse momenta and large invariant mass the hard perturbative qcd analogue of the triple - reggeon limit .
the account of this phenomenon helps to evaluate the triple reggeon vertex near the black disk limit where the leading logarithmic approximations are violated .
a necessary kinematical condition of the applicability of our method is @xmath35 . here
@xmath36 can be determined from the condition that the contribution of one ladder in dipole
dipole scattering at central impact parameters is near the black disk limit in the leading twist approximation .
this inequality is just the condition for the existence of the triple - regge limit and it is the main kinematical limitation of our approach .
the aim of this paper is to show that some of the challenges discussed above can be met for one - scale hard small @xmath0 processes , such as @xmath37 hadrons , @xmath38 , etc .
we restrict ourselves by the consideration of the one - scale hard processes where the contribution into the amplitude of the qcd evolution with scale is suppressed .
consequently the factorization of the information on the wave functions of the projectile and target from the interaction , similar to the one that occurs in the regge pole exchanges , becomes a reasonable approximation .
the rapid increase of the perturbative qcd amplitudes with energy leads to even more rapid increase of the role of the shadowing effects .
it was found long ago that the generalization of the technology of calculations of shadowing effects , encoded in the reggeon field theory , may appear useful for the description of the small @xmath0 processes @xcite .
the generalization of the latter approach , which accounts for any diffractive processes , leads to the effective field theory ( eft ) .
the evident advantage of the eft , as compared to models based on elastic eikonal approximation ( see i.e. ref .
@xcite ) is in the possibility to account for the whole variety of rescattering effects , and to ensure the energy -momentum conservation .
in addition , we will show in this paper , that the transition to the black disk limit ( bdl ) in hard processes has a certain resemblance to a critical phenomena .
note that the soft qcd contribution to the scattering processes is almost absent in the chosen processes .
this suppression is achieved by the choice of the longitudinally polarized highly virtual photons as the projectile and the target .
although the method developed in the paper is inapplicable beyond the one scale hard processes , the new qcd phenomena we found in the paper may appear important for the many - scale hard processes as well . to account for the coherence of the high - energy processes and the rapid increase with energy of the amplitudes of the hard small-@xmath0 processes ,
we construct an effective field theory ( eft ) of the interacting perturbative colorless ladders .
the evident advantage of this approach is the significant reduction of the number of variables in the problem and the mapping of the physics of the coherent processes into the framework resembling the statistical models .
( this approach is in the spirit of the statistical models of critical phenomena that account for the interactions between the major modes only .
the specifics of the physics of the large longitudinal distances is included in the concept of the quasiparticle . )
the interaction between the quasiparticles , when the amplitudes of the hard processes are near the unitarity limit , can be easily evaluated within the wkb approximation .
the use of the wkb approximation is justified by the smallness of the running coupling constant in the perturbative qcd .
we show that account of the running coupling constant helps to establish the dominance of the triple - ladder vertex ( see appendix b ) .
the smallness of the multiladder vertices in pqcd implies that the basic phenomena that characterizes the black disk limit should be insensitive to the restriction by the triple - ladder vertex .
this observation helps to fix the form of the lagrangian of the eft near the unitarity limit . since @xmath39 in pqcd ,
the quasiparticle of the eft is a tachyon : the singularities in the complex angular - momentum plane are located at @xmath40 where @xmath41 .
such a behavior is in the evident conflict with the bound for the total cross section of a small dipole
dipole scattering , that follows from the causality and conservation of probability in qcd .
the causality and unitarity of the @xmath42 matrix , however , are not necessarily valid in the perturbative qcd vacuum . the rapid increase of the amplitude with energy predicted by the perturbative qcd approximations ( i.e. the presence of the tachyon in the eft ) reflects a rapid transition from the false , perturbative vacuum to the nonperturbative , physical vacuum . to visualize this phenomenon
it is useful to consider physical processes in coordinate space @xcite .
so these approximations should fail to predict actual energy dependence of amplitudes of physical processes which is nonperturbative phenomenon .
we find in the paper that account of nonlinear phenomena leads to serious restructuring of produced qcd states besides the onset of black disk limit .
with the lagrangian of the eft at hand , we may identify the order parameter relevant for the critical phenomena in small-@xmath0 processes : the order parameter is the condensate of the quasiparticles , that are the pqcd ladders .
we find the new nonperturbative phenomena in the regime of the strong interaction with a small coupling constant , including the new nonperturbative classical fields , whose effects are @xmath43 , the color inflation due to the tunneling transitions , the zero modes in the dominant classical fields and related massless particles -``phonons '' in the eft , and the formation of a color network where the overlapping quasiparticles ( ladders ) do exchange the colored constituents .
the approach developed in this paper differs from the eft suggested in ref .
the major practical differences are that we restrict ourselves by one scale hard processes only , in the dominance in the wkb approximation of the tunneling transitions and the quantum fluctuations around them , while the contribution of the quantum loops calculated near the vacuum of the pqcd vacuum is negligible .
another fundamental difference from the approach of @xcite as well as from the color glass condensate approaches ( cf .
reviews @xcite ) is in the account of the important role played by the diffusion to large impact parameters . near the unitarity limit this diffusion leads to the fast rise of structure functions with energy @xcite . in our paper
we take into account the universal gribov diffusion related to randomness of radiation @xcite , but neglect the diffusion to large distances due to the running coupling constant and the diffusion to small distances related to the increase of the final state phase volume .
the latter two types of diffusion were shown in ref .
@xcite to be strongly suppressed after an account of the next - to - leading order bfkl type approximation .
there is no systematic evaluation of these effects because of the sensitivity to the restriction of the phase volume of the produced particles and related sensitivity to nlo approximations . in this paper
we heavily use results of analysis of particular preqcd model of reggeon field theory obtained in @xcite .
the major difference from these papers is that dominance of triple ladder vertex is justified in qcd . besides we found that eft predicts significant overlap between ladders near bdl . in the appendix
a we account for the exchange of the constituents between the overlapping ladders . at sufficiently small @xmath0
this leads to the formation of the color network instead of a system of ladders ( see discussion in appendix a ) .
the paper is organized in the following way . in section
2 we introduce the concept of the pqcd ladder as the quasiparticle of the eft .
we explain the hierarchy among the multiladder vertices in pqcd ( see also appendix b ) .
we show that a restriction to the triple - ladder vertex evaluated within the physical vacuum is sufficient for the theoretical description of the phenomena near the unitarity transition , and we construct the lagrangian of the eft . in sections 3 and 4
we discover and analyze the analogies with the critical phenomena . in section 5
we briefly discuss observable phenomena that follow from the solution of the effective field theory .
our results are summarized in the conclusion . in appendix
a we show that color is confined within ladders in the tube of the small transverse size .
we argue that a color network appears once the ladders overlap significantly in space . in appendix
b we present the estimates of the multiladder vertices within the physical vacuum .
the construction of an effective field theory requires the knowledge of the dominant degrees of freedom the quasiparticles and their interactions .
we assume , based on the linear pqcd calculations , that the major degrees of freedom are the pqcd color - singlet ladders with the two - gluon state in the @xmath44 channel , that we will denote the _
`` pomeron''_. ( the four gluon exchange , where each pair of gluons forms an @xmath45 representation of the su(3)@xmath46 , is included in this pomeron . )
such a definition is of common use , although the pqcd `` pomeron '' has no direct relation to the regge - pole pomeron exchange relevant for the successful description of the phenomena dominated by the nonperturbative soft qcd . in principle , the intercept of `` pomeron '' depends on @xmath8 in pqcd .
however at very small @xmath0 the @xmath8 dependence of `` pomeron '' intercept becomes weak @xcite , and the factorization of information on projectile and target is valid for a ladder because of the disappearance of the color transparency phenomenon .
the contribution of the pqcd ladders that have 4 , 6 , @xmath47 gluons in the @xmath44 channel where pairs of gluons form the color representations @xmath48 , 10 , @xmath49 , or 27 can not be topologically and analytically reduced to the contribution of the two - gluon exchange ladder . in principle
it is necessary to introduce the new varieties of quasiparticles into the hamiltonian of the eft and to account for their interactions .
however , the smallness of the running coupling constant leads to the hierarchy of the interactions especially within large @xmath50 limit evaluated within the physical vacuum .
the account of this hierarchy justifies the neglect of this kind of quasiparticles .
in particular , it is well known that the two - gluon colorless ladder has maximal @xmath51 as compared to the colored ladders .
the contributions of all quasiparticles except the ladders with the two gluon color singlet intermediate states in @xmath44 channel can be neglected even near the unitarity limit , where the amplitudes of the high - energy processes are close to the maximum permitted by the probability conservation .
when the contribution of one ladder to the amplitude becomes large it is necessary to take into account the shadowing effects .
the technology of the evaluation of the rescattering effects due to the iterations of the pqcd ladder is the adaptation to pqcd of methods of the reggeon field theory(rft ) @xcite .
the account of the shadowing effects leads to the 2 + 1 dimensional field theory where variables are the position of the leading singularity in the angular momentum plane@xmath52 and the transverse momentum @xmath53 .
the equivalent but more convenient approach to the problem considered in the paper is to use the description in terms of the variables @xmath54 -rapidity and @xmath55 impact parameter .
the effective field theory includes the `` pomeron '' loops and all variety of multi-``pomeron '' vertices .
technologically it is more convenient to find the lagrangian of the eft rather than to evaluate the particular set of diagrams , cf .
consequently , we may restrict ourselves to the brief description of the lagrangian of the eft near the unitarity limit .
the equations of the eft can be derived as the lagrange equations of motion from an effective lagrangian that contains five terms , l = l_0-l_1-l_2-l_3 -l_4 . [ 1 ] the first three terms have a straightforward interpretation in the case of noninteracting quasiparticles , while the other terms describe interactions between quasiparticles and with the virtual photon as the source .
the @xmath56 , @xmath57 and @xmath58 follow from the mellin transformation of the green function of the free quasiparticle , @xmath59^{-1}$ ] , in the plane of the complex angular momentum @xmath60 in the crossed channel .
thus l_0=(q_y p -p_y q ) , [ 2 ] where @xmath61 and @xmath62 are the quasiparticle fields .
we denote @xmath63 where @xmath54 is rapidity and @xmath64 , p is the 4-momentum of the target . the quantity @xmath65 denotes the length of the fragmentation region where there are no @xmath66 factors .
@xmath67 is the `` mass '' term , l_1=-(q^2 ) pq .
[ 3 ] this term accounts for the rise of the structure functions with energy as derived for the scattering of two dipoles within the lt approximation , f_2p(x , q^2 ) , xg_p(x , q^2)~(x_0/x)^(q^2 ) , with @xmath68 .
the intercept @xmath69 has been evaluated in the pqcd in the next - to - leading order ( for a review and references see ref .
@xcite ) , 42_s(q^2)n_c/(1 - 4(_sn_c/)^2/3 - 6.5(_sn_c/)+o(_s)^4/3 + ) , [ t1 ] where @xmath1 is the effective coupling constant and @xmath50 is the number of colors .
the dependence of @xmath70 on @xmath8 is rather weak at sufficiently small x @xcite .
note also slow convergence of pqcd effects .
the term @xmath71 describes the dependence on the collision energy of the essential impact parameters .
we assume that it has the form : l_2=_pp_bq .
[ 4 ] here @xmath55 is a two - dimensional impact parameter , and @xmath72 is the `` pomeron '' slope , @xmath73 is laplace operator in `` b '' space .
the slope @xmath72 is small within the perturbative qcd . within the approximation , that takes into account the @xmath8 evolution ,
the running of the coupling constant , and to some extent the energy - momentum conservation , the fast qcd evolution to small and large distances ( that exists beyond gribov diffusion ) is suppressed @xcite , and we can neglect it .
the gribov diffusion to large impact parameters due to randomness of gluon radiation @xcite is still there .
the parametric estimate gives : @xmath74 . at the same time , near the unitarity limit the additional mechanism ( complementary to gribov diffusion ) of increase with energy of essential impact parameters , due to the fast increase of the amplitudes with energy , begins to play an important role @xcite .
there the effective @xmath72 can not be negligible but it is difficult to evaluate it because of the sensitivity to the uncalculated nonleading order terms in the running coupling constant , etc .
since the exact form of @xmath71 is unknown at present it is chosen in the paper in the simplest form .
the evaluation of the multi - pomeron vertices near the unitarity limit ( see appendix b ) shows that within the wkb approximation , the relative contribution of the fourth and higher multi - pomeron vertices is suppressed by the powers of @xmath1 compared to the triple - ladder term . thus for the description of hard qcd phenomena it should be sufficient to restrict ourselves to the triple - ladder interaction . near the unitarity limit the dominant contribution into the triple-``pomeron '' vertex @xmath75 : l_3=pq(p+q ) , [ 11 ] is given by the component of the virtual photon wave function having transverse size @xmath76 .
really at moderately small @xmath0 soft qcd and pqcd components of the photon wave function give comparable contributions into the total cross section and cross section of the diffraction .
this is due to the conspiracy of small probability of the soft component @xmath77 and the large cross section of the soft component interaction with a target .
the probability of a hard component in the virtual photon wave function is @xmath78 but the cross section of its interaction with a hadron target is @xmath77 - the aligned jet model @xcite .
this conspiracy disappears at sufficiently small @xmath0 as the consequence of the rapid increase with the energy of the cross sections of the hard processes .
so small @xmath0 , and therefore small @xmath75 processes are dominated by hard qcd , for the review and references see ref . @xcite . then at moderately small @xmath0 the vertex @xmath75 is determined by the interplay of the soft and hard qcd dynamics .
however with the decrease of @xmath0 in the kinematics near the black disk limit the vertex @xmath75 should be dominated by the hard qcd . in the absence of a detailed calculation , we limit ourselves to the dimensional estimate of @xmath75 and neglect any possible weak dependence of the coupling @xmath75 on @xmath54 and on @xmath44 . in the lowest order in the coupling
constant , the triple-``pomeron '' vertex , is due to the interaction of ladders via one gluon loop ( see fig . [ sk ] ) , where near the black disk limit hard qcd dominates , as explained in the above discussion .
in addition the account of the running coupling constant , and sudakov form factors suppresses the contribution of almost on shell effects .
our estimate gives .
[ 14 ] here @xmath79 and it is somewhat increasing with the energy increase .
this estimate of @xmath80 differs from the one obtained within the leading order bfkl approximation far from the black disk limit .
the singularity in the dependence of the triple-``pomeron '' vertex on the momentum transfer found in ref .
@xcite reveals an important role of the soft qcd , the appearance of the effects encoded into aligned jet model and represents another challenging problem for the bfkl approximation at moderately small @xmath0 .
we follow the method of calculations of ref .
@xcite where the bfkl approximation arises as the small @xmath0 limit of pqcd series where separation of scales is made before summing them . in this case for single scale hard processes @xmath75 is dominated by hard pqcd effects . besides the nonperturbative contribution into the three `` pomeron '' vertex
is suppressed by sudakov form factors , and by the dominance of the hard regime near the black disk limit , cf . the above discussion .
the existence ( but not the properties ) of the new qcd phenomena is not sensitive to the actual value of @xmath81 which is effectively the characteristic transverse momentum of the constituents of the pqcd ladder where it splits into the two new ones . a tricky point in the evaluation of the `` multi - pomeron '' vertices within
the pqcd is the necessity to account for causality and energy - momentum conservation .
diagrams where singularities are located on the same side of the contour of integration in the energy plane ( in particular eikonal diagrams ) are suppressed by the powers of energy @xcite ) .
so we neglect the eikonal - type inelastic rescatterings since a bare particle may have one inelastic collision and any number of elastic collisions . for the interactions that rapidly increase with the energy , the requirements of the causality , the positivity of the probability for the physical processes , and the energy - momentum conservation can be hardly satisfied within such a set of diagrams @xcite .
in contrast , the contribution of rescatterings due to an inelastic diffraction into the final state with the invariant mass @xmath82 , where @xmath83 , is not too small .
this contribution dominates in the two - scale hard small-@xmath0 phenomena at @xmath84 .
we include this contribution in the scale factor of the source .
( this contribution can be interpreted as the coupling of the `` pomeron '' to secondary reggeon trajectories . ) within the approximations made in this paper the coupling of the pqcd ladder to a hadron can be treated as the interaction with a source .
the actual form of the @xmath85 term is unimportant for the most of the results obtained in this paper .
nonetheless , let us discuss the interaction of the ladders with a virtual photon as an external source in the center - of - mass frame of the reaction .
there is the coupling with the projectile and target virtual photons : l_=_0 ^ 1 d^2b ^2(z , b , q^2 ) q(z , b - b ) c_dipole(y+y ) .
[ la ] here @xmath86 is the total rapidity , and @xmath87 is the total impact parameter . in this formula
we neglect the @xmath8 dependence of the interaction in the vicinity of the black disk limit because , near the black disk limit , the color transparency phenomena and the related decomposition over twists disappear . in the above equation @xmath88
is the square of the dipole wave function of the virtual photon in the transverse parameter space , @xmath89 is the fraction of the photon momentum carried by the constituent , and @xmath90 , where @xmath54 is the rapidity .
the variable @xmath91 is the transverse distance between the constituents in the photon wave function .
the dipole wave function is given by ( for simplicity we restrict ourselves to the collisions of longitudinally polarized photons only ) _
l(b , z , q^2)^2=c(n_c_em)(qz(1-z))^2 ) k_0 ^ 2(bq ) .
[ dipole ] the numerical factor @xmath92 follows from a convention on the normalization of this wave function .
note that this function is localized in the space of relative transverse distances between the constituents as @xmath93 and in the space of rapidities @xmath54 .
since the distribution over @xmath89 is centered at @xmath94 in the case of the wave function of the longitudinally polarized photon , we can safely assume that this contribution into the source corresponds to the finite @xmath54 negligible as compared to the full rapidity span .
so we assume for certainty that @xmath95 .
the contribution of inelastic diffraction that dominates at not extremely small @xmath0 @xcite corresponds to @xmath96 .
since our interest is in @xmath97 we may skip in the analysis of eft the difference between @xmath98 and @xmath99 , although this approximation restricts region of applicability of approximations made in the paper .
altogether , we obtain the lagrangian : @xmath100 & - & c_{\rm dipole}\int \exp(-bq/2 ) q(y,\vec b-\vec b)d^2 b \delta(y+y)\nonumber\\[10pt ] & -&c_{\rm dipole}\int \exp(-bq/2 ) p(y,\vec b-\vec b)d^2b \delta(y -y ) , \nonumber \\[10pt ] \label{lag}\end{aligned}\ ] ] where @xmath101 , and @xmath102 accounts for the normalization of the virtual photon wave function .
the interaction between overlapping ladders due to the exchange by the constituents will be considered in the end of the paper .
the form of the effective lagrangian ( [ lag ] ) relevant for the single scale hard qcd phenomena near the unitarity limit is rather close to preqcd model analyzed in refs .
@xcite ( except for the form of the source term ) . in the further analysis we use the wkb solution of the lagrangian equations of motion and quantum fluctuations around them found in refs . @xcite and adjust it to describe the hard qcd phenomena .
the distinctive property of the lagrangian ( [ lag ] ) is the existence , in addition to the usual perturbative vacua p=0,q=0 , [ v0 ] , of the two new vacua : p=/ , q=0 [ v1 ] and p=0 , q=/. [ v2 ] since lagrangian of eft is nonhermitean it has no vacuum in the usual sense and the term `` vacuum '' really means the critical points of the action ( [ lag ] ) of the eft .
the action ( [ lag ] ) has actually four critical points : the three critical points ( [ v0]),([v1 ] ) and an additional critical point @xmath103 .
however the contribution of the last point to the s - matrix ( see below ) is suppressed by @xmath104 relative to three critical points ( [ v0]),([v1 ] ) . consequently the fourth critical point can be neglected .
having lagrangian ( [ lag ] ) we may deduce the lagrangian equations of motion that will be analyzed first in the classical approximation .
the system of nonlinear partial differential equations in 2 + 1 dimensions is @xmath105 dp / dy&=&\alpha'\triangle p-\mu p+\kappa ( 2pq+p^2).\nonumber\\[10pt ] \label{v4}\end{aligned}\ ] ] this system has kink solutions .
the boundary conditions in the system ( [ v4 ] ) are at @xmath106 and correspond to the vacua ( [ v0]),([v1]),([v2 ] ) .
the detailed analysis of the kink solution is possible in 1 + 1 dimensions only , where the above equations can be reduced to the ordinary differential equation . in refs . @xcite the family of kinks , characterized by a 2d velocity parameter v has been found .
these kinks interpolate between two vacua , say ( [ v1 ] ) and ( [ v2 ] ) .
the action of the kink is finite @xmath107 , where @xmath108 is the field value at the impact parameter value @xmath109 , and @xmath110 is the kink velocity . it is proportional to @xmath111 , where we used the dependence of @xmath51 and @xmath75 on @xmath50 discussed in section 2 . for the value of parameter @xmath112 we obtain critical kinks with zero action .
quantum fluctuations around these kinks are described by a positive quadratic form , cf .
@xcite .
the characteristic property of the kinks , both in unphysical 1 + 1 , and in physical 2 + 1 dimensions , is their step function form .
one of the functions p or q behaves like a step function p ( q)~(v ( y - y_0)-b - b_0 ) , [ cf ] where v is the kink velocity ( a free parameter , v=2 for critical kink ) .
the solution contains an arbitrary parameter @xmath113 that helps to understand why the physics related to the fragmentation can be hidden into the properties of the source .
the arbitrary solution depends also on @xmath114 . the value of @xmath115 is not fixed by the equations .
this is a zero mode relevant for the appearance of the `` phonons '' in quantum fluctuations .
the calculations in 2 + 1 dimensions are significantly more difficult .
however one can easily prove the existence of the kinks with the finite action that tunnel between the vacua ( [ v0 ] ) and ( [ v1 ] ) .
we need to find a solution of eq .
( [ v4 ] ) with boundary conditions @xmath116 y&\rightarrow & -\infty,\,\ , q , p\rightarrow 0.\nonumber\\[10pt ] \label{v4a}\end{aligned}\ ] ] then for @xmath117 it is legitimate to neglect by nonlinear interaction and obtain the usual diffusion equations : @xmath118 dp / dy&=&\alpha'\triangle p-\mu p.\nonumber\\[10pt ] \label{v6}\end{aligned}\ ] ] the general solution , say for q , is given by q(y , b)=(y)(-((b - b)^2/4_py ) ) f(b)d^2b. [ v7 ] here @xmath119 where c is a numerical factor . the boundary condition for
p will be @xmath120 for @xmath121 . then near @xmath122 , we may write @xmath123 , where z is small . the linearized equation for z
is dz / dy=z - z+2q [ z ] the system of equations ( [ v4 ] ) imposes the condition @xmath124 for @xmath121 , since @xmath125 , and @xmath126 constant .
thus for each value of the impact parameter @xmath55 the function q has a maximum .
it follows from eq .
( [ z ] ) that when z is small ( and y is approaching @xmath86 ) , the field p decreases sharply near @xmath86 , so it can be approximated by the step function , while @xmath127 reaches its maximum at @xmath128 and then goes to zero .
it is easy to see that the action of the kink is finite and proportional to @xmath129 in some power .
this solution is similar to the one found in the 1 + 1 dimensional theory .
however there is no analytical expression for the solution giving critical kinks with zero action as well as a full classification of kinks .
so in 2 + 1 dimensions we rely on the results of the numerical simulations made in refs .
@xcite , who found the same properties of kinks for the 2 + 1 dimensional theory as for 1 + 1 dimensional one .
the existence of the step like solutions made it possible to calculate s - matrix for the case of the sources given by eq.([dipole ] ) .
indeed , the classical equations of motion for this case have the form @xmath130 dq / dy&=&\alpha'\triangle q-\mu q+\kappa ( 2pq+q^2)+ c_{\rm dipole}\int d^2b \exp(-bq/2)p(y , b - b ) \delta ( y+y).\nonumber\\[10pt ] \label{v41}\end{aligned}\ ] ] since coupling to dipole for the field q is much smaller than @xmath131 these equations are close to that analyzed in ref .
the field p is small for @xmath132 , and it is the solution of the usual diffusion equation with the boundary condition that follows from the first of eqs .
( [ v41 ] ) , times @xmath133.the field p starts to rise exponentially for @xmath134 .
at this point one can not however neglect the nonlinear terms . since for @xmath121 @xmath135 , we can linearize eqs .
( [ v4 ] ) as @xmath136 .
then for z one gets equation dz / dy=z+z .
[ v8 ] the equation for z is the same equation as above , except that the mass term changes sign , and it is small up to @xmath134 , when it blows up , leading to a step like decrease of p. the point where solutions for p blows up can be estimated from the equation q(b , y)(y - b^2/(r^2 + 4_py ) ) .
[ v10 ] the transition occurs near the point where @xmath137 , i.e. ( r^2 + 4y)y~b^2 .
[ v11 ] here @xmath138 is the transverse scale in the problem .
asymptotically p(b , y)=(/)(2y+-b ) , [ as ] where @xmath139 is a phase shift determined by boundary conditions .
we see that in the classical approximation the equations for p and q effectively decouple .
moreover it can be proved that in the classical limit it is enough to solve the decoupled equations with suitably chosen boundary conditions .
the classical solution leads to the black disk limit , since in the classical theory the s - matrix is given by s~(- d^2 b g(b)p(b , y ) ) , [ rse ] cf .
@xcite . here
@xmath140 characterizes the photon and it is concentrated near @xmath141 . since @xmath142 has a step function form , and g(b ) is concentrated near b , in the classical theory s - matrix is 1 outside the black disk and zero inside ( total absorption ) .
the knowledge of the family of the kink solutions permits the semiclassical quantization of the theory and the calculation of the s - matrix .
the distinctive feature of the semiclassical approximation is the existence of the `` critical '' kinks , with zero action , in particular , the kinks with both the energy and the classical momentum being zero .
classical contributions of these kinks into wave function are not exponentially suppressed .
( energy and momentum of kinks are defined as components of @xmath143 and @xmath144 , where the energy momentum tensor of the eft is found via the nether theorem ) .
the remarkable property of the quantum fluctuations around critical kinks is the existence of the zero modes-``phonons '' in the eft , that are characterized by the linear dispersion : e = i2k=2 p_cl , [ c1 ] where @xmath145 is the total classical momentum derived from the eft action .
we present here plausible reasoning for the existence of such modes , following refs.@xcite .
let us begin from the analysis of 1 + 1 dimensional model .
in this case kinks are f(x - vy ) solutions of the ordinary differential equations .
quantum fluctuations around kinks can be calculated using the standard procedure @xcite .
the zero mode ( [ c1 ] ) leads to the gapless spectrum of excitations ( `` phonons '' ) .
the `` phonon '' wave function satisfies the equation : h_rk>_=i2kk > _ [ c2 ] and p_rk>_=kk > _ , [ c3 ] where @xmath146 and @xmath147 are the total hamiltonian and momentum of the quantized eft ( @xmath148 and @xmath149 , where @xmath150 were calculated via the nether theorem ) .
the corresponding full set of wave functions can be found explicitly : k>_=da exp(-ika)((-(/ ) ( -a)q ( ) ) _ 0>. [ c4 ] here @xmath151 , and @xmath152 is a solution of a classical equation of motion for a kink , @xmath153 is wave function of a perturbative vacuum .
these states propagate in one direction in time @xmath154 to @xmath120 at @xmath155 and in the opposite directions in impact parameter space .
it is straightforward to calculate the quasiparticle correlation function : d(b , y)=<_0p(0,0)(ipb)(-hy ) q(0,0)_0 > , [ cf ] using the set of states ( [ c4 ] ) together with a perturbative and nonperturbative vacua , and get a black disk limit as the asymptotic answer .
let us emphasize the key role of a linear spectrum .
indeed , for the linear spectrum , as it was shown in ref .
@xcite , d(b , y)= ( ikb ) ( iv_0ky)<_0p(0)k > < kq(0)_0>. [ d1 ] it is possible to prove that the above matrix elements are equal to = - , [ d2 ] where @xmath156 defines contour of integration around singularities
. then we obtain : d(b , y)=(/)^2(-b+2 y ) .
[ d3 ] in order to connect this correlator with the amplitudes we have to normalize it properly @xcite : g(b , y)=d(b , y ) .
[ ren ] then g(k , j)= [ d4 ] here @xmath157 is a critical kink speed .
the scattering amplitude is ( @xcite , chapter 16 ) a(s , t)=s^i_-i ( ) g(,k^2)(i+ ( ) ) [ ast ] here @xmath158 .
in addition , there exists a band of low lying states with a dispersion relation @xmath159 , and a band of states with a gap @xmath160 .
it can be argued that the first set of states corresponds to the unshifted quasiparticle fluctuations around the perturbative vacuum @xmath161 in the presence of a kink and viewed from the reference frame moving with a critical speed @xmath162 .
the higher modes can be interpreted as the collective fluctuations of a condensate of ladders , i.e. the quasiparticles interacting with the kinks .
it is easy to prove that these modes do not influence the expanding disk solution asymptotically , although can be important outside the disk , and near the transition to a black disk regime .
the numerical analysis of the spectrum of the quantum fluctuations around the kinks has been performed in refs .
moreover , it was proved on the discrete lattice that the field theory can be continuously connected with the ising model in transverse magnetic field .
two types of collective excitations were found the zero modes with a spectrum @xmath163 and solutions with a gap and spectrum @xmath164 , similar to the ones found in the 1 + 1 dimensional model in the continuum . evidently , only solutions with linear spectrum are relevant for the asymptotic behavior of the high energy processes . to summarize ,
the quasiclassical solution of eft has following distinctive features : \a ) eft has three degenerated `` vacua '' ( [ v0 ] ) , ( [ v1 ] ) , ( [ v2 ] ) ( the word `` vacua '' is in the brackets , since eft is the theory with nonhermitian hamiltonian , and actually means critical points of the action with zero value of the hamiltonian ) .
the true wave function of the physical ground state is a linear combination of these three vacua , and the hamiltonian is diagonalized by `` critical '' kinks with zero action .
if the initial state is a perturbative one interacting with a source , it evolves in rapidity and becomes a condensate of quasiparticles ( ladders ) .
( the term `` ground state '' means here that all states in the relevant physical sector the theory are the excitations of this state ) .
\b ) the s - matrix is given by the functional integral s(b , y , g , f)=dpdq ( -l(b , y , g , f ) ) , [ v11a ] where the corresponding action is calculated via the classical kink solution described above , but includes now the source terms the couplings with the virtual photons . the main contribution to the s - matrix comes from the quantum fluctuations around the critical kinks with the zero action .
the spectrum of the excitations begins from the massless excitations- `` phonons '' .
the contribution of the kinks into the two- quasiparticle green functions that determines the cross - section is : g(b , y)=(b^2 - 4_py^2 ) [ v12 ] for large y , @xmath165 .
this green function defined by eq.([ren ] ) is related in a usual way to a scattering amplitude by the mellin transformation , ( see e.g. ref .
@xcite , chapter 16 and eq .
( [ ast ] ) ) .
we denote the perturbative vacuum ( [ v0 ] ) as @xmath153 .
\c ) the obtained state is described by the asymptotic wave function ( y)=(-d^2bq(b , y ) ( b^2 - 4y^2))_0>. [ slon3 ] here @xmath166 is a perturbative vacuum .
to the extent that the correlations may be neglected the asymptotic vector ( [ slon3 ] ) is a coherent state @xcite and the s - matrix is given by s(b , y)=(-(2y - b ) ) , [ slon31 ] where we explore that the target is localized near the impact parameter @xmath141 . the s - matrix as given by eq .
( [ slon31 ] ) is 1 inside the black disk and suppressed exponentially as @xmath167 , outside .
this is just the bdl behavior .
note the nonperturbative structure of eq .
( [ v12 ] ) , and that the wave function @xmath168 can not be obtained by decomposition over powers of @xmath1 .
\d ) in other words in the limit of the infinite energies @xmath155 the produced state corresponds to a bose - einstein condensate of ladders in the entire space .
however , for finite energies the solution is the black disk of radius @xmath169 .
the equation ( [ slon3 ] ) gives the exact form of the wave function of the bose - einstein condensate of ladders as a function of rapidity y. \e ) the crucial reason why both quantum and classical approximations lead to the black disk behavior is the existence of the `` phonon '' zero mode in the quantum kink hamiltonian , that has been proved on the lattice , and in the 1 + 1 dimensional case .
moreover , the analysis of refs .
@xcite shows that the contribution of the excited states that are different from the `` phonons '' is relevant only in the vicinity of the transition point .
there are two types of states : collective fluctuations near the perturbative vacuum interacting with quantum kinks and the quantum fluctuations of the kink condensate .
none of them influences the asymptotic behavior of the black disk .
some properties of the classical solutions of the effective theory can be understood directly in qcd .
a kink produces action proportional to @xmath170 .
the dependence of the s - matrix ( [ slon31 ] ) , of the critical kink action ( understood as the limit of a family of kinks with nonzero action ) on the coupling constant @xmath1 and existence of `` phonon '' show that this is a novel nonperturbative qcd phenomenon .
let us note here that the 2d translational invariance is broken in the theory at finite rapidity y due to the existence of black disk behavior .
mathematically it reveals itself in the existence of a translational zero mode for the kink solution .
this phenomenon resembles a spontaneous symmetry breaking since the system , after being at moderate rapidities in the perturbative vacuum chooses in the process of space - time evolution the state which is a definite combination of the vacuum states of the system . whether this phenomenon corresponds to the conventional spontaneous symmetry breaking
is unclear at present but the existence of `` phonons '' looks suggestive .
moreover it is easy to show that in the approximation where @xmath171 eft can be mapped by substitution of variables into the quantum mechanics with hermitian double well potential with the two nonperturbative critical points of the action ( [ lag ] ) corresponding to two minima of higgs type potential . in this case
two nonperturbative vacuums of section 3 are real minima of the action and tunneling transitions are relevant for the change of symmetry .
the estimate @xmath172 where transition to the regime of bose - einstein condensation of ladders follows from the inequality @xmath173 where @xmath174 can be found from the requirement of conservation of probability , of unitarity of the s matrix cf . ref .
( additional factor @xmath175 accounts larger energies needed for the applicability of triple `` reggeon''=ladders limit and all ladders should be near unitarity limit . ) assuming @xmath176 cf .
@xcite we obtain @xmath177 . at these rapidities
the bose - einstein condensation of the ladders starts .
the characteristic form of the kink is the step - function in the y space , with the width of the order @xmath178 .
the coherent length relevant for the evolution of the kink is enhanced due to the large lorentz slowing down gamma - factor as t_i ~e / q^2(1/)/q10 ^ 2/q [ v^*1 ] in other words , for sufficiently low x we have @xmath179 , where @xmath180 is coherence length @xmath181 .
this rapid transition to the black disk regime can be called the `` color inflation '' : one ladder due to the tunneling transition blows up during time @xmath182 and creates an entire region of space filled with the gluon ladders . during time @xmath182 approximately @xmath183 ladders are created , where @xmath184 is a black disk radius for a given rapidity @xmath86 . it is easy to evaluate the density of ladders in the coordinate space by solving discussed above diffusion equations ( cf .
similar analysis in ref .
@xcite ) : ( y)>=(-(/)^r(y)_0 d^2b q(b , y ) ) .
[ vec1 ] the conjugate state is = 1 [ dddeer ] consequently , the transverse distance between the ladders is l_t~/~_s /[v * ] on the other hand , the characteristic scale @xmath185 of a ladder in the transverse parameter space is determined by the coefficient in the effective lagrangian in front of the kinetic term , that is equal to @xmath186 , hence d_t^2 ~n_c _ s/^2
[ v * * ] consequently , l_t^2/d_t^2_s/ n_c 1 .
[ v * * * ] it follows from the above estimate that the pqcd ladders overlap significantly . but overlapping ladders can exchange the quarks and the gluons because in the perturbative qcd there are no barriers between the ladders . the resulting color network , as we shall call this object , is macroscopical in the longitudinal direction .
the distinctive features of the color network resemble the quark - gluon plasma . understanding the actual longitudinal structure of the system is not possible within the effective field theory . indeed , even after the phase transition
, the ladders continue to grow till the color network achieves the longitudinal length @xmath187 .
in our analysis we used the fact the qcd ladders can be considered as the effective degrees of freedom up to the scales when they significantly overlap .
this overlap is controlled by the density of ladders since in the perturbative qcd ( see appendix a ) there are no long range forces operating between the ladders .
we predict a variety of the new nonperturbative qcd hard phenomena for the case of the collision of the two small dipoles near the black disk limit .
their very existence shows that transition to the black disk limit is a kind of a critical phenomenon .
the phenomenon of color inflation may significantly change the physics of hard processes at sufficiently small x by softening the parton distribution over the longitudinal momenta and will reveal itself as the threshold like increase of multiplicity .
( we postpone quantitative analysis of such phenomena to the next publications ) . comparatively clean way to identify the onset of the new regime would be to measure mueller - navelet process @xcite : @xmath188 where the distance in the rapidity between the high @xmath189 jets is large .
the expected behavior is the following : initial fast increase of cross section with @xmath54 predicted by the pqcd should change to the fast decrease at larger @xmath54 because of the color inflation , i.e. disappearance of the long range correlations in the rapidity ( coordinate ) space near the black disk limit due to the creation of the color network .
we argue that the absence of the long range forces between the colorless ladders as the consequence of the color screening phenomenon ( see appendix a ) justifies the neglect of the exchanges of the constituents between the colorless ladders in the first approximation .
the colorless ladders should be a dominant degree of freedom even near the black disk limit . as a result it is possible to build the effective field theory where ladders are quasiparticles .
we showed that at sufficiently small x the form of the hamiltonian of this eft is dictated for the single scale hard processes in qcd by the smallness of running coupling constant . moreover smallness of the effective triple ladder vertex : @xmath190 justifies the applicability of the semiclassical approximation .
we show that the effective field theory is solvable within the wkb approximation and leads to the black disk behavior and other qcd phenomena .
the transition to the black disk regime within the effective field theory in one scale hard processes is a kind of a critical phenomena . there
exist classical hard qcd fields relevant for the transition from false to physical vacuum -kinks in the rapidity - impact parameter space .
the transition due to general eft kink is suppressed as @xmath191 , because @xmath192 ( see appendix b ) .
the critical kinks that correspond to actual transitions between vacua are the limiting cases of the families of these noncritical kinks .
the transition to the black disk limit is of the inflationary type : in the case of collisions of two small dipoles the time scale @xmath193 of the transition is significantly smaller than the time needed for the formation of the perturbative ladder @xmath194 .
the nonperturbative transition produces ladders that strongly overlap in the impact parameter space .
due to the exchange of the constituents between the overlapping ladders the system of ladders becomes the color network that resembles the lengthy but narrow pencil .
the difference between the pencil and the jet is the softer distribution of hadrons over transverse and longitudinal momenta .
however to study this color network , we need to be able to describe the bdl transitions ( the kinks ) directly in terms of the qcd language , the goal that is yet not achieved .
to summarize , we were able to show that qcd leads to solvable in quasiclassical approximation effective effective field theory , and this effective field theory has a transition to the black disk limit , that in the qcd language is a color network . the most challenging problems for the future work will be to calculate both the kinks and the `` phonons '' of the eft and to study the condensation of ladders directly in terms of qcd ( quark , gluon ) degrees of freedom and to understand the role of the found qcd phenomena in two scale processes .
it has been known long ago @xcite that ladders are the dominant degrees of freedom in the theoretical description of the high energy processes . in the leading twist approximation
this assumption has been proved in pqcd . at the same time near the black disk limit due to the existence of the triple ladder coupling
the multiladder configurations become important and may overlap .
we explain here that there are no long range forces between the pqcd ladders and quarks and gluons are confined within the ladders , till the latter start to overlap inside the space - time .
indeed , consider the dglap ladder .
the vector potential created by this ladder can be calculated as a^a_(x)=d^4y d^ab_ret ( x - y)j^b_(y ) [ a1 ] here @xmath195 is the retarded gluon propagator in a light - cone gauge , while j is the matrix element of a current for gluon emission by a ladder .
the equation ( [ a1 ] ) can be rewritten in the momentum space as a^a_(x)=d^4s ( ixs ) d^ab_ret , ( s ) j^b_(s ) [ a2 ] the retarded propagator in the light cone gauge has the form : d_ret(s)=(g_-(s^^+s^^)/(s ) ) .
[ a3 ] as usual we take into account for the cancellation of the most singular term in the propagator that follows from the ward identities .
@xmath196 is the current of a soft gluon emission by the ladder and @xmath197 is a light cone vector .
j_(s)=^i = n_i=1 q^_i/(sq_i ) [ a4 ] where the sum is taken over all external lines of a cut ladder .
the contributions from the internal lines cancel at least in the leading order @xcite .
consider integral over s , i.e. @xmath198 . calculating the integral over the residues , we see that the integral is controlled by @xmath199 pole in the propagator . potentially dangerous terms like @xmath200 are far from the mass shell and have no singularity .
thus integral for the vector potential is given by the pole @xmath201 .
the emitted gluon is always on the mass shell and transverse . in other words
no long range forces exist within the dglap approximation .
the same arguments can be applied to bfkl ladder .
the integral over s is also determined by @xmath199 pole , and leads to transverse gluons ( ref .
@xcite ) .
we conclude that the leading order ladders do not create perturbatively long - range fields in the leading twist approximation .
generalization of this statement to the next - to - leading order should be straightforward .
the conventional strategy can be used to evaluate parameters of effective hamiltonian of eft : substitution of variables in the path integral from quarks and gluons to ladders .
however such calculation is too cumbersome at present .
so we restrict ourselves by the qualitative evaluation of multi - ladder vertices near the unitarity limit in the lowest order of pqcd .
the most difficult point is to take into account for causality , i.e. location of singularities in the complex plane of energies in respect to the contour integration .
it has been shown by s.mandelstam @xcite and in the more straightforward way by v.gribov @xcite that the diagrams having no third spectral function @xmath202 in mandelstam representation for the scattering amplitude give no contribution into the leading power of energy .
the transparent interpretation of this result is that bare particle may experience only one inelastic collision within the semiclassical approximation .
= g_3p , [ b1 ] cf .
recent discussion in ref .
@xcite . to evaluate dependence on @xmath50
it is useful to represent vectors in the color space in terms of color spinors and then to find the disentaglment of color contours .
the account of causality ( i.e. of the fact that the bare particle may have only one inelastic collision and any number of elastic ones ) shows that the lowest order diagram for the four ladder vertex corresponds to the attachments of 4 ladders to the 2 gluon loops .
d.gross and f.wilczek , 1,phys .
d8 ( 1973 ) 3633 ; phys .
d9 ( 1974 ) 980 .
dokshitser , sov .
jetp 46 ( 1977 ) h. abramowicz and a. caldwell , rev .
phys . , 71 ( 1999 ) 1275 .
s. brodsky , j.f .
gunion , l. frankfurt , a.h .
mueller , and m. strikman , phys .
d50 ( 1994 ) 3134 , p.c .
collins , l. frankfurt , m .
strikman , phys .
d56 ( 1997 ) 2982 .
h. abramowicz , l. frankfurt , and m. strikman , surv .
high energy physics , 11 ( 1997 ) 51 .
v. guzey , l. frankfurt , m. strikman , m .
mcdermott , phys.rev.lett . , 87 ( 2001 ) 192301 .
e. iancu , hep - ph/0510264 , hep - ph/0510265 .
v. guzey , l. frankfurt , m. strikman , m.mcdermott , eur.j . of physics ,
c16 ( 2000 ) 641 .
e. kuraev , v. fadin , l. lipatov , sov .
jep , 44 ( 1976 ) 443 ; 45 ( 1977 ) 199 . i. balitsky and l. lipatov , sov
. j. nucl .
phys . , 28 ( 1978 ) 822 .
v. fadin and l. lipatov , phys .
lett . , b429 ( 1998 ) 127 .
m. ciafaloni , d. colferai , g.p .
salam and a. stasto , phys .
b541 ( 2002 ) 314 , m. ciafaloni , d. colferai , g.p .
salam , phys .
d60 ( 1999 ) 114036 .
g.p.salam , asymptotics and preasymptotics at small x , hep - ph/0501097 , talk at erice qcd workshop ,
september 2004 .
l. mclerran and r. venugopalan , physical review , d49 ( 1994 ) , 2233,3352 ; d50 ( 1994 ) 2225 ; d59 ( 1999 ) 094002 ; see also l. mclerran , surveys high energy phys . , 18 ( 2003 ) 101 ; nucl .
a702 ( 2002 ) 49 , for the latest reviews on the subject .
j. jalalian - marian , a. kovner , l. mclerran , h. weigert , phys .
rev . , d55 , ( 1997 ) 5414 ; j. jalalian - marian , a .
kovner , h .
weigert , phys .
d59 ( 1999 ) 014015 ; j. jalalian - marian , a .
kovner , a. leonidov , h .
weigert , phys . rev .
d59 ( 1999 ) 014014 ; nucl .
b504 , 415 ; a. kovner , m. giulherme milano , h. weigert , phys . rev .
d62 , ( 2000 ) 114005 .
a. mueller and a.shoshi , small x physics near the saturation regime , hep - ph/0405205 , submitted to 39 recontres de moriond in qcd .
l.mclerran,nucl.phys . , a752 ( 2005 ) 355 ; a. h. mueller , nucl .
b415 ( 1994 ) 373 ; yu . kovchegov , phys .
d60 ( 1999 ) 034008 ; i. balitsky , nucl.phys . , b463 ( 1996 ) 99 .
a. kovner and u. wiedemann , phys.lett . , b551 ( 2003 ) 311 ; phys .
d66 ( 2002 ) 034031 .
e. ferreiro , e. iancou , k. itakura , l. mclerran , nucl .
phys . , a710 ( 2002 ) 373 ; t. ikeda , l. mclerran , nucl .
phys . , a756 ( 2005 ) 385 .
v. gribov and a. migdal , sov .
jetp , 78 ( 1969 ) 784;sov .
j. of nucl .
8 ( 1969 ) 583 .
l. frankfurt , m.strikman , m. zhalov , phys.lett . , b537 ( 2004 ) 51 .
islam , r.j .
luddy , a.v .
prokudin , mod.phys.lett .
, a18 ( 2003 ) 743 .
block , f. halzen , phys.rev.d72:036006,2005 , erratum - ibid.d72:039902,2005 .
b. blok and l. frankfurt , phys .
lett . , b630 ( 2005 ) 49 .
b.blok and l.frankfurt,phys.rev.d70 ( 2004 ) 094003 .
l. landau and e. lifshits , statistical mehanics , oxford , pergamon press , 1980 .
a. gerasimov and s. shatashvili , jhep 0010:034,2000 .
bjorken , j.phys.colloq.34:137-162,1973 .
l. frankfurt and m. strikman , phys .
reports , 160 ( 1988 ) 235 .
l. frankfurt , m. strikman and c. weiss , ann .
rev.of particle and nucl.science 2005 , hep - ph/0507286 .
h. cheng and t.t .
wu , expanding protons : scattering at high energies , mit press , 1987 .
y. dokshitser , d. diakonov and s. troyan , phys . rept .
58 ( 1980 ) 269 .
m. levin , u.maor , phys .
d49 ( 1994 ) 4321 ; phys .
b347 ( 1995 ) 424 .
j. jalalian - marian and yu .
kovchegov , progr .
nucl . part .
physics , 56 ( 2006 ) 104 .
l. lipatov , phys .
repts . , 286 ( 1997 ) 131
. v. n. gribov , space - time description of hadron interactions at high energies , hep - ph/0006158 .
v.n.gribov , theory of complex angular momenta : gribov lectures on theoretical physics , cambridge univ . press , 2003 .
v. fadin , v. t. kim , l. lipatov , g. pivovarov , hep - ph/0207296 to appear in the proceedings of 35th annual winter school on nuclear and particle physics , repino , russia , 19 - 25 feb 2001 .
v. guzey , l. frankfurt , m. strikman , j. phys.,g27 ( 2001 ) r23 - 146 .
j. bartels , m.ryskin and g.p.vacca eur.phys.j.c27,101,2003 .
mandelstam , nuovo cim .
, 30 ( 1963 ) 1148 .
d. amati , m. ciafaloni , le bellac , g. marchesini , nucl .
phys . b112 ( 1976 ) 107 .
d. amati , m. ciafaloni , g. marchesini , g.parisi , nucl .
b114 ( 1976 ) 483 . v. alessandrini , d. amati , m. ciafaloni ,
b130 1977 ) 429 . v. alessandrini , d. amati , r. jengo , nucl phys , b108 ( 1976 ) 425 .
m. ciafoloni and g. marchesini , nucl .
phys . , b105 ( 1976 ) 113 ; b109 ( 1976 ) 261 .
g. thooft , phys .
rev . , d14 ( 1976 ) 3432 .
mueller and h. navellet , nucl .
b282 ( 1987 ) 727 .
v.n.gribov , nucl .
phys . b106 ( 1976 ) 189 .
m. ciafaloni , nucl.phys.b296 ( 1988 ) 49 ; s. catani s. , fiorani f. , marchesini g. , phys.lett.b 234 ( 1990 ) 339 .
kovchegov , hep - ph/0503038 ; yu .
kovchegov , d.h .
rischke , phys .
c56 ( 1997 ) 1084 . | we argue that at sufficiently small bjorken @xmath0 where pqcd amplitudes rapidly increase with energy and violate probability conservation the shadowing effects in the single - scale small @xmath0 hard qcd processes can be described by an effective quantum field theory of interacting quasiparticles perturbative qcd ladders .
we find , within the wkb approximation , that the smallness of the qcd coupling constant ensures the hierarchy among many - quasiparticle interactions evaluated within the physical vacuum and in particular , the dominance in the lagrangian of the triple quasiparticle interaction .
it is explained that the effective field theory considered near the perturbative qcd vacuum contains a tachyon relevant for the divergency of the perturbative qcd series at sufficiently small @xmath0 .
we solve the equations of motion of the effective field theory within the wkb approximation and find the physical vacuum and the transitions between the false ( perturbative ) and physical vacua .
classical solutions which dominate transitions between the false and physical vacua are kinks that can not be decomposed into perturbative series over the powers of @xmath1 .
these kinks lead to color inflation and the bose - einstein condensation of quasiparticles .
the account of the quantum fluctuations around the wkb solution reveals the appearance of the `` massless '' particles `` phonons '' .
it is explained that `` phonons '' are relevant for the black disk behavior of cross sections of small @xmath0 processes .
the bose - einstein condensation of the ladders produces a color network occupying a `` macroscopic '' longitudinal volume .
we discuss briefly the possible detection of new qcd effects .
we outline albeit briefly the relationship between the small @xmath0 hard qcd processes and the coherent critical phenomena . |
let @xmath2 denote the unit disk in the riemann sphere @xmath3 and @xmath4 , where @xmath5 is the closure of @xmath2 .
by @xmath6 we denote the class of all holomorphic univalent functions in @xmath2 normalized by @xmath7 , @xmath8 , and by @xmath9 , the class of all univalent meromorphic functions in @xmath10 normalized by @xmath11 , @xmath12 , @xmath13 stands for all functions from @xmath9 with @xmath14 .
these classes have been one of the principal subjects of research in complex analysis for a long time .
the most inquisitive problem for the class @xmath6 posed by l. bieberbach in 1916 @xcite finally has been solved in 1984 by l. de branges @xcite who proved that @xmath15 for any @xmath16 and the equality is attained only for the koebe function @xmath17 , @xmath18 .
the main tool of the proof turned out to be a parametric representation of a function from @xmath6 by the lwner homotopic deformation of the identity map given by the _ lwner differential equation_. the parametric method emerged almost 80 years ago in the celebrated paper by k. lwner @xcite who studied a one - parameter semigroup of conformal one - slit maps of @xmath2 coming then at an evolution equation called after him .
his main achievement was an infinitesimal description of a semi - flow of such maps by the schwarz kernel that led him to the lwner equation .
this crucial result was generalized , then , in several ways .
attempts have been made to derive an equation that allowed to describe a representation of the whole class @xmath6 .
nowadays , it is rather difficult to follow the correct history line of the development of the parametric method because in the middle of the 20-th century a number of works dedicated to this general equation appeared independently . in particular , p. p. kufarev @xcite studied a one - parameter family of domains @xmath19 , and regular functions @xmath20 defined in @xmath19 .
he proved differentiability of @xmath20 with respect to @xmath21 for @xmath22 from the carathodory kernel @xmath23 of @xmath19 , and derived a generalization of the lwner equation .
pommerenke @xcite proposed to consider subordination chains of domains that led him to a general evolution equation .
we mention here also papers by v. gutlyanski @xcite and v. goryainov @xcite in this direction .
one can learn more about this method in monographs @xcite ( see also the references therein ) .
let us draw reader s attention to goryainov s approach @xcite who suggested to use a method of semigroups to derive several other parametric representations of classes of analytic maps and to apply them to study dynamics of stochastic branching processes .
this approach is based on the study of one - parameter semi - flows on semigroups of conformal maps , their infinitesimal descriptions , and evolution equations ( see also @xcite ) . in 1959
shah dao - shing @xcite suggested a parametric method for quasiconformal automorphisms of @xmath2 . in another form
this method appeared in the paper by f. gehring and e. reich @xcite , and then , in @xcite .
later , cheng qi he @xcite obtained an analogous equation for classes of quasiconformally extendable univalent functions ( to be more precise , in terms of inverse functions ) . unlike the parametric method for conformal maps , its analogue for quasiconformal maps did not receive so much attention .
several attempts have been launched to specialize the lwner - kufarev equation to obtain conformal maps that admit quasiconformal extensions ( see @xcite ) . surprisingly , an analogous equation appeared in fluid dynamics in the study of plane free boundary problems , where the time dependence of the phase domain @xmath19 in a hele - shaw cell was described by a one - parameter chain of univalent maps satisfying an equation that now is known as the polubarinova - galin equation .
it appeared in the pioneering works by p. ya .
polubarinova - kochina @xcite and l. a. galin @xcite ( see surveys @xcite , @xcite ) .
in contrast to the classical lwner - kufarev equation the latter is a non - linear ( even non - quasilinear ) integro - differential equation and many elegant properties of the lwner - kufarev equation are less clear for polubarinova - galin s one .
a typical feature of the hele - shaw flow is that starting with a simply connected phase domain @xmath24 with a smooth boundary possible cusps may be developed during time evolution .
they are caused by vanishing boundary derivatives as well as by topology change .
the principal goal of our paper is to study evolution equations for conformal maps with quasiconformal extensions . in particular ,
we are interested in maps smoothly extendable onto the unit circle .
our approach is based on the study of evolutions on the universal teichmller space @xmath0 and on the manifold @xmath25 embedded into @xmath0 .
another question we are interested in is how a hele - shaw flow is seen on the universal teichmller space as on a general parametric space .
let us consider a _ subordination chain _ of simply connected hyperbolic domains @xmath19 in the riemann sphere @xmath26 , which is defined for @xmath27 .
this means that @xmath28 whenever @xmath29 .
we suppose that all @xmath19 are unbounded and @xmath30 for all @xmath21 . by the riemann mapping theorem
we construct a subordination chain of mappings @xmath31 , @xmath32 , where each function @xmath33 is a meromorphic univalent map of @xmath10 onto @xmath19 for every fixed @xmath21 .
pommerenke @xcite first introduced such chains in order to generalize lwner s equation .
his result says that given a subordination chain of domains @xmath19 defined for @xmath34 with a differentiable real - valued coefficient @xmath35 ( in particular , @xmath36 ) , there exists an analytic regular function @xmath37 such that @xmath38 in @xmath32 and @xmath39 for almost all @xmath34 .
the coefficient @xmath40 is the conformal radius of @xmath19 .
this equation now - a - days is known as the lwner - kufarev equation due to the contribution by k. lwner @xcite and p. p. kufarev @xcite .
we consider two main * questions : * * if @xmath41 is a quasicircle , what is @xmath42 ? * if @xmath41 is a smooth jordan curve , what is @xmath42 ?
we draw reader s attention to the case of smooth boundaries and their connection to free boundary problems of fluid dynamics . in 1898
h. s. hele - shaw @xcite proposed his famous cell that was a device for investigating a flow of viscous fluid in a narrow gap between two parallel plates .
the dimensionless model of a moving viscous incompressible fluid in the hele - shaw cell is described by a potential flow with the velocity field @xmath43 .
the pressure @xmath44 is the potential for the fluid velocity @xmath45 where @xmath46 is the cell gap and @xmath47 is the viscosity of the fluid ( see , e.g. @xcite ) . through the similarity in the governing equations ,
hele - shaw flows can be used to study models of saturated flows in porous media governed by darcy s law . over the years various particular cases of such a flow
have been considered .
different driving mechanisms were employed , such as surface tension or external forces ( suction , injection ) .
we mention here a 600-paper bibliography of free and moving boundary problems for hele - shaw and stokes flows since 1898 up to 1998 collected by k. a. gillow and s. d. howison @xcite .
since the work by hele - shaw several principal steps have been made . among them we distinguish the papers by p. ya .
polubarinova - kochina @xcite and l. a. galin @xcite who suggested in 1945 a complex variable approach that now is one of the basic tools for investigating the hele - shaw evolution .
let us consider the flow of a viscous fluid in a plane hele - shaw cell under injection through a unique well which is placed at infinity .
suppose that at the initial moment the phase domain @xmath48 occupied by the fluid is simply connected and bounded by a smooth analytic curve @xmath49 .
this model can be thought of as a receding air bubble in a viscous flow .
the evolution of the phase domains @xmath19 is described by an auxiliary conformal mapping @xmath31 of @xmath10 onto @xmath19 , @xmath50 , normalized by @xmath51 @xmath52 .
here we denote the derivatives by @xmath53 , @xmath54 , and @xmath21 is the time parameter .
this mapping satisfies the equation @xmath55 = -1,\quad \zeta = e^{i\theta } , \label{pg}\ ] ] under suitable rescaling .
l. a. galin @xcite , p. ya .
polubarinova - kochina @xcite first derived the equation ( [ pg ] ) and stimulated deep investigation in complex variable approach to free boundary problems ( see , e.g. , @xcite and the references therein ) . from ( [ pg ] ) one can derive a lwner - kufarev type equation by the schwarz - poisson formula : @xmath56 where @xmath32 .
the equation ( [ pg ] ) is equivalent to the kinematic condition on the free boundary and , in particular , implies that the phase domains @xmath19 form a subordination chain . in contrary to the classical lwner - kufarev equation ( [ lkp ] ) the equation ( [ pgl ] ) even is not quasilinear and the problem of the short - time existence and uniqueness of the solution is much more difficult .
first it was proved by yu .
p. vinogradov , p. p. kufarev @xcite in 1948 and later in 1993 by m. reissig , l. von wolfersdorf @xcite .
in fact , starting with a smooth domain @xmath48 the solution to ( [ pg ] ) exists and unique locally in time .
it is known that the domains @xmath19 remain to have smooth ( even analytic ) boundaries up to the time @xmath57 when possible cusps are developed or the domain fails to be simply connected .
this means , in particular , that @xmath19 fails to be a quasidisk as @xmath58 .
we ask the following * question * : given an initial smooth phase domain @xmath48 , and the hele - shaw evolution @xmath19 , what kind of evolution it produces on the universal teichmller space as a natural general parametric space ? a general scheme of the proposed study is shown in figure [ fig2 ]
let us consider the family @xmath59 of all quasiconformal automorphisms of @xmath2 .
every such map @xmath60 satisfies the beltrami equation @xmath61 in @xmath2 in the distributional sense , where @xmath62 is a measurable essentially bounded function ( @xmath63 ) in @xmath2 , @xmath64 .
conversely , for each measurable beltrami coefficient @xmath47 essentially bounded as above , there exists a quasiconformal automorphism of @xmath2 , that satisfies the beltrami equation , which is unique if provided with some conformal normalization , e.g. , three point normalization @xmath65 , @xmath66 .
two normalized maps @xmath67 and @xmath68 are said to be equivalent , @xmath69 , if being extended onto the unit circle @xmath70 , the superposition @xmath71 restricted to @xmath70 is the identity map .
the quotient set @xmath72 is called _ the universal teichmller space _ @xmath0 .
it is a covering space for all teichmller spaces of analytically finite riemann surfaces . by definition
we have two realizations of @xmath0 : as a set of equivalence classes of quasiconformal maps and , due to the relation between @xmath72 and the unit ball @xmath73 , as a set of equivalence classes of corresponding beltrami coefficients .
the normalized maps from @xmath59 form a group @xmath74 with respect to superposition and the maps that act identically on @xmath70 form a normal subgroup @xmath75 .
thus , @xmath0 is the quotient of @xmath76 .
if @xmath77 , @xmath78 , then there exists a mbius transformation @xmath46 , such that @xmath79 .
let us denote by @xmath80\in t$ ] the equivalence class represented by @xmath78 .
then , one defines _ the universal modular group @xmath81 _ , @xmath82 , @xmath83 , by the formula @xmath84)=[h\circ
f\circ g^{-1}]$ ] .
its subgroup @xmath85 of right translations on @xmath0 is defined by @xmath86)=[f\circ g^{-1}]$ ] , where @xmath87 .
an important fact ( see ( * ? ? ?
* chapter iii , theorem 1.1 ) ) is that there are real analytic mappings in any equivalence class @xmath80\in t$ ] . given a beltrami coefficient @xmath88 let us extend it by zero into @xmath10 .
we normalize the corresponding quasiconformal map @xmath60 , which is conformal in @xmath10 , by @xmath89 about infinity .
then , two beltrami coefficients @xmath47 and @xmath90 are equivalent if and only if the corresponding normalized mappings @xmath91 and @xmath92 map @xmath10 onto one and the same domain in @xmath93 .
thus , the universal teichmller space can be thought of as the family of all normalized conformal maps of @xmath10 admitting quasiconformal extension .
moreover , any compact subset of @xmath0 consists of conformal maps @xmath60 of @xmath10 that admit quasiconformal extension to @xmath2 with @xmath94 for some @xmath95 .
as we mentioned above , a normalized conformal map @xmath96\in t$ ] defined in @xmath10 can have a quasiconformal extension to @xmath2 which is real analytic in @xmath2 , but on the unit circle @xmath60 may behave quite irregularly .
for example , the resulting quasicircle @xmath97 can have the hausdorff dimension greater than 1 . * remark . * given a bounded @xmath98-quasicircle @xmath99 , @xmath100 , in the plane let @xmath101 denote the minimal number of disks of radius @xmath102 that are needed to cover @xmath99 .
let @xmath103 denote the supremum of the minkowski dimension of curves @xmath99 where @xmath99 ranges over all bounded @xmath98-quasicircles .
the hausdorff dimension of @xmath99 is bounded from above by @xmath104 ( see @xcite ) . in @xcite it was also established several explicit estimates for @xmath104 , e.g. , @xmath105 .
let us denote by @xmath106 the class of those univalent conformal maps @xmath60 defined in @xmath10 which admit a quasiconformal extension to @xmath2 , normalized by @xmath89 .
let @xmath107 and @xmath108 be such that @xmath109 and @xmath110 .
then , _ the teichmller distance _ @xmath111 on @xmath0 is defined as @xmath112 for a given @xmath113 we consider an extremal beltrami coefficient @xmath114 such that @xmath115 .
let us remark that @xmath114 need not be unique . a geodesic on @xmath0 can be described in terms of the extremal coefficient @xmath114 as a continuous homomorphism @xmath116\mapsto t$ ] such that @xmath117 . due to the above remark the geodesic need not be unique as well .
we consider the banach space @xmath118 of all functions holomorphic in @xmath2 equipped with the norm @xmath119 for a function @xmath60 in @xmath9 the schwarzian derivative @xmath120 is defined and nehari s @xcite estimate @xmath121 holds .
given @xmath113 , @xmath122 we construct the mapping @xmath123 and have the homeomorphic embedding @xmath124 by the schwarzian derivative .
the universal teichmller space @xmath0 is an analytic infinite dimensional banach manifold modelled on @xmath118 .
the banach space @xmath118 is an infinite dimensional vector space that can be thought of as the cotangent space to @xmath0 at the initial point ( represented by @xmath125 ) .
more rigorously , let the map @xmath91 be a quasiconformal homeomorphism of the unit disk @xmath2 .
it has a frchet derivative with respect to @xmath47 in a direction @xmath90 .
let us construct the variation of @xmath126 , @xmath127 , with respect to a small parameter @xmath128 : @xmath129 taking the schwarzian derivative in @xmath10 we get @xmath130 locally uniformly in @xmath10 . taking into account the normalization of the class @xmath131 we have ( see , e.g. , @xcite ) @xmath132 the integral formula implies @xmath133 ( subject to the relation for the beltrami coefficient @xmath134 ) for any mbius transform @xmath135 . now let us change variables @xmath136 and reduce the first variation to a holomorphic function in the unit disk by changing @xmath137 to @xmath138 . setting @xmath139 and @xmath140 we have ( see , e.g. , ( * ? ? ?
* section 6.5 , theorem 5 ) ) that @xmath141 so the operator @xmath142 is the derivative of @xmath143 at the initial point of the universal teichmller space with respect to the norm of the banach space @xmath118 .
the reproducing property of the bergman integral gives @xmath144 the latter integral leads us to the so - called harmonic ( bers ) beltrami differential @xmath145 let us denote by @xmath146 the banach space of analytic functions with the finite @xmath147 norm in the unit disk .
we have that @xmath148 is a continuous inclusion ( see , e.g. , ( * ? ? ?
* section 1.4.2 ) ) . on @xmath149 one can define a coupling @xmath150 denote by @xmath151 the space of _ locally trivial beltrami coefficients _ , which is the subspace of @xmath63 that annihilates the operator @xmath152 for all @xmath153 .
then , one can identify the tangent space to @xmath0 at the initial point with the space @xmath154 .
it is natural to relate it to a subspace of @xmath63 .
the superposition @xmath155 acts identically on @xmath146 due to ( [ aaa ] ) .
the space @xmath151 is also the kernel of the operator @xmath142 .
thus , the operator @xmath156 splits the following exact sequence @xmath157 then , @xmath158 .
the coupling @xmath159 defines @xmath146 as a cotangent space .
let @xmath160 denote the banach space of analytic functions @xmath161 with the finite norm @xmath162 and petersson s hermitian product @xcite is defined on @xmath160 as @xmath163 the khlerian weil - petersson metric @xmath164 can be defined on the tangent space to @xmath0 and gives a khlerian manifold structure to @xmath0 .
the universal teichmller space is a smooth manifold on which a lie group @xmath165 of real sense preserving diffeomorphisms is defined .
the tangent bundle is defined on @xmath0 and is represented by the harmonic differentials from @xmath166 translated to all points of @xmath0 .
we will consider tangent vectors from @xmath166 at the initial point of @xmath0 represented by the map @xmath167 .
the weil - petersson metric defines a lie algebra of vector fields on @xmath0 by the poisson - lie bracket @xmath168=\{\nu_2,\nu_1\}-\{\nu_1,\nu_2\}$ ] , where @xmath169 .
one can define the poisson - lie bracket at all other points of @xmath0 by left translations from @xmath165 . to each element
@xmath170 $ ] from @xmath165 an element @xmath171 from @xmath0 is associated as an image of the initial point .
therefore , a curve in @xmath165 generates a traced curve in @xmath0 that can be realized by a one - parameter family of quasiconfromal maps from @xmath131 . for each tangent vector @xmath172
there is a one - parameter semi - flow in @xmath165 and a corresponding flow @xmath173 with the velocity vector @xmath90 . to make an explicit representation we use the variational formula for the subclass @xmath131 of @xmath13 of functions with quasiconformal extension ( see , e.g. ,
@xcite ) to @xmath93 .
if @xmath174 , @xmath172 and @xmath175 then the map @xmath176 locally describes the semi - flow @xmath177 on @xmath0 .
in this section we study a diffeomorphic embedding of the homogeneous manifold @xmath25 into the universal teichmller space @xmath0 .
we denote the lie group of @xmath178 sense preserving diffeomorphisms of the unit circle @xmath70 by @xmath179 .
each element of @xmath179 is represented as @xmath180 with a monotone increasing , @xmath178 real - valued function @xmath181 , such that @xmath182 .
the lie algebra for @xmath179 is identified with the lie algebra @xmath183 of smooth ( @xmath178 ) tangent vector fields to @xmath70 with the poisson - lie bracket given by @xmath184={\phi}_1{\phi}'_2-{\phi}_2{\phi}'_1.\ ] ] fixing the trigonometric basis in @xmath183 the commutator relations take the form @xmath185 & = & \frac{n - m}{2}\sin\,(n+m)\theta+ \frac{n+m}{2}\sin\,(n - m)\theta,\\ \left[\sin\,n\theta , \sin\,m\theta\right ] & = & \frac{m - n}{2}\sin\,(n+m)\theta+ \frac{n+m}{2}\sin\,(n - m)\theta,\\ \left[\sin\,n\theta , \cos\,m\theta\right ] & = & \frac{m - n}{2}\cos\,(n+m)\theta- \frac{n+m}{2}\cos\,(n - m)\theta.\end{aligned}\ ] ] there is no general theory of infinite dimensional lie groups , example of which is under consideration .
the interest to this particular case comes first of all from the string theory where the virasoro algebra appears as the central extension of @xmath183 .
entire necessary background for the construction of the theory of unitary representations of @xmath179 is found in the study of kirillov s homogeneous khlerian manifold @xmath1 , where @xmath186 denotes the group of rotations of @xmath70 .
the group @xmath179 acts as a group of translations on the manifold @xmath187 with @xmath186 as a stabilizer .
the khlerian geometry of @xmath187 has been described by kirillov and yuriev in @xcite .
the manifold @xmath187 admits several representations , in particular , in the space of smooth probability measures , symplectic realization in the space of quadratic differentials .
we will use its analytic representation that is based on the class @xmath188 of functions from @xmath13 which being extended onto the closure @xmath189 of @xmath10 are supposed to be smooth on @xmath70 .
the class @xmath188 is dense in @xmath13 in the local uniform topology of @xmath10 .
let @xmath190 denote the class of all univalent holomorphic maps in the unit disk @xmath191 which are smooth on @xmath70 .
then , for each @xmath192 we have @xmath193 and there is an _ adjoint map _
@xmath194 such that @xmath195 .
the superposition @xmath196 restricted to @xmath70 is in @xmath187 ( see figure [ aafig3 ] ) .
reciprocally , for each element of @xmath187 there exist such @xmath60 and @xmath197 .
a piece - wise smooth closed jordan curve is a quasicircle if and only if it has no cusps .
so any function @xmath60 from @xmath188 has a quasiconformal extension to @xmath2 . by this realization the manifold @xmath187 is naturally embedded into the universal teichmller space @xmath0 .
moreover , the khlerian structure on @xmath187 corresponds to the khlerian structure on @xmath0 given by the weil - petersson metric .
the goluzin - schiffer variational formulae lift the actions from the lie algebra @xmath183 onto @xmath188 .
let @xmath198 and let @xmath199 be a @xmath178 real - valued function in @xmath200 $ ] from @xmath183 making an infinitesimal action as @xmath201 .
let us consider a variation of @xmath60 given by @xmath202 kirillov and yuriev @xcite , @xcite have established that the variations @xmath203 are closed with respect to the commutator and the induced lie algebra is the same as @xmath183 .
moreover , kirillov s result @xcite states that there is the exponential map @xmath204 such that the subgroup @xmath186 coincides with the stabilizer of the map @xmath205 from @xmath188 .
let @xmath206 be a circle quasisymmetric homeomorphism , i.e. , a homeomorphism that possesses a quasiconformal extension into @xmath2 ( for a precise definition see , e.g. , @xcite ) .
then @xmath161 has infinitely many quasiconformal extensions into @xmath2 , one of the most remarkable of which is the beurling - ahlfors extension @xcite . in 1986
douady and earle @xcite defined for any such @xmath206 a conformally natural extension @xmath207 from @xmath59 .
the map @xmath46 is a homeomorphism which is real analytic in the interior .
the idea was to introduce the concept of a conformal barycenter of a measure on @xmath208 .
douady and earle proved that @xmath209 satisfies the functional equation @xmath210 an advantage of this extension is that if @xmath211mb@xmath212 , then the extension of @xmath213 is given by @xmath214 , what is not true for the beurling - ahlfors extension .
the three - point boundary normalization of @xmath74 can be always attained , and thus , the douady - earle extension is compatible with the definition of the universal teichmller space .
later , in 1988 , another proof of douady - earle s result has appeared in @xcite where the authors worked with the inverse function .
the functional equation ( [ de ] ) , in particular , implies that a @xmath178 mapping @xmath161 representing an element from the manifold @xmath187 has a real analytic extension @xmath215 which is @xmath178 on @xmath70 .
let @xmath198 represent an element from @xmath216 .
let @xmath217 be the adjoint map , @xmath218 .
if @xmath46 is the douady - earle extension of @xmath161 , then @xmath219 and @xmath220 is a quasiconformal extension of @xmath198 .
given @xmath216 we construct the mapping @xmath91 that satisfies the normalization of the class @xmath188 and whose beltrami coefficient is @xmath221 with @xmath222 for @xmath32 .
the equivalence class @xmath223 $ ] is a point of the universal teichmller space @xmath0 .
so the douady - earle extension defines an explicit embedding of @xmath187 into @xmath0 .
as it was mentioned , the weil - petersson metric defines a lie algebra of vector fields on @xmath0 by the poisson bracket @xmath168=\{\nu_2,\nu_1\}-\{\nu_1,\nu_2\}$ ] , where @xmath169 .
one can define the poisson bracket at all other points of @xmath0 by left translations of the universal modular group .
we proceed restricting ourselves to @xmath187 embedded into @xmath0 .
the complex form of green s formula implies that ( [ var ] ) for @xmath167 is equivalent to @xmath224 where the distributional derivative @xmath225 is given in the unit disk @xmath2 , @xmath226 is a continuous extension of the @xmath178 function @xmath227 into @xmath2 that has @xmath228 distributional derivatives in @xmath2 , @xmath229 , and @xmath230 is the area element in @xmath2 .
thus , one can extract the elements from @xmath166 that are of the form @xmath231 , where @xmath232 means @xmath233 .
we are going to deduce an exact form of @xmath90 using the douady - earle extension .
for this we start with the variation of the element @xmath234 and @xmath128 is small .
the beltrami coefficient of the extended quasiconformal map @xmath46 has its variation as @xmath235 , where @xmath236 where @xmath237 thus , @xmath238 depends only on @xmath199
. we will give explicit formulae in the next section .
they can be obtained substituting @xmath239 , and taking into account that @xmath240 the lie algebra @xmath183 is embedded into the lie algebra of @xmath166 by ( [ mu1 ] ) , ( [ de1 ] ) .
hence , a flow given on @xmath187 corresponding to a vector @xmath241 is represented as a flow on the universal teichmller space @xmath0 corresponding to the vector @xmath172 given by ( [ mu1 ] ) .
first of all we give an explicit formula that connects the vectors @xmath199 from @xmath183 with corresponding tangent vectors @xmath242 to the universal teichmller space @xmath0 making use of the douady - earle extension .
these vectors give the infinitesimal description of semi - flows on @xmath187 and @xmath0 respectively .
let @xmath227 be the infintesimal description of a flow @xmath161 in @xmath187 .
then , the corresponding infinitesmial description @xmath242 of this flow embedded into @xmath0 is given by the function @xmath243 let @xmath244 and @xmath245 be the douady - earle extension of @xmath161 into the unit disk @xmath2 , @xmath246 by means of ( [ de1 ] ) . if @xmath247 , then @xmath248 . we calculate @xmath249 substituting @xmath247 and @xmath250 we have @xmath251 we will use the properties of the douady - earle extension . let us fix a point @xmath252 and choose two mbius transformations @xmath253 of @xmath2 such that @xmath254 and @xmath255 .
we set @xmath256 .
then , @xmath257 , @xmath258 and @xmath259 so we see that @xmath260 by the property of the douady - earle extension we have that the function @xmath261 , @xmath246 is the extension of @xmath262 by means of ( [ de1 ] ) . if @xmath247 , then @xmath263 .
now we put @xmath264 in ( [ de1 ] ) and calculate variations in @xmath128 @xmath265 then , we can obtain the explicit form of the variation of the beltrami coefficient by ( [ mu1 ] ) as @xmath266 the mbius transformation @xmath267 does not depend on @xmath128 whereas @xmath268 does .
explicitly , we put @xmath269 where @xmath270 .
we denote @xmath271 .
therefore , denoting by @xmath272 we have @xmath273 then , @xmath274 and changing variables in ( [ dot ] ) , we obtain @xmath275 taking into account that @xmath276 we come to the statement of the theorem . if @xmath277}|d(e^{i\theta})|$ ]
, then @xmath278 the formula given in the preceding theorem implies @xmath279 changing variables @xmath280 we obtain @xmath281 next , we obviously estimate @xmath282 as in the statement of the corollary .
as we see , the given estimate is good enough when @xmath283 is not close to 1 .
let us now give an asymptotic estimate for @xmath284 in the case @xmath285 .
there exists a constant @xmath187 independent of @xmath286such that @xmath287 in particular , @xmath288 as @xmath285 .
we integrate by parts the right - hand side in the formula ( [ nnn ] ) twice and come to the following expression @xmath289}{\partial \theta}+\frac{\partial^2 [ e^{i\theta}d(e^{i\theta})]}{\partial \theta^2}\right)d\theta.\ ] ] the absolute value of the above integral is bounded because of the poisson kernel in it and due to the smoothness of the function @xmath290 .
we consider the semigroup @xmath291 of conformal univalent maps from @xmath10 into itself with composition as the semigroup operation .
this makes @xmath291 a topological semigroup with respect to the topology of local uniform convergence on @xmath10 .
we impose the natural normalization for such conformal maps : @xmath292 , @xmath32 , @xmath293 .
the unit of the semigroup is the identity .
let us construct on @xmath291 a one - parameter semi - flow @xmath294 , that is , a continuous homomorphism from @xmath295 into @xmath291 , with the parameter @xmath296 .
for any fixed @xmath296 the element @xmath294 is from @xmath291 and is represented by a conformal map @xmath297 from @xmath10 onto the domain @xmath298 .
the element @xmath294 satisfies the following properties : in particular , @xmath304 .
this semi - flow is generated by a vector field @xmath305 if for each @xmath32 the function @xmath306 , @xmath307 is a solution of an autonomous differential equation @xmath308 with the initial condition @xmath309 .
the semi - flow can be extended to a symmetric interval @xmath310 by putting @xmath311 .
certainly , the latter function is defined on the set @xmath312 . admitting this restriction for negative @xmath128 we define a one - parameter family @xmath294 for @xmath313 . for a semi - flow @xmath294 on @xmath291
there is an infinitesimal generator at @xmath314 constructed by the following procedure .
any element @xmath294 is represented by a conformal map @xmath315 that satisfies the schwarz lemma for the maps @xmath316 , and hence , @xmath317 where the equality sign is attained only for @xmath318 .
therefore , the following limit exists ( see , e.g. , @xcite , @xcite , @xcite ) @xmath319 and the representation @xmath320 holds , where @xmath321 is an analytic function in @xmath10 with positive real part , and @xmath322 in @xcite it was shown that @xmath294 is even @xmath178 with respect to @xmath128 . the function @xmath323 is an infinitesimal generator for @xmath294 at @xmath314 , and the following variational formula holds @xmath324 the convergence is thought of as local uniform .
we rewrite ( [ aa1 ] ) as @xmath325 now let us proceed with the semigroup @xmath326 of quasiconformal automorphisms of @xmath327 .
a quasiconformal map @xmath328 representing an element of @xmath329 satisfies the beltrami equation in @xmath93 @xmath330 with the distributional derivatives @xmath331 and @xmath332 , where @xmath333 is a measurable function vanishing in @xmath10 and essentially bounded in @xmath2 by @xmath334 for some @xmath95 . if @xmath95 is sufficiently small , then the function @xmath335 satisfies the variational formula ( see , e.g. , @xcite ) @xmath336 where @xmath337 stands for the area element in the @xmath338-plane .
now for each @xmath128 small and @xmath339 the mapping @xmath340 is from @xmath131 and represents an equivalence class @xmath341\in t$ ] .
consider the one - parameter curve @xmath173 that corresponds to @xmath341 $ ] and a velocity vector @xmath242 ( that is not trivial ) , such that @xmath342 we take into account that @xmath343 in @xmath10 and is extended up to the identity map of @xmath93 .
the formula ( [ aa3 ] ) can be rewritten for @xmath344 as @xmath345 comparing with ( [ aa2 ] ) we come to the conclusion about @xmath328 : @xmath346 the relations ( [ aa1 ] , [ aa2 ] , [ aa5 ] ) imply that @xmath347 the constants @xmath348 and the function @xmath90 must be such that @xmath349 for all @xmath350 .
let @xmath294 be a semi - flow in @xmath329 .
then it is generated by the vector field @xmath351 , @xmath352 where @xmath242 is a harmonic beltrami differential and the holomorphic function @xmath353 has positive real part in @xmath10 .
in particular , if @xmath362 is a one - parameter semi - flow , then @xmath363 is an evolution family .
we consider a subordination chain of mappings @xmath31 , @xmath32 , @xmath34 , where the function @xmath364 is a meromorphic univalent map @xmath365 for each fixed @xmath21 and @xmath366 for @xmath367 .
let us assume that this subordination chain exists for @xmath21 in an interval @xmath368 .
let us pass to the semigroup @xmath329 .
so @xmath355 now has a quasiconformal extension to @xmath2 and being restricted to @xmath10 is from @xmath291 .
moreover , @xmath360 locally uniformly in @xmath369 as @xmath370 . for each @xmath21 fixed in @xmath368 the map @xmath31 has a quasiconformal extension into @xmath2 ( that can be assumed even real analytic ) .
an important presupposition is that @xmath31 generates a _ nontrivial path _ in the universal teichmller space @xmath0 .
this means that for any @xmath371 , @xmath372 , the mapping @xmath373 , @xmath32 , can not be obtained from @xmath374 by a mbius transform , or taking into account the normalization of @xmath60 , by multiplying by a constant .
we construct the superposition @xmath375 for @xmath34 , @xmath376 .
putting @xmath377 we denote this mapping by @xmath378 . * the maps @xmath31 form a subordination chain in @xmath10 , @xmath34 . *
the map @xmath31 is holomorphic in @xmath10 , @xmath379 , where @xmath52 and differentiable with respect to @xmath21 . *
the map @xmath31 is a quasiconformal homeomorphism of @xmath93 . * the chain of maps
@xmath31 is not trivial . *
the beltrami coefficient @xmath380 of this map is differentiable with respect to @xmath21 locally uniformly in @xmath2 , vanishes in some neighbourhood of @xmath10 ( independently of @xmath21 ) .
the function @xmath378 is embedded into an evolution family in @xmath291 .
it is differentiable with regard to @xmath128 and @xmath21 in @xmath368 , and @xmath381 .
fix @xmath21 and let @xmath382 . then
, there exists @xmath172 such that the beltrami coefficient @xmath47 is of the form @xmath383 in @xmath384 , @xmath385 in @xmath386 , and vanishes in @xmath387 .
we make @xmath128 sufficiently small such that @xmath388 vanishes in @xmath386 too .
therefore , @xmath389 locally uniformly in @xmath369 and @xmath378 is embedded now into an evolution family in @xmath329 .
the identity map is embedded into a semi - flow @xmath390 ( which is smooth ) as the initial point with the same velocity vector @xmath391 that leads to equation ( [ lkp ] ) ( the semi - flow @xmath294 is tangent to the evolution family at the origin ) . actually , the differentiable trajectory @xmath31 generates a pencil of tangent smooth semi - flows with starting tangent vectors @xmath392 ( that may be only measurable with respect to @xmath21 ) .
the projection to the universal teichmller space is shown in figure [ aafig4 ] .
the requirement of non - triviality makes it possible to use the variation ( [ aa4 ] ) .
therefore , the conclusion is that the function @xmath31 satisfies the equation ( [ lkp ] ) where the function @xmath393 is given by @xmath394 and has positive real part .
the existence of @xmath395 comes from the existence of the subordination chain
. we can assign the normalization to @xmath31 controlling the change of the conformal radius of the subordination chain by @xmath36 . in this case , changing variables we obtain @xmath396 , @xmath397 .
[ quasi ] let @xmath31 be a subordination chain of maps in @xmath10 that exists for @xmath34 and satisfies the conditions ( i v ) .
then , there are a real valued function @xmath398 , a complex valued function @xmath399 , and a harmonic beltrami differential @xmath400 , such that @xmath401 for @xmath32 , @xmath402 and @xmath31 satisfies the differential equation @xmath403 in @xmath404 . in the above theorem the function @xmath400 belongs to the space of harmonic differentials .
we ask now about another but equivalent form of @xmath90 as well as whether one can extend the equation @xmath405 onto the whole complex plane .
writing @xmath406 , @xmath407 we calculate the dilatation of the function @xmath378 in @xmath2 .
note that @xmath328 it is differentiable by @xmath408 .
@xmath409 we use that @xmath410 . finally , @xmath62 , @xmath411 , @xmath412 are differentiable by @xmath21 almost everywhere in @xmath34 , locally uniformly in @xmath246 , and @xmath413 where the limit exists a.e . with respect
to @xmath34 locally uniformly in @xmath246 , or in terms of the inverse function @xmath414 sometimes , it is much better to operate just with dilatations , avoiding functions , so we can rewrite the last expression as @xmath415.\ ] ] let us consider one - parameter families of maps in @xmath10 normalized by @xmath417 .
the inverse result to the lwner - kufarev equation states that given a holomorphic function @xmath418 in @xmath32 with positive real part the solution of the equation ( [ aa7 ] ) presents a subordination chain ( see , e.g. , @xcite ) .
this enable us to give a condition for @xmath419 that guarantees a normalized one - parameter non - trivial family of maps @xmath31 to be a subordination chain let @xmath31 be a normalized one - parameter non - trivial family of maps for @xmath32 which satisfies the conditions ( ii v ) and is defined in an interval @xmath368 .
let each @xmath31 be a homeomorphism of @xmath93 which is meromorphic in @xmath10 , is normalized by @xmath417 , and satisfies ( [ aa7 ] ) .
let the quasiconformal extension to @xmath2 be given by a beltrami coefficient @xmath420 which is differentiable with respect to @xmath21 almost everywhere in @xmath421 .
if @xmath422 where @xmath423 is as above and @xmath424 is the complete elliptic integral , then @xmath31 is a normalized subordination chain .
_ let us remark that the function @xmath419 can be unilateraly discontinuous on @xmath70 in @xmath432 , therefore , it is not possible , in general , to use the borel - pompeiu formula to reduce the integral in @xmath44 to a contour integral .
now we discuss the possibility of extending the equation ( [ aa7 ] ) to all of @xmath369 .
we differentiate the function @xmath378 with respect to @xmath128 when @xmath434 .
it follows that @xmath435 this formula can be rewritten in the following form @xmath436 taking into account the equation ( [ aa7 ] ) in @xmath10 we have in the whole plane @xmath437 where @xmath393 is a holomorphic in @xmath10 function with the positive real part by theorem [ quasi ] . the variational formula ( [ aa5 ] ) and differentiation of the singular integral imply that @xmath438 , @xmath246 .
now let us clarify what is @xmath439 .
let us consider @xmath246 .
the pompeiu formula leads to @xmath440 where @xmath441 is a holomorphic function with respect to @xmath286 .
the function @xmath439 is continuous in @xmath432 and by the cauchy theorem @xmath442 to obtain the boundary values of the function @xmath443 , @xmath444 , we will use the second line in ( [ aaqq ] ) .
unfortunately , in general , it is not possible to use the same function @xmath60 in both lines of ( [ aaqq ] ) to obtain boundary values of @xmath439 .
indeed , the mapping @xmath31 is differentiable regarding to @xmath21 a.e . in @xmath445 locally uniformly in @xmath446 , and continuous in @xmath447 for almost all @xmath34 .
therefore , the function @xmath448 , @xmath246 is the extension of @xmath449 , @xmath32 , whereas @xmath411 , @xmath246 is not necessarily an extension of @xmath450 , @xmath32 .
a simple example of this situation is as follows .
let us consider the function @xmath451 where @xmath452 .
this mapping forms a subordination chain with the dilatation @xmath453 that vanishes in @xmath10 and is the constant @xmath454 in @xmath2 .
this chain is trivial , but it is not important for our particular goal here because we do not use at this stage the crucial variation .
then , @xmath455 and it splits into two parts that can not be glued on @xmath70 .
the same is for the derivatives @xmath411 in @xmath2 and @xmath450 in @xmath10 .
if @xmath456 satisfies the condition ( v ) in a neighbourhood of @xmath70 in @xmath432 , then the derivatives @xmath411 , @xmath412 , @xmath246 has a continuation onto @xmath70 and @xmath457 where @xmath458 is thought of as the angular limits that exist a.e . on @xmath70 .
moreover , in a neighbourhood of @xmath70 the derivative @xmath412 vanishes and the function @xmath459 can be written on @xmath70 as @xmath460 . in turn ,
@xmath461 * for @xmath32 there exists a holomorphic function @xmath393 given by theorem [ quasi ] such that @xmath462 * for @xmath246 there exists a continuous in @xmath286 function @xmath459 given by @xmath463 @xmath464 such that @xmath465 dually to the lwner - kufarev partial derivative equation there is the lwner - kufarev ordinary differential equation .
a function @xmath466 is represented as a limit @xmath467 where the function @xmath468 is a solution of the equation @xmath469 almost everywhere in @xmath470 , with the initial condition @xmath471 .
the function @xmath418 is analytic in @xmath10 , measurable with respect to @xmath470 , and its real part @xmath472 is positive for almost all @xmath473 .
the equation ( [ lkord ] ) is known as the _ lwner - kufarev ordinary differential equation_. the solutions to ( [ lkord ] ) form a retracting subordination chain @xmath474 , i.e. , it satisfies the condition @xmath475 , @xmath476 for @xmath477 , and @xmath478 .
the connection between ( [ aa7 ] ) and ( [ lkord ] ) can be thought of as follows .
solving ( [ aa7 ] ) by the method of characteristics and assuming @xmath479 as the parameter along the characteristics we have @xmath480 with the initial conditions @xmath481 , @xmath482 , @xmath483 , where @xmath484 is in @xmath10 .
we see that the equation ( [ lkord ] ) is exactly the characteristic equation for ( [ aa7 ] ) .
unfortunately , this approach requires the extension of @xmath485 into @xmath10 because the solution of the function @xmath31 is given as @xmath485 , where @xmath486 is the solution of the initial value problem for the characteristic equation . * the maps @xmath474 form a retracting subordination chain @xmath488 . * the map @xmath474 is meromorphic in @xmath10 , @xmath379 , where @xmath52 and differentiable with respect to @xmath21 .
* the map @xmath474 is a quasiconformal homeomorphism of @xmath93 . * the chain of maps
@xmath474 is not trivial . *
the beltrami coefficient @xmath489 of this map is differentiable with respect to @xmath21 locally uniformly in @xmath2 .
set @xmath490 where @xmath468 . for each fixed @xmath21
the mapping @xmath474 generates a smooth semi - flow @xmath491 in @xmath329 which is tangent to the path @xmath492 at @xmath247 .
therefore , we use the velocity vector @xmath493 ( that may be only measurable regarding to @xmath21 ) with @xmath468 and obtain @xmath494 as before , the trajectory @xmath474 generates a pencil of tangent smooth semi - flows with the tangent vectors @xmath493 , @xmath495 . since @xmath496 for any @xmath497 , we can consider the limit @xmath498 we have that @xmath499 where @xmath500 is an analytic function in @xmath10 that has positive real part for almost all fixed @xmath21 .
the equation defined by ( [ aa9 ] ) is an evolution equation for the path @xmath474 and the initial condition is given by @xmath471 .
we suppose that all @xmath474 admit real analytic quasiconformal extensions and the family is non - trivial in the above sense .
the function @xmath501 can be extended to a function from @xmath131 and it represents an equivalence class @xmath502\in t$ ] .
there is a one - parameter path @xmath503 that corresponds to a tangent velocity vector @xmath504 such that @xmath505 we calculate explicitly the velocity vector making use of the beltrami coefficient for a superposition : @xmath506 or @xmath507 it is natural to implement an intrinsic parametrization using the teichmller distance @xmath508)=t$ ] , and assume the conformal radius to be @xmath509 that implies @xmath396 .
the assumption of non - triviality allows us to use the variational formula ( [ aa5 ] ) to state the following theorem .
let @xmath474 be a retracting subordination chain of maps defined in @xmath34 and @xmath32 .
each @xmath474 is a homeomorphism of @xmath93 which is meromorphic in @xmath10 , @xmath510 , with a @xmath511-quasiconformal extension to @xmath2 given by a beltrami coefficient @xmath456 that is differentiable regarding to @xmath21 a.e . in @xmath512 .
the initial condition is @xmath513 .
then , there is a function @xmath393 such that that @xmath401 for @xmath514 , and @xmath515 where @xmath516 is given by the formula ( [ aa10 ] ) , @xmath517 , and @xmath468 is a solution to the differential equation @xmath518 with the initial condition @xmath471 .
_ the function @xmath493 has a continuation into @xmath521 given by @xmath522 where the function @xmath523 is a solution to the equation @xmath524 in contrary to the lwner - kufarev equation in partial derivatives , the function @xmath525 is the continuation of @xmath44 in @xmath2 through @xmath70 .
the solution exists by the pompeiu integral and can be written as @xmath526 where @xmath527 , @xmath528 is a holomorphic functions with respect to @xmath338 , that can be written as @xmath529 let us consider the class @xmath535 of functions @xmath536 , @xmath32 , such that being extended onto @xmath70 they are @xmath178 on @xmath70 . repeating considerations of the preceding subsection for the embedding of @xmath187 into the teichmller space @xmath0
we come to the following theorem .
[ smooth ] let @xmath31 be a non - trivial subordination chain of maps that exists for @xmath34 and @xmath32 .
each @xmath31 is a homeomorphism @xmath537 and belongs to @xmath535 for every fixed @xmath21 .
all these maps have quasiconformal extensions to @xmath2 and there are a real - valued function @xmath398 , complex - valued functions @xmath399 , real - valued @xmath178 functions @xmath538 such that @xmath401 for @xmath32 , @xmath539 and @xmath31 satisfies the differential equation @xmath540 theorems [ quasi ] and [ smooth ] are linked as follows . for a given subordination chain of maps @xmath541 , that exists for @xmath34 and @xmath32 ,
there is a @xmath178 function @xmath538 by theorem [ smooth ] and we can construct the function @xmath400 by the douady - earle extension and the formula ( [ mu1 ] ) .
then , the function @xmath31 satisfies the equation of theorem [ quasi ] with @xmath393 defined by such @xmath400 .
let us consider the ordinary lwner - kufarev equation for the functions smooth on @xmath70 .
if the retracting chain @xmath474 is smooth on @xmath70 , then we use again the embedding of @xmath187 into @xmath0 and reach a similar result .
let @xmath474 be a retracting non - trivial subordination chain of normalized maps that exists for @xmath34 and @xmath32 .
each @xmath474 is meromorphic in @xmath10 , smooth on @xmath70 , and @xmath542 , @xmath543 .
an additional assumption is that @xmath544 for each fixed @xmath21 .
then , there are a real - valued function @xmath545 , a complex - valued function @xmath399 , and a smooth real - valued function @xmath538 , such that @xmath401 for @xmath32 , @xmath546 and @xmath468 is a solution to the differential equation @xmath547 with the initial condition @xmath471 .
theorem [ smooth ] is linked to the hele - shaw free boundary problem as follows .
starting with a smooth boundary @xmath49 the one - parameter family @xmath551 consists of smooth curves as long as the solutions exist .
let us consider the equation ( [ pgl ] ) .
under injection we have a subordination chain of domains @xmath19 .
the schwarz kernel can be developed as @xmath552 therefore , in theorem [ smooth ] we can put @xmath553 and @xmath554 apart from the trivial elliptic case there are no self - similar solutions , and therefore the hele - shaw dynamics @xmath31 generates a non - trivial path in @xmath0 .
thus , given a hele - shaw evolution @xmath555 we observe a differentiable non - trivial path on @xmath0 , such that at any time @xmath21 the tangent vector @xmath90 is a harmonic beltrami differential given by @xmath556 the corresponding co - tangent vector is @xmath557 j. becker , _ conformal mappings with quasiconformal extensions_. aspects of contemporary complex analysis ( proc .
study inst .
durham , durham , 1979 ) academic press , london - new york , 1980 , 3777 .
v. v. goryainov , _ fractional iterates of functions that are analytic in the unit disk with given fixed points _ , mat .
* 182 * ( 1991 ) , no . 9 , 12811299 ; engl .
transl . in math .
ussr - sb .
* 74 * ( 1993 ) , no .
1 , 2946 . v. v. goryainov , _ one - parameter semigroups of analytic functions _ , geometric function theory and applications of complex analysis to mechanics : studies in complex analysis and its applications to partial differential equations , 2 ( halle , 1988 ) , pitman res .
notes math .
ser . , 257 , longman sci
harlow , 1991 , 160164 .
v. v. goryainov , _ one - parameter semigroups of analytic functions and a compositional analogue of infinite divisibility .
_ proceedings of the institute of applied mathematics and mechanics , vol .
5 , tr . inst .
prikl . mat .
mekh . , 5 , nats .
nauk ukrainy inst .
prikl . mat .
donetsk , 2000 , 4457.(in russian ) v. ya .
gutljanski , _ the method of variations for univalent analytic functions with a quasiconformal extension _ ,
sibirsk . mat .
zh . * 21 * ( 1980 ) , no .
2 , 6178 ; engl . transl . in siberian math .
j. * 21 * ( 1980 ) , no .
2 , 190204 . | we study one - parameter curves on the universal teichmller space @xmath0 and on the homogeneous space @xmath1 embedded into @xmath0 . as a result
, we deduce evolution equations for conformal maps that admit quasiconformal extensions and , in particular , such that the associated quasidisks are bounded by smooth jordan curves . some applications to hele - shaw flows of viscous fluids are given . |
heusler alloys @xcite have attracted during the last century a great interest due to the possibility to study in the same family of alloys a series of interesting diverse magnetic phenomena like itinerant and localized magnetism , antiferromagnetism , helimagnetism etc @xcite .
the first heusler alloys studied were crystallizing in the @xmath1 structure which consists of 4 fcc sublattices .
afterwards , it was discovered that it is possible to leave one of the four sublattices unoccupied ( @xmath2 structure ) .
the latter compounds are often called half- or semi - heusler alloys , while the @xmath1 compounds are referred to as full - heusler alloys .
nimnsb belongs to the half - heusler alloys @xcite . in 1983
de groot and his collaborators @xcite showed by using first - principles electronic structure calculations that this compound is in reality half - metallic , i.e. the minority band is semiconducting with a gap at the fermi level @xmath3 , leading to 100% spin polarization at @xmath3 . other known half - metallic materials except the half- and full - heusler alloys ( see @xcite and references therein ) are some oxides ( _ e.g _ cro@xmath0 and fe@xmath4o@xmath5 ) @xcite , the manganites ( _ e.g _ la@xmath6sr@xmath7mno@xmath4 ) @xcite , the double perovskites ( _ e.g. _ sr@xmath0fereo@xmath8 ) @xcite , the pyrites ( _ e.g _ cos@xmath0 ) @xcite , the transition metal chalcogenides ( _ e.g _ crse ) and pnictides ( _ e.g _ cras ) in the zinc - blende or wurtzite structures @xcite , the europium chalcogenides ( _ e.g _ eus ) @xcite and the diluted magnetic semiconductors ( _ e.g _ mn impurities in si or gaas ) @xcite . although thin films of cro@xmath0 and la@xmath6sr@xmath7mno@xmath4 have been verified to present practically 100% spin - polarization at the fermi level at low temperatures @xcite , the heusler alloys remain attractive for technical applications like spin - injection devices @xcite , spin - filters @xcite , tunnel junctions @xcite , or gmr devices @xcite due to their relatively high curie temperature compared to these compounds @xcite .
ishida and collaborators studied by means of _ ab - initio _ calculations the full - heusler compounds of the type co@xmath0mnz , where z stands for si and ge , and have shown that they are half - metals @xcite .
later the origin of half - metallicity in these compounds has been largely explained @xcite .
many experimental groups during the last years have worked on these compounds and have tried to synthesize them mainly in the form of thin films and incorporate them in spintronic devices .
the group of westerholt has extensively studied the properties of co@xmath0mnge films and they have incorporated this alloy in the case of spin - valves and multilayer structures @xcite .
the group of reiss managed to create magnetic tunnel junctions based on co@xmath0mnsi @xcite .
a similar study of sakuraba and collaborators resulted in the fabrication of magnetic tunnel junctions using co@xmath0mnsi as one magnetic electrode and al - o as the barrier ( co@xmath9fe@xmath10 is the other magnetic electrode ) and their results are consistent with the presence of half - metallicity for co@xmath0mnsi @xcite . dong and collaborators recently managed to inject spin - polarized current from co@xmath0mnge into a semiconducting structure @xcite .
finally kallmayer _ et al .
_ studied the effect of substituting fe for mn in co@xmath0mnsi films and have shown that the experimental extracted magnetic spin moments are compatible with the half - metallicity for small degrees of doping @xcite .
it is obvious from the experimental results that the full - heusler compounds containing co and mn are of particular interest for spintronics .
not only they combine high curie temperatures and coherent growth on top of semiconductors ( they consist of four fcc sublattice with each one occupied by a single chemical element ) but in real experimental situations they can preserve a high degree of spin - polarization at the fermi level . in order to accurately control their properties
it is imperative to investigate the effect of defects , doping and disorder on their properties .
recently picozzi _ et al .
_ published a study on the effect of defects in co@xmath0mnsi and co@xmath0mnge @xcite followed by an extensive review of the defects in these alloys @xcite .
authors have studied in the recent years several aspects of these half - metallic alloys like the origin of the gap @xcite , properties of surfaces @xcite and interfaces with semiconductors @xcite , the quaternary @xcite , the orbital magnetism @xcite , the exchange constants @xcite , the magneto - optical properties @xcite , the half - metallic ferrimagnetic heusler alloys like mn@xmath0val , @xcite and the fully - compensated half - metallic ferrimagnets or simply half - metallic antiferromagnets @xcite . in this chapter
we will overview some of our results on the defects in the half - metallic full heusler alloys obtained using the full potential nonorthogonal local
orbital minimum basis band structure scheme ( fplo ) @xcite within the local density approximation ( lda ) @xcite and employing the coherent potential approximation ( cpa ) to simulate the disorder in a random way @xcite . in section [ sec2 ]
we present the physics of defects in the ferromagnetic heusler alloys containing co and mn like co@xmath0mnsi @xcite . in section [ sec3 ]
we show the creation of half - metallic ferrimagnets based on the creation of cr and mn antisites in co@xmath0(cr or mn)(al or si ) alloys @xcite and in section [ sec4 ] we expand this study to cover the case of co defects in ferrimagnetic mn@xmath0val and mn@xmath0vsi alloys leading to half - metallic antiferromagnets @xcite . in section [ sec5 ]
we investigate a special case : the occurrence of vacancies in the full - heusler compounds @xcite .
finally in section [ sec6 ] we summarize and conclude .
[ mn@xmath11cr@xmath12si and co@xmath0[mn@xmath11fe@xmath12si for three difference values of the doping concentration @xmath13 .
dos s are compared to the one of the undoped co@xmath0mnsi alloy . in the insets we present the dos for a wider energy range .
we have set the fermi level as the zero of the energy axis .
note that positive values of dos refer to the majority - spin electrons and negative values to the minority - spin electrons . ]
we will start our discussion presenting our results on the defects in the case of co@xmath0mnz alloys where z is al and si @xcite .
the first part of our investigation in this section concerns the doping of co@xmath0mnsi . to simulate the doping by electrons we substitute fe for mn while to simulate the doping of the alloys with holes we substitute cr for mn .
we have studied the cases of moderate doping substituting 5% , 10% and 20% of the mn atoms .
we will start our discussion from figure [ fig1 ] where we present the total density of states ( dos ) for the co@xmath0[mn@xmath11fe@xmath12si and co@xmath0[mn@xmath11cr@xmath12si compounds .
as discussed in reference @xcite the gap is created between states located exclusively at the co sites .
the states low in energy ( around -6 ev ) originate from the low - lying @xmath14-states of the @xmath15 atoms ( there is also an @xmath16-type state very low in energy which is not shown in the figure ) .
the majority - spin occupied states form a common mn - co band while the occupied minority states are mainly located at the co sites and the minority unoccupied states at the mn sites . doping the perfect
ordered alloy with either fe or cr smoothens the valleys and peaks along the energy axis .
this is a clear sign of the chemical disorder ; fe and cr induce peaks at slightly different places than the mn atoms resulting to this smoothening and as the doping increases this phenomenon becomes more intense .
the important detail is what happens around the fermi level and in what extent is the gap in the minority band affected by the doping .
so now we will concentrate only at the enlarged regions around the fermi level .
the blue dashed lines represent the cr - doping while the red dash - dotted lines are the fe - doped alloys .
cr - doping has only marginal effects to the gap .
its width is narrower with respect to the perfect compounds but overall the compounds retain their half - metallicity . in the case of fe - doping
the situation is more complex . adding electrons to the system means that , in order to retain the perfect half - metallicity , these electrons should occupy high - energy lying antibonding majority states .
this is energetically not very favorable and for these moderate degrees of doping a new shoulder appears in the unoccupied states which is close to the right - edge of the gap ; a sign of a large change in the competition between the exchange splitting of the mn majority and minority states and of the coulomb repulsion . in the case of the 20% fe doping this new peak crosses the fermi level and the fermi level is no more exactly in the gap but slightly above it .
further substitution should lead to the complete destruction of the half - metallicity as in the quaternary heusler alloys with a mn - fe disordered site @xcite .
recent ab - initio calculations including the on - site coulomb repulsion ( the so - called hubbard @xmath17 ) have predicted that co@xmath0fesi is in reality half - metallic reaching a total spin magnetic moment of 6 @xmath18 which is the largest known spin moment for a half - metal @xcite .
we expand our theoretical work to include also the case of co@xmath0mnal compound which has one valence electron less than co@xmath0mnsi .
the extra electron in the the latter alloy occupies majority states leading to an increase of the exchange splitting between the occupied majority and the unoccupied minority states and thus to larger gap - width for the si - based compound . in the case of al - based alloy the bonding and antibonding minority d - hybrids almost overlap and the gap is substituted by a region of very small minority density of states ( dos ) ; we will call it a pseudogap . in both cases
the fermi level falls within the gap ( co@xmath0mnsi ) or the pseudogap ( co@xmath0mnal ) and an almost perfect spin - polarization at the fermi level is preserved .
[ mn@xmath11cr@xmath12al and co@xmath0[mn@xmath11fe@xmath12al for three values of the doping concentration @xmath13 .
dos s are compared to the one of the undoped co@xmath0mnal alloy .
details as in figure [ fig1 ] .
[ fig2 ] ] we substitute either fe or cr for mn to simulate the doping by electrons and holes , respectively , in co@xmath0mnal . in figure [ fig2 ] we present the total density of states ( dos ) for the co@xmath0[mn@xmath11(fe or cr)@xmath12al alloys to compare to the perfect co@xmath0mnal alloys . as was the case also for the compounds in reference @xcite and discussed above the majority - spin occupied states
form a common mn - co band while the occupied minority states are mainly located at the co sites and minority unoccupied at the mn sites ( note that the minority unoccupied states near the gap are co - like but overall the mn - weight is dominant ) .
the situation is reversed with respect to the co@xmath0mnsi compound and cr - doping has significant effects on the pseudogap .
its width is larger with respect to the perfect compound and becomes slightly narrower as the degree of the doping increases .
fe - doping on the other hand almost does not change the dos around the fermi level . the extra - electrons occupy high - energy lying antibonding majority states but since co@xmath0mnal has one valence electron less than co@xmath0mnsi half - metallicity remains energetically favorable and no important changes occur upon fe - doping and further substitution of fe for mn should retain the half - metallicity even for the co@xmath0feal compound although lda - based ab - initio calculations predict that the limiting case of co@xmath0feal is almost half - metallic @xcite .
r|c|cccc & & + @xmath13 & ideal & total & co & mn & al + -0.20 & 3.40 & 3.26 & 1.09 & 2.89 & -0.12 + -0.10 & 3.70 & 3.64 & 1.22 & 2.84 & -0.13 + -0.05 & 3.85 & 3.83 & 1.29 & 2.83 & -0.13 + 0.00 & 4.00 & 4.04 & 1.36 & 2.82 & -0.14 + 0.05 & 4.15 & 4.22 & 1.40 & 2.81 & -0.14 + 0.10 & 4.30 & 4.40 & 1.44 & 2.81 & -0.14 + 0.20 & 4.60 & 4.80 & 1.54 & 2.81 & -0.15 + & & + @xmath13 & ideal & total & co & mn & al + -0.20 & 4.40 & 4.40 & 1.92 & 3.19 & -0.06 + -0.10 & 4.70 & 4.70 & 1.95 & 3.15 & -0.08 + -0.05 & 4.85 & 4.85 & 1.96 & 3.14 & -0.08 + 0.00 & 5.00 & 5.00 & 1.96 & 3.13 & -0.09 + 0.05 & 5.15 & 5.15 & 1.99 & 3.10 & -0.10 + 0.10 & 5.30 & 5.30 & 2.00 & 3.09 & -0.10 + 0.20 & 5.60 & 5.60 & 2.03 & 3.05 & -0.11 + finally we shall briefly discuss the case of disorder simulated by the excess of the mn or the @xmath15 atoms . in table [ table1 ] we have gathered the total and atomic spin moments for all cases under study . substituting 5% , 10% , 15% or 20% of the mn atoms by al or si ones , corresponding to the negative values of @xmath13 in the table , results in a decrease of 0.15 , 0.30 , 0.45 and 0.60 of the total number of valence electrons in the cell , while the inverse procedure results to a similar increase of the mean value of the number of valence electrons .
contrary to the si compound which retains the perfect half - metallicity , the al - based compound is no more half - metallic . in the case of co@xmath0mnsi disorder induces states at the edges of the gap keeping the half - metallic character but this is no more the case for co@xmath0mnal compound where no real gap exists @xcite .
in the previous section we have examined the case of defects in half - metallic ferromagnets .
but the ideal case for applications would be a half - metallic antiferromagnet ( hma ) , also known as fully - compensated ferrimagnet @xcite , since such a compound would not give rise to stray flux and thus would lead to smaller energy consumption in devices . also half - metallic ferrimagnetism ( hmfi ) is highly desirable in the absence of hma since such compounds would yield lower total spin moments than the corresponding ferromagnets .
well - known hmfi are the perfect heusler compounds femnsb and mn@xmath0val @xcite .
we will present in this section another route to half - metallic ferrimagnetism based on antisites created by the migration of cr(mn ) atoms at co sites in the case of co@xmath0cral , co@xmath0crsi , co@xmath0mnal and co@xmath0mnsi alloys . for the [ co@xmath11cr@xmath19@xmath0cral ( upper panel ) and [ co@xmath11cr@xmath19@xmath0crsi ( lower panel ) compounds .
[ fig_hmfi ] ] lcccccc + @xmath13 & co & cr(imp ) & cr & al & total & ideal + 0 & 0.73 & & 1.63 & -0.09&3.00 & 3.00 + 0.05 & 0.71 & -1.82 & 1.62 & -0.09 & 2.70 & 2.70 + 0.1 & 0.69 & -1.85 & 1.61 & -0.08 & 2.40 & 2.40 + 0.2 & 0.64 & -1.87 & 1.60 & 0.06 & 1.80 & 1.80 + + @xmath13 & co & cr(imp ) & cr & si & total & ideal + 0 & 0.95 & & 2.17 & -0.06 & 4.00 & 4.00 + 0.05 & 0.93 & -1.26 & 2.12 & -0.06 & 3.70 & 3.70 + 0.1 & 0.91 & -1.26 & 2.07 & -0.05 & 3.40 & 3.40 + 0.2 & 0.87 & -1.25 & 1.96 & -0.04 & 2.80 & 2.80 + + @xmath13 & co & mn(imp ) & mn & al & total & ideal + 0 & 0.68 & & 2.82 & -0.14 & 4.04 & 4.00 + 0.05 & 0.73 & -2.59 & 2.82 & -0.13 & 3.81 & 3.80 + 0.1 & 0.78 & -2.49 & 2.83 & -0.12 & 3.61 & 3.60 + 0.2 & 0.84 & -2.23 & 2.85 & -0.09 & 3.20 & 3.20 + + @xmath13 & co & mn(imp ) & mn & si & total & ideal + 0.0 & 0.98 & & 3.13 & -0.09 & 5.00 & 5.00 + 0.05 & 0.99 & -0.95 & 3.09 & -0.08 & 4.80 & 4.80 + 0.1 & 0.99 & -0.84 & 3.06 & -0.07 & 4.60 & 4.60 + 0.2 & 0.97 & -0.70 & 2.99 & -0.05 & 4.20 & 4.20 + we will start our discussion from the cr - based alloys and using co@xmath0cral and co@xmath0crsi as parent compounds we create a surplus of cr atoms which sit at the perfect co sites . in figure [ fig_hmfi ] we present the total density of states ( dos ) for the [ co@xmath11cr@xmath19@xmath0cral and [ co@xmath11cr@xmath19@xmath0crsi alloys for concentrations @xmath13= 0 and 0.1 , and in table [ table_hmfi ] we have gathered the spin moments for the two compounds under study .
we will start our discussion from the dos .
the perfect compounds show a gap in the minority - spin band and the fermi level falls within this gap and thus the compounds are half - metals . when the sp atom is si instead of al the gap is larger due to the extra electron which occupies majority states of the transition metal atoms @xcite and increases the exchange splitting between the majority occupied and the minority unoccupied states .
this electron increases the cr spin moment by @xmath200.5 @xmath18 and the moment of each co atom by @xmath200.25 @xmath18 about .
the cr and co majority states form a common band and the weight at the fermi level is mainly of cr character .
the minority occupied states are mainly of co character .
when we substitute cr for co , the effect on the atomic dos of the co and cr atoms at the perfect sites is marginal .
the dos of the impurity cr atoms has a completely different form from the cr atoms at the perfect sites due to the different symmetry of the site where they sit .
but although cr impurity atoms at the antisites induce minority states within the gap , there is still a tiny gap and the fermi level falls within this gap keeping the half - metallic character of the parent compounds even when we substitute 20% of the co atoms by cr ones .
the discussion above on the conservation of the half - metallicity is confirmed when we compare the calculated total moments in table [ table_hmfi ] with the values predicted by the slater pauling rule for the ideal half - metals ( the total spin moment in @xmath18 is the number of valence electrons in the unit cell minus 24 ) @xcite .
since cr is lighter than co , substitution of cr for co decreases the total number of valence electrons and the total spin moment should also decrease .
the interesting point is the way that the reduction of the total spin moment is achieved .
co and cr atoms at the perfect sites show a small variation of their spin magnetic moments with the creation of defects and the total spin moment is reduced due to the antiferromagnetic coupling between the cr impurity atoms and the co and cr ones at the ideal sites , which would have an important negative contribution to the total moment as confirmed by the results in table [ table_hmfi ] .
thus the cr - doped alloys are half - metallic ferrimagnets and their total spin moment is considerable smaller than the perfect half - metallic ferromagnetic parent compounds ; in the case of [ co@xmath21cr@xmath22@xmath0cral it decreases down to 1.8 @xmath18 from the 3 @xmath18 of the perfect co@xmath0cral alloy . here we have to mention that if also co atoms migrate to cr sites ( case of atomic swaps ) the half - metallity is lost , as it was shown by miura et al .
@xcite , due to the energy position of the co states which have migrated at cr sites .
mn@xmath19@xmath0mnal and [ co@xmath11mn@xmath19@xmath0mnsi alloys as a function of the concentration @xmath13 : we denote @xmath23 with the solid grey line with shaded region and @xmath24 with a dashed thick blue line . in the insets we present the dos for a wider energy region and for a @xmath23 concentration . ] in the rest of this section we will present results on the [ co@xmath11mn@xmath12@xmath0mnz compounds varying the sp atom , z , which is one of al or si .
we have taken into account five different values for the concentration @xmath13 ; @xmath13= 0 , 0.025 , 0.05 , 0.1 , 0.2 . in figure [ fig_hmfi2 ]
we have drawn the total density of states ( dos ) for both families of compounds under study and for two different values of the concentration @xmath13 : the perfect compounds ( @xmath13=0 ) and for one case with defects , @xmath13= 0.1 . in the case of the perfect co@xmath0mnsi compound
there is a real gap in the minority spin band and the fermi level falls within this gap and this compound is a perfect half - metal .
co@xmath0mnal presents in reality a region of tiny minority - spin dos instead of a real gap , but the spin - polarization at the fermi level only marginally deviates from the ideal 100% and this alloy can be also considered as half - metal .
these results agree with previous electronic structure calculations on these compounds @xcite .
when we create a surplus of mn atoms which migrate at sites occupied by co atoms in the perfect alloys , the gap persists and both compounds retain their half - metallic character as occurrs also for the cr - based alloys presented above . especially for co@xmath0mnsi
, the creation of mn antisites does not alter the width of the gap and the half - metallicity is extremely robust in these alloys with respect to the creation of mn antisites .
the atomic spin moments show behavior similar to the cr alloys as can be seen in table [ table_hmfi ] and the spin moments of the mn impurity atoms are antiferromagnetically coupled to the spin moments of the co and mn atoms at the perfect sites resulting to the desired half - metallic ferrimagnetism .
now , we will present a way to use the defects in perfect half - metallic ferrimagnets in order to create a half - metallic antiferromagnetic material : the doping with co of the mn@xmath0val and mn@xmath0vsi alloys which are well known to be hmfi .
the importance of this route stems from the existence of mn@xmath0val in the heusler @xmath1 phase as shown by several groups @xcite .
each mn atom has a spin moment of around -1.5 @xmath18 and v atom a moment of about 0.9 @xmath18 @xcite .
all theoretical studies on mn@xmath0val agree on the half - metallic character with a gap at the spin - up band instead of the spin - down band as for the other half - metallic heusler alloys @xcite .
prior to the presentation of our results we have to note that due to the slater - pauling rule @xcite , these compounds with less than 24 valence electrons have negative total spin moments and the gap is located at the spin - up band .
moreover the spin - up electrons correspond to the minority - spin electrons and the spin - down electrons to the majority electrons contrary to the other heusler alloys @xcite .
we have substituted co for mn in mn@xmath0v(al or si ) in a random way and in figure [ fig_hma ] we present the total and atom - resolved density of states ( dos ) in [ mn@xmath11co@xmath12@xmath0val ( solid black line ) and [ mn@xmath11co@xmath12@xmath0vsi ( blue dashed line ) alloys for @xmath13=0.1 .
the perfect compounds show a region of low spin - up dos ( we will call it a `` pseudogap '' ) instead of a real gap . upon doping the pseudogap at the spin - up band persists and the quaternary alloys keep the half - metallic character of the perfect mn@xmath0val and mn@xmath0vsi compounds .
co atoms are strongly polarized by the mn atoms since they occupy the same sublattice and they form co - mn hybrids which afterwards interact with the v and al or si states @xcite .
the spin - up co states form a common band with the mn ones and the spin - up dos for both atoms has similar shape .
mn atoms have less weight in the spin - down band since they accommodate less charge than the heavier co atoms .
co@xmath25@xmath0val and [ mn@xmath26co@xmath25@xmath0vsi compounds .
note that the atomic dos s have been scaled to one atom .
positive values of dos correspond to the spin - up ( minority ) electrons while negative values correspond to the spin - down ( majority ) electrons .
[ fig_hma ] ] in table [ table_hma ] we have gathered the total and atom - resolved spin moments for all the co - doped compounds as a function of the concentration .
we have gone up to a concentration which corresponds to 24 valence electrons in the unit cell , thus up to @xmath13=0.5 for the [ mn@xmath11co@xmath12@xmath0val and x=0.25 for the [ mn@xmath11co@xmath12@xmath0vsi alloys . in the last column
we have included the total spin moment predicted by the slater - pauling rule for the perfect half - metals @xcite .
a comparison between the calculated and ideal total spin moments reveals that all the compounds under study are half - metals with very small deviations due to the existence of a pseudogap instead of a real gap .
exactly for 24 valence electrons the total spin moment vanishes as we will discuss in the next paragraph .
co atoms have a spin moment parallel to the v one and antiparallel to the mn moment , and thus the compounds retain their ferrimagnetic character . as we increase the concentration of the co atoms in the alloys , each co has more co atoms as neighbors , it hybridizes stronger with them and its spin moment increases while the spin moment of the mn atom decreases ( these changes are not too drastic ) .
the sp atoms have a spin moment antiparallel to the mn atoms as already discussed in reference @xcite .
lcccccc + @xmath13 & mn & co & v & al & total & ideal + 0 & -1.573 & & 1.082 & 0.064 & -2.000 & -2.0 + 0.05 & -1.580 & 0.403 & 1.090 & 0.073 & -1.799 & -1.8 + 0.1 & -1.564 & 0.398 & 1.067 & 0.069 & -1.600 & -1.6 + 0.3 & -1.484 & 0.456 & 0.953 & 0.047 & -0.804 & -0.8 + 0.5 & -1.388 & 0.586 & 0.782 & 0.019 & @xmath200 & 0 + + @xmath13 & mn & co & v & si & total & ideal + 0 & -0.960 & & 0.856 & 0.063&-1.000 & -1.0 + 0.05 & -0.944 & 0.749 & 0.860 & 0.059 & -0.800 & -0.8 + 0.1 & -0.925 & 0.819 & 0.847 & 0.054 & -0.600 & -0.6 + 0.2 & -0.905 & 0.907 & 0.839 & 0.046 & -0.201 & -0.2 + 0.25&-0.899 & 0.935 & 0.839 & 0.041 & @xmath200 & 0 + the most interesting point in this substitution procedure is revealed when we increase the co concentration to a value corresponding to 24 valence electrons in the unit cell , thus the [ mn@xmath27co@xmath27]@xmath0val and [ mn@xmath29co@xmath30@xmath0vsi alloys .
the slater - pauling rule predicts for these compounds a zero total spin moment in the unit cell and the electrons population is equally divided between the two spin - bands .
our first - principles calculations reveal that this is actually the case .
the interest arises from the fact that although the total moment is zero , these two compounds are made up from strongly magnetic components .
mn atoms have a mean spin moment of @xmath20 - 1.4 @xmath18 in [ mn@xmath27co@xmath27]@xmath0val and @xmath20 - 0.9 @xmath18 in [ mn@xmath29co@xmath30@xmath0vsi .
co and v have spin moments antiferromagnetically coupled to the mn ones which for [ mn@xmath27co@xmath27]@xmath0val are @xmath200.6 and @xmath200.8 @xmath18 , respectively , and for [ mn@xmath29co@xmath30@xmath0vsi @xmath200.9 and @xmath200.8 @xmath18 .
thus these two compounds are half - metallic fully - compensated ferrimagnets or as they are best known in literature half - metallic antiferromagnets .
finally we will also shortly discuss a special case : the occurence of vacancies @xcite . in experiments
a vacancy can appear in all possible sites and we will start our discussion by the case of vacancies ( e ) appearing at the sites occupied by co atoms at the perfect compounds . in figure [ vac_fig ]
we have gathered the total density of states ( dos ) for the four studied compounds , [ co@xmath11e@xmath12@xmath0yz ( y is cr or mn and z is al or si ) when we replace 2.5% ( x=0.025 ) and 10% ( x=0.10 ) of the co atoms . although there is no visible effect on the properties of the gap with respect to the perfect compound @xcite when @xmath13 is as small as 0.025 , as we increase @xmath13 to 0.1 impurity states located at the left edge of the gap appear .
the vacancy carries very small charge with a vanishing dos and the induced minority states are mainly located at the co sites ( mn and si atoms have a very small dos at the same energy range as a result of the polarization of these co - located vacancy - induced minority - states ) .
these states result in a considerable shrinking of the width of the gap of the order of 0.25 - 0.3 ev . in the case where the fermi level is close to the left edge of the gap as in the co@xmath0mnal alloy the shrinking of the gap results in the loss of half - metallicity ( the spin - polarization is reduced to @xmath20 40% for @xmath24 ) . for the four studied alloys [ co@xmath11e@xmath19@xmath0yz where y stands for cr and mn , z for al and si and e the vacant site .
[ vac_fig ] ] in the case when the vacancies occur at the y site occupied in the perfect alloys by cr or mn atoms and the z site occupied by al or si , they again induce minority states within the left edge of the gap leading to a shrinking of its width as in the case discussed in the previous paragraph .
but the effect is much more mild leading to a shrinking of the gap by 0.1 ev approximately when they substitute y - type atoms and even less when they substitute the sp atoms .
moreover , not only the reduction of the width of the gap is smaller , but also the intensity of the minority - spin induced dos within the gap is smaller with respect to the case discussed in the previous paragraph .
this difference in the behavior of the vacancy - induced minority states has its routes to the fact that the gap is created between states exclusively localized at the co sites which due to symmetry reasons do not couple to cr(mn ) or al(si ) orbitals .
when the vacancy substitutes co atoms the effect is much more intense around the fermi level than when it substitutes cr(mn ) or al(si ) atoms which have almost zero weight around the gap .
this is also reflected on the change of the behavior of the atomic spin moments ( see reference @xcite ) for an extended discussion of spin moments ) .
in this chapter we have reviewed our results on the defects in half - metallic heusler alloys .
firstly we have studied the effect of doping and disorder on the magnetic properties of the co@xmath0mnal(si ) full - heusler alloys .
doping simulated by the substitution of cr and fe for mn overall keeps the half - metallicity .
both disorder and doping have little effect on the half - metallic properties of co@xmath0mnsi and it keeps a high degree of spin - polarization .
co@xmath0mnal presents a region of low minority density of states instead of a real gap .
doping keeps the half - metallicity of co@xmath0mnal while disorder simulated by excess of either the mn or al atoms completely destroys the almost perfect spin - polarization of the perfect compound contrary to co@xmath0mnsi .
afterwards , we have studied the effect of defects - driven appearance of half - metallic ferrimagnetism in the case of the co@xmath0cr(mn)al(si ) heusler alloys .
more precisely , based on first - principles calculations we have shown that when we create cr(mn ) antisites at the co sites , these impurity cr(mn ) atoms couple antiferromagnetically with the co and the cr(mn ) atoms at the perfect sites while keeping the half - metallic character of the parent compounds .
this is a promising alternative way to create robust half - metallic ferrimagnets , which are crucial for magnetoelectronic applications .
moreover we have also studied the effect of doping the half - metallic ferrimagnets mn@xmath0val and mn@xmath0vsi .
co substitution for mn keeps the half - metallic character of the parent compounds and when the total number of valence electrons reaches the 24 , the total spin moment vanishes as predicted by the slater - pauling rule and the ideal half - metallic antiferromagnetism is achieved .
thus we can create half - metallic antiferromagnets simply by introducing co atoms in the mn@xmath0val and mn@xmath0vsi half - metallic ferrimagnets .
since crystals and films of both mn@xmath0val and co@xmath0val alloys have been grown experimentally , such a compound maybe is feasible experimentally .
finally , we have studied the effect of vacancies in half - metallic heusler alloys .
we have shown using ab - initio electronic structure calculations that the occurrence of vacancies at the sites occupied by co atoms in the perfect compounds seriously affects the stability of their half - metallic character and the minority - spin gap is rapidly shrinking . on the other hand vacancies at the other sites do not have an important impact on the electronic properties of the half - metallic full - heusler compounds .
thus it is crucial for spintronic applications to prevent creation of vacancies during the growth of the half - metallic full - heusler alloys used as electrodes in the magnetoelectronic devices .
although we have presented several aspects of defects in heusler alloys , many more calculations and experiments are needed ; the aim is to find systems , which either do not lead to states in the gap ( like the defects - driven half - metallic ferrimagnets presented in sections [ sec3 ] and [ sec4 ] ) or systems with particularly high defect formation energies or sufficiently low annealing temperatures ( we have not discussed the energetics of defects in this contribution ) . equally important for realistic applications
is the control of surface and interface states in the gap , the latter are in particular important for interfaces to semiconductors , which we have not been examined in this chapter .
webster and k.r.a .
ziebeck . in : _ alloys and compounds of @xmath31-elements with main group elements .
_ , landolt - brnstein , new series , group iii , vol 19c , ed by h.r.j .
wijn , ( springer , berlin 1988 ) pp 75184 | half - metallic ferromagnets and especially the full - heusler alloys containing co are at the center of scientific research due to their potential applications in spintronics . for realistic devices it is important to control accurately the creation of defects in these alloys .
we review some of our late results on the role of defects and impurities in these compounds .
more precisely we present results for the following cases ( i ) doping and disorder in co@xmath0cr(mn)al(si ) alloys , ( ii ) half - metallic ferrimagnetism appeared due to the creation of cr(mn ) antisites in these alloys , ( iii ) co - doping in mn@xmath0val(si ) alloys leading to half - metallic antiferromagnetism , and finally ( iv ) the occurrence of vacancies in the full - heusler alloys containing co and mn .
these results are susceptible of encouraging further theoretical and experimental research in the properties of these compounds .
_ emails : [email protected] , [email protected] , [email protected]_ |
let us consider a multiparticle interacting system with the internal symmetry taken into account .
the internal symmetry leads to conservation laws which put constraints on the evolution of system .
the problem arises if are there only constraints due to the symmetry conservation ? in the microscopic formulation with symmetry invariant dynamical equations the answer is given by an analysis of corresponding solutions assuming that solutions are known .
we are looking here for global conditions to provide consistency with the overall symmetry of the system .
these conditions should not depend on the exact analytic form of the solutions .
as an example one can take kepler s laws in the classical mechanics which are related to the orbital momentum conservation and can be proved without knowledge of the analytic solutions of newton equations .
another simple example is a case of @xmath0 species of charged particles with individual charges @xmath1 numbers of particles a given by @xmath2 particle numbers are time - dependent but the global charges must be conserved ( exact @xmath3 symmetry ) .
so there is a condition @xmath4 valid for any charge conserving interaction .
our aim@xcite is to find a corresponding condition for non - abelian symmetries . in the non - abelian case
there is a subsidiary condition besides conditions of the type eq .
( [ abcharge ] ) due to the charge conservation .
this is the demand to preserve the internal symmetry group representations during the evolution of the system .
let us consider a system big enough to use methods of statistical physics .
when the system reaches the statistical equilibrium then one can extract contributions from particular irreducible representation of the symmetry group @xcite .
group projection techniques allowed for a consistent treatment of equilibrium systems and gave tools to obtain canonical partition functions corresponding to the system transforming under given representation of the symmetry group .
this technique can also be used for a more general non - static problem .
let us consider a system consisting of particles belonging to multiplets @xmath5 of the symmetry group .
particles from the given multiplet @xmath5 are characterized by quantum numbers @xmath6 related to the symmetry group , and quantum numbers @xmath7 characterizing different multiplets of the same irreducible representation @xmath5 .
the number of particles of the specie @xmath8 is denoted here by @xmath9 these occupation numbers are time dependent until the system reaches the chemical equilibrium .
however , the representation of the symmetry group for the system remains constant in the course of a time evolution .
a multiplicity @xmath10 of the representation @xmath5 in this product is equal to a number of particles which transform under this representation @xmath11 we introduce a state vector @xmath12 in particle number representation .
the probability that @xmath13 particles transforming under the symmetry group representations @xmath14 combine into a state transforming under representation @xmath15 of the symmetry group is given by @xmath16 the projection operator @xmath17 has the form ( see _ e.g. _ @xcite ) @xmath18 here @xmath19 is the character of the representation @xmath15 , @xmath20 is the dimension of the representation , @xmath21 is the invariant haar measure on the group , and @xmath22 is an operator transforming a state under consideration . in particle number representation
the operator @xmath22 is defined as @xmath23 @xmath24 is a matrix elements of the group element @xmath25 corresponding to the representation @xmath26 .
one gets finally @xmath27^{n^{(\alpha_1)}_{\nu_{\alpha_1 } } } \cdots [ d^{(\alpha_n)}_{\nu_n\nu_n}]^{n^{(\alpha_n)}_{\nu_{\alpha_n}}}\,.\nonumber\end{aligned}\ ] ] @xmath24 is a matrix elements of the group element @xmath25 corresponding to the representation @xmath26 and @xmath28 is an overall permutation normalization factor @xmath29 where @xmath30 are partial factors for particles of the kind @xmath31 @xmath32 the permutation factor gives a proper normalization of state vectors reflecting indistinguishability of particles @xmath33 because of the symmetry conservations all weights in eq.([weights ] ) should be constant @xmath34 introducing here the result of eq.([weights ] ) one obtains @xmath35^{n^{(\alpha_1)}_{\nu_{\alpha_1}}}\cdots [ d^{(\alpha_n)}_{\nu_n\nu_n}]^{n^{(\alpha_n)}_{\nu_{\alpha_n}}}\\ & + & \sum_{j=1}^n \sum_{\nu_{\alpha_j}}\ , \frac{d\,n^{(\alpha_j)}_{\nu_{\alpha_j}}}{dt}\int\limits_g d\mu(g)\bar\chi^{(\lambda)}(g ) [ d^{(\alpha_1)}_{\nu_1\nu_1}]^{n^{(\alpha_1)}_{\nu_{\alpha_1 } } } \cdots [ d^{(\alpha_n)}_{\nu_n\nu_n}]^{n^{(\alpha_n)}_{\nu_{\alpha_n } } } \log[d^{(\alpha_j)}_{\nu_j\nu_j}]\nonumber\,.\end{aligned}\ ] ] the integrals which appear in eq.([deriv ] ) can be expressed explicitly in an analytic form for any compact symmetry group .
to write an expression for the time derivative of the normalization factor @xmath36 we perform analytic continuation from integer to continuous values of variables @xmath37 all factorials in eq .
( [ faktor ] ) are replaced by the @xmath38function of corresponding arguments .
we encounter here also the digamma function @xmath39 @xcite @xmath40 this allows to write @xmath41\,.\end{gathered}\ ] ] eq.([deriv ] ) can be written in a form @xmath42 where @xmath43^{n^{(\alpha_1)}_{\nu_{\alpha_1 } } } \cdots [ d^{(\alpha_n)}_{\nu_n\nu_n}]^{n^{(\alpha_n)}_{\nu_{\alpha_n}}}\ ] ] is analytically extended for continuous values of variables @xmath44 this gives @xmath45^{n^{(\alpha_1)}_{\nu_{\alpha_1 } } } \cdots [ d^{(\alpha_n)}_{\nu_n\nu_n}]^{n^{(\alpha_n)}_{\nu_{\alpha_n } } } \log[d^{(\alpha_j)}_{\nu_j\nu_j}]}{\int\limits_g d\mu(g)\bar\chi^{(\lambda)}(g ) [ d^{(\alpha_1)}_{\nu_1\nu_1}]^{n^{(\alpha_1)}_{\nu_{\alpha_1 } } } \cdots [ d^{(\alpha_n)}_{\nu_n\nu_n}]^{n^{(\alpha_n)}_{\nu_{\alpha_n}}}}\,.\end{gathered}\ ] ] eq .
( [ derivp ] ) give a set of conditions related to the internal symmetry of a system .
they are meaningful only for nonzero values of coefficients ( [ weights ] ) .
it is easy to see that coefficients @xmath46 are different from zero only if parameters @xmath47 are consistent with the conservation of the simultaneously measurable charges related to the symmetry group .
a number of such charges is equal to the rank @xmath48 of the symmetry group . for the isospin @xmath49 group that is the third component of the isospin , for the flavor @xmath50 that would be the third component of the isospin and the hypercharge
. in general case one has @xmath48 linear relations between variables @xmath51 what reduces correspondingly the number of independent variables .
let consider a case of isotopic symmetry in a more detailed way .
diagonal matrix elements for the representation @xmath52 in the euler s angles representation have the form @xcite
@xmath53 where @xmath54 and @xmath55 are jacobi polynomials @xmath56\ ] ] the measure @xmath21 for the @xmath49 group in this parametrization has the form @xmath57 = { 1\over{8\pi^2 } } \int\limits^{2\pi}_0d\alpha \int\limits^{2\pi}_0d\gamma \int\limits^{\pi}_0d\beta\sin\beta f[g(\alpha,\beta,\gamma)]\,.\end{aligned}\ ] ] there are only three possible nontrivial hadronic isotopic multiplets . these are spinor @xmath58 , vector @xmath59 , and @xmath60 representations .
corresponding @xmath61 functions are [ dfunctions ] @xmath62 the group theoretic factor ( [ otherp ] ) has the form @xmath63^{n^{(j_1)}_{m_{j_1}}}\cdots [ d^{(j_n)}_{m_{j_n}m_{j_n}}]^{n^{(j_n)}_{m_{j_n}}}\nonumber}\\ & = & \frac{1}{8\pi^2}\sum_{m =- j}^j\int\limits_0^{2\pi}d\alpha\int\limits_0^{2\pi}d\gamma\int\limits_0^\pi d\beta\sin\beta\,{\text{e}}^{i ( n^{(j_1)}_{m_{j_1}}m_{j_1}+\cdots+n^{(j_n)}_{m_{j_n}}m_{j_n}-m)(\alpha+\gamma)}\nonumber\\ & & \times d^{(j)}_{mm}(\cos\beta)\left[d^{(j_1)}_{m_{j_1}m_{j_1}}(\cos\beta)\right]^{n^{(j_1)}_{m_{j_1 } } } \cdots\left[d^{(j_n)}_{m_{j_n}m_{j_n}}(\cos\beta)\right]^{n^{(j_n)}_{m_{j_n}}}\,,\end{aligned}\ ] ] where @xmath64 is a number of particles with the isotopic coordinates @xmath65 .
the nonzero values are obtained only when the arguments of the exponent in eq .
( [ weight1 ] ) vanish @xmath66 for the hadronic system with the given value @xmath67 of the third component of the isospin the factor ( [ otherp ] ) is @xmath68^{n^{(j_1)}_{m_{j_1}}}\cdots [ d^{(j_n)}_{m_{j_n}m_{j_n}}]^{n^{(j_n)}_{m_{j_n}}}\nonumber}\\ & = & \frac{1}{8\pi^2}\,\delta_{\tilde m,\,n^{(j_1)}_{m_{j_1}}m_{j_1}+\cdots+n^{(j_n)}_{m_{j_n}}m_{j_n}}\int\limits_0^{2\pi}d\alpha\int\limits_0^{2\pi}d\gamma\int\limits_0^\pi d\beta\sin\beta\,\nonumber\\ & & \times d^{(j)}_{\tilde m \tilde m}(\cos\beta)\left[d^{(j_1)}_{m_{j_1}m_{j_1}}(\cos\beta)\right]^{n^{(j_1)}_{m_{j_1 } } } \cdots\left[d^{(j_n)}_{m_{j_n}m_{j_n}}(\cos\beta)\right]^{n^{(j_n)}_{m_{j_n}}}\,.\end{aligned}\ ] ] these equations allow to write general forms of the isotopic symmetry factor ( [ otherp ] ) for the hadronic system @xmath69 where @xmath70 the definition ( [ r ] ) and the constraint ( [ m ] ) allow to write @xmath71 this is because @xmath72 and @xmath73 this allows to write eq .
( [ genweight2 ] ) in the form @xmath74 one should notice that this reduction of independent variables is related only to that part of the total weight ( [ weights ] ) which comes from the group theory .
the condition ( [ cond ] ) is obtained when the expression ( [ genweight3 ] ) is multiplied by the combinatoric factor ( [ faktor ] ) .
number of independent variables in the combinatoric factor can be reduced only by conservation laws of corresponding charges of the system . to be more specific
let us consider as an example an iso - singlet system consisting only of pions and nucleons @xcite .
we take as independent variables : @xmath75 - total number of particles , @xmath76 - total baryon number , @xmath77 - total electric charge , @xmath78 - number of neutral pions , @xmath79 - number of neutrons
. then + number of negative pions : @xmath80 number of positive pions : @xmath81 number of protons : @xmath82 + for iso - singlet state @xmath83 and [ partnumb ] @xmath84 let us assume @xcite that this system reaches the chemical equilibrium in the evolution process governed by vlasov - boltzman kinetic equations with the interactions restricted to binary collisions .
then the _ total _ number of particles remains constant but particles ratios are subjected to constraints ( [ cond ] ) . for the given baryon number @xmath76 and the given total number of particles @xmath85 condition @xmath86 gives evolution lines in the @xmath87 plane .
the system evolves along these lines which are here consequences of the isotopic @xmath49 symmetry and baryon number conservation .
the weights ( [ condsingl ] ) expressed by means of variables @xmath88 and calculated according to eqs . ( [ faktorp ] ) and ( [ partnumb ] )
have the form @xmath89 the corresponding evolution lines are shown in fig . [ fig.1 . ] .
we have got relations which are _ necessary _ global conditions to provide consistency with the overall symmetry of the system .
they do not depend on the form of the underlying microscopic interaction .
abelian internal symmetries lead to simple and obvious linear relations as in eq.([abcharge ] ) .
non - abelian internal symmetries lead to nonlinear relations as in eq.([deriv ] ) .
if we knew the solutions of symmetry invariant evolution equations then all those constraints would become identities .
in other case they give a subsidiary information about the system and can be used as a consistency check for approximative calculations . a case of generalized vlasov - boltzman kinetic equations
was considered in @xcite .
new constraints lead to decreasing number of available states for the system during its evolution .
new correlations appear and the change in the thermodynamical behavior can be expected .
this can be observed in hydrodynamical systems formed in high energy heavy ion collisions @xcite .
a challenging point is to find structures which would correspond to chemical potentials when system approaches the equilibrium distribution .
the equilibrium distribution in the presence of constraints can be constructed by the lagrange multipliers method .
the multipliers related to the `` abelian '' constraints , such as eq.([abcharge ] ) , are well known chemical potentials .
multipliers related to the non - abelian " constraints ( [ deriv ] ) are more complicated . because these constraints are nonlinear ones , corresponding multipliers can not be treated as standard additive thermodynamical potentials .
work supported in part by the polish committee for scientific research under contract kbn - 2p03b 030 18 99 l. turko and j. rafelski , _ epj c _ * 18 * , 587 , ( 2001 ) .
+ l. turko and j. rafelski,_macroconstraints from microsymmetries of macrosystems _ , [ hep - th/0011047 ] k. redlich and l. turko , _ z. phys .
_ * c5 * , 201 , ( 1980 ) .
+ l. turko , _ phys .
_ b * 104 * , 153 , ( 1981 ) .
e. p. wigner , _ group theory and its application to the quantum mechanics of atomic spectra _ , ( academic press , new york and london , 1959 ) .
a. r. edmonts , _ angular momentum in quantum mechanics _ , princeton , 1957 m. abramowitz and i. a. stegun ( eds . ) , _ handbook of mathematical functions _ , ( national bureau of standards , applied mathematics series @xmath90 55 , 1964 ) .
elze , j. rafelski and l. turko . , _ phys .
_ * b506 * , 123 ( 2001 ) | internal microscopic symmetry of a many body system leads to global constraints .
we obtain explicit forms of the global macroscopic condition assuring that at the microscopic level the evolution respects the overall symmetry .
pacs numbers : 11.30.-j , 24.10.pa , 25.75gz |
heavy quarkonium or baryon has attracted wide attention due to its special features , which provides a good platform for studying the perturbative qcd and the associated non - perturbative physics in the bound state system [ 1 - 4 ] .
for example , the @xmath0 meson and the @xmath1 baryon can be a fruitful laboratory for testing various potential models and understanding the weak decay mechanism for heavy flavors .
systematic studies for the hadronic production of the doubly heavy quarkonium or baryon at the hadronic colliders as tevatron and lhc have been done in the literature [ 5 - 17 ] . in particular ,
two generators bcvegpy and genxicc have been completed and developed in recent years [ 18 - 24 ] , which can be conveniently implemented into pythia [ 25 ] for simulating the @xmath0 meson and @xmath1 baryon events with high efficiency . in the present paper ,
we present their main idea and their recent progresses .
it is noted that in the case of productions of double heavy baryons the diquark is remaining in the coloured state . in this case
the interactions with hadronic residue part may decrease the results of calculations due to disassociation effect [ 26 ] .
therefore , it is worth to mention , that the results of calculations of cross - sections of hadronic production of double heavy baryons are not absolutely identical to the case of production of @xmath0 meson and the obtained results for the calculations of the cross - sections of production of baryons should be considered with care and may be treated as the estimate of upper bound of the values of the cross - sections .
there are gluon - gluon fusion mechanism , or the light quark - anti - quark mechanism , or the extrinsic or intrinsic heavy quark mechanism for the hadronic production of the doubly heavy meson or baryon respectively .
all of which may provide sizable contribution in specific kinematic regions .
it is noted that in most of the kinematic regions ( e.g. large @xmath2 region ) , the dominant production mechanism is the gluon - gluon fusion mechanism , e.g. for the production of the @xmath3-quarkonium , it stands for the production via the process @xmath4+b+\bar{c}$ ] , where @xmath5 $ ] stands for the fock - states @xmath6\rangle$ ] , @xmath7g\rangle$ ] , @xmath8\rangle$ ] , @xmath9g\rangle$ ] , @xmath10\rangle$ ] and @xmath11\rangle$ ] ( with @xmath12 ) within the nonrelativistic qcd [ 27 ] , respectively .
the lowest fock state @xmath6\rangle$ ] corresponds to the usually called @xmath0 meson .
the cases for the hadronic production @xmath1 , @xmath13 and @xmath14 baryons are similar , which can be obtained from the case of @xmath0 meson production by properly transforming the anti - quark line to be the quark line and by properly dealing with the color flows [ 13 ] .
so , in the following , we shall take the gluon - gluon fusion mechanism for the @xmath0 meson production as an explicit example to explain our calculation technology and to show how to improve the events simulation efficiency .
according to the pqcd factorization , the hadronic production cross section is formulated as @xmath15 x}(x_{1},x_{2},\mu_{f},\mu_{r } ) , \label{pqcdf}\ ] ] where @xmath16 is the gluon distribution function in hadron @xmath17 , @xmath18 is the cross section for the relevant inclusive production ( @xmath19+b+\bar{c}$ ] ) . @xmath20 and @xmath21 stand for the factorization scale and renormalization scale respectively .
the generating of phase space and its integration can be done with the help of the routines rambos [ 28 ] and vegas [ 29 ] .
then , the most important part left is to deal with the hard scattering amplitude of the process . at the lowest @xmath22 order
, there are totally 36 feynman diagrams and hence 36 amplitudes for the gluon - gluon fusion @xmath19+b+\bar{c}$ ] . according to the pqcd factorization
, each amplitude can be factorized into two parts , that of perturbative @xmath23 ( all the quarks are on shell ) and that of the nonperturbative @xmath24 $ ] which can be represented by the universal matrix element for each fock state . using the conventional squared amplitude to deal with the amplitude
is a tedious and time - consuming task . to save the time for events simulation , we can deal with the hard scattering amplitude for @xmath25 directly at the amplitude level . in the literature , two ways
have been suggested , i.e. the improved trace technology [ 30 - 33 ] and the helicity amplitude approach [ 34,35 ] . in our generators , we adopt an improved version for the helicity amplitude approach as suggested in ref.[18 ] to simplify the hard scattering amplitude . .
] main idea of the improved helicity amplitude approach is shown in fig.(1 ) . according to the improved helicity amplitude approach [ 15 ] ,
we first pick out all the basic spinor lines , which are constructed by the @xmath26 structure of fermion line , from the whole amplitude , whose remaining parts include the color factors and the scalar part of the propagators that shall be dealt with separately . these basic spinor lines can be simplified by changing the involved space - like momenta to be light - like ones , and then use the relation @xmath27 ( @xmath28 is light - like ) to transform and simplify all the spinor line into expressions of spinor products .
this intermediate step can be done using the standard helicity amplitude approach , cf.refs.[34,35 ] .
next , we construct the basic feynman diagrams ( nine for the gluon - gluon fusion ) by proper combination of these basic spinor lines .
finally , we adopt the symmetries of the gluon - gluon exchanges , the quark line exchanges to construct all the amplitude of the process . one subtle point .
the authors of ref.[34 ] group the feynman diagrams of the process into gauge - invariant subsets . to achieve the goal , one have to introduce extra terms into the feynman rules for the three - gluon and four - gluon vertices , i.e. @xmath29 where @xmath30 , @xmath31 and etc .
are input momenta for the gluons , @xmath32 and @xmath33 are primary feynman rules @xmath34 and @xmath35 and @xmath36 are modified parts @xmath37 \\ g_{abcd}^{\lambda\mu\nu\delta}(p , q , k_{1},k_{2})&= & -f_{ace}f_{bde}\frac{s_{1}^{\lambda}s_{1}^{\mu}s_{1}^{\nu}s_{1}^{\delta } } { ( s_{1}\cdot k_{1})(s_{1}\cdot k_{2})}-f_{ade}f_{bce}\frac{s_{2}^{\lambda}s_{2}^{\mu } s_{2}^{\nu}s_{2}^{\delta}}{(s_{2}\cdot k_{1})(s_{2}\cdot k_{2})},\end{aligned}\ ] ] where @xmath38 and @xmath39 . because the modified part of the three gluon vertex is symmetric on the interchange of @xmath30 and @xmath31 , we must be careful to choose the sign of modified part in every associated diagrams so as to make all subsets be gauge invariant .
when adding all extra terms together they will result in the required zero contribution .
then by choosing different gauge for each subset , one can obtain a compact result , when all terms of the amplitude are written according to the helicities of the fermions .
this method is most effective for massless cases . while , for the present massive case ( the quark lines are massive ) , the question is much more involved .
it is noted that the sum of all extra terms involving @xmath35 and @xmath36 will have non - zero contributions , which complicates the final results .
so in the improved helicity amplitude approach , we adopt a unique gauge for the whole amplitude .
this gauge choice also avoids numerical cancelations between very large numbers and the numerical results are much more steady when doing the numerical calculations .
this gauge independence can be adopted as one critical point on checking the rightness of the generator .
another subtle point .
we can further make the amplitude more symmetric by applying a proper decomposition of the amplitude . as shown by fig.(2 ) , this is achieved by decomposing the terms involving three- and/or four - gluon vertices , into terms without self - interactions of gluons ( i.e. into qed - like amplitudes ) .
more over , it is noted that for the massive quark case , in addition to the qed - like feynman diagrams , some more additional qed - like terms need to be introduced .
fortunately , for the present subprocess @xmath23 , these ` extra ' terms are just parts of the diagrams with four - gluon vertices ; i.e. they are one of the three terms within the amplitude involving the four - gluon vertex .
this shows that the whole amplitude is just the recombination of all the basic functions already existed in the 36 feynman amplitudes , which is like a magic .
there are totally nine basic qed - like feynman diagrams for the present process .
thus , only the analytic expressions for those ` basic feynman diagrams ' need to be put in the program precisely , while the non - basic ones can be generated by means of linear combination ( the coefficients can be found in ref.[18 ] ) and/or by properly using the symmetry of all those basic ones . as a final subtle point .
as shown by fig.(3 ) , for the feynman diagrams involving two three - gluon vertices , there are usually two ways to transform them into the qed - like diagrams ; i.e. in the second step there are two ways to move the gluon either to the upper quark line or to the lower quark line respectively .
thus , two extra constraints can be found among the nine ` basic feynman diagrams ' , the program can be further simplified and become more compact .
the flowchart for bcvegpy , which is implemented into pythia , is shown in fig.(4 ) .
the bcvegpy ( or genxicc ) is a fortran programme written in a pythia - compatible format and is written in a modularization structure , one may apply it to various situations or experimental environments conveniently .
it is noted that the lhe format is a standard format that is proposed to store process and event information from the matrix - element - based generators .
one can pass these parton - level information to the event generator as pythia for further simulation . and
bcvegpy will generate a standard lhe data file [ 36 ] that contains useful information of the meson and its accompanying partons , which can be conveniently imported into pythia to do further hadronization and decay simulation .
the schematic structure for the generator bcvegpy is shown in fig.(5 ) . in general , the generator is systematically constructed in seven modules according to their purpose .
each module contains necessary files to fulfill the specific tasks for generating events .
the bcvegpy is dominated by two blocks , i.e. the _ vegas _ block ( in module * phase * ) and the _ event _ block ( in module * generate * ) .
the _ vegas _
block is to generate the importance sampling function .
the _ event _ block is to generate events by using pythia , in a way that all the mentioned processes are implemented into pythia as its external processes .
this is achieved by properly setting the two pythia subroutines upinit and upevnt .
here we modify the _
block not only to generate the sampling importance function but also to store an upper bound of the value of the cross sections in each cell [ 37 - 40 ] ; i.e. an improved and more effective method to generate unweighted events has been programmed [ 21,24 ] based on the mint package [ 40 ] .
the importance sampling function is used to increase the simulation efficiency , while the upper bound value will be used to generate unweighted event if the user want to do the experimental analysis and further simulation .
the upper bound value in each cell is an upper bound for the cross sections and also equals a multidimensional stepwise function , according to which it is easy to generate phase - space points . by using this new hit - and - miss technique
, one can generate the points according to the original distribution .
moreover , in doing the initialization , we will call vegas twice to generate the upper bound grid xmax and also a more precise importance sampling function . in vegas
the integral together with its numerical error are related to the sampling numbers ncall and the iteration times itmx .
so , to generate full events , we suggest the user to do a test running first so as to find an effective and time - saving parameters for vegas .
finally , one can generate events by calling the upevnt subroutine according to the probability proportional to the integrand .
the coming lhc experiment shall provide a better platform to check all the theoretical predications and to learn @xmath0 , @xmath1 , @xmath13 and @xmath14 properties in more detail .
recently some new measurements on @xmath0 meson have been done by lhcb and cdf collaborations [ 41 - 45 ] , in which bcvegpy has been adopted for data analysis . due to their high running efficiency ( because of the using of improved helicity amplitude approach ) , bcvegpy and genxicc are very useful for monte carlo simulation and also for theoretical studies . at the present ,
they have been adopted by most of the groups at lhc and tevatron . the suggesting future super z factory , gigaz program , lep3 , and etc .
, shall also provide good platforms for doubly heavy meson and baryon productions , cf .
refs.[46 - 49 ] .
this talk was based on the works done in collaborations with profs.chang ch , wang jx , qiao cf and dr.wang xy .
this work was supported in part by the fundamental research funds for the central universities under grant no.cqdxwl-2012-z002 , by natural science foundation of china under grant no.11075225 and no.11275280 , and by the program for new century excellent talents in university under grant no.ncet-10-0882 . | doubly heavy mesons and baryons provide a good platform for testing pqcd .
two high efficient generators bcvegpy and genxicc for simulating the hadronic production of the doubly heavy mesons and baryons have been developed in recent years . in this talk , we present their main idea and their recent progresses . the dominant gluon - gluon fusion mechanism programmed in those two generators are written based on the improved helicity amplitude approach , in which the hard scattering amplitude are dealt with directly at the amplitude level and the numerical efficiency are greatly improved by properly decomposing the feynman diagrams and by fully applying the symmetries among them .
moreover , in comparison to the previous versions , we have updated the programs in order to generate the unweighted meson or baryon events much more effectively within various simulation environments .
the generators can be conveniently imported into pythia to do further hadronization and decay simulation . |
quantum entanglement has given rise to much work in the condensed matter community as a new way to explore interesting aspects of physical systems .
the kondo problem for instance has been revisited along these lines , with studies addressing the interplay between the impurity screening and the information shared between the impurity and the bath @xcite .
it is certainly reasonable to expect that entanglement together with other quantities inspired by quantum information theory , such as the loschmidt echo or the work distribution might shed new light on , and offer new experimental / numerical probes of , the key physical features of the kondo and other problems @xcite .
a particularly interesting question in this direction is whether the kondo screening cloud which has had so elusive an appearance in standard thermodynamic quantities @xcite might play a bigger role in quantum information aspects .
other aspects of interest in the context of 2 level systems interacting with gapless excitations generalizing the kondo problem apply to the decoherence of qubits interacting with the environment @xcite .
a large part of the work combining entanglement and quantum impurities has been numerical so far . indeed , apart from the scale invariant situations , where conformal invariance techniques have led to spectacular progress @xcite , the general situations involving crossover are very difficult to tackle .
this is mostly because the entanglement is a different kind of quantity , not amenable to simple bethe ansatz calculations , for instance
. there is , however , another reason for the relative lack of analytical results in this area : entanglement , being a zero temperature quantity , is naturally plagued by ir divergences , which make it non perturbative in the impurity strength . in that respect , it does behave somehow like some properties of the kondo screening cloud studied in @xcite . in order to clarify the main features of entanglement in the presence of impurities in particular its scaling properties , and flow from small to strong coupling
we focus in this paper on a couple of representative situations , which we handle by a mix of analytical and numerical techniques .
the lessons learned will be put to use in forthcoming papers , with applications of more direct physical interest .
the paper is organized as follows . in section [ sec : models ] we discuss the basic models we want to study , and define precisely the entanglement entropy . in section
[ sec : uvpert ] we put together the perturbative calculation of the entanglement at small coupling , and show that it is plagued by strong ir divergences . in section [ sec : irdiv ] we discuss this difficulty in a more general context . in section
[ sec : smallcoup ] we show how the non perturbative nature of the entanglement entropy can be obtained using general conformal field theoretic arguments . in section [ sec : largecoup ]
we recall the principles of the large coupling expansion proposed in @xcite and carried out to high order in @xcite .
when the dimension of the perturbation is @xmath1 , we develop in section [ sec : ff ] the form - factor approach using the results of @xcite , and obtain non perturbative approximations for the entanglement extrapolating all the way from the uv to the ir limit .
finally , in section [ sec : numerics ] we compare our results with those of exact numerical calculations on large spin chains .
the conclusion contains some last comments and prospect for future work .
finally , appendix [ sec : relsising ] contains a discussion of the equivalence between our impurity models when the dimension of the perturbation is @xmath1 to the boundary ising model with a boundary magnetic field at special values of the coupling .
the main problem we study in this paper though it has various , mathematically equivalent formulations , see below is the calculation of the entanglement of a region of length @xmath0 centered on an ` impurity ' in an otherwise one dimensional , gapless quantum system .
we characterize this entanglement by the von neumann entropy @xmath2 , where @xmath3 is the reduced density matrix that has been formed by tracing over the degrees of freedom outside of the segment of length @xmath0 .
an example of this setup is obtained by taking two semi infinite xxz chains coupled by a weak link : @xmath4 + \sum_{1}^\infty j\left[s_i^xs_{i+1}^x+s_i^ys_{i+1}^y+\delta s_i^zs_{i+1}^z\right]+j'\left[s_0^xs_{1}^x+s_0^ys_{1}^y+\delta s_0^zs_{1}^z\right]\label{firsth}.\ ] ] the bulk chains are in a gapless luttinger liquid phase for @xmath5 .
we shall consider the case of anisotropy @xmath6 , where the tunneling between the two half infinite chains is a relevant perturbation , and one observes healing at large scales . the case @xmath7 is exactly marginal .
we parametrize @xmath8 , @xmath9 $ ] .
we focus on the physics at energies much smaller than the band - width , where field theoretic results can be applied .
consider then the entanglement of a region of length @xmath0 centered on the modified link .
we can easily surmise what this entanglement will look like in the high and low energy limits from the existing literature .
indeed , at high energy , the system is effectively cut in half . using the well known formula for the entropy of a region _ on the edge _ of a conformal invariant system we have @xmath10={1\over 3}\ln { 2l\over \epsilon}+s_1+\ln g_{\rm uv},\ ] ] where we used that the central charge is unity , @xmath11 is a uv cutoff of the order of the lattice spacing @xmath12 , and @xmath13 where @xmath14 is the boundary entropy @xcite associated with the conformal boundary condition corresponding to an open xxz spin chain .
the remaining constant term @xmath15 is non - universal as it obviously depends on the definition of the cutoff @xmath11 . on the other hand , at low energy ,
healing has taken place , the system behaves just as one ordinary quantum spin chain , and the entropy obeys the general form for a region of length @xmath0 in the bulk of a conformal invariant system : @xmath16 here , we have allowed for a term @xmath17 , which can be thought of as a residual contribution of the weak link at low energy . in general , comparing entanglements for bulk and boundary theories is indeed difficult , since the dependency of the cutoff @xmath11 on the physical cutoff ( the lattice spacing in the spin chain @xmath12 , which is the same in both geometries ) is not universal , and not necessarily the same in the bulk and boundary cases .
this important aspects is discussed in detail in @xcite , see in particular section 6.2.1 in that reference .
the point for us is that the quantity @xmath18 is well defined , and its value @xmath19 can be easily obtained from the folded version of the system ( see eq .
( [ hamilbdr ] ) below ) .
more generally , since the bulk behavior of the entanglement entropy is not modified , it is natural to expect the existence of a scaling relation @xmath20 where the crossover scale , @xmath21 , is expected to be related to the coupling @xmath22 .
@xmath23 should be a monotonic function extrapolating between @xmath24 at small values of the argument and @xmath25 at large values . to proceed , and conveniently describe the field theory limit @xcite
, we first observe that the problem , at low energy , can be turned into a purely chiral one . indeed , in the low energy limit , each half chain is equivalent to a combination of l and r moving excitations , and we formally map via a canonical transformation the l moving sector into a r moving one so as to have two chiral ` wires ' representing the two half chains .
the additional tunneling between the two chains becomes , in this language , a hopping term between two chiral wires .
bosonizing , forming odd and even combinations of the bosons for each wire , one finds that the odd combination decouples , while for the even one obtains the simple hamiltonian , @xmath26 where @xmath27 is the conformal weight of the perturbation , and we have set the fermi velocity @xmath28 .
the dimension of the perturbation being @xmath29^{-\mu}$ ] , we see that @xmath30 .
one can also fold back this problem into the boundary sine - gordon model ( bsg ) with hamiltonian @xmath31 + \lambda \cos{\beta\over 2}\phi(0)\label{hamilbdr}.\ ] ] this shows equivalence to a large variety of other problems , including the one of tunneling between edge states in the fractional quantum hall effect ( fqhe ) @xcite . in this case
, @xmath32 is the filling fraction .
the rg flows from neumann ( @xmath33 ) to dirichlet ( @xmath34 ) boundary conditions ( bc ) , and the boundary entropy associated with these conformally invariant bc satisfy @xmath35 , as claimed earlier .
an interesting variant involves modifying two successive links on the chain : @xmath36 + \sum_{1}^\infty j\left[s_i^xs_{i+1}^x+s_i^ys_{i+1}^y+\delta s_i^zs_{i+1}^z\right]+\nonumber\\ & + & j'\left[s_{-1}^xs_{0}^x+s_{-1}^ys_{0}^y+\delta s_{-1}^zs_{0}^z+s_0^xs_{1}^x+s_0^ys_{1}^y+\delta s_0^zs_{1}^z\right].\end{aligned}\ ] ] this is equivalent to tunneling through a resonant level at the origin .
this time , the dimension of the tunneling operator is half what it is in the previous situation , @xmath22 is always relevant for @xmath5 , and the system is always healed at low energy . the same series of manipulations
` unfolding the two half chains ' , forming odd and even combinations , decoupling the odd one and bosonizing lead to the hamiltonian formulation @xmath37,\label{chiralk}\ ] ] where we recall that @xmath38 ( * post - publication footnote added * ) does not correspond to the xxz spin chain with two successive weak links as claimed here .
the bosonization of the xxz chain with two weak - links leads instead to a field theory with _ two _ chiral bosons ( one for each half - chain ) tunneling through a spin-@xmath39 impurity .
the dimension of the perturbation is also @xmath40 and the predictions of this paper can be applied to that case as well .
the anisotropic kondo hamiltonian ( [ chiralk ] ) can be realized on the lattice as an interacting resonant level model ( irlm ) , with two half - infinite chains of non - interacting fermions tunneling through a resonant level model with coulomb interactions on the dot only . in the xxz language
, this corresponds to an xx spin chain with two weak links , with anisotropy @xmath41 on the two weak links .
all the conclusions of this paper remain unchanged . ] ) . in this case
, the dimension of the perturbation is @xmath40 ( note the factor @xmath42 compared with the first case ) .
the problem can also be folded back into the anisotropic kondo problem @xmath43+\lambda \left[{\mathrm e}^{i{\beta\over2\sqrt{2}}\phi(0)}s^-+{\mathrm e}^{-i{\beta\over2\sqrt{2}}\phi(0)}s^+\right].\label{bdrk}\ ] ] of particular interest is the case @xmath7 , @xmath44 which corresponds to free fermions . while the chain with one weak link is marginal , the chain with two weak links describes an interesting flow , and is in fact equivalent to a widely studied problem that of the resonant level model ( rlm ) .
indeed , fermionization in this case leads to @xmath45 where we have redefined the couplings @xmath46 and @xmath47 .
when going to the continuum limit , the @xmath48 site behaves like a two level impurity , and the hamiltonian reads @xmath49 { \rm d } x+ \int_{0}^\infty i\left[\psi_{2l}^\dagger\partial_x\psi_{2l}- \psi_{2r}^\dagger\partial_x\psi_{2r}\right ] { \rm d } x+ \lambda \left [ \left(\psi^\dagger_{1}(0)+\psi^\dagger_2(0 ) \right)d+ { \rm h.c . }
\right],\label{rlmferm}\end{aligned}\ ] ] with @xmath50 , same for the second species , @xmath51 , corresponds here to @xmath52 in eq ( [ eqrlmlattice ] ) . ] .
in contrast with the case of the xx chain with a single defect , this non - interacting problem _ not scale invariant_. the coupling @xmath53 flows , and the system again exhibits _ healing _ : at low energy , the impurity level is completely hybridized with the two half chains . let us go back to the general case @xmath54 . proceeding like before , we can write the limiting behaviors of the entanglement entropy . at low energy ,
the impurity is hybridized , the system behaves just as _ one _ non chiral wire and a hybridized impurity , and the entropy obeys the general form for a region of length @xmath0 in the bulk of a conformal invariant system decoupled from the two baths , so @xmath55 where once again , we included a term @xmath56 that accounts for the remaining boundary condition at @xmath57 of the hybridized impurity .
meanwhile , at high energy , the impurity is completely decoupled from the wires , and one gets @xmath10={1\over 3}\ln { 2l\over \epsilon}+s_1 + \ln g_{\rm uv}.\ ] ] using the folded ( boundary ) version of the system ( [ bdrk ] ) , one can easily argue that @xmath58 , as a decoupled impurity has two degrees of freedom .
one thus expects a behavior entirely similar to ( [ scaling ] ) , where the crossover scale , @xmath21 , is expected to be proportional to a power of the coupling square , @xmath59 , and @xmath23 should be a monotonic function extrapolating between @xmath60 at small values of the argument and @xmath25 at large values .
finally , we note that in the boundary versions ( [ hamilbdr],[bdrk ] ) , the entanglement impurity we have discussed is now the entanglement of a region of length @xmath61 on the edge of the system with the rest .
if one were to start from an ( anisotropic ) kondo version , this would be the most natural point of view @xcite .
there are of course other variants of the problem , for instance involving a slightly modified link in the antiferromagnetic xxz chain with @xmath62 , interactions in the rlm model , etc . in all these cases
, we should stress that the geometry we are considering is probably not the most interesting : considering the entanglement of the two halves connected by a weak link or a quantum dot is probably more physical .
this latter problem is however significantly more difficult technically
. we will discuss it in our next paper , relying on the present work as a stepping stone .
the most natural to explore the behavior of @xmath63 between the fixed points is to use perturbation theory .
the required calculation is a modification of the one proposed in @xcite . using the well - known replica trick
, one first observes that the entanglement entropy @xmath64 can be obtained from the renyi entropies @xmath65 by considering @xmath66.the renyi entropies in turn can be obtained as @xmath67 , where @xmath68 is the partition function on a @xmath69-sheeted riemann surface @xmath70 with the sheets joined at a cut corresponding to the segment of length @xmath0 .
the difference between the problem at hand and the conformal case is that now there is a perturbation inserted at the origin in the hamiltonian formulation , which corresponds to the insertion of a perturbation along an imaginary time line for each of the @xmath69 sheets .
the modified partition functions @xmath68 can in principle be expanded in powers of the coupling constant , and perturbative corrections to the renyi entropies and the entanglement entropies finally obtained . to see what happens in more detail ,
we consider first hamiltonian ( [ hamil ] ) .
we start with @xmath71 sheets , and write formally the renyi entropy as a functional integral for a pair of chiral bosons as @xmath72[{\cal d}\phi_2 ] \exp\left\{-a[\phi_1]-a[\phi_2]- \lambda\int_{-\infty}^\infty\left[v_1(x=0,y)+v_2(x=0,y)\right]{\rm d}y\right\}\over \left(\int [ { \cal d}\phi]\exp\left\{-a[\phi]-\lambda\int_{-\infty}^\infty v(x=0,y){\rm d}y\right\}\right)^2}\label{def},\end{aligned}\ ] ] where @xmath73 is the perturbation , and @xmath74 = \int { \rm d}^2 x \mathcal{l } [ \phi]$ ] is the free action .
note we are working in the chiral version , but have suppressed the ` r ' label in the fields for simplicity of notation .
finally , the label twist means the functional integral is evaluated with conditions around the cut @xmath75 if @xmath33 , the ratio ( [ def ] ) is nothing but the correlation function of an ( order two ) twist operator corresponding to ( [ eqtwist ] ) , which we will write then @xcite @xmath76 where @xmath77 means that the correlator is to be evaluated in the plane @xmath78 ( worldsheet ) with the lagrangian @xmath79 + { \cal l } [ \phi_2]$ ] .
we also recall that in general , the scaling dimension of the twist operator @xmath80 reads @xmath81 .
we now consider the perturbation expansion in powers of @xmath53 .
for the denominator , we have immediately @xmath82\ ] ] where the factor @xmath83 in the integral comes from the @xmath83 s in the cosines , and we recall that @xmath84 is the conformal weight @xmath85 of the perturbation . for the numerator ,
things are a little more complicated since we have two types of fields on the plane , with @xmath86 and @xmath87 contractions , in the presence of the glueing conditions along the cut .
to proceed , we uniformize .
we start with the complex coordinates @xmath88 , and introduce @xmath89 where @xmath90 and @xmath91 are the complex coordinates of the cut s extremities .
this maps the whole 2-sheeted riemann surface @xmath92 to the @xmath93-complex plane @xmath94 .
we then write ( [ def ] ) as @xmath95 where the spatial integrals in the numerator are now over @xmath92 ( worldsheet ) , and we have a unique boson @xmath96 instead of @xmath97 and @xmath98 . here
the integrals in the numerator correspond to insertions along _ two _ lines , corresponding to the two copies of the theory , so overall there are four possible terms ( contractions ) .
the perturbation @xmath99 is a primary operator , so we can calculate the correlations on @xmath92 by using the conformal mapping ( [ eqconfmap ] ) .
we have @xmath100 here , @xmath101 in ( [ firstorderuv ] ) we have to integrate @xmath102 both over the imaginary axis @xmath103 , but also over the second sheet , which is obtained by sending @xmath104 and same for @xmath105 .
this means we end up with two integrals where @xmath102 are on the same sheet , and two where they are on different sheets . replacing everything by the particular choice of coordinates , and expanding the denominator in ( [ firstorderuv ] ) we get @xmath106 + \dots,\ ] ] where the factor @xmath107 comes
since there are two sheets , and insertions can be on same or different sheets , so @xmath108^{2\mu}\over ( y^2+l^2)^{\mu/2}((y')^2+l^2)^{\mu/2}}- \frac{1}{|y - y'|^{2\mu}},\ ] ] while @xmath109 will be the same expression with a minus in the numerator s bracket , and no subtraction ( the two point function of the fields in different copies in the denominator of course vanish identically ) .
it is convenient to introduce new variables via @xmath110 , so the integral becomes @xmath111 the second integral reads similarly @xmath112 both integrals are uv convergent for a relevant perturbation @xmath113 .
they are however both ir divergent ( here the ir region being @xmath114 ) .
this means that , although formally the perturbation at small coupling looks like it should be an expansion in powers of @xmath115 , this might actually not be the case and , as we shall see later , is not .
in fact , we will see that the entanglement is simply non perturbative in @xmath116 , and can not be obtained via this perturbation theory .
this result could appear as a surprise .
on the one hand , the entanglement is a @xmath117 quantity , and such quantities are often plagued by ir divergences .
on the other hand , we are looking for an @xmath61 dependent quantity , and it would be natural to expect that @xmath61 would act as an effective ir cutoff , rendering the perturbation expansion finite .
this is however definitely not what happens .
the situation is reminiscent of similar divergences encountered in the kondo screening cloud problem @xcite .
we stress finally that the argument applies almost without modification to the hamiltonian ( [ chiralk ] ) .
all that changes is that the perturbation is of the form @xmath118 instead of @xmath119 , so exponentials of opposite signs have to alternate in the imaginary time insertions , modifying some of the numerical coefficients , but not the integrals or their divergences .
to gain a better understanding of the situation , it is useful to start by discussing another observable than the entanglement entropy for ( [ hamil ] ) .
we turn briefly to the boundary formulation ( [ hamilbdr ] ) , and consider the one point function @xmath120 which appears , for instance , in the determination of friedel oscillations for impurities in luttinger liquids .
simple scaling arguments suggest the general form @xmath121 where @xmath122 is a universal function obeying @xmath123 , so the field sees dirichlet boundary conditions , and the bulk normalization has been chosen appropriately . determining the function @xmath122 is also a difficult problem .
the most natural is once again to attempt perturbation theory in @xmath53 .
this would share many features of the calculation of one and two point functions in the bulk sine - gordon theory @xcite .
there , it is well known that ( in fact , the result essentially goes back to coleman ) , provided @xmath124 ( that is , the perturbing operator is not irrelevant ) , there are _ no uv divergences _ in the calculation .
all divergences coming from bringing together two insertions of the perturbing term @xmath125 are exactly canceled by similar divergences coming from the expansion of the denominator ( the partition function and associated bubble diagrams ) . in general , the divergences are indeed controlled by the operator product expansion ( ope ) , @xmath126 with all fields at @xmath57 , @xmath127 the coordinate along the boundary .
the leading order comes from the contribution of the identity operator and leads to a disconnected piece subtracted off by a similar term in the denominator .
the @xmath128 stand for higher orders , or lower orders that vanish after integration .
the overall singularity at order @xmath129 thus behaves as @xmath130 , it comes with dimension @xmath131 and for @xmath132 the integrals are uv finite . other singularities ( when several points are brought together at once etc @xcite ) behave similarly .
however , there will always appear ir divergences at a certain order , depending on the exact value of the conformal weight @xmath133 .
of course , we expect in the end the scaling form to hold , and thus to depend only on @xmath134 , @xmath135 .
what will happen in general is that the divergences in the perturbative expansion have to be resummed before the proper scaling form can be obtained .
the latter , in general , will thus behave non perturbatively in the coupling @xmath53 .
this is nicely illustrated in the case @xmath136 , where the the exact form of the one point function is known , thanks to a mapping to the boundary ising model ( see below ) , together with a very clever argument by chatterjee and zamolodchikov @xcite . one finds @xcite @xmath137 where @xmath138 is the degenerate hypergeometric function .
the asymptotics follows from @xmath139 , where @xmath140 is the usual modified bessel function , so that we find @xmath141 we thus see that this function exhibits a non perturbative dependence at small coupling @xmath53 .
the non analyticity in @xmath53 arises from the ir divergence of the first perturbative integral .
there is a general way to understand the non analyticity of course .
whenever a bulk operator ( of conformal weights @xmath142 ) is sent to the boundary where it becomes a boundary field of weight @xmath143 , one has @xmath144 in our case , the cosine of the bulk field simply goes over to the cosine of the boundary field .
we have thus @xmath145 and @xmath146 , while @xmath147=l^{-h_b}\propto t_b^{h_b}\propto\lambda^{h_b/1-\mu}$ ] .
we thus expect that @xmath148 .
the dependence of the one point function of the boundary field on @xmath53 is non analytic in @xmath53 , and non perturbative - of course , because again of ir divergences .
this problem is the cousin of a similar problem in bulk massive theories , and has been studied in @xcite .
we deduce from this that , to leading order , @xmath149 more generally , we can write @xmath150 so @xmath151 with @xmath152 @xmath153 going back to the case of friedel oscillations , we have therefore @xmath154 . this leading dependence in @xmath53
replaces the expected perturbative one , which would be linear in @xmath53 .
the foregoing argument applies in the generic case .
whenever there are `` resonances '' and the parameter @xmath84 takes special rational values @xmath155 , extra logarithmic terms appear in the one point functions of the operators right on the boundary , which translates in logarithms in the one point functions of operators at @xmath156 as well .
this is the case precisely when @xmath136 .
it is important to stress also that the ir divergences naturally disappear at finite temperature , @xmath157 providing a natural cutoff .
once again this is illustrated in the @xmath136 case , where one finds @xcite , for friedel oscillations at finite temperature @xmath158 here @xmath122 is the usual hypergeometric function , @xmath159 is a function whose existence and value were determined in @xcite .
the right hand side admits a perturbative expansion in powers of @xmath53 , whose leading term , at fixed @xmath160 , goes as @xmath161 when @xmath162 .
the coefficient of @xmath53 thus diverges in the zero temperature limit , in agreement with the fact that the true expansion is then in @xmath163 .
the general ir behavior can easily be investigated .
one finds that at order @xmath164 , there is no ir divergence provided @xmath165 . only when @xmath44 that is , the boundary perturbation is exactly marginal , and the bulk is a fermi liquid are all orders finite . in this case , the friedel oscillations admit a perturbative expansion in powers of @xmath53 @xcite .
while the nature of the divergences is quite generic , the quantities for which they occur depend on the problem at hand .
for instance , for the screening cloud in the ( anisotropic ) kondo model , divergences occur even when the boundary perturbation has dimension one - in that case , it is marginally relevant @xcite .
we now go back to the calculation of the entanglement entropy for hamiltonian ( [ hamil ] ) .
we see that , to obtain the non perturbative uv behavior , we must discuss twist fields and their opes .
we follow the paper @xcite but focus more directly on the question at hand .
imagine we have a single interval for which we want to calculate the entanglement with the rest of the system , and introduce accordingly the @xmath69-sheeted riemann surface ( @xmath69 replicas ) @xmath70 . in the limit
where the interval of length @xmath61 shrinks , we expect the presence of the two sewing points to decompose like an operator product expansion of the form @xmath166 where we allowed for fields inserted at points @xmath167 , the point @xmath93 on the @xmath168 sheet , and the set @xmath169 denotes a complete set of local fields for one copy of the cft .
recall that the cut in the riemann surface @xmath70 corresponds to the insertion of twist fields in the complex plane , so that @xmath170 and ( [ opetwist ] ) should be considered as the ope of these twist fields .
what ( [ opetwist ] ) means more precisely is that , if we have other operators inserted elsewhere , we can expect to have @xmath171 where @xmath172 designates operators inserted on the @xmath168 sheet , and @xmath173 is the @xmath168 copy of the complex plane .
note indeed that the expectation on the right is taken in a fully factorized theory .
restricting now to the @xmath174 that make an orthonormal basis ( so in particular they are all quasiprimary ) , and choosing @xmath175 shows that the structure constant @xmath176 will not vanish only if the average of @xmath177 on the riemann surface @xmath70 does not vanish .
it is useful to make things concrete now , so for instance we see that there is no term with a single primary operator on the right hand side of ( [ opetwist ] ) since the corresponding one point function on @xmath70 vanishes .
there is , however , at least one term with a single operator , the stress energy tensor , since we know that @xmath178 .
apart from this , the most important terms will be those involving the same primary operator on two different sheets @xmath179 , whose average on @xmath70 will be non zero in general . if the field @xmath138 has conformal weights @xmath180 , we will thus have that @xmath181 where @xmath182 is the conformal weight of the twist field .
the crucial point is that @xmath183 involves _
twice _ the scaling dimension of primary fields , in contrast with ordinary opes where only the scaling dimension would appear .
the discussion carries over to the boundary case .
one can , for instance , think of it after unfolding the system so as to keep only chiral fields as in ( [ hamil ] ) .
everything then formally goes through after setting @xmath184 .
the question is then , what kind of fields @xmath185 ( the chiral part of @xmath138 ) can appear in the ope of two twist fields .
the one copy bulk theory is a compact boson which allows for the fields @xmath186 on the boundary .
this means that the radius is @xmath187 , and thus the bulk conformal weights are given by @xmath188 restricting to scalar operators we get @xmath189 or @xmath190 .
for instance , the first values of @xmath191 correspond to fields @xmath192 , or , for the chiral part , @xmath193 .
we now go back to the entropy calculation in the folded , non - chiral theory ( [ hamilbdr ] ) . upon folding ,
the chiral vertex operators @xmath194 become @xmath195 , as @xmath196 .
recall also that the non - chiral twist field in the folded version can be thought of as the chiral part of @xmath197 in the unfolded theory .
hence , going through the discussion of short distance expansions we find , for the non chiral twist field @xmath198 where we used the fact @xcite that the two fields @xmath185 in the twist opes must belong to different copies .
we are only interested in terms whose one point function acquires a non zero value in the presence of the perturbation .
this means the first term with @xmath199 can not contribute , and thus we need @xmath200 , @xmath201 . taking derivative with respect to @xmath69 gives then the leading term for the entanglement correction , which should go as @xmath202 for @xmath136 in particular , this can be corrected by a resonance , and it is tempting to speculate then that one has @xmath203\label{slog}\ ] ] finally , we note once again that the rlm or the various ( anisotropic ) kondo versions will behave identically , the presence of the operators @xmath204 not modifying in any essential way the ope argument but one will have to be careful with the dimensions of the operators involved , and their relationship with @xmath84 . in the end , we find that for the rlm ( [ slog ] ) is expected to hold as well .
while the small coupling expansion is plagued with ir divergences , a large coupling expansion is possible .
it is now finite in the ir , and exhibits uv divergences which are easily taken care of using integrability and analyticity .
let us recall how the calculation goes at leading order in the anisotropic kondo case @xcite ( see also _
the leading ir perturbation is nothing but the stress energy tensor @xmath205 the correction to the renyi entropy can therefore be expressed as @xmath206 the first correction to the entanglement entropy thus reads @xmath207 it is quite remarkable that this result does not depend on the anisotropy parameter @xmath208 ( recall that @xmath209 in the xxz language ) .
it turns out that this ir expansion can be generalized to higher orders @xcite .
the results for the kondo case are as follows @xmath210 where the coefficient @xmath211 has the following dependence on the dimension @xmath212 of the tunneling operator @xmath213 note that in ( [ result : infinite : imp ] ) , the first term in the right hand side has to be truncated at order 6 .
while in principle higher orders in the ir expansion could be determined , the complexity of the calculations increases considerably .
moreover , the convergence properties of this expansion are not clear .
finally , we observe that , in this point of view , the pure bsg case turns out to be quite different , because different operators appear in the ir effective description .
the corresponding result has not even been worked out yet .
making analytical progress therefore requires developing non perturbative approaches .
the problems we are interested in are indeed integrable , at least in their boundary versions .
while it is natural to expect that this can be used in some way , integrability has been mostly used to calculate local properties such as magnetization , energy or impurity entropies .
von neumann entanglement is non local , and therefore much harder to obtain in general .
we will in what follows restrict to the case where the dimension of the perturbation is @xmath1 : this corresponds to @xmath214 ( @xmath215 ) for the problem of tunneling between xxz chains , and to @xmath7 the rlm ( @xmath44 ) for the tunneling through an impurity .
these cases are closely related to the boundary ising model with a boundary magnetic field ( see appendix [ sec : relsising ] ) .
while the problem of calculating the entanglement non perturbatively remains extremely difficult entanglement still involving non local observables in the fermionic language it can be tackled using the idea of form - factors .
it has been known for many years that correlation functions of local observables in massive integrable theories can be calculated using the form - factors approach , where the integrable quasiparticles provide a basis of the hilbert space , and the form - factors ( ff ) that is , the matrix elements of the operators in that basis can be obtained using an axiomatic approach based on the knowledge of the s matrix and the bootstrap . it is a natural idea to extend this approach to the case of entanglement entropy .
indeed , the von neumann entanglement is obtained form the renyi entropy by taking an @xmath69 derivative at @xmath216 , and the renyi entropies can be considered formally as correlation functions of twist operators that live in @xmath69 copies of the theory of interest .
the integrability of a single theory carries over to integrability of the @xmath69 copies , and a calculation similar to the one of ordinary correlators can be set up , after some additional work to determine the form factors of the twist operators @xmath217 @xcite .
we are interested here in a variant where the bulk is massless .
the form - factors technique in this case is more delicate to use , since particles can have arbitrarily low energies , and the convergence of the approach is not guaranteed .
various regularization tricks have to be used in the calculation of local quantities ( eg the charge density for friedel oscillations ) @xcite , and we will see below that the situation for the entanglement is not better . nevertheless , @xmath63 can be calculated for @xmath1 , by using the ising model formulation , and relying heavily on the work @xcite . to fix ideas , and explore the feasibility of form - factors calculations in our problem , we first discuss briefly the bulk case and the massless limit .
one can find in @xcite the first order contribution to the two point function of the bulk ising model twist field in the bulk @xmath218 where @xmath69 is the number of copies , @xmath219 is the rapidity of the @xmath220 particle with energy @xmath221 and momentum @xmath222 , and @xmath223 is the two - particle form factor of the twist field @xmath224 @xmath225 in this last expression , we have used the notation @xmath226 for the usual faddeev - zamolodchikov creation operators ( here , the fermions ) living in the @xmath227 copy . since the theory is integrable , the form factors @xmath223 can be computed exactly and are conveniently expressed using the function @xmath228 which vanishes when @xmath216 .
the other form factors @xmath223 can then be obtained from @xmath229 by shifting appropriately @xmath230 by a factor of @xmath231 .
going to variables @xmath232 one can perform one integration , and be left with @xmath233 where @xmath234 doyon _ et al .
_ then argue the crucial result that @xmath235 taking the derivative of the two point twist correlation function meanwhile should give , at short distances , the entanglement entropy of the cft . since ( [ firstorder ] ) is only a first order approximation where contributions with a larger number of particles have not been included , we get an approximation to the entanglement entropy of a segment of length @xmath236 in the bulk with the rest of the system @xcite @xmath237 and thus the expected factor @xmath238 is approximated by @xmath239 at this order . since in this paper we are interested in bulk cfts
, we need to take an @xmath240 limit .
this corresponds formally to describing the cft using massless particles and massless scattering .
we thus set @xmath241 and send @xmath242 .
only two types of excitations remain at finite energies : those for which @xmath243 with @xmath244 finite . in the first case ,
one obtains right moving particles with @xmath245 and in the second case left moving particles with @xmath246 .
conformal fields factorizing into left and right components are not expected to mix the l and r sectors .
indeed , @xmath247 so only the ll and rr sectors will contribute in the massless limit of ( [ firstorder ] ) . therefore , setting ( say for the r sector ) @xmath248 and introducing @xmath249 we obtain @xmath250 where the @xmath83 coming from the jacobian was canceled by the fact that there are two integrals , the l and the r one . using ( [ deriv ] ) we get the correction to the entanglement entropy as @xmath251 this integral is divergent at small energy , a feature which is quite general in the use of massless form - factors .
we regularize by considering the integral @xmath252 so the finite part of the integral is @xmath253 and thus we recover @xmath254 let us now consider the ising model with a boundary magnetic field as in ( [ ising ] ) , and to start assume that the bulk is massive .
the form factors approach can be applied to this case as well .
the first non trivial contribution reads then @xcite @xmath255 coming from the boundary state @xmath256 { \left|0 \right>},\ ] ] where we recall that @xmath226 are the usual faddeev - zamolodchikov creation operators living in the @xmath227 copy , and @xmath257 is the reflection matrix @xcite of the ising field theory with a boundary magnetic field @xmath258 ( proportional to @xmath53 in [ ising ] ) .
ultimately , we want once again to take the massless limit @xmath259 .
notice that ( [ firstord ] ) involves @xmath260 instead of @xmath261 .
we write @xmath262 and observe that the analytical continuation in @xmath69 is trivial because the particle and its reflection must belong to the same copy . to every order , contributions are linear in @xmath122 .
but there is a lot of similarity eg between the term with four particles here , and the term with two particles in the bulk entropy . in the massless limit case , since the boundary produces as many l as r particles , and since we need both these numbers to be even , only the terms with @xmath263 r movers and @xmath263 l movers contribute .
indeed , the first correction to the entanglement , after taking derivative with respect to @xmath69 at @xmath216 reads explicitly @xmath264 with @xmath265 . to obtain a scaling expression in the massless limit ,
we boost rapidities like in the bulk case , and we obtain @xmath266 a vanishing result - natural , since in this limit , the two - particle form factor factorizes onto one particle form factors ( one for the left , one for the right ) , which both vanish .
we thus need to go to the next order ( corresponding to 4 particles ) , where we use formula eq .
( 3.25 ) in @xcite : @xmath267 we obtain then @xmath268 where we have set @xmath269 and we have shifted the @xmath244 integral .
now the expression ( [ lowestorder ] ) is divergent at low energies , just like ( [ divint ] ) . to regularize it , we consider the difference : @xmath270 the uv value is @xmath271 , to be compared with the exact value @xmath272 .
this indicates we are on the right track . to proceed , we now take equation ( 3.54 ) in @xcite , perform the appropriate limits and rescalings to get @xmath273 \end{aligned}\ ] ] where products and sums run over @xmath274 and
we have set @xmath275 .
the second term is obtained by shifting the contours of integration in the imaginary direction as indicated .
we observe the same divergence at low energy , and the same regularization ( subtracting the formal expression for @xmath276 ) also works like for @xmath277 .
we find @xmath278,\ ] ] where the dots correspond to the two other terms obtained by shifting the contours of integration as in ( [ eqffborder2 ] ) .
the uv value is @xmath279 , so at second order , we have the uv value @xmath280 , to be compared once again with the value @xmath272 we find that higher orders can be dealt with in the same way , and that the uv values can be resummed exactly to yield the exact result @xmath281 as expected .
we now return to the rlm , whose results are obtained simply by multiplying those for ising by a factor of two . in the following , we will allow for an extra multiplicative renormalization to obtain the uv result exactly , that is consider , at lowest order , the ratio @xmath282 the next order approximation can be handled similarly , and we will simply provide the corresponding results on the curves below . ]
we plot our results for the ff approach on figure [ ffir ] where the dashed line is the ir expansion ( see above and @xcite ) , and the full colored lines are form factors approximations .
clearly , on this scale , the ( renormalized ) ff expansion has converged very quickly .
we shall soon see how close it is to the real data from numerical simulations on the xx chain .
we now discuss briefly the uv behavior .
for @xmath277 , standard tables give @xmath288+\ldots\ ] ] where @xmath289 is the euler constant .
this provides a leading order correction in the uv which reads @xmath290 .
note that this is compatible with what was expected from the general discussion about the non perturbative behavior in the uv : we get a term linear in @xmath291 , decorated by logarithmic corrections .
the next order is more difficult to handle analytically , but very accurate numerics shows that it does behave similarly , only leading to a correction of the slope which goes from @xmath292 to @xmath293 .
we plot on figure [ peter3 ] the leading correction divided by @xmath294 as a function of @xmath295 , together with numerical data that shall be discussed in the next section .
we see that the ff expansion converges very well except in the uv , where it seems to still converge , but more slowly , giving us only a rough approximation of the exact leading singularity .
finally , the results for the entanglement for the problem of weakly coupled xxz chains at @xmath214 would be identical but for an overall normalization by a factor @xmath83 .
we now turn to a numerical determination of the entanglement entropy in the rlm , going back to the formulation ( [ eqrlmlattice ] ) where we will now also have to be careful with the overall finite size of the system .
we write @xmath296
so the left and right leads have
@xmath297 sites , and the impurity sits at site 0 .
we can now switch to a representation of symmetric and antisymmetric combination of the lead sites , @xmath298 . since the antisymmetric combination decouples from the rest , its contribution to the impurity entanglement will drop out .
it is therefore sufficient to study a system of @xmath299 sites @xmath300 here a single resonant level couples to a single chain of @xmath301 sites . in order
to compare with field theory we use exactly half filled system to exploit the linear regime of the cosine band . in return
we have to use an even number of @xmath302 sites . the scale of the resonant level with a coupling of @xmath303 is @xmath304 with @xmath305 as we have chosen the normalization @xmath306 . following the recipe of @xcite
we now calculate the reduced single particle matrix @xmath307 for the last @xmath308 sites , where the first site ( labeled @xmath25 ) corresponds to the impurity . in order to obtain the bulk result
we determine the reduced density matrix @xmath309 for the first @xmath61 sites of the chain .
the reason for taking the bulk result from the opposite end of the chain is that we can not just study a chain of @xmath297 sites , as we would then have a degenerate ground state as @xmath297 is odd .
the diagonalization is performed within double precision , while the trace for the entropy is performed using quadruple ( 128 bit ) precision .
the entanglement entropy corresponding to the single particle reduced density matrix is now given by @xmath310 which finally leads to @xmath311 in figure [ peter1 ] we plot the numerical results together with the first three orders of the ir expansion and the first order of the ff expansion .
we would like to remark that on the lattice we get small @xmath312 oscillations on top of the continuum result . ) . ]
figure [ peter2 ] is a similar plot emphasizing the ir behavior . finally , in figure [ peter3 ] as commented already we focus on the singularity in the uv , comparing slopes obtained from the ff expansion .
our numerics on system sizes @xmath313 is consistent with a singularity @xmath314 with a slope @xmath315 . by applying damped boundary conditions ( dbc )
@xcite one can access very small energy scales on system sizes which are accessible by numerics . by looking at systems of @xmath316 sites where we scale down the bulk hopping elements by a factor of @xmath317 on each bond from site 2000 to 3000 and using a bulk hopping element of @xmath318 on the last 1000 sites we find an indication that the singularity is even slightly stronger . while we can exclude an @xmath319 behaviour , we can not rule out the possibility of an @xmath320 contribution .
note that the dbc change the form of the density of states at the fermi surface , for details see @xcite .
it is therefore possible that this additional increase is due to this modification of the level spacing at the fermi surface . due to the slow increase of the logarithm such a clarification is asking for multi precision arithmetic .
this study shows that the entanglement entropy of quantum impurities involved in an rg flow is a quantity which is difficult to access .
it is non perturbative in the uv , and the ir perturbation , while well defined , does not capture the crossover regime very well .
the form - factors approach , on the other hand , is remarkably successful .
it is , however , difficult to develop except in the simple case of the ising model , and more work will have to be done in that direction .
nevertheless , we believe that the essential features of @xmath23 are under control , although it would be useful to check the uv singularities for other values of the coupling ( anisotropy ) . in conclusion , we emphasize that the geometry considered in this paper where the interval for the entropy is centered on the impurity is probably not the most natural physically . to characterize the kondo physics , one would rather be interested in the entanglement of the two wires tunneling through an impurity .
this could be characterized physically by the entropy of an interval with the impurity at its boundary , or for example , by the negativity of two intervals in the different wires ( see _ e.g. _
@xcite for examples related to the kondo problem ) .
this situation is unfortunately much more complicated technically , mostly because the folding procedures described in this paper no longer apply .
however , we still expect the conclusions of this paper to hold in that case as well , namely , we expect the entanglement entropy ( or other entanglement estimators ) to depend non - pertubatively on the coupling to the impurity when this is weak .
we believe that improper regularizations of the ir divergences encountered in perturbation theory led to some confusion in the literature @xcite .
we will report on this together with a correct calculation in a subsequent paper . *
acknowledgments * : we thank ian affleck , benjamin doyon , john cardy and pasquale calabrese for discussions .
hs was supported by the french agence nationale pour la recherche ( anr projet 2010 blanc simi 4 : dime ) and the us department of energy ( grant number de - fg03 - 01er45908 ) .
computations were performed on the compute cluster of the yig 8 - 18 group of peter orth at the karlsruhe institute of technology .
let us begin by discussing the relationship between the rlm and the ising model with a boundary magnetic field .
this can be seen in various ways .
start from the hamiltonian ( [ rlmferm ] ) , unfold the wires to get only right movers , and form the combinations @xmath321 the fermion @xmath322 decouples from the impurity entirely , and we will mostly discard it from now on .
the remaining dynamics is then encoded in the hamiltonian @xmath323\ ] ] we then refold this hamiltonian to map back to a boundary problem , introducing a @xmath324 component : @xmath325{\rm d}x+\lambda\sqrt{2}\left[\psi^\dag(0)d+\hbox{{\rm h.c.}}\right]\ ] ] where @xmath326 .
the next and almost final step is to go to a majorana version of this problem .
we decompose the fermions into real and imaginary parts as @xmath327 where @xmath328 are real and obey @xmath329 etc .
we set similarly @xmath330 with @xmath331 .
the problem then decouples into _ two independent _ majorana problems @xmath332 , with @xmath333{\rm d}x+ \frac{i}{\sqrt{2 } } \lambda \xi(0 ) b , \nonumber\\ h_2=-{i\over 2}\int_{-\infty}^0\left [ \eta_r\partial_x\eta_r-\eta_l\partial_x\eta_l\right]{\rm d}x- \frac{i}{\sqrt{2 } } \lambda\eta(0)a , \label{ising}\end{aligned}\ ] ] and @xmath334 , same for @xmath335 .
the problems correspond of course to two ising models with a boundary magnetic field proportional to @xmath336 ( up to normalizations ) , the boundary spin operator being @xmath337 .
note that this result is compatible with boundary entropy counting .
the flow from uv to ir in the original problem leads to @xmath338 since a dot with two states is screened . in each of the ising models
meanwhile , we have a flow from free to fixed , with @xmath339 , so the product of the two ratios one for each ising copy
is @xmath107 indeed .
turning now to entanglement entropy , we see that the rlm entanglement for a region of size @xmath0 centered around the impurity is exactly twice the entanglement for a region of length @xmath61 on the edge of the system in the boundary ising model .
this has been studied numerically _
e.g. _ in @xcite , and we will present in section [ sec : ff ] an analytical calculation of this quantity .
the boundary sine - gordon problem at @xmath136 is also well known to be equivalent to two boundary ising models @xcite , where this time only one of these models experiences a non zero boundary magnetic field .
hence , we shall also be able to obtain the entanglement entropy for ( [ firsth ] ) for this value of @xmath84 , that is @xmath214 .
r. egger and h. grabert , _
friedel oscillations for interacting fermions in one dimension _ , phys .
lett . * 75 * , 3505 ( 1995 ) , _ friedel oscillations in luttinger liquids _ , quantum transport in semiconductor submicron structures nato asi series * 326 * , 133158 ( 1996 ) . | we study the entanglement entropy of a region of length @xmath0 with the remainder of an infinite one dimensional gapless quantum system in the case where the region is centered on a quantum impurity .
the coupling to this impurity is not scale invariant , and the physics involves a crossover between weak and strong coupling regimes .
while the impurity contribution to the entanglement has been computed numerically in the past , little is known analytically about it , since in particular the methods of conformal invariance can not be applied because of the presence of a crossover length .
we show in this paper that the small coupling expansion of the entanglement entropy in this problem is quite generally plagued by strong infrared divergences , implying a non - perturbative dependence on the coupling .
the large coupling expansion turns out to be better behaved , thanks to powerful results from the boundary cft formulation and , in some cases , the underlying integrability of the problem .
however , it is clear that this expansion does not capture well the crossover physics . in the integrable case which includes problems such as an xxz chain with a modified link , the interacting resonant level model or the anisotropic kondo model a non perturbative approach is in principle possible using form - factors .
we adapt in this paper the ideas of @xcite to the gapless case and show that , in the rather simple case of the resonant level model , and after some additional renormalizations , the form factors approach yields remarkably accurate results for the entanglement all the way from short to large distances .
this is confirmed by detailed comparison with numerical simulations .
both our form factor and numerical results are compatible with a non - perturbative form at short distance . |
seyfert galaxies are generally classified into two groups , based on their spectra and emission line widths in the optical .
seyfert 1 galaxies are characterized by broad permitted lines with typical fwhm velocities @xmath51000 km s@xmath4 , and narrower forbidden lines with fwhm @xmath6500 km s@xmath4 , with a non - stellar underlying continua .
the spectra of seyfert 2 galaxies show only the narrow emission lines , and their non - stellar contribution is normally detected only in polarized light .
the generally accepted unification scheme states that the two classes of galaxies are not disparate , but that seyfert 2 galaxies are seyfert 1s with nuclei that are obscured by a dusty molecular torus .
@xcite discovered broad balmer and fe@xmath7ii lines in the seyfert 2 galaxy ngc 1068 . from polarization arguments they concluded that the lines must be emanating from regions synonymous with the broad - line region ( blr ) of seyfert 1s .
subsequent studies reinforced the notion that seyfert 2 blr are obscured from the line of sight but can be observed via scattered radiation @xcite .
therefore it is now generally accepted that the seyfert designation 1 or 2 is contingent upon the orientation angle as perceived by the observer . + a more dynamical analysis is needed to determine if these two types of active galactic nuclei ( agn ) are indeed similar and powered by the same processes .
kinematic studies of the narrow line region ( nlr ) can provide insights into the intricate dynamical forces occurring within its vicinity . with the stis spectrograph aboard _ hst _
, such studies can be easily accomplished efficiently .
high spatial and spectral resolution data in particular can provide us with detailed diagnostics of the nlr and its vicinity .
we have therefore undertaken an investigation of the kinematics across the nlr of ngc 4151 , one of the nearest seyfert galaxies .
+ this type of observation and kinematic modelling was previously done on ngc 1068 , a bright seyfert 2 galaxy @xcite .
the data were taken with the low resolution ( @xmath8 = 1000 ) g430l grating of stis , and the radial velocity measurements were made using [ o @xmath0iii ] @xmath9 5007 emission along two slit positions [email protected] wide .
they found that a simple biconical outflow model closely matched the observed radial velocities such that emission - line knots accelerate out from the nucleus to @xmath2 1000 km s@xmath4 then decelerate back to the systemic velocity .
the same model was applied to ngc 4151 ( * ? ? ?
* hereafter paper 1 ) , based on stis long - slit low resolution and slitless medium resolution spectra ( hutchings et al .
1999 ; kaiser et al .
2000 ; nelson et al .
2000 ) . in paper 1
they pointed out that the observed velocity field can be well matched by tilting the model bicone by about 40@xmath1 out of the plane of the sky , as opposed to only 5@xmath1 for ngc 1068 .
+ in this paper , we present the same modeling procedures as in the two previous studies , but at a spectral resolving power of 9000 , and five parallel slit positions covering the nlr of ngc 4151 .
this provides the most detailed map of the kinematics in a seyfert 1 nlr , similar to that obtained for the seyfert 2 galaxy ngc 1068 @xcite .
we used a systemic velocity of 997 km s@xmath4 , based on h i observations of the outer regions of the host galaxy @xcite and a hubble constant of 75 km s@xmath4 mpc@xmath4 , which gives us a distance of 13.3mpc to ngc 4151 , with [email protected] corresponding to a projected linear scale of 6.4pc on the sky .
the stis observations were taken on 2000 july 02 over five orbits ( proposal i d 8473 , pi : j. hutchings ) with the _
hst_/stis ccd through a slit of , and dispersed with the medium - resolution g430 m grating , whose spectral resolving power is @xmath29000 .
the stis image scale is [email protected]/pix and the spatial resolution is [email protected]/pix in the cross - dispersion direction .
the five orbits yielded spectra for five parallel slit positions , each at a position angle of 57.8@xmath1 .
there were three exposures per slit with a total of 40 minutes of integration for slit 1 and 48 minutes for the remaining slits .
figure [ intensity ] shows the slit placements and orientation superimposed on a wfpc2 [ o @xmath0iii ] emission - line image of ngc 4151 showing the inner 4 .
slit 1 was centered on the optical continuum nucleus and the remaining slits were placed at [email protected] and [email protected] on either side , parallel to slit 1 .
the image was taken with the [ o @xmath0iii ] filter , which has a central wavelength of 4961 , and a bandpass of 48205100 , which includes the h@xmath10 line and the [ o @xmath0iii ] @xmath11 4959 , 5007 redshifted emission lines .
most of the emission comes from within 2 of the bright central knot , although [ o @xmath0iii ] emission can be traced as far as 6 on either side .
+ figure [ image - oiii - slit1 ] shows a fully reduced stis image of the two [ o @xmath0iii]lines as they appeared through slit 1 .
the bright horizontal line through the image is the continuum emission produced by the nucleus .
the top of the image corresponds to the northeast direction and the horizontal scale is in wavelength .
both [ o @xmath0iii ] lines clearly show an overall trend in that the majority of clouds in the upper region are receding or redshifted and those clouds in the bottom part of the image are approaching or blueshifted .
close to the nucleus , within @xmath2 2 , the clouds appear highly dispersed .
+ we used the idl software developed at nasa s goddard space flight center for the stis instrument definition team to reduce the spectra .
cosmic - ray hits were identified and removed from observations by comparing the three images obtained during each orbit . hot or warm pixels were replaced by interpolation in the dispersion direction .
wavelength calibration exposures obtained after each observation were used to correct the wavelength scale for zero - point shift .
the spectra were geometrically rectified and flux - calibrated to produce a constant wavelength along each slit column and fluxes in units of ergs s@xmath4@xmath12@xmath4 per cross dispersion pixel .
+ each spectral image produced one spectrum per cross - dispersion pixel along the slit , and centered around the two bright doppler - shifted [ o @xmath0iii ] emission lines .
we used the brighter of the two emission lines ( @xmath95007 ) and fitted each [ o @xmath0iii ] component with a local continuum and gaussians .
noisy spectra roughly @xmath5 6 from either side of the nucleus were not fit .
close to the nucleus , and on either side , emission knots showed two and sometimes three major kinematic components ( figure [ gaussfits ] ) , and in such cases we fitted each identifiable peak with gaussians .
+ figure [ gaussfits ] depicts a progression of spectra illustrating how the components fluctuate with position .
the graphs represent spectra taken from slit 1 and range in distance from [email protected] to [email protected] in increments of [email protected] , going away from the nucleus in the northeast direction . in ( a ) , there are two distinct kinematic components of emission , which we fit with two gaussians .
going from ( a ) to ( b ) , a third component emerges .
a possible fourth component emerges in ( c ) , but was not measured by our fitting routine .
the second component shrinks in ( c ) , and disappears in ( d ) , while the fourth component is now detected and labeled ` 2 ' .
this new second component grows stronger while the third shrinks and disappears from ( d ) through ( f ) .
the appearance of the same components in multiple rows along the slit indicates that we are obtaining multiple measurements of spatially - resolved knots , the brightest of which can be seen in figure [ intensity ] . for comparison purposes , we separated these components by relative intensities and plotted them in different colors ( later in the paper ) .
+ the central peaks of the gaussians give us the central wavelengths from which we used the doppler formula to find radial velocities for each component .
there are three sources of uncertainty in the velocity measurements .
the first results from the fact that the [ o @xmath0iii]emission lines are not gaussians as shown in figure [ gaussfits ] . by actually measuring the centroid of a line for a number a spectra
, we found that our gaussian fitting routine introduced a maximum fitting error of 0.75 in the centroid .
the second error comes from emission cloud displacements from the center of the [email protected] slit in the dispersion direction .
we calculated that a [email protected] offset would contribute an error of 0.56 .
noisy spectra introduced the third error .
various noisy spectra was measured several times and the wavelength shifts were found to fluctuate by 0.30 maximum .
the errors were converted to velocities and added in quadrature to produce a total maximum error of @xmath13 60 km s@xmath4 .
we also varied the continuum placement for randomly selected spectra but found that this did not shift the central wavelength significantly .
+ the s / n ratio of the integrated [ o @xmath0iii ] flux was large near the nucleus , approaching values of 1000 , and then decreased rapidly to @xmath14100 at about 1from the nucleus . the s / n ratio reduced to @xmath14 50 at 3 , @xmath1430 at 4 , and @xmath1510 at 6 .
we did not attempt to measure the emission if the s / n ratio was @xmath163 , which typically occured at distances of @xmath176 .
this trend was seen in all five slits .
+ in paper 1 , mostly bright clouds with one or two kinematic components were detected and fitted to a model . in this paper , our data show multiple components ( 1 , 2 , and 3 ) and hence in order to make a comparison to paper 1 , we separated these kinematic components according to total flux in the line .
we assigned the color red to components with the strongest flux ; blue , to components with the the next strongest flux , and black , to components with the weakest flux .
this kind of separation of the components by fluxes incorporates a minor problem .
consider figure [ cartoon - spectra ] showing a progression of spectra from ( a)(f ) .
the components are ordered 1 , 2 , and 3 going from left to right , and the colors that we assign are marked below each component .
notice that in ( b ) , the second component is labeled black ( since it has less flux under the line ) and the third component is labeled blue . for similar reasons , the black and blue colors switched places again in ( c ) , as a new second component emerged .
the third component shrinked in ( d ) , and disappeared in ( e ) .
the relative nature of our separation criteria is made clear in ( e)(f ) ; with only two components detected , they are labeled red and blue , and when we have detected only one component , we labeled those red . in our velocity plots ,
this results in occasional mixing between the low and medium flux clouds , but does not significantly affect our model , since it was based on a general fit for all points .
the plot of fluxes ( bottom plot ) in figures [ slit1-vel - fwhm - flux ] and [ slit-2to5-vel - fwhm - flux ] clearly show the effects of misidentification are small ; as one can see the colors are clearly separated .
+ figures [ slit1-vel - fwhm - flux ] and [ slit-2to5-vel - fwhm - flux ] show the kinematic components in plots of radial velocities , fwhm , and fluxes as a function of projected position from the nucleus .
negative positions corresponds to the southwest direction .
note that the points tend to clump together , which reflects the ability of stis to resolve the clouds . in the upper plots
, the brightest emission line clouds at each position in the northeast are red shifted , while those in the southwest are blue shifted relative to the host galaxy .
the highest velocities occur closer to the nucleus and velocities as high as 800 km s@xmath4 can be seen in approach and recession .
the general trend is an increase in radial velocity from near zero at the nucleus to a maximum redshift or blueshift within @xmath141 of the nucleus . between @xmath142 to @xmath144 the velocities reduce from about 200 km
s@xmath4 to 100 km s@xmath4 and at distances of about 4 or more from the nucleus , the radial velocities taper off to the enlr rotational velocities .
the medium flux clouds generally show the same pattern as the bright ones except in slits 1 and 4 where they are not confined to only blue shifts in the sw .
also in slit 2 , the medium flux clouds show blueshifts in the ne .
the low flux clouds also follow similar patterns as the bright clouds except in slits 1 and 4 where one can see additional flow patterns opposite to that of the bright clouds .
the fwhm in the middle plots show a range of widths at each position , but a decreasing average width with distance from the nucleus at all flux levels . in the bottom plot ,
all three components show a strong decrease in flux with distance and significant structure that corresponds to the emission line knots .
+ @xcite found a number of low - flux , high velocity clouds in ngc 4151 using a combination of wfpc2 and stis images and stis slitless spectra .
some of these faint clouds did not fit into the general picture of biconical outflow ( e.g. redshifts were detected in the sw where blueshifts dominate ) . to check for consistency , we identified hutchings et al.s clouds in our data and compared our velocities with theirs .
our measurements are closely correlated over a wide range of velocites as shown in figure [ hutchings - das - velocity ] .
also shown on top of the figure are the letters of various high velocity knots , labeled according to hutchings scheme .
the positions of knots e and j in hutchings et al.s image should have placed them in our slits 1 and 5 respectively , but they were not found in our data . on average ,
the difference between hutchings et al.s velocity and ours is 134 km s@xmath4 , but there is no significant systemic offset .
in addition to these knots , we found many more high velocity low and medium emission knots .
these can be seen in the velocity plots of figures [ slit1-vel - fwhm - flux ] and [ slit-2to5-vel - fwhm - flux ] .
in @xcite , the radial outflow from the seyfert 2 galaxy , ngc 1068 was accounted for by a simple biconical radial velocity law , @xmath18 , @xmath19 being the distance from the central source and @xmath20 being a constant .
they used stis low - resolution long - slit spectra taken in the [ o @xmath0iii ] emission regime .
the model bicone was evacuated along its axis up to an inner opening angle of 26@xmath1 from which it was assumed to have a filling factor of 1 up to an outer opening angle of 40@xmath1 .
the gas within the filled region accelerates from rest close to the nucleus to about 1300 km s@xmath4 , at a distance 140 pc away . from this point
it decelerates back to systemic velocity at a distance of about 300pc from the nucleus .
the deceleration of the gas followed a law with the velocity decreasing linearly with distance , @xmath21 .
the best - fit model used an inclination of 5@xmath1 ( relative to the plane of the sky ) for the bicone axis . in paper 1 , the same modeling procedures were used to map the kinematics of ngc 4151 .
the bicones were assumed to have identical geometrical and physical properties and a filling factor of 1 within the cones and no absorption of the [ o @xmath0iii ] photons .
interestingly , the velocity laws @xmath18 and @xmath22 further out also worked in this model . the middle column of table [ compare - ngc4151 ] shows the parameters for the best fit model from paper 1 .
+ in this paper we test the model in paper 1 using higher resolution data and more coverage of the nlr of ngc 4151 .
the kinematic modelling code used in paper 1 was updated to test different velocity laws , corresponding to radial outflow .
parameters for the bicone that were allowed to vary in our code include the extent of each cone along its axis ( @xmath23 ) , the minimum and maximum half opening angles ( @xmath24 and @xmath25 ) , the inclination of its axis out of the plane of the sky ( @xmath26 ) , the velocity turnover radius ( @xmath27 ) ( measured radially outward from the nucleus ) , and the maximum velocity at the turnover point ( @xmath28 ) .
all the above parameters were assumed to be the same for both top ( northeast ) and bottom ( southwest ) parts of the bicone .
we tested various velocity laws for the outflow and found that two of them fit the data reasonably well .
the first law was @xmath18 and @xmath29 for the acceleration and deceleration phases of the gas respectively .
the second law was @xmath30 and @xmath31 for the two phases .
we referred to these laws as the @xmath32 and the @xmath33 respectively , where n = 1 for the @xmath32 , and n = 0 for the @xmath34 . ] .
a constant velocity law and velocity proportional to higher powers of radius did not work well .
motion perpendicular to the bicone axis and infall do not work ( see paper 1 ) .
+ our code generates a two - dimensional velocity map and samples this map with slit positions , orientations , and widths that match those of the observations .
figure [ models ] shows our model velocity field projected onto the sky .
the purple color is the highest negative velocity and the red is the highest positive velocity , with intermediate velocities given in gradients .
next to the fields are velocity extractions from slit 1 done to visually aid the reader .
the extraction consists of four parts , two from the back side of the bicone , and two from the front side .
the upper two graphs consists of velocities extracted from the top and bottom parts respectively , of the back side of the bicone .
similarly the lower two graphs represent the front side of the bicone . in each graph , the relevant extracted section is in shaded grey , in addition to the total extraction from slit 1 , which put things into perspective .
radial velocities were extracted along the five slits and plotted as a function of position in arc seconds .
note that the model was tilted out of the plane of the sky so that the extracted radial velocities would best match the general trend of the data .
the other bicone parameters were then adjusted until the best match between model and data was established . in this fit , emphasis was placed on the high and medium flux points .
the low flux points in slits one and four indicate a flow pattern which in some cases is opposite to that determined from the model ; for the rest of the slits however , there is general agreement .
+ figure [ plotshade ] shows all five slits with the models in shaded grey and the observed data points in colors .
zero position corresponds to the nucleus .
these radial velocity plots show that the northeast side is mostly redshifted , and the southwest side is mostly blueshifted , except for some low luminosity clouds ( black points ) , and a few medium luminosity clouds ( blue points ) , which exhibit an unusual flow pattern .
these are shown by the arrows in figure [ plotshade ] and to some extent , between 0 and -2 in figure [ plotshade ] ( e ) .
plots ( a ) through ( e ) represent models with the @xmath32 and correspond to slits 1 through 5 respectively . plot ( f )
shows a fit for slit one with the @xmath33 .
a closer inspection of plots ( a ) and ( f ) shows minuscule differences in the fit , with ( a ) being slightly better .
minor tweaking of the parameters would result in good agreement for the @xmath35 in slit 1 , however the @xmath32 tends to be more consistent across all five slits .
+ the final model parameters for ngc 4151 are presented in table [ compare - ngc4151 ] . for comparison ,
the fitting parameters from paper 1 are also presented .
following the deceleration trend of the knots in all five slits , the model suggests that the knots return to systemic velocity at a larger distance from the nucleus than previously found in paper 1 .
this extends the bicone longer than in paper 1 , to 400pc as opposed to 288pc .
the turnover point is much closer in , 96pc compared to 162pc and our inner opening angle is 15@xmath1 , which makes our model less evacuated along the bicone axis .
the high luminosity knots seem to travel outward from the nucleus at lower inclination and was modeled in paper 1 as such ; however , in this paper we also incorporate medium and low luminosity knots into the plots and hence our match suggests a larger inclination of 45@xmath1 .
the other parameters , inner and outer opening angles , and maximun velocity of the gas , show only minor differences between the two papers .
the maximum velocity before turnover was increased in our paper to account for the fast moving medium and low luminosity clouds seen in figure [ plotshade ] .
these clouds are accelerated to high speeds close to the nucleus and they therefore play a major role in modelling the turnover point , which we found to be @xmath36 96 pc . if we had used only the bright clouds ( red points ) then our turnover point would have been larger and similar to that of paper 1 .
+ figure [ fwhm - vel ] shows a plot of fwhm versus radial velocity for all five slits combined .
there seems to be no definite correlation between the fwhm and the absolute value of the radial velocity .
this indicates that the directions of velocity dispersion are not preferentially along the direction of the velocity vector .
the fwhm is therefore not due to a purely radial gradient .
however , various low luminosity clouds do show some trend .
figure [ vel - fwhm - low - flux ] shows evidence of a correlation for a combined plot of isolated groups of points shown by the arrows in figure [ plotshade ] .
these groups of points do not follow the general trend of the bright clouds and they consist of only medium and low flux points .
this leads us to believe that these particular knots are either preferentially disturbed in the direction of the velocity vector , or that they exhibit turbulence in proportion to their velocity .
note that the faintest clouds have higher fwhm and radial velocity .
we will look at these rogue clouds more carefully later in the paper . + figure [ bicone ] shows a model of ngc 4151 with the bicone as it should appear in the sky to us , using our final model parameters .
the position angle of the galaxy is 22@xmath1 , with an inclination of 20@xmath1 to the line of sight , the sw side being closer to us .
the position angle of the bicone is 60@xmath1 , and its inclination is 45@xmath1 .
the maximum half opening angle of the bicone is shown here to be 33@xmath1 .
note that our line of sight is outside of the bicone , at an angle of 12@xmath1 with respect to the nearest part of the bicone .
the @xmath37 angle , which is the angle between the bicone axis and the normal to the galactic disk , is 36@xmath1 . due to the bicone
s large opening angle , it will intersect a sufficiently thick galactic disk ( e.g. , a scale height of @xmath36 100 pc ) . the ne bicone will also intersect the ne galactic disk due to symmetry .
this intersection produces the observed geometry of the enlr as shown by @xcite .
the kinematics of the [ o @xmath0iii ] clouds in some seyfert galaxies have been interpreted as outflow away from ( perpendicular to ) the jet axis .
several authors have suggested close relationships between the radio jets and the optical emission such that hot gas around the radio lobes expands and propels the nlr clouds to high velocities ( see gallimore et al .
1996 ; capetti et al .
1997b : axon et al .
although this scenario seemed plausible when applied to ngc 1068 , it failed to explain the turnover velocity where emission knots started to decelerate back to systemic velocity , at about 2from the nucleus @xcite .
also the clumpy nature of the radio knots could not explain the orderly large - scale flow pattern of the radial velocities .
the axis outflow was discussed in @xcite in relation to ngc 4151 .
they found that regardless of inclination , the radial velocities should appear both red - shifted and blue - shifted with equal magnitudes at each position along the axis .
however ngc 4151 has mostly blue - shifted clouds in the southwest region and mostly red- shifted clouds in northeast of the nucleus .
therefore , radial outflow matches the data much better .
+ to test whether the radio knots are responsible for locally disturbing the [ o @xmath0iii ] clouds , we generated plots from radio maps taken by the very long baseline array plus the very large array ( see * ? ? ? * and references therein ) together with radial velocity plots from our data , and compared them within the 2 range that the radio data cover .
we extracted radio emission along the same slit positions as our 5 slit positions in figure [ intensity ] .
the radio fluxes were summed across each [email protected] slit width and normalized to the peak of the brightest radio knot , which was in slit 1 .
radio flux maxima in slits 3 through 5 were only about one tenth the value as those in slits 1 and 2 .
+ figure [ radio - slit - new ] shows our slits overlaid on top of the radio map .
the radio jet is not parallel with the [ o @xmath0iii ] axis , but lies close to one edge of the projected bicone .
slits 1 and 2 contained the two brightest radio knots and the entire image spans about 4 .
figure [ radio - slit - all ] shows intensity plots of the radio emission together with the our radial velocities . here
we expect that bright radio knots would produce roughly symmetric red and blue shifts at the positions of the knots if the axis outflow applies to ngc 4151 .
however we see no evidence for such a scenario in any slit .
figure [ radio - slit - all ] ( b ) shows some blue shifted clouds at the position of the radio knot , but no corresponding red shifted clouds are observed .
we see some red - shifted material in the upper bicone at [email protected] , but this is too far away to be affected by the radio knot . near the brightest radio flux in ( c ) and ( d ) , we see some blue and red shifted [ o @xmath0iii ] clouds respectively , but again there are no symmetric counterparts for these clouds .
similar plots of the velocity widths show no evidence for larger fwhm at the positions of the radio lobes .
we conclude that the radio lobes have no discernible effects on the kinematics of the nlr clouds .
we confirm our previous biconical outflow models @xcite , that were used to match the radial velocities of the nlr in ngc 4151 , using higher velocity - resolution stis spectra . in @xcite ,
the model was based on two slit positions at different position angles . here
we use five parallel slits , all at the same position angle , but different from that used in @xcite .
however , the previous model was found to be consistent with ours , except for a few adjustments .
the turnover point of the velocity ( where clouds start to decelerate ) occurs much closer to the nucleus than the previous model showed , and the point where the clouds velocity approaches systemic velocity is further away .
these two differences follow from the fact that we were able to detect fainter , high - velocity knots closer in to the nucleus .
we note that some of the high - velocity low - flux components do not match the model as shown by the arrows in figure [ plotshade ] .
these high - velocity clouds had been detected in @xcite to have velocities of both approach and recession on both sides of the nucleus .
we see similar trends in figure [ plotshade ] ( a ) , ( b ) , ( d ) , and somewhat in ( e ) . the rest of the low - flux clouds are in agreement with the model
. + the sum of the half opening angle and the inclination of the axis puts the near - sided south - west cone at an angle of 14@xmath1 with respect to the line of sight for paper 1 and 12@xmath1 for this paper .
this means that we are still looking outside the bicone , contrary to the seyfert 1 classification , and general seyfert unification models .
the most likely explanation for an unobscured view of the nucleus is that the bicone edges , and therefore the torus edges , are not sharp .
+ the radio jet lobes do not seem to cause any disturbance to the [ o @xmath0iii ] emission and are not directly responsible for the acceleration of the gas , as is evident from figure [ radio - slit - all ] .
one should see both blue and red - shifts roughly at the same position in these plots , but there is no evidence of this .
the same plots with fwhm also show no correlation between position of the radio lobes and kinematics of the [ o @xmath0iii ] knots .
+ we propose possible scenarios to explain the faint clouds that do not fit the biconical outflow pattern , such as in figure [ plotshade ] ( a ) .
one explanation is that of backflow due to shocks , as a result of outflowing bright clouds impacting on an ambient medium .
the low luminosity clouds could be flowing back along the bicone with a velocity that is proportional to the outflow velocity ; that is , the higher the outflow velocity , the higher the backflow velocity .
the top arrow in figure [ plotshade ] ( a ) points to rogue points which may have resulted from fast outflowing gas in the near side of the southwest bicone , with the low emission gas moving in the opposite direction of general outflow , leaving a trail of clouds in the direction of the nucleus ; see location b in figure [ backflow - cartoon ] . because the near side of the cone has a higher projected velocity , we see the rogue clouds trailing back along this side .
the bright clouds are accelerated in the first few tenths of an arcsec and the rogue points mimic the same behaviour , but in the opposite direction .
the same argument holds for clouds pointed at by the second arrow in figure [ plotshade ] ( a ) .
this backflow most likely comes from the far side of the cone ( d in figure [ backflow - cartoon ] ) , since the velocities and position there agree with those of the rogue clouds .
the knots pointed at by the arrow in figure [ plotshade ] ( b ) can be associated with backflows from the near side of the ne bicone where their positions coincide with those of the bright clouds at location a in figure [ backflow - cartoon ] .
clouds pointed at by the lower arrow in figure [ plotshade ] ( d ) can also be explained by backflow from the near side of the ne bicone .
the velocities and positions of the rogue clouds are similar to those of the clouds in this region .
the argument loses some credibility , however , when we consider the upper arrow in figure [ plotshade ] ( d ) .
these high velocity rogue knots seemed to be geometrically associated with the near side region of the sw bicone , ( b in figure [ backflow - cartoon ] ) , but there are no bright clouds in this region to contribute to the backflow behaviour . + there is another possibility which can explain the unusual flow pattern of all the rogue clouds .
the argument is that the rogue clouds are moving in an entirely different geometry flowing away from the nucleus as shown in figure [ backflow - cartoon ] , locations e , f , g , and h. note that backflows from a , b , c , and d , are equivalent to projected velocities from g , f , e , and h , respectively .
this geometrical equivalent of the backflows is but one of the ways in which the projected velocities agree , since the geometry could be positioned anywhere along the line of sight to yield the same projected velocities .
we adopt the geometry whereby clouds emanate from the central nucleus , as this scenario seems to be the most plausible one .
if that is the case then the flows from locations e and f would be at a rather large angle ( near 90 degrees ) with respect to the bicone axis .
this would suggest that clouds are flowing close to the plane of the torus , which might explain their low flux .
the problem with this scenario is that there is no means for the clouds to be ionized unless radiation is coming from holes in the torus .
it is also not clear how these clouds would be accelerated without radiation or winds from the nucleus , as these clouds do indeed attain high velocities .
+ we prefer the first scenario , because the rogue clouds appear at the positions of the turnover velocities of the bright clouds .
this supports the idea ( paper 1 ) that the bright clouds may be plunging into an ambient medium at the turnover point which cause the clouds to slow down and eventually return to zero velocity , at the same time producing the faint clouds in backflow
. the second scenario seems less likely , because there is no other evidence at present for a leaky torus .
we thank c. mundel for providing the radio map of ngc 4151 used in this study .
+ the observations for this research were obtained as part of a nasa guaranteed time observer allocation to the stis science team .
@lll @xmath23 & 288 pc & 400 @xmath39 16 pc + @xmath24 & 20@xmath1&15 @xmath40 2@xmath1 + @xmath25 & 36@xmath1&33 @xmath40 2@xmath1 + @xmath26 & 40@xmath1(sw is closer ) & 45 @xmath40 5@xmath1(sw is closer ) + @xmath28 & 750 km s@xmath4&800 @xmath39 50 km s@xmath4 + @xmath27 & 162 pc & 96 @xmath40 16 pc + | using _ the hubble space telescope s _ space telescope imaging spectrograph ( _ hst s stis _ ) , observations of the [ o @xmath0iii ] emission from the narrow - line region ( nlr ) of ngc 4151 were obtained and radial velocities determined .
five orbits of _ hst _
time were used to obtain spectra at five parallel slit configurations , at a position angle of 58@xmath1 , with spatial resolution [email protected] across and [email protected] along each slit .
a spectral resolving power of @xmath29,000 with the g430 m grating gave velocity measurements accurate to @xmath334 km s@xmath4 .
a kinematic model was generated to match the radial velocities , for comparison to previous kinematic models of biconical radial outflow developed for low - dispersion spectra at two slit positions .
the new high - resolution spectra permit the measurement of accurate velocity dispersions for each radial - velocity component .
the full - width at half - maximum ( fwhm ) reaches a maximum of near the nucleus , and generally decreases with increasing distance to about 100 km s@xmath4 in the extended narrow - line region ( enlr ) , starting at about 6 from the nucleus .
in addition to the bright emission knots , which generally fit our model , there are faint high velocity clouds which do not fit the biconical outflow pattern of our kinematic model .
these faint clouds occur at the turnover points of the outflowing bright clouds .
we suggest possible scenarios that could explain these rogue clouds : ( 1 ) backflow resulting from shocks and ( 2 ) outflow outside of the bicones , although the latter does not explain how the knots are ionized and accelerated .
a comparison of our observations with a high - resolution radio map shows that there is no evidence that the kinematics of the nlr clouds are affected by the radio lobes that comprise the inner jet . |
the density operator encapsulates all the statistical information about the state of a quantum system .
the evolution of the density operator of a closed system is governed by the hamiltonian . in practice
, the hamiltonian can seldom be strictly determined or precisely controlled it fluctuates both in a temporal sense and between repeated realizations , which can be mathematically described by random processes .
therefore , instead of treating any hamiltonian as deterministic in an idealized manner , we would like to take such fluctuations into account explicitly when studying quantum dynamics . our goal is to obtain the _ average _ dynamics in the following sense : suppose an ensemble of systems are prepared in some initial state and subsequently evolve under a randomly fluctuating hamiltonian , how does the density matrix that describes the ensemble as a whole evolve ?
previous work on stochastic average dynamics was done by budini @xcite using a variational calculus method and novikov s theorem , and by guha et al .
@xcite using a non - perturbative cluster cumulant method .
a different kind of average dynamics over the time domain was studied by gamel and james @xcite , assuming the deterministic ( i.e. non - stochastic ) hamiltonian but taking into account the finite time - window of measurements .
the master equation formalism is also widely used in the study of open systems dynamics @xcite .
it should be noted that , despite the formal similarity , our study is on the dynamics of _ closed _ quantum systems and no environment is involved . in this paper , we will first adopt a series expansion approach and derive a time - local master equation that describes the ensemble - average dynamics of a general quantum system .
the general formalism is then used to study a representative class of hamiltonians obeying gaussian statistics . finally , we apply the master equation method to some physical examples and find interesting phenomena such as fluctuation - induced decoherence and decoherence - induced disentanglement . throughout ,
our results are compared to exact dynamics and the validity of the master equation approach is discussed .
consider a closed , but not isolated , system for which the hamiltonian is determined by some classical stochastic quantity @xmath0 .
suppose an experiment is carried out repeatedly with each realization labelled by @xmath1 .
the evolution of the density matrix @xmath2 that describes the quantum system in the @xmath1-th realization is governed by the hamiltonian @xmath3 $ ] , and is given by @xmath4 where @xmath5 is the initial density matrix , which is assumed to be uncorrelated with @xmath0 and thus is the same in all realizations .
the unitary evolution operator @xmath6 obeys the equation of motion , @xmath7 the average density matrix @xmath8 is defined as follows , @xmath9 it can be shown that @xmath8 is hermitian , positive and of unit trace , which is ensured by the properties of the individual density matrices @xmath2 .
thus the operator @xmath8 is indeed a physical density matrix , describing the average statistics of the ensemble of realizations as a whole .
the equation of motion for @xmath8 is formally given by @xmath10=\overline{[\hat{h}(t),\rho(t)]}.\ ] ] however , since the right hand side can not be written as a function of @xmath8 , the equation is not of a closed form and thus not very useful . with the goal of obtaining a closed equation for @xmath8 , we resort to a series expansion approach . following the standard recipe for perturbative expansion @xcite , the unitary operator @xmath6 in a particular realization @xmath1 can be written as a power series in @xmath11 ( a parameter controlling the `` strength '' of the hamiltonian ) : @xmath12 where @xmath13 thus @xmath8 can be expressed in terms of @xmath14 and @xmath11 : @xmath15\equiv\mathcal{e}[\rho_{0}],\label{eq : map}\end{gathered}\ ] ] where @xmath16 $ ] is the time - dependent map defined as @xmath17\equiv\overset{k}{\underset{j=0}{\sum}}\overline{\hat{u}_{k - j}(t , t_{0})\rho_{0}\hat{u}_{j}^{\dagger}(t , t_{0})},\label{eq : define maps}\ ] ] and @xmath18 $ ] is a completely positive linear map @xcite .
although their argument is a density matrix in this instance , @xmath19 and @xmath20 can act on any operator in general .
the map @xmath20 is a linear transformation that maps @xmath5 to @xmath21 .
since both @xmath5 and @xmath21 are operators in the same hilbert space , and thus of the same dimension , it is natural to postulate that an inverse transformation @xmath22 exists that maps @xmath21 to @xmath5 .
that is , @xmath23\equiv\mathcal{f}[\overline{\rho}].\label{eq : inverse map}\ ] ] note that the meaning of `` inverse '' is purely mathematical here : the map @xmath24 is not to be confused with an inverse dynamical evolution in the physical sense .
according to the semigroup property , a completely positive , trace preserving ( cptp ) linear map is physically invertible if and only if it is an unitary map ( see section 3.8 of ref.@xcite ) .
therefore , in general , a cptp map @xmath25 does not have a physical inverse , that is , we can not find another cptp map that gives @xmath26 .
however , the mathematical inverse @xmath24 that serves our purpose here needs not be cptp . since the composition of a transformation and its inverse is the identity transformation , the identity @xmath27=\mathcal{i}[\rho]$ ] holds for an arbitrary operator @xmath28 .
following @xcite , we adopt the ansatz that @xmath29 can be expanded in powers of @xmath11 , @xmath30=\overset{\infty}{\underset{m=0}{\sum}}\lambda^{m}\mathcal{f}_{m}[\rho]$ ] .
then we have @xmath31=\overset{\infty}{\underset{k=0}{\sum}}\lambda^{k}\left(\overset{k}{\underset{j=0}{\sum}}\mathcal{f}_{j}[\mathcal{e}_{k - j}[\rho]]\right)=\lambda^{0}\mathcal{i}[\rho].\ ] ] collecting terms of like powers in @xmath11 , we obtain the set of equations involving @xmath32 and @xmath33 : @xmath34 & = & \mathcal{i}[\rho],\\ \mathcal{f}_{0}[\mathcal{e}_{1}[\rho ] ] & + & \mathcal{f}_{1}[\mathcal{e}_{0}[\rho]]=0,\\ \mathcal{f}_{0}[\mathcal{e}_{2}[\rho ] ] & + & \mathcal{f}_{1}[\mathcal{e}_{1}[\rho]]+\mathcal{f}_{2}[\mathcal{e}_{0}[\rho]]=0,\end{aligned}\ ] ] and so on . solving for @xmath32 in terms of @xmath33 , and making use of @xmath35 as defined in eq.([eq : define maps ] ) , we have @xmath36 & = & \mathcal{e}_{0}[\rho]=\mathcal{i}[\rho],\\ \mathcal{f}_{1}[\rho ] & = & -\mathcal{e}_{1}[\rho],\\ \mathcal{f}_{2}[\rho ] & = & -\mathcal{e}_{2}[\rho]+\mathcal{e}_{1}[\mathcal{e}_{1}[\rho]],\end{aligned}\ ] ] and so on .
differentiating eq.([eq : map ] ) with respect to time and making use of the inverse relation in eq.([eq : inverse map ] ) , we obtain the following equation : @xmath37=i\hbar\dot{\mathcal{e}}[\mathcal{f}[\overline{\rho}(t)]].\ ] ] here , the notation @xmath38 $ ] means first taking time - derivative of the time - dependent transformation @xmath20 to obtain a new transformation denoted by @xmath39 and then letting @xmath39 act on @xmath28 ; the argument @xmath28 is not differentiated whether or not it is time - dependent .
assuming the order of differentiation and summation can be switched , we have @xmath40=\overset{\infty}{\underset{n=0}{\sum}}\lambda^{n}\dot{\mathcal{e}_{n}}[\overset{\infty}{\underset{m=0}{\sum}}\lambda^{m}\mathcal{f}_{m}[\overline{\rho}(t)]]=\overset{\infty}{\underset{k=0}{\sum}}\lambda^{k}\left(\overset{k}{\underset{j=0}{\sum}}\dot{\mathcal{e}_{j}}[\mathcal{f}_{k - j}[\overline{\rho}(t)]]\right);\ ] ] thus the equation of motion can be written as @xmath41\right)\equiv\overset{\infty}{\underset{k=0}{\sum}}\lambda^{k}\mathcal{l}_{k}[\overline{\rho}(t)].\label{eq : me - series expansion}\ ] ] evaluating @xmath32 and @xmath42 explicitly , we find @xmath43 & = & i\hbar\dot{\mathcal{e}_{0}}[\mathcal{f}_{0}[\rho]]=0,\\ \mathcal{l}_{1}[\rho ] & = & i\hbar\dot{\mathcal{e}_{0}}[\mathcal{f}_{1}[\rho]]+i\hbar\dot{\mathcal{e}_{1}}[\mathcal{f}_{0}[\rho]]=\overline{\hat{h}}\rho-\rho\overline{\hat{h}},\\ \mathcal{l}_{2}[\rho ] & = & i\hbar\dot{\mathcal{e}_{0}}[\mathcal{f}_{2}[\rho]]+i\hbar\dot{\mathcal{e}_{1}}[\mathcal{f}_{1}[\rho]]+i\hbar\dot{\mathcal{e}_{2}}[\mathcal{f}_{0}[\rho]]\nonumber \\ & = & \overline{\hat{h}\hat{u_{1}}}\rho-\overline{\hat{h}}\,\overline{\hat{u}_{1}}\rho+\overline{\hat{h}\rho\hat{u}_{1}^{\dagger}}-\overline{\hat{h}}\rho\overline{\hat{u}_{1}^{\dagger}}\nonumber \\ & & -\rho\overline{\hat{u}_{1}^{\dagger}\hat{h}}+\rho\overline{\hat{u}_{1}^{\dagger}}\,\overline{\hat{h}}-\overline{\hat{u}_{1}\rho\hat{h}}+\overline{\hat{u}_{1}}\rho\overline{\hat{h}},\label{eq : l(2)}\end{aligned}\ ] ] and so on . note again that the argument @xmath28 is not to be averaged or differentiated and that terms like @xmath44 are time - dependent just as @xmath45 $ ] are time - dependent transformations . keeping terms up to 2nd order and setting @xmath46 in eq.([eq : me - series expansion ] ) , a time - local master equation is thus obtained for the evolution of @xmath8 : @xmath47+\hat{a}\overline{\rho}(t)-\overline{\rho}(t)\hat{a}^{\dagger}+\mathcal{d}[\overline{\rho}(t)],\ ] ] where @xmath48 and @xmath49\equiv\overline{\hat{h}\rho\hat{u}_{1}^{\dagger}}-\overline{\hat{h}}\rho\overline{\hat{u}_{1}^{\dagger}}-\overline{\hat{u}_{1}\rho\hat{h}}+\overline{\hat{u}_{1}}\rho\overline{\hat{h}}$ ] .
the effective hamiltonian responsible for unitary evolution is @xmath50 with which the master equation can be written in a more insightful way , @xmath51+\frac{1}{2}\left\ { \hat{a}-\hat{a}^{\dagger},\overline{\rho}(t)\right\ } + \mathcal{d}[\overline{\rho}(t)].\label{eq : general me}\ ] ] it can be shown that the right - hand side of the equation can be put into the lindblad form , which ensures hermiticity , complete positivity and trace preservation of the evolution .
this result is formally similar to a previous work on average dynamics @xcite .
however , the physical meaning is different since the derivation in that case is for a time - average density matrix in a _
single _ realization .
incidentally , our result may also be reminiscent of some master equations for the reduced density matrix of open quantum systems .
but it should be emphasized that our derivation is for a closed system and thus quantum entanglement with environment does not play a role here .
note that the above results are formally applicable to an interaction - picture density matrix , though we implicitly assume the schrdinger picture in the derivation .
the only difference is in the interpretation of the density matrix : when we use the interaction - picture density matrix @xmath52 , the expectation value of an observable @xmath53 is given by @xmath54 , where @xmath55 is the interaction - picture operator instead of the original operator in schrdinger picture .
let us first apply our general result to the simple case where the parameters in the hamiltonian are time - independent .
that is , @xmath56 , where @xmath57 represents random variable(s ) instead of random process(es ) .
suppose further that @xmath58 is of zero - mean , which implies a particular choice of `` picture '' : any time - independent , deterministic part of the hamiltonian plus the average component of the stochastic part can be removed by a gauge transformation , that is , by switching to an suitably chosen interaction picture @xcite .
note that , in the case of time - independent random variables , @xmath59 and @xmath60 , thus @xmath61=[\hat{u}_{1}^{\dagger},\,\hat{h}]=0 $ ] .
so we have @xmath62 and thus @xmath63 .
this result is special to the time - independent case , however . as we will see later ,
the effective hamiltonian ( to 2nd order ) is in general non - zero , due to the non - commutativity of @xmath64 and @xmath58 in the time - dependent case .
after simplification , the 2nd - order master equation in this particular case is @xmath65 a class of problems of physical interest has a hamiltonian of the form @xmath66 where @xmath67 are jointly circular complex gaussian random variables of zero mean . substituting eq.([eq : h ] ) into eq.([eq : t - indep me ] ) , we find on the right - hand side of eq.([eq : me ] ) : it is just a result of a time integral , as can be seen in the more general case of time - dependent hamiltonians later .
note that the trace - preserving property of the equation is guaranteed by the lindblad form . ]
@xmath68 where @xmath69 are the correlation functions .
note that the equation is of the familiar lindblad form , which can be further simplified to a diagonal form through a linear transformation of the coefficients . now consider the case of a time - dependent hamiltonian @xmath70 where @xmath71 is a ( real ) gaussian random process .
this is representative of a wide class of problems , for example , the zeeman effect , where @xmath71 is proportional to the external magnetic field and @xmath72 is the @xmath73-component of the total angular momentum @xcite .
the random process in this section is taken to be the most general case , that is , we do not assume any additional property like zero - mean or stationarity .
the ensemble - average dynamics under this hamiltonian is exactly solvable , so let us first work out the exact , analytic result .
the unitary evolution operator in a particular realization @xmath1 is @xmath74 where @xmath75 for an initial state @xmath76 , where @xmath77 is the energy - eigenbasis with @xmath78 , the evolution in a particular realization is @xmath79 thus the ensemble - average is @xmath80 invoking the special properties of gaussian statistics , is a linear filtered gaussian random process , it is a gaussian random process itself .
( see page 83 of ref.@xcite . ) ] it can be shown that @xmath81\right\ } .\ ] ] thus , the exact ensemble - average dynamics is given by the elements of the average density matrix : @xmath82\right\ } .\label{eq : exact off - diag}\end{aligned}\ ] ] now let us solve the same problem by the master equation approach . using the results from eqs.([eq : effective h]-[eq : general me ] ) , the following expression is obtained , @xmath83\\ + i\hbar\left[\overline{a(t)}\intop_{t_{0}}^{t}dt'\overline{a(t')}-\intop_{t_{0}}^{t}dt'\overline{a(t)a(t')}\right]\left(\hat{h}^{2}\overline{\rho}(t)+\overline{\rho}(t)\hat{h}^{2}\right)\\ + 2i\hbar\left[\intop_{t_{0}}^{t}dt'\overline{a(t)a(t')}-\overline{a(t)}\intop_{t_{0}}^{t}dt'\overline{a(t')}\right]\hat{h}\overline{\rho}(t)\hat{h},\end{gathered}\ ] ] which can be simplified to @xmath84+d(t)\left[\hat{h},[\hat{h},\overline{\rho}(t)]\right],\label{eq : me2}\ ] ] where @xmath85 to find the solution to this differential equation , we first write it down in terms of the matrix elements in the @xmath72-eigenbasis : @xmath86\overline{\rho}_{kl}(t),\,\,(k\neq l).\end{aligned}\ ] ] now we have a set of ( de - coupled ) linear ordinary differential equations ( ode s ) _ , _ which is easily solvable , @xmath87\right\ } \nonumber \\ & = & \rho_{kl}(t_{0})\exp\left\ { -i(e_{k}-e_{l})\overline{v(t)}-\frac{(e_{k}-e_{l})^{2}}{2}[\overline{v(t)^{2}}-\overline{v(t)}^{2}]\right\ } .\label{eq : off - diagonal}\end{aligned}\ ] ] the second equality in eq.([eq : off - diagonal ] ) is obtained after some calculation , where @xmath88 is given by eq.([eq : v ] ) .
thus we find the dynamics generated by the 2nd - order master equation _ coincides _ with the exact dynamics in this case .
we observe that the energy population is conserved during the evolution while the coherence between different energy levels decays .
thus the evolution of the average dynamics is pure decoherence , with the `` pointer basis '' @xcite being the energy - eigenbasis .
note that , although the hamiltonian varies with time and across different realizations , the energy - eigenbasis is the same throughout . in the case
where some energy level is degenerate , we readily have a `` decoherence - free subspace '' , in which quantum information can be stably stored @xcite .
incidentally , a derivation in the context of open systems also suggests that energy eigenstates can emerge as `` pointer states '' in the so - called `` quantum limit of decoherence '' @xcite . in that case
, however , the decoherence results from small environmental perturbation to the system , not from the fluctuation of the system hamiltonian itself .
we could have worked out the higher - order terms ( i.e. @xmath89 $ ] for @xmath90 ) explicitly to see how accurate the 2nd - order approximation is . however , since the solution to the 2nd - order master equation coincides with the exact dynamics , we can readily conclude that all higher - order terms must sum up to zero without actually carrying out further calculations .
note that when deriving eq.([eq : me2 ] ) we do not assume anything about the nature of the random process @xmath71 , not even the gaussian statistics .
in other words , the solution to the 2nd - order master equation is given by eqs.([eq : diagonal]-[eq : off - diagonal ] ) in all cases . on the other hand , the exact dynamics eqs.([eq : exact diag]-[eq : exact off - diag ] ) is based on the assumption of gaussian statistics .
if @xmath71 is _ not _ a gaussian random process , then the exact dynamics will be different .
the implication is that , for @xmath71 being non - gaussian , the 2nd - order master equation is not exact .
let us briefly present the results for the more general hamiltonian @xmath91 , where @xmath92 are jointly circular complex gaussian random processes of zero mean .
the 2nd - order master equation in lindblad form is found to be @xmath93,\,\overline{\rho}(t)\right]+\underset{k , l}{\sum}\beta_{kl}(t)\\ \times\left\ { -\hat{h}_{k}\hat{h}_{l}^{\dagger}\overline{\rho}(t)-\overline{\rho}(t)\hat{h}_{k}\hat{h}_{l}^{\dagger}+2\hat{h}_{l}^{\dagger}\overline{\rho}(t)\hat{h}_{k}-\hat{h}_{l}^{\dagger}\hat{h}_{k}\overline{\rho}(t)-\overline{\rho}(t)\hat{h}_{l}^{\dagger}\hat{h}_{k}+2\hat{h}_{k}\overline{\rho}(t)\hat{h}_{l}^{\dagger}\right\ } , \label{eq : me3}\end{gathered}\ ] ] where @xmath94 @xmath95 by comparing with eq.([eq : me ] ) for the time - independent hamiltonian case , we notice a major difference in this case is that the effective hamiltonian is non - zero despite @xmath96 .
this effective unitary evolution results from the fact that the hamiltonian operators at different times do not commute with each other in general .
the 2nd - order master equation yields exact dynamics only for the special case of a single real gaussian random process . in this more general case , eq.([eq : me3 ] ) does not lead to exact dynamics in general .
this can be shown by explicitly evaluating higher - order terms like @xmath97 $ ] to find that they do not vanish in general .
despite the lack of perfect agreement , the master equation is nevertheless of great use in such cases , because the exact dynamics is generally not obtainable and the 2nd - order master equation serves as a good approximation when the higher - order terms ( e.g. @xmath97 $ ] ) are small compared to @xmath98 $ ] .
we will illustrate the general results by applying them to a few examples .
the findings will then be used to gain physical insights , and the validity of the master equation approach will be examined by comparing to the exact dynamics .
first consider a two - level system ( e.g. spin-1/2 ) subject to the hamiltonian @xmath99 , where @xmath100 is the @xmath73-component of pauli operator and @xmath101 a stationary gaussian random process of zero mean .
this falls into the category of hamiltonians ( [ eq : h - single grp ] ) . using eq.([eq : me2 ] )
, the 2nd - order master equation is obtained , @xmath102\right],\ ] ] where @xmath103 .
assuming an auto - correlation function of the form @xmath104 , where @xmath105 , we have @xmath106 for @xmath107 . written in the @xmath100-eigenbasis @xmath108 , the master equation becomes a set of linear ode s : @xmath109 the solutions to which are @xmath110 as already discussed in the general case of a single gaussian random process , the energy population remains constant while the coherence decays .
this can be understood from a more physical perspective .
quantum coherence depends on the relative phase between the two components @xmath111 and @xmath112 . in an individual realization ,
the relative phase factor is @xmath113 .
since @xmath101 is a random process , the quantity @xmath114 becomes increasingly randomized with the passage of time .
when the average is taken over an ensemble , these randomly distributed relative phase factors cancel out , thus suppressing the coherence .
this suggests that quantum interference is difficult to observe because random fluctuation is ubiquitous .
as has been shown in the more general case eqs.([eq : diagonal]-[eq : off - diagonal ] ) , the 2nd - order master equation gives exact dynamics .
when we work out the exact dynamics directly , the result is indeed found to be consistent , though such a direct calculation is more demanding . clearly , calculational convenience is one advantage of the master equation approach .
next consider an example of two atoms in magnetic field , each atom being a two - level system .
the interaction of the spin with the b - field is given by @xmath115 , where @xmath116 is the usual pauli @xmath73-operator of the @xmath117-th atom .
suppose that the frequency @xmath118 , which is proportional to the b - field strength , is a stationary gaussian random process of zero mean and that its auto - correlation is of a markovian type @xmath119 , where the constant @xmath120 has dimension of inverse - time .
applying eq.([eq : me2 ] ) , it can be shown that the 2nd - order master equation is @xmath121\right].\label{eq : me-2 atoms}\ ] ] suppose the system starts in an entangled state between two atoms @xmath122 . since it is an eigenstate of @xmath123 , the right - hand side of eq.([eq : me-2 atoms ] ) is identically zero , thus the system does not evolve ( except possibly to an unobservable global phase ) .
so the two atoms remain entangled over time . indeed ,
notice that @xmath124 and @xmath125 are degenerate eigenstates with the same energy .
thus any arbitrary superposition state of @xmath124 and @xmath125 will remain unchanged over time ; in particular , the coherence between them does not decay .
thus , any state in this degenerate subspace is immune to decoherence , making it a good place to store quantum information @xcite .
let us see what happens if the 2-atom system starts in a different entangled state like @xmath126 . writing the master equation in the energy eigenbasis @xmath127
, we obtain a set of decoupled linear ode s as usual .
the solutions are found to be @xmath128 while the rest of the matrix elements are identically zero .
note that decoherence occurs here , as is expected , since the initial state does not lie in the decoherence - free subspace .
furthermore , as @xmath129 , the coherence is suppressed to zero and @xmath130 .
interestingly , the two atoms become _ disentangled _ , as there is no quantum correlation between them .
in contrast to the general belief that entanglement leads to decoherence , as is widely studied for open quantum systems , here we find that decoherence can result in _ _ disentanglement _ _ in the case of a closed system .
consider an ion with mass @xmath131 and charge @xmath132 in a harmonic binding potential with characteristic frequency @xmath133 .
the ion is driven by a classical electric field @xmath134 , which is a stationary gaussian random process of zero mean .
it is more convenient to work in the interaction picture , in which the easily solvable , deterministic evolution induced by the harmonic potential is treated separately .
the interaction - picture hamiltonian is given by @xmath135 $ ] , where @xmath136 and @xmath137 @xmath138 being the zero - time annihilation ( creation ) operator for the harmonic oscillator .
the evolution is exactly solvable and the analytic results are given in @xcite .
let us derive the 2nd - order master equation for this case .
note that it does not fall into the category of a single real gaussian random process as in eq.([eq : h - single grp ] ) . since @xmath139 , applying eq.([eq : effective h ] )
, the effective hamiltonian is found to be @xmath140\nonumber \\ & = & -\frac{e^{2}}{2m\omega_{0}}\intop_{t_{0}}^{t}dt'\overline{e(t)e(t')}\sin\left[\omega_{0}(t - t')\right]\hat{i}.\end{aligned}\ ] ] in this case , the 2nd - order contribution to the effective hamiltonian is non - zero , a consequence of the non - commutativity of @xmath141 and @xmath142 . however , since @xmath143 is proportional to @xmath144 , the unitary part of the equation of motion results only in an unobservable global phase in this case .
for more general cases , @xmath143 can be different from the identity @xmath144 and can well lead to non - trivial dynamics .
evaluating the remaining terms in eq.([eq : l(2 ) ] ) for this example , we find the following master equation : @xmath145\left((\hat{a}^{\dagger})^{2}\overline{\rho}(t)+\overline{\rho}(t)(\hat{a}^{\dagger})^{2}-2\hat{a}^{\dagger}\overline{\rho}(t)\hat{a}^{\dagger}\right)\\ -e^{-2i\omega_{0}t}\left[\mathcal{c}(t)+i\mathcal{s}(t)\right]\left(\hat{a}^{2}\overline{\rho}(t)+\overline{\rho}(t)\hat{a}^{2}-2\hat{a}\overline{\rho}(t)\hat{a}\right),\end{gathered}\ ] ] where @xmath146 ( @xmath147 ) are proportional to the incomplete cosine ( sine ) transform of the field correlation function , viz @xmath148,\\ \mathcal{s}(t ) & \equiv & \frac{e^{2}}{2m\hbar\omega_{0}}\intop_{t_{0}}^{t}dt'\overline{e(t)e(t')}\sin\left[\omega_{0}(t - t')\right].\end{aligned}\ ] ] assuming @xmath149 and setting @xmath150 for convenience , we find @xmath151 + 1\right\ } $ ] and @xmath152-\omega_{0}t\right\ } $ ] , where @xmath153 is the heating time defined as @xmath154 . unlike the case of eq.([eq
: h - single grp ] ) , the 2nd - order master equation does not generate exact dynamics in this case . to get an approximation of the heating from the ground state ( i.e.
@xmath155 ) for a short period of time , let us write the master equation in the energy eigenbasis of the harmonic oscillator , @xmath156\overline{\rho}_{02}(t)-\sqrt{2}e^{-2i\omega_{0}t}\left[\mathcal{c}(t)+i\mathcal{s}(t)\right]\overline{\rho}_{20}(t).\end{gathered}\ ] ] since @xmath157 , @xmath158 and @xmath159 are all negligibly small for @xmath160 , we have @xmath161 to the lowest order . in the same manner ,
since the depopulation @xmath162 is perturbatively small , we have @xmath163 to the lowest order on the right - hand side .
thus an approximate differential equation is obtained as @xmath164 . solving this ode
, we find , to lowest order , for @xmath165 using approximation to the same order . ]
@xmath166 which holds for short times and agrees with the analytic result in @xcite . to investigate the evolution of the system for longer times , we write down the master equation in the same basis and solve it numerically . since the hilbert space is of infinite dimensions ,
it is not possible to write down the complete set of ode s for the matrix elements .
instead , we truncate it to a set of @xmath167 coupled ode s that includes only the matrix elements of the five lowest energy - eigenstates and their coherence .
the numerical solutions of @xmath168 ( i.e. fidelity of the ground state ) are shown in figure 1 for different sets of parameters .
it can be seen that , as @xmath169 ( i.e. the dimensionless heating time ) increases with @xmath170 ( i.e. the dimensionless coherence time of @xmath171 ) fixed , the numerical result gives better approximation to the exact dynamics .
also note that , for larger values of @xmath170 , the ground state population shows temporary revival against its general trend of decrease .
despite the artificial defect caused by the truncation of the set of ode s , it is of more interest to know the validity of the 2nd - order master equation itself in approximating the exact dynamics .
this is done by comparing the size of higher - order terms to that of the 2nd - order term . using the gaussian moment theorem@xcite , it is easy to show that @xmath89=0 $ ] for all odd numbers @xmath172 , so we are interested in the ratios between the even - number - order terms .
assuming @xmath170 is fixed , it can be shown that @xmath97\propto1/\tau_{1}^{2}\omega_{0}$ ] as opposed to @xmath98\propto1/\tau_{1}$ ] , so @xmath97/\mathcal{l}_{2}[\rho]\propto1/\omega_{0}\tau_{1}$ ] .
the same ratio holds for @xmath173/\mathcal{l}_{4}[\rho]$ ] , etc .
therefore , as long as @xmath174 is small , the higher - order terms become progressively small , lending legitimacy to the 2nd - order approximation .
this is also consistent with the previous observation from the numerical results .
physically , this can be better understood by switching to the schrdinger picture : the external field @xmath175 is treated as a perturbation to the self - hamiltonian of the system @xmath176 . naturally , as the relative size of the perturbing hamiltonian @xmath177 becomes smaller , a perturbative method such as the 2nd - order master equation gives better approximation to the exact dynamics .
in this paper we have presented the derivation of a master equation for closed systems driven by stochastic hamiltonians from an ensemble - average perspective .
the principal result is given in eqs.([eq : effective h]-[eq : general me ] ) .
the validity of this approach is examined and 2nd - order master equation is found to yield either exact dynamics or good approximations to exact dynamics . applying the formalism to various physical examples , we find the ensemble - average dynamics usually contains decoherence terms in addition to the unitary evolution .
decoherence plays an important role in the foundational problems of quantum mechanics , as it gives insights in two aspects of the measurement problem , namely the absence of observable superposition and the problem of preferred basis @xcite .
extensive research has been done on how environmental entanglement causes decoherence in open systems . however , as our findings suggest , decoherence could also be attributed to the random fluctuations of physical quantities in closed systems .
if this is true , then the tension between the classicality of our experience and the quantumness of the underlying laws of physics could be reconciled in some degree by the ubiquitous random fluctuations .
further investigation is needed to find out ( a ) to what extent decoherence is actually caused by random fluctuations and ( b ) whether / how we can distinguish it from the usual entanglement - induced decoherence through physical observation .
the authors would like to thank o. gamel for valuable discussions and c .- h .
chang for comments on the manuscript .
this work is supported by natural sciences and engineering research council of canada ( nserc ) through create and by university of toronto through utea - nse . | we develop a master equation formalism to describe the evolution of the average density matrix of a closed quantum system driven by a stochastic hamiltonian .
the average over random processes generally results in decoherence effects in closed system dynamics , in addition to the usual unitary evolution .
we then show that , for an important class of problems in which the hamiltonian is proportional to a gaussian random process , the 2nd - order master equation yields exact dynamics . the general formalism is applied to study the examples of a two - level system , two atoms in a stochastic magnetic field and the heating of a trapped ion .
_ department of physics , university of toronto , 60 st .
george street , toronto , ontario m5s 1a7 , canada _ pacs
number(s ) : 03.65.ca , 03.65.yz , 05.40.-a , 03.67.lx |
at present there are two main approaches to modelling galaxy evolution .
the `` traditional '' view is that the most massive galaxies the big ellipticals , s0s and early - type spirals formed first and have been evolving with strongly declining star formation rates since at least a redshift of 2 .
the less massive late - type spiral and irregular galaxies formed their stars at a roughly constant rate over a hubble time and in many cases undergo substantial bursts .
alternatively , according to hierarchical theories of galaxy formation , massive galaxies were assembled recently from mergers of smaller subunits .
traditional models adopt a simple `` backwards - in - time '' technique for calculating how the observed properties of galaxies evolve as a function of redshift ( e.g. tinsley 1980 ; bruzual & kron 1980 ; koo 1981 ; shanks et al 1984 ; king & ellis 1985 ; yoshii & takahara 1988 ; guiderdoni & rocca - volmerange 1990 ) .
the two basic inputs are a ) the observed present - day luminosity function divided by morphological type , and b ) a parametrization of the mean star formation rate history of galaxies of a given morphological type .
these star formation histories are tuned to reproduce the spectral energy distributions of nearby spirals and ellipticals .
it is usually assumed that all galaxies form at the same redshift and evolve as closed box systems .
stellar population synthesis models are used to compute how the galaxy luminosity function evolves as a function of lookback time , and predictions are then made for the counts , redshift distributions and colours of faint galaxies .
hierarchical models of galaxy formation follow the formation and evolution of galaxies within a merging hierarchy of dark matter halos ( e.g. white & rees 1978 ; white & frenk 1991 ; lacey et al 1993 ; kauffmann , white & guiderdoni 1993 ; cole et al . 1994 ; somerville 1997 ) .
an analytic formalism allows the progenitors of a present - day object , such as a cluster , to be traced back to arbitrarily early times .
simple prescriptions are adopted to describe gas cooling , star formation , supernova feedback and the merging of galaxies .
finally , stellar population synthesis models are used to generate luminosity functions , counts and redshift distributions for comparison with observations .
there have been many recent papers claiming that either the traditional models or the hierarchical models can provide good fits to available data on high redshift galaxies .
most of the recent papers discussing traditional models invoke a local luminosity function with a `` steep '' ( @xmath11 ) faint - end slope in order to reproduce the observed high number density of faint blue galaxies .
dust extinction is also included to avoid overproducing blue , star - forming galaxies at high redshift ( gronwall & koo 1995 ; campos & shanks 1997 ; pozzetti et al 1996 ) .
the hierarchical models _ predict _ a local luminosity function with a steep faint - end slope ( perhaps too steep ) and hence have little trouble in matching the counts . because bright galaxies have not yet formed at high redshift
, the hierarchical models can reproduce the redshift distributions without requiring dust ( kauffmann , guiderdoni & white 1994 ; cole et al 1994 ; heyl et al 1995 ; baugh , cole & frenk 1996 ) . in this letter , we present an observational test that is designed to _ differentiate _ between the traditional and hierarchical scenarios . to do this
, we concentrate on the aspect that differs the most between the two pictures the redshift evolution of the most massive galaxies .
we show that the observed k - band flux from a galaxy provides a good measurement of its total stellar mass out to redshifts in excess of 2 for all but the most extreme starbursting systems .
moreover , the k - band is only weakly affected by extinction at these redshifts .
we compute the evolution of the observed k - band luminosity function for traditional and hierarchical models , assuming both high - density ( @xmath12 ) and low - density ( @xmath13 ) cosmologies .
we use the new population synthesis models of bruzual and charlot ( in preparation ) , which include updated stellar evolutionary tracks and new spectral libraries . at @xmath0 , both the traditional models and the @xmath12 hierarchical model can give a good fit to the local k - band luminosity functions of gardner et al ( 1997 ) and szokoly et al ( 1998 ) . by redshift 1 ,
the two give very different predictions for bright galaxies . at @xmath7 ,
the traditional models predict 10 times more galaxies with @xmath14 than do the hierarchical models . at @xmath15 ,
the traditional models predict fifty times more galaxies with @xmath1 .
we then calculate the redshift distributions of galaxies with @xmath1 and compare these with the hawaii deep field samples of songaila et al ( 1994 ) and cowie et al .
we show that the observed number of bright galaxies with redshifts greater than 1 is well below the number predicted by the traditional models . in the songaila et al ( 1994 ) sample of 118 galaxies with @xmath16 ,
33 galaxies are predicted to lie at @xmath3 .
only 2 are observed . in the cowie et al ( 1996 ) sample of 52 galaxies with @xmath17 , 28
are predicted to lie at @xmath5 .
only 5 are observed .
both these samples are more than 90% complete .
we conclude that there is already strong evidence that the abundance of massive galaxies has declined substantially at @xmath6 .
this is inconsistent with the traditional models , unless most massive galaxies are extremely heavily obscured by dust at this redshift , but similar to the expectations of hierarchical models .
in figure 1 , we show the observed k - magnitude of a galaxy with @xmath18 of stars as a function of its observed redshift .
we assume @xmath19 , @xmath20 km s @xmath21 mpc@xmath21 .
( note that this is _ not _ an evolutionary plot for any specific galaxy .
rather it shows the apparent brightness of a galaxy which has @xmath18 of stars _ at the redshift when it is observed .
_ ) the three different curves in the plot are for three possible star formation histories of the galaxy .
the solid line shows the predicted k - band magnitude of the galaxy if all its stars formed in a burst at @xmath22 .
the short - dashed line shows the k - magnitude if the galaxy formed its stars at a constant rate from @xmath23 to the redshift of observation .
the long - dashed line is for a galaxy that formed 80% of its stars at a constant rate , and 20% over the @xmath24 years immediately before it is observed .
this is supposed to represent a galaxy that is undergoing a strong starburst . in the lower panel of figure 1 ,
we plot the observed @xmath25 colours for the same three star formation models .
note that we have adopted a scalo imf with upper and lower mass cutoffs of 100 @xmath26 and 0.1 @xmath26 .
a 10 gyr single - age stellar population has a k - band mass - to - light ratio of @xmath27 , in good agreement with the estimated k - band mass - to - light ratios of elliptical galaxies ( mobasher et al 1998 ) .
figure 1 shows that galaxies of the same stellar mass will have roughly the same observed k - magnitude , independent of their star formation histories . the difference in k - luminosity between the initial burst and constant star formation models is less than a factor 2 .
even the starburst model differs from the initial burst model by less than a factor 3 at @xmath28 .
this is in marked contrast to the observed b - band luminosities , which , for high redshift galaxies , are dominated by the flux from young massive stars .
differences in @xmath25 colour of up to 9 magnitudes are obtained for different star formation models . from figure 1
, one can also infer that evolution at @xmath28 in the abundances of galaxies more massive than @xmath18 should be evident even in relatively bright ( @xmath29 ) k - selected samples .
the traditional or pure luminosity evolution ( ple ) model we employ is similar to that described in pozzetti et al .
the present - day luminosity function is divided into a set of 4 hubble types ( e / s0 , sa - sb , sc - sd , sm - im ) using the results of marzke et al ( 1994 ) on the relative abundances of galaxies of differing morphology in the cfa redshift survey . because the schechter - function @xmath30 derived for the cfa survey is considerably fainter than that found for other recent surveys covering a larger area , we have chosen to normalize the total @xmath31-band luminosity function using the schechter function fits derived for the eso slice project ( esp ) redshift survey ( @xmath32 , @xmath33 , @xmath34 mpc@xmath35 ) ( zucca et al 1997 ) . at the bright end
, the esp luminosity function agrees well with the luminosity functions derived for the stromlo - apm and the las campanas redshift surveys ( loveday et al 1992 ; lin et al 1996 ) .
we compute the spectral evolution of e / s0 , sa - sb and sc - sd galaxies using bruzual & charlot ( 1998 ) models with exponentially declining star formation rates with star formation timescales of 1 , 4 and 15 gyr , respectively .
for the adopted scalo ( 1986 ) initial mass function , solar metallicity and formation redshift @xmath36 , the models provide good fits to the spectral features and the colours of nearby galaxies of these types from the ultraviolet to the infrared ( table 1 ; bruzual & charlot 1998 ) .
local sm - im galaxies are too blue to be fit by models with exponentially declining or even constant star formation rates , if we assume a formation redshift of 5 .
instead , we have adopted a model with constant star formation rate seen at a fixed age of 1.4 gyr .
this model best reproduces the colours and spectral properties of the propotypical sm - im galaxy ngc 4449 ( table 1 ; bruzual & charlot 1998 ) .
since sm - im galaxies contribute very little to the luminosity function at bright magnitudes , their treatment in the models does not affect our conclusions .
our results _ are _ affected by the evolutionary model we adopt for the luminous e / s0 and sa - sb galaxies .
we note that the choice of a scalo imf is a conservative one , since this results in milder luminosity evolution than the salpeter imf , which has a higher fraction of massive stars ( pozzetti et al 1996 ) .
we have not attempted to include any metallicity evolution in the traditional models .
this will also not affect our conclusions since massive galaxies , which have short star formation timescales , quickly reach solar metallicity . because of the simplified way in which they treat the many physical processes which play a role in galaxy formation , hierarchical models have considerably more freedom than the traditional models .
on the other hand the properties of the present galaxy population , which are used as an input in the traditional approach , are an output of the hierarchical models and so can be used to test them . as examples of hierarchical models
, we here use a model from our earlier paper on the systematics of elliptical galaxies together with a low density variant .
the reader is referred to kauffmann & white ( 1993 ) , kauffmann , white & guiderdoni ( 1993 ) and kauffmann & charlot ( 1998 ) for more details of our semi - analytic techniques for modelling the formation and evolution of galaxies in a hierarchical universe . in the current paper
we explore two possibilities : 1 . an @xmath12 `` standard '' cdm model with @xmath37 km s@xmath21 mpc@xmath21 , @xmath38 , and @xmath39 . the star formation and feedback parameters are the same as in model a of kauffmann & charlot ( 1998 ) , which was shown to provide a good fit to the slope and scatter of the colour - magnitude relation of elliptical galaxies in clusters
this model includes a prescription for chemical evolution and makes use of the metallicity - dependent spectral synthesis models of bruzual & charlot ( 1998 ) .
an @xmath13 ( @xmath40 ) cdm model with @xmath37 km s@xmath21 mpc @xmath21 and @xmath41 .
chemical evolution is also included in this model .
redshift zero @xmath42-band luminosity functions for the traditional and hierarchical models are shown as thick lines in figure 2 .
the schechter function fit obtained by szokoly et al ( 1998 ) for a sample of 175 galaxies to @xmath42=16.5 is shown as a dotted line on the plot .
the fit obtained by gardner et al ( 1997 ) for a wide - angle survey of 567 galaxies limited at @xmath42=15 , is shown as a thin dashed - dotted line .
the traditional model and the @xmath12 cdm model both fit the observed local @xmath42-band luminosity function well , particularly at the bright end .
note that in the traditional model , the @xmath42-band luminosity function is computed from the observed @xmath31-band luminosity function and the model @xmath25 colors corresponding to the different morphological types .
the fact that the k - band luminosity function agrees well with the data gives us confidence that the the assumptions made in 3.1 are giving self - consistent results .
the @xmath13 cdm model fails to produce enough galaxies at @xmath30 by almost an order of magnitude .
this failure of low - density models has been noted before ( kauffmann , white & guiderdoni 1993 ) .
baugh et al ( 1998 ) fit the number density of galaxies at the `` knee '' of the luminosity function in a low - density ( @xmath43 ) cdm model by adopting a brighter normalization , but they then vastly overproduce galaxies at the bright end of the luminosity function .
( basically , this is equivalent to shifting our curve to the right ) . because of the failure of low - density hierarchical models to reproduce the _ shape _ of the local @xmath42-band luminosity function
, we will not pursue their predictions any further in this letter .
the predicted evolution of the observed differential @xmath42-band luminosity function is shown in figure 3 .
the solid and dotted lines show results from the traditional models , for @xmath19 and @xmath44 respectively .
the dashed line shows results from the @xmath12 cdm model .
all models assume @xmath37 km s@xmath21 mpc@xmath21 . by @xmath7 ,
the predictions of the traditional and the hierarchical models deviate strongly at the bright end of the luminosity function .
this just reflects the fact that in the hierarchical models , the most massive galaxies are still forming today . in the traditional models ,
the massive galaxies were already in place at @xmath7 and were significantly brighter than at present because their stars were less evolved .
complete @xmath42-selected redshift surveys of faint galaxies are not yet large enough to construct accurate luminosity functions at a series of redshifts .
the young bright galaxies predicted by the traditional models should , however , show up as a `` tail '' of high redshift galaxies in these samples . in the b - band
, the presence of high - z tails in the predicted redshift distributions has been shown to depend sensitively on whether dust extinction is included in the modelling ( gronwall & koo 1995 ; campos & shanks 1997 ) . in the @xmath42-band ,
the effects of dust are not likely to be as significant . in figure 4 , we plot the cumulative fraction of galaxies with redshifts greater than @xmath45 for samples limited at @xmath42=18 , 19 and 21 .
the solid and dotted line show results for the @xmath19 and @xmath44 ple models .
the dashed line is the hierarchical model .
the thin solid lines in figure 4 show recent observational results .
the curve in the top panel is computed using the sample of songaila et al ( 1994 ) and the curve in the middle panel is derived from the hawaii deep field sample of cowie et al ( 1996 ) . in both cases ,
we have chosen conservative limiting magnitudes in order to ensure that the redshift data are nearly complete ( @xmath46 90% in both cases ) and that the derived redshift distributions are not biased by selection effects .
there are 118 galaxies with @xmath16 in the songaila redshift sample and 52 galaxies with @xmath17 in the cowie redshift sample . in the plot
, we do not include galaxies without known redshifts ( 11 galaxies in songaila et al . and 4 galaxies in cowie et al . )
it should be noted that all the models produce roughly the right total number of galaxies per unit area on the sky to @xmath47 .
the @xmath19 ple model fits the counts extremely well to @xmath42=21 .
the @xmath44 model counts are a factor of two too high at @xmath42=21 , because of the increased volume at high redshift in a low - density cosmology .
the hierarchical model counts are also a factor 2 - 3 too high at this magnitude , because the faint end of the @xmath42-band luminosity function is too steep in this model ( see figure 2 ) .
this accentuates the tendency for the hierarchical redshift distributions to be dominated by faint , low - z galaxies .
figure 4 shows that the fraction of galaxies predicted to lie at high redshift is very much larger in the ple models than in the hierarchical model .
it is also apparent that the number of galaxies observed at redshifts greater than @xmath48 falls well below the predictions of the ple models in both the songaila and the cowie data samples . for the @xmath2 sample ,
the ple models predict that 33 galaxies should have been detected at @xmath49 , when in fact only 2 such galaxies are found .
a ks test gives the probability that the songaila sample is consistent with the ple redshift distribution as less than 10@xmath50 , even if all the galaxies without redshifts are arbitrarily assigned to @xmath51 .
a similar result is obtained for the cowie et al . sample .
this shows that our conclusions are not being affected by finite sample statistics . in anticipation of future redshift surveys
complete to even fainter limiting magnitudes , we show the predicted redshift distributions for galaxies with @xmath52 in the bottom panel of figure 4 .
finally , we note that similar results have been found by others .
pozzetti et al ( 1996 ) show that their ple models fail to match the songaila et al .
( 1994 ) redshift distributions .
cowie et al ( 1996 ) note that the median redshift of their sample as a function of observed k magnitude falls below the predictions of models with mild luminosity evolution . in our analysis
, we focus our attention on the evolution of the _ bright end _ of the k - band luminosity function , since this where the predictions of the traditional and the hierarchical models diverge most strongly .
in previous work we showed that the abundance of galaxies with colours consistent with passively evolving early - type galaxies was a factor 2 - 3 lower at a redshift of 1 than at present ( kauffmann , charlot & white 1996 ; see also zepf 1997 ) .
it was not clear from our analysis whether this was because present - day ellipticals were forming stars at high redshift and were thus not as red as specified by simple passive evolution models , or because many ellipticals had simply not yet assembled by @xmath7 .
the analysis presented in this letter is considerably more general in that it applies to the evolution of the _ entire population of massive galaxies_. from our comparison of the data with the models , we conclude that the abundance of massive galaxies is substantially below its present value at @xmath7 , in contradiction with the traditional picture of galaxy formation . if massive galaxies were forming stars more rapidly than we have assumed at high redshift , they would appear even brighter at @xmath42 , and the failure of the traditional picture would be even more significant .
the only remaining `` escape route '' would be for a large fraction of today s massive galaxies to be so heavily obscured by dust at @xmath7 , that even the observed @xmath42-band is significantly affected ( franceschini et al 1997 ) .
this does not seem likely to us , because damped lyman - alpha systems , which are good probes of the conditions within high column density gas clouds , contain rather little dust at z @xmath53 ( pei , fall & bechtold 1991 ; pettini et al .
1997 ) we suggest that massive galaxies have continued forming until recent times through a process of merging and accretion , as predicted by hierarchical theories of galaxy formation .
future wide - area @xmath42-selected redshift surveys will enable this buildup of galaxies to be quantified more accurately .
+ we thank simon white for helpful discussions .
this work was carried out under the auspices of eara , a european association for research in astronomy , and the tmr network on galaxy formation and evolution funded by the european commission . | there are two major approaches to modelling galaxy evolution .
the `` traditional '' view is that the most massive galaxies were assembled early and have evolved with steeply declining star formation rates since a redshift of 2 or higher . according to hierarchical theories ,
massive galaxies were assembled much more recently from mergers of smaller subunits .
here we present a simple observational test designed to differentiate between the two .
the observed k - band flux from a galaxy is a good measure of its stellar mass even at high redshift .
it is likely only weakly affected by dust extinction .
we compute the evolution of the observed k - band luminosity function for traditional , pure luminosity evolution ( ple ) models and for hierarchical models . at @xmath0
, both models can fit the observed local k - band luminosity function . by redshift 1 , they differ greatly in the predicted abundance of bright galaxies .
we calculate the redshift distributions of k - band selected galaxies and compare these with available data .
we show that the number of @xmath1 galaxies with redshifts greater than 1 is well below the numbers predicted by the ple models . in the songaila et al ( 1994 ) redshift sample of 118 galaxies with @xmath2 ,
33 galaxies are predicted to lie at @xmath3 .
only 2 are observed . in the cowie
et al ( 1996 ) redshift sample of 52 galaxies with @xmath4 , 28 galaxies are predicted to lie at @xmath5 .
only 5 are observed .
both these samples are more than 90% complete .
we conclude that there is already strong evidence that the abundance of massive galaxies at @xmath6 is well below the local value .
this is inconsistent with the traditional model ( unless most massive galaxies are extremely heavily obscured by dust at redshift 1 ) , but similar to the expectations of hierarchical models .
-0.55 cm -0.55 cm 0.1 mm 16.75 cm the k - band luminosity function at @xmath7 : a powerful constraint on galaxy formation theory + guinevere kauffmann@xmath8 & stphane charlot@xmath9 + + _ @xmath10 institut dastrophysique du cnrs , 98 bis boulevard arago , f-75014 paris , france _
+ keywords : galaxies : formation , evolution ; galaxies : stellar content |
inclusive decays for heavy - quark hadrons such as @xmath1 mesons are computed in a systematic expansion in inverse powers of the heavy quark mass using the operator product expansion ( ope ) .
decay rates are related to the discontinuity of scattering amplitudes at physical cuts , so the calculation is necessarily performed for time - like momentum transfer and is not a short - distance process for which the ope is valid .
nevertheless , one can still use the ope for suitably averaged quantities .
the simplest example is @xmath5 , the inclusive rate for @xmath6 , which is related to the discontinuity in the vacuum polarization amplitude @xmath7 along the real axis for @xmath8 .
the operator product expansion can be applied to @xmath7 far away from the physical cuts , such as at complex @xmath9 , @xmath10 , to give @xmath7 as an expansion in @xmath11 and @xmath12 .
analyticity can then be used to relate @xmath13 to an integral of the form @xcite @xmath14 thus smeared averages of @xmath5 can be computed using perturbation theory .
the @xmath12 corrections to these depend on the smearing width @xmath15 , and are not under control as @xmath16 . in practice @xmath15
is made as small as possible while keeping the radiative corrections under control .
typically @xmath15 is chosen to be @xmath17mev .
one expects that perturbative computations agree with data , provided one averages over a large number of hadronic resonances .
a similar situation exists in inclusive semileptonic @xmath1 decays , @xmath18 @xcite .
the differential decay distribution @xmath19 can be computed using the operator product expansion @xcite . here
@xmath20 is the invariant mass of the lepton pair , and @xmath21 and @xmath22 are the energy of the electron and neutrino , respectively .
the total semileptonic decay rate @xmath23 automatically involves averaging over different hadronic states , and one expects that the ope computation for @xmath23 is valid .
inclusive nonleptonic @xmath24 decays can also be computed using an ope @xcite . in this case
, there is no variable such as @xmath20 to integrate over , so the computation has no intrinsic smearing .
one must then make the assumption of local duality , that the quark and hadron computations agree at a single kinematic point . in the case of @xmath5
, this would mean that one could compute @xmath5 for each value of @xmath25 .
this is clearly false when there are narrow resonances ( e.g. , in the @xmath26 region ) , but is valid elsewhere because the large number of overlapping hadronic resonances provide a natural smearing mechanism .
an important question is the extent to which local duality holds for nonleptonic @xmath1 meson decays .
there is a long - standing claimed discrepancy between theory and experiment in the @xmath1 meson semileptonic branching ratio and in the @xmath27 to @xmath1 lifetime ratio , and the failure of local duality is one way to resolve these . in this paper , we study the violation of local duality using the @xmath3 expansion of qcd @xcite . in this limit ,
we show that the nonleptonic @xmath1 meson decay rate @xmath28 can be expressed in terms of the product of @xmath29 , the invariant mass distribution of the lepton pair in semileptonic @xmath1 decay , and @xmath30 , the hadron mass distribution in nonleptonic @xmath2 decay .
one gets a quantitative measure of the violation of local duality in the large @xmath0 limit by comparing the prediction for @xmath28 using the ope with that using the experimental data for @xmath31 and @xmath30 , and a large @xmath0 factorization formula derived here .
we estimate the accuracy to which the large @xmath0 results should be valid in qcd .
nonleptonic @xmath1 meson decays are dominantly due to the weak hamiltonian decay to the total nonleptonic @xmath1 decay width is small , and we neglect it in this paper .
there are also terms in the weak hamiltonian for @xmath32 decay , which are discussed later . ]
@xmath33\,,\ ] ] where @xmath34 and @xmath35 .
the hamiltonian in eq .
( [ hw ] ) is renormalized at the scale @xmath36 .
the coefficient @xmath37 is the electromagnetic correction .
the coefficients @xmath38 are qcd corrections given by @xmath39 , where @xmath40^{-6/23 } , \qquad c_-(\mu ) = \left[{\alpha_s(\mu ) \over \alpha_s(m_w ) } \right]^{12/23},\ ] ] at one loop order . at next - to - leading order @xcite , @xmath41 and @xmath42 for @xmath43 .
the @xmath44 and @xmath45 terms are effective charged and neutral current interactions , respectively .
nonleptonic @xmath1 decay widths are computed by taking the decay amplitude @xmath46 , squaring , and summing over all possible final states .
a simpler method is to compute the imaginary part of the @xmath47 forward scattering amplitude with an insertion of @xmath48 and @xmath49 .
diagrams involving spectator quarks @xcite , such as pauli interference , weak annihilation , and @xmath50-exchange are of order @xmath51 relative to free quark decay ( since they depend on matrix elements of four - quark operators ) , and can be neglected for the purposes of this paper . in the large @xmath0 limit ,
annihilation effects are enhanced by a factor @xmath0 relative to free quark decay .
while this means that annihilation diagrams are formally dominant at @xmath52 , they can still be neglected for @xmath53 because the @xmath0 enhancement does not compensate for the @xmath54 suppression factor , since the diagram is helicity suppressed and involves four - quark operators . the free quark decay diagram in fig . [ fig:1 ]
is redrawn in fig .
[ fig:2 ] to show the color flows .
the charged and neutral current terms in the weak hamiltonian in eq .
( [ hw ] ) are denoted by the exchange of color singlet charged and neutral `` gauge bosons . ''
this is a convenient way of showing the color flows in the diagrams .
the order in @xmath0 of the various graphs can be computed by using a factor of @xmath0 for each closed color loop , and @xmath55 for each external meson . in the large @xmath0 limit ,
[ fig:2](a ) and ( b ) are @xmath56 , and ( c ) is @xmath57 .
[ fig:2](a ) , ( b ) and ( c ) will be referred to as the charged current , neutral current , and interference diagrams , respectively . in the large @xmath0 limit ,
arbitrary planar gluons must be summed over , which is represented schematically by the shaded regions in the figures .
the leading diagrams have arbitrary planar exchanges within each quark loop , but exchanges between the two quark loops are suppressed by at least two powers of @xmath0 .
thus in the large @xmath0 limit , fig .
[ fig:2](a ) and ( b ) factorize into the product of two non - perturbative matrix elements .
they are : ( i ) the matrix element of two currents in a @xmath1 meson ; ( ii ) the matrix element of two currents in the vacuum .
these will be related to inclusive semileptonic @xmath1 decay and hadronic @xmath2 decay , respectively .
consider the inclusive decay of an initial particle @xmath58 with velocity @xmath59 due to the hamiltonian @xmath60 the decaying particle @xmath58 and the weak currents @xmath61 and @xmath62 will be different for the three cases we need in this paper nonleptonic @xmath1 decay , semileptonic @xmath1 decay , and hadronic @xmath2 decay . in these decays , the weak hamiltonian has factors in addition to those shown explicitly in eq .
( [ hw1 ] ) , such as ckm angles or renormalization group coefficients , which will be included later .
define the tensors @xmath63 and @xmath64 in terms of @xmath50 and @xmath65 , the weak decay rate is given by @xmath66 as long as the diagrams for the process factorize , e.g. , as in fig . [ fig:2](a ) or ( b ) .
this is the factorization formula which we will use in the rest of the paper .
the hadronic tensor @xmath67 is identical to the one that appears in the ope for semileptonic decays @xcite .
it can be expanded as @xmath68 where @xmath69 are functions of @xmath20 and @xmath70 .
@xmath71 can be decomposed into transverse and longitudinal parts @xmath72 where @xmath73 are functions of @xmath20 .
contracting eqs .
( [ wdecompose ] ) and ( [ odecompose ] ) gives @xmath74 \nonumber\\ * & + & \omega_l \left [ -q^2\ , w_1 + ( q\cdot v)^2\ , w_2 + q^4\ , w_4 + 2 q^2 ( q\cdot v)\ , w_5 \right ] .
\label{womeg}\end{aligned}\ ] ] both @xmath67 and @xmath75 are calculable in an operator product expansion .
they are related to the discontinuities across the cut of the amplitudes @xmath76 and @xmath77 defined by @xmath78 i.e. , @xmath79 and @xmath80 .
@xmath76 and @xmath81 have a form factor decomposition similar to @xmath67 and @xmath71 in eqs .
( [ wdecompose ] ) and ( [ odecompose ] ) , with @xmath82 and @xmath83 . in the complex @xmath20 plane , @xmath81 has a cut along the real axis for @xmath84 .
@xmath76 has two cuts along the real axis for @xmath85 corresponding to intermediate states with a charm quark , and for @xmath86 corresponding to intermediate states with a @xmath87 and two @xmath88 quarks .
the weak hamiltonian at @xmath89 for hadronic @xmath2 decay @xmath90 is @xmath91 @xmath92 is the electromagnetic correction .
there are no qcd corrections to eq .
( [ hwtau ] ) since the quark part is a conserved current .
the hadronic @xmath2 decay amplitude clearly factors into hadronic and leptonic parts , irrespective of whether @xmath0 is large or not , since no gluons can be exchanged between the quark and lepton parts of the diagram .
the leptonic part of the diagram is calculable , and gives @xmath93 where @xmath94 .
the hadronic part of the @xmath2 decay diagram is precisely @xmath65 for the charged current nonleptonic @xmath1 decay diagram .
the factorization formula eq .
( [ rate ] ) can be applied to this case , with @xmath95 set equal to @xmath96 .
@xmath65 in @xmath2 decay is defined to be @xmath97 , the sum of @xmath65 s for the weak currents @xmath98 and @xmath99 , so that the ckm angle is included in @xmath100 . integrating over @xmath101 using @xmath102 , gives @xcite @xmath103 , \ ] ] where @xmath20 is the invariant mass - squared of the hadrons in the final state .
inclusive semileptonic @xmath104 decay is due to the weak hamiltonian @xmath105 @xmath106 is the electromagnetic correction .
there are no qcd corrections since the quark part of the hamiltonian is a conserved current . as for hadronic
@xmath2 decay , the semileptonic @xmath1 decay diagram factorizes into a hadronic and leptonic part , irrespective of whether @xmath0 is large or not .
the leptonic part @xmath71 is calculable , @xmath107 where @xmath20 is the invariant mass - squared of the lepton pair .
therefore , @xmath108 gives information on the hadronic part @xmath109 .
using differential @xmath1 decay spectra for @xmath110 , @xmath111 , and @xmath2 , it is possible to measure @xmath112 @xcite .
this seems to be a remote possibility , so in the rest of this paper we neglect lepton masses . in the @xmath113 limit ,
( [ rate ] ) and ( [ wdecompose ] ) give @xmath114 { 1 \over 12 \pi } \,,\ ] ] where the ckm angle @xmath115 has been absorbed into the definition of @xmath116 .
the nonleptonic @xmath1 decay rate can be computed using perturbation theory if one assumes local duality .
to order @xmath117 , the result is @xcite @xmath118 = 3 \gamma_0 \abs{\cnl}^2 & & { } \bigg\ { { c_-^2 + 2c_+^2\over3}\ , \bigg [ \bigg ( 1 + { \lambda_1\over2m_b^2 } \bigg ) + { 3\lambda_2\over m_b^2 } \bigg ( \rho\,{{\rm d}\over{\rm d}\rho } - 2 \bigg ) \bigg ] f(\rho ) \nonumber\\ * & & { } + 4(c_-^2-c_+^2)\ , { \lambda_2\over m_b^2}\ , ( 1-\rho)^3 \bigg\ } + \ldots \ , , \label{bcu } \\ \gnlb [ b \to c \bar c ( d+s ) ] = 3 \gamma_0 \abs{\cnl}^2 & & { } \bigg\ { { c_-^2 + 2c_+^2\over3}\ , \bigg [ \bigg ( 1 + { \lambda_1\over2m_b^2 } \bigg ) + { 3\lambda_2\over m_b^2 } \bigg ( \rho\,{{\rm d}\over{\rm d}\rho } - 2 \bigg ) \bigg ]
g(\rho ) \nonumber\\ * & & { } + 4(c_-^2-c_+^2)\ , { \lambda_2\over m_b^2}\ , h(\rho ) \bigg\ } + \ldots \,,\label{bcc}\end{aligned}\ ] ] where @xmath119 , @xmath120 , and @xmath121 the ellipses in eqs .
( [ bcu ] ) and ( [ bcc ] ) denote corrections of order @xmath122 and @xmath123 .
the total nonleptonic decay rate is the sum of eq .
( [ bcu ] ) and eq .
( [ bcc ] ) .
@xmath12 corrections will be included when we use these results in the next section .
the total nonleptonic decay rate depends on the short distance qcd corrections @xmath44 and @xmath45 .
the validity of local duality depends on long distance effects , such as the masses and widths of hadronic resonances .
one can test local duality in the nonleptonic @xmath1 decay rate by studying a hypothetical world in which @xmath125 and @xmath126 final states are turned off .
this tests local duality only for the charged current part of eq .
( [ hw ] ) . the charged current diagram fig .
[ fig:2](a ) factorizes in the large @xmath0 limit , so eq . ( [ rate ] ) can be used . combining this with eqs .
( [ womeg ] ) , ( [ tauspec ] ) , and ( [ slrate ] ) , one finds @xmath127 @xmath128 has been neglected in this equation because it is proportional to the current quark masses of the @xmath129 , @xmath130 and @xmath25 quarks . in the large @xmath0 limit , one can compute the nonleptonic @xmath1 decay rate in terms of the semileptonic @xmath1 decay rate and the @xmath2 hadronic decay rate .
this is the main result in this paper .
equation ( [ pred ] ) will be used to test local duality in nonleptonic @xmath1 decay . assuming that the ope computation of the total semileptonic decay rate is valid
, one knows that the integral of the experimentally measured spectrum @xmath131 over @xmath20 is equal to the ope value of the total decay rate .
moments of the measured @xmath20 spectrum and of the ope prediction should also agree ; however , the higher order @xmath12 and @xmath132 corrections get larger for higher moments . as a result ,
the integral of the @xmath20 spectra from experiment and from the ope should agree when weighted with a sufficiently smooth and broad function , but not necessarily point by point .
a similar result holds for the @xmath20 spectrum in @xmath2 hadronic decay . in eq .
( [ pred ] ) above , the experimentally measured @xmath20 spectrum in semileptonic @xmath1 decay is multiplied by the experimentally measured @xmath20 spectrum in hadronic @xmath2 decay .
the @xmath2 decay spectrum is not a smooth function ( see fig .
3 ) , which is what makes the test non - trivial .
the test of local duality is to compare @xmath133 in eq .
( [ bcu ] ) calculated using the ope and perturbation theory with @xmath28 in eq .
( [ pred ] ) obtained from experimental data .
before we discuss the result , it is worth examining the errors in the formul we have derived .
the leading corrections to eq .
( [ pred ] ) are factorization violating diagrams involving two - gluon exchange between the two quark loops in fig .
[ fig:2](a ) .
single gluon exchange is forbidden by color conservation .
if both gluons are hard , the diagram can be computed in perturbation theory .
it is order @xmath134 , but is not part of the blm series@xcite @xmath135 , where @xmath136 is the coefficient of the one - loop qcd @xmath137-function .
if both gluons are soft , the correction is of order @xmath138 , and if one gluon is hard and one is soft , the correction is of order @xmath139 .
the leading power suppressed corrections to eq .
( [ pred ] ) are of order @xmath140 or @xmath141 .
the @xmath142 corrections cancel between @xmath28 and @xmath23 .
the first correction in the ope for @xmath77 is order @xmath143 .
both @xmath2 decay and @xmath1 decay involve summing over @xmath144 and @xmath145 final states with weights @xmath146 and @xmath147 , respectively , so both contributions are correctly included in eq .
( [ pred ] ) when one uses the measured @xmath2 hadronic decay rate . in a world with @xmath125 , the largest correction to eq .
( [ pred ] ) is the pauli interference contribution , correction , which has not yet been computed , is large . ] of order @xmath148 .
very little use has been made of the @xmath3 expansion in the theoretical analysis for @xmath125 .
however , the skeptic might argue that the error estimate is based on the ope , which is what is being tested in the first place
. the factorization violation contributions to the decay rate are @xmath149 corrections .
this gives an ope - independent error estimate of @xmath150 .
the hadronic @xmath2 decay spectrum ( i.e. , @xmath100 ) is only measured in the region @xmath151 . in the region @xmath152 , ( with @xmath153 ) , we use the perturbative calculation @xcite , including corrections to @xmath154 up to order @xmath134 with @xmath155 .
higher order terms are negligible for our purposes .
we vary @xmath156 between @xmath157 and check that our conclusions are unaffected by the precise value of @xmath15 .
the region @xmath151 is where the largest violations of local duality are expected in @xmath65 , and experimental data suggests that local duality already holds at some level for @xmath158 . for @xmath95
the largest deviations from local duality are expected near @xmath159 . for the hadron invariant mass spectrum in @xmath2 decay we use a monte carlo distribution that has been tuned to be consistent with cleo data .
this can be used to determine @xmath160 . in fig .
[ figtau ] we have plotted @xmath161 together with the perturbation theory prediction for this ratio ( including corrections up to order @xmath134 ) . the lepton invariant mass spectrum in semileptonic @xmath1 decay has not yet been measured , so we use the cleo monte carlo . the normalized mass spectrum is plotted in fig . [ figbsl ] as a function of @xmath20 together with the prediction of perturbation theory .
the shape of this curve is rather insensitive to the order @xmath12 and @xmath162 corrections@xcite .
it is simplest to normalize the nonleptonic rate to the semileptonic rate .
this eliminates the order @xmath117 corrections and reduces the perturbative corrections as well .
the result for the @xmath1 meson decay rate using the factorization formula eq .
( [ pred ] ) and the cleo monte carlo spectra is @xmath163 / \gslb \bigg|_{\rm cleo } = 3.32 \abs{c_1}^2.\ ] ] the result for the ope calculation at this order is @xmath164/{\rm d}q^2 \over { \rm d } \gslb /{\rm d}q^2 } = 3\abs{c_1\ , \cnl \over \csl}^2 \left [ 1 + { \alpha_s(q^2)\over\pi } + 0.173\beta_0 \left({\alpha_s(q^2)\over\pi}\right)^2 + \ldots \right ] \,.\ ] ] expressing @xmath165 in terms of @xmath166 and neglecting nonperturbative and higher order perturbative corrections , this implies @xmath167 / \gslb = 3 \abs{c_1 \cnl/\csl}^2 \left ( 1 + 0.07 + 0.025 \right ) = 3.21 \abs{c_1}^2 $ ] .
the difference is surprisingly small , at the 3% level , which is less than the 5% theoretical uncertainty of the calculation .
so we conclude that the uncertainties in the charged current contribution to @xmath88 hadron lifetimes due to local duality violation are unlikely to exceed the @xmath168 percent level .
of course , it is very important to repeat this analysis using real data for the lepton mass spectrum in semileptonic @xmath1 decay , @xmath169 , especially since the monte carlo predictions used above ( solid curve in fig .
[ figbsl ] ) differ significantly from perturbation theory in the low @xmath20 region .
in the real world with @xmath170 , there are several effects that restrict our calculational ability . the interference term fig .
[ fig:2](c ) , is a @xmath3 correction and does not factorize .
its contribution is order @xmath171 of the @xmath172 decay rate .
it has been argued that deviations from local duality are likely to be larger for this contribution@xcite .
the neutral current decay fig . [ fig:2](b ) also factorizes .
@xmath95 for this decay can be determined from semileptonic @xmath1 decay due to the ( hypothetical ) neutral current @xmath173 .
this can be related to semileptonic @xmath1 decay due to the weak current @xmath174 by isospin invariance .
@xmath75 for the neutral current decay is the vacuum polarization contribution due to the @xmath175 current , for which there is no experimental information .
the neutral current contribution to the decay width is order @xmath176 .
we have neglected @xmath177 decays .
these can also be factorized in the large @xmath0 limit .
the vacuum polarization @xmath65 is due to @xmath178 and @xmath179 currents , for which there is no experimental information .
thus our calculational method , while valid in principle , is not useful for @xmath177 .
experimentally , one can separate the @xmath180 and @xmath177 contributions to the nonleptonic decay rate by counting the number of charm quarks in the final state , and the results of the previous section can be used for the @xmath181 part .
we have estimated the violation of local duality for nonleptonic @xmath1 decay with @xmath125 to be less than the theoretical uncertainty in our calculation of @xmath182 .
it is important to redo the analysis using experimental data , rather than monte carlo , once it is available .
one can also view the results in a different way : the factorization formula allows one to compute the nonleptonic @xmath1 decay rate using the lepton mass spectrum in semileptonic @xmath1 decay and the hadron mass spectrum in @xmath2 decay without using an ope . in the world
@xmath125 , one can predict this rate with an accuracy of 5% . in the real world
, one can compute the @xmath183 part of the decay rate with approximately 20% accuracy .
the results can not easily be extended to @xmath184 meson decays , because the @xmath185 corrections in the charm sector are @xmath186 of the charm quark decay result .
these reasonably tight limits on local duality violation do not resolve the question of the @xmath27 to @xmath1 lifetime ratio .
of course , duality violation in @xmath187 and in the @xmath188 interference contributions are probably larger . while enhancing the average number of charm in @xmath1 decay seems experimentally disfavored , the @xmath188 interference remains suspect .
it is also possible that spectator effects are anomalously large in @xmath27 decay in fact , there are six - quark operators ( see fig .
5 ) which contribute roughly of order @xmath189 relative to the @xmath88 quark decay .
these have not been analyzed , and may significantly affect the spectator contribution to @xmath27 decay . they do not occur for @xmath1 meson decay , and so contribute to the @xmath4 lifetime difference . in semileptonic @xmath1 decay , there is no @xmath190 correction @xcite .
this is also true in nonleptonic @xmath1 decays , if one uses the ope @xcite . in other processes which do not have an ope
, it is known that one can get arbitrary power suppressed corrections @xcite .
it has been suggested that there might be @xmath190 corrections in inclusive nonleptonic decay rates @xcite . in the large @xmath0 limit with @xmath125 , eq .
( [ pred ] ) shows that the @xmath190 corrections to the nonleptonic decay rate can be computed in terms of the corrections to the lepton mass spectrum in semileptonic @xmath1 decay and the hadron mass spectrum in @xmath2 decay .
at least one of these would have to have a @xmath190 correction if there is one in the nonleptonic decay rate .
we thank e. potter and a. weinstein for providing us with monte carlo simulations , m.e .
luke for some numerical results , and m. gremm , v. sharma , i. stewart , and m.b .
wise for discussions .
this work was supported in part by the department of energy under grant doe - fg03 - 97er40546 , and by the national science foundation under grant phy-9457911 . | in the large @xmath0 limit , the nonleptonic width of the @xmath1 meson is determined by the lepton mass spectrum in semileptonic @xmath1 decay and the hadron mass spectrum in @xmath2 decay .
this result can be compared with the nonleptonic width computed using the operator product expansion and local duality , to estimate the violation of local duality in @xmath1 decay . in the absence of the required data , we use experimental monte carlo predictions and
find that the difference is small , at the few percent level .
we estimate the theoretical uncertainty in our results from higher order corrections ( including @xmath3 effects ) .
we also identify a new contribution to the @xmath4 lifetime difference . #
1#2#3#1 | # 2 | # 3 # 1o ( # 1 ) # 1| # 1 | |
in the standard model ( sm ) and its supersymmetric extensions , the masses of electroweak gauge bosons , leptons , and quarks are generated by interactions with higgs fields @xcite .
the yukawa couplings between higgs particles and fermions therefore grow with the masses @xmath2 of the fermions .
the couplings obey a universal scaling law @xmath3 in the sm , with @xmath4 being the ground - state value of the higgs field . in supersymmetric models , which involve at least two higgs doublets , the size of the yukawa couplings
is also set by the fermion masses , yet the relationship is more complex owing to the mixing among the higgs fields .
the yukawa couplings @xcite of the two cp - even light / heavy higgs bosons @xmath5 and of the cp - odd higgs boson @xmath6 in the minimal supersymmetric extension of the standard model ( mssm ) @xcite , expressed in units of the sm couplings , are determined by the parameters @xmath7 , the ratio of the vacuum expectation values of the two higgs fields generating the masses of up- and down - type particles , and @xmath8 , the mixing angle in the cp - even sector . in the decoupling limit , in which the light higgs mass reaches the maximum value for a given parameter @xmath9 , the @xmath10 yukawa couplings approach the sm values . in general , they are suppressed for up - type fermions and enhanced for down - type fermions ; the enhancement increases with @xmath9 and can therefore be very strong .
higgs radiation off top or off bottom quarks in @xmath1 collisions , @xmath11 , \label{eq : processes}\ ] ] lends itself as a suitable process for measuring the yukawa couplings of all higgs bosons @xcite , particularly for the sm higgs bosons , the light higgs boson @xmath10 and for moderately heavy higgs bosons @xmath12 and @xmath6 . within the sm
it has been demonstrated that the top yukawa coupling can be measured with an accuracy of 23% at linear colliders with @xmath13 gev @xcite .
we present the cross sections for these processes including the next - to - leading order ( nlo ) qcd corrections @xcite . in this analysis
we have calculated the complete set of @xmath14 qcd corrections to all subchannels in ( [ eq : processes ] ) systematically .
the large number of interfering subchannels in supersymmetric theories renders this program more complex than the corresponding calculation in the sm , in particular since the relative weight of the subchannels varies over the supersymmetric parameter space and over the phase space for different mass ratios .
moreover , the higgs - energy distributions haven been obtained at nlo @xcite .
they turn out to be relevant in the separation of the resonant parts and those , which depend on the yukawa couplings . introducing appropriate cuts in the higgs energy the sensitivity to the yukawa couplings
can be increased experimentally .
the qcd corrections can be categorized into five classes .
virtual corrections of the internal quark lines , of the @xmath15quark vertices , of the higgs quark vertices , and box diagrams interfere with the born amplitude .
gluon radiation off internal and external quark lines adds incoherently to the cross sections .
the value of the electromagnetic coupling is taken to be @xmath16 and the weinberg angle as @xmath17 .
the mass of the @xmath18 boson is set to @xmath19 , and the pole masses of the top and bottom quarks are set to @xmath20 @xcite and mass @xmath21 . ]
@xmath22 @xcite , respectively .
the masses of the mssm higgs bosons and their couplings are related to @xmath9 and the pseudoscalar higgs boson mass @xmath23 . in the relation
we use , higher - order corrections up to two loops in the effective - potential approach are included @xcite .
the susy parameters are chosen as @xmath24 and @xmath25 tev ; this simple choice is sufficient to illustrate the main results .
the production of @xmath26 final states can be mediated by resonance channels @xmath27 and @xmath28 .
we describe the resonance structures as breit
wigner forms by substituting @xmath29 in all boson propagators .
the decay widths of the higgs bosons are calculated including higher - order corrections , as described in refs .
@xcite , while the @xmath18 width is set to @xmath30 . for @xmath31 production
the widths can be neglected , since the higgs masses are taken below the @xmath32 threshold .
the renormalization scale of the qcd coupling @xmath33 , which is evaluated in nlo with five active flavors normalized to @xmath34 @xcite , is chosen at @xmath35 , where @xmath36 is the center - of - mass ( cm ) energy squared .
the qcd radiative corrections have been calculated in the standard way .
the feynman diagrams have been evaluated within dimensional regularization .
ultraviolet divergences are consistently regularized in @xmath37 dimensions , with @xmath38 treated naively since no anomalies are involved .
the renormalization of the @xmath39 vertices is connected to the renormalization of the quark masses , which , in the case of the top quark , is defined on shell ( see e.g. refs .
@xcite for details ) . in the case of bottom
quarks large logarithms are mapped into the running mass @xmath40 for the yukawa couplings of the @xmath41 quark , with @xmath42 denoting the squared momentum flow through the corresponding higgs - boson line .
the infrared divergences encountered in the virtual corrections and in the cross section for real gluon emission , are treated in two different ways .
both calculations follow subtraction procedures , one using dimensional regularization and one introducing an infinitesimal gluon mass @xcite .
the results obtained by the two different procedures are in mutual numerical agreement after adding the contributions from virtual gluon exchange and real gluon emission .
a second , completely independent calculation of the qcd corrections to the total cross section was based on the evaluation of all relevant cut diagrams of the photon and @xmath18-boson self - energies in two - loop order , generalizing the method applied to @xmath43 intermediate states in ref .
the results of the two approaches are in numerical agreement .
moreover , the parts that have been calculated in refs .
@xcite , i.e. photon exchange and resonant contributions , are in full agreement with our corresponding partial results .
* a. ) asymptotic behaviour . * the qcd corrections to the top final states @xmath44 can be interpreted easily in two kinematical areas . whenever the invariant mass of the @xmath45 pair is close to threshold , the gluonic sommerfeld rescattering - correction is positive and becomes large . in the threshold region
the @xmath46 factor approaches the asymptotic form @xcite @xmath47 with the maximal quark velocity @xmath48 in the @xmath49 rest frame . for high energies , on the other hand , the qcd corrections are of order @xmath50 .
in the energy region @xmath51 but @xmath52 , which is relevant for the present analysis , the qcd corrections can qualitatively be traced back to vertex corrections and infrared gluon radiation . since scalar higgs bosons
are radiated off top quarks preferentially with small energy [ @xmath53 , as is evident from the leading ( universal ) part of the fragmentation function @xmath54 \,\ , , \label{eq : fragh}\ ] ] the qcd correction of the scalar yukawa vertex , regularized by soft gluon radiation , approaches the value @xcite @xmath55 \to -4\frac{\alpha_s}{\pi } \,\ , .\ ] ] the scalar yukawa vertex is therefore reduced by four units in @xmath50 which are compensated only partly by one unit due to the increase of the @xmath32 production probability , leading in total @xcite to @xmath56 the ellipsis accounts for hard higgs and gluon radiation ( of order @xmath57 ) .
thus , the qcd corrections are expected negative for scalar higgs particles in the high energy continuum . by contrast , the corresponding fragmentation function for the pseudoscalar higgs boson @xcite @xmath58 is hard so that the average of the vertex and ir gluon corrections over the higgs spectrum amounts to @xmath59 \right\rangle \sim -\frac{3}{2 } \frac{\alpha_s}{\pi } \,\ , .\ ] ] adding to this correction the increase of the @xmath32 production probability of one unit , the @xmath46 factor is very close to unity @xmath60 after hard gluon bremsstrahlung is taken into account ( symbolized by the ellipsis ) , the overall qcd corrections for the pseudoscalar higgs boson are therefore expected slightly positive .
[ for ultra - high energies , i.e. @xmath61 , hard gluon bremsstrahlung becomes important .
similarly to the leading terms in the fragmentation functions eqs .
( [ eq : fragh ] ) and ( [ eq : fraga ] ) , the qcd corrections for scalar and pseudoscalar higgs bosons approach each other as a result of chiral symmetry restoration in _ asymptotia _ ; this has been verified in a numerical calculation . ]
similar estimates can be applied to bottom final states which in general are dominated by resonance decays . after absorbing the large logarithms @xmath62 into the yukawa couplings ,
the non - leading effects are positive : @xmath63 also close to the thresholds and in the high - energy limit the qcd corrections remain positive after mapping the large ( negative ) corrections into the running yukawa couplings . since different channels are activated at the same time ,
only a qualitative estimate can be given in the continuum regime , @xmath64 while details must be left to the numerical analysis .
+ * b. ) numerical results .
* the total cross sections for @xmath65 production in the sm are presented in fig .
[ fg : smtot ] as a function of the higgs mass for different collider energies @xcite .
the nlo ( lo ) cross sections are given by the full ( dotted ) curves .
the qcd corrections are large for @xmath66 gev due to the coulomb singularity at threshold and moderate for high energies . at high energies
the results agree with the approximation of eq .
( [ eq : kas ] ) . the cross sections amount to more than about 0.1 fb , which leads to a significant number of events at the tesla collider , being designed to reach integrated luminosities of about @xmath67 ab@xmath68 in three years of operation . the corresponding higgs - energy distribution is presented in fig .
[ fg : smxh ] for @xmath69 gev as a function of the scaled variable @xmath70 @xcite .
the shape of the distribution is shifted towards larger values of @xmath71 due to the coulomb singularity at threshold .
the results for small values of @xmath71 reproduce the approximation of eq .
( [ eq : kas ] ) .
-90 the results for the mssm are exemplified in fig .
[ fig : cs ] for @xmath72 is already excluded for vanishing mixing in the stop sector , this choice exemplifies the complete result for smaller values of @xmath9 and can easily be extended to the non - vanishing stop mixing case .
] and 30 and for the collider energy @xmath73 @xcite .
if required by the size of the cross section , which should not fall below @xmath74 fb in order to be accessible experimentally , we switched to @xmath75 .
the born terms are shown by the dotted curves , while the final results for the cross sections , including qcd corrections , are given by the full curves . -90 -90 -90 -90 for @xmath73 the qcd corrections to higgs - boson production in association with @xmath32 pairs increase the scalar higgs - production cross sections significantly , as can be inferred from fig . [ fig : cs ] .
close to threshold the numerical results clearly exhibit the strong increase of the cross sections due to the coulomb singularity ( [ eq : coulsing ] ) .
moreover , for @xmath76 the cross sections are strongly suppressed except for the regions where the light ( heavy ) scalar higgs mass is close to its upper ( lower ) bound . for @xmath72 the cross section amounts to about 1 fb , which leads to a significant number of events at the tesla collider . for cm energies of 1 tev the qcd corrections to scalar higgs production in association with @xmath32 pairs are of moderate size .
they decrease the cross sections by about 35% .
this is in accordance with the asymptotic form of the @xmath46 factor @xcite . for pseudoscalar higgs production ,
the size of the qcd corrections is slightly positive at 1 tev , in agreement with the qualitative discussion above .
the cross sections for higgs - boson production associated with @xmath77 pairs are significantly larger due to the resonance contributions from on - shell @xmath18 and higgs - boson decays into @xmath77 pairs .
the qcd corrections increase these cross sections by about 525% .
the drop in the @xmath78 production cross section for @xmath72 at @xmath79 and @xmath80 can be attributed to the crossing of the thresholds for resonant @xmath81 and @xmath82 decays , respectively , in @xmath83 final states . without cuts ,
the intermediate resonance decays @xmath84 dominate all @xmath85 production processes , whenever they are kinematically allowed , and it will be difficult to extract the bottom yukawa couplings experimentally in regions where resonant higgs decays to @xmath77 pairs are dominant .
this is the case at large values of @xmath9 for all neutral higgs particles and at small values of @xmath9 for higgs masses below the @xmath86 threshold . in these cases the branching ratios , which determine the size of the @xmath85 cross sections
, will be nearly independent of the bottom yukawa couplings .
resonance decays @xmath87 in the @xmath88 final states can however be eliminated by cutting out the resonance energy of the final - state higgs boson , @xmath89 . after subtracting these resonance parts ,
the non - resonant contributions are suppressed by about one to three orders of magnitude .
the resonances have been removed from the cross sections in the examples of fig .
[ fg : cont ] by subtracting the two - particle cross sections in the breit wigner bands @xmath90 of the energy @xmath91 with the resolution @xmath92 .
this theoretical definition is used for the sake of simplicity ; wider cuts may be required in experimental analyses .
peaks and dips in the cross sections are the result of overlapping breit - wigner bands . for the scalar higgs particles they arise from overlapping @xmath18 and @xmath6 boson bands ; in pseudoscalar higgs production the light and heavy scalar resonance bands overlap for 100 gev@xmath93 $ \sim$}}~}m_a { \raisebox{-0.13cm}{~\shortstack{$<$ \\[-0.07 cm ] $ \sim$}}~}120 $ ] gev for @xmath76 .
the dips occur whenever the two resonance bands touch each other , while the peaks between the dips occur when the resonance masses coincide exactly . as shown in fig .
[ fg : cont ] , the qcd - corrected cross sections are still close to @xmath94 or slightly below , except for heavy masses at small values of @xmath9 .
thus , ensembles of order @xmath95 events can be collected at a high - luminosity collider @xcite .
-90 -90 fig .
[ fg : ttaxa ] presents the pseudoscalar higgs - energy distribution in @xmath96 production at @xmath97 tev for @xmath98 gev and @xmath72 as a function of the scaled higgs energy @xmath99 @xcite . as in the case of the sm higgs
boson the distribution is shifted towards larger higgs energies due to the coulomb singularity at threshold .
moreover , it is clearly visible from the comparison of figs .
[ fg : smxh ] and [ fg : ttaxa ] that the pseudoscalar energy spectrum is harder than the scalar one , which is related to the absence of the @xmath100 enhancement in eq .
( [ eq : fraga ] ) .
-90 the corresponding higgs - energy distributions of @xmath101 production for @xmath102 gev is exemplified in fig .
[ fg : bbhxh ] and for @xmath78 production for @xmath98 gev in fig .
[ fg : bbaxa ] as functions of the corresponding scaled higgs energies for @xmath76 at @xmath66 gev @xcite .
the resonance peaks of @xmath103 are dominating the distribution of @xmath101 production and of @xmath104 the distribution of @xmath78 production .
the continuum contributions are suppressed by several orders of magnitude .
the qcd corrections are moderate for all higgs energies . -90
top- and bottom - yukawa couplings in the sm and supersymmetric theories can be measured at future linear @xmath1 colliders in higgs radiation off top and bottom quarks .
we have presented the total cross sections and higgs - energy distributions including the full nlo qcd corrections .
the corrections turn out to be large for higgs radiation off top quarks at c.m .
energies @xmath66 gev , while they are moderate at larger energies .
the qcd corrections to higgs radiation off bottom quarks are moderate after absorbing large logarithms in the running bottom yukawa couplings .
measurements of the top yukawa coupling in sm higgs radiation off top quarks can be performed quite accurately at the future tesla collider thanks to the high luminosities .
measurements of yukawa couplings in supersymmetric higgs radiation off heavy quarks at @xmath1 linear colliders are difficult .
this is a result of the large number of subchannels contributing to the @xmath105 final states in supersymmetric theories in general , and the contamination by two - higgs final states in particular .
nevertheless , the continuum cross sections appear large enough to allow for a solution of this experimental problem as shown in the present analysis . even though experimental simulations are beyond the scope of this note , it may be concluded from earlier sm analyses that the method will work at least in part of the mssm parameter space , thus providing us with the absolute size of the quark
higgs yukawa couplings in the minimal supersymmetric theory .
+ * acknowledgements .
* i would like to thank s. dittmaier , m. krmer , y. liao and p.m. zerwas for the pleasant collaboration on the topic presented here .
moreover , i am grateful to the organizers of radcor2000 for the stimulating atmosphere during the symposium and financial support .
a. juste and g. merino , proceedings , sitges conference , barcelona au report [ hep - ph/9910301 ] ; h. baer , s. dawson and l. reina , phys .
rev . * d61 * ( 2000 ) 013002 ; s. moretti , phys
. lett . *
b452 * ( 1999 ) 338 ; tesla collaboration , technical design report , to appear . | yukawa couplings between higgs bosons and quarks in the standard model ( sm ) and supersymmetric theories can be measured in the processes @xmath0 . the cross sections and higgs energy distributions of these processes in the sm and minimal supersymmetric model
have been determined including the complete set of next - to - leading order qcd corrections for all channels .
hep - ph/0101314 higgs radiation off quarks + in the standard model and supersymmetric theories + at @xmath1 colliders + 1.cm michael spira 0.8 cm _ paul scherrer - institut , ch5232 villigen psi , switzerland _ presented at the + 5th international symposium on radiative corrections + ( radcor2000 ) + carmel ca , usa ,
1115 september , 2000 |
the densities in the interstellar space ( consisting of 99% gas and 1% dust ) , even in the densest interstellar clouds , are so low that they are beyond the capabilities of the best vacuum systems on earth .
even then , a large number of infrared absorption features are observed toward low and high mass protostars . some of these features are attributed to the solid state interstellar molecules @xcite .
discovery of these interstellar molecules in condensed phase demands a proper explanation for the interstellar chemical composition budget . increasing observational and experimental evidences in the gas phase points to the inadequacy of the gas phase chemistry and
suggests the need to incorporate the grain surface processes .
interstellar dust grains are thought to be consisting of an amorphous silicate or carbonaceous core surrounded by a molecular ice layer @xcite .
presently , very little is known about the morphology or chemistry on these grain surfaces , or the porosity of the grains themselves .
similarly , the chemical and physical processes on the grain surfaces till date are also very poorly known .
a number of theoretical attempts have been carried out to verify the observed abundances of the gas phase and the condensed phase species @xcite .
both the deterministic and stochastic approaches are tested by these studies .
observational and experimental evidences suggest that almost 90% of the grain mantle is covered with h@xmath1o , ch@xmath0oh and co@xmath1 . in our earlier paper @xcite , ( hereafter dac10 ) , we mentioned that it is very difficult to theoretically match the exact observed abundances of all the three molecules simultaneously .
results of the numerical simulations show that only in a narrow region of parameter space all these three molecules are produced within the observed limit .
it is well known that the ism chemistry is highly influenced by the interstellar uv radiations which mainly come from the nearby young stars .
uv radiation affects the chemistry of the ism through photo - dissociation and photo - ionization of the molecules and atoms .
several works are carried out to study the photo - chemistry in the ice containing simple as well as complex molecules .
most of the studies @xcite are concentrated following the irradiation of ice mixtures to reflect the chemical composition of the star forming regions .
these types of studies give us an insight about the photo effects on the interstellar ice mantles .
structure of the interstellar ices under various physical circumstances are yet to fully understood .
water ice is less abundant along some lines of sight . in the present paper ,
we carry out a coupled gas - grain study to fully understand the structure of the interstellar grain mantle around various lines of sight .
gas - grain interactions are considered via accretion from gas phase and evaporation from grain surface .
important surface reactions , which are feasible under the interstellar conditions are considered in our surface chemistry network to have clear insight of the interstellar ice feature . moreover , to mimic the realistic situation , the formation of several mono - layers are considered . as we are interested only around the dense region of the molecular cloud , we assume that the major portion of the gas phase carbon atoms are already converted into carbon monoxide and as in the other models @xcite , we also consider the accretion of h , o and co from the gas phase to follow the time evolution of several chemical species .
recent experimental findings of several new reaction pathways and several interaction energies involved during these process , motivate us to update our earlier dac10 model . our main aim in the present context is to study the formation of ice mantle under the influence of photo - processing and to make a comparative study among different routes to the water formation .
this has been suggested recently by some experiments @xcite .
we also compare our simulation results with other observational , experimental and theoretical results carried out till date . in section 2 , we discuss the physical processes and the method used for the modeling .
extensive discussions along with the results are presented in section 3 . in section 4 , a comparison between the theoretical and observational results are presented .
finally , in section 5 we have summarized our work .
in general , the rate equation method @xcite is widely used to follow the surface process occurring under the interstellar condition .
main advantage of this method is that the computation is very fast .
this method assumes the average concentration of species to follow the chemical evolution .
main drawback with this method is that it does not consider the statistical fluctuations and discrete nature of the species . as a result
, this method is very much dependent on the physical conditions of the interstellar cloud . in the lower accretion rate regime ,
where surface population is comparatively lower , this method tends to overestimate the result @xcite .
moreover as this method deals with the average concentration , it is not able to follow the internal structure of the grain mantle .
however , as the time passes by , grains are expected to grow in size with several surface species . as the rate equation is unable to trace any particular species
, it is not possible for this method to mimic the actual situation in ism .
this is the reason , we use the descrete - time random walk monte carlo method to have more accurate results . in our approach ,
fluctuations in the surface abundance due to the statistical nature of the grain is preserved and as a result this method is much more realistic to study the interstellar surface chemistry . with this method , we can follow movements of each species and evaluate the results at every instant .
details of the processes we follow are already discussed in @xcite . for the sake of completeness ,
we discuss them again briefly .
we assume that the grains are square shaped and similar to an atom on an fcc[100 ] plane , each grain site has four nearest neighbors .
we assume the periodic boundary conditions and also allow the possibility to form multiple layers on a grain surface .
for modeling purposes , we consider classical dust grains , characterized by 1000a@xmath2 and 10@xmath3 surface sites .
following @xcite , we consider the surface density of available sites to be 1.5 @xmath4 @xmath5 .
number density of the grains are taken to be 1.33@xmath6 ( @xmath7 is the concentration of h nuclei in all forms ) and gas to dust ratio is taken to be @xmath8 by mass@xcite .
monte carlo method is computationally much slower than the rate equation method and it is quite difficult to simulate for a grain having @xmath9 sites . to save computational time
, we consider the grain having @xmath10 sites and the final results are extrapolated for the grain having @xmath9 sites .
reason behind choosing the grain having @xmath11 sites is that it is above the limit of the statistical fluctuating size and according to @xcite , above this size , results are scalable and do not differ significantly .
location of the accreting species is dictated by generating a pair of random numbers .
langmuir hinshelwood ( reactions between the surface species via hopping ) and eley rideal ( reaction between the incoming species with the surface species ) , both types of reactions are considered . after landing on a grain , its direction towards any one of the four nearest neighbors
is dictated by generating random numbers .
two possibilities may occur : if the site is occupied then either it will react when the reaction is permitted or it will wait until its next hopping time . if the neighboring site is vacant and it does not have any species just beneath that site then it can roll down on that grid until it touch some existing species in a layer or just the bare grain surface . if the reaction between the species touching the existing species is permitted , a new species is formed and if not , then it can sit on the top of that . for the reactions having barriers
, we are generating random numbers upon each encounter and checking for the possibility of the reaction .
following @xcite probability ( p ) for the reaction to happen upon an encounter is calculated and if the generated random number is less than p then the reaction is allowed and if the generated number is greater than p then reaction is discarded at that instant .
every species has its own hopping time scale . in our simulation
, we consider the hopping time scale of h atom to be the minimum .
after each interval of h hopping time , we are checking for the evaporation ( via thermal or other process ) and photo - dissociation by generating random numbers .
for example , let us consider h@xmath1o .
its evaporation rate is r@xmath12 sec@xmath13 .
that means , one h@xmath1o can evaporate after each @xmath14 seconds of its formation on the grain . since the smallest time scale is the hopping time scale @xmath15 of an h atom , after every @xmath15 , we generate a random number to find out whether the species will evaporate or stick to the grain .
if the generated number is smaller than @xmath16 , then the species can evaporate .
photo - dissociations are also treated similarly , i.e. , if the generated random number is under the probability of photo - dissociation of any species , that species can photo - dissociate .
gas phase compositions are adjusted at every @xmath15 after taking care of the accretion onto and the evaporation from the grains .
a major difference between our previous study and the present study is that we have updated our reaction network by considering several new reactions and have considered the effects of the cosmic rays as well .
cosmic rays have the direct / induced effect on the interstellar grain mantle . in our model , we consider the photo - dissociation and photo evaporation to incorporate the effects of the cosmic rays .
formation of the most abundant molecules such as , h@xmath1o , ch@xmath0oh and co@xmath1 are considered as before . keeping in mind the recent experimental results @xcite , we have also included the formation of h@xmath1o via o@xmath0 and h@xmath1o@xmath1 routes .
all the reactions considered in our present model are given in table 1 .
activation barrier energy used in our previous paper @xcite , and in some other papers @xcite are noted in tab.1 .
we gave some comments on the most recent experimental / theoretical findings in the last column .
for the reaction numbers 4 and 6 , we normally use @xmath17k as the activation barrier energy @xcite .
the most recent findings by @xcite , suggest that the energy barrier could be much lower ( @xmath18k for reaction number 4 and @xmath19k for the reaction number 6 at @xmath20k ) . in our model calculation
, we assume that at @xmath21k , the energy barrier is @xmath22k for the reaction number 4 , and @xmath23k for the reaction number 6 .
so far , barrier energy @xmath24k was used as mentioned in @xcite .
recent experimental findings by @xcite suggest that the reaction number 12 is barrier less .
recent surface chemical modelling by @xcite assumed that the reaction number 14 is also barrier less though it is not experimentally verified . in our model , we assume reaction number 12 is barrier less and reaction number 14 with barrier 1400k @xcite . reaction number 16 is considered to have the activation energy @xmath25k @xcite .
binding energies of the species are the keys for the chemical enrichment of the interstellar grain mantles .
grain surface provides the space for the interstellar gas phase species to land on and to react with the other surface species .
if allowed , chemical species can return back to the gas phase after chemical reactions or in the original accreted form .
reactions between the surface species is highly dependent on the mobility of the surface species , which in turn , depends on the thermal hopping time scale or on the tunneling time scale .
for the lighter species , tunneling time scale is much shorter than the hopping time scale .
so for the lighter species , surface migration by tunneling method might be effective .
when two surface species meet , they can react .
if some activation energy involved for that reaction to happen , they can react with a quantum mechanical tunneling probability ( p ) .
@xcite studied the tunnelling reaction between o and co with direct dynamics employing density functional theory and they obtained the activation energy 2500k for the reaction between o and co. we use only thermal hopping of o atom for its mobility on the grain surface and use the same activation barrier energy , 2500k following @xcite for the reaction number 9 of tab .
1 . in @xcite , a comparative study between tunneling and hopping for the mobility of h atom on the grain surface was carried out and it was noticed that results did not show drastic variation within the accretion regime considered here ( number density 10@xmath26 - 10@xmath3 ) . we expect that as long as the accretion rate is within the limit , this does not significantly affect the final results .
there are two types of binding energies associated with the physical processes at low temperature , namely , physisorption and chemisorption . among them , physisorption energies are more probable in low temperature ( @xmath27k ) condition , i.e. , frigid ( low temperature , @xmath28 10k ) condition .
therefore , in our model calculations , we considered the physisorption energies .
we defined the whole system with a lot of interaction energies .
most of the energies considered here are discussed in detail in @xcite and references therein .
the rest of the species which are updated in our present model are given in table 2 .
it is very difficult to find the exact binding energies for every specific surface . in @xcite ,
olivine surface was considered for the modeling purpose .
presently , we also considered olivine surface as a model substrate . among the updated chemical species , binding energies of ho@xmath1 , o@xmath0 and h@xmath1o@xmath1 are taken from @xcite .
they used the physisorbed energies of these molecules for the carbonaceous grain and here we have included their values as if those are the interaction energies with the bare olivin surfaces . for c atom
, we used the same binding energy as in co. for ch@xmath0 and ch@xmath1oh , we used the same binding energies as in ch@xmath0oh and h@xmath1co respectively .
since the binding energies of these species are very high , we can be sure that our assumptions would not alter the final results significantly .
due to the lack of knowledge of the appropriate interaction energies between various species , we assumed that the binding energies of these species with other surface species are the same as with olivine surfaces .
.surface reactions considered in our network [ cols="<,<,<,<",options="header " , ] + @xmath29 @xcite +
in this paper , we carried out a monte carlo simulation to study the formation and structural information of the interstellar grain mantle .
the main results of our study are as follows : @xmath30 the effects of interstellar radiations are included to extract the information about the structure of the interstellar grain mantle .
the effects of photo - dissociation are studied as a function of the extinction parameter . for lower values of the extinction parameter ,
the photo - dissociation effects dominate . a comparison between the rate equation method and the monte carlo method has been made and difference in results between the two approaches is presented in fig.1 .
we notice that in comparison with the monte carlo method , the rate equation method always overestimates the abundances of species .
stable species like h@xmath1o , ch@xmath0oh and co@xmath1 are photo - dissociating more strongly in the rate equation method as compared to the monte carlo method .
@xmath30 recent chemical models assumed that the interstellar photons can have effects on the first few mono - layers only .
but as the column density of the grain mantle itself is very low , we assume that the effect may be significant also deep inside the mantle .
we notice that the choices of number of exposed layers can affect the chemical composition of the interstellar grain mantle significantly .
@xmath30 column densities are calculated along different lines of sight and it is noticed that deep inside the cloud , the probability of finding different grain surface species are very high .
effects of interstellar photons are extensively studied and it is noticed that beyond @xmath31=10 , its contribution toward the chemical evolution is negligible .
@xmath30 column densities of various grain surface species are studied for static as well as time dependent models .
a static model is considered which is more realistic than the dac10 model , where the number density of the cloud is varied according to the extinction parameter of the cloud . as the initial conditions are completely different from those of the dac10 model , our results are also significantly different .
in the time dependent model , variation of the number density and extinction parameter with time is considered to follow the chemical evolution during the collapsing phase of a proto - star .
it is noticed that since initially the density was low , formation of several species was seriously hindered by the effects of the interstellar photon . as the time passes by , the density and a@xmath32 increases , which in turn enhance the production of several grain surface species .
@xmath30 water is the most abundant species on the grain .
there are several routes by which water can be formed on the interstellar grain .
major parts of h@xmath1o are formed via traditional route(h + oh ) .
but there are also other pathways responsible for the h@xmath1o production on the grain . from our simulation , we find that the formation of h@xmath1o via o@xmath0 and h@xmath1o@xmath1 routes are also have significant contributions , especially in high density regions .
@xmath30 a comparison between the results of our approaches and observational data has been made along different lines of sight .
the initial gas phase composition was chosen from the dac10 model ( dac10a and dac10b ) as well as the modified dac10 model ( i.e. , by considering the variations of density with the extinction parameter , dac10a(m ) and dac10b(m ) ) .
some of our results are found to be in good agreement with different observational results obtained so far .
this work was partly supported by a dst project ( grant no . sr / s2/hep-40/2008 ) .
allamandola , l. j. , sandford s. a. & tielens a.g.g.m .
1992 . , apj , 399 , 134 allen , m. , robinson , g. w. , 1975 . , apj , 195 , 81 allamandola , l. j. , sandford , s. a. , valero , g. j. 1988 , icarus , 76 , 225 andersson , s. , al - halabi , a. , kroes , g. , van dishoeck , e. f. , 2006 , jcp , 124 , 064715 boogert , a.c.a & ehrenfreund , p. , 2004 , aspc 309 , 547 cazaux , s.,cobut , v. , marseille , m. , , spaans , m. , caselli , p. , 2010 ,
a&a , 522 , a74 chang , q. , cuppen , h. , m. , herbst , e. , 2005 , a&a , 434 , 599 charnley , s.b . , 2001 ,
apj , 562l , 99 chakrabarti , s.k . ,
das , a. , acharyya , k. , chakrabarti , s. , 2006 , a&a , 457 , 167 chakrabarti , s.k . , das , a. , acharyya , k. , chakrabarti , s. , 2006 , basi , 34 , 299 cuppen , h. m. , herbst , e. , 2007 , apj , 668 , 294 cuppen , h. m. , van dishoeck e. , f. , herbst , e. , tielens , a. g. g. m. , 2009 , a&a , 508 , 275 dartois , e. , schutte , w. , geballe , t. r. , demyk , k. , ehrenfreund , p. , dhendecourt , l. 1999 a&a , 342l , 32 das , a. , acharyya , k. , chakrabarti , s. , chakrabarti , s. k. , 2008 , a & a , 486 , 209 das , a. , acharyya , k. , chakrabarti , s. k. , 2010 , mnras 409 , 789 das , a. , acharyya , k. , chakrabarti , s. , chakrabarti , s. k. , 2008 , new astronomy , 13 , 457 dhendecourt , l. , et al . , 1996 ,
a&a , 315 , l365 dhendecourt , l. b. , allamandola , l. j. , baas , f. , greenberg , j. m. 1982 , a&a , 109 , l12 draine , b. t. , 2003 , apj 598 , 1017 .
fuchs , g. w. , cuppen , h. m. , ioppolo , s. , romanzin , c. , bisschop , s. e. , andersson , s. , van dishoeck , e. f. , linnartz , h. , 2009 , a&a , 505 , 629 goumans , t. , p. , m. , andersson , s. , 2010 , mnras , 406 , 2213 .
gibb , e. l. , whittet , d. c. b. , boogert , a. c. a. , tielens , a. g. g. m. , 2004 , apjs 151 , 35 hagen , w. , allamandola , l. j. , greenberg , j. m. 1979 , ap&ss , 65 , 215 hollenbach , d. , salpeter , e. e. , 1970 , j. chem .
phys . , 53 , 79 hasegawa , t. , herbst , e. , leung , c.m . , 1992 , apj , 82 , 167 ioppolo , s. , cuppen , h. m. , romanzin , c. , van dishoeck , e. f. , linnartz , h. , 2008 , apj , 686 , 1474 jiang , b. w. , szczerba , r. , & deguchi , s. , 2000 , a&a , 362 , 273 keane , j. v. , boogert , a. c. a. , tielens , a. g. g. m. , ehrenfreund , p. , schutte , w. a. , 2001 , a&a , 375l , 43 _ klemm , r. b. , payne , w. a. , stief , l. j. 1975 , in chemical kinetic data for the upper and lower atmosphere , ed . s. w. benson ( new york : wiley),61 _ lee , h. , h. , herbst , e. , pineau des forets , g. , roueff , e. , le bourlot , j. , 1996 , a&a , 311 , 690 leger , a. , jura , m. , omont , a. , 1985 , a&a , 144 , 147 leith - devlin , m. , a. , williams , d. , a. , 213 , 295 , mnras , 1985 melius , c. f. , & blint , r. j. 1979 , chem .
64 , 183 murakawa , k. , tamura , m. , nagata , t. , 2000 , apjss , 128 , 603 nguyen , t. k. , ruffle , d. p. , herbst , e. , williams , d. a. 2002 , mnras , 329,301 onishi , t. , mizuno , a. , kawamura , a. , ogawa , h. , & fukui , y. 1996 , apj , 465 , 815 roberts , h. , & herbst , e. , 2002 , a&a , 395 , 233 schiff , h. , i. , 1973 , in physics and chemistry of upper atmospheres , ed . b. m. mccormac ( dordrecht : reidel ) , 85 shalabiea , o. m. , greenberg , j. m. , 1994 , a&a , 290 , 266 stantcheva , t. , shematovich , v. i. , herbst , e. , 2002 , a&a , 391 , 1069 talbi , d. , chandler , g. s. , rohl , a. l. , 2006 , chem .
, 320 , 214 tielens , a. g. g. m. , hagen , w. , 1982 , a&a , 114 , 245 vandenbussche , b. , ehrenfreund , p. , boogert , a. , c. , a. , van dishoeck , e. , f. , schutte , w. , a. , gerakines , p , a. , chiar , j. , tielens , a. , g. , g. , m. , keane , j. , whittet , d. , c. , b. , and 2 coauthors , 1999 , a&a , 346l , 57 westley , m. s. , baragiola , r. a. , johnston , r. e. , baratta , g. , a. , 1995 , nat , 373 , 405 watson , w. d. , salpeter , e. e. , 1972 , apj , 174 , 321 whittet , d. c. b. , et al . , 1996 , a&a , 315 , l357 woodall , j.,agundez , m. , markwick kemper , a. j. , millar , t. j. , 2007 , a&a , 466 , 1197 | we study the chemical evolution of interstellar grain mantle by varying the physical parameters of the interstellar medium ( ism ) . to mimic the exact interstellar condition , gas grain interactions via accretion from the gas phase and desorption ( thermal evaporation and photo - evaporation ) from the grain surface
are considered .
we find that the chemical composition of the interstellar grain mantle is highly dependent on the physical parameters associated with a molecular cloud .
interstellar photons are seen to play an important role towards the growth and the structure of the interstellar grain mantle .
we consider the effects of interstellar photons ( photo - dissociation and photo - evaporation ) in our simulation under various interstellar conditions .
we notice that the effects of interstellar photons dominate around the region of lower visual extinction .
these photons contribute significantly in the formation of the grain mantle .
energy of the incoming photon is attenuated by the absorption and scattering by the interstellar dust .
top most layers are mainly assumed to be affected by the incoming radiation .
we have studied the effects of photo - dissociation by varying the number of layers which could be affected by it .
model calculations are carried out for the static ( extinction parameter is changing with the density of the cloud ) as well as the time dependent case ( i.e. , extinction parameter and number density of the cloud both are changing with time ) and the results are discussed in details .
different routes to the formation of water molecules are studied and it is noticed that around the dense region , production of water molecules via o@xmath0 and h@xmath1o@xmath1 contributes significantly . at the end , various observational evidences for the condensed phase species are summarized with their physical conditions and are compared with our simulation results .
astrochemistry , ism : abundances , molecules , evolution , stars : formation , methods : numerical |
to study the inner structures of multiquark states is an interesting subject in hadron physics , because it is obviously beyond the simple @xmath1 or @xmath2 textile which was established by gell - mann more than half century ago .
the molecular structures of the multiquark systems have been studied by many authors , in comparison the tetraquark states have much less been investigated so far . generally , the tetraquark is suggested to be composed of a diquark and an antidiquark which reside in color antitriplet and triplet , respectively @xcite .
because of the different color structures , the dynamics for molecular states and tetraquark would be very different .
in fact , in molecular states , the two constituents are bound by exchanging color - singlet mesons or baryons which can be described by the chiral lagrangian . instead ,
the two constituents in tetraquarks are bound by direct gluon exchange . as for the quarkonium ,
the interaction between quark and antiquark is realized by exchanging gluons , a single - gluon exchange composes the leading contribution which results in a coulomb - type effective potential , however , for the hadron formation , the energy scale is below @xmath3 , therefore the nonperturbative qcd effects would play an important role which induces a confinement piece in the potential . similarly , for the tetraquark case , the diquark and antidiquark interact by exchanging gluons and definitely the single gluon exchange exists and induce a coulomb - like potential whereas the nonperturbative qcd effects should also be introduced .
it is generally believed that such interaction may be described by the color - flux model @xcite . following this scenario
, we will study the dynamics which not only results in the different structures of the two configurations , but also determines their different decay patterns . because the newly observed @xmath0 is very likely to be a tetraquark
, it would be an ideal place for carrying research on the teraquark and molecular states via their decay behaviors .
@xmath0 has been observed in the invariant mass spectra of the @xmath4 channel @xcite , and it is identified as a @xmath5 resonance with mass @xmath6 mev and width @xmath7 mev .
there are many alternative interpretations for the observed peak @xcite , for example , the authors of ref .
@xcite consider that a strong attraction between @xmath8 and @xmath9 bind them together , so that @xmath0 may be interpreted as a baryon - antibaryon molecule . instead , in refs .
@xcite @xmath0 is interpreted as a @xmath10 charmonium state .
also , the @xmath0 is considered to be induced by a threshold effect instead of being a genuine resonance @xcite . among those proposals ,
the suggestion that @xmath0 is a tetraquark state is more favorable @xcite . in ref .
@xcite , @xmath0 is identified as the ground state with its orbital angular momentum @xmath11 . by reanalyzing the @xmath12 and @xmath13 spectra , cotugno _
_ suggested that @xmath0 and @xmath14 @xcite could be the same tetraquark state , and is the first radial excitation of the @xmath15 with @xmath11 @xcite . in ref .
@xcite , the authors studied the open - charm decay @xmath16 by assuming that @xmath0 is a radially excited state of the diquark - antidiquark bound state with hidden charm . by another theoretical physics group @xcite @xmath0
is interpreted as a molecular state made of @xmath17 and @xmath18 .
as is well known , in general the multiquark states may be in tetraquark states which are composed of colored constituents , or in molecular states which are composed of two color singlets , or their mixtures .
therefore , the mechanisms which bind the components in a unique system and induce the multiquark states to decay would be different in those cases . in this work , using the quantum mechanics we analyze the dynamical mechanisms inducing decays of the tetraquarks and the molecular states , where @xmath0 stands as an example for the study .
thus , a numerical analysis on the decay width of channel @xmath19 based on the tetraquark postulate is made and a qualitative discussion about the possible decay process @xmath20 where @xmath0 may be a mixture of tetraquark and molecular state is presented .
the paper is organized as follows : after this introduction , we study the decay of @xmath0 with the tetraquark and molecular interpretations in secs [ sec : tetra ] and [ sec : mole ] respectively , then sec [ sec : conclusion ] is devoted to our discussion and conclusion .
inspired by the fact that the @xmath0 decays into charmed baryon pair , one is tempted to conjecture this resonance as a tetraquark which is made of the diquark - antidiquark @xmath21[\bar c\bar q]$ ] , where @xmath22 is a light quark either @xmath23 or @xmath24 , @xmath21 $ ] resides in a color antitriplet whereas @xmath25 $ ] is in a color triplet . here
we take the diquark - antidiquark picture proposed by brodsky _
_ @xcite that a diquark and an antidiquark are bound together by a gluon - flux - tube into a color singlet tetraquark where the constituents are separated by a substantial distance once they are created .
the interaction for the system can be well described by a generalized cornell potential @xcite since the constituents ( diquark and antidiquark ) are treated as two pointlike color sources in analog to the configuration for an ordinary @xmath26 quarkonium . in terms of the quark pair creation ( qpc ) model where a quark - antiquark pair is excited out from vacuum , we calculated the decay width of @xmath27 @xcite . in that picture , the qcd vacuum is excited and a quark - antiquark pair is created . the quark - antiquark pair would tear " apart the diquark and antidiquark due to strong qcd interaction between quark and diquark ( antiquark and antidiquark ) . then joining the diquark ,
the created quark becomes a constituent of the charmed baryon @xmath8 , and as well for the @xmath9 . in parallel ,
let us consider an alternative way to discuss the production of @xmath12 in the framework of quantum mechanics .
in fact , the color flux - tube results in a potential barrier to forbid the inner constituents ( either the diquark or the antidiquark ) to escape from the bound state . in terms of a modified flux - tube - induced potential ,
the decay of @xmath0 may occur as the diquark - antidiquark bound system falling apart via tunneling through the effective potential barrier .
after tunneling out the barrier , the diquark ( antidiquark ) would immediately attract a quark ( antiquark ) from the vacuum to compose a color singlet @xmath8 ( @xmath9 ) and the hadronization process is somehow similar to the picture frequently used to study the multiparticle production at high energy collision . then the transition probability can be calculated in terms of the wkb ( wentzel - kramers - brillouin ) approximation .
first , we employ a nonrelativistic potential model with a cornell - like potential where some free parameters are obtained by fitting the mass spectrum of heavy quarkonia and generalized to the case for tetraquark , then we get the wave function of @xmath0 by solving the schrdinger equation .
the general hamiltonian of a diquark - antidiquark system ( i.e. a quarkonium - like system ) can be written as @xmath28 where the @xmath29 and @xmath30 are the masses(3-momenta ) of the diquark @xmath31 and antidiquark @xmath32 respectively , we take @xmath33 in this work .
the interaction potential is @xmath34 where @xmath35 is the distance between the diquark and the antidiquark .
the one - gluon - exchange ( oge ) term @xmath36 , which plays the main role at short distances , is @xcite @xmath37 and the confinement part @xmath38 takes the linear form@xcite @xmath39 where @xmath40 and @xmath41 are , respectively , the spin operators and the color factor specific to @xmath42-@xmath43 attraction .
@xmath44 is the string tension and @xmath45 is a global zero - point energy .
@xmath46 is the phenomenological strong coupling constant .
the @xmath47-function in eq.([eq : voge ] ) is replaced by a gaussian smearing function @xcite with a fitted parameter @xmath48 @xmath49
the spin wave functions for the @xmath50 state with @xmath11 is taken as @xmath51 in the basis of @xmath52 @xcite . here
we take the values from ref .
@xcite : @xmath53 gev@xmath54 gev . by adopting the suggestion given by authors of refs .
@xcite that @xmath0 is a radial excitation state of p - wave ( l=1 ) , the ( anti)diquark mass is determined to be @xmath55 gev which is close to the @xmath56 meson mass .
this value is also consistent with that computed by using qcd sum rules in ref .
@xcite where it is @xmath57 gev .
the fitted spectra are presented in fig .
[ fig : spectra ] , and the charmonium spectra calculated by authors of ref .
@xcite are also shown in the figure for a clear comparison . in this framework
, we fit the ground state to be 4235 mev , and such a state is consistent with the observed @xmath58 resonance @xcite and/or @xmath59 resonance @xcite which is also considered as a tetraquark @xcite . the radial wave function of @xmath0 is plotted in fig .
[ fig : wave ] . of @xmath0 with quantum numbers @xmath60 .
, width=245 ] the interaction between the constituents in the tetraquark can not simply be derived from the quantum field theory yet and phenomenologically the dynamics of the nonperturbative qcd effects which determines the confinement behavior may be described by the flux - tube model . moreover , as is well known , when the tension of the flux - tube goes beyond a certain bound , namely , the distance between the diquark and antidiquark gets long enough , the flux - tube will break into two strings and at the new ends a quark - antiquark pair is created @xcite .
one can use a step function to describe the breaking effect as @xmath61\times v(r),\ ] ] where @xmath62 is a parameter corresponding to the strengthening limit of the string at where the probability of the string fragmentation reaches maximum .
a typical scale for nonperturbative qcd is @xmath63 , therefore it is natural to consider @xmath62 should be of order of @xmath64 . just as smearing the delta function , we need also to smear the step function .
in fact @xmath65 so smearing the step function implies that we keep @xmath66 as a nonzero free parameter to be determined .
in fact , if we do not consider breaking of the flux tube , the effective potential is the same as @xmath67 given in eq .
( [ eq : v ] ) .
therefore , taking into account of the breaking " effect does not affect the computation on the @xmath0 spectrum . here
the interaction between the diquark and the antidiquark at a relatively large distance is described by a modified potential as @xcite . @xmath68 .
the dotted line stands for the eigenvalue @xmath69 , and @xmath70 are the turning points at the potential barrier . in this figure , we set @xmath66 and @xmath62 to be 0.03 and 1.8 fm respectively for an illustration.,width=7 ] in this scenario
, we translate the flux - tube induced confinement into the potential barrier , and breaking the tube corresponds to tunneling through the barrier .
the diquark - antidiquark bound system falls apart by tunneling through this effective potential barrier , then is hadronized into a @xmath71 pair . the process is graphically shown in fig .
[ fig : tunnel ] . by means of the wkb approximation
, the transition probability of the tunneling process can be calculated . under this assignment ,
the transition probability is given by @xmath72,\ ] ] where @xmath73 is the reduced mass of the @xmath21$]-@xmath74 $ ] system , @xmath75 and @xmath44 are the turning points , as shown in fig .
[ fig : tunnel ] .
one can obtain the effective velocity @xmath76 of the motion of a particle with the reduced mass inside the system from the average kinetic energy @xmath77 where @xmath78 is the wave function obtained in terms of the pure cornell potential without the flux tube breaking correction .
since the breaking effect only affects the long distance behavior , it does not change the wave function @xmath79 and the decay width @xmath80 is deduced .
dependence of the partial decay width of @xmath19 on @xmath66 , where @xmath0 is supposed to be in @xmath81 state , the belle data are shown in the plot for a comparison .
the dashed black line and the green band correspond to the central value and error for the total width of @xmath0 measured by the belle collaboration ( @xmath7 mev @xcite ) .
the purple , blue and cyan curves correspond to the three different @xmath62 assignments respectively.,width=7 ] assuming the @xmath0 as the first radial excited state , we compute the partial decay width of the channel @xmath82 . in fig .
[ results ] , dependence of the calculated partial width @xmath83 on the free parameter @xmath66 is plotted .
the purple , blue and cyan curves correspond to the cases of @xmath84 fm , respectively . as @xmath66 increases , the predicted decay width increases as shown in fig .
[ h1 ] , where each curve in the figure corresponds to a special @xmath62 value , and it is noted that each curve stops at some value of @xmath66 , because at that @xmath66 value , the diquark and antidiquark are no longer bound by the modified potential .
it sets a constraint on @xmath66 .
the dashed black line and the green band are respectively the central value and the error of the total width of @xmath0 measured by the belle collaboration ( @xmath7 mev ) @xcite .
varying @xmath62 which represents the flux - tube breaking effect does not change the general picture .
it is noticed that the calculated partial decay width changes with @xmath62 , but for any specific value of @xmath62 , there exists a range of @xmath66 which allows the predicted width to be consistent with the experimental data , namely , falls within the error tolerance region set by present measurements , which is rather wide .
indeed , even though the calculated decay width of @xmath19 is smaller than the central value of the measured total width , one still can not conclude that @xmath0 is not a pure tetraquark yet .
our numerical results indicate that the tetraquark picture is able to predict the correct decay width of @xmath0 , even though not completely confident , we believe that its decay mechanism could be considered as the diquark - antidiquark system falling apart via tunneling through an effective potential barrier , and then diquark and antidiquark are respectively hadronized into color singlet hadrons , and the @xmath12 pair should be the main product .
this conclusion is consistent with that of ref .
@xcite , in which a similar result is obtained within the qpc framework .
it is noteworthy that if the future measurement indeed confirms a rather large total width which is larger than our prediction based on the pure tetraquark structure , a possible mixture between tetraquark and molecular state should be taken into account and other decay modes such as @xmath20 may occur with non - negligible fraction .
as is mentioned in the introduction and the above discussion , the hadronic molecular picture @xmath85 is another possible choice for the inner structure of @xmath0 which was proposed by guo _
et al . _
it is noted that the mass and width of @xmath0 are consistent with those for the @xmath14 state ( @xmath86 mev , @xmath87 mev @xcite ) within error tolerance . by taking into account the @xmath88 final state interaction ,
it is found that the @xmath0 may be the same state as @xmath89 , and the resonance can be a @xmath85 molecular bound state . and @xmath18 of @xmath0 residing in the attractive potential well which is supposed to be a hadronic molecular state .
the effective potential is induced by the @xmath48 meson exchange , see ref .
@xcite for details.,width=283 ] now let us study the mechanism which may induce the decay of the molecular state . because the two constituents in the molecular state are color singlets
, they do not interact by directly exchanging gluons , but only via exchanging color - singlet mesons , and the leading contribution is coming from the @xmath90 exchange between @xmath17 and @xmath18 which does not induce a potential barrier as for the tetraquark case .
instead , the interaction provides a potential well and the constituents are confined in the well , as shown in fig .
[ fig : well ] . thus we should have @xmath91 where @xmath92 is the binding energy of the molecular state and roughly is @xmath93 mev . in the traditional framework of quantum mechanics
it is a stable structure , i.e. @xmath0 can not dissolve into on - shell @xmath17 and @xmath18 , however , due to the quantum fluctuation , @xmath18 can jump out the potential well to become an off - shell virtual particle . if the virtual @xmath18 does not transit into two real pions , it would fall back to its original state inside the molecule .
the duration of it being virtual particle can be estimated by the uncertainty principle as @xmath94 in the natural unit system , thus the virtuality time @xmath95 is proportional to @xmath96 where @xmath92 is the binding energy .
obviously the decay amplitude should be proportional to @xmath95 , namely the larger the binding energy @xmath97 is , the shorter @xmath98 is , and then the smaller the decay probability would be . by this principle , we can write out an effective lagrangian which induces the decay of the molecule @xmath99 where the superscript @xmath100 denotes that @xmath18 is an off - shell virtual meson which later transits into two pions .
the effective vertex at @xmath101 is @xmath102 where @xmath103 and @xmath104 correspond to @xmath105 , @xmath18 and @xmath106 is a dimensionless universal coupling constant . by the equation of motion
it is easy to be reduced into @xmath107 which reads as @xmath108 the effective coupling given by authors of ref .
@xcite is @xmath109 where @xmath110 is the reduced mass of the @xmath111 and @xmath18 , @xmath112 has an energy dimension .
thus we can identify the relation @xmath113 with this assumption another decay mode of @xmath0 could be @xmath114 for the off - shell @xmath18 mainly decaying into @xmath115 , as shown in fig .
[ otherdecay ] . with the given lagrangian ,
the decay width was calculated in ref .
@xcite as @xmath116 mev , here we do not repeat it and advise readers to refer to that paper .
the hadronic decay is closely associated with the nonperturbative qcd , and a lot of phenomenological models are proposed to account for its effects .
for example , the qpc model , flux - tube model , qcd sum rules and lattice qcd , etc . have been successfully used to estimate decay rates , even though , with the exception of the lattice calculation , none of them can be directly derived from quantum field theory so far . for the @xmath26 systems ,
the physics picture is clear , even though a phenomenological model must be embedded to reflect the nonperturbative qcd effects and the computation schemes are mature . however , for the four - quark states , the inner structure and dynamics which leads the binding and decay of the state are still not well understood and there are various proposals for them . in this work we study the decay mechanisms of @xmath0 in both tetraquark and molecule pictures in the framework of quantum mechanics .
namely , we use the wkb approximation to calculate the decay width of @xmath0 as it is assumed to be a tetraquark state , and then qualitatively discuss its decay mechanism as it is postulated as a molecular state where @xmath18 jumps out the potential well due to a quantum fluctuation and becomes a virtual particle and later transits into two real pions .
definitely , all of the assignments to the observed resonance at 4630 mev should be tested in the future by more precise measurements . in our other works
@xcite , we study the case that if @xmath0 is a tetraquark , its favorable decay mode should be @xmath82 which would overwhelmingly dominate its width , but due to the inelastic rescattering processes between @xmath8 and @xmath9 , some other final states , such as @xmath117 , @xmath118 , @xmath119 and @xmath120 might be produced with measurable rates , whereas , if @xmath0 is a molecular state , its dominant decay mode would be @xmath121 and due to the decay of @xmath17 and final state interaction , the pattern of the decay products which will be experimentally measured would be completely different from the tetraquark case . thus the measurements would provide more information about the assignment of @xmath0 .
we are lying hope on the future experiments which will be carried out at the belleii , besiii and even lhcb in the coming years .
g. pakhlova _ et al .
_ [ belle collaboration ] , `` observation of a near - threshold enhancement in the @xmath124 cross section using initial - state radiation , '' phys .
* 101 * , 172001 ( 2008 ) [ arxiv:0807.4458 [ hep - ex ] ] .
h. x. chen , w. chen , x. liu and s. l. zhu , `` the hidden - charm pentaquark and tetraquark states , '' arxiv:1601.02092 [ hep - ph ] .
a. esposito , a. l. guerrieri , f. piccinini , a. pilloni and a. d. polosa , `` four - quark hadrons : an updated review , '' int .
j. mod .
a * 30 * , 1530002 ( 2015 ) [ arxiv:1411.5997 [ hep - ph ] ] .
n. lee , z. g. luo , x. l. chen and s. l. zhu , `` possible deuteron - like molecular states composed of heavy baryons , '' phys .
d * 84 * , 014031 ( 2011 ) [ arxiv:1104.4257 [ hep - ph ] ] .
a. m. badalian , b. l. g. bakker and i. v. danilkin , `` the s - d mixing and di - electron widths of higher charmonium @xmath125 states , '' phys . atom .
nucl . *
72 * , 638 ( 2009 ) [ arxiv:0805.2291 [ hep - ph ] ] . j. segovia , d. r. entem and f. fernandez , `` charm spectroscopy beyond the constituent quark model , '' arxiv:0810.2875 [ hep - ph ] .
x. l. wang _ et al .
_ [ belle collaboration ] , `` observation of two resonant structures in @xmath130 via initial state radiation at belle , '' phys .
lett . * 99 * , 142002 ( 2007 ) [ arxiv:0707.3699 [ hep - ex ] ] .
j. d. weinstein and n. isgur , `` the @xmath133 system in a potential model , '' phys .
d * 27 * , 588 ( 1983 ) .
r. t. kleiv , t. g. steele , a. zhang and i. blokland , `` heavy - light diquark masses from qcd sum rules and constituent diquark models of tetraquarks , '' phys . rev .
d * 87 * , 125018 ( 2013 ) [ arxiv:1304.7816 [ hep - ph ] ] .
m. ablikim _ et al . _
[ besiii collaboration ] , `` study of @xmath134 at center - of - mass energies from 4.21 to 4.42 gev , '' phys .
lett . * 114 * , 092003 ( 2015 ) [ arxiv:1410.6538 [ hep - ex ] ] . c. z. yuan , `` evidence for resonant structures in @xmath135 , '' chin .
c * 38 * , 043001 ( 2014 ) [ arxiv:1312.6399 [ hep - ex ] ] . c. z. yuan , `` recent progress on the study of the charmoniumlike states , '' int .
j. mod .
a * 29 * , 1430046 ( 2014 ) [ arxiv:1404.7768 [ hep - ex ] ] .
x. w. liu , h. w. ke , y. b. ding and x. q. li , `` study of the structures of four - quark states in terms of the born - oppenheimer approximation , '' chin .
c * 39 * , 083103 ( 2015 ) [ arxiv:1409.5939 [ hep - ph ] ] .
x. liu , z. g. luo , y. r. liu and s. l. zhu , `` x(3872 ) and other possible heavy molecular states , '' eur .
j. c * 61 * , 411 ( 2009 ) [ arxiv:0808.0073 [ hep - ph ] ] .
x. d. guo , d. y. chen , h. w. ke , x. liu and x. q. li , `` study on the rare decays of @xmath0 induced by final state interactions , '' phys .
d * 93 * , 054009 ( 2016 ) [ arxiv:1602.02222 [ hep - ph ] ] . | the inner structures of the multiquark states are an interesting subject in hadron physics , generally they may be in tetraquark states which are composed of colored constituents , or in molecular states which are composed of two color singlets , or their mixtures . therefore , the mechanisms which bind the constituents in a unique system and induce the multiquark states to decay would be different in those cases . in this work , using the quantum mechanics we analyze the dynamical mechanisms inducing decays of the tetraquarks where @xmath0 stands as an example for the study , we also comment on the molecular states without making numerical computations . |
vladimir kulinskii thanks nwo ( nederlandse organisatie voor wetenschappelijk onderzoek ) for a grant , which enabled him to visit dick bedeaux s group at the leiden university . | we consider the dynamics of systems of self propelling particles with nonholonomic constraints . a continuum model for a discrete algorithm used in works by t. vicsek et al .
is proposed . for a case of planar geometry
the finite flocking behavior is obtained .
the circulation of the velocity field is found not to be conserved as a consequence of the nonholomicity .
the stability of ordered motion with respect to noise is discussed .
an analogy with the kinetics of charges in superconductors is noted . the emergence of ordered structures in dynamic systems is a long standing problem in physics .
generally speaking one deals with dynamic phase transitions governed either by external or internal noise .
most interesting is the arising of ordered motion caused by the internal dynamics of the system .
recently , there has been a growing interest in studying the collective behavior in systems of self - propelling particles ( spp ) .
one may distinguish systems of two types : the first one is for systems of unintelligent particles interacting via real physical forces produced by the background in which they are moving ( micelles , bacteria , etc ) . here
the driving forces are caused by the gradients of chemical factors ( concentrations , chem .
potential etc . ) or physical factors ( light , potential and dissipative fields etc . ) , which influence the motion . in particular ,
the earth magnetic field is vital for the orientation during long distance migration of biological species like birds or turtles @xcite .
it is clear that the absence of conservation of translational and angular momentum is a direct consequence of these external factors @xcite .
the second class is formed by systems of particles , which interact via nonholonomic constraints imposed on their velocities .
we would like to stress that the nonholonomicity of such systems clearly expresses the intelligent nature of the particles , since such constraints need instant exchange of information ( visual or any other sensorial ) between the particles and their environment .
this explains the coherent motion and arising ordered patterns in the dynamics of systems like crowds , traffic or flocking .
usual potential gradients or other physical forces are not relevant for the collective behavior , though the particles need some physical source of energy to sustain the constraints .
the physical origin of the nonholonomicity is the force which acts on the particle due to its interaction with the background ( earth , air , liquid substrate etc . ) . since we are not interested in the dynamic degrees of freedom of the background we lose this information .
such loss of dynamic information leads to the breaking of the conservation of the ( angular ) momentum and effectively to the nonholonomic constraints . in general
it means that the system with such constraints is not closed and therefore its dynamics is not hamiltonian , though it does not mean that the energy dissipates .
rather a redistribution among the dynamic degrees of freedom takes place .
but the very form of the constraint is determined by the intellect of a particle which uses the information about the environment and moves accordingly .
note that the numerical algorithm used in ref .
@xcite modelled that kind of systems . for shortness
we call it the czir@xmath0k - vicsek automaton or algorithm ( cva )
. there are also modifications of this algorithm , which differ from the cva by inclusion of potential interparticle forces @xcite , external regular and stochastic fields @xcite .
we will consider cva as the minimal model for the collective behavior since the main cause for the self - organization in the system is the nonholonomicity of cva dynamic rule . to our knowledge none of the hydrodynamic models proposed earlier
is aimed to reflect the essence of the cva - its conservative and distinctly nonholonimic character . to a great extent
such models are modifications of the navier - stokes equation .
such an approach is certainly valid for microorganisms floating in a medium .
it is , however , hardly adequate for the collective behavior of `` intelligent '' boids , e.g. birds or drivers in traffic flow @xcite , when the individual behavior is determined rather by the instant exchange of information with the environment than by the action of some interparticle forces .
the key point here is the nonholonomicity of the system .
in addition all the models based on the modifications of the navier - stokes equation like ref .
@xcite include additional phenomenological terms which generate the spontaneous transition to the state of ordered motion .
such terms are added much in formal analogy with landau s theory of equilibrium continuous phase transitions and the notion of an order parameter but any firm base for such an analogy is not given .
thus the ordered state introduced in such a model is rather an _
ad hoc _ assumption than the natural consequence of the underlying interparticle interactions .
the interpretation of the viscous term for such systems is also completely unclear . in ref .
@xcite it was noted that the cva can be considered as the dynamic @xmath1-model .
it was concluded also that the cause of the ordering is the convective term .
the @xmath1-model is a hamiltonian dynamic system .
the latter property is the key point since for hamiltonian dynamic flow there is an ergodic measure .
due to nonholonomic character the dynamic rule of the cva even in the static limit @xmath2 , i.e. when there is no any transfer flow term like @xmath3 , has no canonical hamiltonian form .
the general fact is that such systems do not have an invariant measure with respect to the dynamic flow .
in fact the ordered state in the cva appears as the fixed point ( attractor ) of its averaging dynamic rule .
note that from the point of view of the theory of dynamic systems the compactness of the phase space ( i.e. the space of positions and velocities ) is very important so the question about influence of the boundary conditions usually used in simulations on the ordering the system also needs discussion .
from such a nonholonomic point of view the appearance of ordered motion in the cva and similar dynamic systems is a trivial consequence of their nonholonomicity .
indeed , the breaking of conservation of the ( angular ) momentum is due to nonholonomic constraints which as has been mentioned above mean that the system is not closed .
here we discuss a hydrodynamic model which can be considered as the continuous analogue of the discrete dynamic automaton proposed in ref .
@xcite for the spp system .
it manifestly takes into account all local conservation laws for the number of particles and the kinetic energy . the self propelling force and the frictional force
are assumed to balance each other .
the algorithm used in ref .
@xcite corresponds to the following equation of motion of a particle : @xmath4 where @xmath5 is the angular velocity of @xmath6-th boid , which depends on what happens in the neighborhood .
it is assumed also that the number of particles is conserved .
like algorithm in ref .
@xcite , eq . distinctly expresses the conservation of the kinetic energy . from a physical point of view
it is natural that the hydrodynamic model corresponding to the cva is based on the following equations : @xmath7 where @xmath8 and @xmath9 are the number density and the eulerian velocity .
the volume @xmath10 moves along with the velocity field .
the first condition is the conservation of number of particles . as usual
we can rewrite this condition in the differential form : @xmath11 the second constraint eq .
means that the kinetic energy of a lagrange particle is conserved , i.e. @xmath12 the first integrand vanishes due to the conservation of the particle number . as a consequence
the second integral also vanishes for an arbitrary choice of the volume @xmath10 which in view of the natural condition @xmath13 leads to : @xmath14 this implies that a pseudovector field @xmath15 exists such that @xmath16 this equation , which is the continuous analogue of eq . , has now been derived from the conservation of particle number and kinetic energy , eq .
and eq .
. we will model @xmath15 by : @xmath17 which has the proper pseudovector character and heuristically may be considered as the continual analog of the cva dynamic rule .
there are other possible choices like : @xmath18 and combinations of the two .
the averaging kernels @xmath19 and @xmath20 should naturally decrease with the distance in realistic models .
we concentrate our discussion on the case eq . .
equations and obviously have the uniform ordered motion with a constant both density and velocity as trivial solution .
one may consider eq . as the equation of motion of a charge in a magnetic field , where @xmath21 is the charge velocity and @xmath15 is proportional to the magnetic field .
one may even include an `` electric field `` and dissipative ( collision ) terms in eq .
: @xmath22 where @xmath23 is the mean free time .
we will not consider such an extended model here since algorithm used in @xcite is distinctly conservative .
further we exploit the analogy of eq . and the equation of motion for charges in superconductors ( see e.g. ref .
@xcite ) . in our model
a current density corresponding to the particle velocity at some point @xmath24 depends on ' ' magnetic field ''
@xmath15 , i.e. vector potential
, at all neighboring points @xmath25 within some region of coherence
. therefore the relations or can be considered as a corresponding nonlocal relations between the current density and the vector potential in nonlocal pippard s electrodynamics of the superconductors .
the situation in the system under consideration is more complex . in electrodynamics of superconductors
the external magnetic field is the main cause of the vortical motion of charges since their own magnetic field is negligibly small and does not lead to formation of the ordered motion . in our case
the magnetic field
@xmath15 itself depends on the motion of charges ,
i.e. particles , and vise versa , which leads to a nonlinearity of the system . depending on the parameters one can expect either the direct current state since the system is conservative or vortical states like meissner currents or abrikosov vortices . using such an analogy
let us find the conditions for the existence of the stationary vortical states .
we can rewrite eq . in the following form : @xmath26 where @xmath27
thus it follows that if @xmath28 is equal to @xmath29 , then @xmath30 and therefore @xmath31 is independent of the time .
such states are naturally interpreted as stationary translational @xmath32 or rotational @xmath33 regimes of motion . for the model
together with @xmath34 we get the integral equation : @xmath35 which determines such states .
equation gives stationary vortical motion , represented by the vector field with @xmath36 .
from here it follows that the vorticity of the velocity field is an eigenstate of the integral operator with @xmath37 as the corresponding weight factor .
it should be noted that these stationary states do not exhaust all stationary vortical states , since in general @xmath38 .
we further scale @xmath19 by multiplying with some @xmath39 and similarly scale the density by dividing by @xmath39 .
furthermore we restrict our discussion to the simple case of a planar geometry with averaging kernel in as @xmath40-functional : @xmath41 we will call this the local model . for this case eqs . and take the form : @xmath42 these equation can be obtained as a special case of the corresponding one in ref . @xcite . in our work
the foundation of the terms is given .
we do not use general symmetry arguments to take all terms of a certain symmetry into account irrespective of their physical meaning .
the corresponding local model with @xmath40-kernel for may be identified with _ rotor chemotaxis _
( i.e. caused by chemical field ) force introduced in ref .
@xcite if one takes into account simple linear relation between field of food concentration and the number density of boids , which is surely valid for low concentrations of food and bacteria .
the parameter @xmath43 of the local model distinguishes different physical situations concerning the microscopic nonholonomical constraint . to see this we find the stationary radially symmetric solutions of and . as usual , we search for the solutions of the form @xmath44 , @xmath45 .
the continuity equation is satisfied trivially . substituting this into eq .
we finally obtain : @xmath46 where @xmath47 is the cut - off radius of the vortex .
this is the core of the vortex .
the constant @xmath48 is determined by the circulation of the core @xmath49 the spatial character of the solution strongly depends on the parameter @xmath43 .
if @xmath50 the infinitely extended distributions for @xmath51 are allowed , e.g. @xmath52 .
they lead to localized vortices with exponential decay of angular velocity .
if @xmath53 only compact distributions , i.e. @xmath54 outside some compact region , are consistent with the finiteness of the total kinetic energy i.e. they corresponds to finite number of particles @xmath55 . as an example we may give : @xmath56 substituting eq . into eq .
one obtains : @xmath57.\ ] ] the corresponding component of the velocity @xmath58 for such a case is shown on fig .
[ vr ] . in the vortex of the local model with @xmath51 given by eq . at @xmath59 . ] taking the rotation on both sides of eq .
one obtains the following equation for the vorticity in a case of planar geometry : @xmath60 this implies that for @xmath53 the vorticity is damped by compression along the flow and therefore such a flow is stable with respect to the vortical perturbations . for @xmath50 the vorticity is damped by expansion .
the hydrodynamics of nonholonomic fluid under consideration differs essentially from the potential dynamics of ideal fluids @xcite due to the nonpotential character of the equation of motion . in ideal fluids
the nontrivial stationary flows are only possible under the influence of an external force ( pressure gradient or reaction of boundaries ) .
moreover the vorticity in the nonholonomic fluid is not coserved .
this is not a case for usual ideal fluids where the vorticity is the integral of motion . the first term in right - hand side of eq .
shows the influence of compression on the evolution of the vorticity .
the second term represent the modified spatial transfer of the vorticity along the flow . since this term can be excluded locally by the choice of the instant local reference frame which moves accordingly with the flow we consider the first term as the main source of the vorticity .
in such an approximation we can write : @xmath61 the circulation is defined by @xmath62 where the integration contour and the corresponding surface area move along with the velocity field .
the time derivative of the circulation is : @xmath63 thus the circulation does not conserve in contrast to the ideal fluid model , where the microscopic interactions are of holonomic character . the total momentum @xmath64 does not conserve too . for the local model we can write : @xmath65 from eq .
it follows that the damping of vortical part of the velocity leads to the formation of the state of uniform motion with @xmath66 .
here we consider the influence of noise on the stability of the flow with respect to the vortical perturbation .
it is clear that instability with respect to such perturbations drives the system to disordered state .
inclusion of stochastic noise can be done in a way analogous to that used in ref .
@xcite : @xmath67 where @xmath68 is the same contribution as before and @xmath69 is the stochastic contribution .
these fluctuations lead to fluctuations of the density and velocity fields . replacing @xmath70 by an average value @xmath71 plus a fluctuating contribution @xmath72 in eq .
one obtains for the above described local model with @xmath53 : @xmath73 the simplest model for the noise is the gaussian white noise approximation : @xmath74 the stochastic equation has the solution : @xmath75 where @xmath76 is the wiener process @xcite . averaging over the realization of the stochastic process we get the averaged evolution of the vorticity : @xmath77 where @xmath78 is the relaxation time of the circulation in the system . for large enough noise @xmath79
the system becomes unstable . for @xmath80
the system is stable and the circulation decays to zero . when @xmath81 increases to @xmath82 the relaxation time @xmath78 goes to infinity , a result similar to critical slowing down near the critical point .
these estimates are modified when other terms , which have been neglected , are taken into account but we believe that qualitatively the obtained results remain unchanged .
it should be noted that the average kernels can also contain noise contributions . in view of the above results for the character of stationary states and vorticity relaxation , which depends on the sign of parameter @xmath43
, this case needs a more thorough investigation . in conclusion
we have constructed a continuum spp model with particle number and kinetic energy conservation .
we found in 2d that vortical solutions exist for the model and that they show a finite flocking behavior .
these solutions , which qualitatively reproduce some results of ref .
@xcite , were obtained without imposing any boundary conditions on the velocity field .
the nonholonomic constrains were found to lead to a circulation which was not conserved .
the influence of noise on the stability of the system was discussed . |
when quantum particles interact , they can team up to form new particles with fractional charge or spin as observed in two - dimensional electron gases or low - dimensional spin lattices .
a particularly fruitful model system to study emerging new particles and their interactions has been the antiferromagnetic ( af ) @xmath0=@xmath1 heisenberg chain .
it does not order even at @xmath3 due to strong quantum fluctuations and its elementary excitations are spinons carrying fractional @xmath0=@xmath1 , which interact only weakly and are unbound particles.@xcite spinons were observed in several materials , including @xmath4 @xcite , @xmath5@xcite and copper pyrazine dinitrate.@xcite strong interactions between spinons can arise from the breaking of spin rotational symmetry as for example in the three - dimensional coupled chain antiferromagnet where the mean field associated with the long - range ordered ground state restricts spin fluctuations .
this creates an attractive potential for the spinons and at low energies they condense into spin - wave excitations carrying @xmath0=@xmath6,@xcite as observed in @xmath4 @xcite and @xmath5.@xcite recently excitations were observed in the @xmath0=@xmath1 chain antiferromagnet @xmath2 ( cdc ) subject to uniform and staggered magnetic fields which were interpreted as bound spinon states.@xcite in the long - wave - length limit , the experiment demonstrated that the low - energy excitations of a @xmath0=@xmath1 chain in a staggered field correspond to the soliton and breather excitations of the quantum sine - gordon model.@xcite however , the sine - gordon model is only valid for a very restricted range of wave - vectors and does not apply for excitations at smaller length scales where the discreteness of the lattice becomes important .
more importantly , it does not fully describe the mechanism by which the staggered field produces an attractive potential that binds spinons into the long - lived dispersive @xmath0=@xmath6 excitations as observed in the experiment . in this paper
, we analyze the dispersion of the observed bound spinon states @xcite in detail , and we report the observation of a bound - spinon state at high energies in cdc . in this system which has a nearest neighbor exchange @xmath7 , a staggered field of the order of @xmath8 , which corresponds to a zeeman energy @xmath9 ,
qualitatively affects the excitation spectrum to energies more than twice @xmath10 at @xmath11 .
this is in stark contrast to spinon binding in coupled chain magnets where the effects are only apparent for @xmath12.@xcite a simple mean - field theory of fermions carrying @xmath0=@xmath1 in one dimension captures the wave - vector dependence of the bound spinon states , and explains the high - energy excitation through the opening of a gap at the fermi surface .
this model represents a first step towards a comprehensive description of the incommensurate excitations in af @xmath0=@xmath1 chains subject to staggered fields .
cdc was identified as an af @xmath0=@xmath1 chain system in which a staggered @xmath13-tensor and/or dzyaloshinskii - moriya ( dm ) interactions @xcite lead to a staggered field @xmath14 upon application of a uniform field @xmath15 . in cdc ,
a uniform magnetic field @xmath16 along the c - axis generates a staggered field @xmath17 along the a - axis , and the hamiltonian can be written as @xmath18where @xmath19 and @xmath20 are the uniform part of the gyromagnetic tensor along the c and a - axis , respectively .
the staggered field is given by @xmath21where @xmath22 is the staggered gyromagnetic form factor and @xmath23 is the length of the dm vector , which points along the b - axis in cdc .
the nearest - neighbor spin exchange along the chain is accurately known from susceptibility measurements @xcite and inelastic neutron scattering.@xcite the spin chains run along the @xmath24-axis of the orthorhombic crystal structure ( _ pnma_),@xcite with the @xmath25 ions separated by @xmath26 .
wave vector transfer is indexed in the corresponding reciprocal lattice @xmath27 , and we define the wave - vector transfer along the chain as @xmath28 . due to weak inter - chain interactions , cdc has long - range af order in zero field below @xmath29 with an af wave - vector @xmath30 .
an applied field along the c - axis suppresses the ordered phase in a second order phase transition at @xmath31,@xcite indicating that inter - chain interactions favor correlations that are incompatible with the field - induced staggered magnetization.@xcite at fields much greater than @xmath32 , the staggered fields thus arise mostly from a staggered @xmath13-tensor and dm interactions and not from interchain interactions .
the neutron scattering experiments were performed on @xmath33 of deuterated single - crystalline cdc .
the measurements were carried out using the spins triple axis spectrometer and the dcs time of flight spectrometer at the nist center for neutron research .
the spins measurements were performed with a focussing analyzer covering @xmath34 in @xmath35 scattering angle set to reflect @xmath36 to the center of the detector .
a be filter rejected neutrons with energies higher than @xmath37 from the detection system .
the measurements were performed with the strongly dispersive direction along the scattered neutron direction to integrate over wave - vectors along a weakly - dispersive directions .
the experimental configuration for the measurements made using the dcs spectrometer and the conversion of those data to absolute units are described in detail elsewhere.@xcite they were performed with an incident energy @xmath38=@xmath39 and the incident beam parallel to the a - axis for configuration a , and with an incident energy @xmath38=@xmath40 and angle of @xmath41 between the incident beam and the a - axis for configuration b.
figure [ fignewexcitation ] demonstrates the dramatic changes that cdc undergoes upon application of a magnetic field . in zero field ,
the neutron scattering spectrum for @xmath42 consists of a strong peak which corresponds to the two - spinon continuum , whose band width , at this wave vector is narrow and barely distinguishable from the experimental resolution . in a @xmath15=@xmath43 field applied along the c - axis , which also induces a staggered field along the a - axis , the scattering includes two resolution - limited excitations . according to the quantum sine - gordon theory
, the lower - energy excitation corresponds to a bound - spinon state which develops into solitons and breathers at long wave - lengths .
the high - energy excitation at @xmath44 , however , does not have a simple interpretation in terms of the sine - gordon model .
its magnetic nature is apparent on account of its field dependence .
the inset to fig .
[ fignewexcitation ] shows the dispersion of this excitation , which has a maximum at @xmath45=@xmath46 .
the dispersion of the high - field excitations at lower energies , @xmath47 , is illustrated in fig .
[ figotherscans ] for two different chain wave - vectors @xmath45 . for @xmath45=@xmath48 , there are two maxima as a function of energy , corresponding to the well - defined modes developing into the sine - gordon soliton and breather excitations at long wave - lengths . for @xmath45=@xmath46 , only one peak is clearly observed because of the weak intensity of one of the modes in this wave - vector region .
the dispersion of these field - induced resonant modes was determined as a function of the chain wave - vector , @xmath45 , by fitting resolution - corrected gaussian line - shapes to the observed scattering .
the adjusted excitation energies are shown in fig .
[ figdispersion]a , illustrating that magnetic spectral weight generally shifts to lower energies in an applied field
. however , due to the @xmath14-induced gap , this effect is much less pronounced than in a uniform field as the ground state energy increases.@xcite
we now present a simple mean - field theory of interacting @xmath0=@xmath1 fermions which captures both the emergence of a new excitation at high energies and the dispersion of the bound spinon states .
as it is well known , a mean - field approach usually is not adequate to solve a one dimensional system .
this is because , for finite range interactions , the fluctuations are strong enough to preclude a non - zero value of an order parameter at any fine temperature .
moreover , if the order parameter is associated with a continuous symmetry , its mean value is zero even at t=0 .
consequently , a mean field theory that assumes a non - zero value of the order parameter with excitations that are originated by small fluctuations of such quantity can not be a good description of a general one dimensional system . in the present case , the system is invariant under rotations around the @xmath49 axis for @xmath50 and the candidate to be the order parameter is the @xmath51 planar component of the staggered magnetization @xmath52 .
as expected for a continuous symmetry and a gapless spectrum , the system is critical at @xmath53 , i.e. , the order parameter has divergent fluctuations .
however , the staggered field @xmath50 couples linearly with the order parameter @xmath54 along the @xmath55 direction .
consequently , for @xmath56 , the u(1 ) rotational symmetry is explicitly broken and the mean value of @xmath54 becomes non - zero at any temperature .
this also changes the nature of the excitations .
the spinons ( kink and antikinks ) are no longer the low energy quasiparticles of the system since a pair of them is now confined by a linear potential .
a similar effect occurs when we increase the dimension of the system due to the interchain interaction .
therefore , we expect a mean - field treatment to be good approach for high enough values of @xmath14 . for @xmath50 , we know that the theory must be critical with the associated linear soft modes shown in fig.[figdispersion]a .
we also know that one dimensional spin system can be mapped into a fermionic system .
in addition , a non - interacting fermionic hamiltonian has a ground state that is also critical and has linear soft modes like the ones shown in fig.[figdispersion]a .
therefore , it is convenient to use a fermionic representation for the mean - field approach . for this purpose ,
the spin degrees of @xmath57 ions are described in terms of fermionic creation and annihilation operators : @xmath58 where @xmath59 and @xmath60 are the pauli matrices . using this fermionic representation , baskaran , zou , and
anderson @xcite proposed a mean - field theory ( mft ) to treat low dimensional heisenberg spin @xmath61 hamiltonians .
the mft was generalized to su(n ) spin models ( for large @xmath62 ) by affleck and marston.@xcite arovas and auerbach @xcite studied this theory in comparison with the bethe ansatz solution and showed that the fluctuation corrections are important in enforcing the gutzwiller projection . using an extended version of this mft , we will study here the ground state properties and the spin dynamics of the @xmath0=@xmath1 chain in a staggered field given by @xmath63 . apart form an irrelevant constant ,
the expression for @xmath63 in the fermionic representation is : @xmath64with @xmath65 .
since there is one spin per site , there is a constraint on the fermion occupation number : @xmath66 with @xmath67 . for the heisenberg term
, we will use a linear combination of the mean - field decoupling introduced in ref . and the other natural decoupling in the presence of a staggered field along the @xmath55 direction : @xmath68 where the lagrange multiplier or chemical potential , @xmath69 , enforces the constraint of one spin per site at mean - field level .
we are assuming translational invariance for @xmath70 , @xmath71 , and a staggered dependence for the effective field @xmath72 .
this staggered dependence is induced by the field @xmath14 . in momentum space
, this leads to @xmath73\ , , \]]which can be written in the matrix formulation as @xmath74\ , , \ ] ] for @xmath75 .
the eigenvalues of this matrix , @xmath76 are the energies of the quasi - particle operators , @xmath77 with @xmath78 @xmath79 is diagonal in the new basis : @xmath80 the mean - field parameters @xmath81 and @xmath82 are given by the self - consistent equations : @xmath83 dk\ , , \nonumber \\ \delta=\frac { 1}{2\pi } \sum_{\sigma } \int_{-\pi/2}^{\pi/2 } u_{k\sigma } v_{k \sigma } [ \langle \alpha^{\dagger}_{k\sigma } \alpha^{\;}_{k\sigma } \rangle - \langle \beta^{\dagger}_{k\sigma } \beta^{\;}_{k\sigma } \rangle ] dk \ , .
\label{self}\end{aligned}\ ] ] the value of @xmath69 is determined by imposing the average occupation per site to be equal to 1 : @xmath84 dk = 1 \,.\ ] ] when @xmath85 , the integration over the phase fluctuations of the local field @xmath86 @xcite renormalizes the value of @xmath81 given by eq.([self ] ) : @xmath87 .
this renormalization improves considerably the comparison with the exact des - cloizeaux - pearson @xcite two - spinon threshold ( @xmath88 ) . to improve the quantitative comparison of the mft with the experiment and the exact diagonalization results
, we will assume here that the same renormalization factor , @xmath89 , must be applied when @xmath15 and @xmath14 are finite .
the value of @xmath81 is a measure of the effective strength of the spin fluctuations introduced by the heisenberg term .
more specifically , @xmath90 is the fermi velocity that in the original spin language corresponds to the velocity of the spinon excitations .
[ figfermi ] shows the evolution of the fermionic bands when the uniform and the staggered fields are applied . in the absence of these magnetic fields ( fig .
[ figfermi]a ) , there is only one band and the two - spinon cross section is associated with the continuum of particle - hole excitations .
the dispersion relation for the lower branch is @xmath91 . when a uniform magnetic field @xmath92 is applied ( fig .
[ figfermi]b ) , the spin up and down bands are split by the zeeman term . as a consequence ,
there is a change @xmath93 of fermi wave vectors @xmath94 and a corresponding change in the wave vectors of the zero energy modes : the energy of the transverse modes goes to zero at @xmath95 and @xmath96 , while the longitudinal excitations have gapless modes at @xmath97 and @xmath98 . from eq .
( [ eqfermiondispersion ] ) , the main effect of a non - zero staggered field @xmath14 is to open a gap at the fermi level , i.e. , the fermionic system becomes an insulator and the spectrum is gaped for any excitation ( fig .
[ figfermi]c ) .
the gap results from the inter - band scattering which is introduced by the staggered field @xmath14 . according to eqs .
( [ qpo1 ] ) , the degree of mixing is maximum at @xmath99 .
the emergence of this gap is consistent with the experimental data shown in fig .
[ figdispersion ] .
to study the excitations of the new ground state induced by @xmath100 it is necessary to calculate the neutron scattering cross section within our mft .
the neutron scattering cross section for the transverse excitations ( @xmath101 ) at @xmath3 is given by : @xmath102 dk\ , , \label{cross}\end{aligned}\ ] ] where @xmath103 for @xmath104 , @xmath105 for @xmath106 and @xmath107 .
note that in eq.([cross ] ) , @xmath108 must be contracted to the reduced brillouin zone ( @xmath109 ) .
the cross section for longitudinal excitations is : @xmath110 dk\ , .
\label{long}\end{aligned}\ ] ] equations ( [ cross ] ) and ( [ long ] ) reveal well - defined excitations that are determined by the cancelation of @xmath111 , i.e. , the divergence of the jacobian . in fig .
[ mfbr]a , we show the different branches of the transverse excitations .
the lower branch corresponds to the dispersion relation of the low energy excitations . in agreement with the experiment ( see fig .
[ figdispersion ] ) , there are two minima , one is located at the incommensurate wave vector @xmath112 and the other one occurs at @xmath95 .
it is interesting to note that for @xmath104 the intensity of this branch goes to zero at @xmath113 due to a cancelation of the matrix element that multiplies the delta function in the integrand of eq .
( [ cross ] ) .
the black and the blue curves of fig .
[ mfbr]a are the upper boundaries of inter - band particle - hole excitations associated with the transverse modes ( fig .
[ figfermi]b ) . more specifically , the black curve results from excitations in which an electron is annihilated in the lower band and created in the upper band , while for the blue curve the process is the opposite . the green curve of fig .
[ mfbr]b is the upper boundary for the intra - band particle - hole excitations that describes the longitudinal modes .
note that these boundaries appear with dashed lines in the spectrum of excitations with the other polarization ( fig .
[ figfermi ] ) .
this is a consequence of the inter - band @xmath95 scattering which is introduced by the staggered field @xmath14 .
the dashed line just indicates that these `` shadow '' branches have a very small intensity .
the gap in the low energy spectrum is not the only qualitative change introduced by the staggered field @xmath14 within the mean - field approach . since the fermi wave vectors @xmath114 are now extremal points of the new bands ( see fig .
[ figfermi]c ) , we expect the emergence of a new branch of transverse excitations associated with transitions between points that are close to @xmath115 ( @xmath114 ) and points in the proximity of @xmath97 ( @xmath95 ) . this new branch of excitations ( see the red curve in fig . [
mfbr]a ) coincides with the new excitation which is experimentally observed at high energies ( see fig .
[ fignewexcitation ] ) , and thus explains our experimental results . for the longitudinal polarization ,
the new branch of excitations is shifted by @xmath116 relative to the transverse polarization ( see the red curve in fig . [
since the maximum of this longitudinal branch is located at @xmath117 , it is difficult to distinguish this branch from the breathers in the experimental data .
the energy of this maximum is close to @xmath118 according to the mft while the experimental value is @xmath119 .
this is reasonable given that the mft is not an adequate approximation to give a quantitative description of the excitations .
the intensities of the different branches are not properly described by the mean - field equations ( [ cross ] ) and ( [ long ] ) .
for instance , the mft predicts a high intensity for the upper boundary of the two - spinon excitations even at zero field for which very accurate calculations are available.@xcite this is clearly an artifact of the mft . to obtain an accurate description of the intensities and the energies of the different branches we complemented our analytical approach with the exact diagonalization of finite size chains . using the lanczos method
, we obtained the exact ground state of @xmath15 for finite chains of length @xmath120 . having the ground state
, we computed the dynamical magnetic susceptibility , @xmath121 , for all the possible wave vectors @xmath122 of a chain of length @xmath123 using the method introduced in ref . .
the wave vector @xmath95 is present in all of the considered chains .
the small changes in the calculated @xmath124 as a function of @xmath123 indicate that the finite size effects are small for the considered problem . in general
, smaller finite size effects are expected for systems that have an excitation gap because the spin - spin correlation length is finite .
the @xmath3 structure factors were calculated using @xmath125the energy spectra were obtained by convoluting the discrete spectra for finite chains with lorentzian functions with a full width at half maximum , @xmath126 , in order to model the experimental energy resolution .
the intensity of the calculated structure factors is given for a chain of @xmath123 spins and normalized so that @xmath127 as required by the total scattering sum rule .
the calculated neutron scattering for all chain lengths was averaged and the dynamic structure factor for the three different polarizations is shown in fig .
[ figexactdiag ] as a function of wave - vector transfer and energy transfer for @xmath14=@xmath128 on an absolute scale .
the structure factor @xmath129 polarized along the uniform field contains a well - defined excitation with a minimum gap energy at an incommensurate wave - vector .
the structure factors @xmath130 and @xmath131 polarized perpendicular to the uniform field contain well - defined excitations whose dispersion has a minimum at the antiferromagnetic point , with @xmath131 having an excitation at lower energy than @xmath130 .
these excitations correspond to the first and second breather of the quantum sine - gordon model .
[ figexactdiag ] also provides evidence that the excitation spectrum contains a substantial amount of continuum states , as observed in the experiment.@xcite these states lie higher in energy than the well - defined low - energy excitations and extend to high energies . at @xmath14=@xmath128
our numerical calculations yield an expectation value of the hamiltonian per @xmath132 of @xmath133 , increased from the ground state energy at zero field , @xmath134 , by @xmath135 .
the numerical data were binned and gaussian fits were used to obtain excitation energies as a function of wave - vector .
these results are shown in fig .
[ figdispersion]a as a solid line , showing that the numerical calculations reproduce the dispersion relation of the low energy modes .
we calculated the dynamic structure factor for the dcs experiment taking into account wave - vector dependent mixing of the polarized dynamic structure factor @xmath136 .
[ figcomparison ] directly compares the calculated and measured intensities on an absolute scale , showing that there is excellent agreement between the numerical calculations and the experiment .
in addition , figure [ figexactdiag ] shows that a new branch of transverse magnetic excitations with the maximum around @xmath42 and @xmath137 emerges when the staggered field is present .
this provides a quantitative explanation for the high energy peak that appears in the neutron scattering data ( see fig .
[ fignewexcitation ] ) .
et al_. @xcite showed that this branch is not present for @xmath50 by using macroscopic selection rules . for non - zero @xmath14 ,
the total spin @xmath0 and its projection along the @xmath49-direction , @xmath138 , are not good quantum numbers anymore , allowing the emergence of this new excitation .
our mean - field approach is in good agreement with this result .
in summary , we have performed neutron scattering experiments , numerical calculations and an analytical study investigating the af @xmath0=@xmath1 chain in uniform and staggered fields .
we found that the incommensurate bound - spinon states are well described by a mapping to an interacting fermionic model after a renormalization of the energy .
the model also explains the emergence of a new excitations with the application of a staggered field .
our results suggest that the proposed mapping is more powerful than initial results suggested , and that it may also be useful for other quantum spin systems with a relatively short correlation length .
work at jhu was supported by the nsf through dmr-0306940 .
dcs and the high - field magnet at nist were supported in part by the nsf through dmr-0086210 and dmr-9704257 . c.
b. and d. h. r. gratefully acknowledge discussions with a. j. millis concerning the mean - field theory of @xmath0=@xmath1 chains . | we report an experimental and theoretical study of the antiferromagnetic @xmath0=@xmath1 chain subject to uniform and staggered fields . using inelastic neutron scattering ,
we observe a novel bound - spinon state at high energies in the linear chain compound @xmath2 .
the excitation is explained with a mean - field theory of interacting @xmath0=@xmath1 fermions and arises from the opening of a gap at the fermi surface due to confining spinon interactions .
the mean - field model also describes the wave - vector dependence of the bound - spinon states , particularly in regions where effects of the discrete lattice are important .
we calculate the dynamic structure factor using exact diagonalization of finite length chains , obtaining excellent agreement with the experiments . |
the vacuum of the standard model ( sm ) scalar potential is unstable since at high energies the higgs effective quartic coupling is driven to negative values by the renormalization group flow @xcite .
nevertheless , the sm can not be a complete theory of nature for various reasons , one of which is that neutrinos need to be massive in order to account for neutrino oscillation results @xcite . with only the sm fields
, neutrino masses can arise in a model - independent way from a dimension 5 effective operator @xmath0 which gives rise to a @xmath1 neutrino mass after electroweak symmetry breaking @xcite .
this same operator unavoidably provides a correction to the higgs self - coupling @xmath2 below the scale of the mechanism of neutrino mass generation through the diagram in fig .
[ fig:1 ] .
although tiny is suppressed by a factor @xmath3 . ] and negative , it suggests that the mechanism responsible for generating neutrino masses and is potentially relevant for the higgs stability problem .
the quantitative effect of neutrino masses on the stability of the scalar potential will , however , be dependent on the ultra - violet completion of the model . contribution of weinberg s effective operator to the higgs quartic interaction . ] after the historic higgs boson discovery at cern and the confirmation of the brout - englert - higgs mechanism , it is natural to imagine that all symmetries in nature are broken spontaneously by the vacuum expectation values of scalar fields .
the charge neutrality of neutrinos suggests them to be majorana fermions @xcite , and that the smallness of their mass is due to the feeble breaking of lepton number symmetry .
hence we need generalized electroweak breaking sectors leading to the double breaking of electroweak and lepton number symmetries . in this letter
we examine the vacuum stability issue within the simplest of such extended scenarios , showing how one can naturally obtain a fully consistent behavior of the scalar potential at all scales for lepton number broken spontaneously at the tev scale .
note that within the simplest gauge structure lepton number is a global symmetry whose spontaneous breaking implies the existence of a physical goldstone boson , generically called majoron and denoted @xmath4 , which must be a gauge singlet @xcite in order to comply with lep restrictions @xcite .
its existence brings in new invisible higgs boson decays @xcite @xmath5 leading to potentially sizable rates for missing momentum signals at accelerators @xcite including the current lhc @xcite .
given the agreement of the atlas and cms results with the sm scenario , one can place limits on the presence of such invisible higgs decay channels .
current lhc data on higgs boson physics still leaves room to be explored at the next run .
absolute stability of the scalar potential is attainable as a result of the presence of the majoron , which is part of a complex scalar singlet .
indeed , it is well known that generically the quartic coupling which controls the mixing between a scalar singlet and the higgs doublet contributes positively to the value of the higgs quartic coupling ( which we shall call @xmath6 ) at high energies @xcite see diagram a in figure [ fig:2 ] . on the other hand , new fermions coupling to the higgs field @xmath7 , such as right - handed neutrinos @xcite , tend to destabilize @xmath6 not only through the 1-loop effect depicted in diagram @xmath8 of figure [ fig:2 ] , but also in what is effectively a two - loop effect ( diagram @xmath9 ) : through their yukawa interaction with @xmath7 , the new fermions soften the fall of the top yukawa coupling at higher energies , which in turn contributes negatively to @xmath6 .
the model we consider below is a low scale version of the standard type i majoron seesaw mechanism , such as the inverse seesaw type @xcite .
we stress however that , even though our renormalization group equations ( rges ) are the same as those characterizing standard case , the values of the dirac type neutrino yukawa couplings are typically much higher in our inverse seeaw scenario . in models with a complex singlet scalar @xmath10 , such as majoron type - i seesaw schemes , the positive contribution to the rge of the higgs quartic coupling ( diagram a ) is accompanied by the destabilizing effect of right - handed neutrinos through the 1-loop diagram @xmath8 and also through the two - loop diagram @xmath9 . ]
the simplest scalar sector capable of driving the double breaking of electroweak and lepton number symmetry consists of the sm doublet @xmath7 plus a complex singlet @xmath10 , leading to the following higgs potential @xcite @xmath11 in addition to the gauge invariance , @xmath12 has a global u(1 ) symmetry which will be associated to lepton number within specific model realizations .
the potential is bounded from below provided that @xmath13 , @xmath6 and @xmath14 are positive ; these are less constraining conditions than those required for the existence of a consistent electroweak and lepton number breaking vacuum where both @xmath7 and @xmath10 adquire non - zero vacuum expectation values ( @xmath15 and @xmath16 ) . for that to happen , @xmath13 , @xmath6 and @xmath17 need to be all positive for energies up to @xmath18 see @xcite for details . ] .
three of the degrees of freedom in @xmath7 are absorbed by the massive electroweak gauge bosons , as usual .
on the other hand , the imaginary part of @xmath10 becomes the nambu - goldstone boson associated to the breaking of the global lepton number symmetry , therefore it remains massless . as for the real oscillating parts of @xmath19 and @xmath10 ,
these lead to two cp - even mass eigenstates @xmath20 and @xmath21 , with a mixing angle @xmath22 which can be constrained from lhc data @xcite .
we take the lighter state @xmath20 to be the 125 gev higgs particle recently discovered by the cms and atlas collaborations . using the renormalization group equations ( given in the appendix )
we evolved the three quartic couplings of the model imposing the vacuum stability conditions mentioned previously . given that such equations rely on perturbation theory ,
the calculations were taken to be trustable only in those cases where the running couplings do not exceed @xmath23 . and @xmath24 at the @xmath25 scale more precise values with higher order corrections can be found in @xcite .
small changes to these input values ( for example a change of 0.03 in the top yukawa @xmath26 ) do not affect substancially our plots . ] +
in order to assign to the u(1 ) symmetry present in eq .
( [ eq : pot ] ) the role of lepton number we must couple the new scalar singlet to leptonic fields .
this can be done in a variety of ways . here
we focus on low - scale generation of neutrino mass @xcite . for definitiveness
we choose to generate neutrino masses through the inverse seesaw mechanism @xcite with spontaneous @xcite .
the fermion content of the standard model is augmented by right - handed neutrinos @xmath27 ( with lepton number + 1 ) and left - handed gauge singlets @xmath28 ( also with lepton number + 1 ) such that the mass term @xmath29 as well as the interactions @xmath30 and @xmath31 are allowed if @xmath10 carries -2 units of lepton number : which is , in principle , also allowed . ]
@xmath32 the effective neutrino mass , in the one family approximation , is given by the expression @xmath33 which shows that the smallness of the neutrino masses can be attributed to a small ( but natural ) @xmath34 coupling , while still having @xmath35 of order one and both @xmath36 , @xmath37 in the tev range .
in most cases , the stability of the potential is threatened by the violation of the condition @xmath38 , as in the standard model .
instability can be avoided with a large @xmath39 , which might , however , lead to an unacceptably large mixing angle @xmath22 between the two cp - even higgs mass eigenstates @xcite . in such cases , one must rely instead on a heavy @xmath21 see the green region in figs .
[ scan1][scan3 ] . indeed , within the red regions therein , the potential becomes unbounded from below at some high energy scale , just like in the standard model .
this happens for relatively small values of either @xmath22 or @xmath40 . as a result ,
a tight experimental bound on @xmath22 can be used to place a lower limit on the mass of the heavier cp - even scalar . from fig .
[ scan1 ] one can also see that the lepton breaking scale @xmath41 must not be too low , otherwise a big ratio @xmath42 will lead to the existence of a landau pole in the running parameters of the model before the planck scale is reached ( shown in orange ) .
this also accounts for the difference between the two plots in fig .
[ scan2 ] . +
as far as the neutrino sector is concerned , since @xmath34 is taken to be small , this parameter has no direct impact on the potential s stability .
however , it should be noted that in order to obtain neutrino masses in the correct range , the values of both @xmath43 and @xmath35 will depend on the one of @xmath34 . in principle
then , @xmath35 might be large , but not too large , as @xmath44 leads to either unstable or non - perturbative dynamics .
a non - zero dirac neutrino yukawa coupling has a destabilizing effect on the scalar potential which is visible in the recession of the green region to bigger values of @xmath22 and @xmath40 , when comparing the bottom plot in fig .
[ scan2 ] and the one in fig .
[ scan3 ] .
+ another interesting possibility is to have a negligible @xmath45 and potentially sizeable @xmath46 . in this case , if we keep @xmath37 of the order of the tev , we find that the region of stability and perturbativity ( shown in green in fig .
[ scan4 ] ) depends significantly upon the parameter @xmath46 characterizing spontaneous and neutrino mass generation through @xmath47 . to be more precise , as shown in fig .
[ scan4 ] the allowed values for the mass of the heavy scalar boson ( @xmath48 ) varies with this yukawa coupling ; for example , if @xmath48 was to be found to be , say , @xmath49 tev ( @xmath50 tev by assumption here ) , then one would conclude that either @xmath51 or the scalar sector must be strongly interacting .
+ the region stability and perturbativity for the case of non - zero @xmath52 and very small @xmath45 is displayed in green ; the color ordering code is the same as in the scan in fig .
[ scan1].,scaledwidth=48.0% ]
in conclusion , the standard model vacuum is unstable for the measured top and higgs boson masses .
however the theory is incomplete as it has no masses for neutrinos .
we have therefore generalized its symmetry breaking potential in order to induce naturally small neutrino masses from the breaking of lepton number .
we have examined the vacuum stability issue in schemes with spontaneous breaking of global lepton number at the tev scale , showing how one can naturally obtain a consistent behavior of the scalar potential at all scales , avoiding the vacuum instability .
given that the new physics parameters of the theory are not known , it sufficed for us to adopt one - loop renormalization group equations .
since all new particles in the model lie at the tev scale , they can be probed with current experiments , such as the lhc .
invisible decays of the two cp - even higgs bosons , @xmath53 , were discussed in @xcite .
improved sensitivity is expected from the 13 tev run of the lhc .
in addition , we expect enhanced rates for lepton flavour violating processes @xcite . in summary
, schemes such as the one explored in this letter may shed light on two important drawbacks of the standard model namely , the instability associated to its gauge symmetry breaking mechanism and the lack of neutrino mass . + * acknowledgements * + we thank sofiane boucenna for discussions in the early phase of this project .
this work was supported by mineco grants fpa2014 - 58183-p , multidark csd2009 - 00064 and the prometeoii/2014/084 grant from generalitat valenciana .
in this appendix we provide some details on the scalar sector of the model . the potential in equation ( [ eq : pot ] )
is controlled by 5 parameters ( @xmath54 , @xmath55 , @xmath13 , @xmath6 , and @xmath39 ) which one can translate into two vacuum expectation values ( @xmath56 and @xmath57 ) , two mass eigenvalues ( @xmath58 and @xmath40 ) and a mixing angle @xmath22 : @xmath59 with @xmath60 on the other hand , it is well known that the standard model potential is controlled by just two parameters @xmath61 and @xmath2 : @xmath62 for a reasonably small mixing angle @xmath22 , one can consider that the state @xmath20 is mostly made of the real part of the singlet , hence we may integrate out @xmath63 . in this approximation , we note that @xmath64 at the scale of decoupling , meaning in particular that there is a tree - level threshold correction between @xmath6 and the standard model quartic coupling @xmath2 .
for the results in this paper , we neglect altogether the small standard model range between the @xmath65 and @xmath58 scale , starting instead with equations ( [ eq : lsigma])([eq : mu2h ] ) , which already include this threshold effect .
for completeness , we write down here the renormalization group equations of the model parameters which are relevant for the study of the potential s stability . we work with the 1-family approximation , ignoring the bottom and tau yukawa couplings .
these equations were obtained with the sarah program @xcite ( see also @xcite ) and explicitly checked by us using the results in @xcite ; furthermore they are consistent with @xcite . as usual
, @xmath66 stands for the natural logarithm of the energy scale .
38ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1016/j.physletb.2012.08.024 [ * * , ( ) ] , link:\doibase 10.1007/jhep12(2013)089 [ * * , ( ) ] , link:\doibase 10.1103/physrevd.90.093006 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.111.241801 [ * * , ( ) ] , link:\doibase 10.1103/physrevlett.43.1566 [ * * , ( ) ] link:\doibase 10.1103/physrevd.22.2227 [ * * , ( ) ] link:\doibase 10.1103/physrevd.92.025024 [ * * , ( ) ] , @noop * * , ( ) link:\doibase 10.1103/physrevd.25.774 [ * * , ( ) ] link:\doibase 10.1088/1674 - 1137/38/9/090001 [ * * , ( ) ] link:\doibase 10.1016/0550 - 3213(93)90337-o [ * * , ( ) ] @noop * * , ( ) , link:\doibase 10.1140/epjc / s2004 - 02011 - 4 [ * * , ( ) ] , @noop * * , ( ) , @noop ( ) , link:\doibase 10.1103/physrevd.62.053005 [ * * , ( ) ] , @noop ( ) , link:\doibase 10.1007/jhep06(2012)031 [ * * , ( ) ] , link:\doibase 10.1103/physrevd.82.055018 [ * * , ( ) ] , link:\doibase 10.1007/jhep01(2010)053 [ * * , ( ) ] , link:\doibase 10.1140/epjc / s10052 - 012 - 2058 - 2 [ * * , ( ) ] , link:\doibase 10.1007/jhep05(2015)057 [ * * , ( ) ] , link:\doibase 10.1007/jhep09(2015)210 [ * * , ( ) ] , link:\doibase 10.1007/jhep12(2015)050 [ * * , ( ) ] , link:\doibase 10.1016/j.physletb.2015.03.015 [ * * , ( ) ] , @noop * * , ( ) @noop * * , ( ) link:\doibase 10.1103/physrevlett.112.201802 [ * * , ( ) ] , link:\doibase 10.1140/epjc / s10052 - 014 - 2980 - 6 [ * * , ( ) ] , https://cds.cern.ch/record/2007270[__ ] , ( , , ) link:\doibase 10.1007/jhep08(2012)098 [ * * , ( ) ] , link:\doibase 10.1155/2014/831598 [ * * , ( ) ] , @noop * * , ( ) @noop * * , ( ) , @noop * * , ( ) , @noop ( ) , link:\doibase 10.1016/j.cpc.2013.12.002 [ * * , ( ) ] , link:\doibase 10.1103/physrevd.67.065019 [ * * , ( ) ] , | the vacuum of the standard model is known to be unstable for the measured values of the top and higgs masses . here
we show how vacuum stability can be achieved naturally if lepton number is violated spontaneously at the tev scale .
more precise higgs measurements in the next lhc run should provide a crucial test of our symmetry breaking scenario .
in addition , these schemes typically lead to enhanced rates for processes involving lepton flavour violation . |
in the present paper the szeged index and the wiener index of partial cubes are investigated .
our main result extends a parallel result about the szeged index of benzenoid systems @xcite to all partial cubes . in @xcite
the corresponding indices were expressed as the sum of indices of three weighted quotient trees , while in the general case these trees are quotient graphs of a partial cube with respect to a partition coarser than @xmath0-partition .
partial cubes constitute a large class of graphs with a lot of applications and includes , for example , many families of chemical graphs ( benzenoid systems , trees , @xmath1 systems , phenylenes , cyclic phenylenes , polyphenylenes ) .
therefore , our results can be used to calculate the indices of a particular family of chemical graphs in linear time . the wiener index and the szeged index
are some of the most commonly studied topological indices .
their history goes back to @xmath2 , when h. wiener used the distances in the molecular graphs of alkanes to calculate their boiling points @xcite .
this research has led to the wiener index , which is defined as @xmath3 in that paper it was also noticed that for every tree @xmath4 it holds @xmath5 where @xmath6 denotes the number of vertices of @xmath4 whose distance to @xmath7 is smaller than the distance to @xmath8 and @xmath9 denotes the number of vertices of @xmath4 whose distance to @xmath8 is smaller than the distance to @xmath7 .
therefore , a proper generalization to any graph was defined in @xcite as @xmath10 this topological index was later named as the szeged index .
for some recent results on the wiener index and the szeged index see @xcite .
the wiener index , due to its correlation with a large number of physico - chemical properties of organic molecules and its interesting mathematical properties , has been extensively studied in both theoretical and chemical literature .
later , the szeged index was introduced and it was shown that it also has many applications , for example in drug modelling @xcite and in networks @xcite .
the paper reads as follows . in the next section
we give some basic definitions needed later . in section
[ main ] we prove that the problem of calculating the szeged index of a partial cube can be reduced to the problem of calculating the szeged indices of weighted quotient graphs with respect to a partition coarser than @xmath0-partition .
it turns out that such quotient graphs are again partial cubes .
we also consider a similar result for the wiener index , which was proved in @xcite under the name partition distance . in section [ uporaba ]
we show that the mentioned results generalize already known results for benzenoid systems ( see @xcite ) and apply them to @xmath1 systems .
furthermore , we demonstrate with an example that the method can also be used to calculate the corresponding indices by hand .
the _ distance _ @xmath11 between vertices @xmath12 and @xmath13 of a graph @xmath14 is the length of a shortest path between vertices @xmath7 and @xmath8 in @xmath14 . we also write @xmath15 for @xmath11 .
two edges @xmath16 and @xmath17 of graph @xmath14 are in relation @xmath0 , @xmath18 , if @xmath19 note that this relation is also known as djokovi ' c - winkler relation .
the relation @xmath0 is reflexive and symmetric , but not necessarily transitive .
we denote its transitive closure ( i.e. the smallest transitive relation containing @xmath0 ) by @xmath20 .
let @xmath21 be the @xmath20-partition of the set @xmath22 .
then we say that a partition @xmath23 of @xmath22 is _ coarser _ than @xmath24 if each set @xmath25 is the union of one or more @xmath20-classes of @xmath14 ( see @xcite ) .
the _ hypercube _ @xmath26 of dimension @xmath27 is defined in the following way : all vertices of @xmath26 are presented as @xmath27-tuples @xmath28 where @xmath29 for each @xmath30 and two vertices of @xmath26 are adjacent if the corresponding @xmath27-tuples differ in precisely one coordinate . therefore , the _ hamming distance _ between two tuples @xmath12 and @xmath13 is the number of positions in @xmath12 and @xmath13 in which they differ . a subgraph @xmath31 of @xmath14 is called an
_ isometric subgraph _ if for each @xmath32 it holds @xmath33 .
any isometric subgraph of a hypercube is called a _ partial cube_. for an edge @xmath34 of a graph @xmath14 ,
let @xmath35 be the set of vertices of @xmath14 that are closer to @xmath36 than to @xmath37 .
we write @xmath38 for the subgraph of @xmath14 induced by @xmath39 .
the following theorem puts forth two fundamental characterizations of partial cubes : @xcite [ th : partial - k ] for a connected graph @xmath14 , the following statements are equivalent : * @xmath14 is a partial cube . * @xmath14 is bipartite , and @xmath40 and @xmath41 are convex subgraphs of @xmath14 for all @xmath42 . *
@xmath14 is bipartite and @xmath43 .
is it also known that if @xmath14 is a partial cube and @xmath44 is a @xmath0-class of @xmath14 , then @xmath45 has exactly two connected components , namely @xmath40 and @xmath41 , where @xmath46 . for more information about partial cubes
see @xcite .
let @xmath31 and @xmath14 be arbitrary graphs .
then a mapping @xmath47 is an _
isometric embedding _ ,
if @xmath48 for each @xmath32 .
suppose @xmath49 is a partition of the vertices of a graph @xmath14 .
the _ quotient graph _ @xmath50 is a graph with vertex set @xmath49 , and for which distinct classes @xmath51 are adjacent if some vertex in @xmath52 is adjacent to a vertex of @xmath53 .
the _ wiener index _ of a graph @xmath14 is defined as @xmath54 .
let @xmath14 be a graph and @xmath55 an edge of @xmath14 . throughout the paper we will use the following notation : @xmath56 @xmath57 also , for @xmath58
, let @xmath59 using this notation , the _ szeged index _ of a graph @xmath14 is defined as @xmath60 now we extend the above definitions of the indices to weighted graphs as follows .
let @xmath14 be a graph and let @xmath61 and @xmath62 be given functions .
then @xmath63 and @xmath64 are a _
vertex - weighted graph _ and a _ vertex - edge - weighted graph _ , respectively .
the _ vertex - weighted wiener index _ of @xmath63 is defined as @xmath65 furthermore , if @xmath55 is an edge of @xmath14 and @xmath66 , set @xmath67 the _ vertex - edge - weighted szeged index _
is defined as @xmath68 if @xmath14 is a partial cube and @xmath44 a @xmath0-class , then for any two edges @xmath69 and @xmath66 it holds @xmath70 therefore , it is legitimate to define @xmath71 @xmath72 where @xmath73 .
in this section we prove the main result of the paper .
we start with the following basic proposition .
let @xmath14 be a partial cube with @xmath27 vertices and let @xmath74 be its @xmath0-classes .
for @xmath75 , let @xmath76 be the number of vertices in one of the two connected components of @xmath77 .
then @xmath78 and @xmath79 the proof for the wiener index is based on the canonical embedding , for the details see the proof of ( * ? ? ?
* proposition 19.2 ) . since for every two edges @xmath80 of the same @xmath0-class of @xmath14 if holds ( without loss of generality ) @xmath81 and @xmath82
, the proof for the szeged index follows obviously
. the above proposition can be useful for deriving closed formulas for some families of graphs .
but since the number of @xmath0-classes of a partial cube can be very large , in the rest of this section we generalize this result such that the computation is possible for any partition of @xmath22 coarser than @xmath0-partition . as a consequence
, the calculation of these two indices will be much more efficient . to prove the main theorem ,
some preparation is needed .
the following lemma claims that every quotient graph obtained from a partition coarser than @xmath0-partition is again a partial cube .
[ partial ] let @xmath14 be a partial cube .
if @xmath23 is a partition coarser than the @xmath0-partition , then @xmath83 is a partial cube for each @xmath84 , @xmath85 .
let @xmath25 be the union of @xmath0-classes @xmath86 and @xmath87 .
since @xmath14 is a partial cube , the graph @xmath88 has exactly two connected components ; denote these two components by @xmath89 and @xmath90 .
we define function @xmath91 in the following way : @xmath92 for every @xmath93 .
+ let @xmath94 be defined as @xmath95 for each @xmath93 .
+ if @xmath96 , then @xmath97 is the number of @xmath0-classes @xmath98 , @xmath87 , such that @xmath12 and @xmath13 are in different components of the graph @xmath88 .
hence , @xmath99 which is the hamming distance .
obviously , @xmath100 is an isometric embbeding into @xmath101-dimensional hypercube .
therefore , @xmath83 is a partial cube .
let @xmath14 be a partial cube and let @xmath23 be a partition coarser than the @xmath0-partition .
we next extend the quotient graphs @xmath83 , @xmath102 , to weighted graphs as follows : 1 . for @xmath93 ,
let @xmath103 be the number of vertices in the connected component @xmath12 of @xmath104 , 2 . for @xmath105 ,
let @xmath106 be the number of edges in @xmath25 with one end - point in @xmath12 and another in @xmath13 .
now we are ready to state the following lemma .
[ lemica ] let @xmath14 be a partial cube and @xmath23 a partition coarser than the @xmath0-partition .
moreover , let @xmath107 be the weighted quotient graph for some @xmath108 and @xmath100 the function defined in the proof of lemma [ partial ] .
let @xmath25 be the union of @xmath0-classes @xmath86 and @xmath87 .
then @xmath109 is a @xmath0-class of @xmath83 and @xmath110 moreover , if @xmath111 , then @xmath112 that the set @xmath113 is a @xmath0-class of @xmath83 follows directly from the fact that @xmath100 is the hamming distance and @xmath83 is a partial cube ( see lemma [ partial ] ) .
every edge @xmath114 represents @xmath115 edges of @xmath98 .
also , any edge @xmath116 is represented by some edge @xmath114 , since each edge of @xmath14 is in exactly one @xmath0-class of @xmath14 .
hence , @xmath110 let @xmath111 .
every vertex @xmath117 represents @xmath118 vertices of @xmath119 .
conversely , any vertex @xmath120 is represented by exactly one vertex @xmath117 .
therefore , @xmath112 the main result of this paper now follows from lemma [ lemica ] .
[ segi ] let @xmath14 be a partial cube .
if @xmath23 is a partition coarser than the @xmath0-partition , then @xmath121 let all the notation be the same as in lemma [ lemica ] .
we obtain @xmath122 using lemma [ lemica ] it follows @xmath123 in the last step we have taken into account that @xmath113 is a @xmath0-class and therefore , @xmath124 for any @xmath114 and @xmath66 . since @xmath125 is a partition of the set @xmath126 , we finally deduce @xmath127 and the proof is complete . at the end of this section
we also consider the wiener index of partial cubes .
this problem was solved in ( * ? ? ?
* theorem 2.3 ) , where it was proved that the partition distance of a connected ( weighted ) graph can be expressed as the sum of the wiener indices of weighted quotient graphs with respect to a partition coarser than @xmath20-partition .
in a special case , the partition distance is the wiener index . since for the partial cubes it holds @xmath43 , we get the following result .
@xcite [ wini ] let @xmath14 be a partial cube .
if @xmath23 is a partition coarser than the @xmath0-partition , then @xmath128
the results proven in section [ main ] are already well - known for benzenoid systems ( see @xcite ) .
since each benzenoid system is a partial cube , these well - known results also follows directly from the results from section 3 .
therefore , in this section we will concentrate to the @xmath1 systems .
more precisely , for @xmath1 systems we show that the wiener index and szeget index can be computed in linear time .
let @xmath129 be the infinite @xmath1 net and let @xmath130 be a cycle in it .
then a @xmath1 _ system _ is induced by the vertices and edges of @xmath129 , lying on @xmath130 and in its interior . obviously , any @xmath1 system is a bipartite graph , since the whole @xmath1 net is bipartite .
@xmath1 systems are important in chemical graph theory , since they are subgraphs of @xmath131 and @xmath132 nanotubes and nanotori , which are some of the central and most commonly studied families of chemical graphs ( for example , see @xcite ) .
also , the infinite version of a @xmath1 system was considered in @xcite .
in addition , @xmath1 systems can represent chemical compounds , formed by the rings of cyclobutane and cyclooctane . in the next theorem
we show that each @xmath1 system is a partial cube .
let @xmath14 be a @xmath1 system .
two edges of an internal face of @xmath14 are in relation @xmath0 if they are antipodal .
if the edge @xmath133 is contained in two different internal faces @xmath134 and @xmath135 , then @xmath136 is in relation @xmath0 with two different antipodal edges @xmath137 and @xmath138 .
moreover , @xmath139 and @xmath140 are also in relation @xmath0 and therefore , @xmath136 , @xmath139 and @xmath140 are all in relation @xmath0 . again ,
if @xmath139 ( or @xmath140 ) is contained in two different internal faces , then we obtain the edge @xmath141 which is in relation @xmath0 with edges @xmath136 , @xmath139 , and @xmath140 . without loss of generality
we assume that @xmath142 , where @xmath27 is a positive integer , is a maximal set of edges obtained as above .
obviously , @xmath143 is not connected .
moreover , it contains exactly two connected component , namely @xmath144 and @xmath145 .
clearly , these components are convex subgraphs of @xmath14 .
now it follows from theorem [ th :
partial - k ] that @xmath14 is a partial cube .
let @xmath14 be a @xmath146 system and let @xmath147 be its @xmath0-classes
. without loss of generality , assume that @xmath14 contains at least one internal face of length @xmath148 .
based on this assumption we have in @xmath14 four different directions of edges .
let @xmath25 , @xmath149 , be the union of all @xmath0-classes with the same direction .
obviously , @xmath150 is a partition coarser than the @xmath0-partition . from the construction of @xmath151 it follows that @xmath151 is a tree for each @xmath152 ( the existence of a cycle in @xmath151 would imply that @xmath14 is not a @xmath146 system ) .
next , we extend the quotient trees @xmath151 , @xmath153 , to weighted trees @xmath154 and @xmath155 as in section [ main ] .
our next goal is to prove that the wiener index and the szeged of a @xmath1 system can be computed in @xmath158 time .
since the weighted quotient trees can be computed in @xmath158 time , it suffices to show that the wiener indices and the szeged indices of weighted trees can be computed in @xmath158 time . in @xcite
it was mentioned that if @xmath159 is a vertex - weighted tree , then the wiener index can be computed as @xmath160 it was also shown that using this formula , the wiener index of a weighted tree can be computed in linear time .
let @xmath4 be a rooted tree with a root @xmath12 and label the vertices of @xmath4 such that , if a vertex @xmath13 is labelled @xmath100 , then all vertices in the subtree rooted at @xmath13 have labels smaller than @xmath100 . using the standard bfs ( breadth - first search ) algorithm
this can be done in linear time .
next we visit all the vertices of @xmath4 according to this labelling ( we start with the smallest label ) .
furthermore , for every vertex @xmath13 , we will adopt weights @xmath163 and calculate new weights @xmath164 .
obviously , for every vertex @xmath13 of the tree @xmath4 , we can consider the subtree rooted at @xmath13 as a connected component of the graph @xmath168 , where @xmath169 is the up - edge of @xmath13 .
therefore , @xmath170 . let @xmath171 ( this can be computed in linear time ) .
it follows that @xmath172 .
let @xmath173 be the sum of numbers @xmath174 for all down - neighbours @xmath167 of @xmath13 ( and @xmath175 if @xmath13 is a leaf ) .
finally , set @xmath176 if @xmath177 and @xmath178 , if @xmath179 .
it is obvious that @xmath180 and the proof is complete .
99 t. al - fozan , p. manuel , i. rajasingh , r. s. rajan , computing szeged index of certain nanosheets using partition technique , match commun .
72 ( 2014 ) 339353 .
m. arezoomand , random walks on infinite @xmath190 net , nanotube and nanotori , match commun .
65 ( 2011 ) 231240 .
a. r. ashrafi , s. yousefi , computing the wiener index of a @xmath131 nanotorus , match commun .
57 ( 2007 ) 403410 .
v. chepoi , on distances in benzenoid systems , j. chem .
36 ( 1996 ) 11691172 .
v. chepoi , s. klavar , the wiener index and the szeged index of benzenoid systems in linear time , j. chem .
37 ( 1997 ) 752755
. m. diudea , m. stefu , wiener index of @xmath1 nanotubes , match commun .
50 ( 2004 ) 133144 .
i. gutman , a formula for the wiener number of trees and its extension to graphs containing cycles , graph theory notes n. y. 27 ( 1994 ) 915 .
r. hammack , w. imrich , s. klavar , handbook of product graphs , second edition , crc press , taylor & francis group , boca raton , 2011 .
a. heydari , b. taeri , szeged index of @xmath131 nanotubes , european j. combin . 30
( 2009 ) 11341141 .
a. kelenc , s. klavar , n. tratnik , the edge - wiener index of benzenoid systems in linear time , match commun .
74 ( 2015 ) 521532 .
p. v. khadikar , s. karmarkar , v. k. agrawal , j. singh , a. shrivastava , i. lukovits , m. v. diudea , szeged index - applications for drug modeling , lett .
drug . des .
discov . 2 ( 2005 )
606624 . s. klavar , m. j. nadjafi - arani , partition distance in graphs , ( 2016 ) , preprint .
s. klavar , m. j. nadjafi - arani , wiener index versus szeged index in networks , discrete appl .
( 2013 ) 11501153 .
m. knor , r. krekovski , a. tepeh , an inequality between the edge - wiener index and the wiener index of a graph , appl .
comput . 269
( 2015 ) 714721 .
m. knor , r. krekovski , a. tepeh , mathematical aspects of wiener index , ars math . contemp . 11 ( 2016 ) 327352 .
t. pisanski , m. randi , use of the szeged index and the revised szeged index for measuring network bipartivity , discrete appl .
( 2010 ) 19361944 .
n. tratnik , the edge - szeged index and the pi index of benzenoid systems in linear time , match commun .
77 ( 2017 ) , in press . h. wiener , structural determination of paraffin boiling points , j. amer .
69 ( 1947 ) 1720 .
s. yousefi , a. r. ashrafi , an exact expression for the wiener index of a @xmath132 nanotorus , j. math .
42 ( 2007 ) 10311039 . | in this paper we study the szeged index of partial cubes and hence generalize the result proved by v. chepoi and s. klavar , who calculated this index for benzenoid systems .
it is proved that the problem of calculating the szeged index of a partial cube can be reduced to the problem of calculating the szeged indices of weighted quotient graphs with respect to a partition coarser than @xmath0-partition .
similar ( but more general ) result was recently proved by s. klavar and m. j. nadjafi - arani .
furthermore , we show that such quotient graphs of partial cubes are again partial cubes . since the results can be used to efficiently calculate the wiener index and the szeged index for specific families of chemical graphs , we consider @xmath1 systems and show that the two indices of these graphs can be computed in linear time
. szeged index , wiener index , partial cube , benezenoid system , @xmath1 system 92e10 , 05c12 |
the author is grateful to p. bambade , m. berggren and f. richard for the fruitful collaboration on the mass threshold method and to u. martyn for providing material from his study based on analyzing the energy spectrum .
99 p. bambade , m. berggren , f. richard and z. zhang , hep - ph/0406010 .
u. martyn , hep - ph/0406123 .
bennett _ et al_. , _ astrophys .
j. suppl_. 148 ( 2003 ) 1 .
m. battaglia et al . , _ eur .
j. _ * c33 * ( 2004 ) 273 - 296 , hep - ph/0306219 .
g. blanger et al . , _ comput .
_ * 149 * ( 2002 ) 103 - 120 , hep - ph/0112278 .
m. berggren , sgv 2.31 - a fast and simple program for simulating high energy physics experiments at colliding beam detectors . | this talk presents the detection issues for the lightest slepton @xmath0 at a future @xmath1 tev collider given the dark matter constraints set on the susy mass spectrum by the wmap results .
two methods for measuring its mass @xmath2 and the resulting precision on the dark matter density are briefly discussed in the susy msugra scenario with @xmath3-parity conservation when the mass difference between @xmath2 and that of the lightest neutralino is small ( a few gev ) .
the analysis is performed with tesla parameters in both head - on and crossing angle modes .
# 1#2#3#4#1 * # 2 * , # 3 ( # 4 ) * lal 04 - 100 * + * experimental implications for a linear + collider of susy dark matter scenario * + in2p3-cnrs et universit de paris - sud , bt . 200 ,
bp 34 , f-91898 orsay cedex the talk is based on recent studies [ 1,2 ] motivated by the increasing awareness in the community of the role of an @xmath1 linear collider ( lc ) for a precise determination of the susy parameters which are needed to interpret the dark matter ( dm ) content of the universe .
after the wmap results [ 3 ] leading to an accuracy at the @xmath4 level or @xmath5 in two standard deviation range and awaiting for the planck mission in @xmath6 which aims at @xmath7 , it seems appropriate to check that a lc can do its job properly on this essential topic . in the susy scenario with @xmath3-parity conservation ,
the lightest susy particle ( lsp ) is the lightest neutralino @xmath8 .
this particle is considered as the best candidate to satisfy the cosmological constraints on dm in the universe .
dm constraints have been recently re - examined [ 4 ] within the msugra scenario , confronting the precise predictions obtained after the wmap results .
these data imply , for many of the benchmark points retained , a very small difference between the mass of the lightest slepton ( @xmath0 ) , the susy partner of the @xmath9 , and the lsp mass ( @xmath10 ) since one of the preferred mechanism to regulate the amount of dm in the universe is the so - called ` co - annihilation mechanism ' .
previous studies have shown that the masses of both smuon and lsp can be precisely measured using the so - called end - point method [ 2 ] . here
we shall thus concentrating on the measurement of stau mass , which is relevant as the amount of dm depends critically on it .
two methods are presented for this purpose .
the first is appropriate when the stau mass is comparable to the beam energy and the expected cross section of the stau production is small . the second works when the stau mass is significantly smaller than the beam energy and the stau production cross section is large .
we also address detectability issues related to different collision mode either in head - on or with a half crossing angle of @xmath11mrad .
the end - point method could not applied to the stau analysis as there are additional missing energies arising from neutrinos in subsequent @xmath9 decays .
furthermore , the final state particle is very soft , typically a few gev for @xmath12gev taking benchmark point @xmath13 in [ 4 ] as a working point .
another difficulty comes from the fact that the signal cross section is often many orders of magnitude smaller than that of the standard model ( sm ) processes . for point @xmath13 at center - of - mass energy ( @xmath14 ) of @xmath15gev ,
the cross section of the signal process @xmath16 is around @xmath11fb , which is to be compared with @xmath17fb of the dominant sm background processes @xmath18 and @xmath19 .
the spectator @xmath20 in the background process is however predominantly peaked in the forward direction .
therefore an efficient tagging down to lowest possible angle is crucial in rejecting these background events .
quantitative studies show that the current tagging efficiency of the beam monitor calorimeter , lcal , does not allow for a background free analysis .
such an analysis may be achieved [ 1 ] when the lcal is fully efficient in tagging all spectator @xmath20 having a transverse momentum above @xmath21gev and when additional discriminating variables are used .
one such variable is the scalar sum of transverse momentum with respect to the thrust axis in the plane transverse to the beam directions .
the resulting signal efficiency and background contribution from the dominant processes in the head - on collision are summarized in table [ tab : effevtd ] .
.__the efficiency , the signal and dominant background events in the head - on case for benchmark point d@xmath22 at @xmath23gev . _ _ [ cols="^,^,^,^",options="header " , ] the program micromegas [ 5 ] has been used to compute the relative uncertainty on the dm density due to the susy mass error measurements .
this program operates without any assumption , in particular it does not rely on the msugra scheme .
results listed in table [ tab : results ] show , as expected , that @xmath24 depends primarily on the precision on the stau and lsp masses .
the analysis , optimized for the d@xmath22 solution , gives satisfactory results except for point j@xmath22 which is almost beyond detectability .
the second method works at higher beam energies but still below the mass thresholds of other sparticles , the idea being at such large energies , the signal cross section is big enough to collect a large event sample .
the analysis also benefits from using explicitly the polarized beams to enhance the signal over background ratio .
the stau mass can then be determined by analyzing the high energy spectrum .
indeed for the same benchmark point @xmath13 , if @xmath14 could be chosen to be at @xmath25gev , sufficiently higher than the stau mass , a more precise stau mass and therefore dm density could be achieved even with a moderately small integrated luminosity of @xmath26pb@xmath27 .
the same method has also been applied to a sps1a inspired model for three different @xmath28 values . again
the precision on the stau mass and dm density improves as @xmath28 increases . to summarize
, our studies have shown that the detection and the mass measurement of the tau slepton , potentially important in view of the cosmological implications , is challenging in the so - called `` co - annihilation '' scenario .
a forward veto to remove the @xmath29 background down to very small angles is essential to reach an almost background free result , adequate to achieve the accuracy implied by the post - wmap results in a model independent analysis . in our analysis with method one
, we have assumed an ideal detector for particle detection but with realistic detector acceptance as expected from a fast simulation program sgv [ 6 ] , developed and tested at lep .
some of the detector capabilities are not yet fully explored , e.g. , the @xmath30 information of the tracking device and possible different decay lengths ( secondary vertex distributions ) between the signal and the background events .
the analysis in terms of the signal and background separation may still be improved using more sophisticated likelihood methods instead of simple cuts .
nevertheless , the stau analysis in collisions with a crossing angle is likely to be more difficult ( thought still feasible ) than in head - on collision , only possible in the tesla scheme . in a warm machine like nlc
, the same conclusion could be reached provided that there is no degradation due to pile - up of several bunches in the forward veto ( this may require some r@xmath31d for a very fast calorimeter ) . |
the concept of a singular point of a tropical variety is not well established yet .
a natural definition is the following .
let @xmath0 be an algebraically closed field of characteristic @xmath1 with a valuation @xmath2 .
we say that a point @xmath3 in a tropical variety @xmath4 is singular if there exists a singular algebraic subvariety of the torus @xmath5 , with tropicalization @xmath6 , with a singular point of valuation @xmath3 ( see definition [ def : tropical_singularity ] ) . this definition of singularity in terms of the tropicalization of classical algebraic varieties has been considered in @xcite in the case @xmath7 of planar curves , and indirectly in @xcite and @xcite , in the general hypersurface case .
thus , in principle , one should study all the _ preimages _ of @xmath6 under the valuation map to decide whether @xmath6 is singular .
we present an equivalent formulation when @xmath6 is a hypersurface defined by a tropical polynomial with prescribed support @xmath8 and the residue field of @xmath0 has also characteristic @xmath1 ( but our approach can be extended if this hypothesis is relaxed ) .
recall that given a finite set @xmath9 , gelfand , kapranov and zelevinsky @xcite defined and studied the main properties of the @xmath8-discriminant @xmath10 associated to the family of hypersurfaces with support @xmath8 .
let @xmath11 be the variety of laurent polynomials @xmath12 with coefficients in @xmath13 and support in @xmath8 which define a singular hypersurface @xmath14 in the torus @xmath15 .
if @xmath11 has codimension one , then there exists a unique ( up to sign ) polynomial @xmath16 $ ] such that if @xmath17 has a singular point in @xmath15 then @xmath18 .
this polynomial is called the _ @xmath8-discriminant _ and
its locus coincides with the dual variety @xmath19 of the equivariantly embedded toric variety @xmath20 rationally parametrized by the monomials with exponents in @xmath8 .
if codim@xmath21 ( in which case @xmath20 is said to be _ defective _ ) , the polynomial @xmath10 is defined to be the constant polynomial @xmath22 , and we call in this case @xmath8-discriminant the ideal of @xmath19 . the varieties @xmath20 and @xmath19 , as well as the @xmath8-discriminant , are affine invariants of the configuration @xmath8 .
a tropical polynomial @xmath23 with coefficients in @xmath24 defines a singular tropical hypersurface precisely when its vector of coefficients @xmath25 lies in the tropicalization @xmath26 of the @xmath8-discriminant .
the concept of singularities of tropical varieties , as well as the concept of tropical tangency , can thus be addressed via the tropicalization of the @xmath8-discriminant described in @xcite .
we call @xmath27 the integer matrix with columns @xmath28
. theorem 1.1 in @xcite states that the tropical discriminant equals the minkowski sum of the co - bergman fan @xmath29 and the row space of the matrix @xmath30 .
the co - bergman fan @xmath31 is the tropicalization of the kernel of @xmath30 , or equivalently , the tropicalization of the space of affine relations among the vectors @xmath32 . following the notations in section [ sec : euler_derivatives ] ( cf . also the discussion in section [ sec : bergman ] ) , we can give an equivalent appealing definition of singular point of the tropical hypersurface defined by @xmath33 .
let @xmath34 be the linear form with all coefficients equal to @xmath22 , and denote by @xmath35 the tropical hypersurface @xmath36 consisting of those vectors @xmath37 for which the minimum of the coordinates of @xmath38 is attained at least twice
. clearly , @xmath39 .
we then have @xmath40 and the singularities @xmath41 are described by @xmath42 deciding whether a given tropical polynomial defines a singular tropical hypersurface , amounts with this approach to finding a way of writing its vector of coefficients @xmath25 as the sum of an element in @xmath31 plus an element in the rowspan of @xmath30 .
this is particularly involved when @xmath20 is defective , and there was no algorithm known in the general case ( cf .
@xcite , where an algorithm is presented under some geometric assumptions , or the arguments in the proof of ( * ? ? ?
* lemma 3.12 ) ) . instead , we give in theorem [ teo : singular_point_by_derivatives ] a direct characterization of tropical singular points in terms of analogs of euler derivatives of tropical polynomials , which allows us to recover theorem 1.1 in @xcite .
our tropical approach translates into an algorithm to decide whether the tropical variety associated to @xmath43 is singular and to detect all the singular points .
note that , given a laurent polynomial @xmath44 , if the vector of valuations @xmath45 defines a non singular tropical hypersurface , we get a `` certificate '' that @xmath12 defines a non singular hypersurface in the torus @xmath46 .
however , it is not possible to find a simple combinatorial formula to describe all singular points because the situation is not , as one could expect , completely local ( cf .
proposition [ prop : anotherpt ] , and the concept of @xmath47-equivalence by gelfand , kapranov and zelevinsky ) .
we give several combinatorial conditions which characterize singular tropical hypersurfaces . in @xcite or @xcite ,
tropical smooth curves are defined in terms of coherent subdivisions where all points in @xmath8 are marked , and which define primitive triangulations of the convex hull @xmath48 of @xmath8 ( cf . also the concept of singular tropical curves of maximal dimensional type in @xcite ) . if the dual subdivision of @xmath48 induced by a tropical curve @xmath49
is a primitive triangulation , @xmath49 will always be a smooth curve in our sense too .
but our definition allows for certain non primitive triangulations which correspond to smooth tropical hypersurfaces , that is , tropical hypersurfaces that can not be the tropicalization of an algebraic hypersurface with a singularity in the algebraic torus @xmath50 ( see examples [ ex : triang2 ] , [ ex : nontriang2 ] and proposition [ teo : singular_dim_1_and_2 ] ) .
when @xmath8 does not admit a unimodular triangulation or if not all the lattice points in the convex hull of @xmath8 are marked , we can have smooth points at facets where the weight ( as it is currently defined , see @xcite ) is @xmath51 .
note that we concentrate on _ affine _ singular points , that is points in the `` torus '' @xmath24 of the tropical semifield .
our study includes all coherent subdivisions of @xmath8 .
thus , we refine the definition in @xcite and we explore the whole @xmath8-discriminant .
our method also generalizes trivially to hypersurfaces in arbitrary dimension . in the last section ,
we apply our tools to define and study the non - transversal intersections of two tropical curves with fixed monomial support , that is , the tropicalization of mixed discriminants of bivariate polynomials .
we fix throughout the text a finite lattice set @xmath8 in @xmath52 of cardinality @xmath53 .
we will assume without loss of generality that the @xmath54-linear span of @xmath8 equals @xmath52 .
the @xmath24-_affine _ span of a subset @xmath55 in @xmath56 will be denoted by @xmath57 .
we consider the tropical semifield @xmath58 , where @xmath59 and the tropical operations are defined by @xmath60 , @xmath61 .
our object of study are _ tropical polynomials _
@xmath62 $ ] with _ support @xmath8 _ , that is @xmath63 for _ all _ @xmath64 . to simplify the notation , we will write @xmath65 . the _ tropical hypersurface _ defined by a non zero tropical polynomial @xmath43 with support @xmath8 , is the set @xmath66 any tropical hypersurface is a rational polyhedral complex .
for any @xmath67 , its associated cell @xmath68 is the closure of all the points @xmath69 for which @xmath70 and @xmath71 are attained at the same subset @xmath72 of @xmath8 .
each cell @xmath68 comes with a marking , given by the subset @xmath72 .
so , a tropical hypersurface associated with a tropical polynomial with fixed support @xmath8 will be a _ marked rational polyhedral complex_. this marking will be transparent in the notation .
we refer to the beginning of section [ sec : bergman ] for further details
. we will also work with laurent polynomials @xmath12 with support @xmath8 and coefficients in an algebraically closed field @xmath0 of characteristic @xmath1 , that is @xmath73.\ ] ]
we will assume that the field @xmath13 is provided with a rank - one non - archimedean valuation @xmath74 , and that the residue field @xmath75 of @xmath13 is also of characteristic zero .
the _ tropicalization _ of a non zero polynomial @xmath12 as in is the tropical polynomial @xmath76 when the valuation group is not the whole of @xmath24 , we will suppose that the coefficients @xmath77 of a tropical polynomial @xmath78 or a tropical point @xmath3 that we want to lift _ lie in the image _ of the valuation map . to accompany our notions in the classical and tropical settings ,
the elements of @xmath79 and @xmath80 will be denoted systematically by the letters @xmath81 , @xmath82 and the elements of @xmath83 and @xmath84 by the letters @xmath85 .
we will denote by @xmath86 a fixed element of @xmath13 of valuation one .
the elements of @xmath87 with be denoted by the letter @xmath64 .
we introduce now the notion of singular point of a tropical hypersurface .
[ def : tropical_singularity ] let @xmath88 as before .
let @xmath89 $ ] be a tropical polynomial .
let @xmath3 be a point in the tropical hypersurface @xmath90 .
then , @xmath3 is a _ singular point _ of @xmath90 if there is a polynomial @xmath91 $ ] and a point @xmath92 such that @xmath93 , @xmath94 and @xmath95 is a singular point of the algebraic hypersurface @xmath96 defined by @xmath12 .
if @xmath90 has a singular point , we call it a _ singular tropical hypersurface_. for instance , if @xmath97 , with @xmath98 , and @xmath99 , then @xmath100 is always a singular point of @xmath90 since for all @xmath101 there exist univariate polynomials of degree @xmath101 with coefficients of valuation @xmath1 and multiple roots with valuation @xmath1 ( just consider @xmath102 which has a multiple root at @xmath22 ) .
let @xmath103 be an _ integral affine function _ on @xmath56 , @xmath104 where @xmath105 and @xmath106 .
the euler derivative of a tropical polynomial @xmath43 with support in @xmath8 with respect to @xmath107 is defined as follows .
[ def : trop_derivative ] let @xmath108 and @xmath109 be an integral affine function .
the _ euler derivative _ of @xmath43 with respect to @xmath107 is the tropical polynomial @xmath110 we also have the standard euler derivative of classical laurent polynomials .
[ def : euler_derivative ] let @xmath111 $ ] and @xmath109 be an integral affine function .
we associate to @xmath107 the euler vector field @xmath112 , where @xmath113 for all @xmath114 .
the _ euler derivative _ of @xmath12 with respect to @xmath107 is the polynomial @xmath115 it is clear that for any singular point @xmath116 of @xmath96 , it holds that @xmath117 for all integral affine functions @xmath107 .
note that if @xmath107 is the constant function @xmath22 , then @xmath118 .
we relate the derivative of @xmath12 with respect to @xmath107 with the derivative with respect to @xmath107 of its tropicalization .
[ lem : lff ] given a tropical polynomial @xmath43 with support @xmath8 and an integral affine function @xmath107 , the equality @xmath119 holds for any polynomial @xmath12 with support @xmath8 such that @xmath120 .
take any @xmath12 with @xmath121 .
note that the euler derivative of @xmath12 with respect to @xmath107 equals @xmath122 from our assumption that the residue field of @xmath13 is also of characteristic zero , it follows that @xmath123 whenever @xmath124 and @xmath125 otherwise .
the result is then a direct consequence of definition [ def : trop_derivative ] of the euler derivative with respect to @xmath107 in the tropical context .
as @xmath8 is finite , the set @xmath126 , with @xmath107 ranging over all possible integer affine linear functions , is finite for any @xmath43 with support @xmath8 .
[ ex : nsconic ] consider the tropical conic @xmath127 .
let @xmath128 and let @xmath129 @xmath130 be any polynomial with tropicalization @xmath43 .
the associated @xmath8 discriminant @xmath10 equals @xmath131 of the determinant of the matrix @xmath132 which is non zero since @xmath133 is the only term in the expansion of the determinant with lowest valuation @xmath1 .
thus , as one could expect , @xmath90 is a non singular tropical hypersurface according to definition [ def : tropical_singularity ] .
it is straightforward to verify that @xmath134 , @xmath135 , and @xmath136 , @xmath137 , correspond to the standard partial derivatives .
note that @xmath138 is non singular but it also lies in the intersection of the tropical hypersurfaces @xmath139 and @xmath140 .
consider now the affine form @xmath141 .
then , @xmath142 and @xmath143 does not lie in the associated tropical hypersurface @xmath144 since the minimum of the linear forms associated to the @xmath145 terms is attained only once at @xmath3 .
the main result in this section is theorem [ teo : singular_point_by_derivatives ] , which characterizes singular tropical hypersurfaces ( with a given support ) in terms of tropical euler derivatives . as we saw in example
[ ex : nsconic ] , it is not enough to consider the @xmath146 euler derivatives corresponding to the coordinate axes .
it is not difficult to solve this problem by appealing to the notion of a tropical basis @xcite , which we now recall .
[ def : tropical_basis ] let @xmath147 $ ] be an ideal . then
, @xmath148 consists of all those weights @xmath149 which satisfy the following : @xmath150 for every nonzero @xmath151 . by @xcite , @xmath148 coincides with @xmath152 , that is with ( the closure of ) the image under the valuation map of the zeros @xmath153 of @xmath154 in the torus @xmath46 .
a tropical basis of @xmath154 is a finite set of polynomials @xmath155 generating @xmath154 such that @xmath156 . given a finite lattice set @xmath88 with @xmath53 elements , we will identify in what follows the space of polynomials with coefficients in @xmath0 and support @xmath8 with @xmath157 .
denote by @xmath158 the point @xmath159 .
the subvariety @xmath160 of polynomials with support @xmath8 and a singularity at @xmath158 is a linear space .
its closure in @xmath161 equals the dual space to the tangent space at the point @xmath158 of @xmath20 .
see the discussion of this space and the following results in section [ sec : bergman ] .
denote by @xmath162 the set of integer affine functions @xmath163 such that @xmath164 and @xmath165 .
[ prop : tbh1 ] let @xmath166 be variables .
the finite set of tropical linear polynomials @xmath167 is a tropical basis of @xmath168 .
let @xmath169 $ ] be the generic polynomial with support @xmath8 .
note that as @xmath107 runs over all integer affine functions , the euler derivatives @xmath170 are precisely _ all _ integral linear combination of @xmath171 .
@xmath172 is the linear space ( in the variables @xmath173 ) defined by the linear equations @xmath174 , @xmath175 , @xmath176 .
we know that the linear forms vanishing on @xmath172 form a tropical basis of @xmath172 @xcite , and it is enough to consider linear forms with rational ( and a fortiori , integer ) entries .
now , @xmath177 . moreover , by @xcite , the set of linear forms in @xmath172 that have minimal support define a tropical basis of @xmath168 .
this set corresponds to the affine functions @xmath107 such that @xmath178 spans an affine space of maximal dimension @xmath179 .
we have defined a tropical basis of the set of polynomials with a singularity at @xmath158 .
if we have another point @xmath180 , we can easily provide a tropical basis of the variety @xmath181 of hypersurfaces with a singular point at @xmath182 by considering a diagonal change of coordinates .
explicitly , if @xmath183 is a laurent polynomial with coefficients in @xmath184 with a singularity at @xmath158 and such that @xmath185 , then the polynomial @xmath186 has a singularity at @xmath187 and @xmath188 .
we can easily deduce the following .
[ prop : tbh ] let @xmath88 with @xmath54-linear span @xmath189 .
as before , identify @xmath157 with the space of polynomials with support @xmath8 .
consider the incidence variety @xmath190 .
let @xmath191 be the generic polynomial with support @xmath8 , where @xmath192 and @xmath193 are variables .
then the finite set @xmath194 is a tropical basis of @xmath195 .
we have now the tools to prove the following tropical characterization of singular tropical hypersurfaces with fixed support .
[ teo : singular_point_by_derivatives ] let @xmath108 be a tropical polynomial with support @xmath8 .
let @xmath196 be a point in the hypersurface defined by @xmath43 .
then , @xmath3 is a singular point of @xmath90 if and only if @xmath197 for all @xmath107 .
thus , @xmath43 defines a singular tropical hypersurface if and only if @xmath198 this intersection is given by a finite number of euler derivatives of @xmath43 ; for instance , we can take only the affine linear functions @xmath199 defined before proposition [ prop : tbh1 ] .
one implication is trivial .
if @xmath3 is a singular point of @xmath90 there exists a polynomial @xmath183 , @xmath93 with a singularity at a point @xmath200 with @xmath94 .
then , @xmath201 for all @xmath107 , and so @xmath202 for all @xmath107 .
for the converse , let @xmath3 be a point in @xmath203 .
in particular , @xmath196 . then , for any integer affine function @xmath107 , the minimum @xmath204 is attained at least twice .
this happens if and only if for all @xmath107 the point @xmath205 .
it follows from proposition [ prop : tbh ] that @xmath206 belongs to the incidence variety @xmath195 .
so , by kapranov s theorem @xcite , there is a point @xmath207 such that @xmath12 is an algebraic polynomial with support in @xmath8 and a singularity at @xmath200 such that @xmath121 and @xmath208 .
we present two examples that illustrate theorem [ teo : singular_point_by_derivatives ] .
[ ex : first_example ] let @xmath209 .
consider the tropical polynomial @xmath210 .
let us compute its singular points .
let @xmath211 , @xmath212 , @xmath213 , @xmath214 .
the intersection set of these three curves is the segment @xmath55 whose ends are @xmath215 .
consider now @xmath216 that contains all the monomials _ dual _ to @xmath55 ( cf .
the beginning of section [ sec : bergman ] for a more precise explanation of this duality ) .
then , @xmath217 .
the intersection of this tropical curve with the segment @xmath55 is the set of points @xmath218 ( see figure [ fig : derivative1 ] ) .
let us check that these two points are valid singular points .
the polynomial @xmath219 has support @xmath8 .
it defines a curve with a singularity at @xmath220 and @xmath221 . on the other side
, @xmath222 is a polynomial with support @xmath8 .
@xmath223 defines a curve with a singularity at @xmath224 and such that also @xmath225 .
it is not difficult to see that it is not possible to find a single polynomial @xmath12 with two singular points with valuations @xmath226 and @xmath227 .
( bold ) and @xmath228 ( pointed ) in example [ ex : first_example],scaledwidth=40.0% ] [ ex : laface ] let @xmath209 be as in example [ ex : first_example ] and let @xmath229 , so that @xmath230 are all the lattice points in the convex hull of @xmath8 . consider the tropical polynomial @xmath231 with support in @xmath230 .
that is , the coefficients of the @xmath232 new points are given by interpolation of the linear functions defining the subdivision associated to the polynomial @xmath43 in the previous example .
note that all points in @xmath230 are thus marked .
it is easy to check that all points in @xmath233 are singular .
indeed , in this case @xmath234 can be lifted to the polynomial @xmath235 .
we now present theorem [ teo : singular_point_by_derivatives ] into action in a defective example , where the tropical @xmath8-discriminant can be explicitly computed .
[ ex : defective ] let @xmath236 be the configuration @xmath237 .
thus , @xmath8 is the union of two one dimensional circuits and the convex hull of @xmath8 is the lattice tetrahedron with vertices @xmath238 .
note that @xmath8 does not contain any circuit of full dimension @xmath232 .
the zero set of any integer affine function @xmath107 such that the affine span of @xmath239 is equal to @xmath240 , consists of one of the circuits plus one more point .
consider a tropical polynomial @xmath241 with support in @xmath8 .
thus , there exists a singular point @xmath242 if and only if @xmath243 this corresponds to the fact that this configuration is self - dual ; indeed , the dual variety @xmath244 has ( projective ) dimension @xmath232 , it is isomorphic to the toric variety @xmath20 and it is cut out by the binomials @xmath245 , @xmath246 , where @xmath247 are homogeneous coordinates in @xmath248 .
the tropicalization of this binomial ideal is the rowspan of the associated matrix @xmath8 in @xmath249 , which is defined by the equations .
given a tropical polynomial @xmath250 with support @xmath8 , most of the ( finite ) euler derivatives @xmath251 do not provide relevant information to detect singular points of @xmath90 . in this section
we give further conditions and characterizations to detect singular points .
we need to recall the following duality @xcite .
the vector of coefficients @xmath252 of @xmath43 defines a _ coherent marked subdivision _ @xmath253 of the convex hull @xmath48 of @xmath8 .
that is , @xmath25 defines a collection of subsets of @xmath8 ( called _ marked cells _ ) which are in one - to - one correspondence with the domains of linearity of the affine function cutting the faces of the lower convex hull of the set of lifted points @xmath254 in @xmath255 .
assume that a lower face @xmath256 equals the graph of an affine function @xmath257 .
the corresponding marked cell @xmath258 of the subdivision of @xmath48 is the subset of @xmath8 of all those @xmath64 for which @xmath259 .
the marked subdivision @xmath253 is combinatorially dual to the _ marked _ tropical variety @xmath90 .
as we saw , this is a polyhedral complex which is a union of dual cells @xmath260 , where we also record the information of the dual cell @xmath258 , and not only of the geometric information of the vertices of @xmath258 .
more explicitly , the dual cell @xmath68 in @xmath90 of a given cell @xmath72 of @xmath253 equals the closure of the union of the points @xmath261 for all ways of writing @xmath262 , and we also record the information of all the points in @xmath72 , that is , of _ all the points at which the minimum @xmath263 is attained _ for any point @xmath261 in the relative interior of @xmath68 ( which is the _ marking _ of the cell ) .
the sum of the dimensions of a pair of dual cells is @xmath146 . in particular ,
vertices of @xmath90 correspond to marked cells of @xmath253 of maximal dimension @xmath146 .
we now prove that @xmath253 is a ( coherent ) triangulation , then the tropical hypersurface associated to @xmath43 is non singular , as expected .
as we will see , the converse to this statement is not true and involves a complicated combinatorial study . recall that a point configuration is a pyramid , if all but one of its points lie in an affine hyperplane .
[ lem : punto_regular ] let @xmath264 lying in the relative interior of a cell @xmath68 such that the dual cell @xmath72 in @xmath253 is a pyramid . then , @xmath3 is non singular . in particular ,
if @xmath253 is a coherent triangulation , then the tropical hypersurface @xmath49 is non singular .
if @xmath72 is a pyramid , let @xmath107 be a linear functional such that @xmath265 intersects @xmath72 in a facet and leaves out one point .
this means that the minimum of @xmath251 at @xmath3 is attained at exactly one monomial .
hence , @xmath266 and @xmath3 is not a singular point .
[ ex : triang2 ] consider the same configuration @xmath8 of example [ ex : nsconic ] , that is , the six lattice points of the @xmath240-simplex in the plane and the tropical conic @xmath267 .
then the associated marked subdivision has only one cell @xmath268 and it is not singular by lemma [ lem : punto_regular ] , even if there are points in @xmath8 that do not occur in the subdivision and this single cell has lattice volume bigger than one . also , it is straightforward to check that any polynomial @xmath269 @xmath270 $ ] with the given valuations is non singular , because this time the only term in the expansion of the determinant of the matrix with smallest valuation @xmath1 , is the diagonal term @xmath271 .
[ ex : nontriang2 ] consider the tropical polynomial @xmath272 .
we read the support @xmath8 from @xmath43 .
the coefficients of @xmath43 induce the coherent marked subdivision depicted in figure [ fig : cayley2rectas ] , which has two cells of dimension @xmath232 and it is not a triangulation .
, scaledwidth=40.0% ] one of these top dimensional cells is a unimodular 3-simplex .
the second top dimensional cell contains a circuit @xmath273 of dimension @xmath274 and it is a pyramid over the point @xmath275 .
the affine integer function @xmath276 verifies that @xmath277 .
computing @xmath251 , we check that @xmath43 defines a _ non singular _ tropical surface .
we now analyze some further conditions that a point @xmath196 must satisfy in order to be a tropical singular point .
[ teo : char_punto_singular ] let @xmath108 be a tropical polynomial and @xmath278 lying in the interior of a cell @xmath68 .
then , @xmath3 is a singular point if and only if the dual cell @xmath72 is not a pyramid and we have that @xmath279 for all affine linear functions @xmath107 such that @xmath280 and @xmath281 .
so , in the particular case of a vertex @xmath3 of @xmath90 , @xmath3 is singular if and only if @xmath72 is not a pyramid .
if @xmath3 is a singular point , @xmath72 is not a pyramid by lemma [ lem : punto_regular ] .
as @xmath282 for any @xmath107 by definition , in particular this happens for those @xmath107 of the form described in the hypotheses .
suppose now that @xmath3 is not a singular point and let @xmath283 be an affine integer function such that @xmath284 .
let @xmath285 be the unique point of @xmath8 at which @xmath286 is attained .
then , if @xmath72 is not contained in @xmath287 , we have that @xmath288 and it is the unique point of @xmath72 outside @xmath287 , and so @xmath72 is a pyramid .
otherwise , take any integer hyperplane @xmath265 such that @xmath289 , @xmath290 spans an affine space of dimension @xmath179 and @xmath291 .
for any such @xmath107 we have that @xmath292 , as wanted . as a consequence
, we can easily describe the polynomials that define singular hypersurfaces in the case of @xmath22 and @xmath240 variables . recall
that , if @xmath8 is not defective , then @xmath293 is a subfan of the secondary fan of @xmath8 . in the simplest case of one variable , @xmath294
, it holds that @xmath295 equals the union of the non top dimensional cones in the secondary fan ( since the only proper faces of @xmath8 are vertices ) .
hence , a univariate polynomial is singular if and only if the induced marked subdivision is not a triangulation . with our notation , this is a simple case of theorem [ teo : char_punto_singular ] , because all circuits of @xmath8 are of maximal dimension @xmath22 .
the following result , in the smooth case , appears in ( * ? ? ?
[ teo : singular_dim_1_and_2 ] let @xmath296 with @xmath53 elements .
suppose @xmath297 induces a coherent marked subdivision @xmath253 in @xmath8 that is not a triangulation .
then @xmath25 is in @xmath298 ( equivalently , the polynomial @xmath299 defines a singular tropical hypersurface ) in exactly the following situations : * there exists a marked cell of @xmath253 which contains a circuit of dimension @xmath240 . *
[ case : ii ] all circuits contained in a cell of @xmath253 have affine dimension @xmath22 and there exists a marked cell @xmath72 of @xmath253 of dimension @xmath22 and cardinality @xmath300 with the following property : let @xmath107 be an integer affine function such that @xmath301 . then , @xmath302 .
the first item is contained in theorem [ teo : char_punto_singular ] .
with respect to the second item , note that @xmath72 contains a circuit @xmath303 ( of dimension @xmath22 ) and for any integral affine function @xmath107 , @xmath277 if and only if @xmath304 .
the result follows again by theorem [ teo : char_punto_singular ] . in case item
ii ) of proposition [ teo : singular_dim_1_and_2 ] holds , @xmath305 if and only if there is a cell @xmath306 of dimension @xmath240 containing @xmath72 , such that @xmath307 consists of a single point @xmath308 and , assuming @xmath309 , there exists another point @xmath310 with @xmath311 .
this is a particular case of the following more general result .
recall that we always assume that the convex hull of our exponent set @xmath8 is full dimensional .
[ prop : anotherpt ] let @xmath88 .
let @xmath312 such that @xmath253 contains a top dimensional cell @xmath306 which contains a circuit @xmath303 of dimension @xmath179 and it is a pyramid over a point @xmath313 .
let @xmath107 be an affine integer function such that @xmath314 and @xmath315 .
then , there exists a singular point @xmath316 with @xmath317 if and only if there exists another point @xmath318 not in @xmath306 such that @xmath319 .
in particular , if @xmath303 intersects the interior of @xmath48 , then @xmath320 is singular .
we can assume that @xmath321 , with @xmath322 coprime . to make the notation easier
, we apply an invertible affine linear transformation to our configuration @xmath8 so that @xmath323 .
denote by @xmath324 the linear form which interpolates @xmath25 over the cell @xmath306 , that is @xmath325 for all @xmath326 and @xmath327 for all @xmath328 .
thus , @xmath329 defined by @xmath330 defines the same marked subdivision .
so we can assume that @xmath331 for all @xmath332 and @xmath333 otherwise .
therefore , @xmath334 is the vertex of @xmath335 dual to @xmath306 , which is not singular since it does not lie in @xmath336 .
there will be a singular point @xmath337 in @xmath338 with @xmath339 if and only if there exists @xmath340 and two points @xmath341 in @xmath8 such that @xmath342 for all @xmath32 with @xmath343 .
note that as @xmath306 is a pyramid over @xmath313 , for any point @xmath344 in @xmath8 for which @xmath345 it holds that @xmath344 is not in @xmath306 , or equivalently , @xmath346 .
assume first that there is a point @xmath64 in @xmath347 with @xmath348 and let @xmath344 with these properties and such that moreover @xmath349 .
then , it is enough to take @xmath350 and @xmath351 .
reciprocally , assume there exists a singular point @xmath352 with @xmath340 . as @xmath353 , there exist two points @xmath354 such that @xmath355 .
assume @xmath356 .
then , @xmath357
. therefore , @xmath358 , as wanted .
the condition that @xmath303 intersects the interior of @xmath48 guarantees the existence of a point @xmath310 with @xmath358 . note that the point @xmath344 in the statement of proposition [ prop : anotherpt ] does not need to belong to any cell in @xmath253 .
in this section , we relate our definitions to the results and definitions in @xcite . as before ,
@xmath359 $ ] denotes a tropical polynomial with support @xmath8 .
[ def : weight_class ] let @xmath3 be in the interior of a cell @xmath360 . we define the _ flag _ of @xmath43 with respect to @xmath3 as the flag of subsets @xmath361 of @xmath8 defined inductively by : @xmath362 , @xmath363 , and for any @xmath364 : @xmath365 is the subset of @xmath366 where the tropical polynomial @xmath367 attains its minimum at @xmath3 .
the _ weight class _ of the flag @xmath368 are all the points @xmath369 for which @xmath370 theorems [ teo : singular_point_by_derivatives ] and [ teo : char_punto_singular ] provide an algorithm to decide if @xmath278 is singular or not , which is similar to the method presented in @xcite but which works without any restrictive hypothesis on @xmath8 .
the algorithm returns an @xmath107 such that @xmath371 or @xmath3 is a singular point " .
first , we compute @xmath372
. if @xmath72 is a pyramid , there exists @xmath373 such that @xmath374 and we can compute an @xmath107 defining the facet @xmath375 of @xmath376 , which verifies @xmath292 . if this is not the case and the dimension of @xmath377 , we compute @xmath378 and we iterate the procedure .
we stop when we find an @xmath107 that certifies that @xmath3 is not singular or when @xmath379 spans an affine dimension @xmath146 , in which case @xmath3 is singular .
number the elements @xmath380 of @xmath8 and call @xmath381 matrix with these columns .
let @xmath30 , as in the introduction , be the integer matrix with columns @xmath382 .
thus , the vector @xmath383 lies in the row span of @xmath30 .
in fact , as the @xmath8-discriminant is an affine invariant of the configuration @xmath8 , we could assume without loss of generality that @xmath8 has this property , but we prefer to point out the fact the we are interested in affine properties of the configuration @xmath8 , equivalent to linear properties of @xmath30 .
let @xmath384 be an affine linear function .
we can associate to @xmath107 the linear form @xmath385 .
then , the support of @xmath386 is precisely @xmath387 .
moreover , the coefficient vector @xmath388 lies in the row span of @xmath30 , as it is obtained as the product @xmath389 , and all linear forms in the row span of @xmath30 ( which span the ideal of @xmath390 ) , are of this form .
let @xmath391 be a gale dual configuration of @xmath30 .
the linear forms in the row span of @xmath30 with minimal support correspond to the circuits in @xmath392 and to the affine linear forms @xmath107 such that @xmath393 .
denote by @xmath394 ( tropical ) variables .
the tropicalization @xmath395 equals : @xmath396 we recover the fact that @xmath172 is an incarnation of @xmath390 and so @xmath168 equals the co - bergman fan @xmath31 ( cf .
proposition [ prop : tbh1 ] ) .
the flag of sets @xmath397 and the weight classes in definition [ def : weight_class ] coincide for instance with those occurring in ( * ? ? ?
* page 3 ) .
our previous algorithm can be modified to decide whether @xmath398 contains a singular point , that is , to decide whether @xmath25 lies in the tropical @xmath8-discriminant , and in this case , to compute all the singular points .
just notice that , as weight classes induce a fine subdivision on the co - bergman fan @xmath399 , they also induce a finer polyhedral subdivision of @xmath90 .
two points @xmath3 and @xmath400 belong to the relative interior of the same cell of the fine subdivision if and only if @xmath401 belong to the same weight class .
if @xmath72 is a cell of the fine subdivision of @xmath90 , then either every point of @xmath72 is singular or all points are regular .
since the number of cells in this subdivision is finite and computable , we can derive an algorithm to compute all singular points of @xmath90 that uses this information .
[ prop : singcells ] the ( finitely many ) weight classes associated to a tropical polynomial @xmath402 with support @xmath8 , are relatively open polyhedral cells which refine the polyhedral structure of @xmath90 dual to the marked coherent subdivision @xmath253 . if @xmath403 is a cell in this new subdivision , then all points in @xmath403 are singular or all of them are regular .
the previous algorithm applied to any of the points in @xmath403 , allows us to decide if @xmath403 is a set of singular or regular points .
we can thus reprove ( * ? ? ?
* theorem 1.1 ) : a point @xmath404 lies in the tropicalization @xmath405 of the @xmath8-discriminant if and only if there exists a singular point @xmath406 .
this happens if and only if the @xmath53-th dimensional vector @xmath407 defined by the equalities @xmath408 lies in @xmath399 , by the characterization given in theorem [ teo : singular_point_by_derivatives ] ( expressed by the previous algorithm ) .
equivalently , if and only if @xmath409 with @xmath38 in @xmath399 and @xmath410 .
that is , if and only if @xmath25 lies in the minkowski sum of @xmath399 and the row span of @xmath8 , which equals the minkowski sum of @xmath399 and the row span of @xmath30 , since the vector @xmath411 .
it follows that if @xmath412 , the singular points of @xmath413 are those @xmath3 which occur in a decomposition of the form .
we end this section with some examples that exhibit different interesting features of these objects .
the next example shows a tropical polynomial @xmath43 whose coefficients lie in a codimension one cone of the secondary fan of @xmath8 , for which @xmath90 has two singular points .
[ ex : example_discriminant ] let @xmath414 @xmath415 . the subdivision induced by @xmath43 in its newton polygon is a triangulation except for the circuit of exponents in @xmath416 .
the @xmath8-discriminant of the support of @xmath43 is ( with the obvious meaning of the variables @xmath417 ) @xmath418 @xmath419 @xmath420 @xmath421 @xmath422 @xmath423 @xmath424 @xmath425 @xmath426 @xmath427 @xmath428 @xmath429 @xmath430 @xmath431 @xmath432 @xmath433 @xmath434 @xmath435 @xmath436 @xmath437 @xmath438 @xmath439 @xmath440 @xmath441 @xmath442 @xmath443 @xmath444 @xmath445 @xmath446 @xmath447 @xmath448 @xmath449 @xmath450 @xmath451 @xmath452 @xmath453 @xmath454 @xmath455 @xmath456 @xmath457 @xmath458 @xmath459 @xmath460 @xmath461 @xmath462 @xmath463 @xmath464 @xmath465 @xmath466 @xmath467 @xmath468 @xmath469 @xmath470 @xmath471 @xmath472 @xmath473 @xmath459 @xmath474 @xmath475 @xmath467 @xmath476 @xmath477 @xmath478 @xmath479 @xmath480 @xmath481 .
the minimum valuation of the terms in the @xmath8-discriminant is attained for any choice of coefficients @xmath417 with valuations prescribed by the coefficients of @xmath43 , in the five underlined monomials of the @xmath8-discriminant .
three of these monomials @xmath482 lie in the convex hull of the other two @xmath483 .
hence , the exponents of the monomials of the @xmath8-discriminant where the minimum is attained lie on an edge , and the vector of coefficients of @xmath43 belongs to a maximal cell of the tropicalization of the @xmath8-discriminantal variety .
the singular points of this curve are @xmath484 ( see figure [ fig : example_discriminant ] ) .
two lifts of the curve and the singular point are : @xmath485 with a singularity at @xmath486 , and @xmath487 with singularity at @xmath488 .
( in bold ) and @xmath489 ( pointed),scaledwidth=40.0% ] our next example shows that for two vectors of coefficients inducing the same coherent subdivision of @xmath8 , the associated flags need not coincide . [ ex : howdareformed ] let @xmath490 @xmath491 @xmath492 @xmath493 @xmath494 and @xmath495 , with @xmath496 arbitrary . in this case , @xmath497 defines the curve given by @xmath498 .
the marked subdivision @xmath499 induced by any @xmath497 contains three maximal cells : @xmath500 , @xmath501 , @xmath502 .
we claim that all these curves are singular , with a singular point in the cell dual to the marked edge @xmath503 .
however , as we will see , the number and locations of the singular points vary .
cell @xmath504 is dual to the point @xmath226 in the curve and @xmath505 is dual to the point @xmath506 . by theorem [ teo : singular_dim_1_and_2 ]
there is a singularity if there is a point in the segment @xmath507 $ ] that also belongs to the partial derivative @xmath508 . in the segment
, @xmath509 attains its minimum at @xmath226 on the linear form associated to @xmath510 and @xmath509 attains its minimum at @xmath511 on the linear form associated to @xmath512 . since @xmath509 is a continuous function , there must be a point @xmath513 where the minimum of @xmath509 is attained twice , so this point will be a singularity of @xmath43 ( cf .
this reasoning works for any hypersurface in dimension @xmath146 with a circuit in the interior of @xmath8 of dimension @xmath179 .
* if @xmath514 there is a singular point at @xmath515 , the flag with respect to @xmath3 is : @xmath516 @xmath517 @xmath518 @xmath519 . *
if @xmath520 there is a singular point at @xmath515 , the flag with respect to @xmath3 is : @xmath521 * if @xmath522 we get two different singular points : * * @xmath523 with flag with respect to @xmath3 : + @xmath524 . * * @xmath525 with flag with respect to @xmath3 : + @xmath526 . ,
title="fig:",scaledwidth=90.0% ] @xmath527 ( left ) , @xmath528 ( center ) , @xmath529 ( right ) thus , we can take different values of @xmath530 such that , while keeping @xmath531 invariant , it is not possible to find singular points @xmath532 in @xmath533 for which the flags coincide @xmath534 for all @xmath364 .
it is worthwhile to note that the set of singular points in a hypersurface is not , in general , a tropical variety .
let @xmath535 represents a tropical conic .
so , if it is singular , it is a pair of lines .
it happens that this conic is the union of the lines @xmath536 and @xmath537 .
the intersection of these two lines is the ray @xmath538 , @xmath539 .
this is not a tropical variety .
however , any of its points is a valid singular point of the conic .
we can take any point @xmath3 in the intersection set and lift the whole configuration , as it is an acyclic configuration ( see @xcite ) .
in fact , for the point @xmath540 , @xmath541 we can take the lift @xmath542 that has a singularity at @xmath543 .
we fix two finite subsets @xmath544 of @xmath545 with @xmath546 and such that @xmath547 . in this section
we define and study non transversal intersections of two tropical curves associated to tropical polynomials with respective supports @xmath548 . as in definition
[ def : tropical_singularity ] , we will say that the intersection is non transverse when it comes from the tropicalization of a classical non transverse intersection of two curves . in this case , we will see that the standard definitions do not have a straightforward translation to the tropical setting .
[ def : nonstransversal_intersection ] let @xmath549 , @xmath550 be two tropical polynomials in @xmath551 $ ] .
let @xmath3 be a point in the intersection of the tropical curves @xmath552 .
then @xmath3 is a _ non - transversal ( or non - smooth ) intersection point _ of @xmath90 and @xmath553 if there exists two laurent polynomials @xmath554 in @xmath555 $ ] , with respective supports @xmath544 and a point @xmath556 which is a non - transversal intersection of @xmath14 and @xmath557 such that @xmath121 ( that is , @xmath558 for all @xmath559 ) , @xmath560 and @xmath561 .
recall that @xmath562 is a non - transversal intersection point if moreover the jacobian @xmath563 vanishes at @xmath200 .
this jacobian is the determinant of the jacobian matrix ( or the matrix of the differential of the map @xmath564 ) , and also of its transpose @xmath565 .
the condition that @xmath566 is obtained from elimination of variables from the following equivalent fact : @xmath14 and @xmath557 intersect non transversally at @xmath200 if and only if their tangent lines coincide , or equivalently , the matrix @xmath567 has a non trivial kernel , that is , there exists a non trivial vector @xmath568 which is a solution of the system @xmath569 given two laurent polynomials @xmath570 in @xmath555 $ ] , with respective supports @xmath571 , the _
mixed discriminant _ of @xmath12 and @xmath572 is the @xmath8-discriminant associated to the polynomial @xmath573 $ ] with support in the cayley configuration ( cf .
@xcite ) @xmath574 in fact , this is a three dimensional configuration lying in the plane defined by the sum of the two first coordinates equal to @xmath22 .
this mixed discriminant vanishes at the vectors of coefficients @xmath575 whenever @xmath12 and @xmath572 have a non - transversal intersection at a point @xmath576 for which the system has a solution @xmath577 . in particular , note that horizontal and vertical tangents are not necessarily reflected in the mixed discriminant ( cf .
* section 3 ) for a more general definition of discriminants which takes into account different supports ) .
hence , we do not take cover these extremal cases and will only describe the non - transversal intersection points for which the system has a solution @xmath577 .
[ lem : l ] let @xmath578 $ ] be two tropical bivariate polynomials with respective supports @xmath544 .
the tropical plane curves that @xmath43 and @xmath579 define intersect non - transversally at an intersection point @xmath580 if and only if there exists @xmath581 such that @xmath582 belongs to the tropical discriminant associated to the polynomial @xmath583 $ ] with support in the configuration @xmath584 , that is if @xmath585 is a singular point of @xmath586 .
suppose that there is an element @xmath581 such that @xmath589 belongs to the tropical discriminant of @xmath590 .
so , @xmath591 belongs to the incidence variety @xmath195 associated to @xmath592 .
then , there are algebraic polynomials @xmath593 $ ] and a point @xmath594 such that @xmath595 lies in the incidence variety @xmath596 , @xmath121 , @xmath597 , @xmath598 .
hence @xmath599 is a singular point of @xmath600 , and so the partial derivatives must vanish : @xmath601 it follows that @xmath602 and @xmath603 is a non - transversal intersection point of @xmath14 and @xmath557 . to prove the converse ,
suppose @xmath604 is a non transversal intersection point of @xmath43 and @xmath579 .
there exists @xmath605 $ ] with respective supports @xmath544 such that @xmath606 and @xmath607 as in definition [ def : nonstransversal_intersection ] .
then @xmath608 , @xmath609 , and let @xmath568 be as in .
it follows that @xmath610 is a singular point of the surface defined by @xmath611 , so @xmath612 and @xmath613 .
therefore , @xmath614 is a non - transversal intersection point of @xmath590 .
[ ex : twolines ] let @xmath615 and @xmath616 be two tropical lines .
these lines intersect at an infinite number of points ( all the points in the ray @xmath617 . however , two algebraic lines intersect non - transversally if and only if they are the same , and we expect that this also happens in the tropical setting .
since @xmath618 , let us check that they intersect transversally according to our definition .
consider the surface defined by @xmath619 .
both lines have the same support @xmath620 .
the associated mixed subdivision of @xmath621 is precisely the marked subdivision occurring in example [ ex : nontriang2 ] and depicted in figure [ fig : cayley2rectas ] , so @xmath622 is indeed non singular according to lemma [ lem : l ] .
let @xmath623 , @xmath624 .
so , @xmath43 and @xmath579 are in fact univariate polynomials in different variables .
the associated configuration @xmath625 is just the configuration @xmath8 occurring in example [ ex : defective ] . in this case
, the mixed discriminant has codimension bigger than @xmath22 and a point of intersection @xmath626 is non - transversal if and only if @xmath627 is a singular point of @xmath628 and @xmath629 is a singular point of @xmath630 , which is the translation of the fact that @xmath3 is singular if and only if it lies in the rowspan of @xmath8 .
since we are looking for the singular points @xmath631 in @xmath632 , we could use the known tropical basis of the discriminant of this surface to compute them , as in the discussion above .
however , we would like to obtain a method that involves only @xmath43 and @xmath579 and no new variables .
just checking if the jacobian matrix is singular will not work .
let @xmath633 , @xmath634 .
consider the intersection point @xmath635 .
if we would like to use a kind of tropical jacobian matrix , a natural choice would be the matrix when we evaluate at @xmath3 , we get the matrix @xmath636 which is nonsingular because @xmath637
. however , take @xmath638 , @xmath639 , @xmath640 is an intersection point and @xmath121 , @xmath597 , @xmath641 and @xmath642 so the tropical point @xmath226 is a non - transversal intersection point .
the problem is that in general , given a polynomial @xmath643 $ ] and a point @xmath644 , @xmath645 . in our case
@xmath646 is the jacobian @xmath647 and @xmath640 .
we have that @xmath648 , while @xmath649 .
let @xmath650 . @xmath651 and @xmath652 equal @xmath653 and intersect non transversally at any point of @xmath654 .
the toric jacobian @xmath655 is the determinant of the @xmath656 matrix @xmath657 with columns the euler derivatives of @xmath12 and @xmath572 with respect to @xmath658 , and so it is identically zero .
if we consider the matrix @xmath659 obtained by taking the tropicalization of each of the entries of @xmath657 , we get the @xmath656 matrix with two equal rows @xmath660 $ ] .
the tropical determinant of @xmath659 equals the permanent @xmath661 .
if we forget the fact that we have _ twice _ the term @xmath662 , we lose information .
we just get a monomial , which defines an empty tropical curve .
[ teo : non_transversal_intersection_two_vertices ] let @xmath43 , @xmath579 be two tropical bivariate polynomials with respective supports @xmath544 .
let @xmath663 be a tropical point that is a vertex in both @xmath90 and @xmath553 .
then @xmath3 is a non - transversal intersection point of @xmath43 and @xmath579 .
set @xmath664 .
we claim that @xmath582 is a singular point of the tropical surface @xmath586 .
to see this , note that the points in @xmath665 where the minimum in @xmath585 is attained are of the form @xmath666 , for all @xmath64 such that @xmath70 attains its minimum and those of the form @xmath667 , for all @xmath64 where @xmath668 attains its minimum . since @xmath3 is a vertex of both @xmath90 and @xmath553 ,
the minimum of @xmath70 is attained in a @xmath240-dimensional cell @xmath504 in @xmath669 and the minimum of @xmath668 is attained in a @xmath240-dimensional cell @xmath670 in @xmath671 .
let @xmath672 be any integer affine function in three variables , @xmath673
. then @xmath265 can not contain @xmath504 nor @xmath670 , so there are at least two different points of @xmath665 where the minimum of @xmath622 is attained at @xmath585 .
it follows that @xmath674 . in case
@xmath107 is of the form @xmath675 , @xmath265 is disjoint from @xmath665 unless @xmath676 , and in this case either all the monomials corresponding to the points in @xmath677 or @xmath678 occur in the euler derivative with respect to @xmath107 of @xmath679 .
so , by theorem [ teo : singular_point_by_derivatives ] , @xmath585 is a singular point of @xmath622 and @xmath3 is a nontransversal intersection point of @xmath43 and @xmath579 .
[ teo : non_transversal_intersection_one - vertex - and - segment ] let @xmath680 be tropical polynomials with respective supports @xmath544 .
suppose that @xmath3 is a vertex of @xmath90 and lies on a segment or ray of @xmath553 .
moreover , suppose that , locally , @xmath25 is the only local intersection point in @xmath552 .
then @xmath3 is a non - transversal intersection point of @xmath43 and @xmath579 .
the minimum of @xmath43 at @xmath3 is attained in at least three non - collinear monomials associated to the points @xmath503 and the minimum in @xmath579 is attained in at least @xmath240 monomials associated to the points @xmath681 .
take again @xmath682 .
then , @xmath683 of @xmath684 is attained at the monomials corresponding to @xmath685 . by construction ,
@xmath686 , @xmath687 is a simplex .
since the intersection of @xmath90 and @xmath688 around @xmath3 is just the point @xmath3 , it follows that the line through @xmath689 and @xmath690 is not coplanar with any line generated by two different points among @xmath691 . in other words
, @xmath403 is a circuit of dimension @xmath232 . by theorem [ teo : char_punto_singular ]
, @xmath692 is a singular point of @xmath622 and thus @xmath3 is a non - transversal intersection point of @xmath43 and @xmath579 .
[ lem:4types ] let @xmath697 be an intersection point of @xmath90 and @xmath698 and let @xmath283 be an integer affine function in @xmath232 variables .
there exists @xmath699 such that @xmath582 is a point in the tropical surface @xmath700 .
in fact , keeping the notations of the previous lemma , we can make a finer classification of the affine linear forms @xmath283 into @xmath145 types and describe all possible choices of @xmath702 in each case .
we say that @xmath283 is of type 1 if the minimum of @xmath251 at @xmath3 is attained at least twice and exactly once in @xmath703 . in this case , we can take any @xmath704 .
we say that @xmath283 is of type 2 if instead the minimum at @xmath3 is attained at least twice in @xmath705 and once in @xmath251 . in this case , we can take any @xmath706 .
the integer affine linear function @xmath283 is said to be of type 3 if the minimum is attained at least twice in both @xmath251 and @xmath703 . in this case
, we can take any @xmath707 .
finally , we say that @xmath283 is of type 4 if the minimum is attained once in both @xmath251 and @xmath703 . in this case
, the only possible choice is @xmath708 .
[ teo : non_transversal_intersection_nontrivial_case ] let @xmath43 , @xmath579 be two tropical bivariate polynomials with respective supports @xmath544 and @xmath3 an intersection point of @xmath90 and @xmath698 .
consider the set @xmath709 of all @xmath145-tuples where @xmath710 are integer affine functions in two variables and @xmath711 , for which @xmath3 belongs to at _ least one _ of the tropical curves associated to @xmath712 , @xmath713 , @xmath714 , @xmath715 .
let @xmath716 be the set of @xmath145-tuples for which @xmath3 does _ not _ belong to any of these four tropical curves . by theorem [ teo : singular_point_by_derivatives ] ,
a point @xmath697 is a non - transversal intersection point if and only if there is an @xmath702 such that @xmath589 belongs to all the partial derivatives @xmath721 .
assume first that @xmath722 is a non - transversal intersection point and let @xmath581 such that @xmath589 belongs to all the partial derivatives @xmath723 .
let @xmath710 be integer affine functions of @xmath240 variables and @xmath724 .
call @xmath725 .
then , by , we have that the minimum at @xmath726 is attained twice for @xmath727 .
it is not difficult to see that the two conditions in the statement of the theorem hold , separating the arguments for the different types of functions @xmath283 .
reciprocally , assume that these conditions hold for all @xmath145-tuples of the form @xmath728 and let @xmath3 be an intersection point .
then , any @xmath589 belongs to the tropical surfaces defined by @xmath622 and @xmath729 for all constants @xmath730 .
hence , it remains to check the euler derivatives when @xmath283 for which @xmath731 .
let @xmath732 be the maximum of @xmath733 for all @xmath734 of type 1 .
if there are no @xmath283 of type 1 set @xmath735 .
let @xmath736 be the minimum of @xmath737 for all @xmath734 of type 2 .
if there are no @xmath283 of type 2 , set @xmath738 . finally , let @xmath739 be the integral affine functions of type 4 whose supports represent all possible supports with this type , and write @xmath740 .
call @xmath741 .
then , it is clear that there exists an @xmath702 such that @xmath692 belongs to all partial derivatives if and only if @xmath742 we have to translate these conditions into jacobian like equations .
let @xmath743 , @xmath744 be two integral affine linear functions .
if @xmath745 and @xmath746 are of the same type 1 or 2 , or if one of the types is 3 , then it always happens that @xmath3 is a point in the curve @xmath747 , because we are adding two polynomials for which @xmath3 is a `` zero '' .
if @xmath748 is of type @xmath22 and @xmath749 is of type 2 then : @xmath750 and @xmath751 attain its minimum at least twice in @xmath3 , and @xmath752 and @xmath753 attains its minimum once in @xmath3 .
hence @xmath3 is a point in @xmath754 if and only if @xmath755 .
this is equivalent to @xmath756 .
so , @xmath757 . this happens for every pair of integral affine functions @xmath745 of type 1 and @xmath746 of type 2 if and only if @xmath758 .
let @xmath748 be of type 1 and @xmath749 of type 4 .
then @xmath3 is a point in @xmath754 for all @xmath748 of type 1 if and only if @xmath759 .
let @xmath748 be of type 2 and @xmath749 of type 4 .
then @xmath3 is a point in @xmath760 for all @xmath748 of type 2 if and only if @xmath761 .
finally , if both equations are of type 4 , then the condition @xmath762 translates into : @xmath763 .
we deduce that @xmath764 , which ends the proof .
note that in the last equality of the above proof , there is only one monomial on each side of the equality where the minimum is attained .
if this monomial happens to be the same on both sides , we can not ensure that @xmath3 is on the variety @xmath765 .
both items in the statement of theorem [ teo : non_transversal_intersection_nontrivial_case ] are similar but the first one concerns tropical varieties , while the second one needs to deal with equalities of the values taken by two tropical polynomials , that do not represent tropical varieties .
a more homogeneous approach can be given if we use the supertropical algebra introduced by izhakian and then developed by izhakian and rowen , to unify both conditions .
recall that one can consider an _ extended tropical semiring _
@xmath766 constructed from our tropical ring @xmath767 @xcite .
this semiring structure has a partial idempotent addition that distinguishes between sums of similar elements and sums of different elements .
set theoretically , this bigger semiring @xmath768 is composed from the disjoint union of two copies of @xmath24 , denoted @xmath24 and @xmath769 , plus the neutral element for the sum @xmath770 .
there is a natural bijection @xmath771 and the operation @xmath772 verifies @xmath773 , for all @xmath774 .
the elements in @xmath769 are called _ ghosts _ , so @xmath775 is a ghost element .
this terminology reflects the idea that in a field @xmath13 with a valuation , if two elements have the same valuation , we can not predict in general the valuation of their sum .
we refer to @xcite for further details , in particular for the full definition of the operations in this supertropical algebra .
an element @xmath776 lies in the variety @xmath777 defined by a supertropical polynomial @xmath778 with a finite support set @xmath779 , when @xmath780 is a ghost element .
given @xmath781 as before and two supertropical polynomials @xmath782 with respective supports in @xmath544 and coefficients in @xmath24 ( the so called tangible elements of @xmath768 ) , we can mimic our previous definitions .
thus , a point @xmath697 is said to be a non - transversal intersection of @xmath43 and @xmath579 if there exists @xmath783 such that @xmath784 is a ghost element for any integer affine linear form @xmath283 in @xmath232 variables .
[ teo : non_transversal_intersection_supertropical ] let @xmath782 be two supertropical bivariate polynomials with respective supports @xmath544 and let @xmath785 . then @xmath3 is a non - transversal intersection point if and only if for all @xmath145-tuples @xmath786 , it holds that @xmath3 lies in the variety defined by the supertropical polynomial @xmath787 @xmath788 ; that is , the value of this polynomial at @xmath3 is a ghost .
* acknowledgments : * this work started during our stays at the mathematical sciences research institute ( msri ) during the special semester on tropical geometry , where we enjoyed a wonderful working atmosphere .
ad is grateful for the support of the simons foundation during that period .
we thank cristian czubara for useful remarks and antonio laface for stimulating questions .
we also thank the referees for insightful remarks . | we study the notion of singular tropical hypersurfaces of any dimension .
we characterize the singular points in terms of tropical euler derivatives and we give an algorithm to compute all singular points .
we also describe non - transversal intersection points of planar tropical curves . |
the massive quantities of data being generated every day , and the ease of collaborative data analysis and data science have led to severe issues in management and retrieval of datasets .
we motivate our work with two concrete example scenarios . * [ intermediate result datasets ] for most organizations dealing with large volumes of diverse datasets ,
a common scenario is that many datasets are repeatedly analyzed in slightly different ways , with the intermediate results stored for future use .
often , we find that the intermediate results are the same across many pipelines ( e.g. , a _ pagerank _
computation on the web graph is often part of a multi - step workflow ) .
often times , the datasets being analyzed might be slightly different ( e.g. , results of simple transformations or cleaning operations , or small updates ) , but are still stored in their entirety .
there is currently no way of reducing the amount of stored data in such a scenario : there is massive redundancy and duplication ( this was corroborated by our discussions with a large software company ) , and often the computation required to recompute a given version from another one is small enough to not merit storing a new version . * [ data science dataset versions ] in our conversations with a computational biology group , we found that every time a data scientist wishes to work on a dataset , they make a private copy , perform modifications via cleansing , normalization , adding new fields or rows , and then store these modified versions back to a folder shared across the entire group .
once again there is massive redundancy and duplication across these copies , and there is a need to minimize these storage costs while keeping these versions easily retrievable . in such scenarios and many others ,
it is essential to keep track of versions of datasets and be able to recreate them on demand ; and at the same time , it is essential to minimize the storage costs by reducing redundancy and duplication . the ability to manage a large number of datasets , their versions , and derived datasets , is a key foundational piece of a system we are building for facilitating collaborative data science , called datahub @xcite .
datahubenables users to keep track of datasets and their versions , represented in the form of a directed _ version graph _ that encodes derivation relationships , and to retrieve one or more of the versions for analysis . in this paper , we focus on the problem of trading off storage costs and recreation costs in a principled fashion . specifically , the problem we address in this paper is : given a collection of datasets as well as ( possibly ) a directed version graph connecting them , minimize the overall storage for storing the datasets and the recreation costs for retrieving them .
the two goals conflict with each other minimizing storage cost typically leads to increased recreation costs and vice versa .
we illustrate this trade - off via an example
. indicates a storage cost of @xmath0 and a recreation cost of @xmath1 ; ( ii , iii , iv ) three possible storage graphs ] [ fig : version_graph ] indicates a storage cost of @xmath0 and a recreation cost of @xmath1 ; ( ii , iii , iv ) three possible storage graphs , title="fig : " ] ( i ) indicates a storage cost of @xmath0 and a recreation cost of @xmath1 ; ( ii , iii , iv ) three possible storage graphs , title="fig : " ] ( ii ) [ fig : extreme1 ] indicates a storage cost of @xmath0 and a recreation cost of @xmath1 ; ( ii , iii , iv ) three possible storage graphs , title="fig : " ] ( iii ) [ fig : extreme2 ] indicates a storage cost of @xmath0 and a recreation cost of @xmath1 ; ( ii , iii , iv ) three possible storage graphs , title="fig : " ] ( iv ) [ fig : storage_graph ] figure [ fig : version_graph](i ) displays a version graph , indicating the derivation relationships among 5 versions .
let @xmath2 be the original dataset .
say there are two teams collaborating on this dataset : team 1 modifies @xmath2 to derive @xmath3 , while team 2 modifies @xmath2 to derive @xmath4 . then , @xmath3 and @xmath4 are merged and give @xmath5 . as presented in figure [ fig : version_graph ] ,
@xmath2 is associated with @xmath6 , indicating that @xmath2 s storage cost and recreation cost are both @xmath7 when stored in its entirety ( we note that these two are typically measured in different units see the second challenge below ) ; the edge @xmath8 is annotated with @xmath9 , where @xmath10 is the storage cost for @xmath4 when stored as the modification from @xmath2 ( we call this the _ delta _ of @xmath4 from @xmath2 ) and @xmath11 is the recreation cost for @xmath4 given @xmath2 , i.e , the time taken to recreate @xmath4 given that @xmath2 has already been recreated .
one naive solution to store these datasets would be to store all of them in their entirety ( figure [ fig : version_graph ] ( ii ) ) . in this case , each version can be retrieved directly but the total storage cost is rather large , i.e. , @xmath12 . at the other extreme , only one version is stored in its entirety while other versions are stored as modifications or deltas to that version , as shown in figure [ fig : version_graph ] ( iii ) .
the total storage cost here is much smaller ( @xmath13 ) , but the recreation cost is large for @xmath14 and @xmath5 .
for instance , the path @xmath15 needs to be accessed in order to retrieve @xmath5 and the recreation cost is @xmath16 .
figure [ fig : version_graph ] ( iv ) shows an intermediate solution that trades off increased storage for reduced recreation costs for some version .
here we store versions @xmath2 and @xmath4 in their entirety and store modifications to other versions .
this solution also exhibits higher storage cost than solution ( ii ) but lower than ( iii ) , and still results in significantly reduced retrieval costs for versions @xmath4 and @xmath5 over ( ii ) . despite the fundamental nature of the storage - retrieval problem
, there is surprisingly little prior work on formally analyzing this trade - off and on designing techniques for identifying effective storage solutions for a given collection of datasets .
version control systems ( vcs ) like git , svn , or mercurial , despite their popularity , use fairly simple algorithms underneath , and are known to have significant limitations when managing large datasets @xcite .
much of the prior work in literature focuses on a linear chain of versions , or on minimizing the storage cost while ignoring the recreation cost ( we discuss the related work in more detail in section [ sec : related ] ) . in this paper
, we initiate a formal study of the problem of deciding how to jointly store a collection of dataset versions , provided along with a version or derivation graph . aside from being able to handle the scale , both in terms of dataset sizes and the number of versions ,
there are several other considerations that make this problem challenging . * different application scenarios and constraints
lead to many variations on the basic theme of balancing storage and recreation cost ( see table [ table : prob ] ) .
the variations arise both out of different ways to reconcile the conflicting optimization goals , as well as because of the variations in how the differences between versions are stored and how versions are reconstructed .
for example , some mechanisms for constructing differences between versions lead to symmetric differences ( either version can be recreated from the other version ) we call this the _ undirected _ case . the scenario with asymmetric , one - way differences
is referred to as _ directed _ case .
* similarly , the relationship between storage and recreation costs leads to significant variations across different settings . in some cases the recreation cost is proportional to the storage cost ( e.g. , if the system bottleneck lies in the i / o cost or network communication ) , but that may not be true when the system bottleneck is cpu computation .
this is especially true for sophisticated differencing mechanisms where a compact derivation procedure might be known to generate one dataset from another . *
another critical issue is that computing deltas for all pairs of versions is typically not feasible .
relying purely on the version graph may not be sufficient and significant redundancies across datasets may be missed . *
further , in many cases , we may have information about relative _ access frequencies _ indicating the relative likelihood of retrieving different datasets . several baseline algorithms for solving this problem can not be easily adapted to incorporate such access frequencies .
we note that the problem described thus far is inherently `` online '' in that new datasets and versions are typically being created continuously and are being added to the system . in this paper
, we focus on the static , off - line version of this problem and focus on formally and completely understanding that version .
we plan to address the online version of the problem in the future .
the key contributions of this work are as follows .
* we formally define and analyze the dataset versioning problem and consider several variations of the problem that trade off storage cost and recreation cost in different manners , under different assumptions about the differencing mechanisms and recreation costs ( section [ sec : proboverview ] ) .
table [ table : prob ] summarizes the problems and our results .
we show that most of the variations of this problem are np - hard ( section [ sec : complexity ] ) .
* we provide two light - weight heuristics : one , when there is a constraint on average recreation cost , and one when there is a constraint on maximum recreation cost ; we also show how we can adapt a prior solution for balancing minimum - spanning trees and shortest path trees for undirected graphs ( section [ sec : algorithms ] ) .
* we have built a prototype system where we implement the proposed algorithms .
we present an extensive experimental evaluation of these algorithms over several synthetic and real - world workloads demonstrating the effectiveness of our algorithms at handling large problem sizes ( section [ sec : experiments ] ) . [
cols="<,<,<,<,<,<",options="header " , ] [ table : ilp ]
perhaps the most closely related prior work is source code version systems like git , mercurial , svn , and others , that are widely used for managing source code repositories . despite their popularity
, these systems largely use fairly simple algorithms underneath that are optimized to work with modest - sized source code files and their on - disk structures are optimized to work with line - based diffs .
these systems are known to have significant limitations when handling large files and large numbers of versions @xcite . as a result , a variety of extensions like git - annex @xcite , git - bigfiles @xcite , etc .
, have been developed to make them work reasonably well with large files .
there is much prior work in the temporal databases literature @xcite on managing a linear chain of versions , and retrieving a version as of a specific time point ( called _ snapshot _ queries ) @xcite .
@xcite proposed an archiving technique where all versions of the data are merged into one hierarchy .
an element appearing in multiple versions is stored only once along with a timestamp .
this technique of storing versions is in contrast with techniques where retrieval of certain versions may require undoing the changes ( unrolling the deltas ) .
the hierarchical data and the resulting archive is represented in xml format which enables use of xml tools such as an xml compressor for compressing the archive .
it was not , however , a full - fledged version control system representing an arbitrarily graph of versions ; rather it focused on algorithms for compactly encoding a linear chain of versions .
snapshot queries have recently also been studied in the context of array databases @xcite and graph databases @xcite . seering et al .
@xcite considered the problem of storing an arbitrary tree of versions in the context of scientific databases ; their proposed techniques are based on finding a minimum spanning tree ( as we discussed earlier , that solution represents one extreme in the spectrum of solutions that needs to be considered ) .
they also proposed several heuristics for choosing which versions to materialize given the distribution of access frequencies to historical versions .
several databases support `` time travel '' features ( e.g. , oracle flashback , postgres @xcite ) .
however , those do not allow for branching trees of versions .
@xcite articulates a similar vision to our overall datahubvision ; however , they do not propose formalisms or algorithms to solve the underlying data management challenges .
in addition , the schema of tables encoded with flashback can not change .
there is also much prior work on compactly encoding differences between two files or strings in order to reduce communication or storage costs .
in addition to standard utilities like unix diff , many sophisticated techniques have been proposed for computing differences or edit sequences between two files ( e.g. , xdelta @xcite , vdelta @xcite , vcdiff @xcite , zdelta @xcite ) . that work is largely orthogonal and complementary to our work .
many prior efforts have looked at the problem of minimizing the total storage cost for storing a collection of related files ( i.e. , problem 1 ) .
these works do not typically consider the recreation cost or the tradeoffs between the two .
quinlan et al .
@xcite propose an archival `` deduplication '' storage system that identifies duplicate blocks across files and only stores them once for reducing storage requirements .
zhu et al .
@xcite present several optimizations on the basic theme .
douglis et al .
@xcite present several techniques to identify pairs of files that could be efficiently stored using delta compression even if there is no explicit derivation information known about the two files ; similar techniques could be used to better identify which entries of the matrices @xmath17 and @xmath18 to reveal in our scenario .
ouyang et al .
@xcite studied the problem of compressing a large collection of related files by performing a sequence of pairwise delta compressions .
they proposed a suite of text clustering techniques to prune the graph of all pairwise delta encodings and find the optimal branching ( i.e. , mca ) that minimizes the total weight .
burns and long @xcite present a technique for in - place re - construction of delta - compressed files using a graph - theoretic approach . that work could be incorporated into our overall framework to reduce the memory requirements during reconstruction .
similar dictionary - based reference encoding techniques have been used by @xcite to efficiently represent a target web page in terms of additions / modifications to a small number of reference web pages .
kulkarni et al .
@xcite present a more general technique that combines several different techniques to identify similar blocks among a collection files , and use delta compression to reduce the total storage cost ( ignoring the recreation costs ) .
we refer the reader to a recent survey @xcite for a more comprehensive coverage of this line of work .
large datasets and collaborative and iterative analysis are becoming a norm in many application domains ; however we lack the data management infrastructure to efficiently manage such datasets , their versions over time , and derived data products . given the high overlap and duplication among the datasets , it is attractive to consider using delta compression to store the datasets in a compact manner , where some datasets or versions are stored as modifications from other datasets ; such delta compression however leads to higher latencies while retrieving specific datasets . in this paper
, we studied the trade - off between the storage and recreation costs in a principled manner , by formulating several optimization problems that trade off these two in different ways and showing that most variations are np - hard .
we also presented several efficient algorithms that are effective at exploring this trade - off , and we presented an extensive experimental evaluation using a prototype version management system that we have built .
there are many interesting and rich avenues for future work that we are planning to pursue .
in particular , we plan to develop online algorithms for making the optimization decisions as new datasets or versions are being created , and also adaptive algorithms that reevaluate the optimization decisions based on changing workload information .
we also plan to explore the challenges in extending our work to a distributed and decentralized setting .
git uses delta compression to reduce the amount of storage required to store a large number of files ( objects ) that contain duplicated information .
however , git s algorithm for doing so is not clearly described anywhere .
an old discussion with linus has a sketch of the algorithm @xcite .
however there have been several changes to the heuristics used that do nt appear to be documented anywhere .
here we focus on `` repack '' , where the decisions are made for a large group of objects .
however , the same algorithm appears to be used for normal commits as well .
most of the algorithm code is in file : ` builtin / pack - objects.c ` note the name hash is not a true hash ; the ` pack_name_hash ( ) ` function ( ` pack - objects.h ` ) simply creates a number from the last 16 non - white space characters , with the last characters counting the most ( so all files with the same suffix , e.g. , ` .c ` , will sort together ) .
* step 2 : * the next key function is ` ll_find_deltas ( ) ` , which goes over the files in the sorted order .
it maintains a list of @xmath19 objects ( @xmath19 = window size , default 10 ) at all times . for the next object ,
say @xmath20 , it finds the delta between @xmath20 and each of the objects , say @xmath21 , in the window ; it chooses the the object with the minimum value of : ` delta(b , o ) / ( max_depth - depth of b ) ` where ` max_depth ` is a parameter ( default 50 ) , and depth of b refers to the length of delta chain between a root and b. the original algorithm appears to have only used ` delta(b , o ) ` to make the decision , but the `` depth bias '' ( denominator ) was added at a later point to prefer slightly larger deltas with smaller delta chains .
the key lines for the above part : * line 1812 ( check each object in the window ) : + .... ret = try_delta(n , m , max_depth , & mem_usage ) ; .... * lines 1617 - 1618 ( depth bias ) : + .... max_size = ( uint64_t)max_size * ( max_depth - src->depth ) / ( max_depth - ref_depth + 1 ) ; .... * line 1678 ( compute delta and compare size ) : + .... delta_buf = create_delta(src->index , trg->data , trg_size , & delta_size , max_size ) ; .... ` create_delta ( ) ` returns non - null only if the new delta being tried is smaller than the current delta ( modulo depth bias ) , specifically , only if the size of the new delta is less than ` max_size ` argument .
note : lines 1682 - 1688 appear redundant given the depth bias calculations .
* step 3 . *
originally the window was just the last @xmath19 objects before the object @xmath20 under consideration .
however , the current algorithm shuffles the objects in the window based on the choices made .
specifically , let @xmath22 be the current objects in the window .
let the object chosen to delta against for @xmath20 be @xmath23 .
then @xmath23 would be moved to the end of the list , so the new list would be : @xmath24 $ ] .
then when we move to the new object after @xmath20 ( say @xmath25 ) , we slide the window and so the new window then would be : @xmath26 $ ] .
small detail : the list is actually maintained as a circular buffer so the list does nt have to be physically `` shifted '' ( moving @xmath23 to the end does involve a shift though ) .
relevant code here is lines 1854 - 1861 .
finally we note that git never considers / computes / stores a delta between two objects of different types , and it does the above in a multi - threaded fashion , by partitioning the work among a given number of threads .
each of the threads operates independently of the others . | the relative ease of collaborative data science and analysis has led to a proliferation of many thousands or millions of _ versions _ of the same datasets in many scientific and commercial domains , acquired or constructed at various stages of data analysis across many users , and often over long periods of time . managing , storing , and recreating these dataset versions is a non - trivial task .
the fundamental challenge here is the _ storage - recreation trade - off _ : the more storage we use , the faster it is to recreate or retrieve versions , while the less storage we use , the slower it is to recreate or retrieve versions . despite the fundamental nature of this problem ,
there has been a surprisingly little amount of work on it . in this paper
, we study this trade - off in a principled manner : we formulate six problems under various settings , trading off these quantities in various ways , demonstrate that most of the problems are intractable , and propose a suite of inexpensive heuristics drawing from techniques in delay - constrained scheduling , and spanning tree literature , to solve these problems .
we have built a prototype version management system , that aims to serve as a foundation to our datahubsystem for facilitating collaborative data science @xcite .
we demonstrate , via extensive experiments , that our proposed heuristics provide efficient solutions in practical dataset versioning scenarios . |
the lattice quantum chromodynamics calculations predict the existence of a transition between a hadronic gas and a quark gluon plasma at a temperature around 170mev in a domain close to the net - baryon free region @xcite .
this deconfined state of quarks and gluons is expected to be formed in ultra - relativistic heavy ion collisions .
the study of the final hadronic state properties of such collisions essentially dominated by the low @xmath8 part of the particle spectra ( the bulk ) via multi - strange particles may provide information on its dynamics from the early stage to the chemical and thermal freeze - out ( fo ) .
+ strange quarks whose mass is comparable to the temperature of the qgp formation are expected to be abundantly produced in the high temperature qgp phase because of parton rescattering and should achieve equilibration @xcite .
hadronization process has been very well described in the framework of statistical model @xcite by adjusting four free parameters , the chemical temperature @xmath1 , the baryon and strange chemical potential @xmath9 and @xmath10 and the strangeness phase space occupancy factor @xmath2 .
this latter provides a measurement of the degree of strangeness equilibration in the system .
its evolution with centrality as well as @xmath1 evolution gives a quantitative measurement of strangeness evolution in the bulk matter .
+ a study of the collective motion of the collision in the framework of a hydrodynamically - inspired model has previously shown that @xmath3 , k , p particles seem to take part to the same transverse collective flow and freeze - out kinetically at a temperature smaller than @xmath1 suggesting an expansion and the cooling of the system between chemical and thermal fo .
concerning multi - strange baryons , it has been suggested that these particles should not develop such significant transverse radial flow due to their presumably small cross section so that they should decouple much earlier in the collision @xcite .
their observed transverse radial flow would then primarily reflects partonic flow behaviour .
elliptic flow due to the initial asymmetry of the system in non - central collisions has also proven to be a good tool for understanding the properties of the early stage of the collisions @xcite .
thus multi - strange baryon elliptic flow could be a valuable probe of the initial partonic system . as flow is an additive quantity , we present both radial transverse flow and elliptic flow measurements of multi - strange baryons in order to disentangle its hadronic and partonic contributions .
8 cm 0.8 all the data presented in this contribution have been collected by the star detector described in @xcite .
multi - strange particles are identified via the topology of their decay @xmath11@xmath12@xmath13 and @xmath14@xmath12@xmath15 then the subsequent decay @xmath16@xmath12@xmath17 with the respective branching ratios 100@xmath18 , 68@xmath18 and 64@xmath18 . for more details
see @xcite .
corrections for tracking efficiency and detector acceptance were applied .
the final corrected transverse momentum distributions are fitted in order to extract yields and inverse slope parameters .
the results of a statistical fit of particle ratios including @xmath19 , @xmath20 , @xmath5 , @xmath21 , @xmath16 , @xmath22 , @xmath23 , @xmath24 and @xmath25 are presented on figure [ fig : gammas ] .
very good agreement between our data and the model are achieved for each centrality range at @xmath0 at rhic @xcite .
a fit has been performed first including @xmath19 , @xmath20 , @xmath5 , @xmath21 . for the most central collisions ,
the four free parameters of the fit are @xmath26 , @xmath27 , @xmath28 and @xmath29 .
the evolution of @xmath1 and @xmath2 with centrality are represented as square symbols on figure [ fig : gammas ] .
an other fit has also been performed including then all the hadrons @xmath19 , @xmath20 , @xmath5 , @xmath21 , @xmath16 , @xmath22 , @xmath23 , @xmath24 and @xmath25 .
the parameters obtained for the most central collision are @xmath30 , @xmath31 , @xmath32 and @xmath33 .
@xmath1 and @xmath2 evolutions with centrality are represented as circles .
we note no dependence of @xmath1 with centrality .
all the particles seem to chemically freeze - out at a temperature of 160@xmath345mev close to lqcd predictions , @xmath1 seems to be essentially fixed by the most numerous @xmath3 , @xmath4 , @xmath5 particles and does not seem to be dependent on the initial system size . from peripheral to central collisions , including ( multi-)strange hadrons in the fit , @xmath2 increases from 0.8 and saturates at 1 .
this value suggests that in most central collisions at top rhic energy , the phase space is saturated in strange quarks so that the system is close to strangeness equilibration .
this increase of @xmath2 also signs the existence of significant @xmath35@xmath36 production processes at a partonic level such as gluon fusions . a hydrodynamically - inspired fit known as `` blastwave fit '' @xcite assuming all particles
are emitted from a thermal expanding source with a transverse flow velocity @xmath37@xmath38@xmath39 at the thermal freeze - out temperature @xmath40 has been performed on @xmath3 , @xmath4 , @xmath5 spectra together and on @xmath11 and @xmath14 separately .
a velocity profile @xmath41=@xmath42 was used , where @xmath43 is the radius of the source and @xmath44 was determined from the fit to the @xmath3 , @xmath4 and @xmath5 spectra ranging from @xmath44=0.81 for the most central bin to @xmath44=1.42 for the most peripheral . for ( @xmath3 , @xmath4 , @xmath5 ) ,
9 bins of centrality indexed from 1 ( most central ) to 9 ( most peripheral ) have been considered while 5 centrality bins for @xmath6 have been studied . for @xmath7 ,
only the most central bin has been investigated .
the results of the fits are presented on figure [ fig : blastwave ] .
9.cm 0.8 the one and two sigma contours are represented for the best fit values ( @xmath40,@xmath37@xmath38@xmath39 ) .
for the most central bin , we note : 1 ) there is no overlap of the contours for ( @xmath3,@xmath4,@xmath5 ) and ( @xmath11 ) suggesting that ( @xmath3,k , p ) take part to the same collective transverse radial flow different from the one developed by the @xmath11 ; 2 ) concerning @xmath11 , they seem to thermally freeze - out at a temperature of @xmath45 , close to the chemical fo temperature previously obtained from statistical fits to particle ratios whereas @xmath40 for ( @xmath3,@xmath4,@xmath5 ) amounts @xmath46 .
it suggests that multi - strange particles should have decoupled earlier in the collision close to chemical fo ; 3 ) furthermore , the fact that they develop an as significant flow as @xmath16 and that their interaction cross - section is presumably very small , suggests that their flow has been developed prior to chemical fo so prior to the hadronization , probably at a partonic stage of the system and not in the hadronic phase as for ( @xmath3,@xmath4,@xmath5 ) .
otherwise , it is corroborated by the fact that the thermal fo parameters of the multi - strange baryons do not depend on the centrality and that @xmath1 is close to @xmath40 . concerning ( @xmath3,@xmath4,@xmath5 ) , results show that @xmath1@xmath39@xmath40 and that the difference between these two temperatures increases with centrality .
it suggests a longer duration time between these two fo for the lightest particles ( @xmath3,@xmath4,@xmath5 ) essentially due to their rescattering in the hadron phase while the system is cooling down .
these results indicate that au+au collisions with different initial conditions evolve always to the same chemical fo temperature , and then cool down further to a kinetic fo dependent on centrality .
so this radial flow scenario suggests that for multi - strange baryons , a significant fraction ( if not all ) of the transverse flow has been developed probably in a partonic phase of the system so that multi - strange baryons should develop elliptic flow .
figure [ fig : v2pt ] shows the measurement of the elliptic flow @xmath47 of @xmath48 and @xmath49 as a function of @xmath8 for the minimum bias data .
@xmath47 of @xmath50 and @xmath51 previously measured @xcite are also represented for comparison . ' '' '' ' '' '' first we observe that @xmath47 of multi - strange baryons is different from zero and seems to follow the same behaviour as @xmath16 @xmath47 .
that means same shape ( saturation at a @xmath52 ) and same amplitude ( saturation at @xmath53 ) . in the low @xmath8 region ,
@xmath11 @xmath47 is in agreement with hydrodynamic model calculations ( colored zone ) which predict its mass ordering in this @xmath8 region .
however , for a @xmath54 , @xmath47 deviates from hydrodynamic model prediction and shows different behaviour for @xmath50 which saturates at a @xmath55 , at a value around @xmath56 compared to the strange baryons @xmath47 .
it confirms a previously established baryon to meson dependence of the elliptic flow parameter from a particle mass dependence in the intermediate @xmath8 region @xcite .
this particle type dependence is well and `` simply '' explained by quark coalescence or recombination models @xcite in which hadrons are dominantly produced by the coalescence of constituent quarks from a partonic system supporting the idea of a collectivity between partons .
these models predict a universal scaling of transverse momentum @xmath8 and elliptic flow to the number of constituent quarks ( @xmath44 ) .
previously , such scaling has been demonstrated for the mesons @xmath57 and the baryons @xmath16 at intermediate @xmath8 @xcite .
figure [ fig : v2nptn ] shows the superposition of the scaled elliptic flows @xmath58 for @xmath57 , @xmath16 as well as for @xmath6 , supporting that the flow of @xmath35 quarks is close to that of @xmath59 and @xmath60 quarks within error bars .
we have presented the evolution of freeze - out parameters with au+au collision centrality .
the increase of the strange quark phase space saturation factor , @xmath2 , up to 1 for the most central collision suggests that strangeness equilibration is achieved at top rhic energy .
@xmath1 , common for all particles , appears to be independent of the system size .
the kinetic fo parameters obtained from blastwave fit to @xmath3 , @xmath4 , @xmath5 spectra suggest that they are taking part to a same collective flow behaviour with an increasing duration time between chemical and thermal fo with centrality essentially due to hadron rescatterings . for multi - strange baryons
, the collective behaviour seems to be quite different .
since @xmath1 and @xmath40 are close to each other and show no dependence with centrality , it indicates that multi - strange baryons take less part in the evolution dynamics in the hadronic phase and should have decoupled much earlier in the collision than ( @xmath3 , @xmath4 , @xmath5 ) , carrying with them an important partonic flow contribution .
this idea is emphasized by the measurement of their elliptic flow whose scaling by the constituent quarks in the intermediate @xmath8 region is well described by coalescence and recombination models .
9 z. fodor , nucl .
a715 * , 319c ( 2003 ) and references there in ; z. fodor and s.d .
katz , hep - lat/0402006 .
j. rafelski , j. letessier , nucl . phys .
* a715 * , 98 ( 2003 ) .
p. braun - munzinger , i. heppe and j. stachel , phys .
* b465 * , 15 ( 1999 ) . f. becattini and l. ferroni , hep - ph/0407117 .
p. braun - munzinger , k. redlich and j. stachel , nucl - th/0304013 .
h. van hecke , h. sorge and n. xu , phys .
* 81 * , 5764 ( 1998 ) .
y. cheng _ et al .
c*68 * , 034910 ( 2003 ) .
j. y. ollitrault , phys .
rev . d*46 * , 229 ( 1992 ) .
star collaboration , nucl .
a499 * 659 ( 2003 ) . c. adler _
et al . _ ,
star collaboration , phys .
rev . lett.*89 * , 092301 ( 2002 ) .
e. schnedermann , j. sollfrank and u. heinz , phys . rev . *
c48 * , 2462 ( 1993 ) .
j. adams _ et al .
_ , star collaboration , phys .
lett . * 92 * , 052302 ( 2004 ) .
j. castillo , star collaboration , j. phys .
phys . * 30 * s827 - 34 ( 2004 ) d. molnar and s. voloshin , phys .
91 * , 092301 ( 2003 ) .
fries , b. mueller , c. nonaka , s.a .
bass , phys .
lett . * 90 * , 092303 ( 2003 ) . | the study of the multi - strange baryon production in the final state of ultra - relativistic au+au collisions at @xmath0 at rhic gives information on the freeze - out conditions and probably on earlier stages of the collision .
chemical freeze - out temperature @xmath1 as well as the strangeness phase space occupancy factor @xmath2 extracted from statistical models are studied as a function of the collision centrality . @xmath2
saturates at 1 in most central collisions suggesting an equilibration of strange quarks in the medium .
a study of the collective motion of the collision is possible in the framework of a hydrodynamically - inspired model which considers the particles to be emitted from a locally thermalized system . in this view , multi - strange particles seem to develop significant radial transverse flow but smaller than @xmath3,@xmath4,@xmath5 . furthermore , @xmath6 and @xmath7 present an elliptic flow as strong as for previously measured baryons supporting the idea that a fraction of the final collective motion has been developed at an early partonic stage of the collision . |
models of the nambu jona - lasinio ( njl ) type @xcite have a long history and have been extensively used to describe the dynamics of lightest hadrons and the thermodynamic properties of excited matter ( see the review articles @xcite ) . in the `` classical '' versions such schematic models incorporate the chiral symmetry of two - flavor qcd and its spontaneous breakdown at temperatures below the critical one , @xmath1 .
they offer a simple and practical illustration of the basic mechanisms that drive the spontaneous breaking of chiral symmetry , a key feature of qcd in its low - temperature , low - density phase .
however , in spite of their widespread use , the njl models suffer from the major shortcoming that the reduction to global color symmetry prevents quark confinement . in the polyakov - loop - extended njl ( pnjl )
model @xcite@xcite , the quarks are coupled simultaneously to the chiral condensate , to be an order parameter of the chiral symmetry breaking , and to a homogeneous gauge field representing polyakov loop dynamics , which serves as an order parameter for the transition from the low - temperature , symmetric , confined phase to the high - temperature , deconfined phase .
the model has proven successful in reproducing lattice data on qcd thermodynamics @xcite . in this paper
we confront general properties of @xmath2 and @xmath3 mesons as well as thermodynamics at finite temperature @xmath4 and baryon chemical potential @xmath5 calculated within the two - flavor njl model with those of the pnjl one .
to describe the coupling between quarks and the chiral condensate in the scalar - pseudoscalar sectors , the two - flavor njl model @xcite is used with the following lagrangian density @xmath6~,\end{aligned}\ ] ] where @xmath7 is the coupling constant , @xmath8 is the vector of pauli matrices in flavor space , @xmath9 and @xmath10 are the quark fields ( color and flavor indices are suppressed ) , @xmath11 is the diagonal matrix of the current quarks masses , @xmath12 , @xmath13 . the grand canonical thermodynamic potential can be obtained from this lagrangian in systematic approximations . in the mean - field approximation
it has the form @xcite @xmath14 with @xmath15~,\end{aligned}\ ] ] where @xmath16 and @xmath17 with @xmath18 and the inverse temperature @xmath19 .
the deconfinement in pure @xmath20 gauge theory can be simulated by introducing an effective potential for a complex polyakov loop field .
the pnjl lagrangian @xcite@xcite is @xmath21 -\mathcal{u}\left(\phi[a],\bar\phi[a];t\right)~.\end{aligned}\ ] ] here , the notation is the same as in eq .
( [ njl ] ) .
the quark fields are coupled to the gauge field @xmath22 through the covariant derivative @xmath23 .
the gauge field is @xmath24 ( the polyakov gauge ) .
the field @xmath25 is determined by the trace of the polyakov loop @xmath26 @xcite @xmath27 = \dfrac{1}{n_c } \mbox{tr}_c l(\vec{x})~,\nonumber\end{aligned}\ ] ] where @xmath28 $ ] .
the gauge sector of the lagrangian density ( [ lpnjl ] ) is described by an effective potential @xmath29,\bar\phi[a];t\right)$ ] fitted to the lattice qcd simulation results in pure @xmath30 gauge theory at finite @xmath4 @xcite with @xmath31 the parameters of the effective potential ( [ effpot ] ) and ( [ ueff ] ) are summarized in table [ table1 ] .
the parameter @xmath32 in general depends on the number of active flavors and the chemical potental @xcite . in the present work we use @xmath33 mev as in @xcite . the thermodynamic potential for the pnjl model in the mean - field approximation is given by the following equation @xcite @xmath34 where ( in analogy with ( [ potnjl ] ) ) @xmath35~.\end{aligned}\ ] ] here
, @xmath36 is the quark energy , @xmath37 , and @xmath38 , \\ & & n^-_\phi(e_p ) = \left [ 1 + 3\left ( \bar{\phi } + { \phi } e^{-\beta e_p^-}\right ) e^{-\beta e_p^- } + e^{-3\beta e_p^- } \right]~.\end{aligned}\ ] ] since njl - type models are nonrenormalizable it is necessary to introduce a regularization , e.g. , by a cutoff @xmath39 in the momentum integration . following @xcite , we use in this study the three - dimensional momentum cutoff for vacuum terms and extend this integration till infinity for finite temperatures . a comprehensive study of the differences between the two regularization procedures ( with and without cutoff on the quark momentum states at finite temperature ) was performed in @xcite .
in the mean - field approximation , we can obtain the constituent quark mass @xmath40 from the condition that the thermodynamical potential ( eqs .
( [ potnjl ] ) and ( [ potpnjl ] ) , resp . ) shall have a minimum with respect to varying this parameter , @xmath41 .
this condition is equivalent to the gap equation @xcite @xmath42 where the quark condensate is defined as @xmath43 . for the mass gap equation of both models
we get @xmath44~,\end{aligned}\ ] ] with @xmath45 for the njl model , and @xmath46/ n_\phi^+(e_p)~ , \label{fermimod}\\ f^- & = & \left[\left(\bar{\phi}+2{\phi } e^{-\beta e_p^-}\right ) e^{-\beta e_p^- } + e^{-3\beta e_p^- } \right]/ n_\phi^-(e_p ) \label{afermimod}\end{aligned}\ ] ] for the pnjl model .
moreover , for pnjl calculations we should find the values of @xmath25 and @xmath47 by minimizing @xmath48 with respect @xmath25 and @xmath47 @xcite at given @xmath4 and @xmath5 .
one should note , that if @xmath49 , the expressions eqs .
( [ fermimod]),([afermimod ] ) reduce to the standard njl model eqs .
( [ fermi ] ) and ( [ afermi ] ) . for a self - consistent description of the particle spectrum in the mean - field approximation
, the meson correlations have to be taken into consideration .
these correlations are related to the polarization operator of constituent fields . for scalar and pseudoscalar particles the polarization operators
are represented by loop - integrals @xcite .
@xmath50 , \label{polpi } \\
\pi^{ss}_{ab } ( p^2 ) & = & \int \frac{d^4p}{(2\pi)^4 } \mbox{tr}\ , \left[s(p+p ) s(p ) \right ] , \label{polsig}\end{aligned}\ ] ] where the operation @xmath51 is taken over dirac , flavor and color indices of quark fields . from point of view of the polarization operators ,
the pseudoscalar ( @xmath2 ) and scalar ( @xmath3 ) meson masses can be defined by the condition that for @xmath52 @xmath53 the corresponding polarization operator @xmath54 ( @xmath55 ) , leads to a bound state pole in the corresponding meson correlation function @xcite . for mesons at rest ( @xmath56 ) in the medium
, these conditions correspond to the equations @xmath57 in order to solve eqs .
( [ masq ] ) , ( [ masspi ] ) and ( [ masssigma ] ) , a set of model parameters has to be determined : the cutoff parameter @xmath39 , the current quark mass @xmath58 ( in chiral limit @xmath59 ) and the coupling constant @xmath7 .
these parameters are fixed at @xmath60 to reproduce physical quantities : the pion mass @xmath61 gev , the pion decay constant @xmath62 gev and the quark condensate @xmath63 mev .
the obtained parameters are shown in table [ table2 ] .
solutions of the gap - equation ( [ masq ] ) and eqs .
( [ masspi ] ) , ( [ masssigma ] ) at nonzero @xmath4 are presented in fig .
[ masses ] .
the temperature is normalized to the mott temperature , which is defined from the condition @xmath64 . in the pnjl model @xmath65 gev and in the njl model
@xmath66 gev for our parameters .
the modification of the quasiparticle properties is clearly seen in this figure .
up to the mott temperature @xmath67 , the @xmath3 mass practically follows the behaviour of @xmath68 with a drop towards the pion mass signalling chiral symmetry restoration . at @xmath69
the masses of chiral partners become equal to each other , @xmath70 , and then both masses increase with temperature . below the mott temperature ,
the pion mass remains practically constant .
the transition region from the phase with broken chiral symmetry ( @xmath71 ) to chirally symetric phase ( @xmath72 ) is much narrower in the pnjl case when compared to the njl model .
for a recent discussion of this issue within the nonlocal pnjl model , see @xcite .
the thermodynamics of particles is described in terms of the grand canonical ensemble which is related with the hamiltonian @xmath73 as follows : @xmath74 where @xmath75 is the particle number operator and the operator @xmath51 is taken over momenta as well as color , flavor and dirac indices . if @xmath48 is known , the basic thermodynamic quantities - the pressure , the energy and entropy densities , the density of quarks number and heat conductivity - can be defined as follows : @xmath76 the thermodynamic potential in equilibrium corresponds to a global minimum with respect to variations of the order parameter(s ) @xmath77 all these relations describe thermodynamics of the system . for the considered models
the thermodynamic potentials are defined from eqs .
( [ potnjl ] ) and ( [ potpnjl ] ) . from these equations
we can read off the vacuum part @xmath78 this quantity does not vanish at @xmath79 and @xmath80 . therefore , in order to obtain the physical thermodynamical potential which corresponds to vanishing pressure and energy density at @xmath81 , one has to renormalize the thermodynamic potential by subtracting its vacuum expression ( [ omvac ] ) .
this corresponds to the following defintion of the physical pressure @xmath82 within the pnjl model with @xmath83 the reduced pressure and energy density exhibit reasonable behavior consistent with the recent lattice qcd results for the vanishing chemical potential @xcite ( see fig .
[ therm ] ) keeping in mind that the @xmath84 ratio in lattice calculations is still far from that for physical masses @xmath85 .
one should note that both models have the cutoff parameter @xmath39 @xcite .
but the integrals containing the logarithm in eq .
( [ potnjl ] ) and eq .
( [ potpnjl ] ) are both convergent @xcite . in our work
we calculated these integrals with @xmath83 .
it leads to the flattening of pressure at high temperature .
however , most integrals for the pnjl model are convergent too .
this was the reason to consider the thermodynamic functions for @xmath86 .
it was supposed that with increasing temperature the pressure has to reach the stefan - boltzmann limit @xcite , which in the pnjl model is defined as @xmath87 if the regularization @xmath88 0.639 is used , the @xmath4-behaviour of the thermodynamic quantities considered is roughly the same while their absolute values are noticeably lower , being far from the stefan - boltzmann limit . in the njl model both , @xmath89 and @xmath90
are not only underestimated due to the missing gluon contribution , but also essentially shifted toward lower temperatures because of the lack of a confining mechanism for the dynamical quark degrees of freedom . within njl - like models
there are several characteristic temperatures .
the parameter @xmath32 entering the effective potential ( [ ueff ] ) of the pnjl model has been noted above .
three other scales are the pseudo - critical temperature for chiral crossover @xmath91 defined by the maximum of @xmath92 , the pseudo - critical temperature for the deconfinement crossover @xmath93 which can be found from the maximum of @xmath94 , and @xmath95 defined for pnjl model as the average of two transition temperatures @xmath91 and @xmath93 @xcite .
the temperature dependence of the order parameters for the chiral ( @xmath96 ) and deconfinement ( @xmath25 ) phase transitions , are shown in fig .
( [ order ] ) .
the chiral condensate decreases and the polyakov loop potential increases with @xmath4 , demonstrating closeness of the pseudo - critical temperatures @xmath91 and @xmath93 at @xmath97 ( see also table [ table3 ] ) . for @xmath2-mesons ,
the mott temperature @xmath67 is provided by the condition @xmath98 and similarly the @xmath3 meson dissociation temperature @xmath99 is given by the equation @xmath100
all these quantities obtained at @xmath97 are presented in table [ table3 ] .
extending our study of the ( pseudo-)critical temperatures to nonzero chemical potential @xmath5 , we obtain phase diagrams in the @xmath101 plane shown in fig .
[ phasediag ] . the chiral transition line , determined by @xcite @xmath102_{m=0}=0\ ] ] in both njl and pnjl models , is a monotonously decreasing function of the chemical potential . in the limiting chirally symmetric case corresponding to @xmath103
the @xmath91 at large @xmath5 is higher than those for finite mass but both temperatures coincide when @xmath1040 for both the njl and pnjl models . for the case
@xmath1050 both models show the critical end point at the temperature @xmath106 below which the chiral phase transition is of first order . at this point ( @xmath107 ) the phase transition changes from first order to crossover @xcite . for the chirally symmetric case in the pnjl model the first order phase transition ends at a tricritical point above which , for @xmath108 , the chiral transition is of the second order . in the njl model
@xcite the topology of the phase diagram is the same as in the pnjl case , only @xmath106 and @xmath109 are situated at higher temperatures . in accordance with other calculations , the temperature of the tricritical point is above that of the critical end point . within the pnjl model
the positions of critical points are @xmath110(95,320 ) and @xmath111(160,265 ) mev .
these numbers are quite close to those in @xcite for the set with similar parameter values ( set b , the case i ) .
one should emphasize that critical properties of observables are significantly influenced by the chosen parameter set and regularization procedure as was demonstrated in @xcite .
we have compared the thermodynamics of njl and pnjl models . in agreement with previous results , it is shown that the inclusion of coupling between chiral symmetry and deconfinement essentially improves the description of thermodynamic bulk properties of the medium .
the models qualitatively reproduce both , @xmath2 and @xmath3 meson properties in hot , dense quark matter and the rich and complicated phase structure of this medium .
effects of the polyakov loop move the cep to higher @xmath4 and lower @xmath5 than in the njl case @xcite .
the position of the calculated cep in the @xmath101 plane is still far from the predictions of lattice qcd and empirical analysis .
the further elaboration of the presented models may include color superconducting phases and nonlocality of the interaction @xcite as well as effects beyond the meanfield @xcite .
we are grateful to d. blaschke , p. costa and v. v. skokov for useful comments .
v.t . acknowledges financial support from the helmholtz international center ( hic ) for fair within the loewe program . the work of yu .
k. was supported by rffi grant no .
09 - 01 - 00770a .
y. nambu and g. jona - lasinio , phys . rev . *
122 * , 345 ( 1961 ) .
m. k. volkov , annals of physics * 157 * , 282 ( 1984 ) .
u. vogl and w. weise , progr .
. phys . * 27 * , 195 ( 1991 ) .
s. p. klevansky , rev .
phys . * 64 * , 649 ( 1992 ) .
t. hatsuda and t. kunihiro , phys .
* 27 * , 221 ( 1994 ) .
m. buballa , phys . rep . * 407 * , 205 , ( 2005 ) . m. k. volkov and a. e. radzhabov , phys .
usp . * 49 * , 551 ( 2006 ) .
p. n. meisinger , t. r. miller , and m. c. ogilvie , phys . rev . *
d 65 * , 034009 ( 2002 ) .
k. fukushima , phys .
* b 591 * , 277 ( 2004 ) . c. ratti , m. a. thaler and w. weise , phys
d 73 * , 014019 ( 2006 ) .
e. megias , e. ruiz arriola and l. l. salcedo , phys . rev .
* d 74 * , 065005 ( 2006 ) .
s. k. ghosh , t. k. mukherjee , m. g. mustafa and r. ray , phys . rev .
* d 73 * , 114007 ( 2006 ) . m. asakawa and k. yazaki , nucl .
phys . * a 504 * , 668 ( 1989 ) .
j. hfner , s. p. klevansky and p. zhuang , acta phys . pol . * b 25 * , 85 ( 1994 )
. s. rssner , c. ratti and w. weise , phys . rev . *
d 75 * , 034007 ( 2007 ) .
b. j. schaefer , j. m. pawlowski and j. wambach , phys .
d * 76 * , 074023 ( 2007 ) .
h. hansen , w. m. alberico , a. beraudo , a. molinari , m. nardi and c. ratti , phys . rev . *
d 75 * , 065004 ( 2007 ) .
h. j. schulze , j. phys .
* g 21 * , 185 ( 1995 ) . e. quack , p. zhuang , y. kalinovsky , s. p. klevansky and j. hfner , phys . lett . * b 348 * , 1 ( 1995 ) .
d. horvatic , d. blaschke , d. klabucar and o. kaczmarek , arxiv:1012.2113 [ hep - ph ] . w. j. fu and y. x. liu , phys .
d 79 * , 074011 ( 2009 ) .
p. costa , m. c. ruivo , h. hansen and c. a. de sousa , phys .
d 81 * , 016007 ( 2010 ) .
k. kashiwa , h. kuono , m. matsuzaki and m. yahiro , phys . lett .
* b 662 * , 26 ( 2008 ) .
k. fukushima , phys .
d 78 * , 114019 ( 2008 ) . c. sasaki , b. friman and k. redlich , phys .
d 75 * , 054026 ( 2007 ) .
z. zhang and y. x. liu , phys . rev .
* c 75 * , 064910 ( 2007 ) .
a. ali khan at al .
, phys . rev . * d64 * , 074510 ( 2001 ) .
o. scavenius , a. mocsy , i. n. mishustin and d. h. rischke , phys . rev . * c 64 * , 045202 ( 2001 ) .
d. gomez dumm , d. b. blaschke , a. g. grunfeld , n. n. scoccola , phys .
* d73 * , 114019 ( 2006 ) .
d. blaschke , m. buballa , a. e. radzhabov , m. k. volkov , yad .
71 * , 2012 - 2018 ( 2008 ) . | the thermodynamic behavior of conventional nambu - jona - lasinio and polyakov - loop - extended nambu - jona - lasinio models is compared .
particular attention is paid to the phase diagram in the @xmath0 plane . |
after the first direct evidence for the point - like constituents in the nucleon@xcite , identified as the valence quarks with spin-1/2 in the naive constituent quark model ( nqm)@xcite , a lot of experiments have been conducted to probe the structure of the proton in the deep inelastic scattering ( dis ) experiments . surprisingly , the dis results in the early 80s@xcite indicated that the valence quarks of the proton carry only about 30% of its spin and is referred to as the `` proton spin crisis '' in the nqm .
these results provided the first evidence for the proton being composed of three valence quarks surrounded by an indistinct sea of quark - antiquark pairs ( henceforth referred to as the `` quark sea '' ) . in the present day
, the study of the composition of hadrons can be said to be primarily the study of the quark sea and gluons and is considered as one of the active areas in hadronic physics . the conventional expectation that the quark sea perhaps can be obtained through the perturbative production of the quark - antiquark pairs by gluons produces nearly equal numbers of @xmath4 and @xmath5 . until early 90 s a symmetric sea w.r.t . @xmath4 and
@xmath5 was assumed , however , the famous new muon collaboration in 1991@xcite established the quark sea asymmetry of the unpolarized quarks in the case of nucleon by measuring @xmath6 giving first clear evidence for the nonperturbative origin of the quark sea .
this was later confirmed by the drell - yan experiments@xcite which measured a large quark sea asymmetry ratio @xmath7 reminding us that the study of the quark sea is intrinsically a nonperturbative phenomena and it is still a big challenge to perform these calculations from the first principles of qcd .
one approach to account for the observed quark sea asymmetry is the pion cloud mechanism@xcite where the quark sea is believed to originate from process such as virtual pion production .
it is suggested that in the deep inelastic lepton - nucleon scattering , the lepton probe also scatters off the pion cloud surrounding the target proton .
the @xmath8 cloud , dominant in the process @xmath9 , leads to an excess of @xmath5 sea .
however , this effect should be significantly reduced by the emissions such as @xmath10 with @xmath11 cloud .
therefore , the pion cloud idea is not able to explain the significant @xmath12 asymmetry .
this approach can be improved upon by adopting a mechanism which operates in the _ interior _ of the hadron . the chiral constituent quark model ( @xmath0cqm)@xcite can yield an adequate description of the quark sea generation through the chiral fluctuations . the basic idea is based on the possibility that chiral symmetry breaking takes place at a distance scale much smaller than the confinement scale . in this region ,
the effective degrees of freedom are the valence quarks and the internal goldstone bosons ( gbs ) which are coupled to the valence quarks@xcite allowing a simple and intuitive method to investigate the principle features of the hadron structure . in the case of spin
dependent quantities , the @xmath0cqm is not only successful in giving a satisfactory explanation of `` proton spin crisis''@xcite but is also able to account for the baryon magnetic moments@xcite and hyperon @xmath13decay parameters@xcite .
however , in the case of quark distribution functions , the latest developments by the nusea ( e866)@xcite and hermes@xcite to determine the variation of the sea - antiquark ratio @xmath14 and the difference @xmath2 with bjorken scaling variable @xmath1 have renewed considerable interest in the quark sea asymmetries .
recently , there has been substantial theoretical progress to take into account the effects of quark sea in determining the flavor structure of the baryons and the question of sea asymmetry has been investigated by several authors using various phenomenological models .
calculations have been carried out in the meson cloud models@xcite , chiral quark - soliton model@xcite , effective chiral quark model@xcite , statistical models@xcite , bag model@xcite , model for parton densities@xcite , radiative parton model@xcite etc .. however , the inclusion of @xmath15dependence has not yet been successfully included in the quark distribution functions .
therefore , pending further experiments , it would be interesting to examine the flavor structure of the octet baryons at low energy , thereby giving vital clues to the nonperturbative effects of qcd .
the study of @xmath15dependence in the quark distribution functions becomes particularly interesting for the @xmath0cqm where the effects of quark sea and valence quarks can separately be calculated .
the purpose of the present communication is to determine the sea quark distribution functions and their asymmetries in the octet baryons by phenomenologically incorporating @xmath15dependence in the @xmath0cqm .
the extent of contributions coming from the different sea quarks for the octet baryons can also be compared . to understand the relation of the bjorken scaling variable and quark sea
, it would be significant to study its implications in the region @xmath16 which is a relatively clean region to test the quark sea structure as well as to estimate their structure functions and related quantities@xcite .
the key to understand the `` proton spin crisis '' , in the @xmath0cqm formalism , is the fluctuation process@xcite @xmath17 where @xmath18 constitute the `` quark sea''@xcite .
the effective lagrangian describing interaction between quarks and a nonet of gbs , can be expressed as @xmath19 where @xmath20 , @xmath21 and @xmath22 are the coupling constants for the singlet and octet gbs , respectively , @xmath23 is the @xmath24 identity matrix . in terms of the su(3 ) and axial u(1 ) symmetry breaking parameters , introduced by considering @xmath25 , @xmath26 and @xmath27@xcite , the gb field can be expressed as = ( + + & ^+ & k^+ + ^- & - + + & k^0 + k^- & |k^0 & - + ) .
the parameter @xmath28 ) denotes the probability of chiral fluctuation @xmath29 , whereas @xmath30 , @xmath31 and @xmath32 respectively denote the probabilities of fluctuations @xmath33 , @xmath34 , and @xmath35 . for the sake of simplification , the gb field
can also be expressed in terms of the quark contents of the gbs and is expressed as ( _ uu u |u+ _ ud d |d + _ us s |s & _ ud u |d & _ us u |s + _ du d |u & _ duu |u+ _ dd d |d + _ ds s |s & _
ds d |s + _ su s |u & _ sd s |d & _ su u |u + _ sd d |d + _ ss s |s + ) , where _ uu & = & _ dd= + + , _ ss = + , _ us = _ ds= _
su=_sd= - + , + _ du & = & _ ud= - + + , _
ud = _ du = 1,_us = _ ds=_su= _
the quark sea content of the baryon can be calculated in @xmath0cqm by substituting for every constituent quark @xmath36 , where @xmath37 is the transition probability of the emission of a gb from any of the @xmath38 quark and @xmath39 is the transition probability of the @xmath38 quark .
the flavor structure for the baryon of the type @xmath40 is expressed as @xmath41 quark distribution functions ---------------------------- there are no simple or straightforward rules which could allow incorporation of @xmath15dependence in @xmath0cqm . to this end , instead of using an _ ab initio _ approach ,
we have phenomenologically incorporated the @xmath15dependence getting clues from eichten _
et al._@xcite , isgur _
et al._@xcite and le yaouanc _ et al._@xcite .
the @xmath15dependent sea quark distribution functions can be now expressed as @xmath42 , @xmath43 , @xmath44 which together with the valence quark distribution functions give the flavor structure of the baryon as q^b(x)=q^b_val(x)+|q^b(x ) , [ totalquark]where @xmath45 . using the sea quark distribution functions as explained above , the quark sea asymmetries @xmath46 and @xmath3
can also be calculated at different @xmath1 values .
we have already discussed the inclusion of @xmath1-dependence in detail and compared our results with the experimental data for the case of nucleon in ref.@xcite . in the present communication
however , we have extended our calculations to the case of other octet baryons for which experimental data is not yet available .
the @xmath15dependence of the structure functions @xmath47 and @xmath48 can be calculated from f_2^b(x ) = x _ u , d ,
s e^2_q[q^b(x ) + |q^b(x ) ] , f_1^b(x ) = f_2^b(x ) , where @xmath49 is the charge of the quark @xmath38 ( @xmath50 and @xmath51 ) . in terms of the quark distribution functions ,
the structure function @xmath48 for any baryon can be expressed as f^b_2(x)= x(u^b(x)+ |u^b(x ) ) + x(d^b(x)+ |d^b(x)+ s^b(x)+|s^b(x ) ) .
several important quantities can be obtained from the structure functions of different isospin multiplets .
for example , for the case of proton and neutron we have = ( u_val^p(x ) - u_val^n(x ) + 2|u^p(x ) - 2|u^n(x ) ) + ( d_val^p(x ) . .
+ s_val^p(x ) - d_val^n(x)- s_val^n(x ) + 2|d^p(x)+ 2 |s^p(x ) - 2 |d^n(x)- 2 |s^n(x ) ) .
[ ig ] the gottfried integral @xmath52@xcite can be expressed in terms of the sea quarks as follows i^p n_g = _ 0 ^ 1 dx = + _ 0 ^ 1 dx , [ ipn]where we have used the following normalization conditions @xmath53 @xmath54 _ 0 ^ 1 |d^n(|u^n)(x)dx= _
0 ^ 1 |u^p ( |d^p)(x ) dx , _ 0 ^ 1 |s^n(x)dx= _ 0 ^ 1 following normalization conditions _ 0 ^ 1 u_val^^+(x)dx = 2 , _ 0 ^ 1 d_val^^+(x)dx = 0 , _ 0 ^ 1 s_val^^+ dx = 1 , + _
0 ^ 1 u_val^^0(x)dx = 1 , _ 0 ^ 1 d_val^^0(x ) dx=1 , _ 0 ^ 1 s_val^^0(x ) dx = 1 , + _ 0 ^ 1 u_val^^0(x ) dx = 1 , _ 0 ^ 1
d_val^^0(x ) dx = 0 , _ 0 ^ 1 s_val^^0(x ) dx = 2 , [ norm - s ] lead to the other gottfried integrals in terms of the sea quarks i^^+ ^0_g & = & + _ 0 ^ 1 dx , + i^^0 ^-_g & = & + _ 0 ^ 1 dx , +
i^^0 ^-_g & = & + _ 0 ^ 1 dx . [ ioctet]it is clear from eqs .
( [ ipn ] ) and ( [ ioctet ] ) , the flavor symmetric sea leads to the gottfried sum rule @xmath55 with @xmath56=@xmath57 .
after having detailed the contribution of the quark sea and the various asymmetries in the octet baryons of different quark structure , we now discuss the variation of these quantities with the bjorken variable @xmath1 . for the numerical calculation of the sea quark distribution functions of the octet baryons
, we have used the same set of input parameters as detailed in our earlier calculations@xcite .
one can easily find out that @xmath58 @xmath59 @xmath60 @xmath61 showing a clear quark sea asymmetry as observed in the dis experiments@xcite .
these distributions clearly indicate that our results pertaining to the quark sea asymmetry seem to be well in line with the expected results . as already mentioned in the introduction
, the sea quarks do not contribute at higher values of @xmath1 , therefore in this region , the contributions should be completely dominated by the valence quarks . the difference between the various sea distributions
is observed to be maximum at @xmath62 . as the value of @xmath1 increases , the difference between the sea contributions decreases in all the cases which is in line with the observations of other models@xcite .
the general aspects of the variation of the magnitudes of the sea quark distribution functions @xmath63 , @xmath64 and @xmath65 for the octet baryons are able to explain some of the well known experimentally measurable quantities , for example , @xmath66 , @xmath67 and the gottfried integral .
these quantities not only provide important constraint on a model that attempts to describe the origin of the quark sea but also provide a direct determination of the presence of significant amount of quark sea in the low @xmath1 region .
when @xmath1 is small , @xmath68 asymmetries are large implying the dominance of sea quarks in the low @xmath1 region .
in fact , the sea quarks dominate only in the region where @xmath1 is smaller than 0.3 . at the values @xmath69
, @xmath6 tends to 0 implying that there are no sea quarks in this region .
the contribution of the quark sea in the case of @xmath70 is particularly interesting because of its flavor structure which has equal numbers of @xmath71 , @xmath72 and @xmath73 quarks in its valence structure . unlike the other octet baryons , where the @xmath2 asymmetry decreases continuously with the @xmath1 values , the asymmetry in this case first increases and
then for values of @xmath74 it decreases .
however , it is interesting to observe the the asymmetry peak in this case which matches with our other predictions where the contribution of the quark sea is maximum at @xmath62 a measurement of the gottfried integral@xcite for the case of nucleon has shown a clear violation of gottfried sum rule from @xmath75 which can find its explanation in a global quark sea asymmetry @xmath76 .
similarly , for the case of @xmath77 , @xmath78 , and @xmath79 , the gottfried sum rules should read @xmath80 , @xmath81 and @xmath82 if the quark sea was symmetric .
however , due to the @xmath2 asymmetry in the case of octet baryons , a lower value of the gottfried integrals is obtained . in the case of nucleon
the results are in good agreement with the experimental data@xcite as already presented in@xcite .
the quality of numerical agreement in the other cases can be assessed only after the data gets refined .
further , this phenomenological analysis strongly suggests an important role for the quark sea at low value of @xmath1 .
new experiments aimed at measuring the flavor content of the other octet baryons are needed for profound understanding of the nonperturbative properties of qcd .
to summarize , in order to investigate the effects of `` quark sea '' , we have calculated the sea quark distribution functions for the octet baryons in the chiral constituent quark model ( @xmath0cqm ) .
the bjorken scaling variable @xmath1 has been incorporated phenomenologically to enlarge the scope of model and to understand the range of @xmath1 where quark sea effects are important .
implications of the quark sea have also been studied to estimate the quark sea asymmetries like @xmath2 , @xmath3 and gottfried integral .
the results justify our conclusion regarding the importance of quark sea at small values of @xmath1 . in conclusion ,
the results obtained for the quark distribution functions reinforce our conclusion that @xmath0cqm is able to generate qualitatively as well as quantitatively the requisite amount of quark sea .
this can perhaps be substantiated by a measurement of the quark distribution functions of the other octet baryons .
h.d . would like to thank the organizers , gribov80 workshop held at ictp and department of science and technology , government of india , for financial support .
n. isgur , g. karl and r. koniuk , phys .
lett . * 41 * , 1269 ( 1978 ) ; n. isgur and g. karl , phys .
d * 21 * , 3175 ( 1980 ) ; p. geiger and n. isgur , phys .
d * 55 * , 299 ( 1997 ) ; n. isgur , phys .
d * 59 * , 034013 ( 1999 ) .
emc collaboration , j. ashman _ et al .
b * 206 * , 364 ( 1988 ) ; nucl .
b * 328 * , 1 ( 1989 ) ; smc collaboration , b. adeva _
et al . _ ,
b * 302 * , 533 ( 1993 ) ; p. adams _ et al .
d * 56 * , 5330 ( 1997 ) ; e142 collaboration , p.l .
et al . _ ,
* 71 * , 959 ( 1993 ) ; e143 collaboration , k. abe _
et al . _ ,
. lett . * 75 * , 391 ( 1995 ) .
sullivan , phys .
d 5 * , 1732 ( 1972 ) ; a.w .
thomas , phys .
b * 126 * , 97 ( 1983 ) ; j. magnin and h.r .
christiansen , phys .
d * 61 * , 054006 ( 2000 ) .
s. weinberg , physica a * 96 * , 327 ( 1979 ) ; a. manohar and h. georgi , nucl .
b * 234 * , 189 ( 1984 ) .
h. dahiya and m. gupta , phys .
d * 64 * , 014013 ( 2001 ) ; * 67 * , 074001 ( 2003 ) ; int .
jol . of mod .
a , * 19 * 29 , 5027 ( 2004 ) ; h. dahiya , m. gupta , and j.m.s .
rana , int .
jol . of mod .
a , * 21 * , 21 4255 ( 2006 ) ; h. dahiya and m. gupta , phys .
d * 78 * , 014001 ( 2008 ) .
n. sharma , h. dahiya , p.k .
chatley , and m. gupta , phys . rev .
d * 79 * , 077503 ( 2009 ) ; n. sharma , h. dahiya and p.k .
chatley eur .
j. a * 44 * , 125 ( 2010 ) ; n. sharma , h. dahiya , p.k .
chatley and m. gupta phys .
d * 81 * , 073001 ( 2010 ) .
m. alberg , e.m .
henley , and g.a .
miller , phys.lett .
b * 471 * , 396 ( 2000 ) ; s. kumano and m. miyama , phys . rev .
d * 65 * , 034012 ( 2002 ) ; fu - guang cao and a. i. signal , phys .
d * 68 * , 074002 ( 2003 ) ; f. huang , r .-
xu , b .- q .
ma , phys .
b * 602 * , 67 ( 2004 ) ; b. pasquini , s. boffi , nucl . phys .
a * 782 * , 86 ( 2007 ) .
m. wakamatsu , phys .
d * 44 * , r2631 ( 1991 ) ; m. wakamatsu , phys . rev .
d * 46 * , 3762 ( 1992 ) ; h. weigel , phys .
d * 55 * , 6910 ( 1997 ) ; m. wakamatsu and t. kubota , phys .
d * 57 * , 5755 ( 1998 ) ; m. wakamatsu , phys . rev .
d * 67 * , 034005 ( 2003 ) .
trevisan , c. mirez , t. frederico , and l. tomio , eur .
j. c * 56 * , 221 ( 2008 ) ; yunhua zhang , lijing shao , and , bo - qiang ma , phys . lett .
b * 671 * , 30 ( 2009 ) ; yunhua zhang , lijing shao , and bo - qiang ma , nucl .
a * 828 * , 390 ( 2009 ) .
m. gl@xmath83ck , e. reya , and a. vogt , z. phys .
c * 67 * , 433 ( 1995 ) ; m. gl@xmath83ck , e. reya , m.stratmann , and w. vogelsang , phys .
d * 53 * , 4775 ( 1996 ) ; d. de florian , c.a
. garcia canal , and r. sassot , nucl .
b * 470 * , 195 ( 1996 ) . | the effects of nonperturbatively generated `` quark sea '' have been investigated to determine the flavor structure of the octet baryons . the chiral constituent quark model ( @xmath0cqm ) , which is known to provide a satisfactory explanation of the proton spin and related issues in the nonperturbative regime , is able to explain the qualitative generation of the requisite amount of quark sea .
the importance of quark sea has been studied at different values of the bjorken scaling variable @xmath1 by including it phenomenologically in the sea quark distribution functions .
the results for the quark sea asymmetries like @xmath2 , @xmath3 and gottfried integral for the octet baryons strengthen the significance of quark sea at lower values of @xmath1 . |
various methods related to @xmath9 decays have been proposed to determine the ut angle @xmath0 . within them , the one that gives the best error on @xmath0 is the dalitz method @xcite@xcite .
we propose to measure the angle @xmath0 using a dalitz analysis applied to neutral b mesons . in general , since the neutral b mesons mix , interference effects between @xmath10 and @xmath11 decay amplitudes in @xmath12 decays ( for instance into @xmath13 final states ) are studied for the determination of the combination of ut angles @xmath14 . in this case
the tagging technique and a time dependent analysis are required .
this can be avoided if the final states contain a particle which allows to unambiguously identify if a @xmath12 or @xmath15 has decayed . this is the case of neutral b mesons decaying into @xmath16 $ ] final states , where the flavor of the b meson can be determined through the sign of the electric charge of the kaon .
one of the relevant parameter in those kind of analysis is the ratio between the @xmath11 and @xmath10 amplitudes , which can be expressed by : @xmath17 the sensitivity to @xmath0 is proportional to the @xmath18 value . considering the ckm factors in the @xmath10 and @xmath11 transitions and the fact that in both cases the processes are mediated by a color suppressed diagram , we expect @xmath19 to be in the range [ 0.3 - 0.5 ] , larger than for the equivalent ratio in the charged b sector , which has been found to be of the order 0.1 @xcite .
in the analysis on data we would perform an unbinned maximum likelihood fit in order to extract the interesting quantities : the likelihood will contain a product of a yields pdf and a cp pdf ( only the latter has , within its variables , the weak phase @xmath0 ) . in our feasibility study , we assume the yields pdf to be a function of the classic babar analysis variable @xmath20 , where the asterisk denotes evaluation in the @xmath21 cm frame ] and of a variable that is able to discriminate between @xmath22 and continuum events ( @xmath23 events , with @xmath24 ) .
for the distributions of those variables we make some realistic assumptions inspired by babar analysis . in writing the cp pdf , we have an additional difficulty , with respect to the @xmath25 dalitz analysis .
this is due to the fact that the natural width of the @xmath26 is not small ( @xmath2750 ) and hence the interference with the other @xmath28 processes may not be negligible .
we follow the formalism introduced in @xcite and also used in the @xmath29 babar analysis in order to solve this problem .
we introduce the effective quantities @xmath30 , @xmath31 and @xmath32 defined as : @xmath33 where @xmath34 for the schwartz inequality and @xmath35 $ ] . in the limit of a @xmath362-body decay , such as @xmath37 we have : @xmath38 , @xmath39 ( the phase difference between the @xmath40 and the @xmath41 amplitudes of eq.[eq : rb ] ) and k=1 .
@xmath42 , @xmath43 are real and positive .
the subscript @xmath44 and @xmath45 refer to the @xmath46 and @xmath47 transitions , respectively .
the index @xmath48 indicates the position in the phase space of @xmath49 . applying this formalism
, we write the cp pdf as follows : @xmath50k^-\pi^+(k^+\pi^- ) ) \propto \nonumber \\
\label{eq : dalitzrate1}\end{aligned}\ ] ] where @xmath51 is the strong phase difference between @xmath52 and @xmath53 and @xmath54 , @xmath31 and @xmath32 are defined in eqs .
( [ eq : rs_square ] ) and ( [ eq : k ] ) .
we have simplified the notation using @xmath55 and @xmath56 .
the dalitz structure of the decay @xmath57 is well known and it has already been used in dalitz analysis in the charged b sector @xcite @xcite .
we performe a study to evaluate the possible variation of @xmath30 and @xmath31 on the b dalitz plot . for this purpose
we use a b dalitz model as suggested by the recent measurements @xcite , @xcite . the model assumed for the decay parametrizes the amplitude @xmath58 at each point @xmath31 of the dalitz plot as a sum of two - body decay matrix elements and a non - resonant term according to the following expression : @xmath59 we consider a region within @xmath60 48 mev from the nominal mass of the @xmath26(892 ) resonance and we obtain the distribution of @xmath30 , @xmath31 and @xmath61 by randomly varying all the strong phases between [ 0 - 2@xmath62 and the amplitudes up to @xmath63 of their nominal value ( shown in table [ tab : b0 ] ) . only for the @xmath64 resonance ,
the variation is up to @xmath65 of its nominal value .
.list of mass ( @xmath66 ) , widths ( @xmath67 ) and quantum numbers of the resonances considered in our model .
the last two columns present the chosen values of the coefficients @xmath68 for the @xmath41 and @xmath40 transitions respectively . note that the phase @xmath69 are not indicated and their choice is arbitrary . [
cols="<,^,^,^,^,^",options="header " , ] we conclude that , with the now available babar statistics , the measurement strategy would be extracting from the fit @xmath0 as a scan with respect to @xmath18 ( i.e. by performing a likelihood scan on @xmath0 for each value of @xmath18 with @xmath0 , @xmath70 and all the yields parameters floated in the fit ) . in this way we would extract from data the maximum possible information for the available statistics : an information on @xmath0 and not on @xmath18 .
the @xmath0 distribution obtained in that way , already very interesting on its own , would be very precious when combined with an experimental input for @xmath18 .
the coverage tests on toy - mc generated with different values of @xmath30 show that we have no bias in this fit configuration .
in this section we show , on one chosen toy - mc , what would be the impact of a real mesurement . in figure
[ fig : gammascan_rbfix_toy ] we show the output of our fit : a likelihood scan of @xmath0 with respect to @xmath30 .
this distribution , when combined with a fake @xmath30 measurement ( @xmath71 ) gives @xmath0 with an error of @xmath72 .
a measurement of @xmath30 with such an error is what we could expect to obtain , if @xmath73 , on @xmath74 data from an analysis of @xmath75 with the @xmath4 in flavor modes .
clearly , this is just a example on a chosen toy - mc and it is not necessarily representative of what we will find on data .
indeed we found , on toy - mc studies , that in a small percentage of cases we can be much less sensitive to @xmath0 .
that problem has been found to depend on the low number of signal events ( such that in some cases those events happen to be in poorly sensitive regions of the @xmath4 dalitz plot ) and it will therefore disappear with the increasing of the statistics . still , our test shows that we could have a result on @xmath0 that is competitive with the charged b dalitz analysis ( the analysis made on babar data gives today an error on @xmath0 of @xmath76 ) .
99 b. aubert _ et al . _ [ babar collaboration ] , arxiv : hep - ex/0607104 .
a. poluektov _ et al .
_ [ belle collaboration ] , phys .
d * 73 * , 112009 ( 2006 ) [ arxiv : hep - ex/0604054 ] .
m. bona _ et al . _ [ utfit collaboration ] , jhep * 0507 * , 028 ( 2005 ) [ arxiv : hep - ph/0501199 ] .
updated results available at http://www.utfit.org/ m. gronau , _ phys .
lett . _ * b557 * , 198 - 206 ( 2003 ) .
f. polci , m .- h .
schune and a. stocchi [ arxiv : hep - ph/0605129 ] .
g.cavoto et al .
, proceedings of the ckm 2005 workshop ( wg5 ) , uc san diego , 15 - 18 march 2005 , hep - ph/0603019 .
the parameterization given in eq .
( [ eq : dalitzrate1 ] ) can be rewritten in terms of the cartesian coordinates @xmath77 $ ] and @xmath78 $ ] as @xmath79\bar{x_s^0}(x_s^0 ) ) \propto$ ] + @xmath80 + y_{s\mp } im[f_{\mp}f_{\pm}^ * ] \right]$ ] | we present a feasibility study for a new analysis for extracting the angle @xmath0 of the unitarity triangle from the study of the neutral b meson decays .
we reconstruct the decay channel @xmath1 with the @xmath2 and the @xmath3 using a @xmath4 dalitz analysis technique .
the sensitivity to the angle @xmath0 comes from the interference of the @xmath5 and @xmath6 processes contributing to the same final state and by the fact that the @xmath7 can be unambiguosly identified through the sign of electric charge of the kaon from @xmath8 decay .
the impact of the result of such analysis is evaluated for the actual babar statistics . |
experiments @xcite have now confirmed that leptonic flavour is not conserved in nature : the historical observation of neutrino oscillations has changed our picture of fundamental physics . in contrast to the quark sector , neutrino oscillations are characterized by two large mixing angles @xcite .
it is natural to expect that lepton flavour violation ( lfv ) effects also take place among the electrically charged partners of neutrinos under the weak interaction @xmath3 .
the simplest and well - motivated way to induce neutrino lfv effects is through the exchange of neutral leptons involved in generating neutrino masses via various variants @xcite of the simplest type - i seesaw @xcite .
the basic feature of such seesaw picture is that neutrino masses arise only as a result of the exchange of heavy gauge singlet fermions through @xmath4 leading to an effective neutrino mass matrix @xmath5 in the @xmath6 basis , where @xmath7 denote the heavy 21 singlet right - handed neutrino states which are sequentially added to the standard model .
the smallness of neutrino mass follows naturally from the heaviness of @xmath7 . as an alternative to the simplest type - i seesaw , it has long been proposed that , thanks to the protecting @xmath8 global lepton number symmetry , the exchange of heavy neutral dirac fermions implied by the matrix @xmath9 ( in the basis @xmath10 ) will keep the neutrinos massless and yet allow for lfv effects .
this is the idea behind the so called inverse seesaw model @xcite ( for other extended seesaw schemes see , e.g. @xcite ) .
note that , to each of the isodoublet neutrinos @xmath11 two 21 isosinglets @xmath12 are added .
neutrinos get masses only when @xmath8 is broken , for example through a nonzero @xmath13 mass term .
thanks to the lepton number symmetry which arises as @xmath14 the magnitude of @xmath15 can be chosen to be small in a natural way , in the sense of t hooft @xcite . moreover , in specific models , the smallness of @xmath15 may be dynamically preferred @xcite .
after @xmath8 breaking the effective light neutrino mass matrix is given as @xmath16 so that , when @xmath15 is small , @xmath17 is also small , even when @xmath18 lies at the electroweak scale . in other words ,
the smallness of neutrino masses does not require superheavy physics .
we now turn to simple variants of the above schemes where the 21 singlet fermions @xmath7 are replaced by su(2 ) triplets @xmath19 @xcite .
the minimal type iii seesaw model is described by the lagrangian @xmath20 where @xmath21 denotes the hyperchargeless isotriplet fermion , @xmath22 and @xmath23 is the standard model higgs scalar doublet .
the effective neutrino mass matrix is fully analogous to eq .
( [ normal ] ) and its smallness requires a very large isotriplet fermion mass . the charged lepton mass matrix is a @xmath24 matrix , @xmath25 which is brought to diagonal form by a 6 @xmath26 6 unitary matrix @xmath27 of the same dimension , @xmath28 @xmath29 leading to three light fermions , namely @xmath30 and @xmath31 , and three heavy charged fermions @xmath32 with @xmath33 . in analogy with the matrix describing neutrino
nc interactions in general type - i and type - ii seesaw schemes introduced in ref .
@xcite we define the @xmath34 matrix as below @xmath35 the piece @xmath36 characterizes the nc lagrangian of charged leptons in the mass basis @xmath37 for finite @xmath38 values there are non - diagonal elements @xmath39 that induce tree level fcnc among the charged leptons @xmath40 .
in other words the mixing between different isospins implies the violation of the gim mechanism @xcite with amplitude of order @xmath41 where @xmath42 . and
@xmath43 decay branching ratios in normal type - iii seesaw.,title="fig:",scaledwidth=45.0%,height=207 ] and @xmath43 decay branching ratios in normal type - iii seesaw.,title="fig:",scaledwidth=45.0%,height=207 ] the smallness of neutrino masses implies that , for @xmath38 values accessible at the lhc , and barring fine - tuned parameter choices , the expected lfv rates are expected to be too small to be of phenomenological interest .
we have estimated the maximum attainable values for ( i ) the tree level lfv z vertex , which also induces the @xmath44 decays , and ( ii ) for the electromagnetic penguin vertices , which induce the radiative lfv decays and @xmath45 conversion . barring fine - tuning , one finds that they are far from the sensitivities expected in the upcoming lfv searches @xcite . as an example fig .
[ ref : fig1 ] illustrates the expected rates for lfv z - decay process .
similarly lfv processes involving taus are too small .
having discarded normal type - iii seesaw as an interesting model for lepton flavor violation , we turn instead to an inverse type - iii seesaw variant , characterized generically by the lagrangian @xmath46 where , as before , @xmath23 denotes the standard model higgs scalar doublet and now @xmath47 is a hyperchargeless scalar @xmath3-triplet , @xmath48 @xmath49 leading to @xmath50 this leads to six heavy states @xmath51 with @xmath52 and an effective three light majorana eigenstates @xmath53 with @xmath33 .
the light effective neutrino mass matrix is similar to that of the inverse seesaw model with isosinglet instead of isotriplets .
for an alternative inverse seesaw model with two lepton triplets see ref .
@xcite . ] .
@xmath54 the smallness of the parameter @xmath15 may also arise dynamically @xcite and/or spontaneously in a majoron - like scheme with @xmath55 where @xmath56 is a 21 singlet @xcite . in the latter case , for sufficiently low values of @xmath57 there may be majoron emission effects in neutrinoless double beta decay @xcite .
note that now the ratio @xmath58 need not be too small to reproduce acceptably small neutrino masses , since the latter vanish in the limit where the parameter @xmath15 goes to zero @xcite .
the smallness of @xmath15 is not only natural @xcite but also dynamically preferred in some cases @xcite .
the smaller the @xmath15 values the larger can be the @xmath59 .
this implies that when the mass of @xmath19 is accesible at lhc , say of the order of tev , one expects relatively large lfv decay rates .
in fact the situation is completely novel with respect to what one is used to , in the sense that lfv as well as cp violation effects survive even in the limit when neutrinos become massless @xcite @xcite .
clearly now fcnc effects can be naturally enhanced without conflict with the smallness of neutrino masses .
the neutrino mixing angles @xcite indicated by neutrino oscillation experiments @xcite should be explained from first principles . here
we consider the possibility of doing so in the framework of the inverse 21 seesaw mechanism . to this end
we adopt the attractive tribimaximal ( tbm ) ansatz for lepton mixing @xcite @xmath60 which provides a good first approximation to the values indicated by current neutrino oscillation data . here
we propose a simple @xmath0 flavor symmetry realization of the tbm lepton mixing pattern within the inverse type - iii seesaw scheme .
an @xmath0 realization of the tbm in inverse seesaw has already been studied in @xcite .
recall that @xmath0 is the group of the even permutations of four objects .
such a symmetry was introduced to yield @xmath61 and @xmath62 @xcite .
most recently @xmath0 has also been used to derive @xmath63 @xcite .
the group @xmath0 has 12 elements and is isomorphic to the group of the symmetries of the tetrahedron , with four irreducible representations , three distinct singlets @xmath64 , @xmath65 and @xmath66 and one triplet @xmath67
. for their multiplications see for instance ref .
the matter fields are assigned as in table [ tab2 ] .
+ .matter assignment for inverse seesaw model . [ cols="^,^,^,^,^,^,^,^",options="header " , ] decay into @xmath68 _ vs _ the branching of @xmath15 decay into @xmath69 for different values of @xmath70 , 0.6 ( dotted ) , 0.3 ( dot - dashed ) , 0.1 ( dashed ) and 0.03 ( continous ) and @xmath71 tev.,scaledwidth=45.0%,height=226 ]
we have proposed a new inverted version of type - iii seesaw or equivalently , a new type - iii version of the inverse seesaw mechanism . this way the physics responsible for neutrino masses can lie at low - scale and can be accessible at the large hadron collider ( lhc ) , due to : ( i ) the tev scale neutral fermions having large cross sections at the lhc and ( ii ) the tev scale neutral fermions inducing large lfv processes due to the low - scale violation of the glashow - iliopoulos - maiani mechanism , which implies potentially large tree - level fcnc involving charged leptons . by assuming an @xmath0-based underlying flavor symmetry we have implemented a tri - bimaximal lepton mixing pattern to account for the observed neutrino oscillation parameters .
we have studied the phenomenology of the resulting lfv decays and given the typical expectations for their magnitude , in addition to discussing the predictions for their relative rates . in fig.[fig : e3vseg ] we give the correlation between the branching of @xmath72 vs the branching of @xmath73 fixing @xmath74 and for different values of @xmath75 .
clearly neutral heavy fermion states can lie at the tev scale and their production cross section at the lhc is enhanced with respect to that expected in type - i inverse seesaw @xcite
. indeed the much larger production cross sections expected for the type - iii models should encourage detailed dedicated montecarlo simulations @xcite in order to scrutinize the viability of detecting the associated signals .
last but not least , given the underlying flavor symmetry predictions these should also take into account the details of flavor physics which will determine the expected decay pattern of the heavy leptons .
we thank martin hirsch and ioannis papavassiliou for useful comments .
this work was supported by the spanish grant fpa2008 - 00319/fpa and prometeo/2009/091 .
super - kamiokande collaboration , s. fukuda _ et al .
_ , phys . lett . *
b539 * , 179 ( 2002 ) , [ hep - ex/0205075 ] .
sno collaboration , q. r. ahmad _ et al .
lett . * 89 * , 011301 ( 2002 ) , [ nucl - ex/0204008 ] .
kamland collaboration , t. araki _ et al .
lett . * 94 * , 081801 ( 2005 ) .
t. kajita , new j. phys .
* 6 * , 194 ( 2004 ) .
k2k collaboration , m. h. ahn _
et al . _ ,
lett . * 90 * , 041801 ( 2003 ) , [ hep - ex/0212007 ] . t. schwetz , m. tortola and j. w. f. valle , new j. phys .
* 10 * , 113011 ( 2008 ) , [ 0808.2016 ] , for a review see m. maltoni , t. schwetz , m. a. tortola and j. w. f. valle , new j. phys . * 6 * , 122 ( 2004 )
. j. w. f. valle , j. phys .
* 53 * , 473 ( 2006 ) , [ hep - ph/0608101 ] , review based on lectures at the corfu summer institute , september 2005 .
m. gell - mann , p. ramond and r. slansky , ( 1979 ) , print-80 - 0576 ( cern ) ; t. yanagida , ( kek lectures , 1979 ) , ed .
sawada and sugamoto ( kek , 1979 ) .
r. n. mohapatra and g. senjanovic , phys . rev . *
d23 * , 165 ( 1981 ) .
j. schechter and j. w. f. valle , phys . rev . * d22 * , 2227 ( 1980 ) ; phys .
d * 25 * , 774 ( 1982 ) .
g. lazarides , q. shafi and c. wetterich , nucl .
b181 * , 287 ( 1981 ) .
r. n. mohapatra and j. w. f. valle , phys . rev . * d34 * , 1642 ( 1986 ) .
j. bernabeu _ et al .
_ , phys . lett . *
b187 * , 303 ( 1987 ) .
d. wyler and l. wolfenstein , nucl .
b218 * , 205 ( 1983 ) .
e. akhmedov , m. lindner , e. schnapka and j. w. f. valle , phys .
* d53 * , 2752 ( 1996 ) , [ hep - ph/9509255 ] . s. m. barr and i. dorsner , phys . lett .
* b632 * , 527 ( 2006 ) , [ hep - ph/0507067 ] . g. t hooft , lecture given at cargese summer inst . ,
cargese , france , aug 26 - sep 8 , 1979 .
f. bazzocchi , d. g. cerdeno , c. munoz and j. w. f. valle , 0907.1262 .
r. foot , h. lew , x. g. he and g. c. joshi , z. phys .
* c44 * , 441 ( 1989 ) . b. w. lee and r. e. shrock , phys . rev . * d16 * , 1444 ( 1977 ) . y. kuno , aip conf .
proc . * 542 * , 220 ( 2000 ) .
a. van der schaaf , j. phys . * g29 * , 1503 ( 2003 ) .
a. maki , aip conf .
proc . * 981 * , 363 ( 2008 ) .
e. ma , 0905.2972 .
m. c. gonzalez - garcia and j. w. f. valle , phys . lett . * b216 * , 360 ( 1989 ) .
z. g. berezhiani , a. y. smirnov and j. w. f. valle , phys .
b291 * , 99 ( 1992 ) , [ hep - ph/9207209 ] .
g. c. branco , m. n. rebelo and j. w. f. valle , phys .
b225 * , 385 ( 1989 ) ; n. rius and j. w. f. valle , phys .
* b246 * , 249 ( 1990 ) .
p. f. harrison , d. h. perkins and w. g. scott , phys .
b530 * , 167 ( 2002 ) , [ hep - ph/0202074 ] .
m. hirsch , s. morisi and j. w. f. valle , 0905.3056 .
e. ma and g. rajasekaran , phys . rev . * d64 * , 113012 ( 2001 ) , [ hep - ph/0106291 ] .
k. s. babu , e. ma and j. w. f. valle , phys . lett .
* b552 * , 207 ( 2003 ) , [ hep - ph/0206292 ] . m. hirsch , j. c. romao , s. skadhauge , j. w. f. valle and a. villanova del moral , phys . rev . *
d69 * , 093006 ( 2004 ) , [ hep - ph/0312265 ] . g. altarelli and f. feruglio , nucl .
b720 * , 64 ( 2005 ) , [ hep - ph/0504165 ] . g. altarelli and f. feruglio , nucl . phys .
* b741 * , 215 ( 2006 ) , [ hep - ph/0512103 ] .
e. ma , phys .
b671 * , 366 ( 2009 ) , [ 0808.1729 ] .
w. grimus and l. lavoura , jhep * 09 * , 106 ( 2008 ) , [ 0809.0226 ] .
a. zee , phys .
b630 * , 58 ( 2005 ) , [ hep - ph/0508278 ] .
m. hirsch , s. morisi and j. w. f. valle , phys . rev . *
d79 * , 016001 ( 2009 ) , [ 0810.0121 ] .
s. morisi , 0901.1080 ; g. seidl , 0811.3775 .
w. grimus and l. lavoura , jhep * 04 * , 013 ( 2009 ) , [ 0811.4766 ] . m. malinsky , j. c. romao and j. w. f. valle , phys .
lett . * 95 * , 161801 ( 2005 ) , [ hep - ph/0506296 ] , see also m. hirsch et al , phys . rev . * d75 * , 011701 ( 2007 ) , [ hep - ph/0608006 ] .
m. c. gonzalez - garcia and j. w. f. valle , mod .
a7 * , 477 ( 1992 ) .
a. ilakovac and a. pilaftsis , nucl
b437 * , 491 ( 1995 ) , [ hep - ph/9403398 ] .
f. deppisch and j. w. f. valle , phys . rev .
* d72 * , 036001 ( 2005 ) , [ hep - ph/0406040 ] .
f. deppisch , t. s. kosmas and j. w. f. valle , nucl .
b752 * , 80 ( 2006 ) , [ hep - ph/0512360 ] .
l. lavoura , eur .
j. * c29 * , 191 ( 2003 ) , [ hep - ph/0302221 ] .
m. dittmar _ et al .
_ , nucl . phys . *
b332 * , 1 ( 1990 ) ; m. c. gonzalez - garcia , a. santamaria and j. w. f. valle , nucl . phys . *
b342 * , 108 ( 1990 ) .
r. franceschini , t. hambye and a. strumia , phys . rev . *
d78 * , 033002 ( 2008 ) , [ 0805.1613 ] ; j. a. aguilar - saavedra , 0905.2221 . | we present a type - iii version of inverse seesaw or , equivalently an inverse version of type - iii seesaw . naturally small neutrino masses arise at low - scale from the exchange of neutral fermions transforming as hyperchargeless su(2 ) triplets . in order to implement tri - bimaximal lepton mixing we supplement the minimal 21 gauge symmetry with an @xmath0-based flavor symmetry .
our scenario induces lepton flavour violating ( lfv ) @xmath1 decays that can proceed at the tree level , while radiative @xmath2 decays and mu - e conversion in nuclei are also expected to be sizeable .
lfv decays are related by the underlying flavor symmetry and the new fermions are also expected to be accessible for study at the large hadron collider ( lhc ) . |
* experimental setup . * a femtosecond pulse laser ( duration of 100 fs and average power of 165 mw ) impinging on a bbo ( @xmath67-bab@xmath24o@xmath30 , thickness of 2 mm ) crystal generates two horizontally polarized photons at each of signal @xmath69 and idler @xmath70 modes , see fig .
[ fig : setup ] . to eliminate spectral and spatial correlations of the photons , interference filters ( ifs , full width at half maximum bandwidth of 3 nm centered at 780 nm ) and single - mode fibers ( smfs ) are used . by using a half - wave plate h@xmath13 at the angle of @xmath71 , the polarization state of idler photons
is rotated to vertical polarization just prior to the polarizing beam splitter ( pbs@xmath13 ) .
all the photons made to arrive at pbs@xmath13 simultaneously and the quantum state of the photons at the output of pbs@xmath13 is @xmath72 .
alice measures the single - photon reflected at the beam splitter ( bs@xmath13 ) by the single - photon detector ( d@xmath13 ) : the measurement basis is controlled by a phase shifter ( ps@xmath73 ) and a half wave plate ( h@xmath74 ) .
pbs@xmath24 is used for rsp of a pure three - photon entangled state ( eq . ( [ eq : three - photon ] ) ) at bob .
rsp of a three - photon mixed state ( eq . ( [ eq : mixed ] ) )
requires the use of the partial polarizer ( pp ) @xcite .
bob measures the three photons by coincidence detection on d@xmath24 , d@xmath29 , and d@xmath30 .
the measurement basis is controlled by using ps@xmath75 and h@xmath76 .
the phase shifters ps@xmath73 and ps@xmath75 are implemented by rotating a half - wave plate between two quarter wave plates set at the angle of @xmath71
. * chsh inequality test .
* we obtain the chsh parameter in eq .
( [ eq : chsh ] ) by measuring the correlations values from joint measurements on the single - photon @xmath77 and on the three photons @xmath78 : ^^ = , where @xmath79 is the frequency of outcomes @xmath80 by @xmath77 and @xmath81 by @xmath78 . for single - photon measurement
, ps@xmath13 is removed and pbs@xmath24 is used .
the outcomes @xmath82 and @xmath83 of @xmath84 correspond to a click on d@xmath13 when the angle @xmath23 of h@xmath24 is @xmath85 and @xmath86 , respectively . similarly , @xmath82 and
@xmath83 outcomes of @xmath87 are obtained by setting the angle @xmath23 of h@xmath24 with @xmath88 and @xmath89 , respectively . for three - photon measurement
, ps@xmath24 is removed , and coincidence clicks on d@xmath24 , d@xmath29 , and d@xmath30 are recorded : the outcomes @xmath90 are obtained by setting the angle @xmath28 of h@xmath29 at @xmath91 for @xmath92 and at @xmath93 for @xmath94 . | remote state preparation ( rsp ) provides an indirect way of manipulating quantum information based on the nonlocal effect of quantum measurement .
although rsp has been demonstrated in recent years to remotely prepare multi - photon states , quantum measurement on the same number of photons was required , i.e. , to prepare @xmath0-photon states via rsp , quantum measurement on the other n - photons was required , hence significantly limiting practicality and applicability of rsp . here
we report the first experimental demonstration of remote preparation of three - photon entangled states by measuring only a single - photon entangled with the three photons .
we further generalize our protocol to prepare multi - photon entangled states with arbitrary photon number and purity via single - photon measurement . as our rsp scheme relies on the nonlinearity induced by single - photon measurement , it enables quantum state engineering of multi - photon entangled states beyond the linear optical limit .
our results are expected to have significant impacts on quantum metrology and quantum information processing .
entanglement , once conceived as the ` weirdness ' of quantum mechanics @xcite , is now at the heart of quantum technologies , such as , quantum communication @xcite , quantum computing @xcite , and quantum metrology @xcite .
manipulation of quantum information based on the nonlocal nature of quantum measurement on entangled quantum systems can be categorized into two : quantum teleportation @xcite and remote state preparation ( rsp ) @xcite . in quantum teleportation
, alice can transfer an _ unknown _ qubit to bob by sending him two bits of classical information resulting from the bell state measurement @xcite , assuming alice and bob already share a pair of maximally - entangled qubits . in rsp , alice and bob
are still assumed to share a pair of maximally - entangled qubits , but alice can transfer a _ known _ qubit to bob by sending him only one bit of information from her measurement . moreover , as rsp does not require complete bell state measurement , it can be more readily scaled up for larger quantum systems . in photonic systems , rsp has recently been demonstrated for single - photons @xcite , two photons @xcite , and three photons @xcite . in these rsp schemes , however , quantum measurement on the same number of photons was required , i.e. , to prepare an @xmath0-photon state at bob , @xmath0-photon quantum measurement is required at alice , hence significantly limiting practicality and applicability of rsp .
however , rsp does not necessarily require alice to measure the same number of photons as the remotely prepared quantum state on bob .
in fact , if alice and bob share an entangled state consisting of a single - photon and multiple photons , single - photon measurement at alice would be sufficient to remotely prepare a multi - photon state at bob . in this work ,
we demonstrate remote preparation of various three - photon entangled states via single - photon measurement by preparing entanglement between a single - photon and three single - photons .
alice s measurement on the single - photon remotely prepares a three - photon entangled state in the form of @xmath1 at bob .
we also generalize our rsp protocol to prepare multi - photon entangled states of arbitrary photon number and purity via single - photon measurement .
our rsp protocol extends the capability of multi - photon state engineering beyond the linear optical limit via the nonlinearity induced by the single - photon measurement @xcite , allowing us to prepare various multi - photon states required for quantum metrology @xcite and for fundamental studies in quantum optics @xcite .
remote preparation of a three - photon entangled state by single - photon measurement requires alice and bob to share an entangled state between a single - photon and three single - photons of the form , |_ab = , [ eq : entangledstate ] where subscripts @xmath2 , @xmath3 , @xmath4 , and @xmath5 refer to alice , bob , horizontal polarization , and vertical polarization , respectively .
alice measures her single - photon in the basis @xmath6 , where @xmath7 with real parameters @xmath8 , and @xmath9 .
note that @xmath10 .
when alice s single - photon is measured in the basis @xmath11 , bob s three - photon state is then projected onto |_b = |2_h,1_v_b + e^i at alice can remotely prepare the desired three - photon entangled state at bob . to prepare entanglement between a single - photon and three photons
, we employ the experimental setup shown in fig .
[ fig : setup ] .
four photons in @xmath12 impinge on a non - polarizing beam splitter bs@xmath13 , and when a single - photon is reflected to alice and the other three photons are directed to bob ( assured by detecting the corresponding numbers of photons at alice and bob ) , the single - photon and the three photons are in the entangled state in eq .
( [ eq : entangledstate ] ) . to verify entanglement between the single - photon and the three photons
, we test the clauser - horne - shimony - holt ( chsh ) inequality @xcite .
the hilbert space for the quantum state under the test has the dimension 2@xmath142 because , for the single - photon , the basis is @xmath15 and , for the three photons , the basis is @xmath16 .
the chsh parameter is then defined as s _
= |-^^ + ^^ + ^^ + ^^| , [ eq : chsh ] where @xmath17 and @xmath18 are measurements on the single - photon , and @xmath19 and @xmath20 are measurements on the three photons .
the relevant operators are defined as , ^_z & = & |1_h,0_v-|0_h,1_v , + ^_x & = & |1_h,0_v+|0_h,1_v , + ^_z & = & |2_h,1_v-|1_h,2_v , + ^_x & = & |2_h,1_v+|1_h,2_v .
[ eq : sigma ] in experiment , @xmath21 violates the classical limit of 2 by more than 7 standard deviations , summarized in table [ tab : chsh ] .
see methods for details of measurements to obtain @xmath21 . .
* test of chsh inequality between the single - photon and the three photons .
* experimentally measured normalized correlation values and the chsh parameter @xmath21 .
the measurement basis for the single - photon is @xmath15 and that for the three photons is @xmath16 .
thus , the hilbert space dimension for the state under the test is @xmath22 .
the experimental @xmath21 value violates the classical limit of 2 by more than 7 standard deviations , a clear indication that the single - photon and the three photons are entangled .
errors represent one standard deviation . [
cols="^,^,^,^,^",options="header " , ] using the entanglement between the single - photon and the three photons , we carry out rsp of a three - photon entangled - state with an adjustable phase @xmath9 , i.e. , |_b^ = ( |2_h,1_v_b + e^i |1_h,2_v_b ) , [ eq : three - photon_phase ] which is useful in quantum metrology with lossy channels @xcite and in characterizing the kinds of quantum decoherence @xcite . to do this ,
we employ the single - photon measurement at alice in fig . [
fig : setup ] , where the angle @xmath23 of a half wave plate h@xmath24 is set to @xmath25 , the phase shift @xmath9 of ps@xmath13 is adjusted .
a click on d@xmath13 heralds the successful preparation of the three - photon entangled state in eq .
( [ eq : three - photon_phase ] ) . for measuring the phase @xmath9 in the remotely prepared three - photon state , we employ the measurement setup at bob in fig .
[ fig : setup ] , which projects the three - photon state onto @xmath26 ; the phase @xmath27 of ps@xmath24 is varied while the angle @xmath28 of h@xmath29 is set to @xmath25 , and coincidence counts on d@xmath24 , d@xmath29 , and d@xmath30 are recorded . by this measurement , the projection probability for the state @xmath31 is @xmath32 , where the offset of sinusoidal probability as a function of @xmath27 indicates the phase @xmath9 in the three - photon state .
figures [ fig : data](a - c ) show three - photon entangled states having various phases by choosing different @xmath9 values at alice : for @xmath33 , the offset of the sinusoidal oscillation is 0 , but as @xmath9 is adjusted , the offset is shifted accordingly .
we next carry out rsp of a three - photon state at bob having a varying degree of entanglement , i.e. , |_b^ = 2|2_h,1_v_b + 2|1_h,2_v_b .
[ eq : three - photon_amp ] to carry out this rsp , the measurement setup at alice is modified : ps@xmath13 is removed , the angle @xmath23 of h@xmath24 is adjusted , and a click at d@xmath13 is informed to bob . at bob s side , as it is necessary to observe the change of amplitudes in a three - photon state , the projection basis @xmath34 is used .
to do so , ps@xmath24 is removed and coincidence counts at d@xmath24 , d@xmath29 , and d@xmath30 are measured as a function of the angle @xmath28 of h@xmath29 . the projection probability of the state @xmath35 by this measurement is @xmath36 , and thus , the amplitude in the three - photon state can be obtained by finding the maximum probability , which takes place at @xmath37 ( modulo @xmath38 ) .
experimental results in fig .
[ fig : data](d - f ) show the changes of the amplitudes in the three - photon state depending on the single - photon measurement at alice .
additionally , we have confirmed that contributions of @xmath39 and @xmath40 in the generated states are negligible , as shown in fig .
[ fig : contributions ] . , @xmath41 , @xmath42 , and @xmath43 states in experimentally generated three - photon states . *
the maximum contribution from @xmath44 and @xmath43 is below 0.03 .
this result also justifies the choice the hilbert space dimension @xmath22 as contributions from @xmath44 and @xmath43 are negligible .
error bars represent one standard deviation .
, width=325 ] it is important to emphasize that the set of three - photon states in eq .
( [ eq : three - photon_amp ] ) can not be fully prepared by linear optical transformation of any particular three - photon states .
for example , any nontrivial linear optical transformations on @xmath45 necessarily generate undesired quantum states @xmath46 and/or @xmath47 , so that the state @xmath48 can not be prepared . on the other hand , as rsp is based on measurement - induced nonlinearity @xcite , it enables us to access multi - photon states on different orbits determined by linear optical transformations @xcite .
our rsp protocol , thus , extends the capability of multi - photon state engineering beyond the linear optical limit via the nonlinearity induced by the single - photon measurement @xcite , allowing us to prepare various multi - photon states required for quantum metrology @xcite and for fundamental studies in quantum optics @xcite .
we have so far experimentally demonstrated rsp of pure three - photon states , but the scheme can be further generalized to prepare general multi - photon entangled states of arbitrary purity and photon number via single - photon measurement . to prepare a mixed three - photon state at bob , alice uses a partial polarizer pp @xcite instead of pbs@xmath24 in fig .
[ fig : setup ] .
this then implements a partial projection on the single - photon at alice of the form , ( |_a|,p ) = p |_a| + ( 1-p)i_a /2 , [ eq : partialpol ] where @xmath49 is the strength of the projection and @xmath50 is the identity density operator @xmath51 @xmath52 .
consequently , the three - photon state at bob becomes a mixed state of the following form , _ b & = & + & = & p |_b| + ( 1-p ) i_b /2 , [ eq : mixed ] where @xmath53 @xmath54 .
in other words , an arbitrary amount of noise ( @xmath55 ) can be added to any three - photon pure state @xmath56 via alice s partial projection measurement .
note that a pure three - photon state can not be transformed to such a mixed state by applying the distinguishability - based decoherence schemes @xcite , which is commonly adopted to generate a mixed single - photon state .
this is due to the fact that the distinguishability - based decoherence schemes cause emergence of non - trivial multi - photon state components @xcite ; e.g. , introducing @xmath57 , @xmath58 , and @xmath59 , where @xmath60 represents the vertically polarized @xmath61-photon state in a physically distinguishable mode ( i.e. , time , path , etc ) from that of @xmath62 .
now , for preparing a multi - photon state with arbitrary photon number , we consider @xmath61-pair photon generation in the spontaneous parametric down - conversion .
the quantum state at the output of pbs@xmath13 in fig .
[ fig : setup ] is then given as @xmath63 .
if we then consider , after bs@xmath13 , the case in which a single - photon is reflected to alice and 2@xmath61 - 1 photons are directed to bob , the quantum state shared by alice and bob is an entangled state of the form , & & |^(2n)_ab = 1/ + & & ( |1_h,0_v_a |n-1_h , n_v_b + |0_h,1_v_a |n_h , n-1_v_b ) . [ eq : entangled2n ] to carry out rsp at bob , alice measures the single - photon in the projection basis @xmath11 . this will then project the quantum state of @xmath64 photons at bob to become , @xmath65 where the amplitudes @xmath66 and @xmath67 as well as the phase @xmath9 are determined by the single - photon measurement at alice . to prepare a mixed state , similarly to above , it is necessary to introduce partial projection measurement @xmath68 in eq .
( [ eq : partialpol ] ) .
this then results in the multi - photon state of the form , ^(2n-1)_b & = & p |^(2n-1)_b| + ( 1-p ) i^(2n-1)_b /2 , [ eq:2n-1mixed ] where i^(2n-1)_b & = & |n_h,(n-1)_v_b_h,(n-1)_v| + & + & |(n-1)_h , n_v_b(n-1)_h , n_v| .
the rsp by single - photon measurement can , therefore , be applied to multi - photon states , which also provides enhanced engineering capability on multi - photon states . in summary , we report the first experimental demonstration of remote preparation of three - photon entangled states by measuring only a single - photon entangled with the three photons .
we have also generalized the rsp protocol to prepare a multi - photon entangled state with arbitrary photon number and purity via single - photon measurement .
in addition to fundamental interest , our rsp protocol extends the capability of multi - photon state engineering beyond the linear optical limit via the nonlinearity induced by the single - photon measurement @xcite , allowing us to prepare various multi - photon states required for quantum metrology @xcite and for fundamental studies in quantum optics @xcite .
we further anticipate that our work will stimulate fundamental studies on entanglement between a single particle and multiple particles , such as , nonlocality tests between a single - particle state and a multiple - particle state , quantum teleportation of a single - particle state to a multi - particle state , etc .
this work was supported by the national research foundation of korea ( grant no .
2013r1a2a1a01006029 ) .
we thank s .- w .
lee for helpful discussions . |
consider the abelian - higgs field theory on a background surface @xmath0 with metric @xmath1 . at critical coupling the static energy functional satisfies a bogomolny bound @xmath2 where the topological invariant @xmath3 ( the ` vortex number ' ) is the number of zeros of @xmath4 counted with multiplicity @xcite . in the notation of @xcite we have taken @xmath5 .
equality in is attained when the fields satisfy the bogomolny vortex equations , which are obtained by completing the square in . in complex
coordinates @xmath6 these are @xmath7 this set of equations has smooth vortex solutions . as we explain in section [ introduction ] , analytical results are most readily obtained when @xmath0 is hyperbolic , having constant negative curvature @xmath8 , a case which is of interest in its own right due to the relation between hyperbolic vortices and @xmath9-invariant instantons , @xcite . the aim of this paper is to construct solutions to the vortex equations on _ compact _ surfaces @xmath0 whose universal cover has a hyperbolic metric . in section [ hypertilings ]
we will see how such surfaces can be obtained as quotients of the hyperbolic plane .
the problem of finding one vortex on @xmath0 is then equivalent to finding a regular lattice of vortices in the hyperbolic plane , which may be of interest as a hyperbolic version of the abrikosov vortex lattice observed experimentally in superconductors @xcite .
our approach involves the construction of the higgs field @xmath4 from a holomorphic function @xmath10 satisfying certain periodicity conditions . the bulk of this paper ( section [ solutions ] )
is concerned with constructing this map for an especially symmetric genus @xmath11 surface , the bolza surface .
section [ highergenus ] extends the construction to certain higher genus surfaces and we wrap up in section [ conclusions ] with some ideas for future work .
the bogomolny equations for abelian - higgs vortices at critical coupling on a curved surface can be combined to give the taubes equation @xmath12 the right hand side provides sources for the vortices , where the higgs field @xmath4 vanishes .
the higgs field can be written as @xmath13 with @xmath14 and @xmath15 real functions of @xmath16 and @xmath17 .
the phase @xmath15 depends on the gauge choice and increases by @xmath18 around a unit charge vortex .
expressed in terms of these functions , the components of the @xmath19-form gauge potential and magnetic flux density are @xmath20 where @xmath21 is the naive euclidean laplacian .
+ + it was first observed by witten , @xcite , that on a background of constant negative gaussian curvature @xmath22the taubes equation becomes the liouville equation and can be solved exactly .
we work with the poincar disk model of the hyperbolic plane , with metric @xmath23geodesics for this metric are circular arcs which intersect the boundary of the unit disk at right angles , and the geodesic distance between two points is @xmath24 the general solution to the taubes equation constructed by witten is a higgs field @xmath25 so @xmath26 is a ratio of two hyperbolic metrics of equal curvature @xcite , @xmath27 the function @xmath28 is holomorphic with @xmath29 on the interior of the disk and @xmath30 on the boundary @xmath31 , so @xmath32 is a map from the poincar disk to itself .
the map has ramification points at the vortex positions , where , for a vortex of unit multiplicity , @xmath4 has a simple zero and @xmath10 behaves locally quadratically . solving the taubes equation on hyperbolic space thus boils down to finding a holomorphic map @xmath33 with ramification points of order one greater than the multiplicity of @xmath4 at each vortex position . the schwarz - pick lemma guarantees that @xmath34 on the interior of the disk , with @xmath35 everywhere on the boundary ( where the higgs field takes its vacuum value ) .
let us consider the case where the domain of @xmath36 is a compact surface @xmath37 of area @xmath38 and genus @xmath39 .
then the target surface @xmath40 has a finite area which is obtained by pulling back the area form on @xmath41 by @xmath36 .
integrating the taubes equation over @xmath0 and using the topological invariant @xmath42 gives @xmath43 where @xmath3 is the number of zeros of @xmath4 counted with multiplicity .
positivity of @xmath44 implies the bradlow bound on the number of vortices on a compact surface : @xmath45 where the relation between the area and genus of the surface @xmath0 follows from the gauss - bonnet theorem @xmath46 .
the vacuum solution with @xmath47 has @xmath48 everywhere , while at the bradlow limit @xmath49 .
the interpretation is that each vortex takes up an effective area @xmath18 .
, the bradlow inequality becomes @xmath50 .
however , this has the effect of changing the curvature of the target space to @xmath51 , and the taubes equation is no longer integrable . ]
it is important to check that there is no contribution from the cross term arising when completing the square in .
this term is a total derivative and vanishes identically on a compact surface .
multivaluedness of @xmath36 means that the area of each sheet of @xmath41 is in fact the value of @xmath44 given in divided by one plus the number of vortices on the interior of @xmath0 .
this is in essence the riemann - hurwitz theorem and will be important in sections [ solutions ] and [ highergenus ] , where @xmath41 is considered as a tessellation of @xmath52 . for a finite number of vortices , @xmath3 , on @xmath52 ,
the function @xmath10 is a blaschke product @xcite @xmath53 in general the relation between the complex numbers @xmath54 and the vortex positions ( at the critical points of @xmath36 ) is not obvious .
constructing a vortex solution from the vortex positions can only be done analytically for symmetric configurations , such as a cyclic ring of vortices .
the parameters needed for an arbitrary prescribed vortex arrangement must be found numerically .
a vortex solution on the hyperbolic cylinder @xmath55 was presented in @xcite .
the solution was constructed via a map @xmath36 from a half annulus in the upper half plane to a euclidean rectangle , giving rise to a doubly periodic function . in poincar
disk coordinates , @xmath36 is given by a ratio of jacobi elliptic functions , @xmath56 vortices ( i.e. zeros of @xmath57 ) are located at @xmath58 with @xmath59 and @xmath60 , where @xmath61 is the complete elliptic integral of the first kind with elliptic modulus @xmath62 , and @xmath63 .
notice that there is a double zero of @xmath10 at every other zero of @xmath4 .
this suggests that @xmath36 can be expressed as an infinite blaschke product @xmath64 equality between and is easily checked at @xmath58 by known infinite products .
the _ blaschke condition _
@xmath65 holds for @xmath66 ( the sum can be evaluated as a @xmath67-hypergeometric function ) , thereby ensuring convergence of the product . using the closed form expression , the higgs field near the vortex at the origin and at the midpoint between vortices is @xmath68periodicity of @xmath69 follows from the fact that a period shift in @xmath16 is equivalent to a fractional linear transformation of @xmath36 : @xmath70a symmetry which will be discussed in more detail in section [ fproperties ] .
the hyperbolic plane has infinitely many tessellations by regular polygons whose edges are geodesic arcs .
such tessellations are denoted by their schlfli symbol @xmath71 , where @xmath72 is the number of sides of the fundamental polygon and @xmath67 the number of such polygons meeting at each vertex .
these polygons tile the hyperbolic plane if @xmath73 and , in contrast to the case of tessellations of the euclidean plane , have a fixed area @xmath74 where in our conventions @xmath8 .
we follow @xcite in defining the euler characteristic of any regular tessellation to be @xmath75 as discussed below , a given tessellation describes a smooth compact surface of genus @xmath39 if @xmath15 is a negative even integer , @xmath76 . the surface is obtained by identifying pairs of edges of the fundamental polygon .
+ + the uniformization theorem guarantees that a compact surface @xmath0 of genus @xmath77 possesses a global hyperbolic metric .
the universal covering space of such a surface is the hyperbolic plane , with a tessellation whose fundamental domain is obtained as the quotient @xmath78 , where @xmath79 is the first homotopy group of the surface .
the usual application of the gauss - bonnet theorem tells us that the area of a compact genus @xmath39 surface with curvature @xmath8 is @xmath80 .
note from that a tessellation with @xmath81 has the correct area for a surface of genus @xmath39 . for given @xmath39 ,
there is a @xmath82-complex - dimensional family of surfaces parametrised by the lengths of @xmath82 non - intersecting closed geodesics on the fundamental polygon , together with ` twist ' data describing the gluing of the edges .
these parameters are known as the fenchel - nielsen coordinates on teichmller space , @xcite .
for ease of computation we specialise to the most symmetric surfaces , corresponding to tessellations by regular polygons . as an example
, a smooth genus @xmath11 surface can be constructed from any of the tessellations @xmath83 of the tessellations above , our focus will be mainly on the highly symmetric tessellation by regular octagons in the @xmath84 configuration with opposite edges identified .
such a tessellation is obtained as the quotient of the unit disk by the freely acting subgroup @xmath85 of the full automorphism group @xmath86 of the tessellation .
@xmath87 is a fuchsian group generated by eight hyperbolic mbius transformations @xmath88 .
this is a discrete subgroup of @xmath89 with the fixed - point - free condition @xmath90 .
note that @xmath86 also contains elliptic elements with fixed points in the interior of the disk ; these elements are excluded from the quotient in order to obtain a smooth surface .
if we orient the tessellation such that there are octagon centres at the origin and along the coordinate axes , then the eight mbius transformations @xmath91 with @xmath92 take the form @xmath93 where @xmath94 is the euclidean distance from the origin to one of the neighbouring octagon centres and we have the relation @xmath95 .
this identification of the tessellation amounts to defining the fundamental octagon , @xmath96 , by @xmath97 where the hyperbolic distance function was defined in .
note that hyperbolic tessellations do not , in general , correspond to compact surfaces , but a fundamental polygon can still be defined as in , and the tessellation has a fractional euler characteristic defined by . the fundamental polygon of the regular @xmath84 tessellation with opposite edges identified is the most symmetric genus @xmath11 surface , known as the bolza surface .
it has an automorphism group of order @xmath98 ( @xmath99 if one includes reflections ) and can be thought of as a double cover of @xmath100 with ramification points at the six vertices of an inscribed regular octahedron . on the hyperbolic octagon
, the ramification points correspond to points of @xmath101 symmetry .
there are also @xmath102 points with @xmath103 symmetry and @xmath104 with @xmath105 symmetry , as can be seen by dividing the octagon into @xmath99 triangles with angles @xmath106 .
most of this symmetry is broken when a vortex is placed on the surface . however ,
when the vortex is placed at the origin , the @xmath101 symmetry about both the origin and the vertex is preserved , which will be key to our analytical construction . the fundamental domain for the @xmath84 tessellation ,
together with the positions of the symmetric points , is sketched in figure [ figoct ] . as a hyperelliptic curve
, the genus @xmath11 bolza surface can be expressed as @xmath107we will return to the representation of the bolza surface as a hyperelliptic curve in the context of compact surfaces of higher genus in section [ highergenus ] .
tessellation of the hyperbolic plane ( right ) where opposite edges of the fundamental octagon are identified .
there is a @xmath101 symmetry at each of the branch points , which are indicated by dots .
the shaded equilateral triangle with angles @xmath108 on the sphere is covered by the two shaded equilateral triangles with angles @xmath109 on the bolza surface ( one for each branch ) .
the north and south poles of the sphere correspond to the centre and vertex of the octagon , and hence to antipodal points on the bolza surface . ]
tessellation of the hyperbolic plane ( right ) where opposite edges of the fundamental octagon are identified .
there is a @xmath101 symmetry at each of the branch points , which are indicated by dots .
the shaded equilateral triangle with angles @xmath108 on the sphere is covered by the two shaded equilateral triangles with angles @xmath109 on the bolza surface ( one for each branch ) .
the north and south poles of the sphere correspond to the centre and vertex of the octagon , and hence to antipodal points on the bolza surface . ]
the aim of this section is to compute the higgs field of a vortex placed at the centre of the period octagon . as described in section [ introduction ] , we proceed by looking for a holomorphic function @xmath33 ramified at the centre of every octagonal cell in the tessellation . restricted to the fundamental octagon , the map @xmath28 is conformal except at the vortex position , where angles are doubled .
the image is precisely a square in the @xmath110 tessellation , and the map @xmath36 is two - to - one , providing a double cover of the square with double winding around the centre .
pairs of identified opposite edges of the octagon map to the same edge of the square .
it should be noted that opposite edges of the square are not identified , as the @xmath110 tessellation does not correspond to a compact surface .
a mbius transformation mapping one vortex to the next corresponds to a rotation about one of the four edge centres of the fundamental square , @xmath111 which we deduce by considering the effect of @xmath36 on the vertices of the octagon .
the fact that @xmath36 transforms by a mbius transformation shows that it is _ not _ , as might have been expected , an automorphic form . instead ,
if @xmath16 and @xmath10 both transform by fractional linear transformations @xmath112with @xmath113 , then @xmath114 transforms as @xmath115 for the specific case in question , with the fractional linear transformations and , we have @xmath116the factor multiplying @xmath114 can be thought of as a gauge transformation of the vortex fields , which does not affect @xmath69 . despite this complicated transformation property of @xmath36 and @xmath114
, we will see in section [ mappingbetweenpolygons ] that automorphic forms do in fact appear in our computation of the vortex fields . the area of the image square follows from the riemann - hurwitz theorem .
denote by @xmath117 and @xmath118 the areas of the fundamental polygons in the domain and image tessellations , which are related to the euler characteristic of the tessellation via . the map @xmath36 has a single ramification point of order @xmath11 , so @xmath119 in agreement with , which gives @xmath120 and @xmath121 .
more generally , if @xmath3 vortices are placed in the interior of a domain polygon representing a compact genus @xmath39 surface , then the map @xmath36 has @xmath3 ramification points and is of degree @xmath122 . by the riemann - hurwitz theorem ,
the area of the image polygon is given by @xmath123 the requirement that @xmath124 implies that the number of vortices permitted on a genus @xmath39 surface is constrained by @xmath125 the bradlow bound again .
the map we have described between the octagon and the square can hence be generalised to more vortices on a higher genus surface ( see also sections [ comphiggsfield ] and [ highergenus ] ) . before proceeding any further , we should investigate whether the map @xmath36 can be expressed as an infinite blaschke product with zeros of @xmath36 occurring at the same density as vortices , mimicking the way this was done for the one - dimensional vortex array ( [ jacobi ] , [ infprod ] ) . for @xmath36 of the form with @xmath126 , this will be the case if the blaschke condition holds .
consider a circle @xmath127 in the poincar disk with @xmath128 .
as each octagon in the tessellation has area @xmath129 , the interior of the circle will contain @xmath130 vortices .
then the number of vortices in an annulus of thickness @xmath131 is @xmath132 . using this measure , the sum can be converted to an integral if we assume that the @xmath133 have the density of the tessellation : @xmath134the right hand side diverges logarithmically
, hence there is no convergent blaschke product with the symmetries of the octagonal tessellation . for the hyperbolic cylinder discussed in section [ introduction ] the number of vortices as a function of distance from the origin is @xmath135 , so @xmath136 and the corresponding integral converges .
the map @xmath36 between the polygonal tessellations of the poincar disk described in section [ fproperties ] is easy to implement because the position of the vortex is consistent with a regular tiling of the fundamental polygon by congruent triangles with a shared vertex at the vortex position .
this procedure avoids mapping directly between the octagon and the square , which would introduce a large number of additional _ accessory parameters _ which would then need to be determined .
we thus divide the fundamental octagon into @xmath102 triangles , each of which maps , via an auxiliary half plane , to an eighth of the image square .
the fundamental domains @xmath137 ( with poincar disk coordinate @xmath16 ) and @xmath138 ( with coordinate @xmath139 ) are shown in figure [ mapf ] . by schwarz reflection
, the hyperbolic plane is covered by repeated reflection of the triangle @xmath137 in its edges . on a triangle @xmath137 within the fundamental octagon @xmath96 , via an auxiliary upper half plane .
the image square and triangle are denoted @xmath140 and @xmath138 . in our conventions ,
@xmath141 maps to the vertex of the shaded triangles at the origin of the tessellations , while @xmath142 and @xmath143 map to the other two vertices , respectively the edge midpoint and vertex of the polygons .
the real axis of the upper half plane maps to the boundary of the triangle .
the angles of the black triangles are @xmath144 and @xmath145 .
the higgs field at a vertex of the octagon is easier to compute if we instead use the dashed triangles , which has the effect of swapping @xmath146 and @xmath147 . ]
the existence of the maps in figure [ mapf ] is guaranteed by the riemann mapping theorem . in the following paragraphs
we sketch how the riemann map is constructed for curvilinear polygons before specialising to the case of triangles , where one can obtain explicit results .
then in section [ comphiggsfield ] we will focus on the specific triangles in figure [ mapf ] .
the riemann mapping theorem states the existence of a unique conformal map between the upper half plane and the interior of a simply connected domain of @xmath148 .
specialising to the case of an @xmath149-sided polygon bounded by circular arcs , the map is bijective and extends to a map from the boundary of the upper half plane to the boundary of the polygon .
it is conformal everywhere except at the @xmath149 points @xmath150 on the real axis which map to the vertices of the polygon .
the requirement that the image of the map should extend to the entire poincar disk by a series of reflections leads to an @xmath151-invariant differential equation satisfied by the function @xmath152 representing the map .
this third order non - linear ordinary differential equation is solved by a ratio of linearly independent solutions to a second order linear ordinary differential equation with @xmath149 regular singular points at @xmath153 , where the exponents depend on the angle at each corresponding vertex of the polygon .
the interested reader is referred to @xcite for details . from now on we specialise to the case of triangles , in which case the riemann map is known as the schwarz triangle map . in order to make use of the riemann mapping theorem we introduce an auxiliary upper half plane with complex coordinate @xmath154 .
then there are two schwarz triangle maps , @xmath155 and @xmath156 , mapping the upper half plane to each of the triangles in figure [ mapf ] .
the map @xmath36 between the triangles @xmath137 and @xmath138 is given by the composition @xmath157 .
note that when the images are extended to the entirety of the poincar disk ( by reflections both in the triangle edges and from the auxiliary upper half plane to the lower half plane ) , the functions @xmath158 and @xmath159 are multivalued . on the other hand ,
the inverse maps @xmath160 and @xmath161 are automorphic forms invariant under the elements of the automorphism group of the octagon , and are dense in both zeros and poles as the boundary of the disk is approached . in general
there is no closed form expression for @xmath160 , so we use a parametric approach : picking an arbitrary point on the @xmath154 plane will tell us the value of @xmath10 at @xmath162 .
as we shall see , there are some particular points on @xmath137 at which we can expand @xmath10 as a series , allowing @xmath163 to be computed analytically using . by convention , we take the points @xmath164 on the upper half plane to map to the vertices of a triangle with angles @xmath165 , @xmath166 , @xmath167 , respectively . the map from the upper half plane to this triangle is given by a ratio of two linearly independent solutions of the hypergeometric equation @xcite @xmath168 the two solutions at @xmath141 have exponents @xmath169 and @xmath170 , and are given by the hypergeometric functions @xmath171where @xmath172 , @xmath173 , @xmath174 and @xmath175 are related to the angles of the triangle by @xmath176 the schwarz triangle function is then defined as @xmath177where the normalisation factor @xmath178 is chosen so as to ensure that the triangle has the correct size , such that geodesic extensions of the circular edges intersect the unit disk at right angles .
this calculation was performed in @xcite .
putting everything together and simplifying , we have @xmath179 in order to calculate the higgs field , we use the two triangle maps @xmath158 and @xmath159 as depicted in figure [ mapf ] .
the derivative of @xmath36 is obtained by the chain rule .
then substituting into gives @xmath180 recall that @xmath181 are the angles in the triangle ( divided by @xmath182 ) , and @xmath183 are defined through .
the triangles of interest ( figure [ mapf ] ) have @xmath184 and @xmath185 .
the higgs field of a vortex placed at the origin can be computed analytically at the vertices of the fundamental triangle @xmath137 . in order to do this
, we expand @xmath186 near the singular points : rcl s()&=&^+ [szero ] + s(1+)&=&(1-^(1-a)(1-b)(1-a)(1-b)+ ) + [ sone ] the formula then gives the higgs field at the points on the octagon corresponding to the vertices of @xmath137 .
near @xmath141 we have from @xmath187using and the expansion of @xmath158 to change from @xmath154 to @xmath16 coordinates gives @xmath188 at the vertex of @xmath137 corresponding to @xmath142 , the expansion is of the form @xmath189with coefficients that can read off from .
applying gives @xmath190evaluating and for the triangles of interest , we find rcll ||_&=&|z|+(|z|^3)&1.768|z|+(|z|^3),[phiexacto ] + & & & + & & & + an expansion of the hypergeometric functions about the point at infinity can be avoided by a redefinition of the fundamental triangle , as explained in the caption to figure [ mapf ] , thereby allowing the higgs field at the vertex to be computed using and , albeit with different values of the parameters . away from the points where
the behaviour of the higgs field can be studied analytically by known series expansions of the hypergeometric function , contours of @xmath26 are obtained numerically and plotted using the auxiliary coordinate @xmath154 as a parameter .
this method leads to difficulty when sampling in the vicinity of the vertex of the octagon , since this is the image of the point at infinity in the @xmath154 plane .
a more effective procedure is to divide the octagon into eighths rather than sixteenths , so @xmath191 .
the preimage of this triangle under @xmath158 is the entire upper half plane .
splitting the triangle into two further isosceles triangles as shown in figure [ mapf2 ] , we find we need only sample within the region @xmath192 ( its complement in the upper half plane mapping to the other half triangle ) .
after interpolating between gridpoints , the solution is analytically continued from @xmath137 to @xmath96 by schwarz reflection , giving the contour plot of figure [ contours ] ( left ) . to
that shown in figure [ mapf ] .
the resulting higgs field is unchanged by this map .
shading indicates regions which map into each other , and the remainder of the hyperbolic plane is covered by reflections in the sides of the fundamental triangles .
the advantage of this redefinition is improved numerical sampling ( as the parameter @xmath154 need only take values in a semi - disk ) , although it is harder to compute the value of @xmath69 at the midpoint of an edge analytically as in . the angles of the triangles are @xmath193 and @xmath194 . ]
recall from section [ hypertilings ] that the octagon vertex and the origin are antipodal points on the bolza surface .
placing a vortex at the origin breaks most of the symmetry of the bolza surface , preserving a @xmath101 symmetry both at the vortex position and at its antipodal point .
there should thus be a @xmath101 symmetry at the vertex of the octagon when a vortex is placed at the origin , but this is not easily seen . to see it more explicitly ,
let us shift the vortex to the vertex of the octagon .
there should still be a @xmath101 symmetry at the origin .
recall that there is a double winding of @xmath36 about the vortex position , which for the vortex at the origin maps from the @xmath84 to the @xmath110 tessellation .
angles measured at the vortex position are doubled as illustrated in figures [ mapf ] and [ mapf2 ] . when the vortex is placed at a vertex , however , it is the angles at the vertices of @xmath96 which are doubled ( from @xmath109 to @xmath108 ) , while the map is conformal on the interior .
the image is then an octagon in the @xmath195 tessellation .
the absence of ramification points on the interior of the octagon means the degree of the map is @xmath19 and the riemann - hurwitz theorem tells us that the area of the image octagon is @xmath196 .
note that instead of , the more general formula for the area of the image polygon when @xmath3 vortices are placed at the vertex is @xmath197which still implies the bradlow bound . dividing the domain and image polygons into triangles as before gives the contours of @xmath26 shown in figure [ contours ] ( right ) . for a vortex at the centre of the fundamental octagon ( left ) and at a vertex ( right ) .
@xmath26 has a zero at the vortex position ( marked by black dots ) and reaches a maximum value of @xmath198 .
the two configurations are equivalent due to the two preserved @xmath101 symmetries of the fundamental octagon ( figure [ figoct ] ) .
the leading term of @xmath26 near the centered vortex is at order @xmath199 and is circularly symmetric , while non - trivial angular dependence enters at order @xmath200 . ] for a vortex at the centre of the fundamental octagon ( left ) and at a vertex ( right ) .
@xmath26 has a zero at the vortex position ( marked by black dots ) and reaches a maximum value of @xmath198 .
the two configurations are equivalent due to the two preserved @xmath101 symmetries of the fundamental octagon ( figure [ figoct ] ) . the leading term of @xmath26 near
the centered vortex is at order @xmath199 and is circularly symmetric , while non - trivial angular dependence enters at order @xmath200 . ]
we now proceed to check consistency of the magnetic flux .
a single vortex has flux @xmath18 , and we thus expect to obtain this value when @xmath201 is integrated over the fundamental octagon . using the same notation as in the flux through the fundamental octagon @xmath96 can be computed using stokes theorem , that @xmath69 is invariant under transformations by the elements of the fuchsian group @xmath87 .
it follows from the bogomolny equations that the physical magnetic flux density @xmath202 is invariant under these transformations , although neither @xmath201 nor @xmath203 is invariant on its own .
the physical magnetic flux is thus @xmath204 , where @xmath205 is the area form on the poincar disk .
care must be taken to include the total derivative term @xmath206 which affects the total flux , but does not alter the magnetic field locally . ]
@xmath207where @xmath208 denotes an infinitesimal line element of the boundary of the octagon , @xmath209 is the tangential component of the gauge potential , and @xmath210 and @xmath211 are derivatives tangent to and normal to the boundary @xmath212 .
we note from figure [ contours ] that the midpoints of the edges of the octagon are saddle points of @xmath26 .
combined with the reflection symmetry across the edges of the octagon we conclude that @xmath213 vanishes everywhere on the boundary .
let us now consider the topological contribution @xmath214 . on a segment of the boundary of @xmath96
, it evaluates to the difference in @xmath15 at the endpoints , @xmath215 . from we have , with @xmath216 , @xmath217dividing the octagon into sixteenths as in figure [ mapf ] , we carry out a rotation @xmath218 to take us from one vertex of the fundamental octagon to the next . from
, this leads to @xmath219 , hence the contribution to the flux from each edge of the octagon is @xmath109 , and the total flux is @xmath18 .
mapping between opposite edges of the fundamental octagon has the effect @xmath220 , where the change in sign agrees with our traversing of the edge in the opposite direction .
it is interesting to compare our results with those for a single vortex placed at the origin of the poincar disk , @xmath52 . from
, this vortex has @xmath221 and higgs field @xmath222 near the origin , @xmath223 , which should be compared to the result for the regular octagonal tessellation , where the leading coefficient of @xmath69 at the vortex is approximately @xmath224 . to get a feel for the change in the fields near the origin
when a single vortex on @xmath52 is replaced by an @xmath84 tessellation of vortices , we integrate the flux of the single vortex over the fundamental octagon of figure [ figoct ] . as a fraction of the total flux , we find @xmath225showing that the lattice of vortices compresses the flux by about @xmath226 .
the vortex equations on surfaces with @xmath227 can be studied using similar arguments to those given in the preceding sections .
the obvious extension of our construction is to a vortex at the origin of the poincar disk in the hyperbolic tessellation with schlfli symbol @xmath228 .
quotienting by the fuchsian group of the tessellation gives a vortex on a compact surface of genus @xmath39 and area @xmath229 .
the fundamental @xmath230-gon is divided into @xmath231 congruent triangles with angles @xmath232 .
the map @xmath36 is a two - to - one map from the @xmath230-gon to a @xmath233-gon in the @xmath234 tessellation , and the image of the fundamental triangle is a triangle with angles @xmath235 .
applying the construction described in section [ mappingbetweenpolygons ] gives the maximum value of @xmath69 , measured at the vertex of the @xmath230-gon , @xmath236 and the coefficient of the leading term of @xmath69 near the vortex at the origin , @xmath237 as the genus is increased , the area available to a vortex grows linearly with @xmath39 .
the resulting increase in separation between adjacent vortices in the tessellation suggests that for large @xmath39 , @xmath69 should approach the solution @xmath238 for a single vortex in @xmath52 , which has @xmath239 and @xmath240 .
evaluating and the leading coefficient of for some specific values of @xmath39 , we do indeed approach this limit : r|ccccc & @xmath241&@xmath242&@xmath243&@xmath244&@xmath245 + + @xmath246&0.841&0.965&0.986 & & 1 + @xmath247&1.768&1.955&1.984 & & 2 the principal interest of higher genus surfaces is that the bradlow bound ( [ bradlow ] , [ bradlowagain ] ) allows them to hold more than one vortex .
for example , for @xmath242 the regular @xmath248 tessellation with opposite edges identified allows one , two or three vortices placed at the origin or a vertex of the dodecagon .
the higgs field can again be computed by dividing the dodecagon into congruent triangles .
placing @xmath249 vortices at the origin and @xmath250 at the vertex of the dodecagon requires a map @xmath36 with ramification of order @xmath251 at the origin and @xmath252 at the vertex .
thus the image tessellation is of type @xmath253 whose fundamental polygon has an area @xmath118 given by either or the riemann - hurwitz theorem : @xmath254the bradlow bound requires that @xmath255 .
this ensures that @xmath256 and @xmath257 both divide @xmath258 and that the image tessellation is hyperbolic .
the euclidean tessellations of types @xmath259 and @xmath260 do not occur , because they would both require @xmath261 .
it is easy to check using the expansion of the schwarz triangle function that @xmath262 behaves like @xmath263 , with angular dependence at order @xmath264 .
note that on more general @xmath71 tessellations , integer winding of @xmath36 imposes divisibility constraints on @xmath249 and @xmath250 .
we remark that the surface described by the @xmath248 tessellation with opposite edges identified is not the most symmetric genus @xmath265 surface , which is the so - called klein surface that can be represented by a @xmath266 tessellation .
however , our construction only allows us to place a single vortex on this surface , necessarily at the origin of the @xmath267-gon ( as @xmath268 the centres and vertices of the polygon are not equivalent ) .
the image of the @xmath267-gon is a heptagon in the @xmath269 tessellation , and we obtain a vortex with @xmath270 symmetry with @xmath271 and @xmath272 . a case of special interest is the @xmath248 tessellation with one vortex placed at the origin of the dodecagon and one at the vertex .
the image tessellation is of type @xmath273 .
the leading behaviour of the higgs field near each vortex is @xmath274 we find the same leading coefficient when placing a single vortex at the origin of a dodecagon in the @xmath275 tessellation .
this result is due to the existence of a fixed - point - free @xmath276 action on the genus @xmath265 surface which identifies the origin and vertex of the @xmath248 tessellation .
quotienting by this group gives a smooth compact genus @xmath11 surface with one vortex on it .
this procedure is only possible when the vortex locations are compatible with the @xmath276 symmetry , and the quotient surface contains half the number of vortices .
figure [ dodecagon ] illustrates the relation between the two tessellations .
tessellation with opposite edges identified ( solid line ) can be cut along the dotted line to give a dodecagon of half the area in the @xmath275 tessellation . identifying edges as shown in the right hand diagram
yields a smooth compact genus @xmath11 surface .
placing a vortex at the origin of the @xmath248 cell and another at the vertex is therefore equivalent to a single vortex at the origin of a @xmath275 cell .
three distinct vertices are indicated by dots . ]
tessellation with opposite edges identified ( solid line ) can be cut along the dotted line to give a dodecagon of half the area in the @xmath275 tessellation . identifying edges as shown in the right hand diagram
yields a smooth compact genus @xmath11 surface .
placing a vortex at the origin of the @xmath248 cell and another at the vertex is therefore equivalent to a single vortex at the origin of a @xmath275 cell .
three distinct vertices are indicated by dots . ]
this observation can be understood by explicitly looking for a free @xmath276 action , an involution @xmath277 on the hyperelliptic curve .
let us generalise our discussion to a genus @xmath39 surface represented by a regular @xmath228 tessellation with opposite edges identified , with a vortex at the centre and another at the vertex . as a riemann surface ,
this genus @xmath39 surface is a double cover of @xmath100 with @xmath278 ramification points .
the ramification points at the north and south poles correspond to the centre and vertex of the polygon , and there are @xmath233 equally spaced ramification points on the equator .
generalising what was discussed in section [ hypertilings ] for the @xmath241 bolza surface , the cyclic @xmath279 symmetry has fixed points at the polygon centre and at the vertex , while there are @xmath39 @xmath280 symmetries with pairs of fixed points on the equator .
the symmetry group of the surface is a double group of the dihedral group @xmath281 , and the hyperelliptic curve representing this setup is ) , the symmetry is enhanced to @xmath282 , a double cover of the cubic group . ]
@xmath283the desired involution must map between the origin of the polygon and the vertex ( i.e. between the poles of the sphere @xmath284 and @xmath285 ) and have no fixed points .
this is only possible for @xmath39 odd ( the case @xmath39 even has fixed points at @xmath286 ) , in which case it is achieved by the involution @xmath287the quotient of the branched cover of @xmath100 by this involution is a double covering of the northern hemisphere with a particular identification of the segments of the equator between ramification points . in the polygon picture , there are @xmath230 edges to pair up .
consistency with the transformation @xmath288 gives two possible pairings for each edge , due to the choice of branch .
we also require the quotient surface to be smooth , which will be the case if there is no angle deficit at the identified vertices .
the total internal angle of the quotient @xmath289 polygon is @xmath290 , hence there must be @xmath39 distinct vertices when the edges are paired .
this imposes a further constraint on our choice of edge identifications .
there is also the possibility that edge pairs can be twisted , although it is not clear whether our construction , which relies on @xmath10 extending from a fundamental triangle to all of @xmath52 by reflections , works in the twisted case .
once the correct identifications have been made , the resulting surface has half the area of the original surface and genus @xmath291 .
the identification described in the preceding paragraphs is illustrated for the @xmath242 case in figure [ dodecagon ] .
the genus @xmath11 surface in the @xmath275 tessellation has different fenchel - nielsen parameters from the @xmath84 tessellation , and they therefore represent different surfaces .
an immediate consequence of this is the different leading behaviour of the higgs field , as is seen by contrasting and .
because a @xmath11-holed surface has no freely acting rotations by @xmath182 , the genus can not be reduced further .
in this paper we have shown how to construct abelian - higgs vortices on compact hyperbolic surfaces which can be represented by tessellations of the hyperbolic plane by regular polygons .
placing vortices on the surface breaks most of the symmetry , and our method requires the vortex positions to be chosen in such a way that there is a preserved symmetry allowing the surface to be covered by congruent triangles sharing vertices at the vortex positions .
in particular , multivortex solutions can be constructed with vortices located at antipodal points on the surface , corresponding to the origin and vertex of the polygonal tessellation .
the higgs field is computed by identifying a holomorphic function on the fundamental polygon which is conformal everywhere except at the vortex positions , where there is a ramification point of order one greater than the number of vortices at that location .
this function can be expressed implicitly as the composition of two schwarz triangle functions , allowing numerical evaluation of the higgs field at any given point on the surface .
there are several immediate generalisations of this procedure which it would be interesting to investigate further : * the tessellation can be deformed such that the angles are fixed but the edge lengths change ( see figure [ def ] ) . in this case
we require a map from a quadrant of the fundamental octagon to half a kite .
the construction is analogous to that of the schwarz triangle map described in section [ solutions ] , with the hypergeometric differential equation replaced by the heun differential equation , which has four regular singular points .
the difficulty with this construction is the appearance of an _ accessory parameter _ which characterises the deformation , and which requires numerical computation .
a different deformation of the regular octagon was studied in @xcite .
+ octagon with angles of @xmath109 which can be split into four quadrilaterals , each of which maps to half of a deformed image quadrilateral . ] + octagon with angles of @xmath109 which can be split into four quadrilaterals , each of which maps to half of a deformed image quadrilateral . ] * the construction described in this paper does not allow the vortex to be placed at an arbitrary position on the surface .
the riemann map from a hyperbolic polygon to the upper half plane is analytically continued to cover the poincar disk by reflection in the edges of the polygon .
if the vortex is not positioned symmetrically then the fields can not be constructed by maps between regular polygons .
as it stands , our construction only enables us to compute the fields at isolated points in the moduli space . *
a final outstanding question is how to place an arbitrary number @xmath3 of vortices ( with @xmath292 ) at the origin of a symmetric tessellation describing a genus @xmath39 surface .
an @xmath293-fold map from the fundamental polygon to a smaller polygon is only possible if @xmath122 divides the number of edges of the initial polygon .
for example , we can place a single vortex on the klein surface ( with a map from a @xmath267-gon to a heptagon ) , but not two or three vortices .
this work was supported by the uk science and technology facilities council under grant number st / j000434/1 .
9 a. jaffe , c. taubes , _ vortices and monopoles _ , birkhuser ( 1980 ) j. m. baptista , _ vortices as degenerate metrics _ ,
lett . math .
* 104 * ( 2014 ) 731 , http://arxiv.org/abs/1212.3561[`arxiv:1212.3561 [ hep - th ] ` ] e. witten , _ some exact multipseudoparticle solutions of classical yang - mills theory _ , phys .
* 38 * ( 1977 ) 121 u. essmann , h. truble , _ the direct observation of individual flux lines in type ii superconductors _ , phys .
lett . * a 24 * ( 1967 ) 526 n. s. manton , n. a. rink , _ vortices on hyperbolic surfaces _
, j. phys .
a * 43 * ( 2010 ) 434024 , http://arxiv.org/abs/0912.2058[`arxiv:0912.2058 [ hep - th ] ` ] w. j. floyd , s. p. plotnick , _ growth functions on fuchsian groups and the euler characteristic _ , invent . math .
* 88 * ( 1987 ) 1 a. v. nazarenko , _ two - parametric hyperbolic octagons and reduced teichmller space in genus two _ , http://arxiv.org/abs/1301.5446[`arxiv:1301.5446 [ math - ph ] ` ] z. nehari , _ conformal mapping _
, mcgraw - hill ( 1952 ) m. harmer , g. martin , _ conformal mappings from the upper half plane to fundamental domains on the hyperbolic plane _ , department of maths report series 499 , university of auckland , n.z . | we construct , for the first time , abelian - higgs vortices on certain compact surfaces of constant negative curvature .
such surfaces are represented by a tessellation of the hyperbolic plane by regular polygons .
the higgs field is given implicitly in terms of schwarz triangle functions and analytic solutions are available for certain highly symmetric configurations . |
our understanding of the intergalactic medium ( igm ) has undergone a paradigm shift in recent years , largely as a result of powerful hydrodynamical simulations . when the ly@xmath0 forest was first observed in the late 1960s , the rich field of absorption blueward of the qso s
ly@xmath0 emission was interpreted as the detection of discrete intergalactic clouds ( e.g. lynds & stockton 1966 ; lynds 1971 ; sargent et al .
1980 ) . in order to understand the existence of such isolated absorbers , various theories of cloud confinement , including self - gravity ( melott 1980 ) , cold dark matter minihaloes ( rees 1986 ) and the presence of an inter - cloud medium ( e.g. sargent et al . 1980 ; ostriker & ikeuchi 1983 ) were proposed . however , the advent of hydrodynamical simulations , which model the growth of structure in the high redshift universe , provided a significant revision to our picture of the igm ( see the recent review by efstathiou , schaye and theuns 2000 ) .
it has been found that in the presence of a uv ionizing background , the ` bottom - up ' hierarchy of structure formation knitted a complex , but smoothly fluctuating ` cosmic web ' in the igm ( e.g. cen et al .
1994 ; hernquist et al .
1996 ; bi & davidsen 1997 ) .
the absorption in the ly@xmath0 forest is caused not by individual , confined clouds , but by a gradually varying density field characterized by overdense sheets and filaments and extensive , underdense voids .
the advance in theoretical simulations has been matched by increasingly high quality data , as comprehensively reviewed by rauch ( 1998 ) .
one of the major discoveries concerning the igm has been the identification of metal absorption lines associated with many of the ly@xmath0 forest clouds ( cowie et al .
1995 ; tytler et al .
thus , whilst the ly@xmath0 forest was once thought to be chemically pristine , it has now been well - established that a large fraction of the high column density ly@xmath0 clouds ( @xmath3 ( ) @xmath6 14.5 ) , associated with collapsing , over - dense structures , contain metals ( most notably ) the signature of enrichment by the products of stellar nucleosynthesis ( songaila & cowie 1996 ) . the presence of metals in the ly@xmath0 forest may be reasonably explained either by in - situ enrichment ( local star formation in the cloud itself or in a nearby galaxy ) or by early pre - enrichment by a high redshift episode of population iii stars . whilst the effects of supernova feedback are still not fully understood and therefore the spatial extent of wind - driven ejecta is poorly constrained , enrichment by galactic winds and superbubbles is unlikely to be an efficient way to distribute metals over distances large in comparison with the mean separation between galaxies ( maclow & ferrara 1999 ) . instead
, models have appealed to larger scale processes such as merging and turbulent diffusion as the dominant mixing mechanisms ( gnedin & ostriker 1997 ; gnedin 1998 ; ferrara , pettini & shchekinov 2000 ) .
such processes would take time to smooth out the metallicity of the igm so that at @xmath7 the metal enrichment is still expected to be very patchy . whilst the deep potential wells of galaxies inhibit efficient
, widespread distribution of metals far from their sites of formations , small regions of star formation at high redshift may be able to eject their nucleosynthetic products for more homogeneous mixing ( e.g. nath & trentham 1997 and references therein ) .
an episode of population iii star formation may therefore have spread metals far from their sites of formation , seeding ` sterile ' regions of the igm with metals ( ostriker & gnedin 1996 ) .
clearly , distinguishing between in - situ and pop iii scenarios has important implications for understanding not only the first generation of stars , but also the mechanisms by which metals are mixed and distributed from their stellar birthplaces .
several papers ( cowie & songaila 1998 ; lu et al . 1998 ; ellison et al .
1999a , hereafter paper i ) have addressed this question by attempting to probe the low density regions of the igm where the difference in metallicity predicted by in - situ and pop iii enrichment may be most marked .
the detection of @xmath81548,1550 associated with low column density ly@xmath0 clouds ( log @xmath3 ( ) @xmath9 14.0 ) is observationally challenging , even with the capabilities of keck , due to the extreme weakness of the metal lines .
analysis techniques have therefore been developed to effectively enhance the sensitivity of the data beyond the normal equivalent width limits of the spectra .
two in particular have received recent attention , namely the production of a stacked spectrum ( lu et al .
1998 ) and the use of individual pixel optical depths of ly @xmath0 and ( cowie & songaila 1998 ) .
in paper i we attempted to reconcile the apparently conflicting results obtained from these two techniques with an analysis of a very high signal - to - noise ratio ( s / n ) spectrum of the ultra - luminous bal quasar apm 08279 + 5255 .
rigorous testing of the analysis procedures revealed that both methods suffered from hitherto unrecognised limitations and it was concluded that the question of whether or not the low density regions of the igm have been enriched remains unanswered . in order to probe deeper into the low density igm ,
we have obtained an exceptionally high s / n spectrum of the well - known lensed qso , q1422 + 231 , using the hires instrument ( vogt 1994 ) on the kecki telescope .
although this qso has been well studied in the past ( e.g. songaila & cowie 1996 ; songaila 1998 ) , we have roughly doubled the exposure time of earlier spectra , allowing us to probe the metallicity of the ly@xmath0forest to more sensitive levels than has previously been achieved .
the spectrum is much better suited to the present work than the the much more complex bal quasar apm 08279 + 5255 ( ellison et al .
we present a careful and extensive analysis of the systems in order to determine the extent of metal enrichment in the igm .
this paper is organised as follows . in
2 we describe the observations , the data reduction procedures and the voigt profile fitting process used to determine column densities .
we briefly discuss in 3 the suitability of q1422 + 231 for this work and define the redshift interval over which we will perform the analysis . in 4 , we determine the column densities of the 34 detected absorption systems in the spectrum and investigate the column density distribution of these absorbers .
finally , we critically assess the two methods of analysis , described in sections [ stacksec ] and [ tausec ] , which have been developed to probe low column density ly@xmath0 clouds to very sensitive levels .
we utilize a suite of simulation techniques to fully test these methods and quantify the potential inaccuracies in our analysis .
we adopt @xmath10 = 1.0 throughout .
observations of q1422 + 231 were made with the hires spectrograph ( vogt 1994 ) on the kecki telescope in february 1999 , using the @xmath11 arcsec slit .
the resultant resolution is @xmath12 .
individual exposure times varied between 30 min and 40 min , for a total exposure of 630 min in two ( overlapping ) grating settings to provide total wavelength coverage between the quasar s
ly@xmath0 and emission lines .
the data were reduced as described in songaila ( 1998 ) and added to the spectrum of q1422 + 231 obtained under similar conditions by songaila and cowie ( 1996 ) .
the total integration time for the two datasets is 1130 minutes and has a s / n of 200 300 redward of the qso s
ly@xmath0 emission , a quality superior to all previously published data . after extraction and sky subtraction
, we noted that the cores of the saturated absorption lines contain a very small systematic residual flux .
this error in the zero level is possibly due to light from a foreground source or a small underestimate of the background sky .
the correction required to bring the line cores to zero was found to vary slightly with wavelength , ranging from 0.8% for the bluest absorption lines ( @xmath13 ) to 0.2% at @xmath14 .
no correction could be estimated redward of the qso s
ly@xmath0 emission since the absence of saturated lines gave us no basis for estimating the adjustment required .
however , since the correction factor required appears to diminish with increasing wavelength and is already very small at 5500 , and given that such an adjustment will make little difference to the relatively weak lines studied in the red , we consider any residual eror to be unimportant for the analysis presented here .
the continuum fit was achieved using the starlink package dipso with a cubic spline polynomial applied to windows of spectrum judged to be free from absorption .
voigt profiles were fitted to the entire normalized ly@xmath0 forest using vpfit ( webb 1987 ) to decompose the complex absorption into individual components defined by a column density ( @xmath3 ( ) ) , a @xmath1-value ( doppler width ) and a redshift .
this task , though time consuming , is an important feature of our analysis since synthetic spectra can be simulated based on the line list of voigt profile parameters .
for example , additional associated with the fitted can be included to these synthetic spectra according to any desired enrichment recipe . as will be discussed later in this paper
, this is an essential step for testing the analysis techniques used here .
q1422 + 231 ( @xmath15 ) is a well - studied quasar and actually consists of four closely spaced lensed images with separations of 0.5 1.3 arcsec ; the lensing galaxy is at @xmath16= 0.34 ( patnaik et al .
1992 ; kundic et al 1997 ; tonry 1998 ) .
gravitational lenses enhance the emission from high redshift qsos making them more powerful probes of the intergalactic and interstellar medium , but obviously the sight lines will sample different spatial regions of the intervening material . its redshift and luminosity ( v=16.5 ) have made q1422 + 231 an ideal candidate with which to probe intervening material through closely spaced sightlines ( petry et al .
1998 ; rauch et al . 1999 ) .
the observations reported here are of the closely spaced a and b components . in our observations
the sightlines are unresolved .
it is well established that multiple lines of sight through quasar pairs separated by several arcsecs show coherence between ly@xmath0 clouds on scales @xmath6 100 kpc ( e.g. bechtold et al .
1994 ; dinshaw et al .
this is consistent with the scenario that has emerged from hydrodynamical simulations that portrays structure in the igm not as discrete localized clouds but as a smoothly fluctuating medium . for metal line systems ,
whilst there may be slight differences in the individual components that constitute complexes , rauch et al .
( 1998 ) have shown that the total system column density remains largely unchanged over @xmath17 10 kpc .
the small linear separations probed by q1422 + 231 ( @xmath18 kpc h@xmath5 for the systems in the @xmath19 range considered here ) are therefore unlikely to give rise to by line of sight differences so large as to compromise our column density determinations . in order to avoid confusion between ly@xmath0 and higher order lyman lines ,
we restrict our analysis of the ly@xmath0 forest in q1422 + 231 to the interval 4740 @xmath20(ly@xmath0)@xmath21 which corresponds to a redshift range of 2.90 @xmath22 3.54
. the lower limit of this interval is determined by the onset of ly@xmath23 absorption and the upper limit is enforced in order to avoid effects due to quasar proximity ( e.g. lu , wolfe & turnshek 1991 ) and corresponds to a velocity separation of @xmath24 km s@xmath5 relative to the emission redshift , @xmath25 = 3.625 . in order to improve the statistics and
s / n of stacked data ( 5 ) , we have , in some of the work presented here , included the spectrum of apm 08279 + 5255 , recently analysed in paper i. apm 08279 + 5255 is an ultra - luminous broad absorption line ( bal ) quasar with a systemic redshift @xmath25 = 3.911 and a broad band magnitude r = 15.2 .
the data obtained for this quasar ( as presented in paper i ) are also of excellent quality , though the broad absorption line makes some portions of the spectrum less ideal for this problem than the spectrum of q1422 + 231 , and taken together with q1422 + 231 they represent a premium data set for this work .
the wavelength region used for apm 08279 + 5255 is 4995 @xmath20(ly@xmath0)@xmath26 ( 3.11 @xmath27 3.70 ) which in addition to the selection criteria applied to q1422 + 231 , takes into account the bal nature of this quasar . finally , a region contaminated by atmospheric absorption from 6865 @xmath20 ( )
@xmath9 6940 ( corresponding to @xmath28 for ) was excluded in both spectra .
q1422 + 231 has been the target of several observing campaigns over the years , which have resulted in spectra of various s / n ratios . whilst the primary motivation for obtaining such a high s / n spectrum was our scientific goal of probing the low column density ly@xmath0 forest , continued focus on this target has produced important results for igm enrichment on a range of column density scales . in this section
we present all of the detected systems within the redshift range defined in 3 , several of which have not been detected in previous spectra and we show that the power law column density distribution established for strong absorbers with @xmath3()@xmath6 12.75 ( songaila 1997 ) continues to significantly lower column densities .
we detect 34 systems within our defined wavelength interval , 29 of which are associated with saturated ly@xmath0 clouds ( log @xmath3 ( ) @xmath29 14.5 ) .
these systems exhibit a wide variety of column densities and complexity ( i.e. number of components ) as can be seen in figure [ civsys ] .
the column densities , @xmath1-values and redshifts determined for each component using the voigt profile fitting program vpfit , are presented in table [ civlist ] .
examples of systems detected in our spectrum which have not been observed in other published spectra ( songaila & cowie 1996 ; songaila 1998 ) are the weak systems at @xmath30 and 3.518 ( c24 , c25 and c33 in figure [ civsys ] ) .
previous studies of absorbers in a variety of quasar sightlines has established a power law column density distribution ( with index @xmath31 , see eqn .
1 ) complete down to log @xmath3 ( ) @xmath32 12.75 at @xmath33 and 12.25 at @xmath34.(songaila 1997 ) .
the column density distribution function is defined as @xmath35 where , @xmath36 is the number of systems per column density interval per unit redshift path .
the redshift path ( used instead of @xmath19 in order to account for co - moving distances ) is given by @xmath37 $ ] for our adopted cosmology .
an important question is how much of the iceberg has been exposed ? to what limit
does this power law distribution continue ?
as spectral data have improved , early determinations of the column density distribution of absorbers have been shown to be incomplete at low column densities ( compare for example petitjean & bergeron 1994 and songaila 1997 ) .
the superb quality of this single spectrum can address whether the established power law continues to lower @xmath3 ( ) , in which case the apparent fall - off towards low @xmath3 ( ) seen previously is due to incompleteness , or whether there is a real turnover in number density . in figure [ f_n ]
we show the column density distribution of derived from the systems in q1422 + 231 , assumed to be a power law of the form of eqn .
1 . a maximum likelihood fit to the data ( binned in figure [ f_n ] for display purposes only ) gives a power law index of @xmath2 , consistent with other recent estimates .
this high quality spectrum , however , clearly uncovers more of the ` iceberg ' than previous studies and the power law continues down to log @xmath3 ( ) @xmath17 12.3 ( figure [ f_n ] , solid points ) . below this column density
, @xmath36 shows an apparent departure from the power law which may be due to incompleteness or alternatively reflect a real turnover in the @xmath3 ( ) distribution .
the formal 5@xmath38 detection limit for in our spectrum is log @xmath3 ( ) = 11.6 , for the median @xmath1-value of the observed absorption lines , @xmath39 km s@xmath5 .
this detection limit is for _ one line _ and is based on the @xmath401548 doublet component . however ,
identification of a suspected system is dependent on confirmation from the weaker @xmath401550 line whose oscillator strength is only half of the @xmath41-value of the 1548 line .
this effectively reduces our sensitivity for detecting _
systems _ by a factor of two to a 5@xmath38 detection limit of log @xmath3 ( ) = 11.9 .
moreover , lines with @xmath1-values significantly larger than the median @xmath1=13 km s@xmath5 , may not be detected . in order to estimate the incompleteness for systems with log @xmath3 ( ) @xmath9 12.3 due to large @xmath1
, 40 doublets were simulated for the two bins in figure [ f_n ] which show a departure from the established power law , i.e. @xmath3 ( ) = 12.05 and @xmath3 ( ) = 11.75 .
the @xmath1-value for each line was drawn at random from the real distribution of doppler widths and noise was added at the appropriate level . each simulated line
was then inspected to assess whether it would have been identified in the original spectrum so that a correction factor could be determined to estimate incompleteness . for @xmath3 ( )
= 12.05 a correction factor of 2.4 was determined ( 17/40 test systems detected in the incompleteness trial ) , with the largest @xmath1-value in the lines detected being 13 km s@xmath5 . in the lowest column density bin , @xmath3 ( ) = 11.75 only 9 out of 40 lines ( with @xmath42 8 km s@xmath5 ) were identified , corresponding to a correction factor of 4.4 .
clearly , these correction factors assume that the @xmath1-value distribution does not significantly change with decreasing column density .
the two data points adjusted for incompleteness are represented by open circles in figure [ f_n ] .
thus , it appears that the power law distribution of column densities continues down to at least log @xmath3 ( ) = 11.75 .
recently , hydrodynamical simulations have been used to predict the expected scatter in individual / ratios within strong ly@xmath0 absorbers for a fixed [ c / h ] ( hellsten et al .
1997 ; rauch , haehnelt & steinmetz 1997 , dav et al . 1998 ) . when compared with observations ,
these models show that the data are consistent with a mean igm metallicity [ c / h ] = @xmath43 with variations of up to a factor of 10 around this average value . in these models ,
the systems associated with high column density ly@xmath0 clouds are dominated by metals from in - situ star formation , since simulations show that these metal enriched clouds are found within a few tens of kpc from collapsed , dense clumps at @xmath7 ( e.g. haehnelt 1998 and references therein ) .
we would then expect the metallicity of such clouds to be variable , being dependent on the local star formation history , and therefore the scatter amongst individual / ratios at a given @xmath3 ( ) to be larger than that predicted for a homogeneous [ c / h ] .
we can calculate / ratios for 19 systems in our spectrum of q1422 + 231 . of the 34 detected systems in the spectrum presented here ,
29 are associated with saturated ly@xmath0 lines . although it is not possible to determine accurate column densities from these lines alone , for 14 of the systems higher order lyman lines are both accessible and suitably uncontaminated by blends so that an accurate @xmath3 ( ) can in fact be derived .
in addition , 5 of the systems in q1422 + 231 are associated with clouds with log @xmath3 ( ) @xmath9 14.5 which are not saturated so that the column densities can be determined from the voigt profile fit of ly@xmath0only . the and column densities determined for these 19 systems are presented in table [ hifits ] .
dav et al . (
1998 ) have also extensively studied the / ratios of q1422 + 231 , as determined from the spectrum of songaila & cowie ( 1996 ) .
they constructed a mock spectrum of q1422 + 231 from hydrodynamic simulations for comparison with the data .
having both simulated and observed spectra at their disposal , they measured / ratios in both datasets in a _ consistent manner _ ( using the autovp voigt profile fitter , dav et al .
1997 ) , and found that an intrinsic scatter of approximately 0.5 dex in metallicity is required to fit the data .
however , the observed range of / ratios is the result of many complex effects , which include not only spatial variations in the temperature - density relationship , but also , for example , fluctuations in the ionizing background .
these complex effects have not yet been fully incorporated into models and therefore we should be mindful that the scatter in measured values of / is an upper limit to the variations in [ c / h ] when compared with homogeneously enriched simulations with a uniform ionizing background .
dav et al . (
1998 ) also found that the most robust diagnostic for determining the mean carbon abundance in detected systems is @xmath44log @xmath3()@xmath45 , although this statistic is clearly dependent on the sensitivity of the data .
as a consistency check , we determine the @xmath44log @xmath3()@xmath45 for those systems identified in our spectrum of q1422 + 231 whose column densities are above the detection limit of the songaila & cowie ( 1996 ) spectrum ( estimated to be log @xmath3 ( @xmath6 12.0 ) .
we calculate @xmath44log @xmath3()@xmath45 = 12.77 which compares well with the value of 12.72 determined by dav et al .
( 1998 ) , considering the errors associated with column density determinations of individual systems .
this demonstrates that the different line finding and fitting procedures used on different spectra reproduce the same answer , even when the same systems are observed with a higher s / n . for voigt profile fitting
, this is an important point to make in a process that is sometimes not clear - cut .
it is also an illustration that the detection limit of the spectrum used by dav et al .
( 1998 ) was well determined and did not suffer from serious incompleteness .
ideally , one would like to see how the mean [ c / h ] varies as we detect progressively weaker systems .
however , a simple extrapolation to lower column densities is not possible , since the relation between @xmath44log @xmath3()@xmath45 and [ c / h ] presented in dav et al .
( 1998 ) is tailored specifically for the detection limits of their data . in order to apply this technique to the full sample of systems detected in the new q1422 + 231 spectrum presented here
, it would be necessary to re - create a model spectrum to match our data .
however , we remind the reader that the caveats mentioned in relation to the scatter of / are also applicable here , a fact of which one should be aware when comparing observational and simulated results and interpreting the conversion to metallicity .
we can summarise the results from this section with the following conclusions . by obtaining an ultra - high s / n spectrum of the @xmath46 quasar q1422 + 231 ,
we have detected 34 absorption systems with 2.91 @xmath22 3.54 .
some hitherto undetected weak systems are reported whose column densities are consistent with the established @xmath36 distribution showing that this power law ( @xmath2 ) continues down to at least log @xmath3 ( ) = 11.75 , a factor of 10 deeper than the previous determination at @xmath33 in songaila ( 1997 ) . by fitting down the lyman series , we are able to determine accurate / ratios for 19 of the identified systems , rather than relying on a statistical estimate of the median metallicity as has often been done .
dav et al .
( 1998 ) have previously determined that the scatter in / in the q1422 + 231 spectrum of songaila & cowie ( 1996 ) requires an intrinsic scatter in the [ c / h ] of a factor of @xmath17 3 . finally , by considering only the systems above the detection limit
, we obtain the same @xmath44log @xmath3()@xmath45 as dav et al .
( 1998 ) , a statistic used to infer that the mean [ c / h ] = @xmath43 . however , we stress that these simulations do not take into account the full complexity of the physical processes determining the / ratio in the igm so that both the mean [ c / h ] and its scatter may change with future more detailed work .
as discussed in previous sections , the / ratio in low @xmath3 ( ) systems may hold the key to understanding metal enrichment mechanisms in the igm . since one of the major limitations in detecting weak lines
is s / n , stacking many sections of the spectrum is one way to circumvent this problem ( e.g. norris , peterson & hartwick 1983 ; tytler et al .
1995 ) . in our application of this technique , we select ly@xmath0 lines with 13.5 @xmath9 log @xmath3 ( ) @xmath9 14.0 whose corresponding spectrum shows no obvious metal lines or contamination from other absorption features .
each section is de - redshifted to the rest frame , re - binned to the dispersion of the lowest @xmath19 system , weighted according to its s / n , stacked and finally re - normalized .
the optimal weighting used to co - add the sections is given by @xmath47 where @xmath48 is the variance of the data . to achieve maximum sensitivity , we use low column density lines from both the apm 08279 + 5255 and q1422 + 231 spectra . within the ranges
defined in 3 , a total of 67 low @xmath3 ( ) lines were identified in the two qso sightlines . _
the resulting composite spectrum has a s / n = 1250 and shows no absorption at the rest wavelength of , @xmath49 _ , as can be seen from the top panel of figure [ data_stack ] . in order to place a significance limit on this non - detection , a synthetic spectrum was created , re - producing the ly@xmath0 forest of the 2 quasars from the fitted line lists .
was included for lines with log @xmath3 ( ) @xmath9 14.5 assuming a fixed / ratio and @xmath1 ( ) = @xmath50(ly@xmath0 ) ( representing a combination of thermal and bulk motion , as found in paper i ) .
figure [ data_stack ] shows the results of stacking synthetic spectra with log / = @xmath51 .
the resultant absorption in this stack has an equivalent width of 0.15 m and represents a 4@xmath38 feature which we therefore adopt as the detection limit for our stacked data .
several analyses ( e.g. songaila & cowie 1996 ; paper i ) have determined the / ratio in high column density ly@xmath0 clouds to be in the range from log / = @xmath52 to @xmath53 .
the detection limit of the synthetic stack is almost factor of two lower than the metal - poor limit of this range , and thus it would appear to indicate a drop in the / ratio at lower column densities .
it must be understood , however , that this technique has several potential problems which may compromise its efficiency in detecting weak absorption features .
we now discuss these problems in turn . * in order to co - add each section , the data must be re - binned , usually to the dispersion of the lowest redshift system .
this could smooth out a weak feature , although the scale of smoothing is very small compared with the width of the expected line so that this is not likely to be a major effect . * even for a fixed carbon abundance there will be a scatter in the values of / ( as discussed in the previous section ) . therefore ,
if absorption is detected in the composite spectrum it will be averaged over a range of column densities that can not be recovered individually . moreover , depending on the scatter of / values , the absorption may be dominated by the strong tail end of this metallicity range . in fact , it is conceivable that residual absorption could be caused by only a few relatively strong lines , since this method relies on an average which is very sensitive to a non - gaussian tail . interpreting a residual signal in the composite spectrum is therefore not straightforward , although in the analysis presented here we find that there is no absorption in the stacked data and so we can determine a useful detection limit .
more serious problems for the present analysis are : * the sensitivity of this method to errors in continuum fitting , anomalous pixels and other forms of contamination .
the usual procedure is to visually inspect each section before adding it to the stack in order to ensure that it is ` clean ' .
this will filter out major contamination by , for example , uncorrected cosmic ray events or absorption due to systems other than .
however , small errors in the continuum fit or deviant pixels could seriously compromise the efficiency of the stack .
in addition , a re - normalization of the composite spectrum is usually required , since the small errors in the original continuum fit have now been compounded by stacking . *
stacking the individual sections of spectrum in order to build up a signal is pivotal upon centering the absorption feature in the composite .
if there is a significant error in the stack center , i.e. if an offset exists between the redshift of the parent ly@xmath0 line and its associated complex , the composite signal would be smeared out .
depending on the magnitude of the offset , this effect may lead us to under - estimate the amount of absorbing material .
this problem could be exacerbated by the afore - mentioned need for a re - normalization because it may be difficult , if not impossible , to distinguish a weak smeared signal from the compound continuum errors . in paper
i , we showed that there is indeed a random redshift offset ( @xmath54 ) between the measured position of the ly@xmath0 and corresponding lines . in that work
it was found that the redshift offset had a distribution with @xmath55 4 @xmath56 equivalent to a velocity difference of 27 km s@xmath5 , although this statistic was based on a relatively small number of systems .
the addition of a second quasar to the analysis has improved the statistics and for a total of 56 systems in the two quasars we now determine a @xmath57=2.6 @xmath56 ( @xmath5817 km s@xmath5 for an average redshift of 3.45 ) .
this redshift offset was determined in two different ways . in both cases
the redshift of the was estimated by taking the centroid of the system since it is often complex , consisting of several blended components . for most of the corresponding ly@xmath0 lines ,
the absorption is well represented by a single component and so in the first instance the redshift of the cloud was obtained from the voigt profile fit . for comparison , the redshift of ly@xmath0 was also determined using the same centroid method applied to the systems and it was found that both determinations yielded almost identical results for the distribution of @xmath54 , shown in figure [ offsets ] . since it is the strongest component in each complex that first emerges from the noise in the weak systems , this will be the feature enhanced by the stacking procedure .
therefore , we also investigated the distribution of offsets between the fitted @xmath19 of ly@xmath0 and the redshift of the deepest trough and once again the value @xmath58 17 km s@xmath5 was found .
we investigated how seriously the composite spectrum would be affected by this offset which effectively shifts the expected positions of the lines by a random amount .
again , synthetic spectra were produced , this time including a redshift offset applied to the position of the line with @xmath54 drawn at random from a gaussian distribution with @xmath57=2.6@xmath59 .
we find that in order to reproduce a 4@xmath38 detection in the presence of a redshift offset , twice as much must be included in the ly@xmath0 forest clouds , i.e. log ( / ) = @xmath60 , see the bottom panel of figure [ data_stack ] .
this is consistent with the measured / in log @xmath3()@xmath614.5 lines and is therefore not a sufficiently sensitive limit to establish whether the low column density ly@xmath0 clouds contain significantly less that the stronger lines .
we conclude that a factor of at least two improvement in sensitivity is required in order to show conclusively whether the low column density absorbers are more metal deficient than their high column density counterparts .
alternatively , it must be shown that the redshift offset in these lines is @xmath61 17 km s@xmath5 and dilution of absorption from smearing the stack is unimportant . the redshift offset in saturated lines could have two possible explanations , one physical and the other an observational artefact
. there could be an intrinsic redshift difference between the and ly@xmath0 absorbers , caused by , for example , ionization effects or outflows .
an additional effect could be a redshift offset caused by the blending of strong ly@xmath0 lines which , when saturated , can not be distinguished into separate components .
two examples of this are shown in figure [ lya_breakdown ] .
plotted in velocity space are two examples of ly@xmath0 absorbers ( black solid line ) and corresponding ( dashed line ) .
both saturated ( systems ` b ' and ` c ' ) and weak ( systems ` a ' and ` d ' ) ly@xmath0 clouds are shown and in the top panel the ly@xmath23 ( in gray ) is also included .
the system ` c ' is associated with an apparently monolithic ly@xmath0 absorber centered at @xmath62 = 0 and exhibits a redshift offset of approximately 25 km s@xmath5 . however , as seen from the ly@xmath23 absorption , this system is made up of more than one component and the clearly has a redshift that more closely matches the strongest of these . the unsaturated ly@xmath0 absorber associated with the system `
a ' does not break down into a multi - component system in ly@xmath23 and the redshift offset is correspondingly small ( @xmath63 1 km s@xmath5 ) .
however , there are other examples of systems associated with weak ly@xmath0 that _ do _ show a significant offset , for example system ` d ' in the lower panel of figure [ lya_breakdown ] . unfortunately ,
higher order lyman lines are not available for this particular case .
we re - measured the redshift offset for each of the 19 systems in table [ hifits ] for which an accurate @xmath3 ( ) had been obtained by tracing down the lyman series . for 16 of these systems ,
the appeared to be associated with a single component ( i.e. not obviously blended ) , as seen from ly@xmath23 and/or ly@xmath64 .
the offsets associated with the sub - cloud whose redshift most closely matches that of the are shown in the lower panel of figure [ offsets ] . given the small number of systems for which the lyman series can be traced , at least to ly@xmath23 , the statistics of the offset distribution are not very meaningful . however , the scatter is now clearly smaller , although the high @xmath3 ( ) ly@xmath0 lines at the top end of the column density distribution may still have several sub - components that are unresolved in our data and the offset may be further reduced if the lyman series could be traced down to subsequent transitions .
we investigated a possible solution to the problem of an unknown offset from the predicted position of systems discussed above .
the technique involves scanning for the maximum absorption around the predicted position of absorption and re - centering the stack at this wavelength . in practice , each ( de - redshifted ) data section is scanned @xmath65 from the predicted line center ( i.e. for @xmath54 = 0 ) and the section re - centered on the pixel with the maximum optical depth ( @xmath66 ) prior to stacking .
figure [ preshift ] shows an example of how this technique is clearly able to improve upon a direct stack in the presence of a @xmath54 for a synthetic spectrum containing lines with log / = @xmath672.0 . it can be seen that without re - centering , whilst the overall level of the continuum is below unity for the ` smeared ' stack , all profile information is lost
such a broad depression may be mistaken for compound errors in the continuum level and lost in the subsequent renormalization of the stacked spectrum . executing a re - center on the @xmath66 pixel before stacking recovers almost all of the original signal in this simulation ,
the characteristic line profile is prominent and will not be lost when the post - stacking continuum is re - fitted .
however , as one attempts to detect progressively weaker absorption , the likelihood of centering on a noise pixel becomes higher and this technique is no longer efficient . in order to determine whether a ` re - center ' is a viable improvement to the stacking method , given the s /
n of the data and the low column density of the targeted features , we performed a feasibility study in which test absorption lines were created over a range of / ratios .
the results of this study are shown in figure [ test_shift ] where we plot the distribution of @xmath57 caused by line center misidentification ( i.e. where @xmath66 is a noise feature rather than line trough ) as a function of / ratio .
clearly , when the @xmath38 of the offset caused by trough misidentification exceeds the value of the intrinsic offset we are trying to overcome , this re - centering technique is no longer useful .
however , these are conservative estimates for how well the centering would perform on real data , since we would expect some trough misidentification from contaminating ( i.e. anything but ) lines in addition to the effects of noise investigated here .
since we need to determine whether the low column density ly@xmath0 lines have the same metallicity as their high @xmath3 ( ) counterparts , we must ideally reach levels of sensitivity deeper than log /=@xmath52 .
figure [ test_shift ] shows that for log / @xmath682.6 , this technique no longer compensates adequately for the intrinsic offset and is therefore unable to improve the stacking technique in the search for weak lines .
smoothing the spectrum over several pixels before locating @xmath66 improves the pre - shifting slightly as shown by the gray squares in figure [ test_shift ] .
however , even with smoothing , this re - centering procedure can not compensate for a @xmath57 = 2.6 @xmath56 below a log / = @xmath69 . in summary , by stacking together the regions associated with low column density ly@xmath0 lines in q1422 + 231 and
apm 08279 + 5255 , we have obtained a s / n = 1250 composite spectrum which shows no residual absorption .
using simulated spectra we find that for log / = @xmath51 in the weak ly@xmath0 lines we would expect a 4@xmath38 detection of the composite absorption .
the shortcomings of this technique are discussed and we focus on an observed redshift offset between the position of ly@xmath0 and its associated . the redshift offset determined from the detected systems has a dispersion @xmath57 = 2.6@xmath70 which corresponds to a @xmath4 = 17 km s@xmath5 .
we have investigated how this offset would affect the stacking procedure by using simulated spectra and find that a @xmath4 = 17 km s@xmath5 will reduce the sensitivity of this method by a factor of two and that a metallicity of log / = @xmath60 is now required to achieve a 4@xmath38 detection .
we are unable to know whether the same redshift offset persists in the low @xmath3 ( ) clouds targeted by the stacking technique , but a large range of offsets are measured over the full column density range of detected systems , so one must clearly take into account the possible repercussions when interpreting the stacked data .
a potentially more robust way of measuring weak absorption in high s / n spectra is to analyse the optical depths in each ly@xmath0 forest pixel and its corresponding pixel ( cowie & songaila 1998 ) .
this technique also has the advantage that by tracing down the lyman series , one can determine the / ratio in each pixel over a very large range of @xmath71(ly@xmath0 ) . in paper
i we investigated the potential of this method with apm 08279 + 5255 and critically analysed its performance on synthetic spectra . specifically , we investigated the effect of including a redshift offset and how the results depended on the @xmath1-value of .
we concluded that , despite the excellent quality of the spectrum , the results from the spectrum of apm 08279 + 5255 were inconclusive and could not determine whether the metallicity of the igm is constant or diminished at low values of , because at the lowest values of @xmath71(ly@xmath0 ) both scenarios were consistent with the observations .
the spectrum of q1422 + 231 presented here not only has a significantly higher s / n than that of apm 08279 + 5255 , but has many of its higher order lyman lines accessible for analysis and considerably less contaminating absorption by , for example , systems ( ellison et al .
this spectrum therefore represents the best data yet obtained for this analysis of the low column density ly@xmath0 forest .
briefly , the optical depth technique consists of stepping through the ly@xmath0 forest measuring @xmath71(ly@xmath0 ) for each pixel .
the noise ( @xmath38 ) array is used to determine which pixels are included in the analysis via a series of optical depth criteria to account for effects such as saturation .
using the values in the noise arrays rather than fixing the rejection criteria provides maximum flexibility for this technique so that it can be readily applied to spectra of different s / n ratios . for pixels with a residual flux ( @xmath72 for a normalised spectrum )
@xmath73 above the zero level , we trace down the lyman series since there is too little residual flux in these saturated pixels to determine an accurate optical depth from ly@xmath0 alone . however , the danger here is that higher order lines may be contaminated by lower redshift ly@xmath0 .
we therefore use @xmath71(ly@xmath0 ) = minimum(@xmath71(ly@xmath74)@xmath75 / @xmath76 ) over all observed higher order lines using the higher order pixel if @xmath77(ly@xmath78 .
this minimizes the effect of contamination and maximizes the number of usable pixels and range of @xmath71(ly@xmath0 ) which can be considered for analysis .
if no @xmath77(ly@xmath79 pixels are found , then the pixel is discarded .
the position of the associated @xmath81548 , 1550 lines are calculated and , again to avoid contamination , we use @xmath71(1548)= minimum(@xmath80,@xmath81(1550))-values ( oscillator strengths ) in the doublet is 2:1 ] if the flux in the 1550 component @xmath82 , otherwise only @xmath71(1548 ) is considered .
the optical depths are then binned according to their corresponding @xmath71(ly@xmath0 ) and the median determined for each interval .
taking the median of a large number of pixel optical depths not only provides a statistical advantage over considering a relatively small number of lines ( as for the stacking method ) , but also is much less susceptible to non - gaussian effects . in order to estimate 1 @xmath38 errors for the optical depth determinations , we used bootstrap re - sampling with 2sections of the ly@xmath0 forest and corresponding , i.e. we drew @xmath74 random sections from the complete set of @xmath74 sections ( in this case @xmath83 ) that comprise the original data , with replacement .
this procedure was repeated 250 times and the 1@xmath38 error was taken to be the dispersion of these 250 realizations .
figure [ 1422_tau ] shows the results obtained for the optical depth analysis of q1422 + 231 .
the shape of the optical depth distribution appears consistent with a constant level of / ( i.e. parallel to the dashed line ) over optical depths from @xmath71(ly@xmath0 ) @xmath17 100 down to @xmath17 2 3 , below which @xmath71 ( ) flattens off to an approximately constant value .
each of the low optical depth bins ( @xmath71(ly@xmath0 ) @xmath9 3 ) contains approximately 10% of the pixels .
this percentage decreases with increasing @xmath71(ly@xmath0 ) with only approximately 2% of pixels in the @xmath71(ly@xmath0 ) = 10 bin .
in addition , we determine optical depths in pixel pairs separated by the @xmath81548 , 1550 doublet ratio _ regardless of @xmath71(ly@xmath0 ) _ over the entire range considered for absorption .
this is done using the same method of doublet comparison as before to eliminate contamination .
the median of this reference distribution is plotted in figure [ 1422_tau ] as a dotted line and represents the median absorption for an effectively random set of pixel pairs separated by doublet ratio and will include the effects of noise and low level contamination expected to affect our results .
this median optical depth is higher than the observed @xmath71 ( ) for all bins with @xmath71(ly@xmath0)@xmath84 1 .
this is a significant result which indicates that the absorption for these optical depths is less than that expected by selecting random pixel pairs .
the key question here is whether the signal in the low optical depth pixels is due to absorption in low density regions of the igm or whether the flattening of the data is caused by some limiting factor in our analysis such as contamination or noise .
as we saw in the previous section , thorough simulations of the analysis technique are vital for interpreting our results . here
, we have taken the ly@xmath0 forest directly from the data and artificially enriched it with to test our methods and address several questions . in paper
i we explored the effect of @xmath1-values , redshift offset and noise on synthetic spectra in order to determine whether a break in the / distribution could be distinguished from a constant ratio in the spectrum of apm 08279 + 5255 . to review these findings and to give a visual impression of the optical depth analysis
, we show the results of this technique on four synthetic spectra in figure [ 2panel ] and compare them with the results from q1422 + 231 ( solid points ) .
the top panel ( ` a ' ) of this figure shows the optical depth analysis of 2 spectra , one of which has a constant log / = @xmath53 in all ly@xmath0 lines ( shown as a dotted line ) and a second spectrum which has log / = @xmath53 only in log @xmath3 ( ) @xmath6 14.5 lines ( dot - dashed line ) , with no noise added to either spectrum redward of ly@xmath0 .
the bottom panel shows the same spectra as panel ` a ' except that noise has now been added to both spectra , based on the error array of the actual data ( typically s / n = 200 redward of ly@xmath0 ) .
all four spectra use the real ly@xmath0 forest of q1422 + 231 and have had added to the synthetic spectrum based on the fitted linelist . in addition , all four synthetic spectra in figure [ 2panel ] have a @xmath54 = 2.6 @xmath59 and @xmath1 ( ) = @xmath50(ly@xmath0 ) .
the dashed line indicates constant log / = @xmath53 .
there are several points to note here .
firstly , in the absence of noise , there is a clear drop in @xmath71 ( ) below @xmath71(ly@xmath0 ) @xmath85 if no is added in ly@xmath0 lines with log @xmath3 ( ) @xmath9 14.5 .
however , this steep decline is much less drastic when noise is included , and although @xmath71 ( ) shows a steady decrease down to @xmath71(ly@xmath0 ) @xmath17 1 , below this value it flattens off to an approximately constant value .
the inclusion of noise also causes the same apparent flattening in the spectrum of constant / , although the @xmath71 ( ) in this spectrum is consistently above the value measured for the dot - dashed line .
there is also a small contribution to this flattening from line blending which changes the overall slope of the expected / , an effect that is exacerbated for larger @xmath1 ( ) .
this shows that even in relatively high s / n spectra , distinguishing a break in the / distribution is very difficult . instead , as we shall see later in this section , one of the main uses of this method is to determine whether the optical depths measured with this technique can be adequately accounted for with the detected systems or whether there must be significant amount of still below our current detection limit .
we also note that for large @xmath71(ly@xmath0 ) the measured @xmath71 ( ) is always less than expected from the dashed line , given the input ratio of /.
this is due to two effects .
firstly , since this technique considers @xmath71(ly@xmath0 ) as the minimum value obtained by tracing down the lyman series when a line is saturated , if contamination is successfully removed we will tend to under - estimate @xmath71(ly@xmath0 ) at these optical depths due to noise ( remember that the real ly@xmath0 forest is used in the simulated spectra so that this will be an effect even in panel ` a ' of figure [ 2panel ] ) .
secondly , the that is included in the synthetic spectra is based on a linelist of fitted values which will probably not be accurate for saturated lines
. these effects also account for the drop of @xmath71 ( ) at very large @xmath71(ly@xmath0 ) which contain @xmath9 1% of the total pixels and are therefore very uncertain . clearly , the exact results of the optical depth analysis are sensitive to the combination of several factors including blending , @xmath1-values and noise , which we will discuss further later in this section .
therefore , rather than attempting to directly fit the observed distribution of optical depths in q1422 + 231 we aim to determine whether the optical depths measured in the spectrum can be explained solely by the relatively strong absorbers detected directly and presented in 4 or whether the results from this analysis are indicative of additional . if the latter is true , are these metals in the low density igm ? we must also investigate whether our results can be explained in terms of contamination , noise or some other limiting factor in the data . finally , we consider the effect of scatter and redshift offset ( @xmath54 ) , which have been found to compromise the efficiency of the stacking method .
three synthetic spectra were produced with ly@xmath0 forest absorption taken directly from the data ( i.e. not reconstructed from a linelist ) and the 34 detected systems re - produced from the voigt profile parameters in table [ civlist ] .
spectrum ` a ' has had no further metals added and therefore shows the results expected of an optical depth analysis if we had already uncovered all of the in the spectrum .
spectra ` b ' and ` c ' have both been enriched with additional metals . in spectrum `
b ' , is included in all strong ( log @xmath3 ( ) @xmath6 14.5 ) lines with @xmath3 ( ) = 12.0 , i.e. the detection limit for directly identifiable systems .
this spectrum therefore represents the maximum amount of that could be ` hidden ' in high column density ly@xmath0 lines .
in addition to this ` hidden ' in strong lines and the fitted in table [ civlist ] , spectrum ` c ' has been enriched with a constant log / = @xmath53 in weak ly@xmath0 lines ( @xmath3 ( ) @xmath9 14.5 ) .
the results from analysis of these three synthetic spectra are compared with the data in figure [ testabc ] .
for all three spectra , we include noise taken from the @xmath38 error array , a random redshift offset in the position of ( @xmath86 ) and take @xmath1 ( ) = @xmath87 @xmath1(ly@xmath0 ) .
the conclusions that can be drawn from figure [ testabc ] are as follows .
first , there is clearly more in the data than we have directly identified in 4 , since the dotted line of the synthetic spectrum in the top panel is well below the solid line at all but the very highest @xmath71(ly@xmath0 ) points . for @xmath71(ly@xmath0 ) @xmath29 3 , the optical depths can be recovered when additional metals are added into strong ly@xmath0absorbers , but below our current detection limit ( spectrum ` b ' ) .
this supports the results of 4 where we determine that @xmath36 is consistent with a power law distribution that continues down to log @xmath3 ( ) = 11.75 , with no evidence for a turnover in column density distribution .
however , adding at the limit of our detection in log @xmath3()@xmath6 14.5 lines with no extra metals in weaker ly@xmath0 clouds can not reproduce the measured @xmath71 ( ) in @xmath71(ly@xmath0)@xmath9 3 pixels .
the results from spectrum ` c ' in the bottom panel of figure [ testabc ] which include log / = @xmath53 in weak ly@xmath0 lines show that the observed optical depths in the low @xmath71(ly@xmath0 ) pixels are consistent with the presence of some metals in significantly lower column density clouds , although for @xmath71(ly@xmath0 ) @xmath6 1 @xmath71 ( ) exceeds the observed value . from figure [ 2panel ]
we have also seen that a constant log / = @xmath53 in all ly@xmath0 lines is a good approximation to the data .
we stress here that these simulations are not an attempt to fit the observed distribution of optical depths , rather they are tests to determine whether the systems in table [ civlist ] can account for the measured absorption . if not , then the objective is to investigate in which column density regime additional metals could be added in order to achieve the observed quota .
the results from these simulations show that the analysis of optical depths in the spectrum of q1422 + 231 is consistent with hitherto undetected in both the strong and weak ly@xmath0 lines . whilst the results from our determination of the column density distribution from detected systems
is consistent with a power law function @xmath36 that continues down to at least log @xmath3 ( ) = 11.7 , these optical depth results provide _ direct evidence _ that there are more metals lurking below our current detection limit .
we investigate the effect that contaminating lines may have on this result .
the analysis may be affected not only by low level contamination from other metal lines such as or telluric absorption ( strong absorbers will have been rejected by the doublet strength comparison discussed in 6.1 ) , but also effects due the non - uniformity of ionizing sources and noise / fluctuations in the continuum level due to fitting errors .
this latter effect will cause fluctuations around the true continuum which will be important only if they are greater than the level of the noise .
an additional ( possibly systematic ) error may be present if the low order polynomial fit to the continuum is consistently over or under - estimated . to test the continuum fit of the regions , sections of the data redward of ly@xmath0 deemed to contain no obvious absorption were examined . from a total of
@xmath17 5000 pixels , a mean flux of 1.0002 and a median of 1.0001 were determined , vaules one order of magnitude lower than @xmath71 ( ) @xmath88 .
it therefore seems unlikely that a systematic error in the continuum fit is the cause of the observed flattening of optical depths seen in figure [ 1422_tau ] .
the distribution of noise pixels would also suggest that there is no significant effect from weak telluric lines .
errors in the continuum fit blueward of ly@xmath0 emission , however , are likely to be a more serious effect , firstly because selection of absorption - free zones is difficult and secondly because the s / n is lower ( 50 150 ) .
the result of continuum fitting errors in the forest in this analysis will be to classify pixels with small @xmath71(ly@xmath0 ) into the wrong optical depth bin .
this will effectively associate with the wrong @xmath71(ly@xmath0 ) and averaging out the metal absorption by this ` mis - binning ' could explain the observed flattening of the measured optical depths .
the effect of a small continuum error is therefore similar to the effect of noise , affecting only those pixels with optical depths smaller than the fitting error .
we also find that including a large number of weak contaminating lines such as or could increase the measured @xmath71 ( ) at low @xmath71(ly@xmath0 ) to the constant level determined .
however , whilst realistic column densities do have a small effect on the optical depth result , it is unlikely that weak contaminating lines account for all of the observed absorption .
as discussed previously , a scatter is expected in the observed / values even if [ c / h ] and the ionizing background remain uniform . to investigate the importance of this effect , we simulated a ` control ' spectrum by re - producing directly the ly@xmath0forest of q1422 + 231 and using @xmath1 ( ) = @xmath50(ly@xmath0 ) and an input log / = @xmath53 with no scatter and no redshift offsets . for comparison
, we then produced a second spectrum in the same way , but rather than adopting a constant / , a gaussian scatter was introduced with @xmath38(/ ) = @xmath89 .
the comparison between the two simulated spectra is presented in figure [ scatter ] where the ` control ' spectrum is shown as open squares connected with a solid line and the spectrum containing a scatter of metallicities is shown as solid diamonds connected with a dotted line .
the points have been offset from one another in the figure for clarity .
the points at high @xmath71(ly@xmath0 ) are again very uncertain due to the reasons previously discussed with regards to figure [ 2panel ] .
the error bars are estimated using the same bootstrap technique as employed for the data .
the results from the 2 spectra are entirely consistent with one another and indicate that a gaussian scatter will therefore not affect the overall median of @xmath71 ( ) . whilst the extent and magnitude of the scatter in the distribution of / values has not yet been fully investigated in simulations down to column densities below log @xmath3 ( ) @xmath17 14.0 13.5 , this result should not be significantly affected unless the scatter becomes highly non - gaussian at low @xmath3 ( ) . also plotted in figure [ scatter ] are the results of the optical depth analysis if a redshift offset is included in the position of the line . as before , metals are added to all ly@xmath0 lines with a metallicity log / = @xmath53 .
there is no scatter in these values , but an offset in the position of has been included , drawn at random from a gaussian distribution with @xmath90 ( @xmath4 = 17 km s@xmath5 ) .
these points are also consistent with the control spectrum , so we can conclude from this simulation batch that neither scatter in the metallicity nor redshift offset will have a large effect on the outcome of our optical depth analysis .
finally we note that more spectra are required in order to provide a representative analysis of the high redshift igm , since our view of the enrichment of the ly@xmath0 forest probed with a single sightline is clearly blinkered . from the list of systems in table [ civlist ]
it can be seen that these absorbers are not uniformly distributed in redshift .
for example , splitting this spectrum in two by redshift produces vastly different optical depth distributions due to a relative dearth of systems that spans over 300 in this spectrum in the range @xmath91 .
with more spectra , there is the possibility of not only developing a more representative study of these high redshift absorbers but also investigating the redshift evolution of systems .
in this paper , we have addressed the enrichment history of the igm by studying the ly@xmath0 forest and its associated systems in a very high s / n ( @xmath92 , high resolution ( @xmath17 8 km s@xmath5 ) spectrum of the well - known lensed quasar , q1422 + 231 obtained with keck / hires .
the numerous systems associated with high column density ly@xmath0 absorbers are fitted with voigt profiles defined by a redshift , @xmath1-value and column density for each component line .
we investigate the column density distribution , @xmath36 , to very sensitive levels and detect several weak systems which had not been previously identified in lower quality spectra of the same object .
we determine a power law index @xmath2 which continues down to log @xmath3 ( ) = 12.2 before starting to turnover . by simulating synthetic absorption lines with @xmath1-values taken at random from the observed distribution
, we estimate a correction factor to account for incompleteness and find that the corrected data points now indicate that the power law continues down to at least log @xmath3 ( ) = 11.75 , a factor of ten more sensitive than previous measurements ( e.g. songaila 1997 ) .
this shows that even at these low column densities there is no evidence for a flattening of the power law and therefore there are probably many more systems that lie below the current detection limit .
we investigate two methods with which it may be possible to recover these weak systems .
firstly , we select 67 ly@xmath0lines with 13.5 @xmath9 @xmath3 ( ) @xmath9 14.0 in q1422 + 231 and apm 08279 + 5255 and produce a stacked spectrum centered on the predicted position of @xmath401548 .
the composite stack has a s / n = 1250 and shows no residual absorption ; we use synthetic stacked spectra to determine a 4@xmath38 upper limit of log / = @xmath51 .
we critically assess the accuracy of this method by performing the stacking procedure on a suite of simulated , synthetic spectra and identify several associated problems .
we investigate , in particular , the effect of a redshift offset between the position of the ly@xmath0 line and its associated . with improved statistics ,
we refine the redshift offset determined in paper i and find that a @xmath4=17 km s@xmath5 is present in the systems which we detect directly . by including a random redshift
offset drawn from a gaussian distribution with @xmath4=17 km s@xmath5 in our stacking simulations , we find that log / = @xmath60 is now required to achieve a 4@xmath38 detection .
this limit is still consistent with current measured metallicities in higher column density ly@xmath0 clouds and is therefore not sufficiently sensitive to determine whether the / ratio drops in low @xmath3 ( ) lines .
a feasibility study is performed to assess the effectiveness of a ` re - center ' on the maximum optical depth pixel prior to stacking for removing the effect of an unknown offset .
we find that this technique can not improve the quality of the stack result in the / regime that we are targeting .
it is not yet clear whether the observed redshift offset persists in the low column density ly@xmath0 clouds , but it must be considered a factor . moreover ,
the effects of contamination , continuum fitting errors and anomolous pixels also pose problems for this technique , although in simulations the redshift offset appears to be a major effect . therefore ,
if this technique is to be pursued further to reach a meaningful detection limit , an improvement in s / n by at least a factor of two is required . the optical depth technique introduced by cowie & songaila ( 1998 )
is considered as an alternative approach .
this technique is shown to exhibit several advantages over the stacking method such as its insensitivity to redshift offsets and its ability to exclude contamination from other absorption features .
we develop this technique as a method that can be used to test whether the detected systems represent the full tally of absorbers .
the data have optical depths consistent with an almost constant log / @xmath93 down to @xmath71(ly@xmath0)@xmath17 23 , below which @xmath71 ( ) flattens off to an approximately constant value .
it is unlikely that this flattening is real and is most probably caused by the effect of noise and/or continuum errors , even at such high s / n ratios as have been achieved in this spectrum .
given the many effects that may alter the measured @xmath71 values , such as blending and noise , we do not attempt to fit the observed distribution of optical depths . instead , our strategy is to test whether the detected systems are sufficient to reproduce the measured @xmath71 ( ) and if not , determine how much additonal may be present below our current detection limit . by simulating synthetic spectra with different enrichment recipes , we have shown that the systems detected directly in the spectrum are not sufficient to reproduce the results of the optical depth analysis of q1422 + 231 .
this is in agreement with the conclusions drawn from the column density distribution of , i.e. that the data are consistent with a continuous power law @xmath36 down to at least @xmath3 ( ) = 11.75 and that there is therefore likely to be a large number of weak metal lines not yet directly detected .
this agrees with the conclusions of cowie & songaila ( 1998 ) . in order to interpret the results from the optical depth method
, we have simulated synthetic spectra with a range of input / ratios .
we find that including associated with strong ly@xmath0 lines ( @xmath3 ( ) @xmath614.5 ) but below the current detection limit , in addition to the 34 identified systems , can reproduce the optical depths measured in the observed spectrum for @xmath71(ly@xmath0 ) @xmath6 3 . for smaller values of @xmath71(ly@xmath0 ) , some additional metals are required and we find that including in low column density lines ( @xmath3 ( ) @xmath9 14.5 ) with log / @xmath94 produces optical depth results consistent with those measured in the data .
however , determining the precise / in the low @xmath3 ( ) ly@xmath0 clouds and the density to which the enrichment persists is still uncertain due to effects such as noise and continuum fluctuations and it is therefore not possible to say whether the low column density forest is pristine . nevertheless , we find that even in the high optical depth pixels ( which will not be seriously affected by noise or small continuum errors ) the detected systems are not sufficient to cause all the measured absorption and clearly there are more metals in the igm than we can currently detect
. the authors would like to thank bob carswell , alberto fernandez - soto , matthias steinmetz and the anonymous referee for discussions and useful suggestions .
we are grateful to len cowie for his support and enthusiasm towards this project .
sle and js are supported by a pparc postgraduate award and js acknowledges additional funding from the isaac newton trust .
we are also grateful to the expert assistance of the staff at the keck i telescope for their help in obtaining the spectra presented here .
haehnelt , m. g. , 1998 , the young universe : galaxy formation and evolution at intermediate and high redshift . edited by s. dodorico , a. fontana , and e. giallongo .
asp conference series ; vol .
146 , p.249 cccr c1 & 2.90969 & 12.44 & 8.8 + & 2.91005 & 11.90 & 11.4 + c2 & 2.94522 & 11.74 & 4.2 + & 2.94551 & 12.32 & 12.6 + c3 & 2.94754 & 12.47 & 7.2 + c4 & 2.96065 & 12.50 & 13.4 + & 2.96110 & 12.68 & 13.6 + & 2.96146 & 12.61 & 7.6 + & 2.96197 & 13.28 & 19.3 + & 2.96235 & 12.80 & 9.7 + c5 & 2.97143 & 12.23 & 13.3 + c6 & 2.97584 & 12.46 & 35.3 + & 2.97622 & 12.76 & 9.2 + c7 & 2.99922 & 12.66 & 16.7 + & 2.99959 & 11.61 & 6.2 + c8 & 3.03505 & 12.14 & 8.3 + c9 & 3.03672 & 12.35 & 28.4 + c10 & 3.06338 & 12.98 & 24.5 + & 3.06433 & 12.66 & 11.0 + & 3.06383 & 12.01 & 5.6 + c11 & 3.07101 & 12.43 & 7.9 + c12 & 3.08666 & 12.95 & 12.8 + c13 & 3.08990 & 12.59 & 7.9 + & 3.09020 & 13.33 & 30.5 + & 3.09051 & 13.10 & 47.8 + & 3.09108 & 12.91 & 8.8 + c14 & 3.09468 & 12.17 & 14.1 + c15 & 3.11930 & 11.87 & 11.1 + & 3.11973 & 12.12 & 14.4 + c16 & 3.13252 & 12.46 & 26.0 + c17 & 3.13379 & 12.80 & 17.5 + & 3.13409 & 12.27 & 6.4 + & 3.13448 & 13.00 & 16.2 + c18 & 3.13712 & 12.89 & 16.9 + & 3.13799 & 12.33 & 27.5 + c19 & 3.19143 & 12.09 & 6.5 + c20 & 3.23330 & 11.77 & 2.6 + c21 & 3.24047 & 12.53 & 40.4 + c22 & 3.25716 & 12.50 & 42.5 + c23 & 3.26564 & 12.65 & 26.9 + & 3.26584 & 11.95 & 9.8 + c24 & 3.27596 & 12.02 & 13.0 + c25 & 3.31710 & 12.35 & 13.0 + c26 & 3.33410 & 11.84 & 7.7 + c27 & 3.37994 & 12.55 & 15.5 + & 3.38045 & 12.30 & 12.3 + & 3.38135 & 11.87 & 5.0 + & 3.38167 & 13.27 & 11.6 + & 3.38223 & 13.23 & 14.0 + & 3.38271 & 12.61 & 8.1 + & 3.38316 & 12.22 & 20.4 + c28 & 3.41080 & 12.20 & 10.7 + & 3.41149 & 12.90 & 21.2 + c29 & 3.44691 & 12.97 & 8.5 + & 3.44736 & 13.46 & 13.4 + c30 & 3.47963 & 12.85 & 31.7 + [ civlist ] cccr c31 & 3.49488 & 12.46 & 11.4 + c32 & 3.51465 & 12.83 & 7.5 + & 3.51497 & 12.59 & 15.3 + c33 & 3.51770 & 11.87 & 10.2 + c34 & 3.53490 & 12.57 & 34.7 + & 3.53505 & 12.69 & 16.0 + & 3.53549 & 12.43 & 7.7 + & 3.53599 & 13.69 & 19.9 + & 3.53659 & 13.12 & 21.7 + & 3.53738 & 13.22 & 17.2 + & 3.53872 & 13.58 & 9.2 + & 3.53937 & 13.58 & 14.1 + & 3.54005 & 12.82 & 25.6 + & 3.53848 & 13.33 & 25.0 + & 3.54140 & 12.10 & 4.4 + cccc 2.947 & [email protected] & 12.47 & -2.71 + 2.999 & [email protected] & 12.70 & -3.07 + 3.035 & [email protected] & 12.14 & -2.50 + 3.037 & [email protected] & 12.34 & -2.29 + 3.063 & [email protected] & 13.18 & -2.18 + 3.071 & [email protected] & 12.43 & -1.42 + 3.132 & [email protected] & 12.46 & -1.28 + 3.134 & [email protected] & 13.26 & -2.45 + 3.137 & [email protected] & 13.00 & -2.93 + 3.191 & [email protected] & 12.09 & -2.80 + 3.276 & [email protected] & 12.02 & -2.05 + 3.318 & [email protected] & 12.35 & -1.62 + 3.334 & [email protected] & 11.84 & -2.94 + 3.382 & [email protected] & 13.68 & -3.02 + 3.411 & [email protected] & 12.98 & -2.21 + 3.479 & [email protected] & 12.85 & -2.44 + 3.515 & [email protected] & 13.02 & -2.32 + 3.518 & [email protected] & 11.87 & -1.42 + 3.539 & [email protected] & 14.29 & -1.88 + [ hifits ] | we have obtained an exceptionally high s / n , high resolution spectrum of the gravitationally lensed quasar q1422 + 231 . a total of 34 systems are identified , several of which had not been seen in previous spectra .
voigt profiles are fitted to these systems and to the entire ly@xmath0 forest in order to determine column densities , @xmath1-values and redshifts for each absorption component .
the column density distribution for is found to be a power law with index @xmath2 , down to at least log @xmath3 ( ) = 12.3 .
we use simulations to estimate the incompleteness correction and find that there is in fact no evidence for flattening of the power law down to log @xmath3 ( ) = 11.7 a factor of ten lower than previous measurements . in order to determine whether the enrichment extends to even lower column density clouds , we utilize two analysis techniques to probe the low column density regime in the ly@xmath0forest .
firstly , a composite stacked spectrum is produced by combining the data for q1422 + 231 and another bright qso , apm 08279 + 5255 .
the s / n of the stacked spectrum is 1250 and yet no resultant absorption is detected .
we discuss the various problems that affect the stacking technique and focus in particular on a random velocity offset between and its associated which we measure to have a dispersion of @xmath4 = 17 km s@xmath5 .
it is concluded that , in our data , this offset results in an underestimate of the amount of present by a factor of about two and this technique is therefore not sufficiently sensitive to probe the low column density ly@xmath0 clouds to meaningful metallicities .
secondly , we use measurements of individual pixel optical depths of ly@xmath0 and corresponding lines .
we compare the results obtained from this optical depth method with analyses of simulated spectra enriched with varying enrichment recipes . from these simulations , we conclude that more than is currently directly detected in q1422 + 231 is required to reproduce the optical depths determined from the data , consistent with the conclusions drawn from consideration of the power law distribution . |
in this paper we prove a theorem about homogeneous continua inspired by a result about finitely presented groups ( @xcite ) .
[ continua ] let @xmath0 be a locally connected , simply connected , homogeneous continuum .
then no arc separates @xmath0 .
we recall that an arc in @xmath0 is the image of a 1 - 1 continuous map @xmath1\to x$ ] .
we say that an arc @xmath2 separates @xmath0 if @xmath3 has at least two connected components .
we say that @xmath0 is simply connected if it is path connected and every continuous map @xmath4 can be extended to a continuous @xmath5 ( where @xmath6 is the 2-disc and @xmath7 its boundary circle ) .
the proof of theorem [ continua ] relies on alexander s lemma for the plane ( see @xcite ) and our generalization of this lemma to simply connected spaces ( see sec .
2 ) . there is a family air between continua theory and group theory .
this became apparent after gromov s theory of hyperbolic groups ( @xcite ) .
gromov defines a boundary for a hyperbolic group which is a continuum on which the group acts by a ` convergence action ' .
the classic ` cyclic elements ' decomposition theory of whyburn was extended recently by bowditch ( @xcite ) in this context and it gave deep results in group theory . `
asymptotic topology ' , introduced by gromov ( @xcite ) and developed further by dranishnikov ( @xcite ) , shows that the analogy goes beyond the realm of hyperbolic groups .
the ` philosophy ' of this is that topological questions that make sense for continua can be translated to ` asymptotic topology ' questions which make sense for groups ( see @xcite for a dictionary between topology and asymptotic topology ) .
one wonders whether theorem [ continua ] holds in fact for all locally connected , homogeneous continua of dimension bigger than 1 : let @xmath0 be a locally connected , homogeneous continuum of dimension 2 .
is it true that no arc separates @xmath0 ?
we note that by a result of krupski ( @xcite ) , homogeneous continua are cantor manifolds .
it follows that no arc separates a homogeneous continuum of dimension bigger than 2 .
krupski and patkowska ( @xcite ) have shown that a similar property ( the disjoint arcs property ) holds for all locally connected homogeneous continua of dimension bigger than 1 which are not 2-manifolds .
we remark that by @xcite if a cantor set separates a homogeneous continuum @xmath0 then @xmath8 .
so question 1 is equivalent to the following question : is it true that if no cantor set separates a locally connected , homogeneous continuum @xmath0 , then no arc separates @xmath0 ?
restated in this way the question makes sense also for boundaries of hyperbolic groups .
in fact a similar question can be formulated for finitely generated groups too ( see @xcite ) .
not much is known about locally connected , simply connected homogeneous continua .
one motivation to study them is the analogy with finitely presented groups .
another reason is that one could hope for a classification of such continua in dimension 2 : are the 2-sphere and the universal menger compactum of dimension 2 the only locally connected , simply connected , homogeneous continua of dimension 2 ?
we recall that @xmath7 and the universal menger curve are the only locally connected homogeneous continua of dimension 1 ( @xcite ) .
prajs ( @xcite , question 2 ) asks whether @xmath9 is the only simply connected homogeneous continuum of dimension 2 that embeds in @xmath10 .
a related question about locally connected , simply connected continua that makes sense also for finitely presented groups is the following : let @xmath0 be a locally connected , simply connected homogeneous continuum which is not a single point . does @xmath0 contain a disc ?
we remark that in the group theoretic setting the answer is affirmative for hyperbolic groups ( @xcite ) . by a result of prajs ( @xcite ) a positive answer to this
would imply that @xmath9 is the only locally connected , simply connected , homogeneous continuum of dimension 2 that embeds in @xmath11 .
we refer to prajs list of problems ( @xcite ) for more questions on homogeneous continua .
let @xmath0 be a metric space . a path @xmath12 is a continuous map @xmath13\to x$ ] . a simple path or an arc @xmath14 , is a continuous and @xmath15 map @xmath16\to x$ ] .
we will identify an arc with its image . for a path @xmath12
we denote by @xmath17 the set of its endpoints , i.e. @xmath18 .
an arc @xmath14 separates @xmath0 if @xmath19 has at least two connected components .
if @xmath20 we say that an arc @xmath14 separates @xmath21 from @xmath22 if @xmath14 separates @xmath0 and @xmath23 belong to distinct components of @xmath19 .
let @xmath14 be an arc of @xmath0 .
on @xmath14 we define an order @xmath24 as follows : if @xmath25 then @xmath26 if and only if @xmath27 .
we denote by @xmath28}_{{\alpha}}$ ] the set of all @xmath29 such that @xmath30 .
similarly we define @xmath31 and @xmath32_{{\alpha}}$ ] .
when there is no ambiguity we write @xmath33 $ ] instead of @xmath28}_{{\alpha}}$ ] and @xmath34 instead of @xmath26 .
finally , if @xmath35 $ ] we denote by @xmath36 the point @xmath37 ( where @xmath38 ) .
we recall alexander s lemma from plane topology ( see theorem 9.2 , p.112 of @xcite ) .
let @xmath39 be closed sets on the plane such that either @xmath40 or @xmath41 is connected and at least one of @xmath39 is bounded .
let @xmath42 .
if there is a path joining @xmath23 in @xmath43 and a path joining @xmath23 in @xmath44 then there is a path joining @xmath23 in @xmath45 .
it is easy to see that alexander s lemma also holds for the closed disc @xmath6 in the case that @xmath40 .
in fact this implies that this lemma holds in general for every simply connected space .
in particular we have the following : let @xmath0 be a simply connected space , @xmath39 disjoint closed subsets of @xmath0 and let @xmath46 .
if there is a path joining @xmath23 in @xmath47 and a path joining @xmath23 in @xmath48 then there is a path joining @xmath23 in @xmath49 .
let @xmath50 be paths joining @xmath23 such that @xmath51 .
we consider the closed path @xmath52 and let @xmath4 be a parametrization of this path . since @xmath0 is simply connected , @xmath53 can be extended to a map @xmath54 .
then @xmath55 and @xmath56 are disjoint , closed subsets of @xmath6 .
clearly , neither @xmath55 nor @xmath56 separates @xmath23 , therefore , using alexander s lemma for the closed disc , we have that there is a path @xmath12 that joins @xmath23 without meeting @xmath57 .
this implies that @xmath58 is a path from @xmath21 to @xmath22 that does not meet @xmath59 .
for the rest of this paper we assume that @xmath0 is a simply connected , locally connected continuum .
[ a.0 ] let @xmath60 be a connected open subset of @xmath61 be a connected component of @xmath62 and let @xmath63 such that @xmath64 and @xmath65 .
then there is a path @xmath12 in @xmath60 connecting @xmath21 to @xmath22 such that @xmath12 is contained in the @xmath66-neighborhood of @xmath62 .
let @xmath67 be the union of the open balls @xmath68 with center @xmath69 and radius @xmath66 .
let @xmath70 be the connected component of @xmath67 containing @xmath71 .
clearly @xmath72 so there is a path in @xmath0 joining them that does not intersect @xmath73 . on the other hand @xmath63 so there is a path in @xmath0 joining them that does not intersect @xmath62 .
since @xmath74 and @xmath75 are closed , applying alexander s lemma for the simply connected space @xmath0 , we have that @xmath12 is a path lying in @xmath0 joining @xmath23 that intersects neither @xmath62 nor @xmath76 .
clearly @xmath12 is contained in @xmath60 and lies in the @xmath66-neighborhood of @xmath62 .
[ simplycon ] let @xmath14 be an arc that separates @xmath0 and let @xmath77 be a connected component of @xmath78 .
then @xmath79 is simply connected and @xmath80 is connected .
let @xmath81 .
we will show that this map can be extended to a map @xmath82 .
@xmath0 is simply connected , so there is a map @xmath83 such that @xmath84 .
furthermore , @xmath85 is an open set , therefore @xmath86 and since @xmath87 is a continuous extension of @xmath53 it follows that @xmath88 ( where by @xmath89 we denote the boundary of @xmath90 in @xmath0 )
. let @xmath91 be the restriction of @xmath87 in @xmath92 .
then , applying tietze s extension theorem , we obtain an extension for @xmath93 : @xmath94 finally we define @xmath95 as follows : @xmath96 this shows that @xmath79 is simply connected .
suppose now that @xmath97 is not connected and let @xmath12 be a path that joins two different components of @xmath98 , such that if @xmath99 are the endpoints of @xmath12 , then @xmath100 .
let @xmath101 and @xmath102 .
( figure [ fig : conn ] ) .
[ fig : conn ] we set @xmath103}_{{\alpha}}\cap s$ ] and @xmath104 .
it is clear that @xmath39 are disjoint closed subsets of @xmath0 and that neither @xmath105 nor @xmath106 separates @xmath21 from @xmath22 .
then alexander s lemma ( for the simply connected space @xmath0 ) implies that there is a path joining @xmath23 in @xmath107 , a contradiction .
[ a.0 ] let @xmath14 be an arc that separates @xmath108 and @xmath109 with @xmath110 .
then for every connected component @xmath77 of @xmath19 such that @xmath111 there are points @xmath112 with @xmath113 and a path @xmath114 that joins @xmath115 and is contained in the @xmath66-neighborhood of @xmath28}_{{\alpha}}$ ] .
let @xmath116 and @xmath117 be balls of center @xmath21 and radius @xmath118 and @xmath66 respectively .
we consider the connected components of @xmath119 and we restrict to those that are not contained in @xmath117 ( here we denote by @xmath120 the open ball ) .
it is clear that there are finitely many such components , so we denote them by @xmath121 .
let @xmath122 for every @xmath123 .
similarly , let @xmath124 be the connected components of @xmath125 that are not contained in @xmath126 and let @xmath127 for every @xmath128 .
let @xmath129 . from lemma [ simplycon ]
, we have that @xmath130 is simply connected , therefore lemma [ a.0 ] , for @xmath131 , implies that there is a path @xmath132 that joins a point of @xmath117 with a point of @xmath126 and lies in the @xmath133-neighborhood of @xmath14 ( figure [ fig : path ] ) .
we will show that there is a subpath of @xmath134 that has the required properties .
[ fig : path ] we assume that none of the @xmath135 contain @xmath28}_{{\alpha}}$ ] , since otherwise we are done .
thus , without loss of generality , let @xmath136 be the connected component of @xmath137 that is contained in @xmath28}_{{\alpha}}$ ] .
we denote by @xmath138}_{{\alpha}})$ ] the open @xmath133-neighborhood of @xmath28}_{{\alpha}}$ ] .
suppose that there is a connected component @xmath139_q$ ] of @xmath140}_{{\alpha}})$ ] with @xmath141 and @xmath142 .
then there is an @xmath143 such that @xmath144 . indeed ,
if not then for every @xmath143 there is an @xmath145 such that @xmath146 .
thus @xmath147 . but @xmath148 , a contradiction .
so there is an @xmath143 such that for every @xmath149 we have @xmath150 , therefore @xmath151 , which is not possible .
this contradiction proves the lemma .
we will prove the theorem by contradiction .
* remark : * since @xmath0 is locally connected and compact , it follows that every open connected subset of @xmath0 is path connected ( see theorem 3.15 , p.116 of @xcite ) .
in particular the closure of every component of @xmath78 is path connected .
let @xmath152 be arcs that separate @xmath0 .
we say that @xmath153 crosses @xmath154 at @xmath155 if for any neighborhood of @xmath21 in @xmath154 , @xmath156 , there are @xmath157 separated by @xmath153 . more generally , if @xmath158 $ ] is a connected component of @xmath159 , which is contained in @xmath160 , we say that @xmath153 crosses @xmath154 at @xmath158 $ ] if for any neighborhood of @xmath158 $ ] in @xmath154 , @xmath161 , there are @xmath162 separated by @xmath153 . in this case , the endpoints @xmath163 are also called cross points of @xmath152 . if @xmath164 , @xmath165 are intervals of @xmath152 containing @xmath21 in their interior , we say that @xmath166 cross at @xmath21 .
similarly we define what it means for two intervals to cross at a common subarc .
we call @xmath21 ( respectively @xmath158 $ ] ) a cross - point ( respectively cross - interval ) of @xmath152 .
we say that @xmath166 cross if they cross at some point @xmath21 or at some interval @xmath158 $ ] .
suppose that this is not the case , so let @xmath14 be an arc that separates @xmath0 in more than two components .
since @xmath0 has no cut points there are two connected components of @xmath78 , say @xmath167 , such that @xmath168 is a subarc ( which is not a point ) of @xmath14 .
clearly @xmath169 separates @xmath0 . to simplify notation we denote by @xmath167 the components of @xmath170 that satisfy @xmath171 .
let @xmath172 be another component of @xmath170 .
by lemma [ simplycon ] we have that @xmath173 is connected , so @xmath174 is a subarc of @xmath169 , which separates @xmath0 .
suppose that there is an arc @xmath176 that separates @xmath0 and crosses @xmath175 at @xmath177 .
then there are @xmath178 that are separated by @xmath176 .
let @xmath179 .
since @xmath180 is path connected , it follows that @xmath181 .
we denote by @xmath182 the connected component of @xmath183 that contains @xmath21 . as in lemma
[ simplycon ] , we may show that @xmath184 is simply connected .
we show then that alexander s lemma for @xmath184 implies that @xmath185 has a connected component that separates @xmath23 in @xmath184 .
this can be achieved as follows : let @xmath186 be the connected component of @xmath187 that contains @xmath22 .
it is clear that @xmath188 .
then no proper closed subset of @xmath189 separates @xmath21 from @xmath22 . indeed , suppose that there is a closed @xmath190 that separates @xmath21 from @xmath22 and let @xmath191 .
let @xmath67 be an open neighborhood of @xmath192 such that @xmath193 .
it is obvious that @xmath67 intersects every component of @xmath194 , therefore , there are paths @xmath195 that join @xmath23 with points @xmath196 respectively
. however @xmath67 is path connected and since @xmath193 , it follows that there is also a path @xmath197 that joins @xmath198 with @xmath199 .
thus @xmath200 is a path joining @xmath23 without meeting @xmath71 , a contradiction .
therefore , @xmath201 is connected and separates @xmath21 from @xmath22 .
we note now that @xmath202 does not cross @xmath175 . indeed , suppose that there is an @xmath203 in which @xmath204 crosses @xmath175 .
let @xmath205 be sufficiently small neighborhood of @xmath206 such that @xmath207 .
we denote by @xmath208 the connected component of @xmath209 that contains the point @xmath206
. then we can pick points @xmath210 with @xmath211 in @xmath175 that are separated by @xmath176 .
let @xmath212 be connected neighborhoods of @xmath198 and @xmath199 respectively , such that @xmath213 .
applying now lemma [ a.0 ] for the component @xmath172 and for @xmath214 , we have that every point of @xmath215 can be joined with every point of @xmath216 by a path in @xmath172 which lies in the @xmath66-neighborhood of @xmath217}_{\gamma}$ ] .
let @xmath218 and let @xmath134 be a path that joins @xmath177 and @xmath219 as above .
we note now that @xmath220 and @xmath221 are path connected , so there are paths @xmath222 and @xmath223 joining the endpoints of @xmath134 with the points @xmath198 and @xmath199 respectively .
clearly then the path @xmath224 joins @xmath198 with @xmath199 without meeting @xmath176 .
this is however impossible , since @xmath198 and @xmath199 are separated by @xmath176 .
therefore , @xmath202 does not cross @xmath175 .
thus @xmath225 is contained in a single component of @xmath226 , a contradiction .
we return now to the proof of lemma [ a.1 ] : let @xmath227 be the group of homeomorphisms of @xmath0 . for every @xmath228 we have that @xmath229 separates @xmath0 and from the previous lemma
it follows that @xmath229 does not cross @xmath175 .
let @xmath230 .
clearly @xmath98 is uncountable .
let @xmath231 be a countable dense set of @xmath0 .
we define a map @xmath232 as follows : let @xmath233 and @xmath234 be three connected components of @xmath235 . for every @xmath236 we pick an @xmath237 and we associate @xmath233 the triple @xmath238 .
we remark that @xmath239 is @xmath15 map , which is a contradiction .
this completes the proof of lemma [ a.1 ] .
let @xmath175 be an arc that separates @xmath0 in exactly two components @xmath240 with @xmath241 .
we denote by @xmath227 the group of homeomorphisms of @xmath0 .
let @xmath242 .
it is clear that @xmath98 is uncountable and that every arc @xmath243 also separates @xmath0 in exactly two components @xmath244 such that @xmath245 . for an arc @xmath243
we will denote these two components by @xmath246 and @xmath247 ( figure [ fig : separc ] ) .
suppose that there are @xmath248 such that @xmath153 crosses @xmath154 at @xmath21 but @xmath154 does not cross @xmath153 at @xmath21 .
then there is an interval @xmath249 containing @xmath21 at its interior that lies in the closure of one of the components of @xmath250 , say @xmath251 .
clearly then we have that @xmath252 .
let @xmath205 be sufficiently small neighborhood of @xmath21 such that @xmath253 .
we denote by @xmath208 the connected component of @xmath254 that contains the point @xmath21 .
we pick two points @xmath255 with @xmath256 in @xmath154 which are separated by @xmath153 and let @xmath257 be connected neighborhoods of @xmath206 and @xmath258 respectively such that @xmath259 . as in proof of lemma [ subl ] , for @xmath260 , we can find a path @xmath12 that joins @xmath206 with @xmath258 without meeting @xmath153 , a contradiction .
we argue similarly if @xmath153 crosses @xmath154 at an interval @xmath158 $ ] .
let @xmath276 and @xmath277 such that @xmath278 . by the definition of @xmath98 it is not possible that @xmath279 , since @xmath14 separates @xmath0 in exactly two connected components .
assume now that @xmath2 does not cross @xmath280 .
we denote by @xmath271 the connected component of @xmath281 that contains @xmath206 and let @xmath282 .
we distinguish two cases : suppose that @xmath274 .
let @xmath283 such that @xmath284 lies in the closure of one of the components of @xmath78 , say @xmath246 .
let @xmath285 and @xmath109 such that @xmath286 . by theorem [ ef ]
there is a @xmath287 such that @xmath288 .
let @xmath289 ( figure [ befef1 ] ) .
suppose now that @xmath296 .
we consider the homeomorphism @xmath290 of the previous case . if @xmath297 , then clearly lemma [ cross ] is proved .
so let @xmath298 and @xmath299 .
as before , by theorem [ ef ] , there is a @xmath300 such that @xmath301 .
let @xmath302 ( figure [ befef2 ] ) .
then there is an @xmath303 with @xmath304 and @xmath305 for every @xmath267 .
it is obvious now that @xmath14 crosses @xmath306 .
let @xmath268\in s$ ] .
by lemma [ cross ] there is an arc @xmath269 $ ] that crosses @xmath14 at @xmath307 . without loss of generality ,
suppose that @xmath21 is the endpoint of a cross interval @xmath202 of @xmath308 .
let @xmath309 and @xmath277 such that @xmath310 .
we denote by @xmath271 the connected component of @xmath311 that contains @xmath312 and similarly by @xmath313 the component of @xmath314 that contains @xmath312 .
clearly if @xmath315 , then lemma [ aaaa ] is proved .
otherwise , we note that if @xmath271 contains one of the endpoints of @xmath14 , then @xmath316 , since @xmath202 is a cross interval of @xmath308 .
so in this case , the required arcs are @xmath280 and @xmath169 .
let @xmath268,\beta=[b_1,b_2]\in s$ ] be paths as in lemma [ aaaa ] , that is @xmath270 and if @xmath271 is the connected component of @xmath272 that contains @xmath273 , then @xmath315 .
let @xmath317 such that @xmath318 and let @xmath50 be paths joining @xmath319 in @xmath320 and @xmath321 respectively ( the points @xmath322 exist since @xmath315 ) .
we pick @xmath323 such that @xmath324 has exactly two connected components neither of which intersects @xmath169 ( this can be achieved using lemma [ a.0 ] for @xmath325 , where @xmath326 is the connected component of @xmath327 that contains @xmath322 and @xmath328 is its boundary in @xmath14 ) .
let @xmath109 with @xmath329 .
as in proof of lemma [ cross ] , using theorem [ ef ] , we can find a homeomorphism @xmath290 such that @xmath330 crosses @xmath14 at @xmath331 , with @xmath332
. then we remark that @xmath333 and that the subarc of @xmath334 with endpoints @xmath21 and @xmath335 does not intersect @xmath336 .
we pick now points @xmath337 and @xmath338 which are separated by @xmath334 so that they satisfy the following : if @xmath22 is a cross point of @xmath14 and @xmath334 , lying in @xmath339}_{{\alpha}}$ ] , then the subarc @xmath340}_{h\beta}$ ] does not intersect the paths @xmath50 .
such points exist by definition of @xmath21 ( figure [ aftef ] ) .
we consider now the closed paths @xmath341_{\alpha } $ ] and @xmath342_{\alpha } $ ] .
let @xmath343 be discs and let @xmath344 , @xmath345 be maps so that @xmath346_{\alpha } $ ] , @xmath347_{\alpha } $ ] ( such maps exist , since @xmath348 and @xmath349 are simply connected by lemma [ simplycon ] ) .
we ` glue ' @xmath343 along @xmath350_{\alpha } $ ] and we obtain a disc @xmath351 and a map @xmath352 with @xmath353 .
more precisely , we consider the disc @xmath354 , where @xmath355 is defined as follows : @xmath356 if and only if @xmath357 and @xmath358 . finally , we define @xmath352 as : by abuse of notation we identify points of @xmath350_{\alpha } $ ] in @xmath351 with their image under @xmath53 .
we note that the interior , say @xmath67 , of @xmath351 is homeomorphic to @xmath360 and since @xmath361 are separated by @xmath334 in @xmath0 , it follows by alexander s lemma that @xmath361 are separated in @xmath67 by a connected component of @xmath362 .
we call this component @xmath71 ( figure [ diskos ] ) .
clearly @xmath363 is a subarc of @xmath334 that contains cross points or cross intervals of @xmath334 with @xmath364_{\alpha } $ ] .
let @xmath312 be such a cross point
. then we can write @xmath363 as @xmath365 , where @xmath366 , are ( connected ) subarcs of @xmath334 , such that @xmath367 . furthermore , at least one of @xmath166 does not intersect @xmath336 ( this is by our choice of @xmath368 and @xmath312 ) .
it follows that at least one of @xmath369 is compact .
we set @xmath370 .
we will define two sets @xmath39 such that the following are satisfied : @xmath39 are closed subsets of @xmath67 that contain @xmath371 and @xmath372 respectively , @xmath41 is connected contained in @xmath373 and @xmath374 .
let now @xmath378 be a connected component of @xmath375 different from @xmath77 and suppose that @xmath379 .
we consider the closure of @xmath71 , @xmath380 , in @xmath381 .
@xmath380 is connected thus the closure of the component of @xmath382 containing @xmath378 intersects @xmath383 .
indeed , we consider the set @xmath384_{\alpha})$ ] as an open subset of the continuum @xmath385 .
let @xmath386 be the component of @xmath387_{\alpha})$ ] that contains @xmath378 .
we recall that if @xmath67 is an open subset of a continuum and @xmath77 is a component of @xmath67 then the frontier of @xmath67 contains a limit point of @xmath77 ( theorem 2.16 , p.47 of @xcite ) .
it follows that the closure of @xmath386 intersects @xmath350_{\alpha}$ ] .
we have a similar conclusion if @xmath390 . therefore if a connected component of @xmath375 different from @xmath77 lying in @xmath376 intersects the closure of both @xmath391 we have that an open interval of @xmath136 around @xmath312 lies in @xmath348 .
this is impossible since @xmath312 is a cross point .
we argue similarly for connected components of @xmath375 contained in @xmath377 .
we conclude that the union of the components of @xmath375 which lie in @xmath376 , intersect exactly one of @xmath391 .
clearly the same is true for the union of the components of @xmath375 contained in @xmath377 . in particular
exactly one of the following two holds : \1 . if @xmath378 is a connected component of @xmath375 different from @xmath77 lying in @xmath376 then the component of @xmath392 containing @xmath378 intersects @xmath371 , while if @xmath378 lies in @xmath377 the component of @xmath393 containing @xmath378 intersects @xmath372 .
\2 . if @xmath378 is a connected component of @xmath375 different from @xmath77 lying in @xmath376 then the component of @xmath392 containing @xmath378 intersects @xmath372 , while if @xmath378 lies in @xmath377 the component of @xmath393 containing @xmath378 intersects @xmath371 .
assume that we are in the first case .
then we define @xmath105 to be the union of the components of @xmath375 intersecting @xmath376 together with @xmath371 .
we define @xmath106 to be the union of the components of @xmath375 intersecting @xmath377 together with @xmath372 .
it is clear that @xmath39 are closed and that @xmath394 , @xmath374 .
since @xmath71 is connected , @xmath39 are connected too .
we define @xmath39 similarly in the second case .
we note now that at least one of @xmath39 is compact subset of @xmath67 , thus bounded in @xmath67 .
since @xmath71 separates @xmath395 and @xmath39 are closed subsets of @xmath351 , applying alexander s lemma for the plane we have that at least one of @xmath39 separates @xmath395 in @xmath67 .
it follows that either @xmath396 or @xmath397 separates @xmath395 .
we remark that the same argument holds in the case @xmath312 is replaced by a cross interval @xmath208 : we have then that @xmath398 with @xmath399 and as before either @xmath396 or @xmath397 separate @xmath395 in @xmath67 .
now we can continue subdividing intervals along cross points ( cross intervals ) that lie in @xmath364_{\alpha } $ ] as follows : let s say that @xmath396 separates @xmath395 .
we have that there is a connected component of @xmath396 , say @xmath400 , that separates them .
we note that @xmath401 is a subinterval of @xmath136 and if there is a cross point ( or cross interval ) of @xmath402_{{\alpha}},h\beta$ ] contained in its interior , we repeat the previous procedure replacing @xmath71 by @xmath400 . if not we have a contradiction .
therefore , either @xmath395 are separated in @xmath67 by the inverse image of an interval @xmath363 of @xmath334 which does not contain in its interior any cross point of @xmath403 lying in @xmath364_{\alpha } $ ] , or by iterating this procedure we conclude that the inverse images under @xmath53 of intervals of @xmath404 of arbitrarily small diameter separate @xmath219 from @xmath177 in @xmath67 .
it is clear that both are impossible , so the theorem is proven . | we show that locally connected , simply connected homogeneous continua are not separated by arcs .
we ask several questions about homogeneous continua which are inspired by analogous questions in geometric group theory . |
the data of the first run of the large hadron collider ( lhc ) at cern have led to the exciting discovery of the higgs boson @xcite and are , within the current level of precision , globally consistent with the picture of the standard model ( sm ) .
the next run of the lhc at almost the double centre - of - mass energy of the colliding protons , which will start in spring 2015 , will open various new opportunities in the search for new physics ( np ) @xcite .
these will be both in the form of direct searches for new particles at the atlas and cms experiments , and in the form of high - precision analyses of flavour physics observables at the lhcb experiment .
concerning the latter avenue , also the belle ii experiment at the kek @xmath14 super @xmath15 factory will enter the stage in the near future @xcite .
the current lhc data suggest that we have to prepare ourselves to deal with smallish np effects , and it thus becomes mandatory to have a critical look at the theoretical assumptions underlying the experimental analyses . concerning measurements of cp violation , the @xmath0 and @xmath1 decays play outstanding roles as they allow determinations of the @xmath16@xmath17 mixing phases @xmath18 and @xmath8 , respectively .
these quantities take the forms @xmath19 where @xmath20 is the usual angle of the unitarity triangle ( ut ) of the cabibbo
maskawa ( ckm ) matrix @xcite and @xmath21 in the sm @xcite .
the @xmath22 and @xmath23 are two of the wolfenstein parameters @xcite of the ckm matrix .
the cp - violating phases @xmath24 , which vanish in the sm , allow for np contributions entering through @xmath25@xmath26 mixing .
the theoretical precision for the extraction of @xmath18 and @xmath8 from the cp asymmetries of the @xmath0 and @xmath1 decays is limited by doubly cabibbo - suppressed penguin contributions .
the corresponding non - perturbative hadronic parameters can not be calculated in a reliable way within qcd .
however , in the era of high - precision measurements , these effects have to be controlled with the final goal to match the experimental and theoretical precisions @xcite . as was pointed out in ref .
@xcite , @xmath4 is related to @xmath0 through the @xmath3-spin symmetry of strong interactions , and allows a determination of the penguin corrections to the measurement of @xmath18 . concerning the @xmath1 channel ,
an analysis of cp violation is more involved as the final state consists of two vector mesons and thus is a mixture of different cp eigenstates which have to be disentangled through an angular analysis of their decay products @xcite . in this case , the decays @xmath9 @xcite and @xmath10 @xcite are tools to take the penguin effects into account .
the lhcb collaboration has very recently presented the first polarisation - dependent measurements of @xmath8 from @xmath1 in ref .
we shall discuss the implications of these exciting new results in detail .
since a measurement of cp violation in @xmath4 is not yet available , we use the @xmath5 flavour symmetry and plausible assumptions for various modes of similar decay dynamics to constrain the relevant penguin parameters . following these lines ,
we assess their impact on the measurement of @xmath18 and predict the cp - violating observables of @xmath4 . in our benchmark scenario ,
we discuss also the determination of cp - conserving strong amplitudes , which will provide valuable insights into non - factorisable @xmath3-spin - breaking effects through the comparison with theoretical form - factor calculations .
concerning the @xmath10 channel , measurements of cp violation are also not yet available . however , in the case of @xmath9 , the lhcb collaboration has recently announced the first results of a pioneering study @xcite , presenting in particular a measurement of mixing - induced cp violation and constraints on the penguin effects .
this new experimental development was made possible through the implementation of the method proposed by zhang and stone in ref .
we shall have a detailed look at these exciting measurements and discuss important differences between the penguin probes @xmath9 and @xmath10 .
we extract hadronic parameters from the @xmath9 data , allowing insights into @xmath5-breaking and non - factorisable effects through a comparison with theory , and point out a new way to combine the information provided by the @xmath10 , @xmath9 system in a global analysis of the @xmath1 penguin parameters .
the outline of this paper is as follows : in section [ sec : hadr ] , we introduce the general formalism to deal with the penguin effects . in section [ sec : psip ]
, we explore the constraints of the currently available data for the penguin contributions to the @xmath27 system , while we turn to the discussion of the most recent lhcb results for @xmath1 and the penguin probes @xmath9 , @xmath10 in section [ sec : psiv ] . in section [ sec :
road ] , we outline a roadmap for dealing with the hadronic penguin uncertainties in the determination of @xmath18 and @xmath8 .
finally , we summarise our conclusions in section [ sec : concl ] .
for the neutral @xmath28 decays ( @xmath29 ) discussed in this paper , the transition amplitudes can be written in the following form @xcite : @xmath30\ : , \label{ampl}\\ a(\bar b^0_q\to f)\equiv \bar a_f = \ : & \eta_f{\cal n}_f\left[1- b_f e^{\rho_f}e^{-i\gamma}\right]\:.\label{ampl - cp}\end{aligned}\ ] ] here @xmath31 is the cp eigenvalue of the final state @xmath32 , @xmath33 is a cp - conserving normalisation factor representing the dominant tree topology , @xmath34 parametrises the relative contribution from the penguin topologies , @xmath35 is the cp - conserving strong phase difference between the tree and penguin contributions , whereas their relative weak phase is given by the ut angle @xmath36 . the parameters @xmath33 and @xmath34 depend both on ckm factors and on hadronic matrix elements of four - quark operators entering the corresponding low - energy effective hamiltonian .
in order to extract information on @xmath37 , cp - violating asymmetries are measured @xcite : @xmath38 where the dependence on the decay time @xmath39 enters through @xmath25@xmath26 oscillations , and and denote the mass and decay width differences of the two @xmath28 mass eigenstates , respectively .
using eqs . and , the direct and mixing - induced cp asymmetries @xmath40 and @xmath41 take the following forms @xcite : and @xmath42 decays is needed to determine an observable , as is the case for cp asymmetries or untagged branching ratios , we use the notation @xmath43 and @xmath44 . ]
@xmath45 @xmath46\:,\ ] ] while the observable @xmath47 is given by @xmath48\:.\ ] ] for the discussion of the penguin effects , the following expression will be particularly useful ( generalising the formulae given in ref .
@xcite ) : @xmath49 where @xmath50 yielding @xmath51\:.\ ] ] it should be emphasised that @xmath52 is a phase shift which depends on the non - perturbative parameters @xmath34 and @xmath35 and can not be calculated reliably within qcd . in the case of @xmath53 ,
the following simple situation arises : @xmath54 allowing us to determine @xmath37 directly from the mixing - induced cp asymmetry . since in the decays
@xmath0 and @xmath1 the parameters corresponding to @xmath34 are doubly cabibbo - suppressed , eq .
is approximately valid .
however , in the era of high - precision studies of cp violation , we nonetheless have to control these effects .
as the corresponding penguin parameters are cabibbo - allowed in the @xmath4 and @xmath55 , @xmath10 decays , these modes allow us to probe the penguin effects . making use of the @xmath5 flavour symmetry
, we may subsequently convert the penguin parameters into their @xmath0 and @xmath1 counterparts , where in the latter case also plausible dynamical assumptions beyond the @xmath5 are required .
channels , where @xmath56 , @xmath57 and @xmath58 represents any of the @xmath59 , @xmath60 , @xmath61 , @xmath62 , @xmath63 , @xmath64 or @xmath65 mesons.,title="fig:",scaledwidth=45.0% ] channels , where @xmath56 , @xmath57 and @xmath58 represents any of the @xmath59 , @xmath60 , @xmath61 , @xmath62 , @xmath63 , @xmath64 or @xmath65 mesons.,title="fig:",scaledwidth=45.0% ] in the sm , the decay @xmath0 into a cp eigenstate with eigenvalue @xmath66 originates from a colour - suppressed tree contribution and penguin topologies with @xmath67-quark exchanges ( @xmath68 ) , which are described by cp - conserving amplitudes @xmath69 and @xmath70 , respectively , and illustrated in fig .
[ fig : feynman_tree ] .
the primes are introduced to remind us that we are dealing with a @xmath71 quark - level process . using the unitarity of the ckm matrix
, the @xmath0 decay amplitude can be expressed in the following form @xcite : @xmath72\:,\ ] ] where @xmath73\ ] ] and @xmath74\ ] ] are cp - conserving hadronic parameters .
the wolfenstein parameter @xmath22 takes the value @xmath75 @xcite , and @xmath76 are combinations of ckm matrix elements .
the parameter @xmath77 measures the size of the penguin topologies with respect to the tree contribution , and is associated with the cp - conserving strong phase @xmath78 . a key feature of the decay amplitude in eq .
is the suppression of the @xmath79 term by the tiny factor @xmath80 .
consequently , @xmath18 can be extracted with the help of eq . up to corrections of @xmath81 . as was pointed out in ref .
@xcite , the decay is related to through the @xmath3-spin symmetry of strong interactions .
it originates from transitions and therefore has a ckm structure which is different from in analogy to eq . , we write @xmath82\:,\ ] ] where the hadronic parameters are defined as their @xmath0 counterparts .
in contrast to eq . , there is no @xmath83 factor present in front of the second term , thereby magnifying " the penguin effects . on the other hand , the @xmath22 in front of the overall amplitude suppresses the branching ratio with respect to @xmath0 .
the @xmath3-spin symmetry of strong interactions implies @xmath84 in the factorisation approximation the hadronic form factors and decay constants cancel in the above amplitude ratios @xcite , i.e. @xmath3-spin - breaking corrections enter @xmath85 through non - factorisable effects only . on the other hand ,
the relation @xmath86 is already in factorisation affected by @xmath5-breaking effects , entering through hadronic form factors as we will discuss in more detail below .
it is well known that the factorisation approximation does not reproduce the branching ratios of @xmath87 decays well , thereby requiring large non - factorisable effects .
furthermore , the qcd penguin matrix elements of the current current tree operators , which are usually assumed to yield the potential enhancement for the penguin contributions , vanish in naive factorisation for @xmath87 decays . consequently , large non - factorisable contributions may also affect the penguin parameters @xmath88 and @xmath85 , thereby enhancing them from the smallish values in factorisation , and eq .
( [ a - rel ] ) may receive sizeable corrections despite the cancellation of form factors and decay constants in factorisation . making the replacements @xmath89 we may apply the formalism introduced in section [ sec : hadr ] , yielding the following phase shifts :
the expansions in terms of the penguin parameters show an interesting feature : the phase shifts are maximal for a strong phase difference around @xmath92 or @xmath93 .
conversely , the penguin shifts will be tiny for values around @xmath94 or @xmath95 , even for sizeable @xmath96 .
the @xmath97 and @xmath98 enter @xmath99 in the expressions corresponding to eq . .
these effective " mixing phases are convenient for the presentation of the experimental results @xcite . the @xmath4 decay channel has been observed by the cdf @xcite and lhcb @xcite collaborations , and measurements of the time - integrated untagged rate @xcite @xmath100 with @xmath101 were performed , resulting in the world average @xcite @xmath102 information on the penguin parameters is also encoded in this observable , thereby complementing the cp asymmetries . in view of the sizeable decay width difference @xmath103 of the @xmath44-meson system , which is described by the parameter @xcite
@xmath104 the experimental " branching ratio has to be distinguished from the theoretical " branching ratio defined by the untagged decay rate at time @xmath105 @xcite .
the conversion of one branching ratio concept into the other can be done with the help of the following expression @xcite : @xmath106 \mathcal{b}\left(b_s\to j/\psi k_{\rm s}^0\right)\:.\ ] ] the observable @xmath107 depends also on the penguin parameters , as can be seen in eq . ( [ eq : adg ] )
. the effective lifetime @xmath108\end{aligned}\ ] ] allows us to determine @xmath107 , thereby fixing the conversion factor in eq .
the lhcb collaboration has performed the first measurement of this quantity @xcite : @xmath109 corresponding to @xmath110 in view of the large uncertainty of this measurement , we shall rely directly on eq . with eq .
in the numerical analysis performed in section [ ssec : constr ] . in order to utilise the branching ratio information , we construct the observable @xmath111 where @xmath112 is the @xmath28 lifetime and @xmath113 denotes the phase - space function for these decays @xcite . in terms of the penguin parameters ,
we obtain @xmath114 where we also give the relation to the direct cp asymmetries of the decays at hand . keeping @xmath115 and @xmath116 as free parameters , the following lower
bound arises @xcite : @xmath117 which corresponds to @xmath118 for @xmath119 .
the determination of @xmath120 from the experimentally measured branching ratios is affected by @xmath3-spin - breaking corrections which enter through the ratio @xmath121 .
consequently , @xmath120 is not a particularly clean observable . on the other hand
, the analysis of the direct and mixing - induced cp asymmetries does not require knowledge of @xmath121 .
if we complement the ratio @xmath120 with the direct and mixing - induced cp asymmetries of the @xmath4 channel , we have sufficient information to determine @xmath36 and the penguin parameters @xmath115 and @xmath116 by means of the @xmath3-spin relation in eq . @xcite . in this strategy
, @xmath8 serves as an input , where we may either use its sm value in eq . or the value extracted from experimental data , as discussed in section [ sec : psiv ] .
we advocate the latter option since it takes possible cp - violating np contributions to @xmath6@xmath7 mixing into account .
although @xmath36 can be extracted with this method at the lhcb upgrade , the corresponding precision is not expected to be competitive with other strategies @xcite .
it is therefore advantageous to employ @xmath36 as an input . using data from pure tree decays of the kind @xmath122
, the following averages were obtained by the ckmfitter and utfit collaborations : @xmath123 for the numerical analysis in this paper , we shall use the ckmfitter result in view of the larger uncertainty . by the time of the lhcb upgrade and belle ii era , much more precise measurements of @xmath36 from pure tree decays will be available ( see section [ ssec : scene ] ) .
once the direct and mixing - induced cp asymmetries of the @xmath4 channel have been measured , eqs . and can be used with eq . to determine @xmath115 and @xmath116 in a theoretically clean way .
employing the @xmath3-spin relation allows us to convert these parameters into the phase shift @xmath98 , and thus to include the penguin effects in the determination of @xmath18 . as a measurement of cp violation in @xmath4
is not yet available , the @xmath3-spin strategy sketched above can not yet be implemented in practice .
however , in order to already obtain information on the size of the penguin parameters @xmath115 and @xmath116 and their impact on high - precision studies of cp violation , we may use experimental data for decays which have dynamics similar to @xmath4 . if we replace the strange spectator quark with a down quark , as proposed in ref .
@xcite , we obtain the @xmath124 decay @xcite , which is the vector pseudo - scalar counterpart of the vector
vector mode @xmath9 .
the @xmath124 mode has contributions from penguin annihilation and exchange topologies , illustrated in fig .
[ fig : feynman_additional ] , which have no counterpart in and are expected to be small .
they can be probed through the @xmath125 decay ( and @xmath126 for @xmath9 ) @xcite .
first measurements of cp violation in @xmath124 were reported by the babar and belle collaborations : @xmath127 @xmath128 the results for the mixing - induced cp asymmetry are not in good agreement with each other , with the babar result lying outside the physical region .
the heavy flavour averaging group ( hfag ) has refrained from inflating the uncertainties in their average , giving @xmath129 @xcite .
the belle ii experiment will hopefully clarify this unsatisfactory situation .
the charged counterpart @xmath130 of @xmath124 also has dynamics similar to @xmath4 but as it is the decay of a charged @xmath15 meson does not exhibit mixing - induced cp violation .
it receives additional contributions from an annihilation topology , illustrated in fig .
[ fig : feynman_additional ] , which arises with the same ckm factor @xmath131 as the penguin topologies with internal up - quark exchanges , contributing similarly to the penguin parameter @xmath132 ( defined in analogy to eq .
( [ eq : penguin_def ] ) ) .
if this parameter is determined from the charged @xmath130 , @xmath133 decays and compared with the other penguin parameters , footprints of the annihilation topology could be detected .
in view of the present uncertainties , we neglect the annihilation topology , like the contributions from the exchange and penguin annihilation topologies in @xmath124 . in appendix [ app : annihil ] , we give a more detailed discussion of the annihilation contribution and its importance based on constraints from current data , which do not indicate any enhancement .
we shall also add data for the @xmath133 ( neglecting again the corresponding annihilation contribution ) and @xmath134 modes to the global analysis , although the penguin contributions are doubly cabibbo - suppressed in these decays . using the @xmath5 flavour symmetry and assuming both vanishing non - factorisable corrections and vanishing exchange and annihilation topologies
, the decays listed above are characterised by a universal set of penguin parameters @xmath135 , which can be extracted from the input data through a global @xmath136 fit .
the resulting picture extends and updates the previous analyses of refs .
@xcite .
observables which can be constructed from the available branching ratio information for modes . the label
dir " indicates that @xmath120 is determined from direct branching fraction measurements , whereas the label rat " is used for @xmath120 observables calculated from a ratio of branching fractions .
the inner uncertainty bars indicate the statistical uncertainty whereas the outer ones give the total uncertainty , including the common uncertainty due to the form factors .
the red band indicates the average @xmath120 observable of the modes .
the hatched blue region is excluded by eq .
.,scaledwidth=75.0% ] a first consistency check is provided by the ratios @xmath137 involving decays which originate from the same quark - level processes but differ through their spectator quarks @xcite .
neglecting exchange and annihilation topologies and assuming perfect flavour symmetry of strong interactions , these ratios equal one . within the uncertainties ,
this picture is supported by the data , as shown in fig .
[ fig : su3_comparison ] . in this compilation ,
the @xmath15-factory branching ratio measurements are corrected for the measured pair production asymmetry between @xmath138 and @xmath139 @xcite at the @xmath140 resonance . note that the branching ratios for decays into final states with @xmath62 or @xmath59 mesons have to be multiplied by a factor of two in eq . to take the @xmath62 and @xmath59 wave functions into account .
let us now probe the penguin parameters through the various branching ratios . to this end , we use ratios defined in analogy to @xmath120 in eq . .
the extraction of these quantities from the data requires knowledge of the amplitude ratio @xmath121 , which is given in factorisation as follows @xcite : @xmath141 the corresponding form factors have been calculated in the literature using a variety of techniques . for our analysis , we take the results from light cone qcd sum rules ( lcsr ) , which are typically calculated at @xmath142 .
the relevant form factors are @xcite , @xcite and @xcite , where the first two describe transitions for both the @xmath143 and the @xmath144 mesons .
the @xmath145 dependence of these form factors is parametrised by means of the bgl method described in ref .
@xcite .
using these form factors and neglecting non - factorisable @xmath5-breaking effects , we obtain the various @xmath120 observables compiled in fig .
[ fig : hobs ] . with exception of the last entry , all @xmath120 observables share the same ratio @xmath146 .
consequently , their central values and uncertainties are highly correlated . however , even restricting the comparison to the statistical uncertainties shows an excellent compatibility between the various @xmath120 results .
the corresponding ratios are related to each other through the isospin symmetry ( neglecting additional topologies ) , and we obtain a consistent experimental picture .
the agreement with the last entry , which involves the decay @xmath147 instead of the @xmath148 modes , suggests that non - factorisable @xmath5-breaking effects and the impact of additional decay topologies are small , thereby complementing the picture of fig .
[ fig : su3_comparison ] .
the uncertainties are still too large to draw definite conclusions . for the global @xmath136 fit to extract the penguin parameters @xmath115 and @xmath116 we use the input quantities summarised in table [ tab : inputs_cp_b2pv ] , and add the ckmfitter result for @xmath36 in eq .
as an asymmetric gaussian constraint .
as far as the @xmath120 observables are concerned , we employ the average of the @xmath149 combinations , which involve the same set of form factors ( see fig .
[ fig : hobs ] ) , and the @xmath120 observable of the @xmath150 system .
the branching ratios entering the @xmath120 observables are complemented by the corresponding direct cp asymmetries . in order to add the mixing - induced cp asymmetry of the @xmath124 channel to the fit ,
the @xmath151@xmath152 mixing phase @xmath18 is needed as an input .
however , the measured cp - violating asymmetries of the @xmath0 decay allow us to determine only the effective mixing phase , which is an average of @xmath0 and @xmath153 data @xcite . ]
@xmath154 from eq . .
but if we express the phase shift @xmath98 in terms of the penguin parameters we may add this observable to our analysis .
.input quantities for the global @xmath136 fit to the penguin parameters @xmath115 , @xmath116 and @xmath18 .
[ cols="<,^,>",options="header " , ] the global fit yields @xmath155 for four degrees of freedom @xmath156 , indicating good agreement between the different input quantities .
it results in the solutions @xmath157 and @xmath158 while @xmath36 is constrained to the input in eq . .
in fig . [
fig : phid - a ] , we show the correlation between @xmath18 and @xmath115 . the value of @xmath18 in eq
. will serve as an input in section [ sec : psiv ] .
following ref .
@xcite , we illustrate the various constraints entering the fit through contour bands of the individual observables in fig .
[ fig : b2pv ] . for the @xmath159 range
, we have used the value of @xmath18 in eq . .
in comparison with the analysis of ref .
@xcite , the penguin parameters are now constrained in a more stringent way .
the penguin parameters in eq .
result in the following penguin phase shift : @xmath160 with confidence level contours shown in fig .
[ fig : b2pv_deltaphi ] .
@xmath152 mixing phase @xmath18 and the penguin parameter @xmath115 arising from the @xmath136 fit to current data as described in the text.,scaledwidth=75.0% ] and @xmath116 through intersecting contours derived from cp asymmetries and branching ratios of @xmath161 decays .
we show also the confidence level contours obtained from a @xmath136 fit to the data . to improve the visualisation ,
the allowed range for @xmath115 has been extended to 1.,scaledwidth=75.0% ] as a function of the strong phase @xmath116 arising from the @xmath136 fit to the data . superimposed
are the contour levels for the penguin parameter @xmath115.,scaledwidth=75.0% ] fit to the present data as discussed in section [ ssec : constr].,scaledwidth=75.0% ] let us conclude the analysis of the penguin effects in @xmath0 by discussing a future benchmark scenario pointing to the lhcb upgrade era . using the results in eq . and assuming the sm value for @xmath8 in eq .
, we obtain the following predictions : @xmath162 the associated confidence level contours for @xmath163 and @xmath164 are shown in fig .
[ fig : prediction ] .
moreover , the penguin parameters in eq .
yield @xmath165 in agreement with the experimental result in eq . .
in order to illustrate the potential of the @xmath4 decay to extract the penguin parameters at the lhcb upgrade , let us assume that @xmath36 has been determined in a clean way from pure tree decays @xmath122 as @xmath166 and that the @xmath6@xmath7 mixing phase has been extracted from the @xmath1 angular analysis and the application of the strategies discussed in sections [ sec : psiv ] and [ sec : road ] to control the penguin effects as @xmath167 the experimental uncertainty projections for the lhcb upgrade are discussed in ref .
we consider our assessment of the theoretical uncertainty of @xmath8 in eq . as conservative .
let us assume that the cp - violating asymmetries of the @xmath4 channel have been measured as follows : @xmath168 i.e. with the central values of eqs . and . in order to estimate the uncertainties , current lhcb measurements of cp violation in @xmath169 modes @xcite have been extrapolated to the lhcb upgrade era , correcting for the @xmath170 event yield @xcite . a @xmath136 fit to these observables would then yield @xmath171 the corresponding confidence level contours
are shown in fig .
[ fig : bench_bs ] . and
@xmath116 from the cp asymmetries of the decay.,scaledwidth=75.0% ] in contrast to the fit in fig .
[ fig : b2pv ] , this future " determination of @xmath115 and @xmath116 is theoretically clean . using the @xmath3-spin relation , these parameters can be converted into the penguin phase shift of @xmath0 .
it is only at this point that potential @xmath3-spin - breaking effects enter .
they can be included by introducing parameters @xmath172 and @xmath173 as follows : @xmath174 assuming @xmath175 and @xmath176 , the results for @xmath115 and @xmath116 in fig .
[ fig : bench_bs ] yield @xmath177^{\circ},\ ] ] with the corresponding contours shown in fig .
[ fig : bench_bd ] . in this benchmark scenario ,
the experimental and theoretical uncertainties are of the same size , with a total uncertainty of @xmath178 if added in quadrature . by the time such measurements will be available , we should have better experimental insights into @xmath3-spin - breaking effects . as we will see in section [ sec : b2jpsirho ] , already the currently available data for @xmath55 decays do not favour large effects in the @xmath55 , @xmath1 system . from the cp asymmetries of the decay .
the confidence level contours assume a 20% uncertainty for @xmath3-spin breaking effects , parametrised through eq .
.,scaledwidth=75.0% ] it is important to emphasise that the observable @xmath120 is _ not _ required in this analysis . assuming eq .
, it can rather be determined with the help of eq . .
as @xmath77 enters there in combination with the tiny @xmath83 factor , the @xmath3-spin - breaking corrections have a negligible effect in this case . using eq .
, we obtain @xmath179 the comparison with eq . then allows us to extract the ratio @xmath180 in order to illustrate the corresponding future experimental precision , we use the central values of the penguin parameters in eq . and combine them with information on the ratio of branching fractions .
the systematic uncertainty of all @xmath44 branching ratio measurements is limited by the ratio @xmath181 @xcite of fragmentation functions , which is required for normalisation purposes @xcite . at the lhcb upgrade , the experimental precision of the ratio of branching fractions entering eq .
will be governed by that of @xmath182 .
assuming no further improvement in the determination of this parameter , which is conservative , would result in the measurement @xmath183 the experimental uncertainty is about five times smaller than the current theoretical uncertainty of the factorisation result @xmath184 using lcsr form factors ( see the discussion of eq . ) .
consequently , the experimental determination of @xmath121 is yet another interesting topic for the lhcb upgrade
. it will provide valuable insights into possible non - factorisable @xmath3-spin - breaking effects and the hadronisation dynamics of the system .
in the case of the @xmath1 and @xmath9 modes , the final states are mixtures of cp - even and cp - odd eigenstates . for the analysis of cp violation
, these states have to be disentangled with the help of a time - dependent angular analysis of the @xmath185 and @xmath186 , @xmath187 decay products @xcite . to this end , it is convenient to introduce linear polarisation amplitudes @xmath188 , @xmath189 and @xmath190 @xcite , where the @xmath191 and @xmath192 final state configurations are cp - even while @xmath193 describes a cp - odd state .
a detailed discussion of the general structure of the various observables provided by the angular distribution in the presence of the penguin contributions was given in ref .
the linear polarisation states are also employed for the theoretical description of the @xmath10 decay , which has a flavour - specific final state @xcite .
the decay is the @xmath6-meson counterpart of . assuming that the @xmath194 meson is a pure @xmath195 state , i.e. neglecting @xmath196@xmath194 mixing ( for a detailed discussion , see ref .
@xcite ) , this transition arises if we replace the down spectator quark of @xmath0 by a strange quark . in analogy to eq . , the sm decay amplitude takes the following form @xcite : @xmath197\:,\ ] ] where the label @xmath198 distinguishes between the different configurations of the final state vector mesons .
we have to make the replacements @xmath199 in order to apply the formalism introduced in section [ sec : hadr ]
. the hadronic phase shift @xmath200 can be obtained from eq . .
the penguin parameters @xmath201 are in general expected to differ for different final - state configurations @xmath32 .
however , applying simplified arguments along the lines of factorisation , the following picture emerges @xcite : @xmath202 the reason giving rise to the polarisation - independent parameters is the feature that form factors , which may depend on the final - state configuration @xmath32 , cancel in the @xmath203 ratios of penguin to tree amplitudes .
it is an interesting question to test eq . with experimental data , in particular in view of the discussion in the paragraph after eqs .
( [ a - rel ] ) and ( [ aprime - a - rel ] ) .
the parameters @xmath204 and @xmath205 in eq . may differ from their @xmath0 counterparts in eq . due to the different hadronisation dynamics and non - factorisable effects .
the lhcb collaboration has recently presented the first results for the effective mixing phases for the different final - state polarisations @xcite : @xmath206 within the uncertainties , no dependence on the final - state configuration is revealed . moreover , eq .
is in excellent agreement with the sm value in eq . .
using eq . as a guideline for the size of possible hadronic phase shifts in @xmath1 ,
the current precision is not yet high enough for resolving such effects .
however , the lhcb analysis of ref .
@xcite has a pioneering character , and it will be very interesting to monitor the polarisation - dependent measurements as the precision increases . assuming a universal value of @xmath207 , i.e. the relations in eq . , the following result is obtained from the time - dependent analysis of the @xmath208 \phi[\to k^+k^-]$ ] angular distribution @xcite : @xmath209 the lhcb collaboration has also reported first polarisation - dependent results for the following quantities : @xmath210 in this expression , cp violation in @xmath6@xmath7 oscillations , which is a tiny effect @xcite , has been neglected , like in eq . with eqs . , and .
the lhcb measurements are given by @xmath211 enters with @xmath212 .
if we use the fit result in eq . as a guideline and
assume @xmath213 , we obtain @xmath214 , which sets the scale of the required precision to resolve possible footprints of the penguin contributions in these measurements . assuming that the parameters @xmath215 do not depend on the final - state configuration of the vector mesons , the lhcb collaboration has extracted the following result from the @xmath208 \phi[\to k^+k^-]$ ] data : @xmath216 level which could well be an experimental fluctuation would be surprisingly large in view of the discussion given above . in fig .
[ fig : bspsiphi ] , we convert this result into a contour band in the @xmath205@xmath204 plane .
as expected , the central value would correspond to penguin effects too large to be consistent with the other constraints . assuming the sm value of @xmath8 in eq .
, we may also show the experimental result in eq . as a band in this figure .
this analysis illustrates the observation we made in the context with eqs . and : in order to ensure a small phase shift of @xmath8 for large penguin parameters , strong phases around @xmath217 are needed .
interestingly , the data for @xmath9 also suggest such a picture for the strong phases .
@xmath218 plane following from the effective @xmath6@xmath7 mixing phase in eq . and
@xmath219 in eq . .
here we assume the sm value of @xmath8 in eq .
.,scaledwidth=75.0% ] and @xmath220 from intersecting contours derived from the cp observables in superimposed are the confidence level contours obtained from a @xmath136 fit to the data .
the contour originating from the direct cp violation in ( see also fig .
[ fig : bspsiphi ] ) has been added for visual comparison , but is not taken into account in the fit.,title="fig:",scaledwidth=45.0% ] and @xmath220 from intersecting contours derived from the cp observables in superimposed are the confidence level contours obtained from a @xmath136 fit to the data .
the contour originating from the direct cp violation in ( see also fig .
[ fig : bspsiphi ] ) has been added for visual comparison , but is not taken into account in the fit.,title="fig:",scaledwidth=45.0% ] and @xmath220 from intersecting contours derived from the cp observables in superimposed are the confidence level contours obtained from a @xmath136 fit to the data .
the contour originating from the direct cp violation in ( see also fig .
[ fig : bspsiphi ] ) has been added for visual comparison , but is not taken into account in the fit.,title="fig:",scaledwidth=45.0% ] and @xmath220 from intersecting contours derived from the cp observables in superimposed are the confidence level contours obtained from a @xmath136 fit to the data .
the contour originating from the direct cp violation in ( see also fig .
[ fig : bspsiphi ] ) has been added for visual comparison , but is not taken into account in the fit.,title="fig:",scaledwidth=45.0% ] in analogy to @xmath4 and @xmath124 , the decay @xmath9 originates from @xmath221 quark - level transitions and has a decay amplitude of similar structure @xcite : @xmath222\:,\ ] ] where the factor of @xmath223 is due to the wave function of the @xmath63 . in analogy to @xmath1 , the @xmath9 decay also shows mixing - induced cp violation , where an analysis of the @xmath224 and @xmath225 decay products is required to disentangle the cp - even and cp - odd final states . in order to apply the formalism of section [ sec : hadr ]
, we have to make the replacements @xmath226 in particular , we then obtain expressions for the effective " mixing phases @xmath227 by applying eq . .
it should be emphasised that the corresponding penguin shifts are not doubly cabibbo - suppressed . if we rescale the result in eq . by @xmath228
, we expect hadronic penguin shifts of @xmath229 in the @xmath9 channel . however , as we noted after eqs . and , and as we will see below , also the strong phases play an important role for the numerical values .
the hadronic parameters in the @xmath230 system and in @xmath9 are generally expected to differ from one another .
the lhcb collaboration has recently reported the first experimental results for cp violation in the @xmath9 channel @xcite .
the measurements of the polarisation - dependent effective @xmath151@xmath152 mixing phases are given as follows : @xmath231 within the uncertainties , no dependence on the final - state configuration @xmath32 is detected .
assuming penguin parameters independent of @xmath32 , i.e.@xmath232 in analogy to the relations in eq . , the phase @xmath233 and the cp asymmetries @xmath234 are extracted from the experimental analysis of the time - dependent angular distribution of the @xmath235 \rho^0[\to \pi^+\pi^-]$ ] decay products . the formulae in section [ sec : hadr ] allow the conversion of these results into the @xmath9 penguin parameters .
to this end , we assume again the ckmfitter value of @xmath36 in eq . . moreover , we need the @xmath151@xmath152 mixing phase @xmath18 as an input for the analysis of the mixing - induced cp asymmetry
. however , as we have actually extracted @xmath18 from the global fit discussed in section [ ssec : constr ] , we shall use the value in eq .
for the @xmath9 analysis . in the lhcb study of ref .
@xcite , corrections from penguin contributions to @xmath0 were not taken into account .
the main results of the @xmath136 fit to the data read as follows : @xmath236 in fig .
[ fig : b2vv ] , we show the corresponding confidence level contours with the bands of the individual observables . it is interesting to note that the current experimental measurement of @xmath219 from is in slight tension with the results from
should this turn out not to be a mere fluctuation of the data , which seems unlikely , the effect can not be explained by penguin effects alone .
we have also explored a polarisation - dependent analysis of the penguin effects in using the same strategy as for the above fit .
the resulting confidence level contours are shown in fig .
[ fig : b2vv ] .
they are compatible with the polarisation - independent results in eq . , but the current uncertainties are too large to draw further conclusions . this analysis should be seen as an illustration and motivation for experimentalists to perform more precise polarisation - dependent measurements , which are the method of choice in the long run . neglecting exchange and penguin annihilation topologies ( see fig . [
fig : feynman_additional ] ) , the @xmath5 flavour symmetry allows us to convert the hadronic parameters of the @xmath9 decay into their @xmath1 counterparts @xcite : @xmath237 allowing us to convert the penguin parameters in eq . into the hadronic phase shift of the @xmath1 decay . parametrising possible @xmath5-breaking effects as in eq . with @xmath238 and @xmath239
, we obtain @xmath240^{\circ}\:,\ ] ] which is statistics limited , even when assuming larger @xmath5-breaking uncertainties .
the power of mixing - induced cp violation in @xmath9 for this determination is remarkable @xcite .
it should be compared with the current value of @xmath207 in eq . , which is affected by significantly larger experimental uncertainties .
the contours in fig .
[ fig : b2vv ] do not rely on information from decay rates and are theoretically clean . as in the discussion of the @xmath4 benchmark scenario in section [ ssec : scene ]
, we may use the penguin parameters extracted from the cp asymmetries of @xmath9 to determine the ratio of cp - conserving strong amplitudes , in analogy to eq . .
the only conceptual difference is that polarisation - dependent studies should be performed in the @xmath9 and @xmath1 systems .
following these lines , we obtain the amplitude ratios @xmath241 which are still consistent with the limit of no @xmath5-breaking corrections .
these results can be compared with qcd calculations , such as the recent results obtained in ref .
@xcite within the perturbative qcd ( pqcd ) approach . within naive factorisation , the lcsr form factors of ref .
@xcite ( see table 8) yield @xmath242 although the uncertainties are still very large , these numbers are consistent with the results in eqs .
, and imply @xmath243 consequently , either the non - factorisable contributions @xmath244 themselves or the difference ( due to @xmath5-breaking effects ) between the ratios @xmath245 and @xmath246 is small . in view of the discussion after eqs . and
, the latter option is favoured .
a similar picture also arises for @xmath5-breaking effects in @xmath247 , @xmath248 , @xmath249 decays , which exhibit a different decay dynamics @xcite . in view of this observation , we get confidence in the first relation in eq .
( and the uncertainties assumed in eq . ) .
it is interesting to note that the experimental uncertainties of the ratios in eqs .
are already smaller or of similar size than the uncertainties of the theoretical calculations , which are challenging to improve .
the decay @xmath10 originates from @xmath221 quark - level processes and is the @xmath6-meson counterpart of the @xmath9 mode .
the cdf @xcite and lhcb @xcite collaborations have measured the branching ratio . in the sm , the decay amplitude takes the form @xmath250\:,\ ] ] where we have introduced the tilde to distinguish the hadronic @xmath10 parameters from their @xmath9 counterparts . using @xmath5 flavour symmetry arguments and neglecting penguin annihilation and exchange topologies in @xmath9
, we obtain the relations @xmath251 in order to apply the formalism of section [ sec : hadr ] , we have to make the substitutions @xmath252 in contrast to the @xmath9 channel , the @xmath10 decay does _ not _ exhibit mixing - induced cp violation as the @xmath253 final state is flavour specific , i.e. the pion and kaon charges of @xmath254 and @xmath255 distinguish between initially present @xmath6 and @xmath7 mesons , respectively . consequently , in order to determine the penguin parameters , we have to rely on direct cp violation and decay rate information @xcite .
for each of the final - state configurations @xmath198 , we have a direct cp asymmetry @xmath256 and an observable corresponding to eq .
: @xmath257 where @xmath258 is the polarisation fraction of the @xmath259 channel with @xmath260 . also for the vector vector modes
the @xmath261 observables use the `` theoretical '' branching ratio concept , which for @xmath44 decays differs from the experimentally measured time - integrated branching ratio @xcite .
the conversion factors are similar to eq . but become polarisation dependent .
the measurement of these observables , which depend on @xmath262 and @xmath263 as well as @xmath36 , requires again an angular analysis of the decay products . since @xmath36 is an input , we may determine the penguin parameters for the different final state configurations @xmath32 @xcite .
in contrast to the analysis of @xmath55 , where mixing - induced cp violation plays the key role , this method is affected by hadronic uncertainties which enter through the @xmath264 ratios .
the extraction of these quantities from the data involves ratios of strong amplitudes , which depend on hadronic form factors and non - factorisable effects . in ref .
@xcite , a detailed analysis of these quantities has been performed within the pqcd approach .
we shall return to this topic below .
measurements of direct cp asymmetries of the @xmath10 decay have not yet been performed .
using eq . , we expect them to equal those of @xmath9 : @xmath265 where the cp asymmetries are defined as in ref .
. it will be interesting to confront these numbers with future experimental results .
in the era of the lhcb upgrade and belle ii , there will be a powerful interplay of the different decay channels discussed in this paper .
the measurement of cp violation in the @xmath4 decay will allow us to extract the corresponding penguin parameters in a theoretically clean way at lhcb and to control the penguin effects in the extraction of @xmath18 from @xmath0 with the help of the @xmath3-spin symmetry @xcite . at belle ii
, it will be important to measure cp violation in @xmath124 and to resolve the current discrepancy between the babar and belle measurements of the mixing - induced cp asymmetry ( see eq . ) .
the penguin parameters can be determined in analogy to the @xmath4 strategy @xcite . however , whereas the @xmath3-spin symmetry is sufficient in the case of the @xmath230 system , the @xmath124 mode is affected by further uncertainties due to penguin annihilation and exchange topologies , which arise in @xmath124 but have no counterpart in @xmath0 . some of these amplitudes are isospin suppressed ( and thus expected to be very small ) but not those competing with the penguin contributions . the annihilation and exchange topologies can be probed through @xmath125 @xcite .
the lhcb collaboration does not see any evidence for the @xmath126 channel in the current data @xcite .
following these lines , the @xmath151@xmath152 mixing phase @xmath18 can be extracted with unprecedented precision .
the key question is whether the comparison with the sm value @xmath266 will result in a discrepancy , thereby indicating a cp - violating np phase @xmath267 . here
the interplay between @xmath36 and the side @xmath268 of the ut is crucial @xcite : @xmath269 the precision will be governed by @xmath268 @xcite .
future data collected at the belle ii experiment and theoretical progress will hopefully resolve the discrepancy between the determination of @xmath268 from inclusive and exclusive semileptonic @xmath15 decays @xcite .
the angle @xmath36 can be determined with high precision from @xmath122 decays , as given in eq . .
measurements of the cp violation in @xmath1 will play a key role for the determination of @xmath8 .
it will be important to have polarisation - dependent analyses of @xmath270 available with a precision much higher than the pioneering lhcb results reported recently in ref .
different values would signal the presence of penguin effects and a violation of the relations for the penguin parameters in eq . .
measurements of the direct and mixing - induced cp - violating observables of the @xmath9 channel allow us to determine the corresponding penguin parameters in a clean way @xcite . here
the value of @xmath18 determined from the @xmath27 system is needed as an input .
also in the @xmath9 analysis it will be important to make final - state - dependent measurements .
the experimental results of ref .
@xcite provide a fertile ground for these analyses .
using then the relations in eq .
allows us to determine the phase shifts @xmath271 and to extract the values of @xmath8 from the effective mixing phases @xmath270 of the @xmath1 channel .
, @xmath10 and @xmath1 modes to simultaneously determine the penguin parameters , the ratio of @xmath5-breaking strong amplitudes , and the cp - violating @xmath6@xmath7 mixing phase @xmath8.,scaledwidth=102.0% ] the penguin effects can also be probed by the @xmath10 decay @xcite .
this channel provides direct cp asymmetries but no mixing - induced cp violation as the final state is flavour - specific . in order to make use of the branching ratio information , ratios of strong amplitudes
@xmath272 are needed which introduce hadronic form - factor and non - factorisable uncertainties into the analysis . however
, these ratios can actually be fixed through experiment . from the @xmath9 , @xmath1 analysis
, we may determine the ratios @xmath273 in a theoretically clean way , as we discussed in eqs .
for the current data . using the relation in eq .
, we obtain @xmath274 which allows us to convert the @xmath10 rate measurements into the @xmath264 observables . finally , using also the relation @xmath275 it is possible to make a simultaneous @xmath136 fit to the experimental data offered by the @xmath1 , @xmath9 , @xmath10 system as illustrated in the flow chart in fig .
[ fig : big - fit - flow ] .
this global analysis allows us to combine all the information offered by the penguin control channels in an optimal way and provides valuable insights into strong interactions as a by - product . even though the direct cp asymmetry measurements in @xmath10 are at present not yet available , we can already implement this strategy and extend the fits in fig .
[ fig : b2vv ] to include branching ratio information from @xmath1 , @xmath9 and @xmath10 .
the results of this analysis are @xmath276 we observe that with the current experimental precision the additional branching ratio information does not have any impact on the determination of @xmath277 and @xmath220 with respect to the fits to the @xmath9 system only .
the information is fully used to constrain the amplitude ratios @xmath278 , which were previously not included in the fit . to observe any impact on @xmath277 and @xmath220 , the combined experimental precision on the @xmath120 observables needs to be improved by at least an order of magnitude .
numerical differences in @xmath278 compared to eqs .
arise due to the added information originating from the @xmath10 system .
this extended fit may be further refined by adding information from @xmath126 to probe exchange and penguin annihilation topologies .
@xmath17 mixing phases and the channels needed to control the penguin contributions in the former measurements.,scaledwidth=40.0% ] there is actually an interplay between the high - precision determinations of @xmath18 and @xmath8 .
the point is that @xmath18 is needed as an input for the analysis of mixing - induced cp violation of @xmath9 whereas @xmath8 is required for the analysis of mixing - induced cp violation of @xmath4 .
we have illustrated these cross links in fig .
[ fig : interplay ] .
consequently , it will be advantageous to eventually perform a simultaneous analysis of the @xmath230 and @xmath1 , @xmath9 , @xmath10 systems . and
@xmath18 for @xmath279 models with flavour - universal cp - violating np phases characterised by eq .
: we show the current experimental situation ( left ) and extrapolate to the lhcb upgrade era ( right).,title="fig:",scaledwidth=45.0% ] and @xmath18 for @xmath279 models with flavour - universal cp - violating np phases characterised by eq .
: we show the current experimental situation ( left ) and extrapolate to the lhcb upgrade era ( right).,title="fig:",scaledwidth=45.0% ] for the search of np in the era of the lhcb upgrade @xcite and belle ii @xcite , it will be important to have determinations of both @xmath18 and @xmath8 available with the highest possible precision .
we obtain an interesting correlation between these mixing phases if their np phases in eq .
take the same value : @xmath280 this relation , which was considered in refs .
@xcite on a phenomenological basis , arises actually in extensions of the sm going beyond minimal flavour violation " ( mfv ) , which are characterised by flavour - universal cp - violating np phases ( for an overview , see ref .
@xcite ) . in this specific class of np , referred to as @xmath279 models , we obtain the following correlation : @xmath281 which allows an experimental test . in fig .
[ fig : non - mfv ] , we illustrate this relation both for the current situation and for the expected situation in the lhcb upgrade era .
the future uncertainty of the value of @xmath266 will be fully governed by @xmath268 ( see eq . ( [ s2b - expr ] ) ) , which enters also the band representing the relation in eq . .
it will be interesting to confront these considerations with experimental data in the next decade .
the picture emerging from run i of the lhc suggests that we have to prepare ourselves to deal with smallish np effects . for the determination of the @xmath16@xmath17 mixing phases @xmath18 and
@xmath8 from cp violation measurements in @xmath282 and @xmath283 , respectively , this implies that controlling higher order hadronic corrections , originating from doubly cabibbo - suppressed penguin topologies , becomes mandatory . in this paper
, we have outlined strategies to accomplish this task using the @xmath5 flavour symmetry of qcd .
the penguin contributions to @xmath282 can be controlled with the help of its @xmath3-spin partner @xmath284 .
as the required cp violation measurements of the latter mode are not yet available , we have performed a global fit to current data for cp asymmetries and branching ratios of @xmath285 modes with similar dynamics to already constrain the hadronic penguin shift affecting the @xmath282 channel . for the future lhcb upgrade era
we have illustrated the potential of the @xmath284 mode , which represents the cleanest penguin probe , with a benchmark scenario .
in addition , we have discussed a strategy to probe non - factorisable @xmath3-spin - breaking effects in the @xmath286 system .
the penguin contributions to @xmath283 can be controlled with the help of the modes @xmath287 and @xmath288 .
we have analysed the first lhcb measurement of cp violation in @xmath287 , taking into account possible penguin effects in the required input for @xmath18 . in view of the excellent precision that can already be obtained in this analysis
, the @xmath287 mode is expected to play the key role for the control of the penguin effects in the determination of @xmath8 .
we have proposed a new strategy to add the @xmath288 data to this analysis in a global fit , which does not require knowledge of form factors for the interpretation of the decay rate information .
it rather allows us to determine also hadronic parameters , which then provide insights into non - factorisable @xmath5-breaking effects .
adding @xmath289 to the analysis , also the impact of penguin annihilation and exchange topologies , which are expected to be small , can be probed through experimental data . finally , we propose a combined analysis of the @xmath286 and @xmath283 , @xmath287 , @xmath288 systems in order to simultaneously determine the mixing phases @xmath18 and @xmath8 , taking into account the cross - correlations between these modes in the control of the penguin effects . for the search of new sources of cp violation in the era of the lhcb upgrade and belle ii , simultaneous high - precision measurements of @xmath18 and @xmath8 are crucial ingredients . in extensions of the sm , such as @xmath279 models ,
characteristic correlations between @xmath18 and @xmath8 arise which can then be tested .
while the sm prediction of @xmath8 has already a precision much smaller than the lhcb upgrade sensitivity , the major limitation for @xmath290 is given by the determination of @xmath291 entering the ut side @xmath268 .
future progress on this long - standing challenge would be very desirable to complement the cutting - edge analyses of cp violation .
we look forward to moving to the high - precision frontier !
we would like to thank patrick koppenburg for discussions and comments on the manuscript and are grateful to sheldon stone for correspondence .
the framework introduced in section [ sec : hadr ] can be extended to allow for annihilation topologies @xmath292 .
the amplitude of the decay @xmath293 can be written as @xmath294,\ ] ] where @xmath295\ ] ] is defined as in eq .
( [ aprime - def ] ) , whereas @xmath296 with @xmath297\ ] ] and @xmath298.\ ] ] the penguin parameter @xmath299 is defined in analogy to eq . , while the relative contribution from the annihilation topology is probed by @xmath300 .
the direct cp asymmetry in then takes the form @xmath301 whereas the ratio depends on @xmath302 and @xmath303 as @xmath304 assuming @xmath306 and universal penguin parameters , i.e.@xmath307 the annihilation parameters @xmath302 and @xmath303 can be obtained from a @xmath136 fit to the two direct cp asymmetries and the two @xmath308 ratios listed above .
including the observables @xmath36 ( from eq . ) , @xmath115 and @xmath116 ( from eq . ) as gaussian constraints results in the solution @xmath309 with the corresponding confidence level contours shown in fig .
[ fig : annihilation_abs ] .
the result is compatible with @xmath310 , which is consistent with our assumption to neglect contributions from annihilation topologies in the main @xmath136 fit . and
@xmath303 , which probe annihilation topologies in @xmath130 and @xmath311 decays , through intersecting contours corresponding to the current data for the cp asymmetries and branching ratio information .
we show also the confidence level contours following from a @xmath136 fit.,scaledwidth=75.0% ] the results in eq .
assume external input for the penguin parameters @xmath115 and @xmath116 , and therefore do not take into account the back reaction of a non - zero value of @xmath312 on @xmath85 .
the annihilation topologies could lead to effects of similar size as the exchange and penguin annihilation topologies , which can be probed through the @xmath125 decay . in the future , with stringent constraints on the branching ratio of this channel , an extended fit could be made , including all additional topologies .
but then we expect to have also high - precision measurements of the cp violation in @xmath147 available , allowing us to implement the strategy discussed in the main part of the paper .
the extended fit would nevertheless offer an interesting cross - check to complement the picture of the penguin parameters .
99 g. aad _ et al . _ [ atlas collaboration ] , phys .
b * 716 * ( 2012 ) 1 http://arxiv.org/abs/1207.7214[[arxiv:1207.7214 [ hep - ex ] ] ] .
s. chatrchyan _ et al . _
[ cms collaboration ] , phys .
b * 716 * ( 2012 ) 30 http://arxiv.org/abs/1207.7235[[arxiv:1207.7235 [ hep - ex ] ] ] .
j. ellis , talk at beauty 2014 , edinburgh , united kingdom , 1418 july 2014 http://arxiv.org/abs/1412.2666[[arxiv:1412.2666 [ hep - ph ] ] ] .
_ et al . _ [ belle - ii collaboration ] , http://arxiv.org/abs/1011.0352[arxiv:1011.0352 [ physics.ins-det ] ] .
n. cabibbo , phys .
* 10 * ( 1963 ) 531 .
m. kobayashi and t. maskawa , prog .
* 49 * ( 1973 ) 652 .
j. charles _ et al . _
[ ckmfitter collaboration ] , phys . rev .
d * 84 * ( 2011 ) 033005 http://arxiv.org/abs/1106.4041[[arxiv:1106.4041 [ hep - ph ] ] ] ; for updates , see http://ckmfitter.in2p3.fr .
l. wolfenstein , phys .
* 51 * ( 1983 ) 1945 .
r. fleischer , eur .
j. c * 10 * ( 1999 ) 299 http://arxiv.org/abs/hep-ph/9903455[[arxiv:hep-ph/9903455 ] ] .
r. fleischer , phys .
d * 60 * ( 1999 ) 073008 http://arxiv.org/abs/hep-ph/9903540[[arxiv:hep-ph/9903540 ] ] .
r. fleischer , nucl .
instrum .
a * 446 * ( 2000 ) 1 http://arxiv.org/abs/hep-ph/9908340[[arxiv:hep-ph/9908340 ] ] .
m. ciuchini , m. pierini and l. silvestrini , phys .
* 95 * ( 2005 ) 221804 http://arxiv.org/abs/hep-ph/0507290[[arxiv:hep-ph/0507290 ] ] ; http://arxiv.org/abs/1102.0392[arxiv:1102.0392 [ hep - ph ] ] .
s. faller , r. fleischer , m. jung , and t. mannel , phys .
d * 79 * ( 2009 ) 014030 http://arxiv.org/abs/0809.0842[[arxiv:0809.0842 [ hep - ph ] ] ] .
s. faller , r. fleischer and t. mannel , phys .
d * 79 * ( 2009 ) 014005 http://arxiv.org/abs/0810.4248[[arxiv:0810.4248 [ hep - ph ] ] ] .
m. jung , phys .
d * 86 * ( 2012 ) 053008 http://arxiv.org/abs/1206.2050[[arxiv:1206.2050 [ hep - ph ] ] ] .
x. liu , w. wang and y. xie , phys .
d * 89 * ( 2014 ) 094010 http://arxiv.org/abs/1309.0313[[arxiv:1309.0313 [ hep - ph ] ] ] .
a. s. dighe , i. dunietz , h. j. lipkin and j. l. rosner , phys .
b * 369 * ( 1996 ) 144 http://arxiv.org/abs/hep-ph/9511363[[arxiv:hep-ph/9511363 ] ] .
a. s. dighe , i. dunietz and r. fleischer , eur .
j. c * 6 * ( 1999 ) 647 http://arxiv.org/abs/hep-ph/9804253[[arxiv:hep-ph/9804253 ] ] .
r. aaij _ et al .
_ [ lhcb collaboration ] , http://arxiv.org/abs/1411.3104[arxiv:1411.3104 [ hep - ex ] ] .
r. aaij _ et al . _
[ lhcb collaboration ] , http://arxiv.org/abs/1411.1634[arxiv:1411.1634 [ hep - ex ] ] .
l. zhang and s. stone , phys .
b * 719 * ( 2013 ) 383 http://arxiv.org/abs/1212.6434[[arxiv:1212.6434 ] ] .
r. fleischer , phys .
* 370 * ( 2002 ) 537 http://arxiv.org/abs/hep-ph/0207108[[arxiv:hep-ph/0207108 ] ] .
t. aaltonen _ et al . _
[ cdf collaboration ] , phys .
d * 83 * ( 2011 ) 052012 http://arxiv.org/abs/1102.1961[[arxiv:1102.1961 [ hep - ex ] ] ] .
r. aaij _ et al . _
[ lhcb collaboration ] , nucl .
b * 873 * ( 2013 ) 275 http://arxiv.org/abs/1304.4500[[arxiv:1304.4500 [ hep - ex ] ] ] .
i. dunietz , r. fleischer and u. nierste , phys .
d * 63 * ( 2001 ) 114015 http://arxiv.org/abs/hep-ph/0012219[[arxiv:hep-ph/0012219 ] ] .
k. a. olive _ et al . _
[ particle data group collaboration ] , chin .
c * 38 * ( 2014 ) 090001
. y. amhis _ et al .
_ [ heavy flavor averaging group collaboration ] , http://arxiv.org/abs/1207.1158[arxiv:1207.1158 [ hep - ex ] ] ; for updates , see http://www.slac.stanford.edu / xorg / hfag/. k. de bruyn , r. fleischer , r. knegjens , p. koppenburg , m. merk and n. tuning , phys .
d * 86 * ( 2012 ) 014027 http://arxiv.org/abs/1204.1735[[arxiv:1204.1735 [ hep - ph ] ] ] .
r. fleischer and t. mannel , phys .
d * 57 * ( 1998 ) 2752 http://arxiv.org/abs/hep-ph/9704423[[arxiv:hep-ph/9704423 ] ] .
r. fleischer and s. recksiegel , phys .
d * 71 * ( 2005 ) 051501 http://arxiv.org/abs/hep-ph/0409137[[arxiv:hep-ph/0409137 ] ] .
k. de bruyn , r. fleischer and p. koppenburg , eur .
j. c * 70 * ( 2010 ) 1025 http://arxiv.org/abs/1010.0089[[arxiv:1010.0089 [ hep - ph ] ] ] .
a. bevan , _ et al .
_ [ utfit collaboration ] , http://arxiv.org/abs/1411.7233[arxiv:1411.7233 [ hep - ph ] ] ; for updates , see http://www.utfit.org .
s. e. lee _ et al .
_ [ belle collaboration ] , phys .
d * 77 * ( 2008 ) 071101 http://arxiv.org/abs/0708.0304[[arxiv:0708.0304 [ hep - ex ] ] ] .
b. aubert _ et al . _ [ babar collaboration ] , phys .
* 101 * ( 2008 ) 021801 http://arxiv.org/abs/0804.0896[[arxiv:0804.0896 [ hep - ex ] ] ] .
a. bharucha , jhep * 1205 * ( 2012 ) 092 http://arxiv.org/abs/1203.1359[[arxiv:1203.1359 [ hep - ph ] ] ] .
a. khodjamirian , t. mannel , a. a. pivovarov and y .- m .
wang , jhep * 1009 * ( 2010 ) 089 http://arxiv.org/abs/1006.4945[[arxiv:1006.4945 [ hep - ph ] ] ] .
g. duplancic and b. melic , phys .
d * 78 * ( 2008 ) 054015 http://arxiv.org/abs/0805.4170[[arxiv:0805.4170 [ hep - ph ] ] ] .
p. ball , phys .
b * 644 * ( 2007 ) 38 http://arxiv.org/abs/hep-ph/0611108[[arxiv:hep-ph/0611108 ] ] .
r. aaij _ et al . _
[ lhcb collaboration ] , eur .
j. c * 73 * ( 2013 ) 4 , 2373 http://arxiv.org/abs/1208.3355[[arxiv:1208.3355 [ hep - ex ] ] ] .
r. aaij _ et al . _
[ lhcb collaboration ] , jhep * 1411 * , 060 ( 2014 ) http://arxiv.org/abs/1407.6127[[arxiv:1407.6127 [ hep - ex ] ] ] .
r. aaij _ et al . _
[ lhcb collaboration ] , phys .
d * 86 * ( 2012 ) 071102 http://arxiv.org/abs/1208.0738[[arxiv:1208.0738 [ hep - ex ] ] ] .
r. aaij _ et al . _
[ lhcb collaboration ] , jhep * 1304 * , 001 ( 2013 ) http://arxiv.org/abs/1301.5286[[arxiv:1301.5286 [ hep - ex ] ] ] ; b. storaci _ et al . _
[ lhcb collaboration ] , http://cds.cern.ch/record/1559262[lhcb-conf-2013-011 ] .
r. fleischer , n. serra and n. tuning , phys .
d * 82 * ( 2010 ) 034038 http://arxiv.org/abs/1004.3982[[arxiv:1004.3982 [ hep - ph ] ] ] .
j. l. rosner , phys .
d * 42 * ( 1990 ) 3732 .
a. lenz , talk at ckm 2014 , vienna , austria , 812 september 2014 http://arxiv.org/abs/1409.6963[[arxiv:1409.6963 [ hep - ph ] ] ] .
p. ball and r. zwicky , phys .
d * 71 * ( 2005 ) 014029 http://arxiv.org/abs/hep-ph/0412079[[arxiv:hep-ph/0412079 ] ] .
r. fleischer and r. knegjens , eur .
j. c * 71 * ( 2011 ) 1532 [ arxiv:1011.1096 [ hep - ph ] ] .
r. aaij _ et al . _
[ lhcb collaboration ] , phys .
d * 89 * ( 2014 ) 9 , 092006 http://arxiv.org/abs/1402.6248[[arxiv:1402.6248 [ hep - ex ] ] ] .
a. j. buras and r. fleischer , adv .
direct . high energy phys . * 15 * ( 1998 ) 65 http://arxiv.org/abs/hep-ph/9704376[[arxiv:hep-ph/9704376 ] ] .
p. ball and r. fleischer , eur .
j. c * 48 * ( 2006 ) 413 http://arxiv.org/abs/hep-ph/0604249[[arxiv:hep-ph/0604249 ] ] .
a. j. buras and d. guadagnoli , phys .
d * 78 * ( 2008 ) 033005 http://arxiv.org/abs/0805.3887[[arxiv:0805.3887 [ hep - ph ] ] ] .
a. j. buras and j. girrbach , rept .
* 77 * ( 2014 ) 086201 http://arxiv.org/abs/1306.3775[[arxiv:1306.3775 [ hep - ph ] ] ] . | measurements of cp violation in @xmath0 and @xmath1 decays play key roles in testing the quark - flavour sector of the standard model .
the theoretical interpretation of the corresponding observables is limited by uncertainties from doubly cabibbo - suppressed penguin topologies . with continuously increasing experimental precision , it is mandatory to get a handle on these contributions , which can not be calculated reliably in qcd . in the case of the measurement of @xmath2 from @xmath0 ,
the @xmath3-spin - related decay @xmath4 offers a tool to control the penguin effects . as the required measurements are not yet available , we use data for decays with similar dynamics and the @xmath5 flavour symmetry to constrain the size of the expected penguin corrections .
we predict the cp asymmetries of @xmath4 and present a scenario to fully exploit the physics potential of this decay , emphasising also the determination of hadronic parameters and their comparison with theory . in the case of the benchmark mode @xmath1 used to determine the @xmath6@xmath7 mixing phase @xmath8 the penguin effects can be controlled through @xmath9 and @xmath10 decays .
the lhcb collaboration has recently presented pioneering results on this topic .
we analyse their implications and present a roadmap for controlling the penguin effects .
nikhef-2014 - 031
* kristof de bruyn@xmath11 and robert fleischer@xmath12 * @xmath11_nikhef , science park 105 , nl-1098 xg amsterdam , netherlands _
@xmath13_department of physics and astronomy , vrije universiteit amsterdam , + nl-1081 hv amsterdam , netherlands _
december 2014 |
a wealth of experiments have explored the competition between different types of fluctuation which disrupt ordered phases @xcite .
a typical combination is thermal fluctuations and quenched random fluctuations due to disorder .
the strength of the thermal fluctuations is controlled via the temperature while the strength of the quenched fluctuations is controlled by dilution of the constituents . since both forms of fluctuation perturb the order they are never in profound competition . here , by contrast , we study the interaction between quantum fluctuations , which give rise to an order - from - disorder transition @xcite , and quenched random fluctuations . as the quantum fluctuations drive the ordering in the chosen material the two sets of fluctuations are in direct competition and the results
can be expected to be fascinating .
ba@xmath0cu@xmath1o@xmath2cl@xmath0 is a lamellar cuprate with two cu sublattices @xcite .
the cu@xmath3 sublattice has cuo@xmath0 layers and orders antiferromagnetically at @xmath4 k. the cu@xmath5 sublattice interpenetrates the cu@xmath3 sublattice with one cu@xmath6 ion at the center of every second plaquette .
the position of the cu@xmath5 sublattice relative to a cu@xmath3 plaquette is shown in fig .
[ coupling](a ) .
the plaquettes are staggered from one plane to the next .
the cu@xmath5 moments order antiferromagnetically at @xmath7 k. the isotropic couplings between the cu@xmath3 and cu@xmath5 sublattices are completely frustrated and this means that more subtle interactions are manifest @xcite .
collinear ordering of the second sublattice with respect to the first is preferred due to the contribution of quantum fluctuations . as a result
the cu@xmath5 system has ising class critical properties .
detailed studies of the pure _ 2342 _ materials have been carried out in recent years @xcite . in an applied magnetic field , in the temperature range @xmath8 , sr@xmath0cu@xmath1o@xmath2cl@xmath0 develops a small magnetic moment on the cu@xmath5 sites .
chou and coworkers studied the behavior as a function of temperature , field strength and field orientation @xcite . combined with the known behavior of the cu@xmath3 order parameter it was possible to determine the nature of the coupling between cu@xmath3 and cu@xmath5 moments .
the isotropic average coupling @xmath9 mev and may originate from superexchange and direct exchange .
the anisotropic coupling is @xmath10ev and is pseudo - dipolar .
noro and coworkers studied the weak ferromagnetism in ba@xmath0cu@xmath1o@xmath2cl@xmath0 in the pure material and with zn or ni substitution @xcite .
the behavior of the pure material has similar magnetization , ferromagnetism and susceptibility to the sr material .
neither substitution of the spin system resulted in any new phase transitions or glassy behavior . adding zn reduced the ordered moment by a small amount . adding
ni reduced the ordered moment by almost a factor of 2 without altering the transition temperatures . apart from very close to the two ordering transitions
the ferromagnetic component is entirely suppressed .
furthermore the anisotropy in the in - plane susceptibility disappears .
these effects suggest that ni substitution weakens the coupling between the two sublattices .
kim and coworkers @xcite made a comprehensive elastic , quasi - elastic and inelastic neutron scattering study of sr@xmath0cu@xmath1o@xmath2cl@xmath0 .
the role of quantum fluctuations in coupling cu@xmath3 and cu@xmath5 subsystems were demonstrated via the dramatic variation in the energies of the spin - wave modes as the cu@xmath5 subsystem orders .
the behavior is in quantitative agreement with theoretical predictions @xcite .
the dispersion of a cu@xmath5 mode was measured out to the zone boundary and this provided precise verification of theoretical calculations .
the major interactions were found and are indicated in fig .
[ coupling](a ) . for the sr material @xmath11 mev while @xmath12 mev ( @xmath13 and @xmath14 had not been determined by chou and coworkers @xcite ) .
the similar transitions and transition temperatures suggest that the coupling strengths will be very similar for the ba material . here
we are studying this intriguing two sublattice antiferromagnet with the cu moments mixed with @xmath15 co. this is seen to modify some of the magnetic behavior substantially .
similar magnetic systems have been considered theoretically and were called a _ decorated _ model .
villain considered whether spin - glass behavior could occur for an insulator with only antiferromagnetic interactions @xcite .
he first demonstrated that competing ferromagnetic and antiferromagnetic interactions emerge for a two sublattice ising antiferromagnet with spin dilution of one of the sublattices .
subsequently a number of promising scenarios were considered including local canted states : an antiferromagnetic ground state is perturbed when an interstitial magnetic impurity is introduced .
the neighboring moments ca nt due to their interaction with the interstitial moment and this produces a local canted state .
the local canted state is characterized by an ( n-1 ) component magnetic moment which is perpendicular to the ordering direction of the antiferromagnetic ground state .
if there are many local canted states in a system they will interact and undergo a spin - glass transition .
hence different components of the magnetic moment order separately via antiferromagnetic and spin - glass transitions . using an alternative approach , henley considered the role of thermal , quantum and concentration fluctuations in a frustrated antiferromagnet @xcite .
he considered a two sublattice antiferromagnet , in contrast to ba@xmath0cu@xmath1o@xmath2cl@xmath0 , both sublattices were on the same sized lattice .
he , following others @xcite , argued that thermal and quantum fluctuations of one sublattice give rise to collinear ordering in the other .
this is because the largest fluctuations are transverse to the spin direction for both sublattices .
the coupling between these fluctuations reduces the energy .
the fluctuations in composition change the magnitude of the magnetic moment at random and this is equivalent to a longitudinal fluctuation . in order to couple transverse fluctuations to these longitudinal fluctuations
the neighboring spins on the other sublattice align perpendicular to them . if the perturbation due to variations in composition is too small the thermal and quantum fluctuations will maintain the ising type symmetry .
the composition fluctuations will then give random exchange fields which will pin the ordering direction in various locations and may prevent the formation of a long - range ordered state .
experimental studies of ba@xmath0cu@xmath16co@xmath17o@xmath2cl@xmath0 are described and the results presented ( sec .
[ sec : results ] ) and discussed ( sec .
[ sec : discussion ] ) below .
a single crystal of ba@xmath0cu@xmath18co@xmath19o@xmath2cl@xmath0 with @xmath20 was prepared by a direct melt method from powder samples @xcite .
the laminar single crystal had dimensions @xmath21
mm@xmath22 and a well developed ( 0 0 1 ) plane .
the crystal is tetragonal @xcite ( space group _ i4/mmm _ ) with lattice constants a = 5.514 and c = 13.711 at t = 30 k. the cu@xmath3 ordering was studied using neutron scattering with the crystal mounted in the ( h 0 l ) plane while the cu@xmath5 magnetism was observed in the ( h h
l ) scattering plane .
the location of reflections in these planes is shown inset to fig .
[ tdependence](a ) .
the neutron scattering measurements were performed on the bt7 and bt9 instruments at the national institute for standards and technology , center for neutron research ( ncnr ) .
the incident neutrons of energy 14.7 mev were selected with a ( 0 0 2 ) graphite monochromator and the collimation was generally [ 60-40-40-80 ] on bt7 and [ 40-40-40-80 ] on bt9 . a pyrolytic graphite filter was used to suppress higher order contamination . an ( 0 0 2 ) graphite analyzer was used to improve the instrumental resolution and to suppress background counts . for the quasi - elastic study of the diffuse scattering a two - axis configuration
was employed .
the magnetization was measured using a squid magnetometer over a temperature range 4 to 400 k. the temperature dependence of the neutron scattering intensity of the ( 1 0 1 ) reflection is shown in fig . [ tdependence](a ) .
the additional component to the scattering below @xmath23 k is due to the cu@xmath3 antiferromagnetic structure .
the solid line shows the order parameter behavior for pure ba@xmath0cu@xmath1o@xmath2cl@xmath0 . for this line the exponent @xmath24 and
this is close to , but somewhat less than , the value @xmath25 expected for a 3d xy phase transition .
the addition of co is seen to have very little effect on this transition .
new behavior occurs at lower temperatures involving the cu@xmath5 spins .
figure [ tdependence](b ) shows the neutron scattering observed at the ( @xmath26 @xmath26 @xmath27 ) position for temperatures above @xmath28 k. the choice of a large @xmath29 component to the wave vector enhances the scattering from magnetic moments lying within the planes .
the scattering was weak with no clear peak and the results presented are integrated counts .
the measurements were made in two - axis configuration in order to reduce the energy resolution .
the intensity is observed to decrease slowly with decreasing temperature with a more substantial drop beginning around @xmath30 k. this diffuse scattering is different in character to that observed in the pure material @xcite . in the pure case rods of scattering are observed in the vicinity of @xmath31 that reach a peak intensity at the transition and diminish in the ordered phase .
the diffuse scattering observed here corresponds to very short - range correlations which appear to freeze out well above @xmath31 .
the results were confirmed to be similar via more sparse measurements at the ( @xmath26 @xmath26 @xmath32 ) position .
the addition of 2% of co increases the effective interaction strength between the cu@xmath5 spins .
one quarter of the cu@xmath5 moments have a co as a nearest neighbor on one of the two sublattices . for these moments
the _ transition temperature _ effectively triples - in line with the diffuse scattering .
inset to fig .
[ tdependence](b ) is the scattering measured at the ( @xmath26 @xmath26 @xmath32 ) position below @xmath33 k superimposed on the temperature dependence of the diffuse scattering .
the enhanced intensity of scattering results from the ordering of the cu@xmath5 moments . in this case
as seen here the temperature dependence of the order parameter is very similar to that of the pure material @xcite where @xmath35
. both of these are close to the 2d ising value of @xmath36 .
figure [ two - d ] demonstrates the short - range nature of the low temperature magnetic structure perpendicular to the planes . within the cu@xmath1o@xmath2 planes
the ( @xmath26 @xmath26 @xmath37 ) peak width is comparable to the instrumental resolution .
by contrast the magnetic reflection is seen to be broader than the instrumental resolution out of the plane .
the magnetism is correlated over @xmath38 in the [ 0 0 l ] direction which is a little over ten unit cells .
this result suggests that two - dimensional ising behavior is retained .
magnetization measurements have been carried out for ba@xmath0cu@xmath1o@xmath2cl@xmath0 and for two levels of co substitution ( x = 0.05 , 0.1 ) .
one of these is similar to the sample described above .
figure [ noro](a ) shows the magnetization along the [ 1 1 0 ] direction .
the two traces for each sample are for zero - field cooling ( zfc ) and cooling in a 0.1 t field ( fc ) .
all samples exhibit a transition at @xmath39k corresponding to ordering on the cu@xmath3 sites .
the transition shifts up in temperature with co substitution due to the increase in the average moment . for the pure ba@xmath0cu@xmath1o@xmath2cl@xmath0 sample
the zfc curve shows a sharp discontinuity at @xmath40k corresponding to the onset of antiferromagnetic order .
the key feature of these data is the evidence of a transition at intermediate temperature induced by the co substitution . for x
= 0.05 the first transition occurs at @xmath41k and appears rounded ; a second discontinuity is observed at @xmath40k .
these results suggest that order is occurring on the cu@xmath5 sites in two steps .
the magnetization observed is greatly enhanced as a result of substitution of a small amount of co. the dependence of magnetization on the applied field strength can be fitted by @xmath42 , indicating the existence of a weak ferromagnetic component @xmath43 .
the temperature dependence of @xmath43 is shown in fig .
[ noro](b ) and the susceptibility in fig .
[ noro](c ) .
the results for pure ba@xmath0cu@xmath1o@xmath2cl@xmath0 along [ 1 1 0 ] shown here and also along other directions have been published previously by noro and coworkers @xcite .
in contrast to zn and ni substitution , co greatly strengthens the weak ferromagnetism .
figure [ noro](b ) shows @xmath44 along [ 1 1 0 ] .
the pure ba@xmath0cu@xmath1o@xmath2cl@xmath0 trace indicates that a ferromagnetic component is developed at @xmath39k and quickly saturates .
co substitution evidently has two effects : firstly , the transition temperature for the appearance of the cu@xmath3 structure rises .
secondly , the ferromagnetic moment never saturates but continues to rise steadily down to low temperatures .
a 2% substitution triples the ferromagnetic moment with an increase up to 4% giving a six - fold enhancement ! in sr@xmath0cu@xmath1o@xmath2cl@xmath0
it was shown that the ferromagnetic moment tracked the behavior of the order parameter for the cu@xmath3 sites @xcite . here
the order parameter behavior is shown in fig .
[ tdependence](a ) . evidently the ferromagnetic contribution departs from this .
the ferromagnetic contribution to the magnetic structure of pure ba@xmath0cu@xmath1o@xmath2cl@xmath0 is attributed to a canting of the magnetic moments .
this canting of the moments is augmented , both by co substitution and then by decreasing temperature .
figure [ noro](c ) shows the magnetic susceptibility as a function of temperature and co substitution along the [ 1 1 0 ] direction .
the overall shape changes little .
the magnitude of the susceptibility is observed to increase with co substitution and this is the first indication of the presence of the co moments on the cu@xmath5 sites . on top of this
the susceptibility climbs strongly at the lowest temperatures with increasing co substitution .
we have made a study of the magnetic properties of ba@xmath0cu@xmath16co@xmath17o@xmath2cl@xmath0 using neutron scattering and magnetization measurements .
for the pure material the behavior for @xmath45 is marked by a single phase transition .
here we observe evidence of a new transition for @xmath30 k and a transition very similar to that of the pure material at around @xmath31 .
the co substituted material also exhibits a much larger ferromagnetic moment than the pure material .
the cu@xmath3 spins with co substitution order in the same way as they do in the pure material , albeit with an increased average moment .
unlike in the pure material case there is now an uncompensated moment component that is experienced by the cu@xmath5 sites .
there are two useful ways to think about the effect on the cu@xmath5 sites .
firstly , the co substitution can be thought of as creating new , effective , interactions between the cu@xmath5 spins .
secondly , the co substitution can be thought of as longitudinal fluctuations which interact with the thermal and quantum fluctuations of the cu@xmath5 spins .
effective interactions between cu@xmath5 spins give rise to the new behavior we observe in ba@xmath0cu@xmath16co@xmath17o@xmath2cl@xmath0 .
figure [ coupling](a ) shows the cancellation of the cu@xmath3 contributions at the cu@xmath5 sites . a co moment on the central site in this figure
would strengthen the coupling between cu@xmath5 moments but would not create anything new .
the effective interactions arise from the co moments on the cu@xmath3 sites pictured in fig .
[ coupling](b ) .
the existence of the uncompensated moment introduces an effective ferromagnetic interaction j@xmath0 between cu@xmath5 moments .
as pointed out by villain , ferromagnetic interactions at random locations in an antiferromagnetic material would give rise to spin - glass behavior in an ising system @xcite .
in the case of xy or heisenberg symmetry the frustration between ferromagnetic and antiferromagnetic interactions can be resolved by canting of the cu@xmath5 moments close to the co site .
this situation is pictured in fig .
[ coupling](c ) and exhibits simultaneous ferromagnetic and antiferromagnetic order .
villain considered a similar case with interstitial magnetic impurities : a _ decorated _ model @xcite . in that case antiferromagnetic coupling between the host and interstitial were chosen .
for ba@xmath0cu@xmath16co@xmath17o@xmath2cl@xmath0 the local canted states have a non - zero moment as is evident in our magnetization results . in villains model the local canting compensates the impurity moment and a quadrupolar field results @xcite .
the substituted co moments on the cu@xmath3 sites are longitudinal fluctuations in the field experienced by the cu@xmath5 moments .
there is a coupling between cu@xmath3 fluctuations and the fluctuations of the cu@xmath5 moments . in the pure material
the coupling gives rise to collinear ordering @xcite .
this is because the fluctuations are predominantly transverse to the moment direction for both cu@xmath3 and cu@xmath5 spins . here
the co substituted cu@xmath3 spins have large static longitudinal fluctuations .
in order for the cu@xmath5 thermal and quantum fluctuations to couple to the longitudinal fluctuations it is necessary for the cu@xmath5 spins to align perpendicular to the cu@xmath3 spins .
this model was studied by henley @xcite .
canting similar to fig .
[ coupling](c ) is the prediction of considerations of the effective interaction and the fluctuations .
evidently longitudinal fluctuations also exist due to the co substitution on the cu@xmath5 sublattice .
these are at random fixed locations and are dilute .
these fluctuations will couple to the thermal and quantum fluctuations of the cu@xmath3 moments and hence these co moments will also favor perpendicular alignment .
our elastic neutron diffraction results indicate that the cu@xmath3 and cu@xmath5 sublattices of ba@xmath0cu@xmath16co@xmath17o@xmath2cl@xmath0 order in a very similar manner to the pure material ( fig .
[ tdependence ] ) .
we assume that the low temperature ordering of the cu@xmath5 spins is collinear to the ordering of the cu@xmath3 spins as in the pure material .
this assumption is supported by the observation that substitution with co has hardly changed the ordering temperature and critical properties .
if there were competition between collinear and perpendicular ordering , due to the co substitution , it might have been expected that the low temperature transition would have moved away from ising critical properties .
the quasi - elastic measurements ( fig .
[ tdependence](b ) ) revealed diffuse scattering at intermediate temperatures which is associated with very short - range correlations .
this scattering is not the same as that observed in the pure material @xcite .
here the correlations are due to the local ordering close to the co moments on both sublattices .
the co moments on the cu@xmath3 sites create a preferred orientation for the antiferromagnetic fluctuations which is perpendicular to the co moment .
the local canted states give rise to the short - range order .
it appears likely that the drop in quasi - elastic intensity is associated with a freezing of the perpendicular spin components while the collinear component retains the pure material behavior .
it is the interactions between canted regions that lead to spin - glass freezing .
other magnetic effects also occur in the same temperature range where we observe changes in the diffuse scattering .
the development of antiferromagnetic cu@xmath3 domains in the vicinity of 100 k has been studied in detail by parks and coworkers @xcite .
this is a rare magnetization study of antiferromagnetic domains and was possible because of the ferromagnetic moment developed on the cu@xmath5 sites due to the pseudo - dipolar interaction .
two kinds of cu@xmath3 domains are possible and the orientation of the associated ferromagnetic moments are mutually orthogonal and so the domain populations can be studied via the magnetization signal .
the application of a high magnetic field leads to a single domain dominating the sample . as the field
is reduced domains of the other type begin to nucleate . at 100
k the susceptibility for domain wall formation is maximum .
additional studies were made using neutron scattering showing the decrease in intensity of the ( 3 2 1 ) reflection at around 100 k. we do not believe that the decrease in scattering intensity at the ( @xmath26 @xmath26 @xmath27 ) position at around 100 k is associated with the proliferation of domain walls in the cu@xmath3 spin system .
this is true in principle because we did not use an applied magnetic field to create a monodomain sample .
it is also true in practice as can be observed from our measurements : at this temperature the cu@xmath5 antiferromagnetic fluctuations at ( @xmath26 @xmath26 @xmath27 ) are very short ranged ; there is no peak in reciprocal space .
by contrast , the cu@xmath3 antiferromagnetic structure has long - range order .
evidently there is no correspondence between the size of the cu@xmath3 and cu@xmath5 antiferromagnetic domains at these temperatures .
figure [ noro](a ) shows that the three regimes we have observed via neutron scattering are also observed in the magnetization studies presented here .
the first regime is the high temperature cu@xmath3 ordering
. then the spin - glass freezing transition emerges at intermediate temperatures on diluting with co. importantly , the behavior in the middle regime is seen to be new : the divergence between field - cooled and zero field - cooled magnetization measurements for the @xmath46 measurements has no parallel in the pure material ( fig .
[ noro](a ) ) . at low temperatures
the ordering of the cu@xmath5 sublattice persists ( presumably for those moments not pinned by a co moment ) .
figure [ noro](b ) shows that in a magnetic field applied along the [ 1 1 0 ] direction the ferromagnetic moment m@xmath47 is substantially larger than that in the pure material and has different temperature dependence . in the pure material m@xmath47 tracks the cu@xmath3 order parameter and
is induced by pseudo - dipolar coupling between sublattices . in the co substituted system
the applied field aligns the co moments on the cu@xmath3 sites and the canted contributions on the cu@xmath5 sites as pictured in fig .
[ coupling](c ) .
some canting of the cu@xmath3 moments will also occur @xcite .
the combination leads to a large ferromagnetic contribution . figure [ noro](c )
shows that these local large moments also give an increase in the susceptibility .
magnetization measurements have been made for ba@xmath0cu@xmath1o@xmath2cl@xmath0 with the cu diluted with zn , ni and co. in the case of zn and ni no new transitions appear @xcite whereas with co spin - glass behavior results .
these differences stem from the magnitude of the effective interactions induced .
the magnetic interaction between the spin of a cu@xmath48 ion on a cu@xmath3 site and the cu@xmath5 spins is canceled by the other spins on the cu@xmath49 sublattice . ni@xmath48
and zn@xmath48 have moments that differ from cu@xmath48 by @xmath26 .
hence zn and ni on cu@xmath3 sites result in an uncompensated spin one - half moment which is apparent to the cu@xmath5 moments .
co@xmath48 has a moment that differs from cu@xmath48 by 1 hence for co the effect is doubled . since j@xmath5 and j@xmath50 are equal and opposite only co is likely to give rise to new and dominant interactions between the cu@xmath5 spins .
our neutron scattering and magnetization results are consistent with the co substitution leading to a new transition in the cu@xmath5 spin system .
spin - glass freezing at @xmath51 100 k precedes the antiferromagnetic ordering at @xmath51 30 k. we interpret our observations in - terms of the involvement of both the parallel and perpendicular spin components on the cu@xmath5 sites ( fig .
[ coupling](c ) ) .
the change in behavior is due to the influence of the co moments located on the cu@xmath3 sites ; this can be explained in terms of new effective interactions or alternatively coupling between fluctuations . either way the cu@xmath5 moments
are caused to form local canted states around the co moments .
the local canted states frustrate each other and lead to spin - glass freezing .
the local cluster has a substantial moment and this is reflected in the magnetization measurements . at lower temperatures the portion of the cu@xmath5 moment that is collinear with the cu@xmath3 ordering direction ( and not pinned by a co moment ) orders antiferromagnetically as in the pure material .
this system exhibits a delicate balance of quenched random and quantum fluctuations .
the former controls the magnetic behavior at intermediate temperatures while the latter yields order at lower temperatures .
we are grateful to p.m. gehring for invaluable help with the neutron scattering measurements and to y.j .
kim for sharing expertise concerning the properties of pure _ 2342 _ materials .
we acknowledge the support of the national institute of standards and technology , us .
department of commerce , in providing the neutron research facilities used in this work .
funding in toronto was provided by the natural science and engineering research council and in edinburgh by the epsrc ( grant gr / s10377/01 ) .
kim , r.j .
birgeneau , f.c .
chou , m. greven , m.a .
kastner , y.s .
lee , b.o .
wells , a. aharony , o. entin - wohlman , i.ya .
korenblit , a.b .
harris , r.w .
erwin , and g. shirane , phys .
b * 64 * , 024435 ( 2001 )
. schematic illustration of the magnetic couplings .
open ( closed ) circles indicate cu@xmath3 ( cu@xmath5 ) sites .
( a ) @xmath13 couples the cu@xmath3 moments and @xmath14 couples the cu@xmath5 moments . the isotropic coupling @xmath52 between sublattices is completely frustrated .
( b ) when co is added ( large arrow ) there is a new effective coupling @xmath53 between the cu@xmath5 moments .
( c ) schematic illustration of the canted structure .
the co moment in the center induces a ferromagnetic ordering in nearby cu@xmath5 moments .
antiferromagnetic order can form independently in the orthogonal direction . ]
( a ) the temperature dependence of the scattered intensity at the ( 1 0 1 ) position .
the growth in intensity corresponds to the ordering of the moments on the cu@xmath3 sublattice .
the solid line is the known behavior of the pure material @xcite .
inset : reciprocal lattice diagrams of the ( h 0 l ) and ( h h
l ) planes explored in these experiments .
the full circles are the nuclear bragg reflections ; the open circles are the cu@xmath3 magnetic bragg reflections and the full ( up and down ) triangles are the magnetic bragg reflections from the two cu@xmath5 domains .
( b ) the temperature dependence of the short - range correlations at the ( @xmath26 @xmath26 @xmath27 ) position .
the intensity drops for @xmath54 k. inset : the triangles and the circles show the temperature dependence of the ( @xmath26 @xmath26 @xmath32 ) and ( @xmath26 @xmath26 @xmath27 ) positions respectively .
the growth of intensity at the ( @xmath26 @xmath26 @xmath32 ) position is due to antiferromagnetic order involving the cu@xmath5 sites . ]
scans of neutron scattering intensity versus wave - vector transfer for ( a ) @xmath55 = ( h h 3 ) and ( b ) @xmath55 = ( @xmath26 @xmath26 l ) .
the reflection corresponds to the antiferromagnetic order of the cu@xmath5 sublattice .
the dotted line is the instrumental resolution .
correlation between the cu@xmath5 ordering on different planes is evidently short range . ]
temperature dependence of ( a ) the magnetization along [ 1 1 0 ] in ba@xmath0cu@xmath18co@xmath19o@xmath2cl@xmath0 single crystals , measured after field cooling ( 0.1 t ) and zero - field cooling .
the magnetic hysteresis shows the presence of the short - range order .
( b ) the weak ferromagnetic moment m@xmath47 along [ 1 1 0 ] in ba@xmath0cu@xmath18co@xmath19o@xmath2cl@xmath0 single crystals . the canting angle of cu@xmath5 and cu@xmath3 moments increases steeply due to the doping of co ions .
( c ) the magnetic susceptibility along [ 1 1 0 ] in ba@xmath0cu@xmath18co@xmath19o@xmath2cl@xmath0 single crystals .
the increase in the paramagnetic susceptibility by co doping is associated with the formation of local canted states . ] | ba@xmath0cu@xmath1o@xmath2cl@xmath0 has two inter - penetrating square cu sublattices , one with square root 2 times the in - plane spacing of the other .
isotropic magnetic interactions between the two sublattices are completely frustrated .
quantum fluctuations resolve the intrinsic degeneracy in the ordering direction of the more weakly coupled sublattice in favor of collinear ordering .
we present neutron scattering and magnetization studies of the magnetic structure when the cu ions are substituted with co. the co spins create new magnetic interactions between the two sublattices .
the ordering behavior of both cu sublattices is retained largely unmodified . between the phase transitions of the two sublattices spin - glass behavior
is observed .
magnetization results show a strong enhancement to the ferromagnetic aspect of the magnetic structure .
the combination of glassy behavior and large moments strongly suggest that the co moments induce the formation of local canted states . |
the study of the _ moduli spaces _ is a traditional research subject in algebraic geometry .
they are schemes or stacks whose points bijectively correspond to fixed kinds of algebro - geometric objects , say curves , sheaves on a fixed variety , etc .
these moduli spaces are interesting not only in algebraic geometry , but also in connection with other research fields such as number theory , differential geometry and string theory . in general
it is not easy to study the geometric properties of the moduli spaces .
instead one tries to construct and study the invariants of the moduli spaces , e.g. their ( weighted ) euler characteristics , virtual poincar or hodge polynomials , integration of the virtual cycles via deformation - obstruction theory .
it has been observed that the best way to study such invariants is taking the generating series . sometimes the generating series defined from the moduli spaces have beautiful forms and properties .
let us observe this phenomenon for some rather amenable examples .
for a quasi - projective variety @xmath0 ( in this article , we always assume that the varieties are defined over @xmath1 ) , the _ hilbert scheme of n - points _ denoted by @xmath2 is the moduli space of zero dimensional subschemes @xmath3 such that @xmath4 .
it contains an open subset corresponding to @xmath5-distinct points in @xmath0 , and the geometric structures of its complement is in general complicated .
nevertheless if @xmath0 is non - singular , the generating series of the euler characteristics of @xmath2 have the following beautiful forms @xcite , @xcite @xmath6 in the case @xmath7 , the torus localization shows that @xmath8 coincides with the number of @xmath9-dimensional partitions of @xmath5 .
the resulting product formulas are consequences of enumerative combinatorics , as in @xcite .
a general case is reduced to the case @xmath7 .
the study of the invariants of the moduli spaces of curves inside a variety is more important and interesting , because of its connection with _ world sheet counting _ in string theory .
a particularly important case is when @xmath0 is a _ calabi - yau 3-fold _ , i.e. @xmath0 has a trivial canonical line bundle with @xmath10 , as the string theory predicts our universe to be the product of the four dimensional space time with a calabi - yau 3-fold . similarly to @xmath2 ,
we denote by @xmath11 the _ hilbert scheme of curves _ inside @xmath0 , that is the moduli space of projective subschemes @xmath12 with @xmath13 , @xmath14=\beta$ ] and @xmath15 .
the following result was obtained by the author in 2008 , and plays a key role in this article : [ intro : thm ] let @xmath0 be a smooth projective calabi - yau 3-fold .
then for fixed @xmath16 , the quotient series @xmath17 is the laurent expansion of a rational function of @xmath18 , invariant under @xmath19 .
note that the denominator of ( [ intro : hilb ] ) is given by the formula ( [ hilb0 ] ) .
a typical example of a rational function of @xmath18 invariant under @xmath20 is @xmath21 .
we remark that the invariance of @xmath20 does not say the invariance of the generating series after the formal substitution @xmath22 , but so after taking the analytic continuation of the function ( [ intro : hilb ] ) from @xmath23 to @xmath24 .
the above result was conjectured in @xcite as the _ unweighted version _ of the rationality conjecture of rank one donaldson - thomas ( dt ) invariants by maulik - nekrasov - okounkov - pandharipande ( mnop ) @xcite .
the rationality conjecture was proposed in order to formulate the _ gromov - witten / donaldson - thomas correspondence conjecture _ comparing two kinds of curve counting invariants on calabi - yau 3-folds .
the dt invariant was introduced by thomas @xcite , as a holomorphic analogue of casson invariants of real 3-manifolds .
it counts stable coherent sheaves on a calabi - yau 3-fold , and is a higher dimensional generalization of donaldson invariants on algebraic surfaces . for a calabi - yau 3-fold @xmath0 , an ample divisor @xmath25 on @xmath0 and a cohomology class @xmath26 , the dt invariant @xmath27 is defined to be the degree of the zero dimensional virtual fundamental cycle on the moduli space of @xmath25-stable coherent sheaves @xmath28 on @xmath0 with @xmath29 .
it also coincides with the weighted euler characteristic with respect to the behrend s constructible function on that moduli space @xcite .
the dt invariants were later generalized by joyce - song @xcite and kontsevich - soibelman @xcite so that they also count strictly @xmath25-semistable sheaves .
the generalized dt invariants involve the behrend functions and the motivic hall algebras in the definition , and they are @xmath30-valued .
+ the hilbert scheme of points or curves on a calabi - yau 3-fold is also interpreted as a moduli space of stable sheaves , by assigning a subscheme @xmath12 with its ideal sheaf @xmath31 . the resulting dt invariant is the weighted euler characteristic of the hilbert scheme of points or curves , and in particular it is independent of @xmath25 . in this sense ,
the invariant @xmath32 is the unweighted version of the dt invariant , which coincides with the honest dt invariant up to sign if @xmath11 is non - singular .
the result of theorem [ intro : thm ] for the weighted version was later proved by bridgeland @xcite .
the rationality property and the invariance of @xmath20 of the series ( [ intro : hilb ] ) are not visible if we just look at the moduli spaces of curves or points .
such hidden properties of the series ( [ intro : hilb ] ) are visible after we develop new moduli theory and invariants of objects in the derived category of coherent sheaves . recall that for a variety @xmath0 , the bounded derived category of coherent sheaves @xmath33 is defined to be the localization by quasi - isomorphisms of the homotopy category of the bounded complexes of coherent sheaves on @xmath0 .
the derived category is no longer an abelian category , but has a structure of a triangulated category .
it was originally introduced by grothendieck in 1960 s in order to formulate the relative version of serre duality theorem , known as grothendieck duality theorem .
later it was observed by mukai @xcite that an abelian variety and its dual abelian variety , which are not necessary isomorphic in general , have the equivalent derived categories of coherent sheaves .
this phenomena suggests that the category @xmath33 has more symmetries than the category of coherent sheaves , as the latter category is known to reconstruct the original variety .
such a phenomena has drawn much attention since kontsevich proposed the _
homological mirror symmetry conjecture _ in @xcite .
it predicts an equivalence between the derived category of coherent sheaves on a calabi - yau manifold and the derived fukaya category of its mirror manifold , based on an insight that the derived category @xmath33 is a mathematical framework of d - branes of type b in string theory .
there have been several developments in constructing mukai type derived equivalences between non - isomorphic varieties @xcite , @xcite , @xcite , @xcite , and non - trivial autequivalences @xcite , @xcite , based on the ideas from mirror symmetry .
furthermore such mukai type equivalences have been discovered beyond algebraic geometry .
for instance , derived mckay correspondence @xcite gives an equivalence between the derived category of finite group representations and the derived category of coherent sheaves on the crepant resolution of the quotient singularity .
this is now interpreted as a special case of equivalences between usual commutative varieties and non - commutative varieties in the context of van den bergh s non - commutative crepant resolutions @xcite .
there also exists an orlov s equivalence @xcite between the derived category of coherent sheaves on a calabi - yau hypersurface in the projective space and the category of graded matrix factorizations of the defining equation of it .
this result , called landau - ginzbrug / calabi - yau correspondence , was also motivated by mirror symmetry .
now it is understood that the derived categories have more symmetries than the categories of coherent sheaves .
our point of view is to make the hidden properties of the generating series of dt type invariants visible via symmetries in the derived categories .
the idea of applying derived categories to the study of generating series of dt type invariants suggests an importance of constructing moduli spaces and invariants of objects in the derived categories .
note that in constructing the original dt invariants , we need to fix an ample divisor on a calabi - yau 3-fold @xmath0 , and the associated stability condition on @xmath34 in order to construct a good moduli space of stable sheaves .
the notion of stability conditions on triangulated categories , in particular on derived categories of coherent sheaves , was introduced by bridgeland @xcite as a mathematical framework of douglas s @xmath35-stability @xcite in string theory . for a triangulated category @xmath36 , a bridgeland stability condition on it roughly consists of data @xmath37 for a group homomorphism
@xmath38 called the _ central charge _ , and the collection of subcategories @xmath39 for @xmath40 whose objects are called @xmath41__-semistable objects with phase _
_ @xmath42 .
the main result by bridgeland @xcite is to show that the set of ` good ' stability conditions on @xmath36 forms a complex manifold .
this complex manifold is in particular important when @xmath43 for a calabi - yau manifold @xmath0 . in this case
, the space of stability conditions @xmath44 is expected to contain the universal covering space of the moduli space of complex structures of a mirror manifold of @xmath0 .
so far the space @xmath44 has been studied in several situations , e.g. @xmath0 is a curve @xcite , @xcite , @xmath0 is a k3 surface @xcite , @xmath0 is a some non - compact calabi - yau 3-fold @xcite , @xcite , @xcite , @xcite . on the other hand ,
there has been a serious issue in studying bridgeland stability conditions on projective calabi - yau 3-folds which are likely to be the most important case : we are not able to prove the existence of bridgeland stability conditions on smooth projective calabi - yau 3-folds . in @xcite
, the existence problem is reduced to showing a conjectural bogomolov - gieseker type inequality evaluating the third chern character of certain two term complexes of coherent sheaves .
however proving that inequality conjecture seems to require a new idea .
let @xmath0 be a smooth projective calabi - yau 3-fold .
we expect that , for a given @xmath45 and @xmath26 , there exists the dt type invariant @xmath46 which counts @xmath41-semistable objects @xmath47 with @xmath29 .
as we mentioned , there is a serious issue in constructing a bridgeland stability condition on projective calabi - yau 3-folds , but let us ignore this for a while . for an ample divisor @xmath25 on @xmath0 , we expect that the classical @xmath25-stability appears as a certain special limiting point in @xmath44 called the _
large volume limit_. if we take @xmath45 near the large volume limit point , then we expect the equality @xmath48 .
on the other hand , suppose that there is an autequivalence @xmath49 of @xmath33 and @xmath50 so that the equality @xmath51 holds for any @xmath52 .
then the generating series of the invariants @xmath53 is preserved by the variable change induced by @xmath54 .
if we are able to relate @xmath55 and @xmath53 , then it would imply the hidden symmetry of the generating series of classical dt invariants @xmath56 with respect @xmath54 .
the relationship between @xmath55 and @xmath53 is studied by the wall - crossing phenomena : there should be a wall and chamber structure on the space @xmath44 so that the invariants @xmath57 are constant on a chamber but jumps if @xmath58 crosses a wall .
the wall - crossing formula of the invariants @xmath57 should be described by a general framework established by joyce - song @xcite , kontsevich - soibelman @xcite , using stack theoretic hall algebras .
+ however as we mentioned , we are not able to prove @xmath59 , so the above story is the next stage after proving the non - emptiness .
the idea of proving theorem [ intro : thm ] was to introduce ` weak ' bridgeland stability conditions on triangulated categories , and apply the above story for the space of weak stability conditions on the subcategory of @xmath33 generated by @xmath60 and one or zero dimensional sheaves .
the latter subcategory is called the category of _ d0-d2-d6 bound states_. the notion of weak stability conditions is a kind of limiting degenerations of bridgeland stability conditions , and it is a coarse version of bayer s polynomial stability conditions @xcite , the author s limit stability conditions @xcite .
it is easier to construct weak stability conditions and enough to prove theorem [ intro : thm ] applying the above story .
the derived dual @xmath61 , an autequivalence of @xmath33 , turned out to be responsible for the hidden symmetric property of @xmath20 of the series ( [ intro : hilb ] ) in the above story .
+ the idea of proving theorem [ intro : thm ] has turned out to be useful in proving several other interesting properties of dt type invariants , say dt / pt correspondence @xcite , @xcite conjectured by pandharipande - thomas @xcite . we refer to @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite for other works relating the above story . in section [ sec : dt ] , we review and survey recent developments of donaldson - thomas theory . in section [ sec : br ] , we survey the developments on bridgeland stability conditions . in section [ sec : con ] , we discuss open problems on dt theory and bridgeland stability conditions .
let @xmath0 be a smooth projective variety and @xmath25 an ample divisor on @xmath0 . for an object @xmath62 , its hilbert polynomial is given by @xmath63 for @xmath64 by the riemann - roch theorem .
here @xmath65 and @xmath9 is the dimension of the support of @xmath28 .
the reduced hilbert polynomial @xmath66 is defined to be @xmath67 .
an object @xmath62 is @xmath25__-(semi)stable _ _ if for any subsheaf @xmath68 , we have @xmath69 and the inequality @xmath70 holds for @xmath71 . note that if @xmath28 is torsion free , then @xmath72 where @xmath73 , @xmath74 , and @xmath75 is some constant .
hence the @xmath25-(semi)stability is the refinement of @xmath25-slope ( semi)stability defined by the slope function @xmath76 .
let @xmath77 be the 2-functor from the category of complex schemes to the groupoid , whose @xmath78-valued points form the groupoid of flat families of coherent sheaves on @xmath0 over @xmath78 .
the 2-functor @xmath77 forms a stack , which is known to be an artin stack locally of finite type , but neither finite type nor separated .
the situation becomes better if we consider the substacks for @xmath79 @xmath80 here @xmath81 is the substack of @xmath25-(semi)stable @xmath62 with @xmath29 , which is an open substack of @xmath77 .
the stack @xmath82 is of finite type but not separated in general .
the stack @xmath83 is of finite type , separated , and a @xmath84-gerb over a quasi - projective scheme @xmath85 .
the scheme @xmath85 is projective if @xmath86 .
let @xmath0 be a smooth projective 3-fold .
we say it is a _ calabi - yau 3-fold _ if @xmath87 and @xmath10 .
a typical example is a quintic hypersurface in @xmath88 .
let @xmath25 be an ample divisor on @xmath0 , @xmath52 an element in @xmath89 , and consider the moduli scheme @xmath85 . a standard deformation theory of sheaves ( cf .
@xcite ) shows that the tangent space at @xmath90 \in m_h^{\rm{s}}(v)$ ] is given by @xmath91 , and the obstruction space is given by @xmath92 .
the calabi - yau condition and the serre duality implies that the latter space is dual to @xmath91 . hence the virtual dimension at
@xmath90 $ ] , defined to be the dimension of the tangent space minus the dimension of the obstruction space , is zero which is independent of @xmath28 . based on this observation , thomas @xcite constructed two term complex of vector bundles @xmath93 on @xmath85 and a morphism @xmath94 in @xmath95 , giving a symmetric perfect obstruction theory in the sense of behrend - fantechi @xcite , @xcite . by the construction in @xcite , there is the associated zero dimensional virtual cycle @xmath96^{\rm{vir}}$ ] on @xmath85 , and we are able to take its degree if @xmath85 is projective .
the dt invariant is defined as follows : if @xmath86 holds , we define @xmath27 to be the degree of @xmath96^{\rm{vir}}$ ] .
the above construction via the virtual cycle easily shows that the dt invariant is invariant under deformations of complex structures of @xmath0 .
however in practice , it is more convenient to describe the dt invariant in terms of behrend s constructible function @xcite .
the behrend function is easily described if we use the following result by joyce - song @xcite : [ thm : cs ] for any @xmath97 , there is an analytic open subset @xmath98 , a complex manifold @xmath99 and a holomorphic function @xmath100 such that @xmath101 is isomorphic to @xmath102 . using the above result
, the behrend function @xmath103 on @xmath85 is described as @xmath104 where @xmath105 is the milnor fiber of @xmath106 at @xmath107 .
the function @xmath103 is shown to a be well - defined constructible function on @xmath85 .
if @xmath86 holds , we have the equality @xmath108 in particular if @xmath85 is non - singular and connected , the invariant @xmath109 coincides with @xmath110 up to sign .
based on the above description of the dt invariant , joyce - song @xcite and kontsevich - soibelman @xcite constructed the generalized dt invariant @xmath111 without the condition @xmath86 .
the construction uses the stack theoretic hall algebra @xmath112 of @xmath34 , and its well - definedness is highly non - trivial . a very rough description of it may be @xmath113 the ` log ' is taken in the algebra @xmath112
. some more explanation of a specific case is available in @xcite .
we can define another invariant @xmath114 by formally putting @xmath115 in the definition of @xmath109 . if @xmath86 , it coincides with the usual euler characteristic @xmath110 .
when we say a result as a _ weighted ( resp .
an unweighted ) version _ , it means the result for the invariants @xmath109 ( resp .
@xmath116 ) . in what follows , we identify @xmath117 , @xmath118 with @xmath119 , @xmath30 respectively by the poincar duality . given @xmath16 and @xmath120 , it is easy to show that @xmath121 is isomorphic to @xmath85 for @xmath122 by the assignment @xmath123 .
the resulting invariant @xmath124 is independent of @xmath25 , and it counts one or zero dimensional subschemes @xmath12 with @xmath14=\beta$ ] , @xmath125 . for @xmath16 , the series @xmath126 is defined to be @xmath127 [ exii ] ( i ) if @xmath128 , we have @xcite , @xcite , @xcite @xmath129 \(ii ) if @xmath130 is a birational contraction whose exceptional locus is @xmath131 with normal bundle @xmath132 , we have @xcite @xmath133}(x)t^m= \prod_{k \ge 1 } ( 1-(-q)^k)^{-k \chi(x ) } \prod_{k\ge 1}(1-(-q)^k t)^k . \end{aligned}\ ] ] the above example indicates that the quotient series @xmath134 is the honest curve counting series with homology class @xmath135 .
the following conjecture was proposed by mnop @xcite : [ mnop ] ( i ) the quotient series @xmath136 is the laurent expansion of a rational function of @xmath18 , invariant under @xmath19 .
\(ii ) after the variable change @xmath137 , we have the equality @xmath138 here @xmath139 is the gromov - witten invariant counting stable maps @xmath140 from projective curves @xmath141 with at worst nodal singularities with @xmath142 , @xmath143=\beta$ ] . the variable change @xmath137 makes sense by the rationality conjecture ( i ) .
the above conjecture was first proved for toric calabi - yau 3-folds in @xcite .
as we mentioned in the introduction , the result of theorem [ intro : thm ] is the unweighted version of conjecture [ mnop ] ( i ) .
the weighted version was proved in @xcite .
we have the following result @xcite , @xcite ( unweighted version ) , @xcite ( weighted version ) : [ thm : main ] there exist invariants @xmath144 , @xmath145 satisfying * @xmath146 for any ample divisor @xmath25 on @xmath0 , * @xmath147 , and it is zero for @xmath148 , such that we have the following formula : @xmath149 the proofs for the unweighted version in the author s papers @xcite , @xcite can be modified to show the weighted version , if once a similar result of theorem [ thm : cs ] for the moduli spaces of complexes in @xcite , @xcite is shown to be true ( cf .
this is also applied for the results below .
the rationality conjecture is an easy consequence of theorem [ thm : main ] : conjecture [ mnop ] ( i ) is true .
there exist geometric meanings of @xmath150 and @xmath151 .
the former invariant is nothing but the generalized dt invariant @xmath152 , which counts one or zero dimensional @xmath25-semistable sheaves @xmath153 on @xmath0 with @xmath154=\beta$ ] , @xmath155 .
a priori , @xmath150 is defined using the ample divisor @xmath25 , but the resulting invariant is shown to be independent of @xmath25 .
the latter invariant @xmath151 is more interesting .
it counts certain two term complexes @xmath156 ( indeed they are perverse coherent sheaves in the sense of @xcite , @xcite ) satisfying @xmath157 , which are semistable with respect to a derived self dual weak stability condition on it .
the result of theorem [ thm : main ] is proved along with the idea stated in subsection [ subsec : new ] .
a similar idea also proves pandharipande - thomas conjecture @xcite relating the quotient series of rank one dt invariants with the invariants counting stable pairs .
the definition of stable pairs is given as follows : a _
stable pair _ is data @xmath158 where @xmath153 is a pure one dimensional sheaf on @xmath0 , @xmath159 is a morphism which is surjective in dimension one . a typical example of a stable pair is @xmath160 , where @xmath12 is a smooth curve
, @xmath161 is an effective divisor and @xmath162 is a natural composition @xmath163 . for given @xmath16 and @xmath164 ,
the moduli space @xmath165 of stable pairs @xmath158 with @xmath154=\beta$ ] , @xmath155 is a projective scheme with a symmetric perfect obstruction theory @xcite .
the pt invariant @xmath166 is defined to be the degree of the zero dimensional virtual fundamental cycle @xmath167^{\rm{vir}}$ ] on @xmath165 .
the invariant @xmath168 is deformation invariant , and coincides with the weighted euler characteristic with respect to the behrend function on @xmath165 .
the following conjecture was proposed by @xcite , its unweighted version was proved in @xcite , @xcite , and the weighted version was proved in @xcite : [ thm : dtpt ] for fixed @xmath169 , we have the equality of the generating series @xmath170 finally in @xcite , pandharipande - pixton proved conjecture [ mnop ] ( ii ) for large class of calabi - yau 3-folds including quintic hypersurfaces in @xmath88 : [ thm : pp ] conjecture [ mnop ] ( ii ) is true if @xmath0 is a complete intersection calabi - yau 3-fold in the product of projective spaces . indeed what they proved is the correspondence between gromov - witten invariants and stable pair invariants .
combined with theorem [ thm : dtpt ] , the result of theorem [ thm : pp ] was proved .
their proof relies on the degeneration formula of gw and pt invariants , and the torus localization formula .
the dt theory can be also constructed for non - commutative varieties or algebras .
let @xmath171 be a quasi - projective 3-fold which admits two crepant small resolutions giving a _ flop _ : @xmath172 in this situation , van den bergh @xcite constructed sheaf of non - commutative algebras @xmath173 on @xmath171 and derived equivalences @xmath174 so that their composition gives bridgeland s flop equivalence @xcite . for @xmath120 and @xmath16 , let @xmath175 be the moduli space of surjections @xmath176 in @xmath177 such that @xmath153 has at most one dimensional support and @xmath178=\beta$ ] , @xmath179 .
if @xmath0 is a smooth projective calabi - yau 3-fold , there is a symmetric perfect obstruction theory on @xmath175 , and the degree of its zero dimensional virtual fundamental cycle defines the _ non - commutative dt ( ncdt ) invariant _ @xmath180 .
alternatively , @xmath181 is defined to be the weighted euler characteristic of the behrend function on @xmath175 .
we set @xmath182 to be @xmath183 the following result was proved in @xcite for the unweighted version , and @xcite for the weighted version , basically along with the argument in subsection [ subsec : new ] : [ thm : flop ] we have the following identities : @xmath184 let @xmath185 be the conifold singularity , and take two crepant small resolutions ( [ small ] ) by blowing up at the ideals @xmath186 and @xmath187 . in this case , the algebra @xmath173 is the path algebra of the following quiver @xmath188|{a_2 } \ar@/^1.5pc/[rr]|{a_1 } & & \bullet \ar@/^/[ll]|{b_1 } \ar@/^1.5pc/[ll]|{b_2 } } \end{aligned}\ ] ] with relation given by the derivations of the super potential @xmath189 .
although @xmath0 is not projective in this case , the ncdt invariant @xmath190 } \in \mathbb{z}$ ] makes sense , and coincides with the weighted euler characteristic of the moduli space of framed @xmath173-representations with dimension vector @xmath191 .
the proof of theorem [ thm : flop ] also works in this situation .
using example [ exii ] ( ii ) , the first identity of theorem [ thm : flop ] becomes @xmath192}q^n t^{m } = \prod_{k\ge 1}(1-(-q)^k)^{-2k } \prod_{k\ge 1}(1-(-q)^k t)^k \prod_{k\ge 1}(1-(-q)^k t^{-1})^k.\end{aligned}\ ] ] the above formula was first conjectured by szendri @xcite , and later proved by young @xcite , nagao - nakajima @xcite . in general for a quiver @xmath193 with a super potential @xmath194 , we are able to define the ncdt theory for @xmath195 .
a mutation of the pair @xmath195 defines another quiver with a super potential @xmath196 .
the relationship between ncdt invariants on @xmath195 and @xmath196 is described in terms of cluster transformations .
we refer to @xcite , @xcite for the detail .
we recall the definition of bridgeland stability conditions on a triangulated category @xmath36 .
we fix a finitely generated free abelian group @xmath197 with a norm @xmath198 on @xmath199 together with a group homomorphism @xmath200 .
a typical example is that @xmath43 for a smooth projective variety @xmath0 , @xmath197 is the image of the chern character map @xmath201 , and @xmath202 . by taking the dual of @xmath203
, we always regard a group homomorhpism @xmath204 as a group homomorphism @xmath205 . [
def : bst ] a _ stability condition _ on @xmath36 is data @xmath37 , where @xmath206 is a group homomorphism ( called _ central charge _ ) , @xmath39 is a full subcategory ( called @xmath41-_semistable objects with phase _
@xmath42 ) satisfying the following conditions : * for @xmath207 , we have @xmath208 .
* for all @xmath40 , we have @xmath209 $ ] . * for @xmath210 and @xmath211
, we have @xmath212 . *
( harder - narasimhan property ) : for each @xmath213 , there is a collection of distinguished triangles @xmath214 $ ] , @xmath215 with @xmath216 and @xmath217 .
another way defining a stability condition is to use a t - structure as follows : [ lem : tst ] giving a stability condition on @xmath36 is equivalent to giving data @xmath218 , where @xmath206 is a group homomorphism , @xmath219 is the heart of a bounded t - structure , satisfying @xmath220 together with the harder - narasimhan property : for any @xmath221 , there exists a filtration @xmath222 such that @xmath223 is @xmath224-semistable with @xmath225 for all @xmath226 . here
@xmath221 is @xmath224-_semistable _ if for any subobject @xmath68 , we have @xmath227 . the correspondence is as follows : given @xmath228 , the corresponding heart @xmath229 is the extension closure of @xmath230 for @xmath231 .
conversely given @xmath218 , the category @xmath230 is defined to be the category of @xmath224-semistable objects @xmath221 with @xmath232 .
other @xmath230 are defined by the rule @xmath209 $ ] .
let @xmath141 be a smooth projective curve , @xmath233 , @xmath234 and @xmath235 .
we set @xmath206 to be @xmath236 .
then @xmath237 is a stability condition , whose @xmath224-semistable objects coincide with classical semistable sheaves on @xmath141 . the space of stability conditions is defined as follows : we define @xmath238 to be the set of stability conditions on @xmath36 satisfying the _ support property _
, i.e. there is a constant @xmath239 such that @xmath240 holds for any @xmath241 .
the main result of bridgeland @xcite shows that the set @xmath238 has a structure of a complex manifold . if @xmath43 for a smooth projective variety @xmath0 , @xmath242 and @xmath202 , we set @xmath243 .
let @xmath244 be the group of exact autequivalences of @xmath33 .
the space @xmath44 admits a left action of @xmath244 and a right action of @xmath1 .
the latter action is given by @xmath245 for @xmath246 .
we are interested in the double quotient stack @xmath247 . \end{aligned}\ ] ] the conjecture by bridgeland @xcite is that if @xmath0 is a calabi - yau manifold , the above double quotient stack contains the stringy khler moduli space of @xmath0 , that is the moduli space of complex structures of a mirror manifold of @xmath0 .
\(i ) if @xmath141 is an elliptic curve , then ( [ double ] ) is shown in @xcite to be isomorphic to the modular curve @xmath248 .
this is compatible with the fact that @xmath141 is self mirror .
\(ii ) let @xmath249 be the total space of @xmath250 , which is a non - compact calabi - yau 3-fold . in this case ,
@xmath44 is defined to be @xmath238 where @xmath36 is the bounded derived category of compact supported coherent sheaves on @xmath0 , @xmath197 is the image of @xmath251 in @xmath252 , and @xmath253 . then the quotient stack ( [ double ] )
contains @xmath254 $ ] by @xcite .
the latter stack is the parameter space @xmath255 of the mirror family of @xmath0 given by @xmath256 the space ( [ double ] ) is most interesting for projective calabi - yau 3-folds , e.g. quintic hypersurfaces in @xmath88 . even in the quintic 3-fold case ,
the space ( [ double ] ) is very difficult to study . in this case ,
bridgeland s conjecture @xcite is stated in the following way : [ conj : quintic ] let @xmath257 be a smooth quintic 3-fold , and set @xmath258 $ ] .
then there is an embedding @xmath259 . \end{aligned}\ ] ] the above embedding should be given by the solutions of the picard - fuchs equation which the period integrals of the mirror family of @xmath0 satisfy .
its explicit description is available in @xcite .
however in the projective calabi - yau 3-fold case , it is even not known that whether @xmath44 is non - empty or not .
the first issue in solving conjecture [ conj : quintic ] is to construct stability conditions , which we discuss in the next subsection .
it has been a serious issue to construct bridgeland stability conditions on projective calabi - yau 3-folds .
contrary to the one dimensional case , it turns out that there is no stability condition @xmath45 of the form @xmath260 if @xmath261 . from the arguments in string theory , we expect the following conjecture : [ conj : bw ] let @xmath0 be a smooth projective variety and take @xmath262 with @xmath263 ample class
. then there exists the heart of a bounded t - structure @xmath264 on @xmath33 such that the pair @xmath265 determines a point in @xmath44 , where @xmath266 is given by @xmath267 here @xmath268 is defined to be @xmath269 .
the resulting stability conditions are expected to form a _ neighborhood at the large volume limit _ in terms of string theory .
the above conjecture is known to hold if @xmath270 . in the @xmath271 case ,
the heart @xmath264 is constructed to be a certain tilting of @xmath34 , which we are going to review .
let @xmath0 be a @xmath9-dimensional smooth projective variety and take @xmath262 with @xmath263 ample .
the @xmath263-slope function on @xmath34 is defined to be @xmath272 here @xmath273 if @xmath274 .
an object @xmath62 is @xmath275__-(semi)stable _ _ if for any non - zero subobject @xmath68 , we have @xmath276 .
we define the pair of subcategories @xmath277 of @xmath34 to be @xmath278 here @xmath279 means the extension closure .
the existence of harder - narasimhan filtrations with respect to the @xmath275-stability implies that the pair @xmath277 is a torsion pair ( cf .
@xcite ) of @xmath34 .
its tilting defines another heart @xmath280 , \mathcal{t}_{b , \omega } \rangle \subset d^b \mathrm{coh}(x ) .
\end{aligned}\ ] ] the following result is due to @xcite , @xcite , @xcite . if @xmath271 , then @xmath281 . here is a rough sketch of the proof : if @xmath271 , then @xmath282 is written as @xmath283 the construction of @xmath284 immediately implies @xmath285 for any @xmath286 .
we need to check that @xmath287 implies @xmath288 .
this property can be easily deduced from the the classical bogomolov - gieseker ( bg ) inequality in theorem [ thm : bg ] below .
the following bg inequality played an important role : [ thm : bg ] let @xmath0 be a @xmath9-dimensional smooth projective variety , and @xmath28 a torsion free @xmath275-semistable sheaf on @xmath0 .
then we have the following inequality : @xmath289 suppose that @xmath0 is a smooth projective 3-fold , and @xmath290 , @xmath263 are defined over @xmath30 . in this case , the central charge @xmath266 is written as @xmath291 contrary to the surface case , the heart @xmath284 does not fit into a stability condition with central charge @xmath266 . in @xcite , bayer , macri and
the author constructed a further tilting of @xmath284 in order to give a candidate of @xmath264 in conjecture [ conj : bw ] .
the key observation is the following lemma , which also relies on theorem [ thm : bg ] : for any @xmath286 , one of the following conditions hold : \(i ) @xmath292 .
\(ii ) @xmath293 and @xmath294 .
\(iii ) @xmath295 and @xmath288 .
the above lemma indicates that the vector @xmath296 behaves as if it were @xmath297 on coherent sheaves on surfaces . in @xcite , this observation led to the following slope function on @xmath284 : @xmath298 here @xmath299 if @xmath300 . the above lemma shows that @xmath301 satisfies the weak see - saw property , and it defines a slope stability condition on @xmath284 . in @xcite , it was called _ tilt - stability _ : an object @xmath302 is tilt ( semi)stable if for any subobject @xmath68 , we have @xmath303 .
we can show the existence of harder - narasimhan filtrations with respect to the tilt stability .
similarly to the surface case , the pair of subcategories @xmath304 of @xmath284 defined to be @xmath305 is a torsion pair . by tilting ,
we have another heart @xmath306 , \mathcal{t}_{b , \omega } ' \rangle \subset d^b \mathrm{coh}(x ) . \end{aligned}\ ] ] by the construction , we have @xmath307 for any @xmath308 . in @xcite , we proposed the following conjecture : if @xmath309 , we have @xmath310 .
our double tilting construction led to a bg type inequality conjecture evaluating the third chern characters of tilt semistable objects .
[ conj : bmt ] let @xmath0 be a smooth projective 3-fold .
then for any tilt semistable object @xmath302 with @xmath311 , i.e. @xmath312 , we have the inequality @xmath313 in order to show @xmath314 satisfies the property ( [ z ] ) , it is enough to show the weaker inequality @xmath315 .
if this is true , the existence of hn filtrations is proved in @xcite , while the support property remains open .
the stronger bound in conjecture [ conj : bmt ] was obtained by the requirement that the equality is achieved for tilt semistable objects with zero discriminant .
it seems to be a hard problem to show conjecture [ conj : bmt ] even in concrete examples .
so far , it is proved when @xmath316 by macri @xcite , @xmath0 is a quadric 3-fold by schmidt @xcite , and @xmath0 is a principally polarized abelian 3-fold with picard rank one by maciocia - piyaratne @xcite , @xcite .
another kind of evidence is that assuming conjecture [ conj : bmt ] implies some open problems in other research fields . in @xcite , it was proved that conjecture [ conj : bmt ] implies ( almost ) fujita conjecture for 3-folds : for any polarized 3-fold @xmath317 , @xmath318 is free and @xmath319 is very ample . in @xcite , it was also proved that conjecture [ conj : bmt ] implies a conjectural relationship between two kinds of dt type invariants inspired by string theory .
this result will be reviewed in theorem [ thm : dm ] .
it may be worth pointing out that , in both of the above applications , assuming a weaker inequality , say @xmath315 , does not imply anything .
the stronger evaluation in conjecture [ conj : bmt ] is crucial for the proofs of the applications .
although the existence of bridgeland stability conditions on projective calabi - yau 3-folds remains open , we are able to modify the definition of bridgeland stability conditions so that the story in subsection [ subsec : new ] works .
the notion of _ weak stability conditions _ in @xcite is one of them . in the situation of subsection [ subsec : defi ]
, we further fix a filtration @xmath320 such that each subquotient @xmath321 is a free abelian group . instead of considering a group homomorphism @xmath206
, we consider an element @xmath322 given an element ( [ zweak ] ) , we set @xmath323 for @xmath324 as follows : there is a unique @xmath325 such that @xmath326 , where @xmath327 .
then @xmath328 is defined to be @xmath329 ) \in \mathbb{c}$ ] where @xmath330 $ ] is the class of @xmath52 in @xmath331 . a weak stability condition on @xmath36 with respect to the filtration @xmath332 is data @xmath228 , where @xmath224 is as in ( [ zweak ] )
, @xmath39 is a full subcategory , satisfying the same axiom in definition [ def : bst ] .
similarly to lemma [ lem : tst ] , giving a weak stability condition is equivalent to giving @xmath218 , where @xmath224 is as in ( [ zweak ] ) , @xmath219 is the heart of a bounded t - structure , satisfying the same conditions in lemma [ lem : tst ] .
we denote by @xmath333 the set of weak stability conditions on @xmath36 with respect to @xmath332 with a support property .
this set also has a structure of a complex manifold , and coincides with @xmath238 if @xmath334 , i.e. the filtration @xmath332 is trivial . in general , it is easier to show the non - emptiness for the space @xmath333 with a non - trivial filtration @xmath332 .
the result of theorem [ thm : main ] was obtained by the wall - crossing formula in the space of weak stability conditions on the following triangulated category @xmath335 here @xmath336 is the category of one or zero dimensional sheaves on @xmath0 , and @xmath337 is the triangulated closure .
the relevant data is @xmath338 with the map @xmath203 given by @xmath339 . here @xmath119 is identified with @xmath117 via poincar duality .
the result of theorem [ thm : main ] is proved along with the wall - crossing argument of subsection [ subsec : new ] with respect to the one parameter family of weak stability conditions on @xmath340 @xmath341 from @xmath342 to @xmath343 . here
@xmath263 is an ample divisor on @xmath0 , @xmath344 are given by @xmath345 the heart @xmath346 is obtained as the extension closure of objects @xmath60 and @xmath347 $ ] .
we are able to construct dt type invariant @xmath348 which counts @xmath349-semistable objects @xmath350 with @xmath157 .
it is shown that @xmath351 where @xmath151 is the invariant in theorem [ thm : main ] .
the wall - crossing formula describes the difference between @xmath168 and @xmath151 .
a similar wall - crossing phenomena also implies the relationship between @xmath352 and @xmath168 in theorem [ thm : dtpt ] . combined them ,
we obtain the result of theorem [ thm : main ] .
some more detail is also available in @xcite .
although conjecture [ mnop ] ( i ) is proved , a stronger version of the rationality conjecture remains open .
it was proposed by pandharipande - thomas @xcite , and predicts the product expansion ( called _ gopakumar - vafa form _ ) of the generating series of pt invariants : @xmath353 for some @xmath354 . using theorem [ thm : main ] and theorem [ thm : dtpt ] ,
the above strong rationality conjecture is proved in @xcite to be equivalent to the following conjecture : let @xmath357 $ ] be a homogeneous polynomial of degree @xmath9 . by definition ,
a graded matrix factorization consists of data @xmath358 where @xmath359 are graded free @xmath360-modules of finite rank , @xmath361 are homomorphisms of graded @xmath360-modules , @xmath362 is the shift of the grading , satisfying @xmath363 .
the triangulated category @xmath364 is defined to be the homotopy category of graded matrix factorizations of @xmath194 .
it has a structure of a triangulated category , and related to @xmath33 for the hypersurface @xmath365 .
for instance if @xmath366 , there is an equivalence @xcite @xmath367 as an analogy of gieseker stability on @xmath34 , we expect the existence of a natural stability condition on @xmath364 .
based on the earlier works @xcite , @xcite , the following conjecture is proposed in @xcite : [ conj : gep ] there is a bridgeland stability condition @xmath368 on @xmath364 whose central charge @xmath369 is given by @xmath370 and the set of semistable objects satisfy @xmath371 , where @xmath372 is the graded shift functor @xmath373 .
if @xmath374 , i.e. @xmath0 is a quintic 3-fold , a stability condition above is expected to correspond to the orbifold point @xmath375 in conjecture [ conj : quintic ] called gepner point . by this reason , a stability condition in conjecture [ conj : gep ]
is called _
gepner type_. some evidence of conjecture [ conj : gep ] is available in @xcite , @xcite , @xcite .
suppose that conjecture [ conj : gep ] is true for @xmath374 .
then as an analogy of fan - jarvis - ruan - witten theory @xcite in gw theory , we may define the dt type invariant @xmath376 which counts @xmath377-semistable graded matrix factorizations @xmath378 with @xmath379 . here
@xmath380 is the zero - th hochschild homology of @xmath364 , and @xmath381 is the chern character map on graded matrix factorizations ( cf .
because of the property of @xmath377 , the invariant ( [ dtg ] ) should satisfy @xmath382 . under the orlov equivalence ( [ orlov ] ) , the equivalence @xmath372 on the lhs corresponds to the equivalence @xmath383 on the rhs , where @xmath384 is the seidel - thomas twist @xcite associated to @xmath60 . along with the argument in subsection [ subsec : new ] , the existence of the invariant ( [ dtg ] )
should imply a hidden symmetry of the generating series of the original dt invariants on the quintic hypersurface @xmath385 with respect to the equivalence @xmath383 .
let us recall the original s - duality conjecture by vafa - witten @xcite .
it predicts the ( at least almost ) modularity of the generating series of euler characteristics of moduli spaces of stable torsion free sheaves on algebraic surfaces with a fixed rank and a first chern class .
we refer to @xcite for the developments on the s - duality conjecture so far . instead of stable torsion free sheaves on algebraic surfaces ,
we consider semistable pure two dimensional torsion sheaves on calabi - yau 3-folds , and dt invariants counting them .
let @xmath0 be a smooth projective calabi - yau 3-fold , @xmath25 an ample divisor on @xmath0 and fix a divisor class @xmath386 .
we consider the following generating series @xmath387 here each coefficient counts @xmath25-semistable @xmath62 whose mukai vector ( _ not _ chern character ) satisfies @xmath388 . as a 3-fold version of the s - duality conjecture , we expect that the series ( [ intro : dt ] ) satisfies a modular transformation property of ( almost ) jacobi forms .
( we refer to @xcite for a basic of jacobi forms . )
some computations of the invariants @xmath389 are available in @xcite , @xcite .
also the transformation formula of the series ( [ intro : dt ] ) under a flop is obtained in @xcite .
let us consider a flop diagram ( [ small ] ) with @xmath171 projective , and @xmath263 an ample divisor on @xmath171 .
we assume that the exceptional locus @xmath141 , @xmath390 of @xmath106 , @xmath391 are isomorphic to @xmath392 with @xmath393 .
let @xmath394 be the scheme theoretic length of @xmath395 at the generic point of @xmath141 .
[ thm : intro ] there exist @xmath396 for @xmath397 such that we have the following formula : @xmath398 here @xmath399 is the variable change @xmath400 , @xmath401 is the dedekind eta function and @xmath402 is the jacobi theta function , given as follows : @xmath403 although @xmath404 is not ample , it is shown that the invariants @xmath405 are well - defined .
recall that @xmath401 is a modular form of weight @xmath406 , @xmath402 is a jacobi form of weight @xmath406 and index @xmath406 .
the result of theorem [ thm : intro ] shows that the series ( [ intro : dt ] ) transforms under a flop by a multiplication of a meromorphic jacobi form , which gives evidence of the s - duality conjecture for dt invariants . in string theory ,
the osv conjecture @xcite predicts a certain approximation @xmath407 where the lhs is the partition function of black hole entropy , and the rhs is the partition function of topological string . a version of the above conjecture is mathematically stated as an approximation between the generating series of dt invariants counting torsion sheaves on calabi - yau 3-folds , and the generating series of gw invariants .
in @xcite , denef - moore proposed a relationship among the series ( [ intro : dt ] ) and the generating series of @xmath352 , @xmath168 in order to give a derivation of ( [ osv ] ) .
a mathematical refinement of denef - moore conjecture is stated in @xcite . for simplicity ,
suppose that @xmath408 is generated by an ample divisor @xmath25 . for @xmath409
, we define the following cut off series @xmath410 here @xmath411 . moreover , we define the cut off generating series of d6-anti - d6 brane counting @xmath412 even if conjecture [ dm ] is proved , still the relationship ( [ osv ] ) is not obvious .
if we follow the arguments in @xcite , at least we need to prove s - duality conjecture for dt invariants in the previous subsection and mnop conjecture .
moreover we need to make a mathematical understanding of the approximation @xmath416 in ( [ osv ] ) .
although the relationship ( [ osv ] ) is motivated by string theory , it seems to involve deep and interesting mathematics .
to3em , _ generating functions of stable pair invariants via wall - crossings in derived categories _ , adv . stud . pure math . * 59 * ( 2010 ) , 389434 , new developments in algebraic geometry , integrable systems and mirror symmetry ( rims , kyoto , 2008 ) . | we survey recent developments on donaldson - thomas theory , bridgeland stability conditions and wall - crossing formula .
we emphasize the importance of the counting theory of bridgeland semistable objects in the derived category of coherent sheaves to find a hidden property of the generating series of donaldson - thomas invariants .
primary 14n35 ; secondary 18e30 .
donaldson - thomas invariants , bridgeland stability conditions . |
[ sec : intro ] in their well - known work @xcite kardar , parisi , and zhang argue that surface growth through random ballistic deposition can be modeled by a stochastic continuum equation , which in the case of a one - dimensional substrate reads @xmath0 here @xmath1 is the height at time @xmath2 at location @xmath3 relative to a suitable reference line .
@xmath4 is space - time white noise of strength @xmath5 ,
@xmath6 , and models the randomness in deposition .
@xmath7 is a not further detailed smoothening mechanism .
the important insight of @xcite is to observe that the growth velocity is nonlinear , in general , and is relevant for the large scale properties of the solution to ( [ eq : kpz1d ] ) . to simplify
, the growth velocity is expanded in the slope .
the first two terms can be absorbed through a suitable choice of coordinate frame .
the quadratic nonlinearity in ( [ eq : kpz1d ] ) is relevant and higher orders can be ignored , unless @xmath8 . the one - dimensional kpz equation ( [ eq : kpz1d ] ) is regarded as exactly solved in the usual terminology . in fact , what can be obtained is the dynamic scaling exponent @xmath9 @xcite .
no other universal quantity has been computed exactly so far . in our contribution we will improve the situation and explain how to extract the scaling function for the stationary two - point function .
a few other universal quantities can be computed as well .
but they have been discussed already elsewhere @xcite . in @xcite a mode - coupling equation for the two - point function
is written down , in essence following the scheme from critical dynamics and kinetic theory . at the time
only @xmath9 and a few qualitative properties could be extracted from the mode - coupling equation . in @xcite this equation
is solved numerically .
such computations are repeated in @xcite with greatly improved precision and using a more convenient set of coordinates .
thus for the 1d kpz equation we are in the unique position of an exact solution and an accurate numerical solution to the mode - coupling equation with _ no _ adjustable parameters . as will be explained below , given the uncontrolled approximation , mode - coupling does surprisingly well . to attack ( [ eq : kpz1d ] ) directly
does not seem to be feasible , a situation which is rather similar to the one for two - dimensional models in equilibrium statistical mechanics .
for example , the ginzburg - landau @xmath10-theory is given through the ( formal ) functional measure @xmath11\ ] ] for the scalar field @xmath12 .
( [ eq : phifour ] ) is not the proper starting point for computing the exact two - point scaling function at the critical coupling @xmath13 .
rather one discretizes through the lattice @xmath14 and replaces the @xmath12-field by ising spins @xmath15 .
then , following e.g. @xcite , the scaling function at and close to criticality can be obtained . by universality
this scaling function is the one of ( [ eq : phifour ] ) .
( while certainly true , to establish universality is difficult and carried out in a few cases only @xcite . ) in the same spirit we replace ( [ eq : kpz1d ] ) by a discrete model , where the most convenient choice seems to be the polynuclear growth ( png ) model . before explaining the png model
let us review the standard scaling theory for ( [ eq : kpz1d ] ) .
if the initial conditions @xmath16 of the kpz equation are distributed according to two - sided brownian motion , then formally the distribution of @xmath17 is again two - sided brownian motion .
therefore it is natural to define the stationary time correlation @xmath18 where from the height difference the average displacement is subtracted . by assumption @xmath19 with roughness amplitude @xmath20 to ensure stationarity in time . if @xmath9 , then @xmath21 scales as @xmath22 with a universal scaling function @xmath23 having the asymptotics @xmath24 for @xmath25 and @xmath26 for @xmath27 . in order to define @xmath28 as a dimensionless function we fix the proportionality constants in ( [ eq : generalscalingform ] ) as appropriate combinations of @xmath29 and @xmath30 , @xmath31 where the particular choice of numerical prefactors is in principle arbitrary .
the factor @xmath32 in the denominator is chosen in order to conform with the convention for the gue tracy - widom distribution @xcite .
the factor @xmath33 in the argument of the numerator differs from the convention used by baik and rains @xcite by a factor @xmath34 but conforms with the definition of the closely related airy process @xcite and has the further advantage to absorb a lot of prefactors in the equations defining @xmath23 .
note however , that the exponents for the parameters @xmath35 $ ] , @xmath36 $ ] , and @xmath37 $ ] are fixed uniquely by dimensional reasoning .
we remark that the slope @xmath38 is space - time stationary in the usual sense . for fixed @xmath2
, @xmath39 is white noise with strength @xmath30 .
since @xmath40 , the standard 2-point function is @xmath41 this relation and the asymptotic behavior of @xmath28 , @xmath42 as @xmath43 , motivates the definition of a second scaling function , @xmath44 which by definition has integral normalized to one and which will be shown to be positive in the next section . in the sequel
we will analyze the distribution function for the height differences in the stationary png model .
as shown in @xcite , they can be represented in terms of certain orthogonal polynomials , which lead to recursion relations connected to the painlev ii differential equation @xcite .
the asymptotic analysis is carried out in @xcite .
our own contribution is twofold : ( i ) we observe that the stationary png model maps to a last passage percolation with boundaries @xcite .
( ii ) the expressions in @xcite are given in terms of certain differential equations and the extraction of the scaling function @xmath28 requires a careful numerical integration .
this is one central point of our article .
we will provide then plots of the structure function and give a comparison with the mode - coupling theory .
[ sec : png ] the polynuclear growth ( png ) model is a model for layer - by - layer growth through deposition from the ambient atmosphere .
the surface is parameterized by a time dependent integer - valued height function @xmath1 , @xmath45 , above a one - dimensional substrate , @xmath46 .
thus the height function consists of terraces bordered by steps of unit height .
the up - steps move to the left and the down - steps to the right with speed @xmath47 .
steps disappear upon collision .
in addition to this deterministic dynamical rule new islands of unit height are nucleated randomly with space - time density @xmath34 on top of already existing terraces . the corresponding stochastic process @xmath1 is well defined even in infinite volume ( cf .
@xcite for the closely related hammersley particle process ) . of interest to us here is the stationary growth process , which means that the slope @xmath48 is stationary in space - time .
one can think of @xmath49 as the density of a particle / antiparticle process .
the particles are located at the up - steps and thus move with velocity @xmath50 , the antiparticles are located at the down - steps and move with velocity @xmath47 . upon collision particle / antiparticle pairs annihilate .
in addition , with space - time density @xmath34 , a particle / antiparticle pair is created with the particle moving to the left , the antiparticle to the right . to make @xmath51 stationary
, one prescribes at @xmath52 up - steps poisson distributed with density @xmath53 and down - steps independently poisson distributed with intensity @xmath54 such that @xmath55 this measure for steps is stationary under the png dynamics .
the mean slope is given by @xmath56 which is the only remaining free parameter .
for fixed @xmath2 , @xmath57 is a ( two - sided ) random walk with rate @xmath58 for a jump from @xmath59 to @xmath60 .
it has average @xmath61 and variance @xmath62 , which implies for the roughness amplitude @xmath63 for the growth velocity one obtains @xmath64 given @xmath49 the height @xmath1 is determined only up to a constant which we fix as @xmath65 . to emphasize that only height differences count
, @xmath66 is sometimes kept in the formulas .
the stationary process with slope @xmath61 transforms to the stationary process with slope @xmath67 through the lorentz transformation @xmath68 with the speed of `` light '' equal to @xmath47 and the velocity parameter @xmath69 .
thus it suffices to restrict ourselves to @xmath70 which we do from now on . in particular
@xmath72 and @xmath73 refer to the stationary density field at slope @xmath70 .
the central objects are the height - height correlation @xmath74 and the closely related two - point function for the density , @xmath75 they are related as @xmath76 the height correlation is convex , equivalently @xmath77 to prove this property we show that the structure function @xmath78 can be regarded as the transition probability for a second class particle starting at the origin .
its initial velocity is @xmath15 with probability @xmath79 , as for the `` first - class '' up / down - steps .
in contrast to an ordinary step the second class particle is never destroyed upon colliding with another step .
rather it eats up the step encountered and , by reversing its own direction of motion , continues along the trajectory of the absorbed step , cf .
figure [ fig : secondclass ] .
let @xmath49 be a given realization of the png process .
the second class particle is added as @xmath80 @xmath81 evolves to @xmath82 with nucleation events identical to the one for @xmath49 . by construction , if @xmath83 denotes the position of the second class particle at time @xmath2 , @xmath84 noting that by the poisson property @xmath81 is given by @xmath85 conditioned on the presence of either an up - step ( @xmath86 ) or down - step ( @xmath87 ) at the origin , we obtain @xmath88 & = & \textstyle\frac12{\mathbb{e}}\big(\rho(x , t)\rho(0,0)\big)=\textstyle\frac12s(x , t).\end{aligned}\ ] ] for arbitrary slope @xmath61 the normalization of @xmath78 would be given by @xmath89 and the mean of @xmath90 evolves along the characteristics of the macroscopic evolution equation @xmath91 .
for the png model the distribution function for the height difference @xmath93 satisfies certain recursion relations , which are the tool for analyzing the scaling limit when @xmath94 and @xmath95 with @xmath96 .
the second moments yield @xmath21 and therefore by ( [ eq : twopointdensity ] ) also @xmath78 .
since the nucleation events are poisson , @xmath93 depends only on the events in the backward light cone @xmath97 and the initial conditions at @xmath52 . along the line
@xmath98 the down - steps are poisson distributed with line density @xmath99 and correspondingly for the up - steps along the line @xmath100 .
this property can be deduced from the uniqueness of the stationary state at given slope and the lorentz invariance ( [ eq : lorentztransf ] ) in the limit @xmath101 . thus @xmath93 is determined by the nucleation events in the rectangle latexmath:[$r_{x , t}=\{(x',t')\in{\mathbb{r}}^2,\
, conditions .
@xmath93 can be reexpressed as a directed last passage percolation according to the following rules : inside @xmath103 there are poisson points with density @xmath34 .
along the two lower edges of @xmath103 there are independently poisson points with line density @xmath99 . a directed passage from @xmath104 to @xmath105
is given through a directed path ( polymer ) .
it is a piecewise linear path in the plane , starts at @xmath104 and ends at @xmath105 , alters its direction only at poisson points , and is time - like in the sense that for any two points @xmath106 , @xmath107 on the path one has @xmath108 .
note that , once the directed path leaves one of the lower edges to move into the bulk , it can never return . by definition
the length of a directed path equals the number of poisson points traversed . with these conventions
@xmath109 we remark that in general there are several maximizing paths , their number presumably growing exponentially with @xmath2
. under the lorentz transformation ( [ eq : lorentztransf ] ) the distribution for the height differences ( [ incrsubsequ ] ) does not change .
therefore , we might as well transform @xmath110 to a square . by an additional overall scaling by @xmath99 one arrives at a @xmath111 square , @xmath112 , with bulk density @xmath47 and the line densities @xmath113 for the lower left , resp .
@xmath114 for the lower right edge . in this way we have recovered precisely the setting in @xcite with @xmath2 replaced by @xmath115
baik and rains derive an explicit expression for the height distribution in terms of toeplitz determinants , which can be further simplified by means of corresponding orthogonal polynomials .
let us state the result for the distribution function of @xmath93 , @xmath116 for fixed @xmath115 the functions @xmath28 and @xmath117 are given in terms of the monic polynomials @xmath118 , which are pairwise orthogonal on the unit circle @xmath119 with respect to the weight @xmath120 .
their norm @xmath121 is given by @xmath122 with @xmath123 .
one has @xmath124 where @xmath125 itself is the distribution function of the maximal length of a directed path in the case @xmath126 . thus @xmath127 and @xmath128 defining the dual polynomials @xmath129 , the second equality in ( [ eq : ydef ] ) is an easy consequence of the christoffel - darboux formula @xcite , @xmath130 valid for @xmath131 , @xmath132 and extended by lhospital s rule to @xmath133 , and the trivial relation @xmath134 taking only the leading order of @xmath135 in ( [ eq : cd ] ) one obtains the well - known relations @xmath136
\pi^*_{n+1}(z)&=&z\,p_{n+1}\pi_n(z)+\pi^*_n(z ) , \\[2 mm ] n_{n+1}&=&n_{n}\big(1-p_{n+1}^2\big ) \nonumber\end{aligned}\ ] ] which are closed given @xmath137 for @xmath138 .
for the particular weight function @xmath120 one can derive a nonlinear recursion relation for the @xmath139 s , @xmath140 with initial values @xmath141 , @xmath142 .
@xmath143 is the modified bessel function of order @xmath144 and thus @xmath145 .
( [ eq : piip ] ) is the discrete painlev ii equation .
it has been derived in the context of orthogonal polynomials for the first time in @xcite , and later on more or less independently in @xcite .
the differential equations for @xmath146 , @xmath147 , @xmath148 can be shown to hold by a tedious but straightforward induction , using ( [ eq : orthpolrecurs ] ) and ( [ eq : piip ] ) .
they are implicitly derived in @xcite , from which we learned their actual form . in @xcite
an integral expression is obtained for the derivative of orthogonal polynomials on the circle with respect to ( up to some technical conditions ) an arbitrary weight function .
specializing to the weight @xmath120 results in a differential - difference equation equivalent to ( [ eq : diffeq ] ) .
of course , the mean of the probability distribution @xmath149 is @xmath150 and its variance , the correlation function ( [ eq : pngcorr ] ) , is given by @xmath151 thus to establish ( [ eq : generalscalingform ] ) , one has to understand the scaling properties of the distribution function @xmath152 .
let us introduce the new variables @xmath153 , @xmath154 defined by @xmath155 where @xmath112 is regarded as fixed when varying @xmath59 and @xmath3 . in @xcite
the different scaling variable @xmath156 is used , which leads to a string of factors of @xmath34 , avoided by our convention . setting @xmath157 we rewrite ( [ eq : piip ] ) as @xmath158 under the scaling ( [ eq : scaledvars2 ] ) , @xmath159 ,
it becomes the painlev ii equation @xmath160 in the limit @xmath161 .
the starting value @xmath162 is consistent with the left asymptotics of @xmath163 only if @xmath164 which singles out the hastings - mcleod solution to ( [ eq : pii ] ) @xcite .
this particular solution will be denoted by @xmath163 and we conclude that @xmath165},\ ] ] provided the limit exists ( a complete proof is the main content of @xcite ) . @xmath166 and @xmath61 has the right asymptotics @xmath167 at this point we can derive the scaling limit for @xmath125 .
it is the gue tracy - widom distribution function @xcite @xmath168 which appears already as the limiting height distribution for nonstationary curved selfsimilar growth @xcite . since @xmath169 , one has @xmath170}\to v(s)$ ] and @xmath171)\to f_{\text{gue}}(s)$ ] as @xmath161 .
next we turn to the scaling limit for the orthogonal polynomials . for it to be nontrivial we set @xmath172 ( [ eq : scaledvars2 ] ) implies @xmath173 . setting @xmath59 as in ( [ eq : scaledvars1 ] ) , we claim that @xmath174 if so , the limit functions @xmath135 , @xmath175 satisfy the differential equations @xmath176 as a consequence of ( [ eq : orthpolrecurs ] ) , and @xmath177 as a consequence of ( [ eq : diffeq ] ) . from ( [ eq : orthpolrecurs ] ) one immediately obtains @xmath178 .
one has the limit @xmath179 since @xmath180 has an interpretation as a probability distribution function @xcite .
therefore the initial conditions to ( [ eq : aby ] ) are @xmath181 the scaling limit of @xmath182 as defined in eq .
( [ eq : ydef ] ) is the function @xmath183 where the second equality can be verified by differentiation with respect to @xmath153 and using the identity @xmath184 itself being a direct consequence of ( [ eq : aby ] ) and ( [ eq : abinit ] ) . putting these pieces together we obtain as scaling limit for the distribution functions @xmath152
, @xmath185 the shift in ( [ eq : fydef ] ) by @xmath186 comes from the fact that @xmath187 is evaluated for constant @xmath188 .
in conclusion we arrive at the scaling function @xmath23 as defined in the introduction . from ( [ eq : dimensionlessscalingform ] ) , with @xmath189 and @xmath190 for the png model , and ( [ eq : pngcorr2 ] ) we obtain @xmath191 as already mentioned , except for ( [ eq : diffeq ] ) , all our relations are derived in @xcite , and the existence of limits is proven with riemann - hilbert techniques . for completeness
let us collect some more properties of @xmath135 shown in @xcite : @xmath192 therefore @xmath187 is asymptotically gaussian and we recover @xmath193 for large @xmath154 .
[ sec : num ] the key object in determining the scaling functions @xmath23 , @xmath194 is the hastings - mcleod solution @xcite to painlev ii , @xmath163 , which is the unique solution to @xmath195 with asymptotic boundary conditions ( [ eq : uleftas ] ) and ( [ eq : rightasympt ] ) .
tracy and widom @xcite integrate ( [ eq : piiagain ] ) numerically with conventional differential equation solvers using the known asymptotics at @xmath196 .
the precision achieved with this technique does not suffice for our purposes , since we need @xmath163 as starting values ( [ eq : abinit ] ) for the differential equations ( [ eq : aby ] ) .
we develop here a different method to obtain @xmath163 , in principle with arbitrary precision . next the functions @xmath197 and @xmath198 have to be determined , which directly leads to values for the distribution functions @xmath187 .
they have to be further integrated with respect to @xmath153 in order to obtain their variance , which is the desired scaling function @xmath23 .
the taylor expansion method to be explained intrinsically produces not only function values at a point but also higher derivatives .
therefore we obtain @xmath194 not by numerically differentiating @xmath23 but rather by direct calculation via the knowledge of @xmath199 . in a first step , to obtain reliable approximations to the hastings - mcleod solution , we need to guess its initial data at some finite @xmath200 by using asymptotic expansions around @xmath201 .
any initial data @xmath202 , @xmath203 give rise to a maximal solution , @xmath204 of ( [ eq : piiagain ] ) , which admits analytic continuation to a meromorphic function on @xmath205 .
the only essential singularity for painlev ii solutions lies at @xmath206 .
if @xmath204 is close to @xmath163 we can estimate the difference @xmath207 by linearizing ( [ eq : piiagain ] ) around @xmath163 .
one obtains that @xmath208 is of order @xmath209 if @xmath210 is in the unstable subspace and of order @xmath211 for the stable subspace , where @xmath212 for @xmath213 and @xmath214 for @xmath215 .
note that on the exponential scale we are looking at , derivation with respect to @xmath153 leaves invariant the order .
therefore the exponential orders of @xmath216 and @xmath217 are the same . since generically initial values at @xmath200 have a component in both subspaces one obtains that @xmath218 is of order @xmath219 in the range of validity of the linear approximation , @xmath220 .
it turns out that the left asymptotic expansion in @xmath221 , optimally truncated at large negative @xmath200 , gives rise to initial values with @xmath222 only of order @xmath223 .
thus control of the approximation always breaks down near @xmath224 .
approximations of the right asymptotics on the other hand allow a , in principle , arbitrary precision on any given finite interval .
for @xmath225 the deviations of @xmath163 from the airy function can be expanded in an alternating asymptotic power series with exponentially small prefactor , @xmath226 the coefficients are @xmath227 , @xmath228 , @xmath229 , @xmath230 , and can be obtained via the recursion relation @xmath231 with initial conditions @xmath232 .
@xmath233 are the coefficients in the asymptotic expansion of @xmath234 and @xmath235 are the coefficients of the asymptotic expansion of the airy function itself @xcite , @xmath236 empirically we observe that for @xmath237 the optimal truncation in ( [ eq : rightapproximant ] ) is @xmath238 leading to an exponentially improved ( relative ) precision @xmath239 the linear perturbation argument for @xmath204 with initial values @xmath240 , @xmath241 , @xmath242 $ ] , is now valid for a slightly smaller interval than @xmath243 $ ] .
for example , by choosing the interval @xmath244 $ ] and gluing @xmath204 at the boundaries smoothly to the optimally truncated asymptotic expansions , the maximal relative error of @xmath204 with respect to the hastings - mcleod solution is of order @xmath245 . for our purpose
it turns out that we do not need values with @xmath246 . on the other hand , to access large values of @xmath154 in ( [ eq : fydef ] ) , because of the shift in ( [ eq : fydef ] ) , we need @xmath163 for large @xmath153 with high precision .
@xmath247 is numerically costly to evaluate for large @xmath153 , so we finally choose @xmath248 and integrate in the interval @xmath249 $ ] .
this requires a maximal working precision of @xmath250 digits and , given the integration of ( [ eq : piiagain ] ) is precise enough , @xmath204 and @xmath163 coincide in the first @xmath251 digits for @xmath252 $ ] and still up to @xmath253 digits at @xmath254 , where @xmath255 . in the sequel we drop the distinction between @xmath163 and its numerical approximation .
the arithmetic computing is done partially with mathematica@xmath256 and for the computationally intensive tasks with the c++-based multiprecision package mpfun++ @xcite . to solve initial value problems for ordinary differential equations highly sophisticated iteration schemes are available , like runge - kutta , adams - bashford and multi - step methods . for arbitrary high ( but fixed ) precision results ,
all these methods become ineffective , since the step size is a decreasing function of the required precision goal for the solution and tends to become ineffectively small .
the only remaining choice is to taylor expand the solution at a given point .
the step size is limited by the radius of convergence only and the precision is controlled by the error made in truncating the taylor series at some order @xcite .
@xmath163 is expanded at @xmath200 as @xmath257 for the painlev ii equation the expansion coefficients @xmath258 at @xmath200 are determined by @xmath259 , @xmath260 and @xmath261 where @xmath262 are the expansion coefficients of @xmath263 at @xmath200 , @xmath264 .
we include the factorial into the expansion coefficients instead of taking the bare taylor coefficients , in order to reduce the workload from multiplications by binomials when multiplying two expansions numerically .
numerically we find that the hastings - mcleod solution does not have any pole in a strip @xmath265 . to have a safety margin we choose a step size one for the extrapolation of the expansion ( [ eq : piiexpansion ] ) .
we take the starting values @xmath266 , @xmath267 from ( [ eq : rightapproximant ] ) at @xmath248 .
the coefficients of the functions @xmath268 , @xmath269 , see ( [ eq : fgue ] ) and ( [ eq : abinit ] ) , when expanded around @xmath200 , as in ( [ eq : piiexpansion ] ) , are given by @xmath270 and @xmath271 , leaving unspecified the yet unknown integration constants @xmath272 and @xmath273 .
the higher expansion coefficients of @xmath61 , @xmath274 , @xmath275 , and @xmath276 are independent of the values for @xmath277 and @xmath278 and are calculated with the recursion relations ( [ eq : piitaylorrecursion ] ) and ( [ eq : uandvexpansion ] ) .
the values of @xmath61 , @xmath274 , @xmath275 , and @xmath276 at @xmath279 are extrapolated , the expansion coefficients at @xmath280 iterated
. then values are calculated at @xmath281 by extrapolation , and so on .
a posteriori we assign to @xmath282 and @xmath283 values , such that @xmath284 .
the numerical errors from iterating ( [ eq : piitaylorrecursion ] ) and from truncating ( [ eq : piiexpansion ] ) can be neglected compared to the uncertainty originating from the initial conditions . at the end of this first step we have at our disposal the expansion coefficients for @xmath285 at the integers in the interval
@xmath249 $ ] . for
the convenience of the interested reader let us just state the values at @xmath224 up to @xmath253 digits , @xmath286 which might be used as starting values for a quick conventional integration of painlev ii to reproduce parts of our results with much less effort but also less precision .
tables can be found at @xcite .
the next step is to determine @xmath197 , @xmath198 at @xmath287 in the interval @xmath288 $ ] employing ( [ eq : aby ] ) and ( [ eq : abinit ] ) . setting @xmath289 ( [ eq : aby ] )
becomes a recursion relation for the expansion coefficients , @xmath290 @xmath138 , allowing one to determine @xmath291 , @xmath292 , @xmath138 upon the knowledge of @xmath293 , @xmath294 .
we integrate along @xmath295 with an extrapolation step size of @xmath296 . from ( [ eq : abs ] ) one obtains the recursions @xmath297 the expansion coefficients @xmath298 of @xmath299 at @xmath300 , are determined from ( [ eq : g ] ) as @xmath301 where @xmath302 , @xmath303 are the corresponding expansion coefficients of @xmath135 and @xmath175 at @xmath304 . to finally determine @xmath23 and its derivatives we write @xmath305 @xmath306 are the expansion coefficients of @xmath307 at @xmath300 , @xmath308 here @xmath309 are the expansion coefficients of @xmath310 and @xmath311 are the expansion coefficients of @xmath312 . numerically the sum over @xmath200 in ( [ eq : gsplit ] ) is truncated to values inside @xmath313 $ ] , since outside contributions turn out to be negligible at the chosen precision goal . after accomplishing this program
we keep values for @xmath23 at @xmath314 $ ] and for @xmath315 , @xmath316 , at @xmath317 $ ] with an accuracy of about @xmath318 digits ( a table in ascii format is available online at @xcite ) . for interpolating these values
we used the interpolation - function of the mathematica@xmath256 package yielding best results due to the high precision data with an interpolation order of @xmath319 .
there have been numerous attempts to approximately determine @xmath320 @xcite . for historical reasons a different scaling function , @xmath321 , is analyzed in some of these works .
the relation to our scaling function @xmath320 is @xmath322 note that by ( [ eq : dimensionlessscalingform ] ) the large @xmath154 behavior of @xmath28 is fixed by definition as @xmath323 .
the special value @xmath324 is the baik - rains constant @xcite . in the literature
the universal amplitude ratio @xmath325 and the universal coupling constant @xmath326 have been investigated .
approximate values have been determined by means of monte - carlo simulations for the single step model @xcite , numerically within a mode - coupling approximation @xcite , and even experimentally for slowly combusting paper @xcite yielding estimates for @xmath327 within reasonable ranges around the ( numerically ) exact value indicated . in figure
[ fig : fy ] the scaling function @xmath328 is versus @xmath154 in a semilogarithmic plot . the dotted line @xmath329 is drawn as a guide to the eye for the large @xmath154 asymptotics of @xmath330.__,width=377 ] shown as determined by the multiprecision expansion method explained in the previous section .
we estimate its large @xmath154 asymptotics as @xmath331 the cubic behavior is very robust and numerical fits yield about @xmath332@xmath333 quite independently of the assumed nature of the finite size corrections .
the prefactor @xmath334 has a relatively high uncertainty because of the unknown subleading corrections .
even though unaccessible in nature we estimate the error term , as indicated in ( [ eq : fasympt ] ) , to be sublinear or even only logarithmic from the numerical data .
possibly , the exact asymptotic behavior could be extracted from a refined asymptotic analysis of the riemann - hilbert problem .
colaiori and moore @xcite tackled the same scaling function by completely different means .
starting from the continuum version of the kpz equation they numerically solved the corresponding mode - coupling equation @xcite , which contains an uncontrolled approximation , since diagrams which would renormalize the three - point vertex coupling are neglected .
nevertheless a qualitative comparison of their result with the exact scaling function @xmath194 shows reasonable similarity , cf .
figure [ fig : fvsmc ] .
both functions are normalized to integral @xmath47 by definition .
the mode compared to the mode coupling result of colaiori and moore @xcite(dotted line ) .
both functions are even.__,width=377 ] coupling solution oscillates around @xmath67 for @xmath335 , whereas @xmath336 for the exact solution .
we do not know whether this is a numerical artifact or an inherent property of the mode - coupling equation .
on the other hand , the second moments are reasonably close together , @xmath337 for @xmath194 , and @xmath338 for the mode - coupling approximation . so is the value of the baik - rains constant @xmath339 for which mode - coupling predicts the value @xmath340 . from the solution to the mode - coupling equations one
does not directly obtain @xmath194 , but rather its fourier transform .
the function @xmath341 from @xcite is defined through @xmath342 moore and colaiori predict a stretched exponential decay of @xmath341 as @xmath343 @xcite and numerically find a superimposed oscillatory behavior on the scale @xmath344 @xcite . in figure
[ fig : fk ] of the scaling function @xmath194.__,width=377 ] on a semilogarithmic scale .
the dotted line is a heuristic fit , shifted by a factor @xmath345 for visibility.__,width=377 ] @xmath346 is plotted as obtained by a numerical fourier transform of @xmath194 .
indeed it exhibits an oscillatory behavior as can be seen in figure [ fig : fklog ] where the modulus of @xmath346 is shown on a semilogarithmic scale .
the dotted line in the plot is the modulus of the function @xmath347 shifted by a factor of @xmath345 for visibility , which fits @xmath346 very well in phase and amplitude for @xmath348 .
this behavior is not in accordance with the results of colaiori and moore , since the oscillations and the exponential decay of @xmath341 for the exact solution are apparently on the scale @xmath349 and not @xmath350 .
note that @xmath346 is the scaling function for the intermediate structure function @xmath351 by fourier transforming with respect to @xmath2 we determine the dynamical structure function , @xmath352 where @xmath353 and @xmath354 has the convenient representation @xmath355 the correlation function ( [ eq : pngcorr ] ) in fourier space is given by @xmath356 describing the asymptotic behavior at @xmath357 .
note that @xmath358 by definition , since @xmath359 for @xmath360 .
the anomalous scaling behavior in real space is reflected by the exponents for the divergence of @xmath361 at @xmath362 . in the linear case , the edwards - wilkinson equation @xmath8 in ( [ eq : kpz1d ] ) , one easily obtains @xmath363 a 3d - plot of @xmath361 is shown in figure [ fig : ckw ] . in fourier space.__,width=453 ] its striking features are the smooth behavior away from @xmath357 , especially on the lines where @xmath364 and @xmath365 and the two symmetric maxima of @xmath366 for constant @xmath367 .
our numerical data yield for the singular behavior at @xmath364 , @xmath365 , @xmath368
for systems close to equilibrium many properties valid in generality rely on detailed balance , amongst them in particular the link between correlation and response functions .
the kpz equation does not satisfy detailed balance , since the growth is directed .
however , it has been speculated that in @xmath369 dimensions detailed balance is recovered in the scaling regime . with our exact scaling function at hand
, such a claim can be tested .
detailed balance implies that the eigenvalues of the generator in the master equation lie on the negative real axis .
thus autocorrelations in the form @xmath370 can be written as the laplace transform of a positive measure .
the structure function @xmath371 at fixed @xmath144 is such an autocorrelation .
using the scaling form ( [ eq : structure ] ) detailed balance would imply @xmath372 with @xmath373 .
in particular @xmath374 . from ( [ eq : fittofhat ] ) we know that @xmath371 oscillates around zero . definitely , at @xmath375 there is a negative dip , cf .
[ fig : fk ] .
thus ( [ eq : detbal ] ) can not be correct .
the bethe ansatz @xcite indicates that , for large system size , the density of states is concentrated on an arc touching @xmath67 .
if so , the integration in ( [ eq : detbal ] ) would have to be replaced by a corresponding line integral .
it is not clear to us how to extract from the numerical knowledge of @xmath371 such a representation .
our main result is the exact scaling function @xmath330 , see figure [ fig : fvsmc ] , for the two - point function of the stationary kpz equation in @xmath369 dimensions .
`` exact '' must be qualified in two respects .
firstly @xmath330 is given indirectly through the solution of certain differential equations , which can be solved numerically only with considerable effort .
the errors are well controlled , however .
secondly , we rely on universality , in the sense that the scaling function is derived for the png model , which is one rather particular model within the kpz universality class . of course
, it would be most welcome to establish the scaling limit also for other models in this class .
the kpz equation ( [ eq : kpz1d ] ) is a two - dimensional field theory and , in spirit , belongs to the same family as two - dimensional models of equilibrium statistical mechanics , one - dimensional quantum spin chains , and other @xmath376-dimensional quantum field theories at zero temperature . while in the latter cases
, there are a number of models for which the two - point function can be computed , in the dynamical context such solutions are scarce . in addition
, the kpz equation does not satisfy the condition of detailed balance .
such nonreversible models are known to be difficult and we believe that the png model is the first one in the list of exact solutions , disregarding noninteracting field theories . for the nonstationary kpz equation with a macroscopic profile of nonzero curvature
the analogue of @xmath377 is the tracy - widom distribution function . in that case
the full statistics of @xmath378 for large but fixed @xmath2 is available @xcite .
it is conceivable that an extension of the techniques used there also admits a more detailed study of , say , the joint distribution of @xmath93 , @xmath379 . on the other hand the joint distribution of @xmath380
, @xmath381 does not seem to be accessible . in the representation through the directed polymers it means that space - like points , even several of them , can be handled , whereas time - like points remain a challenge .
+ * acknowledgments . *
we greatly enjoyed the collaboration with jinho baik at the early stage of this project and are most grateful for his important input .
we also thank mike moore for providing us with the numerical solution of the mode - coupling equation .
m. prhofer and h. spohn .
current fluctuations for the totally asymmetric simple exclusion process . in sidoravicius v. ,
editor , _ in and out of equilibrium _ , volume 51 of _ progress in probability _
, pages 185204 .
birkhuser boston , 2002 .
d. barton , i. m. willers , and r. v. m. zahar .
taylor series methods for ordinary differential equations an evaluation . in john
rice , editor , _ mathematical software _ , pages 369390 . academic press , new york , 1971 . | we determine the stationary two - point correlation function of the one - dimensional kpz equation through the scaling limit of a solvable microscopic model , the polynuclear growth model .
the equivalence to a directed polymer problem with specific boundary conditions allows one to express the corresponding scaling function in terms of the solution to a riemann - hilbert problem related to the painlev ii equation .
we solve these equations numerically with very high precision and compare our , up to numerical rounding exact , result with the prediction of colaiori and moore @xcite obtained from the mode coupling approximation . |
let @xmath7\rightarrow\lbrack0,1]$ ] and @xmath8\rightarrow \lbrack0,1]$ ] be continuous and differentiable , and such that @xmath9 , @xmath10 @xmath11 let the derivatives @xmath12 @xmath4 be uniformly bounded below by @xmath13 .
fig1.eps for @xmath14 $ ] we define @xmath0\rightarrow\lbrack0,1]$ ] by @xmath15\ni x\mapsto\left\ { \begin{array } [ c]{c}w_{0}(x)\text { if } x\in\lbrack0,\rho]\\ w_{1}(x)\text { otherwise}. \end{array } \right.\ ] ] see figure [ figure1 ] . similarly , we define @xmath16\rightarrow \lbrack0,1]$ ] by replacing @xmath17 $ ] by @xmath18
. let @xmath19 .
let @xmath20 have the product topology induced from the discrete topology on @xmath21 . for @xmath22 write @xmath23 where @xmath24 for all @xmath25 .
the product topology on @xmath26 is the same as the topology induced by the metric @xmath27 where @xmath28 is the least index such that @xmath29 .
it is well known that @xmath30 is a compact metric space .
we define a total order relation @xmath31 on @xmath32 and on @xmath33 for any @xmath34 , by @xmath35 if @xmath36 and @xmath37 where @xmath28 is the least index such that @xmath38 . for @xmath39 and @xmath34
we write @xmath40 .
@xmath26 is the appropriate space in which to embed and study the itineraries of the family of discontinuous dynamical systems @xmath0\rightarrow\lbrack0,1]$ ]
. for @xmath41 let @xmath42 denote @xmath43 composed with itself @xmath28 times , for @xmath25 , and let @xmath44 .
we define a map @xmath45\rightarrow i^{\infty}$ ] , using all of the orbits of @xmath46 by @xmath47 where @xmath48 equals @xmath49 or @xmath50 according as @xmath51,$ ] or @xmath52,$ ] respectively .
we call @xmath53 the _ itinerary _ of @xmath54 under @xmath5 , or an _ address _ of @xmath54 , and we call @xmath55)$ ] an _ address space _ for @xmath3 $ ] .
similarly , we define @xmath56\rightarrow i^{\infty}$ ] so that @xmath57 equals @xmath58 , or @xmath50 according as @xmath59 or @xmath60,$ ] respectively ; and we define @xmath61)$ ] .
note that @xmath5 , @xmath62 , @xmath63 @xmath64 @xmath65 , and @xmath66 all depend on @xmath67 .
the main goals of this paper are to characterise @xmath6 and to show that there exists a value of @xmath67 such that @xmath6 is symmetric .
[ irtheorem ] @xmath6 is the maximal attractor of the iterated closed relation @xmath68 defined by@xmath69 where @xmath70 and @xmath71 .
the chain recurrent set of @xmath72 is @xmath73 where @xmath74 is a transitive attractor for @xmath72 , with basin of attraction @xmath75 and dual repeller @xmath76 .
we write @xmath77 to denote the closure of a set @xmath78 .
but we write @xmath79 . for @xmath80
we write @xmath81 to mean @xmath82 and @xmath83 .
define a symmetry function @xmath84 by @xmath85 where @xmath86 for all @xmath28 .
[ symmetrythm ] there exists a unique @xmath14 $ ] such that @xmath87 .
theorem [ irtheorem ] tells us that @xmath6 is fixed by itineraries of two inverse images of the critical point @xmath67 , and provides the basis for a stable algorithm to determine @xmath6 .
it relates the address spaces of dynamical systems of the form of @xmath5 to the beautiful theory of iterated closed relations on compact hausdorff spaces @xcite , and hence to the work of charles conley .
theorem [ symmetrythm ] is interesting in its own right and also because it has applications in digital imaging , as explained and demonstrated , in the special case of affine maps , in @xcite .
it enables the construction of parameterized families of nondifferentiable homeomorphisms on @xmath3 $ ] , using pairs of overlapping iterated function systems , see proposition [ proposition7 ] .
theorem [ symmetrythm ] generalizes results in @xcite to nonlinear @xmath1 s .
the proof uses symbolic dynamics in place of the geometrical construction outlined in @xcite . the approach and
results open up the mathematics underlying @xcite and @xcite . to tie the present work into @xcite , note that @xmath65 is a section , as defined in @xcite , for the hyperbolic iterated function system @xmath88;w_{0}^{-1},w_{1}^{-1})\text{.}\ ] ] our observations interrelate to , but are more specialized than , the work of parry @xcite .
our point of view is topological rather than measure - theoretic , and our main results appear to be new .
the following list of properties is relatively easy to check . below the list we elaborate on points 1 , 2 , and 3 . 1 .
@xmath89 is piecewise differentiable and its derivative is uniformly bounded below by @xmath90 ; each , except the leftmost branch of @xmath89 , is defined on an interval of the form @xmath91.$ ] @xmath92 is piecewise differentiable and its derivative is uniformly bounded below by @xmath90 ; each , except for the rightmost branch of @xmath93 is defined on an interval of the form @xmath94 2 .
if @xmath95 is the interior of the definition domain of a branch of @xmath89 ( and of @xmath96 then @xmath97 is constant on @xmath91 $ ] , @xmath98 is constant on @xmath99 , and @xmath100 for all @xmath101 .
3 . the boundary of the definition domain of a branch of @xmath89 is contained in @xmath102 ; by ( 1 ) , the length of such a domain is at most @xmath103 .
the set @xmath104 is dense in @xmath3.$ ] this follows from ( 3 ) .
@xmath105 unless @xmath106 in which case @xmath107
both @xmath53 and @xmath108 are strictly increasing functions of @xmath109 $ ] and @xmath110 .
this follows from ( 4 ) and ( 5 ) .
7 . for all @xmath109,$ ]
@xmath53 is continuous from the left , @xmath108 is continuous from the right .
moreover , for all @xmath111@xmath112 these assertions follow from ( 2 ) , ( 3 ) and ( 4 ) . 8 .
each @xmath113 such that @xmath97 is constant , moves continuously with respect to @xmath67 with positive velocity bounded above by @xmath114 this follows from ( 1 ) .
9 . for @xmath115 @xmath100 is locally constant with respect to @xmath116 moreover , this holds if @xmath54 depends continuously on @xmath67 .
this follows from ( 2 ) , ( 3 ) and ( 6 ) .
the symmetry function @xmath84 is strictly decreasing and continuous .
11 . for any @xmath117 @xmath34 , the set @xmath118:\tau(x)|_{n}=\sigma|_{n}\text { or } \tau^{+}(x)|_{n}=\sigma|_{n}\},\ ]
] is either empty or a non - degenerate compact interval of length at most @xmath114 this follows from ( 2 ) , ( 3 ) and ( 6 ) .
the projection @xmath119 $ ] is well - defined by@xmath120:\tau^{+}(x)\preceq\sigma\}=\inf \{x\in\lbrack0,1]:\tau(x)\succeq\sigma\}\text{.}\ ] ] this follows from ( 6 ) .
the projection @xmath119 $ ] is increasing , by ( 6 ) ; continuous , by ( 11 ) ; and , by ( 7),@xmath121,\\ \tau(\hat{\pi}(\sigma ) ) & \preceq\sigma\preceq\tau^{+}(\hat{\pi}(\sigma))\text { for all } \sigma\in i^{\infty}.\end{aligned}\ ] ] 14 .
let @xmath122 denote the left - shift map @xmath123 . for all @xmath39 such that @xmath124 or @xmath125@xmath126 also @xmath127 and @xmath128 .
these statements follow from ( 7 ) . here
we elaborate on points ( 1 ) , ( 2 ) and ( 3 ) .
consider the piecewise continuous function @xmath129 , for @xmath130 .
its discontinuities are at @xmath67 and , for @xmath131 , other points in @xmath132 , each of which can be written in the form @xmath133 for some @xmath134 for some @xmath135 .
we denote these discontinuities , together with the points @xmath58 and @xmath136 , by @xmath137 where @xmath138 . for each @xmath139 one of the @xmath140 s
is equal to @xmath141 for @xmath142 we have @xmath143 for @xmath144 $ ] and @xmath145 for @xmath146 .
similarly @xmath147 for all @xmath148 $ ] and @xmath149 for @xmath150 $ ] . for all @xmath151 @xmath152 , @xmath153 @xmath154 for some fixed @xmath155 .
we refer to @xmath156 as the _ address of the interval _
@xmath157 we say @xmath158 that `` _ _ has address _
_ @xmath156 '' , and we write , by slight abuse of notation , @xmath159 let @xmath160 consider two adjacent intervals,@xmath161 $ ] and @xmath162 $ ] for @xmath163 and @xmath131 .
let the one on the right have address @xmath164 and the one on the left have address @xmath165 @xmath166 .
then @xmath165 @xmath167 and we have@xmath168\text{,}\\ \tau^{+}(x)|_{k-1 } & = \eta_{0}\eta_{1} ... \eta_{k-1}\text { for all } x\in\lbrack d_{k , m-1},d_{k , m}),\\ \tau(x)|_{k-1 } & = \theta_{0}\theta_{1} ... \theta_{k-1}\text { for all } x\in(d_{k , m},d_{k , m+1}],\\ \tau^{+}(x)|_{k-1 } & = \theta_{0}\theta_{1} ... \theta_{k-1}\text { for all } x\in\lbrack d_{k , m},d_{k , m+1}).\end{aligned}\ ] ] in particular , @xmath169 and @xmath170 are constant and equal on each of the open intervals @xmath171 and have distinct values at the discontinuity points @xmath172 .
in this section we characterize @xmath173 and @xmath174 as certain inverse limits , and we characterize @xmath6 as an attractor of an iterated closed relation on @xmath26 .
these inverse limits are natural and they clarify the structures of @xmath173 and @xmath175 they are implied by the shift invariance of @xmath173 and @xmath174 . recall that @xmath176 denotes the left - shift map @xmath177 .
[ shiftprop](i ) @xmath178 and @xmath179 for all @xmath109 $ ] .
\(ii ) @xmath180 and @xmath181 \(i ) this follows at once from the definitions of @xmath65 and @xmath182 ( ii ) this follows from ( i ) together with @xmath183)=w_{+}([0,1])=[0,1].$ ] we say that @xmath184 is _ closed from the left _ if , whenever @xmath185 is an non - decreasing sequence of points in @xmath186 , @xmath187 .
we say that @xmath184 is _ closed from the right _ if , whenever @xmath185 is non - increasing sequence in @xmath186 , @xmath187 . for @xmath188 where @xmath189 or @xmath3,$ ] we write @xmath190 _ there is _ _ a non - decreasing sequence _ @xmath191 _ with _ @xmath192to denote the closure of @xmath193 from the left .
analogously , we define @xmath194 for the closure of @xmath193 from the right .
[ closedlemma ] ( i ) @xmath173 is closed from the left and @xmath174 is closed from the right ; \(ii ) @xmath195 proof of ( i ) : by ( 6 ) @xmath45\rightarrow i^{\infty}$ ] is monotone strictly increasing . by ( 7 ) @xmath65 is continuous from the left .
let @xmath196 be a non - decreasing sequence of points in @xmath197 let @xmath198 .
let @xmath199 $ ] .
since @xmath65 is continuous from the left , @xmath200 .
it follows that @xmath173 is closed from the left .
similarly , @xmath174 is closed from the right .
proof of ( ii ) : let @xmath201:\tau(x)=\tau^{+}(x)\}$ ] .
then by ( 4 ) @xmath202.$ ] also , by ( 5 ) , @xmath203)\cap\tau^{+}([0,1])=\tau(q)=\tau^{+}(q).\ ] ] hence @xmath204 finally , @xmath205 @xmath206 .
we define @xmath207 by @xmath208 @xmath209 note that both @xmath210 and @xmath211 , are contractions with contractivity @xmath212 .
we write @xmath213 to denote the set of all subsets of @xmath214 for @xmath215 we define@xmath216 & : = \{\zeta\in i^{\infty}:\sigma\preceq\zeta \preceq\omega\},\\ ( \sigma,\omega ) & : = \{\zeta\in i^{\infty}:\sigma\prec\zeta\prec\omega\},\\ ( \sigma,\omega ] & : = \{\zeta\in i^{\infty}:\sigma\prec\zeta\preceq\omega\},\\ \lbrack\sigma,\omega ) & : = \{\zeta\in i^{\infty}:\sigma\preceq\zeta \prec\omega\}.\end{aligned}\ ] ] [ omegaprop]let @xmath217 and @xmath218 \(i ) @xmath219)$ ] where @xmath220 is defined by @xmath221)\cup s_{1}(\lambda\cap(\beta,\overline{1}]).\ ] ] \(ii ) @xmath222)$ ] where @xmath223 is defined by @xmath224).\ ] ] \(iii ) @xmath225)$ ] where @xmath226 is defined by@xmath221)\cup s_{1}(\lambda\cap\lbrack\beta,\overline{1}]).\ ] ] proof of ( i ) : let @xmath227 denote the domain and range restricted shift map .
it is readily found that the branches of @xmath228 are @xmath229\cap\omega\rightarrow\omega$ ] where @xmath230\cap\omega,\ ] ] and @xmath231\cap\omega\rightarrow\omega$ ] where @xmath232\cap\omega.\ ] ] ( note that @xmath233 @xmath234 and @xmath235 it follows that @xmath236)\cup s_{1}(\lambda\cap(\beta,\overline{1}])=\psi(\lambda)\ ] ] for all @xmath237 .
since @xmath238 $ ] it follows that @xmath239).\ ] ] also , since @xmath240)\subset\lbrack\overline { 0},\overline{1}]$ ] it follows that @xmath241)\}$ ] is a decreasing ( nested ) sequence of sets , each of which contains @xmath242 hence@xmath243)\text{.}\ ] ] it remains to prove that @xmath244)$ ] .
we note that @xmath245)=[\overline{0},\tau(\rho)]$ ] and @xmath246)=(\tau^{+}(\rho),\overline{1}],$ ] from which it follows that @xmath247)=\bigcap\limits_{k\in\mathbb{n}}\{\sigma\in i^{\infty}:s^{k}(\sigma ) \in\lbrack\overline{0},\tau(\rho)]\cup(\tau^{+}(\rho),\overline{1}]\}. \label{nameeq}\ ] ] let @xmath248)$ ] .
suppose @xmath249 .
let @xmath250 so that@xmath251 but @xmath252 ( since @xmath173 is closed from the left ) , so@xmath253 note that , since @xmath254 @xmath255 , and @xmath174 is closed from the right , we have @xmath256 .
let @xmath257 . then @xmath258 and we must have @xmath259 and @xmath260 . so@xmath261 therefore @xmath262\cup(\tau^{+}(\rho),\overline{1}]\}$ ] which ,
because of ( [ nameeq ] ) , contradicts our assumption that @xmath263)$ ] .
hence @xmath264 and we have @xmath265)\text{.}\ ] ] this completes the proof of ( i ) .
proof of ( ii ) : similar to the proof of ( i ) , with the role of @xmath266 $ ] played by @xmath267 and the role of @xmath268 $ ] played by @xmath269 $ ] .
proof of ( iii ) : similar to the proofs of ( i ) and ( ii ) . it is helpful to note that the addresses @xmath270 and @xmath271 in proposition [ omegaprop ] obey@xmath272 let @xmath273 be such that @xmath274 .
it follows from the discussion at the end of section [ basicsec ] that @xmath275 and @xmath276 .
[ addresscor]let @xmath139 @xmath70 , @xmath277 and let @xmath273 be such that @xmath274 .
the set of addresses @xmath278 is uniquely determined by @xmath279 and @xmath280 .
for some @xmath281 such that @xmath282 , @xmath283 and @xmath284 .
the set of addresses @xmath285 are uniquely determined by @xmath286 and @xmath287 ; for example , @xmath288 @xmath289 , and @xmath290 .
it follows from proposition [ omegaprop ] that the set of addresses at level @xmath28 , namely @xmath291 is invariant under the following operation : put a @xmath292 in front of each address that is less than or equal to @xmath270 , then truncate back to length @xmath293 take the union of the resulting set of addresses with the set of addresses obtained by : put a @xmath294 in front of each address that is greater than or equal to @xmath271 , and drop the last digit .
[ lemma1 ] @xmath295 , @xmath296 and @xmath297 this is an immediate consequence of proposition [ omegaprop ] .
[ corollary2 ] @xmath6 is symmetric if and only if @xmath298 ( or equivalently @xmath299 ) . [ lemma3 ]
the maps @xmath300 and @xmath301 are strictly increasing as functions of @xmath302 $ ] to @xmath26 . note that @xmath300 depends both implicitly and explicitly on @xmath141 let @xmath303 be such that @xmath304 @xmath305 .
observe that @xmath306 @xmath307 .
assume first that there is a largest @xmath308 such that @xmath309 @xmath310 . then @xmath311 and @xmath312 which implies @xmath313 ( we write @xmath314 when we want to note the dependence on @xmath67 . )
we may assume that @xmath315 is constant on @xmath316 $ ] for otherwise we can restrict to a smaller interval with a strictly smaller value of @xmath317 .
as a consequence , at every iteration , we apply the same branch @xmath318 or @xmath319 to @xmath320 to compute compute @xmath321 for all @xmath322 $ ] .
therefore @xmath323 is continuous with derivative at least @xmath324 which contradicts ( [ inequalityeq])@xmath325 the only remaining possibility is that @xmath326 .
we may assume that @xmath300 is constant on @xmath316,$ ] otherwise we can reduce the problem to the previous case .
this would mean that for arbitrarily large @xmath327 the image of the interval @xmath316 $ ] under @xmath323 is at an interval of size least @xmath328 a contradiction .
essentially the same argument , with the role of @xmath65 played by @xmath66 and the role of played by @xmath62 , proves that @xmath301 is strictly increasing as a function of @xmath329 $ ] to @xmath214 [ corollary4 ] the map @xmath330 is left continuous and the map @xmath331 is right continuous .
fix a parameter @xmath332 and let @xmath333 .
then by ( 7 ) there is @xmath334 which is not a preimage of @xmath332 for any order and such that @xmath335 by ( 9 ) , for any @xmath34 there exists @xmath336 such that the prefix @xmath337 is constant when @xmath338 .
let @xmath317 be such that @xmath339 , and let @xmath340 and @xmath341 .
we have that @xmath342 and @xmath343 combining the two inequalities we obtain@xmath344 and by lemma [ lemma3 ] we also have@xmath345 the distance @xmath346 has the property that if @xmath347 then @xmath348 and @xmath349 .
this shows that @xmath330 is left continuous .
the right continuity of @xmath331 admits an analagous proof . as a consequence of corollary [ corollary2 ] , lemma [ lemma3 ] and ( 10 )
, we obtain the unicity of @xmath67 for which @xmath6 is symmetric.as a consequence of corollary [ corollary2 ] , lemma [ lemma3 ] and ( 10 ) , we obtain the unicity of @xmath67 for which @xmath6 is symmetric .
[ corollary5 ] there is at most one @xmath14 $ ] such that @xmath350 .
[ proof of theorem [ symmetrythm].]by lemma [ lemma3 ] and ( 10 ) , we may define@xmath351:\tau(\rho)\preceq\tau^{+}(\rho)^{\ast } \}=\inf\{\rho\in\lbrack1-b , a]:\tau(\rho)^{\ast}\preceq\tau^{+}(\rho)\}.\ ] ] assume @xmath352 .
it is straighfoward to check @xmath353 .
there is a largest @xmath354 such that @xmath355 .
observe that @xmath356 and @xmath357 . if neither @xmath358 nor @xmath359 belongs to @xmath360 then by ( 9 ) both @xmath361 and @xmath362 are constant on a neighborhood of @xmath332 which contradicts the definition of @xmath332 .
let us consider the projection @xmath363 . if @xmath364 then by the continuity of @xmath318 , of @xmath365 by ( 13 ) , of @xmath366 ( corollary [ corollary4 ] ) there is a @xmath367 such that @xmath368 . by ( 6 ) and ( 13 ) this implies @xmath369 , which again contradicts the definition of @xmath332 . as @xmath365 is increasing , ( 13 ) and @xmath370 we have @xmath371 .
let @xmath372 be minimal such that @xmath373 @xmath374 or @xmath375
. we may apply ( 14 ) @xmath376 times and obtain@xmath377 as @xmath378 , if @xmath375 then we have@xmath379 which by ( 6 ) and equation ( [ equation2 ] ) implies @xmath380 .
therefore @xmath381 as both are periodic of period @xmath382 and have the same prefix of length @xmath383 , a contradiction .
if @xmath384 then @xmath380 thus by ( 13 ) , ( 10 ) and equality ( [ equation2 ] ) we obtain@xmath385 by ( 6 ) , this means that @xmath386 so in fact@xmath387 as @xmath388 @xmath389 @xmath390 where @xmath391 , as @xmath392 .
we can write @xmath393 therefore @xmath394 by ( 6 ) and the previous inequality . by induction
we get @xmath395 @xmath396 so @xmath397 a contradiction .
the case @xmath398 is analagous by the symmetric definition of @xmath332 , therefore @xmath399 is symmetric .
[ proposition7 ] if @xmath350then the map @xmath400\rightarrow\lbrack0,1]$ ] defined by @xmath401 is a homeomorphism and @xmath402 on @xmath26 .
first by corollary [ corollary2 ] , we have @xmath403 and points @xmath54 for which @xmath404 are exactly preimages of @xmath67 . in this case
, there is @xmath405 such that @xmath53 and @xmath108 have the same initial prefix @xmath406 and @xmath407 @xmath408 @xmath409 .
therefore , by ( 13 ) , for all @xmath109,$ ] we have @xmath410 thus @xmath411 . by ( 6 ) ,
( 10 ) and ( 13 ) , @xmath412 is also decreasing .
therefore @xmath400\rightarrow$ ] @xmath3 $ ] is a homeomorphism .
it is readily verified , by direct substitution of @xmath413 into @xmath414 , that @xmath415 for all @xmath39 .
theorem [ irtheorem ] follows from proposition [ omegaprop ] , but some extra language is needed . in explaining this language we describe the conley - mcgehee - wiandt decomposition theorem , ( * ? ? ?
* theorem 13.1 ) .
for @xmath416 a compact hausdorff space , let @xmath417 be the subsets of @xmath416 .
a _ relation _ @xmath72 on @xmath416 is simply a subset of @xmath418 a relation @xmath72 on @xmath416 is called a _ closed relation _
if @xmath72 is a closed of @xmath418 for example the set @xmath68 defined in theorem [ irtheorem ] , namely @xmath419 is a closed relation . following @xcite , a relation @xmath420 provides a mapping @xmath421 defined by@xmath422 notice that the image of a nonempty set may be empty .
iterated relations are defined by @xmath423 and , for all @xmath25,@xmath424 the _ omega limit set _ of @xmath425 under a closed relation @xmath426 is@xmath427 where @xmath428 by definition , an _ attractor _ of a closed relation @xmath72 is a closed set @xmath429 such that the following two conditions hold : given an attractor @xmath429 for a closed relation @xmath72 on a compact hausdorff space @xmath416 , there exists a corresponding _ attractor block _ , namely a closed set @xmath437 such that @xmath78 contains both @xmath429 and @xmath438 in its interior , and @xmath439 .
also , there exists a unique _ dual repeller _ @xmath440 this repeller is an attractor for the transpose relation @xmath441 .
the set of connecting orbits associated with the attractor / repeller pair @xmath442 @xmath443 is @xmath444 .
if @xmath72 is a closed relation on a compact hausdorff space @xmath416 , then @xmath445 is called _ chain - recurrent _ for @xmath72 if for every closed neighborhood @xmath446 of @xmath72 , @xmath54 is periodic for @xmath446 ( i.e. there exists a finite sequence of points @xmath447 such that @xmath448 @xmath449 and @xmath450 for @xmath451 ) .
the chain recurrent set @xmath452 for @xmath72 is the union of all the points that are chain recurrent for @xmath72 .
a transitive component of @xmath452 is a member of the equivalence class on @xmath452 defined by @xmath453 when for every closed neighborhood @xmath446 of @xmath72 there is an orbit from @xmath54 to @xmath454 under @xmath446 ( i.e. there exists a finite sequence of points @xmath455 such that @xmath448 @xmath456 and @xmath457 for all @xmath458 . )
[ * proof of theorem [ irtheorem]*]this follows at once from proposition [ omegaprop ] together with the definitions associated with the theory of iterated closed relations as summarized above , but see @xcite .
we note the following .
@xmath6 can be embedded in @xmath3\subset \mathbb{r}$ ] using the ( continuous and surjective ) coding map @xmath462 $ ] associated with the iterated function system @xmath463;x\mapsto x/2,x\mapsto(1+x)/2).$ ] this coding map @xmath464 is defined for all @xmath465@xmath466 @xmath464 provides a homeomorphism between @xmath6 and @xmath467 .
the point @xmath468 is uniquely and unambiguously represented by the binary real number @xmath469 . in the representation provided by @xmath470
the map @xmath471 becomes the action of the iterated closed relation @xmath472\times\lbrack0,1]\subset\mathbb{r}^{2}$ ] defined by@xmath473\}\cup\{(x,(x+1)/2):x\in \lbrack\pi(\beta),1]\}\ ] ] on subsets of @xmath3 $ ] .
it follows from proposition [ omegaprop ] ( iii ) that @xmath474 is the maximal attractor , as defined in @xcite , of @xmath475 .
the corresponding dual repeller is the empty set .
the chain recurrent set of @xmath475 is @xmath476 , the transitive components are @xmath477 @xmath478 and @xmath479 , and the points @xmath480 and @xmath478 are repulsive fixed points and comprise the boundary of the basin of the attractor @xmath481 which contains no invariant subsets . | we consider a one parameter family of dynamical systems @xmath0\rightarrow \lbrack0,1]$ ] constructed from a pair of monotone increasing diffeomorphisms @xmath1 , such that @xmath2 @xmath3\rightarrow\lbrack0,1],$ ] @xmath4 .
we characterise the set of symbolic itineraries of @xmath5 using an attractor @xmath6 of an iterated closed relation , in the terminology of mcgehee , and prove that there is a member of the family for which @xmath6 is symmetrical . |
the mass of hot interstellar gas that cools over a hubble time in a massive elliptical galaxy is typically 5 - 7 percent of the total baryonic mass of these galaxies .
several observations indicate that this gas cools in a significant volume of the inner galaxy : ( 1 ) the total cooled mass significantly exceeds the mass of central massive black holes ( magorrian et al .
1998 ) , ( 2 ) although much of the interstellar gas within @xmath1 is ejected from old galactic stars which typically have a systematic galactic rotation , the inner x - ray images of rotating elliptical galaxies are nearly circular , implying that the gas cools before it flows very far toward the rotation axis ( hanlan & bregman 2000 ; brighenti & mathews 2000b ) , ( 3 ) direct evidence of diffuse cooled gas at @xmath2 k is invariably observed in large elliptical galaxies in @xmath3 ( e.g. macchetto et al .
1996 ) , and ( 4 ) evolutionary models for the hot interstellar gas are in better agreement with x - ray observations when cooling mass loss is included ; most of this gas cools within an effective radius ( brighenti & mathews 2000a ) . in view of the very small masses of cold hi or h@xmath4 gas that have been observed in elliptical galaxies , far less than the mass of hot gas that has cooled
, it is usually assumed that the cooled interstellar gas is continuously forming into a population of very low mass ( brown dwarf ) stars ( fabian , nulsen & canizares 1982 ) .
such stars add mass but no light .
we show here that if the cooled gas is optically dark , the fundamental plane for elliptical galaxies is likely to be perturbed beyond observed limits . alternatively ,
if the mass to light ratio of the forming stars is not infinite but is comparable to that of the old stars , then some of the cosmic scatter or tilt of the fundamental plane from the virial plane could be due to the presence of this young stellar population .
support for a continuously forming , optically luminous stellar population in elliptical galaxies is provided by the nearly solar mass of neutral cores in interstellar clouds at which gravitational collapse first occurs , @xmath5 @xmath6 ( mathews & brighenti 1999a ) , and by the apparent young stellar ages of many elliptical galaxies based on the h@xmath7 photometric index ( worthey 1994 ; worthey , trager & faber 1996 ; trager 1997 ; mathews & brighenti 1999b ) . in the following discussion , however , we adopt the usual hypothesis that newly forming stars in elliptical galaxies are optically dark . in this case radial variations in the total stellar mass to
light ratio must occur .
furthermore , the total mass of cooled gas varies with galactic luminosity , resulting in rather profound excursions from the fundamental plane , evidently in violation of the remarkable thinness and regularity of this plane .
if elliptical galaxies were perfectly homologous stellar systems with identical stellar populations , then the virial theorem @xmath8 would imply simple correlations among observable parameters . here
@xmath9 is the `` central '' velocity dispersion , @xmath10 is the stellar ( or galactic ) mass , @xmath1 is the effective ( half - light ) radius and @xmath11 is a constant of proportionality .
the three observables @xmath12 , @xmath13 , and @xmath1 define a homologous plane ( hp ) in this ( logarithmic ) three - dimensional parameter space . instead
, elliptical galaxies satisfy a different , slightly steeper fundamental plane ( fp ) relation @xmath14 $ ] , implying that @xmath15 ( dressler et al .
1987 ; djorgovski & davis 1987 ) . in view of the importance of elliptical galaxies for @xmath16 distance determinations ,
considerable effort has gone into understanding the physical origins of the `` tilt '' of the fp relative to the hp .
interpretations of this tilt are constrained by the regularity of the fp - hp deviation across the entire region observed and the remarkably small observational scatter ( @xmath17% ) about the fp relation ( renzini & ciotti 1993 ) .
generally , the fp - hp tilt has been interpreted as some combination of two limiting effects : either ( 1 ) stellar populations or dark matter contributions vary systematically across the fp so that @xmath18 or ( 2 ) homology fails so that @xmath19 .
this relation depends mildly on photometric band ( e.g. scodeggio et al .
1998 ; pahre & djorgovski 1997 ) . several types of stellar population variation ( age , metallicity or imf ) have been considered along the fp ( renzini & ciotti 1993 ; pahre , djorgovski & carvalho 1995 ; pahre & djorgovski 1997 ) . these attempts to interpret the fp - hp tilt
have not been entirely successful since the fp is nearly invariant to observed wavelength over the @xmath20 photometric bands .
the small systematic wavelength trend of the fp slope is consistent with the color - magnitude relation for giant elliptical galaxies ( scodeggio et al 1998 ; pahre , de carvalho & djorgovski 1998 ) .
moreover , if the stellar population changes across the fp , differential stellar evolution can in some cases introduce a change in the fp slope at large redshifts , but no such effect has yet been observed . it is generally asserted that the full fp tilt can not be understood solely by metallicity variations or changing stellar populations along the fp since this would require a high degree of fine tuning in the imf . if @xmath21 is assumed to be constant with galactic radius , the mass to light ratios for elliptical galaxies found from dynamic models vary in a manner similar to the fp deviation , @xmath22 ( van der marel 1991 ) . a wide variety of non - homologous structural and/or kinematic models
have been proposed to explain all or part of the fp - hp tilt ( e.g. , capelato et al .
1995 ; djorgovski 1995 ; hjorth & madsen 1995 ; ciotti , lanzoni & renzini 1996 ) . in these types of models
the stellar @xmath23 does not vary so the fp tilt should be strictly independent of wavelength apart from mutually non - homologous stellar and dark matter distributions .
graham & colless ( 1997 ) fit the stellar mass and light distribution with sersic @xmath24 laws and note that @xmath25 increases with galactic mass ( by @xmath26 ) in a non - homologous fashion across the fp .
if these galaxies are mapped onto the fp with an @xmath27 law in the usual way , the resulting fp , @xmath28 , tilts even further from the hp , @xmath29 , where @xmath30 is the surface brightness .
however , a dynamical non - homology must accompany a structural non - homology . solving the jeans equation for elliptical galaxies modeled with @xmath24 profiles ( ciotti 1991 ; ciotti & lanzoni 1997 ) , graham & colless ( 1997 ) and graham ( 1998 )
determine infinite - aperture @xmath31 which when fit to @xmath24 galaxies result in @xmath32 , which is somewhat closer to the hp .
busarello et al .
( 1997 ) consider a different dynamic non - homology based on partial rotational support within @xmath1 .
using a modified @xmath33 which incorporates both rotation and random velocities , they find that more than half of the fp - hp tilt is accounted for , @xmath34 .
busarello et al .
claim that the remaining discrepancy can be understood with spatial non - homology and stellar population differences ( e.g. prugniel & simien 1997 ) .
another approach is the possibility that dark ( halo ) matter makes a variable contribution along the fp ( ciotti , lanzoni , & renzini 1996 ) , resulting in the fp - hp tilt .
however , these authors stress again the tightness of observational scatter about the fp so that any systematic trend in the dark matter content must be rather finely tuned . this high degree of coordination may seem unlikely if elliptical galaxies are produced by a variety of random mergers among stochastically different galaxies of lesser mass .
the amount and radial distribution of dark halo matter in the centers of elliptical galaxies is complicated by the baryonic winds in first generation galaxies and the dissipational baryonic concentration expected as giant elliptical galaxies form by mergers ; these counteracting effects respectively contract and expand the dark matter distribution relative to the baryonic stellar population .
there is at present no clear winner among the various explanations for the fp - hp tilt . instead , a patchwork of small contributions
stellar population variations , rotational support and non - homology are required to bring fp into agreement with hp ( ciotti , lanzoni , & renzini 1996 : prugniel & simien 1997 ; pahre , de carvalho & djorgovski 1998 ) .
however , there is no compelling reason why these quite different and presumably independent contributions should combine in a coordinated fashion to produce the full fp - hp tilt observed across the entire fp .
moreover , a multiplicity of coordinated explanations for the tilt , none entirely dominant , makes the tightness of the fp even more remarkable .
in this paper we introduce a new complication to the fp - hp controversy : a variation in the dynamical @xmath35 with galactic mass due to the deposition of cooled interstellar gas in @xmath36 .
the importance of cooled gas in the galactic mass budget follows from the universality of hot interstellar gas in giant elliptical galaxies and the direct relationship of the cooled mass to the loss of energy implied by the observed x - ray emission @xmath37 .
for example , interstellar gas in the large virgo elliptical galaxy ngc 4472 with temperature @xmath38 k emits @xmath39 ergs s@xmath40 in the _ einstein _ band .
the bolometric x - ray luminosity , @xmath41 , is the product of the mass cooling rate @xmath42 and the specific enthalpy of the gas , therefore @xmath43 since @xmath37 was somewhat larger in the past , over a hubble time @xmath44 @xmath6 of hot gas will have cooled .
although this amounts to only about @xmath45 percent of the total stellar mass in ngc 4472 , @xmath46 @xmath6 , most of the cooled gas is likely to be deposited near the galactic center @xmath36 where the hot gas density and x - ray emissivity are greatest and where its contribution to the dynamical @xmath35 determined from stellar velocity dispersions is most apparent .
( in evaluating @xmath47 we assume @xmath48 , as explained below . )
although there is little doubt that interstellar cooling has occurred , two main questions must be addressed : ( 1 ) what is the final physical state of the cold gas ? and ( 2 ) what is the radial distribution of the cooled mass in the galaxy ? for many years it has been assumed that low mass stars are the most likely final state for cooled gas ( e.g. fabian , nulsen & canizares 1982 ) . in a recent paper ( mathews & brighenti 1999a )
we describe how the hot interstellar gas cools ( initially ) to @xmath2 k where it can be heated and ionized by stellar uv photons and made visible in optical emission lines .
this first stage of cooling occurs at many ( @xmath49 ) cooling sites distributed over the inner galaxy .
however , when the radius of partially cooled gas at @xmath50 k exceeds the local stromgren radius at a cooling site , a cold neutral atomic or molecular core forms with @xmath51 k and @xmath52 @xmath53 in pressure equilibrium with the ambient hot interstellar gas .
as the mass of the neutral core @xmath54 increases further , it becomes jeans unstable when @xmath55 @xmath6 so only stars of this or lower mass can form from the available gas ( ferland , fabian & johnstone 1994 ; mathews & brighenti 1999a ) .
cooling gas is processed within hii cooling site clouds with star - forming cold cores in time scales ( @xmath56 years ) that are short compared to their hydrodynamic disruption times in the cooling flow atmosphere ( mathews & brighenti 1999a ) .
the high efficiency of low mass star formation nicely accounts for the absence of observable cold hi or molecular gas in large elliptical galaxies , e.g. for ngc 4472 @xmath57 @xmath6 .
we conclude that two stellar populations coexist in giant elliptical galaxies : an old luminous stellar component with an imf similar to salpeter , and a second population of low mass stars continuously forming from the interstellar medium and having a bottom heavy imf .
although the detailed character of this second imf is unknown , for the purposes of the following discussion we adopt the usual assumption that it is restricted to very low masses , i.e. its mass to light ratio is essentially infinite .
the second question concerning the radial distribution of cooled gas mass can be answered in an approximate way by comparing hydrodynamic models with the radial properties of interstellar gas observed in x - rays .
the production of low mass stars at many distributed sites can only occur if the interstellar gas is inhomogeneous .
radiative cooling is locally enhanced in regions of low entropy ( low @xmath58 , high density @xmath25 ) . as gas in the galactic cooling flow slowly proceeds inward , regions of progressively higher entropy cool at smaller galactic radii ; regions having low entropy cool further from the galactic centers .
distributed h@xmath59 + [ nii ] line emission observed in the central regions of giant elliptical galaxies illuminate the cooling sites where this process is occurring .
interstellar entropy variations can be created by a variety of complex events : stellar disturbances ( e.g. stellar winds or type ia supernovae ) , interactions with merging dwarf galaxies or differential type ii supernova heating in the early universe during the earliest phases of star formation .
because of our poor understanding of these interstellar entropy fluctuations , it is not possible to determine from first principles where most of the hot gas cools and deposits its mass into low mass stars , although the observed optical line emission is a visible tracer of this process .
the best currently available means of setting limits on the radial distribution of cooling gas in elliptical galaxies is to compare the observed radial distribution of interstellar x - ray emission with the computed x - ray emission from gas - dynamic models for the cooling flow gas in which cooling dropout is assumed to occur according to various radial prescriptions .
we have recently computed a series of models for the large elliptical galaxy ngc 4472 with a variety of assumed mass dropout functions ( brighenti & mathews 2000a ) . in the following section
we describe the cooling dropout properties of models that best fit the x - ray observations , then we discuss the implications of this dropout mass for the fundamental plane .
in view of the uncertainty concerning the cooling dropout , in our gas dynamical models of galactic cooling flows we parameterize the radial mass dropout of cooling gas by introducing an adjustable sink term in the equation of continuity , @xmath60 here @xmath61 and @xmath62 are the gas density and velocity in the cooling flow .
the sink term for cooling flow dropout is characterized by a dimensionless parameter @xmath63 and @xmath64 is the time for gas at temperature @xmath65 to cool locally by radiative losses at constant pressure ( see , e.g. sarazin & ashe 1989 ) .
the source term @xmath66 represents the rate that interstellar gas is supplied by stellar mass loss from the old stellar population having density @xmath67 . since @xmath68 for a wide range of power law imfs , the total stellar mass ejected @xmath69 is nearly identical for elliptical galaxies with the same @xmath70 but different imfs ( brighenti & mathews 2000a ) . following the approach in brighenti & mathews ( 2000a )
, we consider galactic cooling flow models for ngc 4472 using a parameterized dropout coefficient @xmath71 where @xmath72 , @xmath73 and @xmath74 are adjustable parameters .
for all sets of dropout parameters considered we found that the total dropout mass in @xmath75 is a substantial fraction ( 0.10 - 0.35 ) of the mass of luminous stars in this region .
the radial distribution of the dropout mass , assumed to remain at the cooling site , does not follow the space density corresponding to the de vaucouleurs light profile .
if low mass stars formed from cooling dropout contribute mass but no appreciable light , the hypothesis that we want to test , it follows that the intrinsic `` stellar '' mass to light ratio of the luminous stellar population in ngc 4472 must be somewhat lower than values determined from the observed motion of luminous stars : @xmath76 ( van der marel 1991 ) or @xmath77 ( faber et al .
the most satisfactory mass dropout coefficients @xmath63 generate hydrodynamic solutions for the interstellar gas at the current time ( @xmath78 gyrs ) that reproduce best the observed x - ray image of ngc 4472 . using this approach , we find that the dynamical mass to light ratio in successful hydrodynamic models requires a somewhat lower intrinsic mass to light ratio for the luminous stars @xmath79 . for the low mass dropout stellar population
we assume @xmath80 .
ideally , it would be desirable to determine if the same ( or appropriately scaled ) mass dropout coefficient @xmath63 used in gas dynamical models for elliptical galaxies of varying @xmath70 , spread across the fundamental plane , can fit the observed x - ray surface brightness distributions @xmath81 in these same galaxies .
unfortunately , at the present pre-_chandra _ time sufficiently accurate @xmath81 are known only for several bright elliptical galaxies in virgo ( e.g. , trinchieri , fabbiano & canizares 1986 ) , all having similar @xmath70 .
approximate @xmath81 distributions are available only for about a dozen elliptical galaxies altogether ( mathews & brighenti 1998a ) .
in addition , emission from low mass binary stars begins to dominate the x - ray luminosity in elliptical galaxies several times less luminous than ngc 4472 , masking the x - ray emission from the cooling flows .
finally , even though @xmath82 approximately holds among brighter elliptical galaxies , there is an enormous spread in values of @xmath37 for given @xmath70 .
we believe that most of this spread is due to environmental interactions : ( i ) tidal exchange of halo hot gas ( and dark matter ) in the group environments in which these elliptical galaxies formed ( mathews & brighenti 1998b ) or ( ii ) ram pressure stripping . with these difficulties in mind , we chose instead to explore the relative influence of optically dark interstellar dropout not directly on the fp but on a homologous galactic plane ( hp ) that is scaled from ngc 4472 for which good x - ray observations are available .
we consider a sequence of three massive elliptical galaxies , ngc 4472 and two similar model galaxies , 4472/4 and 4472/16 , having optical luminosities @xmath70 that are respectively 4 and 16 times less than ngc 4472 .
to explore only the structural non - homology caused by cooling dropout , we also assume that the old stellar population is invariant along our hp , i.e. its @xmath35 is constant . for perfect homology
we also require that the mass and scale radius of the dark halo distribution scale with the baryonic mass .
finally , we solve for the evolution of hot gas in each galaxy along the hp using identical ( or scaled ) parameters ( @xmath72 , @xmath73 and @xmath74 ) that give good agreement with the x - ray observations for ngc 4472 , @xmath83 and @xmath84 ; this procedure is described in more detail in brighenti & mathews ( 2000a ) . by this means we can determine the differential , non - homologous distribution of optically dark dropout mass and assess the deviation of these three idealized galaxies from the hp relation .
we expect that these deviations from the hp are similar to deviations of galaxies of similar @xmath70 from the local fp . if our procedure were perfectly correct
i.e. , elliptical galaxies are exactly homologous and contain identical old stellar populations we would find that the non - homology introduced by cooling dropout would explain the `` tilt '' of the observed fp from the hp .
unfortunately , this goal has not yet been achieved . for perfect homology and simplicity
we assume that 4472 and its two clones of lower mass are spherical with de vaucouleurs stellar profiles and scaled nfw dark halos ( navarro , frenk , & white 1996 ) . the effective radius of ngc 4472 is @xmath85 kpc ( for distance @xmath86 mpc ) and its total ( old ) stellar mass is @xmath87 @xmath6 , based on @xmath48 .
the nfw halo in ngc 4472 has a virial mass of @xmath88 @xmath6 .
these parameters give a good fit to the x - ray mass profile of ngc 4472 assuming hydrostatic equilibrium although the nfw profile is found to be somewhat too centrally peaked ( brighenti & mathews 2000a ) .
for the smaller comparison galaxies ( 4472/4 and 4472/16 ) we choose identical stellar parameters but scale the size of the stellar distribution downward by @xmath89 or @xmath90 where the exponent is found from the observed @xmath91-@xmath1 relation for massive elliptical galaxies , @xmath92 , ( faber et al .
the virial mass of the nfw halos for the smaller galaxies are reduced by 4 or 16 and the scale radius for these halos varies as @xmath93 , as required by our chosen cosmology for which the nfw concentration varies as @xmath94 ( navarro , frenk , & white 1996 ) . as explained in more detail elsewhere ( brighenti & mathews 1999a )
, our gas dynamical models begin with an overdensity in a simple einstein - de sitter cosmology ( @xmath95 , @xmath96 , and @xmath97 km s@xmath40 mpc@xmath40 ) toward which baryonic and dark matter are gravitationally attracted .
galactic stars are assumed to form at time @xmath98 gyr and , soon afterwards , type ii supernova energy based on a salpeter imf ( slope : @xmath99 , mass limits : @xmath100 and @xmath101 @xmath6 ) heats the ambient gas within the accretion shock with 80 percent efficiency ( see brighenti & mathews 1999a for details ) . at @xmath102 gyrs the de vaucouleurs stellar profile and potential is constructed , simulating the results of mergers .
for this approximate exploration we ignore the possibility that elliptical galaxies of lower luminosity may have formed slightly earlier . in some models
we allow the secondary inflow of cosmic gas to continue throughout the calculation . for comparison , in other models
we ignore the secondary infall entirely , crudely simulating a tidal truncation and removal of halo material in a relatively dense group of newly formed galaxies , a process more likely in elliptical galaxies of moderate or low mass .
in table 1 we list the main results of a variety of hydrodynamic cooling flow models for 4472 , 4472/4 and 4472/16 with various assumed dropout parameters and assumed amounts of accumulated cosmic gas .
of particular interest are the mass to light ratios evaluated at @xmath103 and at a fixed distance of 480 pc from the galactic centers . recall that @xmath48 is assumed for the luminous stars in all three galaxies and that the observed value for ngc 4472 is @xmath104 ( van der marel 1991 ) or @xmath105 ( faber et al .
models in which the ( stellar plus dropout ) mass to light ratio for ngc 4472 falls in the range @xmath106 can be regarded as reasonably successful .
however , the @xmath107 entries in table 1 are enhanced by about 15 percent due to a small amount of nfw dark halo material in this region . since our fits to the x - ray surface brightness of ngc 4472 are improved if the dark halo is less centrally peaked than nfw ( also see kravtsov et al . 1998 ; burkert & silk 1997 ; navarro & steinmetz 2000 ) , we believe that its contribution to @xmath35 in this region should be removed . therefore , for comparison with @xmath107 values in table 1 which contain this nfw excess , the range of acceptable models for ngc 4472 should be revised upward to @xmath108 ( i.e. @xmath109 = 0.99 - 1.10 $ ] ) .
values of @xmath35 at 480 pc may be more relevant for comparison with @xmath35 determinations that are often based on stellar velocity dispersions measured within a fixed aperture of a few arcseconds about the galactic centers . in cooling flows without distributed dropout ( @xmath110 )
the flow proceeds entirely to the origin before cooling and the mass to light ratios at 480 pc greatly exceed observed values for 4472 such models can certainly be rejected .
evolutionary gas dynamic models with uniform @xmath111 at all radii and with dropout parameters ( @xmath112 ) = ( 4 , 2 kpc , 1 ) both fit the observed distribution of interstellar gas temperature and density in ngc 4472 reasonably well at the present time ( @xmath78 gyrs ) as illustrated in figure 1 . while the agreement in figure 1 is acceptable , it is by no means perfect .
the apparent gas density ( enhanced by dropout emission ) exceeds observed densities in the central regions ( @xmath113 kpc ) of ngc 4472 , suggesting that some additional non - thermal support is present ; ngc 4472 has a weak extended radio source in this region with an appreciable equipartition pressure ( ekers & kolanyi 1978 ) . at large galactic radii ( @xmath114 kpc )
our models in figure 1 are slightly lower than the azimuthally - averaged values observed in ngc 4472 ; this density mismatch can be corrected with small changes in the galaxy formation time @xmath115 or the efficiency of snii energy release ( brighenti & mathews 1999a ) . however , our interest here is not to achieve absolute perfection in fitting the observations of ngc 4472 , but to make a differential comparison of 4472 with its less massive , homologous clones ; 4472/4 and 4472/16 .
the additional model with stronger but uniform dropout @xmath116 , also included in table 1 , results in apparent gas temperatures at @xmath78 gys that are considerably lower than those observed ( brighenti & mathews 2000a ) ; this is due to enhanced x - ray emission from cooling regions throughout the cooling flow .
several interesting details are apparent in table 1 . since the dropout mass is deposited non - homologously , the dynamic mass to light ratio varies with galactic radius , i.e. @xmath117 .
even when the dropout parameters are scaled homologously from 4472 [ ( @xmath112 ) = ( 4 , 2 kpc , 1 ) ] to 4472/4 [ ( @xmath112 ) = ( 4 , 0.667 kpc , 1 ) ] , the resulting @xmath35 are _ not _ equal , i.e. dropout does not scale homologously .
some models in table 1 were performed with no additional secondary infall gas following the time @xmath115 of galaxy formation ( no in column 5 ) .
this procedure is justified by the strong correlation of normalized x - ray sizes @xmath118 with @xmath119 noted by mathews & brighenti ( 1998 ) . we explained this correlation with models of tidally truncated galactic dark halos and ( mostly secondary infallen ) halo gas in smaller galaxies and the transfer of this material to halos of more massive , centrally located elliptical galaxies , a process which is dynamically expected in galaxy groups where elliptical galaxies are formed .
the removal of secondary infallen halo gas is therefore more relevant to 4472/4 and 4472/16 than for 4472 ( which may have acquired additional halo gas from neighboring galaxies ) ; note that this late arriving cosmic gas also has a larger influence on @xmath35 for the smaller galaxies . the non - homologous dropout behavior listed in table 1
is illustrated in figure 2 in which the 4472 @xmath120 4472/4 @xmath120 4472/16 variation of @xmath107 and @xmath121 are plotted for each model .
the local fundamental plane , @xmath122 , shown with dashed lines , has a slope @xmath123 that is intermediate between values for the b - band fp taken from jorgensen , franx , & kjaergaard ( 1996 ) and scodeggio et al .
the dashed lines are normalized to @xmath124 ( from van der marel 1991 ; long dashed line ) and @xmath125 ( from faber et al .
1997 ; short dashed line ) respectively .
the asterisks show the constant stellar @xmath48 assumed for luminous stars in all three galaxy models .
clearly , none of the model pairs exhibit a variation of @xmath107 as steep as the fp relation , but the dropout contribution to the fp slope is nevertheless consequential at this radius .
however , the variation of @xmath35 at @xmath126 pc for the @xmath111 models follows the fundamental plane quite accurately over the range of @xmath70 between 4472 and 4472/4 .
note that the pair of variable dropout models with homologously scaled parameters [ @xmath127 for 4472 and @xmath128 for 4472/4 ] produce @xmath129 that vary in a sense opposite to the fp trend , but only when all cosmic gas from secondary infall is included .
our mass to light values for 4472 may be lower limits since we have not considered models in which unusually massive galaxies receive more than their cosmological allotment of halo material by tidal acquisitions from less massive , unmerged group galaxies .
the constant @xmath130 models can explain much of the fp tilt in 4472 @xmath120 4472/4 with or without secondary infall .
however , when extended to 4472/16 , values of @xmath129 clearly exceed the fundamental plane for all models .
while the dropout contribution to @xmath129 in the @xmath131 model ( without cosmic inflow ) decreases with @xmath70 over the range of 16 shown in figure 2 ( i.e. vertical distance between open squares and asterisks ) , the dropout mass does not disappear entirely for the 4472/16 model .
in fact the assumed @xmath48 for the luminous stars alone ( asterisks in fig .
2 ) also lies above the fp for 4472/16 . within the framework of our optically dark dropout assumption
, this indicates that an independent variation of @xmath35 for the old stars would be required in addition to that provided by the dropout .
results shown in the upper panel of figure 2 suggest that the fp could be fit over the full 4472 @xmath120 4472/16 range if we had chosen a somewhat lower value of @xmath35 for the luminous stars , but the x - ray data for ngc 4472 become difficult to fit if @xmath132 .
however , dynamically significant galactic rotation is expected for elliptical galaxies having b - band luminosities less than that of 4472/4 , @xmath133 @xmath134 , which lies near the transition between massive , boxy elliptical galaxies with cores and low mass , rotationally supported , disky elliptical galaxies with power law profiles ( faber et al .
rotation can reduce the mass dropout in smaller galaxies since the cooled gas will be deposited in a relatively larger galactic volume , reducing @xmath21 values for 4472/16 in figure 2 . although the excess interstellar density in our models within 1 - 3 kpc ( figure 1 ) can be understood in terms of @xmath135 g magnetic fields , it could also indicate that we have somewhat overestimated the gas density in this region and also the mass of cooled gas .
perhaps the best resolution of the large @xmath21 in our models of 4472/16 would be to allow some of the cooled gas to form a luminous stellar population .
using this assumption , we have shown that the anomalously low ages for bright elliptical galaxies found from observations of the h@xmath7 photometric index ( e.g. , trager 1997 ) can be quite naturally produced by a young population with a salpeter imf extending up to @xmath5 @xmath6 but no further ( mathews & brighenti 1999a ) . in models having luminous dropout stars , about 10 percent of the optical light is produced by the young stars .
the stellar disk component characteristic of rotating , low luminosity elliptical galaxies ( rix , carollo , & freeman 1999 ) , can be formed from younger cooling dropout stars ( brighenti & mathews 1997b ) , and there is some evidence that these galaxies also have enhanced h@xmath7 indices ( de jong & davies 1997 ) . in our models
we determine the detailed radial variation of the mass to light ratio within @xmath1 ; this variation depends most sensitively on our choice of @xmath63 . in general , optical observations of stellar velocity dispersion profiles can not determine the detailed variation of @xmath21 with radius within @xmath1 . only with considerable effort
have observers been able to detect the presence of dark matter at @xmath136 ( e.g. , carollo et al . 1995 ; rix et al .
1997 ) , although this dark matter is immediately apparent from x - ray observations ( e.g. brighenti & mathews 1997a ) .
the insensitivity of optical observations to the detailed radial variation of @xmath21 is probably due to the generic stellar orbital bias toward radial anisotropy in @xmath36 ; this results in @xmath21 determinations that more nearly represent orbital averages rather than local values . for completeness , in the final column of table 1 we list the rosat x - ray luminosity of each of our models .
in general our results are slightly flatter than the observed trend for luminous elliptical galaxies , @xmath137 with @xmath138 , in which the x - ray emission is interstellar ( e.g. eskridge , fabbiano & kim 1995 ) .
however , we have shown elsewhere ( mathews & brighenti 1998 ) that @xmath119 correlates with the size of the x - ray image and that this correlation can be explained by the tidal transfer of halo material in galaxy groups from subordinate elliptical galaxies to the dominant central elliptical in the group .
galactic rotation can also influence @xmath119 .
because of these environmental influences , the values of @xmath37 in table 1 should not be regarded as a strong test of a particular model .
the full extent of the fundamental plane for elliptical galaxies spans a range of @xmath70 about @xmath139 times greater than the factor of 16 shown in figure 2 ( burstein et al .
1997 ) . maintaining
the fp - hp tilt over this larger range would be difficult if cooling dropout is an important contributor of dark mass .
nevertheless , the presence of appreciable dropout mass seems unavoidable .
the total mass of cooled interstellar gas present in elliptical galaxies could be reduced by supernova - driven galactic winds or by the formation of stars that have since evolved . for our models
we assume a type ia rate snu@xmath140 snu@xmath141@xmath142 supernovae per 100 years per @xmath143 @xmath70 with @xmath144 and snu@xmath145 ( brighenti & mathews 2000a ) .
winds produced by more numerous type ia supernovae at early times ( i.e. @xmath146 ) can not be invoked to expel some of the dropout material from galaxies like 4472/4 or 4472/16 since these same higher type ia rates in more massive galaxies would greatly overproduce iron in the current interstellar gas , in violation of x - ray observations .
our value snu@xmath145 was chosen to be consistent with the low interstellar iron abundances for elliptical galaxies reported in most of the recent literature , @xmath147 solar ( e.g. , tanaka et al .
if the abundances are more nearly solar ( buote 1999 ) , as is likely , then a proportionally larger snu@xmath141 is indicated ; this can result in galactic winds in the 4472/16 model at late times @xmath148 .
therefore , the total mass of interstellar gas that cools can be reduced by snii - driven galactic outflow at early times ( when most of the dropout occurs ) and ( less efficiently ) by snia - driven outflow at late times
. however , the large @xmath37 observed in virtually all bright elliptical galaxies clearly rules out ongoing galactic winds in these galaxies .
the current amount of dropout mass could also be reduced if the imf of the forming stars was more luminous in the past ( i.e. more massive stars ) .
while none of these means of reducing the dropout mass is securely understood , some significant mass of cooled gas must be present .
the large mass of cooled interstellar gas expected within @xmath1 together with the alarming number of possible ways of enhancing or moderating the amount of cooled gas makes the coherency and tightness of the fp quite baffling .
our main objective in presenting these results is to emphasize the importance of cooling dropout mass to the fp - hp controversy .
the results that we have shown here are based on the commonly held assumption that only optically dark stars of very low mass can form in cooling flows .
however , optically dark dropout by itself appears to be an unpromising explanation for the full fp tilt , although with some well chosen parameters that might be possible .
what we see in figure 2 is an unwelcome and quite unaesthetic disturbance of the @xmath21 relation , one that involves the uncertain physics of star formation and the origin of entropy fluctuations in the hot gas .
nevertheless , the mass contribution from cooled interstellar gas can not be dismissed by those concerned with the deviation of the fundamental plane from a truly homologous plane . like other effects proposed to explain the fp - hp discrepancy ,
if the cooling dropout mass is a dominant contributor , the mass dropout must vary in a smooth , regular fashion across the fp , proportional to some power of @xmath70 .
such regularity could conceivably result from the similar profiles of interstellar gas density and temperature observed in the central regions of elliptical galaxies ( @xmath36 ) , provided the entropy fluctuations are also similar .
our apparent success in explaining the the anomalously young ages of elliptical galaxies indicated by the h@xmath7 index from an optically luminous dropout stellar population ( mathews & brighenti 1999b ) strongly suggests that a better match to the fp - hp tilt could be achieved if the optically dark assumption used here is relaxed . in this case
the imf of the continuously forming stars is likely to extend to higher maximum masses in elliptical galaxies of lower @xmath70 because of the lower interstellar pressure in low luminosity elliptical galaxies .
studies of the evolution of hot gas in elliptical galaxies at uc santa cruz is supported by grants from nasa and the nsf for which we are very grateful .
in addition fb is supported in part by grant murst - cofin 98 .
brighenti , f. & mathews , w. g. 1997a , apj 486 , l83 brighenti , f. & mathews , w. g. 1997b , apj 490 , 592 brighenti , f. & mathews , w. g. 1999a , apj 512 , 65 brighenti , f. & mathews , w. g. 2000a , apj ( in press ) ( astro - ph/0002385 ) brighenti , f. & mathews , w. g. 2000b , apj ( submitted ) buote , d. a. 1999 , mnras 309 , 685 burkert , a. & silk , j. 1997 , apj 488 , l55 burstein , d. , bender , r. , faber , s. , & nolthenius , r. 1997 , aj , 114 , 1365 busarello , g. , capaccioli , m. , capozziello , s. , longo , g. & puddu , e. 1997 , a&a 320 , 415 capelato , h. v. , de carvalho , r. r. , & carlberg , r. g. 1995 , apj 451 , 525 carollo , c. m. , de zeeuw , p. t. , van der marel , r. p. , danziger , i. j. & qian , e. e. 1995 , apj 441 , l25 ciotti , l. 1991 , a&a 249 , 99 ciotti , l. & lanzoni , b. 1997 , a&a 321 , 724 ciotti , l. , lanzoni , b. & renzini , a. 1996 , mnras 282 , 1 de jong , r. s. & davies , r. l. 1997 , mnras 285 , l1 djorgovski , m. 1995 , apj 438 , l29 djorgovski , m. & davis , s. 1987 , apj 313 , 59 dressler , a. , lynden - bell , d. , burstein , d. , davies , r. l. , faber , s. m. , terlevich , r. j. & wegner , g. 1987 , apj 313 , 42 ekers , r. & kolanyi , c. a&a 67 , 47 eskridge , p. b. , fabbiano , g. & kim , d .- w .
1995 , apjs 97 , 141 faber , s. m. , tremaine , s. , ajhar , e. a. , byun , y - i . , dressler , a. , gebhardt , k. , grillmair , c. , kormendy , j. , lauer , t. r. , & richstone , d. 1997 , aj 114 , 1771 fabian , a. c. , nulsen , p. e.
j. & canizares , c. r. 1982 , mnras 201 , 933 ferland , g. j. , fabian , a. c. & johnstone , r. m. 1994 , mnras 266 , 399 graham , a. w. 1998 , mnras 295 , 933 graham , a. w. & colless , m. 1997 , mnras 287 , 221 hanlan , p. c. & bregman , j. n. 2000 , apj ( in press ) ( astro - ph/9910210 ) hjorth , j. & madsen , j. 1995 , apj , 445 , 55 irwin , j. a. & sarazin , c. l. 1996 , apj 471 , 683 kravtsov , a. v. , klypin a. a. , bullock , j. s. & primack , j. r. 1998 , apj 502 , 48 macchetto , f. et al .
1996 , a&as 120 , 463 magorrian , j. et al .
1998 , apj 115 , 2285 mathews , w. g. & brighenti , f. 1998a , apj 493 , l9 mathews , w. g. & brighenti , f. 1998b , apj 503 , l15 mathews , w. g. & brighenti , f. 1999a , apj 526 , 114 mathews , w. g. & brighenti , f. 1999b , apj 527 , l31 navarro , j. f. , frenk , c. s. & white , s. d. m. 1996 , apj , 462 , 563 navarro , j. f. & steinmetz , m. 2000 , apj 528 , 607 pahre , m. a. , de carvalho , r. r. & djorgovski , s. g. 1998 , aj 116 , 1606 pahre , m. a. , djorgovski , s. g. & de carvalho , r. r. 1995 , apj 453 , l17 pahre , m. a. & djorgovski , s. g. 1997 , in _ the nature of elliptical galaxies , proceedings of the second stromlo symposium _ , eds . m. arnaboldi , g. s. da costa & p. saha , asp conf .
116 , ( san francisco : asp ) , 154 prugniel , p. & simien , f. 1997 , a&a 321 , 111 renzini , a. & ciotti , l. 1993 , apj 416 , l49 rix , h .- w .
, de zeeuw , p. t. , cretton , n. , van der marel , r. p. & carollo , c. m. 1997 , apj , 488 , 702 rix , h .- w . , carollo , m. & freeman , k. 1999 , apj 513 , l25 sarazin , c. l. & ashe , g. a. 1989 , apj 345 , 22 scodeggio , m. , gavazzi , g. , belsole , e. , pierini , d. , & boselli , a. 1998 , mnras 301 , 1001 tanaka , y. , inoue , h. & holt , s. s. 1994 , pasj 46 , l37 trager , s. 1997 , phd thesis , university of california santa cruz trinchieri , g. , fabbiano , g. & canizares , c. r. 1986 , apj 310 , 637 van der marel , r. p. 1991 , mnras , 253 , 710 worthey , g. 1994 , apjs 95 , 107 worthey , g. , trager , s. , & faber , s. m. 1996 , in asp conf .
86 , _ fresh views of elliptical galaxies _ , eds .
buzzoni & a. renzini ( san francisco : asp ) , 203 lcccccrrr 4472/16 : & & & & & & & & & 0.493 & 0 & & & no & 9.32 & 8.57 & 0.10 & 0.493 & 1 & @xmath149 & & no & 7.29 & 7.13 & 0.10 & 0.493 & 4 & 0.317 & 1 & no & 8.25 & 8.17 & 0.25 4472/4 : & & & & & & & & & 1.97 & 0 & & & yes & 11.76 & 19.27 & 9.35 & 1.97 & 0 & & & no & 8.97 & 12.37 & 1.56 & 1.97 & 1 & @xmath149 & & yes & 7.95 & 8.41 & 8.00 & 1.97 & 1 & @xmath149 & & no & 7.40 & 7.65 & 1.52 & 1.97 & 4 & 0.667 & 1 & yes & 11.05 & 10.91 & 14.57 & 1.97 & 4 & 0.667 & 1 & no & 8.65 & 8.54 & 2.57 & 1.97 & 4 & 2 & 1 & yes & 8.16 & 7.27 & 10.78 & 1.97 & 4 & @xmath149 & & yes & 7.03 & 6.79 & 6.00 4472 : & & & & & & & & & 7.89 & 0 & & & yes & 10.47 & 32.29 & 20.08 & 7.89 & 0 & & & no & 9.97 & 28.53 & 8.59 & 7.89 & 1 & @xmath149 & & yes & 8.83 & 11.40 & 31.56 & 7.89 & 1 & @xmath149 & & no & 8.64 & 10.96 & 12.56 & 7.89 & 4 & 2 & 1 & yes & 9.74 & 7.99 & 28.11 & 7.89 & 4 & 2 & 1 & no & 9.37 & 7.88 & 13.73 & 7.89 & 4 & 0.8 & 2 & yes & 10.47 & 13.08 & 36.44 & 7.89 & 4 & @xmath149 & & yes & 7.88 & 7.20 & 39.82 | we explore here the possibly important influence of cooled interstellar gas on the fundamental plane of elliptical galaxies .
interstellar cooling is described by a parameterized sink term in the equation of continuity .
parameters that give the best fits to the x - ray observations of elliptical galaxy ngc 4472 are used as a template for other galaxies .
these spatially - dependent cooling parameters are then extended homologously to elliptical galaxies of lower mass for which x - ray observations are currently unavailable or impossible to observe because of stronger relative x - ray emission from binary stars .
most of the interstellar gas cools within an effective radius where it can contribute an additional 10 - 30 percent to the mass of the old stellar population .
the absence of observed cooled gas and simple theoretical arguments support the hypothesis that the cooled gas is forming into low mass stars , with implications that depend on the imf of this young stellar population .
if the imf includes only stars of very low mass , @xmath0 as is commonly believed , the cooled mass is optically dark . for this imf
the mass to light ratios determined from stellar velocities systematically overestimate that of the old stellar population .
furthermore , the total mass and spatial distribution of the optically dark stellar population does not scale homologously with galactic luminosity or radius so the total stellar mass to light ratio is expected to vary with galactic radius .
these variations in stellar population and non - homology can introduce pronounced deviations from the fundamental plane .
we investigate cooled gas perturbations to the mass to light ratio for several idealized homologous elliptical galaxies and show that they may be incompatible with the observed thinness of the fundamental plane . however ,
if the imf of the stellar population produced from cooled interstellar gas is optically luminous , its influence on observed stellar mass to light ratios and the fundamental plane would be lessened .
if the imf of a young , continuously forming and optically luminous stellar population were sufficiently well - behaved , it could account for about ten percent of the optical light from large elliptical galaxies within @xmath1 , possibly having important implications for understanding the fundamental plane .
.2 in |
recently the use of metamaterials as wave absorbers has been attracting the interest of many researchers from the viewpoint of feasible metamaterial applications using current technology . since the first experimental metamaterial work was reported by smith et al .
@xcite , various metamaterials have been proposed to realise their exotic properties , such as negative permittivity , negative permeability and negative refractive index ( nri ) , from the microwave region to the optical region @xcite . at the same time
many metamaterial applications have also been suggested ; the superlens @xcite , invisible cloaking @xcite and optical transformation @xcite to name but a few . however , there are obstacles in the realisation of these schemes .
most metamaterial applications rely on the negative properties described above . at high frequencies , such as terahertz and optical bands ,
the properties have non zero imaginary parts of refractive index and wave impedance mainly due to the conductive loss of the metal components , making the applications less attractive or sometimes unrealistic . for this reason
the conductive loss issue is recognised as a problem to be solved @xcite and many researchers are trying to address this matter , for example , by the use of gain mediums @xcite or by optical transformation designs that include the conductive loss @xcite .
in this paper we seek to exploit these losses for efficient absorber design . in metamaterial absorbers
the use of the conductive loss can be taken into account as part of the absorber design to achieve strong levels of absorptance .
thus the practical use of metamaterials as wave absorbers is becoming popular for applications in areas such as emc , crosstalk reduction etc .
so far metamaterial absorbers have exhibited attractive features which are not available from conventional wave absorbers .
one example is in the manipulation of the electromagnetic properties . in conventional absorbers design options
are limited by substrate properties and mixture ratios to realise desired permittivity and permeability values .
this is not easy to accomplish . in this paper
we aim at systematic design techniques for metamaterial absorbers to enable us to modify their electromagnetic properties with relative ease ( e.g. changes in metallisation length and geometry ) .
more importantly the use of metamaterials makes the absorber substrate thicknesses significantly smaller . in the conventional absorbers the total thickness of the structures need to be comparable to a quarter wavelength @xmath1 of the operating frequency .
however , the @xmath2 thickness can limit the range of application , for example , where available working space is restricted .
this issue has successfully been solved by use of thin metamaterials , which allowed us to relax the limitation of the physical size remarkably ( e.g. in @xcite 1 mm thickness at operating frequency 6 ghz , i.e. approximately one fiftieth of the wavelength ) , and more flexible design options became possible .
an outstanding issue is the limitation of the absorption band width . in most cases absorption
is concentrated in a narrow frequency band .
although narrow band absorbers may be useful in some applications , the real need is for broadband absorbers with customised characteristics to meet specific user needs .
this paper shows that highly customisable broadband absorption is possible using metamaterials as wave absorber .
different lengths of conductive and lossy cut wires ( cw ) ( see insets of figs .
[ fig : modelfield ] ( a ) and ( b ) as examples of cws ) are deployed as a single periodic metamaterial absorber unit .
highly customisable absorption is achieved by independent conductive loss optimisation in each cw . in addition , this absorption can be obtained for arbitrary polarisations without any serious absorptance reduction when other pairs of cws are introduced .
furthermore , it is also demonstrated that not only enhancing the absorptance but also reducing it is possible by use of additional cws .
these results indicate that the cws make it possible to design very flexible absorptance characteristics . before the customisable absorber configurations
are explained in detail , this paper introduces characteristics of lossless cw metamaterials in section [ sec : material ] then examines the basic properties of cw metamaterials in sections [ sec : singleandpaired ] . in section
[ sec : cwpec ] the customised broadband absorber designs are introduced and conclusion is made in section [ sec : conc ] .
previous work is described in @xcite which was done using limited numerical resolution .
this work is extended here to develop customised absorbers .
this paper uses a thin rectangular metal form , the so called cw , as a base metal geometry for the following reasons .
firstly , the cw metamaterials are very simple structures and the periodic unit is composed of only a dielectric substrate and a single or paired cw . due to their simplicity ,
cw metamaterials can be transformed into other types of metamaterials @xcite , so they play an important role in basic metamaterial research .
secondly , the cw metamaterials do not require any extra metal components to yield a nri .
for example , the first metamaterial @xcite is composed of split ring resonators , yielding a negative value of permeability , and long metal strips , yielding a negative value of permittivity , and hence two types of metal structures are necessary to produce a nri .
however , the cw metamaterials can exhibit both negative permittivity and permeability through a simply paired cw .
thirdly , cw structures are suitable for introducing stronger losses .
as is reported in @xcite , compared to fishnet structures @xcite which have some similar properties with those cw metamaterials have , the cw metamaterials can exhibit greater losses , which although undesirable in most metamaterial applications they are needed in absorber applications . moreover , from a fabrication viewpoint , the cw form is straightforward to manufacture due to its simple geometry . finally , cws with conductive loss easily produce multi absorptance peaks as will be demonstrated in this paper . in absorber applications described in this paper conductive loss
is used instead of dielectric loss which is assumed to be a primary resource in most metamaterial absorbers ( e.g. @xcite ) .
one of the reasons is that materials with suitable conductive losses are readily available through recently developed technologies , e.g. conductive inks @xcite .
also , independent optimisation of the metal resistances gives us more flexible options for absorption characteristic designs , e.g. multi peak broadband absorption , as will be demonstrated in section [ sec : cwpec ] .
the default structures of the simulated cw metamaterials are illustrated in the insets of figs .
[ fig : modelfield ] ( a ) and ( b ) for single and paired cws and the simulations were performed by using the transmission line modelling ( tlm ) method @xcite .
to simplify the situation and consider the absorption effect due only to the conductive loss of the cw metals , the relative permittivity @xmath3 of the substrate was set @xmath4 and all other loss mechanisms were removed .
the cw structure had a width of 0.3 mm along the @xmath5 axis ( @xmath6field ) and a length of 5.1 mm along the @xmath7 axis ( @xmath8field ) .
the dimension of the unit cell was @xmath9 mm .
this structure was modelled using cubical tlm unit cells having edge lengths of @xmath10 mm .
the sheet resistance @xmath11 ( @xmath12 ) used to quantify the conductive loss was allocated so that the scattering coefficients from the metal are determined by @xmath13 where @xmath14 and @xmath15 respectively represent reflection coefficient and transmission coefficient both multiplied by the incident voltage ( the incident wave ) in each tlm unit cell , @xmath16 is the impedance in vacuum and @xmath17 .
the scattering parameters were calculated using observation planes 15 mm away from the absorber surface and a gaussian pulse was excited on a plane one cell above the observation plane for the reflection coefficient . since periodic boundaries were applied for the @xmath18 and @xmath19 plane boundaries , the cw metamaterial absorber unit modelled was assumed to belong to infinite array on the bottom @xmath20 plane . the simulated paired cw metamaterials are generally known to exhibit two types of resonances : electric resonances and magnetic resonances @xcite .
these resonances are explained in fig . [
fig : modelfield ] where the reflection coefficient , the electric fields and the conduction currents of the cw metamaterials all calculated by tlm method are illustrated .
when an incident wave is excited with the electric field parallel to the direction of the cw metals , the conduction currents flow in opposite directions in each cw at the magnetic resonance ( see fig .
[ fig : modelfield ] ( d ) ) .
this is the minimum @xmath21 at 24.89 ghz in fig .
[ fig : modelfield ] ( b ) . in this case
the electric field is anti
parallel at the two edges of the cw metamaterial , fig .
[ fig : modelfield ] ( c ) , and a magnetic field concentrated within the area bounded by the paired cw is generated .
this magnetic field plays the role of an artificial magnetic dipole and can lead to a negative value of permeability . on the other hand , at the electric resonance found at around an @xmath21 peak ( 26.71 ghz in fig .
[ fig : modelfield ] ( b ) ) , conduction currents in the cws become dominant , resulting in a strong unidirectional electric field at the edges of the cws , figs .
[ fig : modelfield ] ( e ) and ( f ) . similarly with the magnetic resonance this electric field behaves as an artificial electric dipole so as to produce a negative value of permittivity .
a similar electric resonance is found in the @xmath21 peak of the single cw metamaterial ( 26.65 ghz of fig .
[ fig : modelfield ] ( a ) ) as well .
the positions of the two types of the resonances can be manipulated @xcite and in subsection [ ssec : pairedcw ] the influence of the resonant frequency positions on the absorptance peak will be discussed . of ( a ) lossless single cw and ( b ) lossless paired cw .
( c ) electric field and ( d ) conduction current of the magnetic resonance frequency ( 24.89 ghz ) .
( e ) electric field and ( f ) conduction current of the electric resonance frequency ( 26.71 ghz ) . ] of ( a ) lossless single cw and ( b ) lossless paired cw .
( c ) electric field and ( d ) conduction current of the magnetic resonance frequency ( 24.89 ghz ) .
( e ) electric field and ( f ) conduction current of the electric resonance frequency ( 26.71 ghz ) . ] of ( a ) lossless single cw and ( b ) lossless paired cw .
( c ) electric field and ( d ) conduction current of the magnetic resonance frequency ( 24.89 ghz ) .
( e ) electric field and ( f ) conduction current of the electric resonance frequency ( 26.71 ghz ) . ] of ( a ) lossless single cw and ( b ) lossless paired cw .
( c ) electric field and ( d ) conduction current of the magnetic resonance frequency ( 24.89 ghz ) .
( e ) electric field and ( f ) conduction current of the electric resonance frequency ( 26.71 ghz ) . ] of ( a ) lossless single cw and ( b ) lossless paired cw .
( c ) electric field and ( d ) conduction current of the magnetic resonance frequency ( 24.89 ghz ) .
( e ) electric field and ( f ) conduction current of the electric resonance frequency ( 26.71 ghz ) . ] of ( a ) lossless single cw and ( b ) lossless paired cw .
( c ) electric field and ( d ) conduction current of the magnetic resonance frequency ( 24.89 ghz ) .
( e ) electric field and ( f ) conduction current of the electric resonance frequency ( 26.71 ghz ) . ]
this subsection investigates basic absorptance characteristics of single cw metamaterial absorbers .
the default size of the periodic structure is illustrated in fig .
[ fig : alllossysingle ] ( a ) .
first of all , conductive loss is distributed in one of three ways : allocations to all the metal part , both of the edges ( 0.9 mm for each edge and 1.8 mm in total ) and one of the edges ( 1.8 mm or 1.2 mm ) .
secondly , the absorptances of different cw lengths are investigated and they are combined as one metamaterial unit to achieve broadband absorption . in all the simulations the value of the sheet resistance , @xmath11 , is constant . in fig .
[ fig : alllossysingle ] ( b ) the absorptance @xmath22 of the single cw composed of only lossy metal is shown for various sheet resistance values .
the absorptance was calculated from @xmath23 .
it is found from fig .
[ fig : alllossysingle ] ( b ) that the positions of the maximum absorptance are fixed at around 26.59 ghz , which is close to the resonant frequency found in fig .
[ fig : modelfield ] ( a ) .
in addition it turns out that the absorptance peak curve reaches the maximum values with a sheet resistance @xmath24 , the inset of fig .
[ fig : alllossysingle ] ( b ) .
this absorptance peak dependence can be easily explained with the equivalent circuit drawn in the inset of fig . [
fig : alllossysingle ] ( c ) . at the resonant frequency
the circuit impedance becomes @xmath25 . in terms of the conduction current @xmath26
, numerical simulations revealed its dependence on @xmath11 as the curve with closed circles shown in fig .
[ fig : alllossysingle ] ( c ) .
therefore , the power dissipated in the circuit , @xmath27 ( @xmath28 ) , as illustrated by the curve with open squares in fig .
[ fig : alllossysingle ] ( c ) , well corresponds to the result in the inset of fig .
[ fig : alllossysingle ] ( b ) . at 26.59 ghz
is illustrated .
( c ) the conduction current @xmath26 and the dissipated power @xmath27 ( @xmath29 ) in the cw with various values of @xmath11 .
( f ) absorptances of the single cw composed of both lossless metal part and lossy metal part .
the lossy metal parts were restricted to both edges ( 0.9 mm per each edge and 1.8 mm in total ) in ( d ) and one edge in ( e ) ( 1.8 mm ) and in ( f ) ( 1.2 mm ) . ] at 26.59 ghz is illustrated .
( c ) the conduction current @xmath26 and the dissipated power @xmath27 ( @xmath29 ) in the cw with various values of @xmath11 .
( f ) absorptances of the single cw composed of both lossless metal part and lossy metal part .
the lossy metal parts were restricted to both edges ( 0.9 mm per each edge and 1.8 mm in total ) in ( d ) and one edge in ( e ) ( 1.8 mm ) and in ( f ) ( 1.2 mm ) . ] at 26.59 ghz is illustrated .
( c ) the conduction current @xmath26 and the dissipated power @xmath27 ( @xmath29 ) in the cw with various values of @xmath11 .
( f ) absorptances of the single cw composed of both lossless metal part and lossy metal part .
the lossy metal parts were restricted to both edges ( 0.9 mm per each edge and 1.8 mm in total ) in ( d ) and one edge in ( e ) ( 1.8 mm ) and in ( f ) ( 1.2 mm ) . ] at 26.59 ghz is illustrated .
( c ) the conduction current @xmath26 and the dissipated power @xmath27 ( @xmath29 ) in the cw with various values of @xmath11 .
( f ) absorptances of the single cw composed of both lossless metal part and lossy metal part .
the lossy metal parts were restricted to both edges ( 0.9 mm per each edge and 1.8 mm in total ) in ( d ) and one edge in ( e ) ( 1.8 mm ) and in ( f ) ( 1.2 mm ) . ] at 26.59 ghz is illustrated .
( c ) the conduction current @xmath26 and the dissipated power @xmath27 ( @xmath29 ) in the cw with various values of @xmath11 .
( f ) absorptances of the single cw composed of both lossless metal part and lossy metal part .
the lossy metal parts were restricted to both edges ( 0.9 mm per each edge and 1.8 mm in total ) in ( d ) and one edge in ( e ) ( 1.8 mm ) and in ( f ) ( 1.2 mm ) . ] at 26.59 ghz is illustrated .
( c ) the conduction current @xmath26 and the dissipated power @xmath27 ( @xmath29 ) in the cw with various values of @xmath11 .
( f ) absorptances of the single cw composed of both lossless metal part and lossy metal part .
the lossy metal parts were restricted to both edges ( 0.9 mm per each edge and 1.8 mm in total ) in ( d ) and one edge in ( e ) ( 1.8 mm ) and in ( f ) ( 1.2 mm ) . ]
similar maximum absorptance peaks were obtained when the conductive loss area was restricted to one or both of the edges of the cw metal ( see figs .
[ fig : alllossysingle ] ( d ) to ( f ) ) . however , further increase of the loss amount exhibited different absorptance peaks at higher frequencies .
these absorption positions were expected to correspond to resonances of cws shortened by the lossy metal part(s ) , since the conductive losses were too high causing a highly distorted current distribution .
numerical simulations confirmed that the 3.3 mm length cw , whose length correspond to the lossless parts of figs .
[ fig : alllossysingle ] ( d ) and ( e ) , has resonant frequency at 38.06 ghz , which is in the vicinity of the absorptance peaks at the higher frequencies in figs .
[ fig : alllossysingle ] ( d ) and ( e ) . similarly the 3.9 mm
length cw , whose metal length is same with the lossless part of fig .
[ fig : alllossysingle ] ( f ) , was found to have resonant frequency at 33.55 ghz , which is also close to another absorptance peak of fig .
[ fig : alllossysingle ] ( f ) .
although the absorptance effect obtained from these structures is weak , one point noticed from figs .
[ fig : alllossysingle ] ( e ) and ( f ) is that the decrease of the absorptance at the frequency range between the two peaks can be reduced by restricting the lossy metal part to the narrow area .
secondly , the absorptances of different cw lengths were calculated .
the calculation results are summarised in fig .
[ fig : difflength ] ( a ) where the simulated cws had lengths of 3.3 to 5.1 mm by 0.6 mm steps .
it is found from fig . [ fig : difflength ] ( a ) that the absorptance peak depends on the cw length .
next several of the cws used in fig .
[ fig : difflength ] ( a ) were combined as one metamaterial unit and the resultant absorptance curves are illustrated in fig .
[ fig : difflength ] ( b ) .
the inset in fig .
[ fig : difflength ] ( b ) describes a structure consisting of four cws . in this structure
the distance between the cws was 0.3 mm and one end of the cws were positioned at a same @xmath7 axis position ( 0.3 mm from one of @xmath18 plane boundaries ) . in the other structures only those lengths noted were included . compared to the results in fig .
[ fig : difflength ] ( a ) , the results in fig .
[ fig : difflength ] ( b ) illustrate broadband behaviour .
, respectively . ] , respectively . ]
this subsection examines absorptance characteristics of paired cw metamaterials . as was explained in section [ sec : material ] ,
the paired cw metamaterials have two types of resonance : electric resonance and magnetic resonance .
these resonant frequency positions can be manipulated and in @xcite , for example , one of the cw pairs is shifted a distance @xmath30 ( see fig .
[ fig : offset ] ( a ) ) . in this case
the magnetic resonance frequency is increased , while the electric resonance frequency is reduced .
this effect is shown in fig .
[ fig : offset ] ( b ) where the dependence of the reflection coefficient curves on the various offset lengths @xmath30 are shown .
it is known that , when the magnetic resonance frequency is greater than the electric resonance frequency , a nri is obtained in the paired cw metamaterial due to overlap of the negative permeability with the negative permittivity @xcite .
this paper utilises the geometrical asymmetry introduced in @xcite to change the resonance positions and implement loss in the structure so that the absorption characteristics in the paired cw metamaterial absorber are investigated in more detail . before the asymmetrically paired cw is simulated ,
the absorption characteristics of the symmetrically paired cws are investigated .
the calculation results are summarised in fig .
[ fig : offset ] ( c ) in which various sheet resistance values , @xmath11 , are applied equally to the front and back cws . in fig .
[ fig : offset ] ( c ) we have marked the electric and magnetic resonant frequencies ( @xmath31 and @xmath32 ) of the symmetrically paired cw shown in fig .
[ fig : offset ] ( b ) and it is seen that the absorptance peak shifts from @xmath32 to @xmath31 as @xmath11 increases . it is also found that the symmetrically paired cw shows the stronger absorption ( @xmath33 ) than that of the single cws ( @xmath34 in fig .
[ fig : alllossysingle ] ) .
next , the absorptance of the paired cw metamaterial absorber with various values of @xmath30 was optimised .
the calculation results are summarised in table [ tab : offset ] and fig .
[ fig : offset ] ( d ) .
again , the same resistance values were applied for both front cw and back cw .
it turns out in this case that the maximum absorptance of the symmetrically paired cw is increased by nearly 10 % for @xmath35 mm .
moreover the absorptance of the paired cw was further enhanced by using different sheet resistance values for the front cw and the back cw . in figs .
[ fig : offset ] ( e ) , ( f ) and ( g ) the absorptances for @xmath36 0.0 mm , 1.2 mm and 2.4 mm are illustrated .
each distribution was calculated at the frequency corresponding to each absorptance peak ; 26.35 , 26.23 and 27.20 ghz , respectively . in these figures ,
the sheet resistance values were varied from 0 to 12 @xmath12 in 2 @xmath12 steps .
the intermediate values between the calculated results were estimated using spline interpolation . despite the coarse resolution
the simulations confirm that the absorptance is further increased in each case by the use of different sheet resistance values .
in addition the distribution of the absorptance depends on the offset length @xmath30 .
although the absorptance of the symmetrically paired cw metamaterial was enhanced by the manipulation of the resonant frequency positions and by use of different resistance values , the maximum value was still far from perfect absorption ( i.e. @xmath37 ) . in the next section the absorptance of the cw metamaterial absorber
is significantly improved and approaches perfect absorption by placing a single cw on a perfect electric conductor ( pec ) wall .
of the lossless paired cw having the various geometrical offset length @xmath30 .
( c ) absorptance of the symmetrically paired cw with various @xmath11 .
( d ) absorptances of the paired cw with various @xmath30 . from ( b ) to ( d ) @xmath30 is expressed in each legend . in these figures
same values of @xmath11 are used for the front cw and the back cw . in ( e ) to ( g ) various combinations of @xmath11 are applied for both cws .
( e ) , ( f ) and ( g ) show the absorptances of no offset at 26.35 ghz , 1.2 mm offset at 26.23 ghz and 2.4 mm offset at 27.20 ghz , respectively .
the dots in the figures represent the calculated sheet resistance patterns and the intermediate values were estimated by using spline interpolation . ] of the lossless paired cw having the various geometrical offset length @xmath30 .
( c ) absorptance of the symmetrically paired cw with various @xmath11 .
( d ) absorptances of the paired cw with various @xmath30 . from ( b ) to ( d ) @xmath30 is expressed in each legend . in these figures
same values of @xmath11 are used for the front cw and the back cw . in ( e ) to ( g ) various combinations of @xmath11 are applied for both cws .
( e ) , ( f ) and ( g ) show the absorptances of no offset at 26.35 ghz , 1.2 mm offset at 26.23 ghz and 2.4 mm offset at 27.20 ghz , respectively .
the dots in the figures represent the calculated sheet resistance patterns and the intermediate values were estimated by using spline interpolation . ] of the lossless paired cw having the various geometrical offset length @xmath30 .
( c ) absorptance of the symmetrically paired cw with various @xmath11 .
( d ) absorptances of the paired cw with various @xmath30 . from ( b ) to ( d ) @xmath30 is expressed in each legend . in these figures
same values of @xmath11 are used for the front cw and the back cw . in ( e ) to ( g ) various combinations of @xmath11 are applied for both cws .
( e ) , ( f ) and ( g ) show the absorptances of no offset at 26.35 ghz , 1.2 mm offset at 26.23 ghz and 2.4 mm offset at 27.20 ghz , respectively .
the dots in the figures represent the calculated sheet resistance patterns and the intermediate values were estimated by using spline interpolation . ] of the lossless paired cw having the various geometrical offset length @xmath30 .
( c ) absorptance of the symmetrically paired cw with various @xmath11 .
( d ) absorptances of the paired cw with various @xmath30 . from ( b ) to ( d ) @xmath30 is expressed in each legend . in these figures
same values of @xmath11 are used for the front cw and the back cw . in ( e ) to ( g ) various combinations of @xmath11 are applied for both cws .
( e ) , ( f ) and ( g ) show the absorptances of no offset at 26.35 ghz , 1.2 mm offset at 26.23 ghz and 2.4 mm offset at 27.20 ghz , respectively .
the dots in the figures represent the calculated sheet resistance patterns and the intermediate values were estimated by using spline interpolation . ] of the lossless paired cw having the various geometrical offset length @xmath30 .
( c ) absorptance of the symmetrically paired cw with various @xmath11 .
( d ) absorptances of the paired cw with various @xmath30 . from ( b ) to ( d ) @xmath30 is expressed in each legend . in these figures
same values of @xmath11 are used for the front cw and the back cw . in ( e ) to ( g ) various combinations of @xmath11 are applied for both cws .
( e ) , ( f ) and ( g ) show the absorptances of no offset at 26.35 ghz , 1.2 mm offset at 26.23 ghz and 2.4 mm offset at 27.20 ghz , respectively .
the dots in the figures represent the calculated sheet resistance patterns and the intermediate values were estimated by using spline interpolation . ] of the lossless paired cw having the various geometrical offset length @xmath30 .
( c ) absorptance of the symmetrically paired cw with various @xmath11 .
( d ) absorptances of the paired cw with various @xmath30 . from ( b ) to ( d ) @xmath30 is expressed in each legend . in these figures
same values of @xmath11 are used for the front cw and the back cw . in ( e ) to ( g ) various combinations of @xmath11 are applied for both cws .
( e ) , ( f ) and ( g ) show the absorptances of no offset at 26.35 ghz , 1.2 mm offset at 26.23 ghz and 2.4 mm offset at 27.20 ghz , respectively .
the dots in the figures represent the calculated sheet resistance patterns and the intermediate values were estimated by using spline interpolation . ] of the lossless paired cw having the various geometrical offset length @xmath30 .
( c ) absorptance of the symmetrically paired cw with various @xmath11 .
( d ) absorptances of the paired cw with various @xmath30 . from ( b ) to ( d ) @xmath30 is expressed in each legend . in these figures
same values of @xmath11 are used for the front cw and the back cw . in ( e ) to ( g ) various combinations of @xmath11 are applied for both cws .
( e ) , ( f ) and ( g ) show the absorptances of no offset at 26.35 ghz , 1.2 mm offset at 26.23 ghz and 2.4 mm offset at 27.20 ghz , respectively .
the dots in the figures represent the calculated sheet resistance patterns and the intermediate values were estimated by using spline interpolation . ] .absorptance peaks of paired cw with various @xmath7 axis offset @xmath30 . [
cols="^,^,^,^",options="header " , ]
this section introduces cw metamaterial structures which exhibit very effective absorption characteristics . in addition
, it is shown that broadband absorption is achieved when different lengths of cws are combined in one periodic metamaterial absorber unit .
strong absorptances were observed when single cw metamaterial was placed on a pec surface , as illustrated in the inset of fig .
[ fig : backmetalmulti ] ( a ) . in these simulations
the cw positions are the same with those used in the inset of fig .
[ fig : difflength ] ( b ) , while the resistance values were optimised to extract the maximum absorptances .
the calculation results are shown in fig .
[ fig : backmetalmulti ] ( a ) and the strong absorptances of @xmath37 are seen .
considering the results obtained , the case of the cw with the metal backing gives the highest absorptance ( @xmath38 ) followed by two cws without the metal backing ( @xmath39 ) .
finally a single cw gives the lowest absorption ( @xmath40 ) .
we note that the conduction currents in each case are different as shown in fig .
[ fig : modelfield ] ( d ) .
next the same pattern of cws was deployed along the @xmath5 axis to reduce polarisation dependence .
the calculation results are shown in fig .
[ fig : backmetalmulti ] ( b ) , where the sheet resistance of each cw were again optimised . the inset of fig .
[ fig : backmetalmulti ] ( b ) describes the simulated structure . as a result of the comparison with fig .
[ fig : backmetalmulti ] ( a ) , strong absorptance of @xmath37 is still maintained despite the use of another orthogonal cw . according to @xcite , straight conductors orthogonal to the electric field
do not react with the external field significantly so that the electromagnetic properties of the whole structure are not significantly affected . to obtain broadband behaviour some or all of the pairs of the cws used in fig .
[ fig : backmetalmulti ] ( b ) are combined as one metamaterial unit .
the inset of fig .
[ fig : backmetalmulti ] ( c ) illustrates the structure .
the calculation results are shown in fig .
[ fig : backmetalmulti ] ( c ) .
the results in this figure indicate that the absorptance properties in the high frequency region are improved by use of additional lossy pairs of cw .
the addition of extra cws introduces additional absorption peaks , as shown in fig .
[ fig : backmetalmulti ] ( d ) where two cw pairs of 5.1 and 3.3 mm show two absorptance peaks corresponding to those of the individual cws .
although such behaviour has been recently reported @xcite , the advantage of using the lossy cws introduced here is to easily customise the absorptance characteristics for _ both _ polarisations .
furthermore , as is explained below , it is also easy to realise several absorptance peaks by adding extra pairs of cw .
the absorption characteristics may be further enhanced by optimising the sheet resistance values .
this is shown in fig .
[ fig : backmetalmulti ] ( e ) in which the three cw pairs of 5.1 , 3.9 and 3.3 mm are deployed . in this figure
the sheet resistance value of only the 3.3 mm cw is varied , while the values of the other cws are the same as those of fig .
[ fig : backmetalmulti ] ( b ) .
it is found from fig .
[ fig : backmetalmulti ] ( e ) that the absorption characteristics are improved by use of adjusted resistance values , leading to a triple absorptance peak ( 0.993 at 24.28 ghz , 0.975 at 31.17 ghz and 0.981 at 36.29 ghz ) .
again , this structure will interact with both polarisations , compared with that of @xcite and is an important advantage of the lossy cw metamaterial absorber .
although only the triple absorptance peak is illustrated , further absorptance peaks are possible by using extra cw pairs .
use of additional pairs of cw enables us not only to increase absorption , but also to decrease it . in fig .
[ fig : backmetalmulti ] ( f ) , the absorptance of the cw metamaterial absorber composed of 5.1 and 3.9 mm cw pairs with the sheet resistance values used in fig .
[ fig : backmetalmulti ] ( b ) is illustrated .
this absorptance peaks can be enhanced by use of optimised sheet resistance values as is described in the figure .
in addition to these two cw pairs , when the lossless cw pair of 3.9 mm is deployed ( see the inset of fig .
[ fig : backmetalmulti ] ( f ) ) , the absorptance magnitude between the two absorption peaks is markedly reduced due to the resonance of the 3.9 mm cw pair ( where the sheet resistance values of 5.1 and 3.9 mm cws were modified again ) . as expected , the centre of the absorptance reduction is close to the resonant frequency of the 3.9 mm cw ( c.f .
[ fig : backmetalmulti ] ( b ) ) .
this allows for a fully customised absorptance characteristic .
this paper demonstrated by numerical simulation that highly customisable broadband absorption for arbitrary polarisation is possible by use of conductively lossy cws as metamaterial absorbers . to begin with , the basic properties of the conductively lossy cw metamaterial absorbers were investigated .
a dependence of absorptance peak on the conductive loss was explained with a simple equivalent circuit . in paired cw metamaterials ,
absorptance peaks were improved by manipulating the two resonance frequency positions and by using independent sheet resistance values for the front and back cws .
the absorptance of the conductively lossy cw metamaterial was further improved , when single cw metamaterials were placed on a pec wall .
moreover , when different lengths of cws were combined as one metamaterial unit , broadband absorption was exhibited .
the deployment of orthogonal pairs of cws showed that the structure simulated here works for both polarisations . due to the flexible absorptance characteristics
, the idea of using conductively lossy cw pairs adds additional advantages to metamaterial absorbers and opens up a new area for metamaterial applications .
the interesting properties of cw metamaterials may be further improved , when the structures are designed to absorb off normal incident waves .
this may be possible by using multiple cws whose lengths and sheet resistance values are suitably optimised . | this paper shows that customised broadband absorption of electromagnetic waves having arbitrary polarisation is possible by use of lossy cut wire ( cw ) metamaterials .
these useful features are confirmed by numerical simulations in which different lengths of cw pairs are combined as one periodic metamaterial unit and placed near to a perfect electric conductor ( pec ) .
so far metamaterial absorbers have exhibited some interesting features , which are not available from conventional absorbers , e.g. straightforward adjustment of electromagnetic properties and size reduction .
the paper shows how with proper design a broad range of absorber characteristics may be obtained .
99 d. r. smith , w. j. padilla , d. c. vier , s. c. n. nasser , and s. schultz , composite medium with simultaneously negative permeability and permittivity , `` phys .
rev .
lett . *
84*(18 ) , 41844187 ( 2000 ) . c. caloz and t. itoh , _
electromagnetic metamaterials : transmission line theory and microwave applications _ ( wiley ieee press , new jersey , 2006 ) .
k. busch , g. von freymann , s. linden , s. f. mingaleev , l. tkeshelashvili , and m. wegener , periodic nanostructures for photonics , '' phys . rep . *
444 * , 101202 ( 2007 ) . c. m. soukoulis , j. zhou , t. koschny , m. kafesaki , and e. n. economou , the science of negative index materials , `` j. phys . condens .
matter *
20*(30 ) , * 304217 * ( 2008 ) .
n. fang , h. lee , c. sun , and x. zhang , sub diffraction limited optical imaging with a silver superlens , '' science * 308 * , 534537 ( 2005 ) .
j. b. pendry , d. schurig , and d. r. smith , controlling electromagnetic fields , `` science * 312 * , 17801782 ( 2006 ) .
a. al@xmath0 and n. engheta , achieving transparency with plasmonic and metamaterial coatings , '' phys .
rev .
e * 72 * , * 016623 * ( 2005 ) .
y. lai .
j. ng , h. y. chen , d. han , j. xiao , z. zhang , and c. t. chan , illusion optics : the optical transformation of an object into another object , `` phys .
rev .
lett . * 102 * , * 253902 * ( 2009 ) .
r. f. service , next wave of metamaterials hopes to fuel the revolution , '' science * 327 * , 138139 ( 2010 ) .
a. a. govyadinov , v. a. podolskiy , and a. noginov , active metamaterials : sign of refractive index and gain assisted dispersion management , `` appl .
. lett . * 91 * , * 191103 * ( 2007 ) .
x. luo , t. yang , y. gu , h. chen , and h. ma , conceal an entrance by means of superscatterer , '' appl .
. lett . * 94 * , * 223513 * ( 2009 ) .
d. j. kern and d. h. werner , a genetic algorithm approach to the design of ultra thin electromagnetic bandgap absorbers , `` microwave opt .
. lett . * 38*(1 ) , 6164 ( 2003 ) .
h. wakatsuchi , j. paul , s. greedy , and c. christopoulos , contribution of conductive loss to cut wire metamaterial absorbers , '' presented at 2010 asia pacific radio science conference ( ap rasc10 ) , toyama , japan , 2226 sept . 2010 .
j. zhou , e. n. economon , t. koschny , and c. m. soukoulis , unifying approach to left handed material design , `` opt .
lett . * 31*(24 ) , 36203622 ( 2006 ) .
h. guo , n. liu , l. fu , h. schweizer , s. kaiser , and h. giessen , thickness dependence of the optical properties of split ring resonator metamaterials , '' phys .
status solidi b * 244*(4 ) , 12561261 ( 2007 ) .
g. dolling , c. enkrich , m. wegener , c. m. soukoulis , and s. linden , low
loss negative index metamaterial at telecommunication wavelengths , `` opt .
lett . * 31*(12 ) , 18001802 ( 2006 ) .
h. tao , c. m. bingham , a. c. strikwerda , d. pilon , d. shrekenhamer , n. i. landy , k. fan , x. zhang , w. j. padilla , and r. d. averitt , highly flexible wide angle of incidence terahertz metamaterial absorber : design , fabrication , and characterization , '' phys .
rev .
b * 78 * , * 241103 ( r ) * ( 2008 ) .
k. kords , t. mustonen , g. tth , h. jantunen , m. lajunen , c. soldano , s. talapatra , s. kar , r. vajtai , p. ajayan , inkjet printing of electrically conductive patterns of carbon nanotubes , `` small * 2 * , 10211025 ( 2006 ) . c. christopoulos , _ the transmission line modeling method : tlm _ ( ieee press , new york , 1995 ) .
s. n. burokur , a. sellier , b. kant , and a. de lustrac , symmetry breaking in metallic cut wire pairs metamaterials for negative refractive index , '' appl .
. lett . *
94 * , * 201111 * ( 2009 ) .
j. b. pendry , a. j. holden , d. j. robbins , and w. j. stewart , low frequency plasmons in thin
wire structures , `` j. phys . condens .
matter *
10*(22 ) , 47854809 ( 1998 ) .
q. y. wen , h. w. zhang , y. s. xie , q. h. yang , and y. l. liu , dual band terahertz metamaterial absorber : design , fabrication , and characterization , '' appl .
. lett . * 95 * , * 241111 * ( 2009 ) .
h. tao , c. m. bingham , d. philon , k. fan , a. c. strikwerda , d. shrekenhamer , w. j. padilla , x. zhang , and r. d. averitt , a dual band terahertz metamaterial absorber , " j. phys .
d : appl .
phys . *
43*(22 ) , * 225102 * ( 2010 ) . |
ngc 6240 is a nearby luminous infrared galaxy ( e.g. , wright , joseph & meikle 1984 ; joseph & wright 1984 ) , comprising two merging nuclei ( north and south nuclei : hereafter , n and s nuclei , respectively ) separated by 1.9 ( condon et al . 1982 ; fried & schulz 1983 ; eales et al . 1990 ; thronson et al .
1990 ; herbst et al . 1990 ; colbert , wilson , & bland - hawthorn 1994 ; sugai et al .
1997 ; tacconi et al .
1999 ; ohyama et al .
2000 ; scoville et al .
2000 ; tecza et al .
2000 ; beswick et al . 2001 ) , and is well known for its unusually luminous h@xmath0 emission in the near infrared ( e.g. , rieke et al . 1985 ; depoy , becklin , & wynn - williams 1986 ) .
although many excitation mechanisms for the h@xmath0 have been proposed to date [ shock excitation ( rieke et al . 1985 ; depoy et al .
1986 ; lester , harvey , & carr 1988 ; elston & maloney 1990 ; herbst et al .
1990 ; van der werf et al . 1993 ; sugai et al .
1997 ) ; uv fluorescence ( tanaka , hasegawa , & gatley 1991 ) ; x - ray heating ( draine & woods 1990 ) ; and formation pumping ( mouri & taniguchi 1995 ) ] , h@xmath0 line ratio analyses have revealed that shock excitation is responsible for almost all the h@xmath0 emission ( sugai et al .
1997 ; ohyama et al .
2000 ; tecza et al .
however , because of its huge h@xmath0 luminosity and exceptionally large h@xmath0/br @xmath3 intensity ratio , the origin of ngc 6240 s h@xmath0 emission has been the subject of considerable debate ; star formation activity is not strong enough to power all the observed h@xmath0 luminosity ( e.g. , rieke et al .
1985 ; draine & woods 1990 ) .
interestingly , most of the h@xmath0 emission comes from the region between the nuclei , although both br @xmath3 and [ fe ii ] ( originating from ob type stars and supernovae in star - forming regions ) are detected from each nucleus ( herbst et al . 1990 ; van der werf et al .
1993 ; sugai et al .
1997 ; ohyama et al .
2000 ; tecza et al .
two possible models have been proposed to explain this intense , off - nuclear , shock - excited h@xmath0 emission . in the first
, the emission is caused by shock heating at the interface between the two colliding galaxies ( tecza et al .
2000 and references therein ) , while in the second it is due to the interaction between the hot gas associated with the superwind activity of the s nucleus and the tidally - produced molecular gas concentration between the two nuclei ( ohyama al .
2000 ) .
recent high - resolution , sensitive co observations of ngc 6240 at radio wavelengths have revealed that a significant fraction of the system s molecular gas is concentrated not at the two nuclei but rather occupies the region between them , forming a massive thick disk with highly turbulent motion ( tacconi et al .
hereafter we refer to the co gas between the nuclei as the off - nuclear molecular gas or molecular gas concentration .
a violent tidal disturbance during the merging process of the two nuclei could be responsible for the formation of such an off - nuclear molecular gas disk , although the creation of such a massive concentration of gas away from the merging nuclei has no theoretical explanation ( e.g. , barnes 2002 and references therein ; see also tecza et al .
the peak of the h@xmath0 emission is located between the s nucleus and the co peak ( ohyama et al . 2000 ; tecza et al . 2000 ) . the first step in exploring the reason for the source s unusually intense h@xmath0 emission was the analysis of the velocity field and the excitation conditions of the h@xmath0 lines ( ohyama et al .
2000 ; tecza et al . 2000 and
references therein ) . in this paper
we take the second step by conducting a detailed comparison of the morphological and kinematical conditions of h@xmath0 and co lines .
this work is motivated by the fact that h@xmath0 and co are molecular species with different excitation conditions ; comparisons of their emission lines should give us some new insights on the conditions under which high h@xmath0 emitting efficiency ( h@xmath0 flux per molecular gas content ) is likely .
such a comparison has only recently become possible , as a result of the availability of sub - arcsecond , sensitive mapping observations of both co and h@xmath0 .
tecza et al .
( 2000 ) conducted three - dimensional h@xmath0 mapping observations of this galaxy under sub - arcsec seeing conditions , while tacconi et al .
( 1999 ) carried out high - resolution co ( @xmath4 ) mapping observations with a sub - arcsec beam .
both data have been re - sampled to make a direct comparison between the cold and warm molecular gas at the same scale ( tecza et al .
based on these data , we discuss how h@xmath0 emission can be excited efficiently , focusing on the h@xmath0/co intensity ratio as an indicator of the h@xmath0 emitting efficiency .
all data that we analyze in this paper were reproduced from figure 3 of tecza et al .
( 2000 ) , in which grids of both the co ( @xmath4 ) and h@xmath0 velocity profiles were presented .
these profiles were reproduced from the original data of both tecza et al .
( 2000 ) and tacconi et al .
( 1999 ) for h@xmath0 and co , respectively .
the velocity resolution of the original h@xmath0 data was 150 km s@xmath5 , while the spatial resolution was 0.8 - 1.0 , the seeing conditions at the time of observation ( tecza et al . 2000 ) . in the co observations ,
the full width at half maximum ( fwhm ) of the elliptical gaussian was beam 0.9@xmath6 0.5 , while the velocity resolution was 6.5 km s@xmath5 .
these data were re - sampled onto @xmath7 arcsec@xmath8 sub - regions ( @xmath9 pc@xmath8 ) around the double nuclei over @xmath10 arcsec@xmath8 ( @xmath11 kpc@xmath8 ) regions , and re - sampled at 140 km s@xmath5 velocity resolution ( tecza et al
these data enabled us to compare the co and h@xmath0 velocity profiles directly , with a resolution that was sufficient to separate the two merging nuclei and the surrounding regions .
the velocity profiles were integrated over blue and red velocity regions ( @xmath12 km s@xmath5 and @xmath13 km s@xmath5 for blue and red , respectively , where @xmath14 is the systemic velocity ) in each sub - region to measure the blue and red intensities of co and h@xmath0 [ @xmath15(co ) , @xmath16(co ) , @xmath15(h@xmath0 ) , and @xmath17(h@xmath0 ) ] .
the velocity cut - offs at @xmath18 km s@xmath5 were applied in the integration because it is necessary to measure the intensities in the same velocity range for all sub - regions in order to make a fair comparison of their spatial distribution .
we found that some sub - regions had a weak co intensity at larger velocity offsets from the systemic velocity , so that the intensity could not be accurately measured out to the velocity cut - off .
we eliminated the integrated intensity data of the sub - regions from the following analyses if the normalized intensity of co ( normalized to be unity at the co peak position : see figure 3 of tecza et al .
( 2000 ) and figure 2 of this paper ) was 0.05 ( the typical noise level ) or less at the cut - off velocity .
the cut - off velocity ( @xmath18 km s@xmath5 ) was chosen as a compromise : although a larger cut - off velocity is desirable for studying the properties of the high - velocity gas components , applying a larger ( @xmath19 or @xmath20 km s@xmath5 ) velocity cut - off would discard a larger number of sub - regions due to a lack of useful integrated intensity data . note that since the observed line widths ( @xmath21 km s@xmath5 and @xmath22 km s@xmath5 for co and h@xmath0 at full width at zero intensity , respectively : tacconi et al . 1999 ; tecza et al .
2000 ) are wider than the applied velocity cut - off ( at @xmath18 km s@xmath5 ) , about 19% and 33% of the total co and h@xmath0 fluxes fall outside the velocity range for the integration , and were not counted in the analysis that follows .
single gaussian profile fitting was used to measure the velocities of both lines [ @xmath23(co ) and @xmath23(h@xmath0 ) ] , although some areas show more complicated emission line profiles , such as a double - peaked profile ( tecza et al .
figure 1 shows the total ( blue+red ) integrated intensity and the velocity maps of both co and h@xmath0 [ @xmath15(co)@xmath24(co ) , @xmath25(h@xmath0)@xmath24(h@xmath0 ) , @xmath23(co ) , and @xmath23(h@xmath0 ) ] .
the peaks of both lines are located between the two nuclei and slightly east of the s nucleus ( @xmath26 ) .
map presented by tecza et al .
( 2000 ) and other previous h@xmath0 maps ( e.g. , sugai et al .
1997 ) showed the h@xmath0 peak between the double nuclei , although our figure 1 shows the peak at a slightly different position .
this is an artefact that was probably caused by the re - sampling of the data onto the @xmath27 sub - regions adopted in producing figure 3 of tecza et al .
( 2000 ) and our figure 1 .
a similar artefact is also seen in the map of co. the adopted velocity range for integrating the profiles ( from @xmath28 km s@xmath5 to @xmath29 km s@xmath5 ) may also affect the positions of the co and h@xmath0 peaks in our maps .
] the velocities of both lines change along pa@xmath30 deg , in the sense that they are blueshifted around the s nucleus and are redshifted in both the ne and sw regions .
simple models of the velocity field , such as a rotation or a pure expansion / contraction , have difficulty reproducing these observations . to compare the properties of co and h@xmath0 emissions , we here introduce a new index , the h@xmath0 emitting efficiency .
the efficiency is defined as the intensity ratio of h@xmath0 and co [ @xmath2(h@xmath0)/@xmath2(co ) ] , and was calculated over the entire velocity range on each sub - region ( figure 2 ) .
note , however , that these ratios are shown only at velocities where the normalized co intensity is larger than 0.05 ( the typical noise level ) to avoid obtaining erroneous ratios in regions or velocities where co is weak .
next , the integrated h@xmath0 emitting efficiencies for blue [ @xmath31 , integrating over the range @xmath28 to @xmath32 km s@xmath5 ] , for red ( @xmath34 , integrating over the range @xmath35 to @xmath36 km s@xmath5 ) , and for blue plus red ( total ) ( @xmath37 , integrating over the range @xmath28 to @xmath36 km s@xmath5 ) were calculated , in order to examine the spatial and velocity distributions of the efficiencies . here
@xmath38 is the velocity range for the integration , i.e. , @xmath39 km s@xmath5 for blue- and red- , and @xmath40 km s@xmath5 for total - integrated h@xmath0 emitting efficiencies .
a velocity cut - off ( at @xmath18 km s@xmath5 ) was applied in integrating the h@xmath0 emitting efficiency , in the same way as for the calculation of the integrated intensities .
also , the integrated h@xmath0 emitting efficiencies in sub - regions where the normalized intensity of co at the cut - off velocity was less than 0.05 were not used in the following analysis .
the blue - to - red ratio of the h@xmath0 emitting efficiency ( @xmath41/@xmath34 ) could then be calculated .
these values ( @xmath37 and @xmath41/@xmath34 ) are mapped together with the velocity difference between co and h@xmath0 [ @xmath42(h@xmath0)@xmath43(co ) ] in figure 3 .
figure 3 illustrates our main results : ( 1 ) the total h@xmath0 emitting efficiency ( @xmath37 ) is smaller around the co peak , and becomes higher with distance from the peak ; ( 2 ) the blue - to - red ratio of the h@xmath0 emitting efficiency ( @xmath41/@xmath34 ) is smaller around the co peak , and becomes larger with distance from the peak ; and ( 3 ) the velocity difference ( @xmath44 ) is smaller around the co peak , while the h@xmath0 emission becomes more blueshifted with respect to co with distance from the peak .
all these trends are seen almost systematically as a function of position , and are evident well over our resolution ( @xmath45 in space and 140 km s@xmath5 in velocity ) .
it seems noteworthy that the peak of h@xmath0 emitting efficiency corresponds neither to the peak position of co nor to that of h@xmath0 .
we searched for correlations between the measured quantities ( @xmath46 and @xmath41/@xmath34 ) ; the results are shown in figures 4 and 5 . in these plots ,
data are marked with different symbols for different groups of sub - regions , which were divided according to their spatial positions around the co intensity peak : the central region with higher h@xmath0 and co intensities between the two nuclei ( nuclear region ) , the region northeast of the central region ( ne region ) , and the southwestern region ( sw region ) .
two other sub - regions , the most southwesterly regions where h@xmath0 intensity increases with distance from the nucleus ( the sw sub - group ) and the most northeasterly regions with double - peaked profiles ( the ne sub - group ) , are grouped separately .
these sub - regions are also indicated in figure 2 .
we found a tight negative correlation between @xmath2(co ) and @xmath37 ( figure 4 ) .
a spearman rank - order test for the correlation between log[@xmath37 ] and log[@xmath2(co ) ] showed that the probability @xmath48 of consistency with the null hypothesis ( i.e. , no correlation between the two quantities ) is less than 0.001 for any sub - grouping ( all the regions , all the regions except the sw sub - group , and all regions except for both sw and ne sub - groups ) .
therefore we consider that we have definitely shown that the total h@xmath0 emitting efficiency is enhanced in the outer regions with weaker co emission .
two other positive correlations were found , between @xmath49 , @xmath41/@xmath34 , and @xmath37 ( figure 5 ) .
although these correlations look less tight , a spearman rank order test showed that the probability @xmath48 of the null hypothesis is 0.20 , 0.02 , and 0.04 for the above three sub - groupings , respectively , in @xmath44 vs. @xmath41/@xmath50 diagram , and 0.02 , 0.001 , and @xmath510.001 for the same three sub - groupings , respectively , in @xmath44 vs. @xmath46 diagram , indicating the presence of significant correlations between these quantities at greater than the 98% significance level if we exclude the sw sub - group .
our analysis has shown that the h@xmath0 emitting efficiency and the kinematical conditions of the warm gas are closely related .
such a correlation is a good indication of shock heating , in which bulk kinetic energy is converted to energy radiated through the h@xmath0 line .
although our analyses were carried out only for fluxes within the velocity range of @xmath52 km s@xmath5 , it seems reasonable to interpret the correlation as an indication of the major mechanism of the excitation of h@xmath0 , because this velocity range contains 81% and 67% of the total fluxes of co and h@xmath0 , respectively ( see section 2 above ) . in some inner sub - regions where we can trace the h@xmath0 emitting efficiency at even bluer velocities ( @xmath53 km s@xmath5 )
, we found that the efficiency is generally higher at bluer velocity ( @xmath54 km s@xmath5 ) than at lower velocity ( @xmath55 km s@xmath5 ) , and this fact supports a model in which shock excitation of h@xmath0 is a major excitation mechanism around ngc 6240 .
although other h@xmath0 excitation mechanisms ( e.g. , uv fluorescence , x - ray heating , and formation pumping ) can not be rejected , their contribution to the total must be less important . we note that the same conclusion has previously been reached based on a line - ratio analysis of the near - infrared spectra ( sugai et al . 1997
; tecza et al . 2000 ; and ohyama et al . 2000 ) .
since correlations between the h@xmath0 emitting efficiency and the kinematical conditions of the molecular gas were found not only on the sw but also on the ne side of the co peak , on scales of @xmath56 , the molecular gas concentration seems likely to be surrounded by gas that has a simple global motion .
the motion is probably directed predominantly toward the observer , and the interaction between the gas that has this blueshifted motion and the molecular gas concentration would increase the h@xmath0 emitting efficiency on the bluer velocity side of the co emission .
the question then arises : what is the origin of this blueshifted gas motion ? in the following , we discuss two possible explanations .
one possibility is that the merging nuclei are interacting with the off - nuclear molecular gas as they rotate around each other .
although there are no good theoretical explanations for the formation of such an off - nuclear molecular gas concentration in merging galaxies ( e.g. , barnes 2002 and references therein ; see also tecza et al .
2002 ) , here we assume that such concentration has formed as a result of tidal effects during the merging process .
since the h@xmath0 emitting efficiency is relatively higher at the sw portion of the concentration , a collision between the s galaxy and the concentration might contribute to the excitation of h@xmath0 emission .
the extinction maps ( scoville et al .
2000 ; tecza et al .
2000 ) have revealed the presence of a dusty region at 0 1.0 nne of the s nucleus , and the position of the maximum extinction ( @xmath57 0.2 0.3 nne of the s nucleus ) coincides spatially with the co peak .
this indicates that the s galaxy is behind the dusty molecular gas concentration from our point of view . since the s nucleus is known to be moving toward us around the concentration ( fried & ulrich 1985 ; tecza et al .
2000 ) , a violent collision between the two would enhance the blueshifted h@xmath0 emission at the sw side of the co peak .
if this is the case , however , the redshifted h@xmath0 emission , rather than the blueshifted component , would be enhanced on the opposite side of the co peak , because the n galaxy is rotating away from us on the opposite side of the concentration .
this model therefore can not explain the fact that blue - enhanced h@xmath0 emitting efficiency is observed on both sides of the co peak .
the alternative model is that an outflow associated with the superwind activity of the s nucleus interacts with the off - nuclear molecular gas concentration .
the superwind activity of this galaxy has been recognized as a result of its large - scale ( @xmath58 kpc ) shock - excited filamentary nebulae ( e.g. , armus , heckman , & miley 1990 ; keel et al . 1990 ) and the expanding motion of the h@xmath0 emitting clouds around the nuclear region ( ohyama et al .
2000 ) . since both of the individual galaxies are known to show flat rotation curves in their stellar velocity fields ( tecza et al .
2000 ) , disk structures are likely to be present around each nucleus , even though the system is an advanced merger .
therefore we expect any superwind outflow to exhibit a bi - directional conical shape pointing toward the disk polar direction ( e.g. , tomisaka & ikeuchi 1988 ; heckman et al .
1990 ; suchkov et al .
1994 ; strickland & stevens 2000 ) , although the shape of the outflow might be distorted due to the inhomogeneous distribution of the medium around the nucleus .
an @xmath59band image of the s nucleus reveals an elongated structure in the nw - se direction , suggesting the presence of a disk whose minor axis is directed ne - sw at an inclination angle of @xmath60 deg ( scoville et al .
2000 ; tecza et al . 2000 ) . also , as mentioned earlier , the s nucleus seems to be located behind the molecular gas concentration . assuming that the off - nuclear molecular gas concentration formed during the merging process of the two nuclei , before the superwind started to blow
, we would expect that , if a bi - conical superwind outflow emanated from the s nucleus , it would interact with the gas concentration from the rear .
this interaction would produce blueshifted h@xmath0 emission around the co peak ( i.e. , not only at the sw but also at the ne sides of the co peak ) if the opening angle of the outflow were wide enough to cover the whole molecular gas concentration .
the oppositely directed cone of the superwind , pointing away from us , would not contribute significantly to the h@xmath0 luminosity , as it would not interact with the off - nuclear molecular gas .
since this model has no major difficulties in reproducing the observations , we regard it as more plausible than the model in which the s nucleus is colliding with the molecular gas concentration . a schematic picture of the superwind model is shown in figure 6 .
note that the real situation would be much more complicated due to the complicated medium distribution around the merging nuclei . following heckman et al .
( 1993 ) , clouds accelerated by the superwind outflow will have a terminal velocity of @xmath61 km s@xmath5 , where @xmath62 is the bolometric luminosity in units of @xmath63 @xmath64 , @xmath65 is the initial distance of the cloud from the nucleus , and @xmath66 is the column density of the interstellar clouds in units of @xmath67 @xmath68 .
adopting @xmath69 ( sanders & mirabel 1995 ) , @xmath70 kpc ( the projected distance between the s nucleus and the co peak multiplied by a de - projection factor of @xmath71 ) , and @xmath72 as a minimum acceptable value corresponding to the typical value for the milky way .
we also adopt @xmath73 as a maximum acceptable value corresponding to one tenth of the total column density inferred from molecular gas observations of @xmath74 ( tacconi et al .
, this equation implies that the clouds would be blown off at a speed of between a few and several hundreds of km s@xmath5 , which is almost comparable to the observed line width of h@xmath0 ( @xmath75 km s@xmath5 in fwhm : e.g. , ohyama et al .
2000 ; tecza et al .
2000 ) and co ( @xmath76 km s@xmath5 in fwhm : tacconi et al .
1999 ) . since the column density is likely to decline outward around the co peak , falling off in a similar way to the co surface brightness , ( figure 1a ) , i.e. , the column density of each cloud is larger near the co peak and is smaller at the outer region , so that clouds in the outer part of the molecular gas concentration would be accelerated up to faster speeds than those near to the co peak .
therefore the observed trend in the relatively blueshifted h@xmath0 emission with respect to the co emission in the outer part of the molecular gas concentration can be explained naturally by the model of superwind - molecular gas concentration interaction .
this fact further suggests that kinematical disturbance due to the superwind also raises the h@xmath0 emitting efficiency within the molecular gas concentration , given the positive correlation between the h@xmath0 emitting efficiency and @xmath44 .
it has been pointed out that a c - type shock ( @xmath77 km s@xmath5 ) , rather than a j - type shock with faster shock velocity , is responsible for exciting most of the h@xmath0 emission , although most of the emission comes from clouds within a faster global velocity field ( @xmath78 km s@xmath5 ) ( sugai et al .
one way to resolve this apparent contradiction is to suggest that internal cloud - cloud collisions within the clumpy medium , at speeds of @xmath79@xmath80 km s@xmath5 , excite the h@xmath0 emission , while the clouds move along the global velocity field as a whole ( tecza et al .
since the molecular gas concentration is known to form a turbulent thick rotating disk ( tacconi et al .
1999 ) , it is likely that the concentration is actually composed of numerous smaller clouds ( tecza et al .
a high - resolution velocity dispersion map of the co ( figure 4 of tacconi et al . 1999 ) reveals the presence of regions with higher velocity dispersion just around the co peak ( especially to the se and nw sides of the co peak ) .
therefore it seems very likely that the interaction of small clouds within the molecular gas concentration with the external superwind provides an enhanced random velocity at the outer part of the co concentration , as well as the blueshift of the h@xmath0 with respect to the co. one of the merits of utilizing cloud - cloud collisions within the molecular gas concentration to give rise to the h@xmath0 emission is that a higher efficiency of energy deposition into the molecular gas clouds from the supernova explosions can be expected as a result of the cloud - crushing mechanism ( cowie , mckee , & ostriker 1981 ) , and the observed huge h@xmath0 luminosity can be explained ( draine & woods 1990 ; elston & maloney 1990 ) .
this therefore raises the question of whether the observed trends in h@xmath0 emitting efficiency , as well as the total h@xmath0 luminosity , can be explained in terms of the cloud - crushing mechanism . in this model
, we may consider , as a simple case , that some fraction of the kinetic energy of the small clouds , randomly moving within the bulk motion of the rotating molecular gas concentration , is dissipated through internal shocks as a result of cloud - cloud collisions , and gives rise to h@xmath0 emission .
then , the h@xmath0 surface brightness at a velocity of @xmath81 can be expressed as @xmath82 , where @xmath83 is the surface mass density of the molecular gas concentration whose velocity is @xmath81 , @xmath84 is the mass fraction of the small clouds that give rise to h@xmath0 emission through the cloud - crushing mechanism at a velocity @xmath81 , @xmath85 is the mean of the squared internal random velocity of the small clouds , @xmath86 is the fraction of the energy dissipated in the shock which is radiated in the h@xmath0 1 - 0 @xmath87(1 ) line , and @xmath88 is the timescale of the cloud - cloud collisions . since the surface mass density of the clouds @xmath83 is proportional to the observed co surface brightness [ @xmath89 at the velocity of @xmath81 , the h@xmath0 emitting efficiency can be expressed as @xmath90 , by definition .
since we are considering a situation in which the cloud - crushing mechanism is responsible for _ all _ the h@xmath0 emission , we may substitute a constant @xmath86 ( @xmath91 ) and a constant internal velocity ( @xmath92 km s@xmath5 ) into the equation , representing an extreme case of the cloud - crushing mechanism working at its maximum efficiency ( draine & woods 1990 ) .
then , the h@xmath0 emitting efficiency can be expressed as @xmath93 .
this expression indicates that the h@xmath0 emitting efficiency is higher when ( 1 ) a larger mass fraction of the clouds is involved in the production of h@xmath0 emission through the cloud - crushing mechanism , and/or ( 2 ) the collision frequency is higher ( i.e. , the collision timescale is shorter ) . the observed close relationship between the h@xmath0 emitting efficiency and the kinematical conditions in the molecular gas clouds
can be explained by this expression , if the cloud - crushing mechanism is efficient : the outer part of the off - nuclear molecular - gas concentration is likely to be more strongly influenced kinematically by the interaction with the external superwind ( see section 3.2 above ) , and the h@xmath0 emitting efficiency is expected to be higher , since a larger mass fraction of the clouds will produce h@xmath0 as a result of internal cloud - cloud shocks and a higher frequency of the cloud - cloud collisions . in addition , the efficiency is expected to be higher at bluer velocities , because these are likely to be caused as a result of a more violent kinematical influence from the superwind .
therefore , the cloud - crushing mechanism within a molecular gas concentration disturbed by the interaction with the external superwind outflow seems to be able to reproduce the observed trends of the h@xmath0 emitting efficiency in both space and velocity , as well as the higher - than - normal h@xmath0 luminosity . although the mechanism by which the off - nuclear molecular gas concentration between the two merging nuclei is created is not clear , the presence of such a gas concentration is necessary if the model is to reproduce the observed intense and efficient h@xmath0 emission .
we look forward to future theoretical studies of the origin of such a gas concentration . | we report on our new analysis of the spatial and kinematical distribution of warm and cold molecular gas in the nearby , prototypical , luminous infrared galaxy ngc 6240 , which was undertaken to explore the origin of its unusually luminous h@xmath0 emission .
the gas components are known to be distributed between the two merging nuclei , forming an off - nuclear molecular gas concentration . by comparing three - dimensional emission - line data ( in space and velocity ) of co ( j=@xmath1 ) in the radio and h@xmath0 in the near infrared , we are able to search for the spatial and kinematical conditions under which efficient h@xmath0 emission is produced in much more detail than has previously been possible . in particular ,
we focus on the h@xmath0 emitting efficiency , defined in terms of the intensity ratio of h@xmath0 to co [ @xmath2(h@xmath0)/@xmath2(co ) ] , as a function of velocity .
we derive this by utilizing the recent high - resolution three - dimensional data presented by tecza et al .
( 2000 ) .
the integrated h@xmath0 emitting efficiency is calculated by integrating the velocity profile of h@xmath0 emitting efficiency in blue , red , and total ( blue + red ) velocity regions of the profile .
we find that ( 1 ) both the total h@xmath0 emitting efficiency and the blue - to - red ratio of the efficiency are larger in regions surrounding the co and h@xmath0 intensity peaks , and ( 2 ) the h@xmath0 emitting efficiency and the kinematical conditions in the warm molecular gas are closely related to each other .
we compare two possible models that might explain these characteristics : a large - scale collision between the molecular gas concentration and the merging nuclei , and a collision between the molecular gas concentration and the external superwind outflow from the southern nucleus .
the latter model seems more plausible , since it can reproduce the enhanced emitting efficiency of blueshifted h@xmath0 around the molecular gas concentration , if we assume that the superwind blows from the southern nucleus toward us , hitting the entire gas concentration from behind . in this model ,
internal cloud - cloud collisions within the molecular gas concentration are enhanced by the interaction with the superwind outflow , and efficient and intense shock - excited h@xmath0 emission is expected as a result of the cloud - crushing mechanism .
the observed spatial distribution of the h@xmath0 emitting efficiency can be explained if there is a greater kinematical disturbance in the outer part of the molecular gas concentration , as a result of the interaction with the superwind outflow , and also more frequent cloud - cloud collisions in the region .
in addition , the kinematical influence of the superwind on the molecular gas concentration should be larger at bluer velocities , and the collision frequency is expected to be larger at bluer velocities , explaining the relationship between velocity and the h@xmath0 emitting efficiency . |
entanglement is a fundamental property of quantum systems and a basic resource for quantum information and computation .
however , its detection and quantification are solved only in some simple cases , like the one of two - qubit systems .
discriminating separability from entanglement for higher dimensional non - pure two - particle states seems to be in the complexity class np - hard @xcite .
finding good measures of entanglement is a related nontrivial problem , even for the next simplest possibilities : non - pure two - particle states in arbitrary dimensions or three - qubit non - pure states . for three - qubit pure states some authors @xcite ,
@xcite divide the infinite set of equivalence classes under local unitary ( lu ) transformations in subsets characterized by particular values of some lu invariants : these invariants are not directly related with entanglement measures , and therefore this is not exactly a classification of entanglement types .
other authors @xcite , @xcite classify three - qubit states in equivalence classes under stochastic local operations and classical communication ( slocc ) ; some of the types considered in other classifications , like the _ star - shaped states _ introduced in @xcite , @xcite do not have their own class in this scheme .
several entanglement measures for pure states of three - particle systems have been proposed @xcite , @xcite , @xcite , @xcite , @xcite , @xcite ; as we will discuss in [ sec:3 ] , these measures do not distinguish adequately between fully entangled and separable states .
extensions to mixed states are even more problematic ; we will discuss a measure that improves on these results .
the paper is organized as follows . in [ sec:1 ]
we summarize some known results for the entanglement of two and three - qubit systems , including definitions and notation that we will use in the following sections . in [ sec:2 ]
we introduce a classification for arbitrary three - qubit states in terms of their three - qubit and reduced two - qubit entanglements , with a graphic representation of the different types for pure states . in [ sec:3 ]
we comment several previous measures of tripartite entanglement , and introduce a multiplicative generalization of two - particle _ negativity_. in [ sec:4 ] we apply a generalized schmidt decomposition ( gsd ) @xcite to pure three - qubit states and specify the form of the states in each of our classes ; this could be used to give a physical interpretation to the abstract classes of @xcite . in [ sec:5 ] , we calculate the tripartite negativity of some families of mixed states and compare our results with previously published ones . we conclude in [ sec:6 ] with a summary of our results .
arbitrary states of pure two - qubit systems @xmath0 , @xmath1 @xmath2 being a basis of the hilbert space @xmath3 for qubits a and b , can always be converted in @xmath4 , by means of lu transformations @xmath5 , being @xmath6 and @xmath7 unitary operators acting on @xmath8 and @xmath9 respectively .
each pair @xmath10 of this schmidt decomposition define a lu equivalence class ; this infinite set can be divided in two big subsets : _ separable states _ if one of the @xmath11 is zero , and _ entangled states _ otherwise ; generalization to pure states in arbitrary dimensions is easy . for pure states , separability equals factorizability . for the non - pure case ,
a state @xmath12 is separable @xcite if it can be written as @xmath13 , where @xmath14 and @xmath15 are state operators of subsystems a and b respectively , @xmath16 @xmath17 , @xmath18 ; otherwise @xmath12 is entangled .
a list of criteria for separability in two - particle systems can be found in @xcite .
sometimes @xcite , @xcite , separable but not factorizable mixed states are qualified as _
classically correlated _ ; we will not use this distinction , because we are interested only in quantum correlated states .
we shall cite only three of the several entanglement measures of pure two - qubit systems : _ von neumann s entropy _
@xcite of reduced states , _ wootters concurrence _ @xcite , and _ negativity _ @xcite .
the von neumann s entropy of a state @xmath12 is defined in information theory as @xmath19 , @xmath20 being the eigenvalues of @xmath12.the reduced states are @xmath21 and @xmath22 .
von neumann s entropy of reduced states is the simplest measure of two - particle entanglement for pure states , but its extension to non - pure states discriminates between separable and entangled states only if the _ mutual entropy _ or _ correlation index _ is zero @xcite .
therefore , it is not a good measure of entanglement for general mixed states .
concurrence is defined as @xmath23 where @xmath24 @xmath25 are the eigenvalues of @xmath26 in decreasing order , with @xmath27 , @xmath28 being the complex conjugate of the state operator @xmath12 . in arbitrary dimensions ,
simple computable generalizations of concurrence are known only for pure states @xcite ; there is no efficient way known @xcite to calculate sophisticate generalizations like biconcurrence @xcite .
the negativity of a bipartite state @xmath12 was introduced in @xcite ; we will use the convention of @xcite , that is twice the value of the original definition : @xmath29 where @xmath30 are the negative eigenvalues of the partial transpose @xmath31 of the total state @xmath12 respect to the subsystem a , defined as @xmath32 , a and b denoting the two subsystems . for pure bipartite states of arbitrary dimensions
the negativity ( 1 ) is equal to the concurrence @xcite .
the negativity is not additive @xcite and some authors prefer to use the logarithmic negativity , which is additive but not convex @xcite .
the negativity can be evaluated in the same way for pure and non - pure states in arbitrary dimensions , although there are entangled mixed states with zero negativity in every dimensions except @xmath33 and @xmath34 @xcite , @xcite .
no measure discriminating separable from entangled states in the general non - pure case is known .
however , non - zero negativity is a sufficient condition for entanglement . for three qubits
, there is no ternary schmidt decomposition of pure states in a strict sense , but there is a generalized schmidt decomposition ( gsd ) @xcite : arbitrary states can always be converted to states that contain at most five of the eight @xmath35 terms of the hilbert basis , by means of a very simple algorithm .
each set of coefficients @xmath36 define a lu equivalence class ; in [ sec:2 ] and [ sec:4 ] we deal with the problem of dividing the infinite set of lu equivalence classes in big subsets corresponding to the different types of entanglement in our classification . for three - particle systems ,
a pure state @xmath37 is _ fully separable _ if it can be written as @xmath38 , _ biseparable _ if it is not fully separable but can be written as @xmath39 with @xmath40 entangled ( the index a can denote any of the three subsystems ) , and _ fully inseparable _ otherwise ; in this last case we will say that the state has _ full tripartite entanglement_. for three - particle systems a non - pure state operator @xmath12 may be fully separable , biseparable or fully inseparable ; biseparable states can be _ simply biseparable _ or _
generalized biseparable_. fully separable states can be written as @xmath41 .
are not fully separable but can be written as @xmath42 where @xmath43 runs from a to c and @xmath44 from @xmath45 to @xmath46 respectively and at least one @xmath47 is entangled @xcite , @xcite ; simply biseparable states have @xmath48 only for a single value of @xmath43 ( one single qubit is separable from the other two , that are entangled ) ; generalized biseparable states are convex sums of states of the previous kind ( non vanishing coefficients @xmath49 for more than one @xmath43 ) . fully inseparable states are those not fully separable nor biseparable .
not all authors agree with this classification of mixed three - qubits : for some of them , biseparable states were only those that we call simply biseparable , and generalized biseparable states were included in the fully inseparable class @xcite , @xcite .
fully separable estates contain no quantum entanglement .
simply biseparable states have bipartite entanglement in a single pair of qubits ; they have _ partial bipartite entanglement_. generalized biseparable states contain bipartite entanglement in more than one pair of qubits ; they have _ distributed bipartite entanglement_. fully inseparable states have full tripartite entanglement ( true nonclassical 3-particle correlations , in the notation of @xcite ) .
the characterization of separability or biseparability for non - pure three - particle states , and the measure of their full tripartite entanglement are nontrivial , even in the simplest case of three qubits .
an ideal measure of tripartite entanglement would discriminate fully entangled from fully separable or biseparable states ; no such measure is known yet , even for three qubits . finally , entanglement of the reduced two - particle states will be called the _ reduced entanglement _ of the pair .
they show how much two - qubit entanglement remains when the third qubit is not observed .
as we will show in the next section , for non - pure states there are unexpected results ; for instance , some simply biseparable mixed states have no reduced entanglement when the separable qubit is traced over .
we propose a classification of three - qubit states based in the presence of no entanglement ( full separability ) , bipartite entanglements only ( simple or generalized biseparability ) , or full tripartite entanglement ( true 3-qubit entanglement , full inseparability ) , both for pure and non - pure states , with subtypes based on the number of entangled reduced two - qubit states .
reduced , bipartite and full tripartite entanglements have direct physical meaning and are lu invariants : local operations can not create entanglement , and therefore all the invertible local operations ( unitary ones in particular ) must leave these entanglements invariant .
thus , a state in any of our subtypes can not be transformed by a lu to a state in any other subtype .
this classification is equally valid for pure and non - pure states , although there are several subtypes that exist only for non - pure ones .
the practical implementation of the classification is also more difficult for mixed states , since the detection of tripartite entanglement is not completely solved in this case .
another difference is that for pure states we can relate these subtypes with the coefficients of a gsd , as we will show in [ sec:4 ] , while no similar decomposition exists for non - pure states .
fig.1 summarizes the different types and subtypes in this classification for pure states ; with the same conventions , we could draw a similar graphic for the subtypes that exist only for mixed states ; we omit it for simplicity .
[ fig:1 ] we will briefly comment now the different subtypes for general states ( pure or not ) . * type @xmath50 : fully separable states , no quantum entanglement .
* type 1 : biseparable states , bipartite entanglements only .
* subtype @xmath51 : simply biseparable states with no reduced entanglement .
two of the qubits are entangled when the total three - qubit state is considered , but when the other qubit is traced over , the reduced two - particle state is separable .
this subtype exists only for non - pure states : for instance , a biseparable state + @xmath52 , where @xmath53 are bell states , has a reduced state @xmath54 that is separable : @xmath55 ( fig.1 ) without the segment connecting the lower two qubits .
* subtype @xmath56 : the three - qubit state is simply biseparable , and when the separable qubit is traced over , the other two remain entangled .
all pure biseparable states are of this subtype : any pure state @xmath57 , with @xmath58 entangled , gives an entangled reduced state , @xmath59 there are also non - pure states of this subtype : for instance , the family of biseparable states @xmath60 , with negativities @xmath61 ; therefore these reduced states are entangled for @xmath62 and the three qubit state @xmath63 is in subtype @xmath56 ; for @xmath64 , @xmath65 is the state previously considered as an example of subtype @xmath51 ; note that this @xmath51 state is a continuous limit ( @xmath66 ) of states of subtype @xmath56 . *
subtype @xmath67 : generalized biseparable states with bipartite entanglements in two pairs of qubits , for instance @xmath68 , with @xmath69 and at least one @xmath47 entangled for each kl . according to the posible entanglements of their reduced states the following subtypes are possible in principle : @xmath70 , @xmath71 , @xmath72 .
we omit the graphic representation of these subtypes .
* subtype @xmath73 : generalized biseparable states with bipartite entanglements in the three pairs of qubits , for instance : @xmath74 , with @xmath75 and the three @xmath76 entangled .
according to the posible entanglements of their reduced states the following subtypes are possible in principle : @xmath77 , @xmath78 , @xmath79 , @xmath80 .
we omit also the graphic representation of these subtypes .
generalized biseparable states of subtypes @xmath67 and @xmath73 exist only in the non - pure case . *
type 2 : fully inseparable states , non - zero full tripartite entanglement .
subtypes @xmath81 to @xmath82 exist for pure and non - pure states . * subtype @xmath81 : their three reduced entanglements are zero .
we will call them _ ghz - like states _ , because the well known ghz states ( see [ sec:4 ] ) belong to this subtype .
the entanglement of ghz - like states disappears if any of the three qubits is traced over ; their entanglement is _
fragile_. * subtype @xmath83 : one reduced entanglement is non - zero , and the other two are zero . * subtype @xmath84 : two reduced entanglements are non - zero . for pure states they have been called _
star shaped states _
subtype @xmath82 : their three reduced entanglements are non - zero .
we will call them _ w - like states _ because the well known @xmath85 states @xcite ( see also [ sec:4 ] ) belong to this subtype .
the entanglement in these states survives the loss ( tracing over ) of any of the three qubits ; it is _
robust_. for pure states
our classification is very easy to implement .
a qubit is factorizable if and only if the reduced state of the other two qubits is pure ( see for instance chapter 8 of @xcite ) . in the affirmative case ,
this pure two - qubit state is factorizable if and only if their reduced one - qubit states are pure ; the total state is then fully separable ( subtype @xmath50 ) ; otherwise the reduced two - qubit state is entangled and the total state is biseparable ( type @xmath56 ) .
if no qubit is factorizable , the total state is fully entangled ( type 2 ) ; the subtypes @xmath81 , @xmath83 , @xmath84 , or @xmath82 can be ascertain calculating the negativities or concurrences of their reduced two - qubit states to determine the number of entangled pairs .
no ambiguity remains in the qualitative application of our classification to pure states . in [ sec:4 ]
we will show the results for all possible gsd s of pure three - qubit states , obtaining the explicit form of the vectors in each one of our classes , and the quantitative values of their entanglements .
for non - pure states the situation is more complicated .
we do not know of any measure that would identify unambiguously entanglement in general non - pure three - qubit states .
therefore , the discrimination between fully separable , biseparable and fully entangled states remains open in general , although it can be answered in some particular cases , like those of the non - pure examples of subtypes @xmath51 and @xmath56 given previously .
in [ sec:3 ] we will propose a measure that , even if it is not a complete solution , in some cases improves on previous results , as we shall see in the examples of [ sec:5 ] .
there are families of states depending on one or several parameters , such that by continuous variations of these the state goes from one type to another . or more generally , there are states @xmath86,@xmath87 , in two different types or subtypes such that the distance @xmath88 is arbitrarily small ; we will show examples in section 5 .
parts of this classification have antecedents in the literature . in @xcite , a classification of pure three - qubit states based only in the number of the entangled reduced states was proposed . but with only this criterion , fully separable states ( type @xmath50 in our classification ) and ghz - like states ( type @xmath81 ) are in the same class ; types @xmath89 ( i=1,2,3 ) and @xmath83 are also in another common class , and so on . this classification leaves out the more important property of the entanglement of three - qubit states : no entanglement , bipartite entanglements only , or full tripartite entanglement . in @xcite , the same authors introduced new classes for non - pure states , taking into account the existence of classical correlations ( separable but non - factorizable states ) ; in this paper we shall restrict ourselves to quantum correlations . in @xcite a classification of three - qubit entanglement in fully separable , biseparable ( in the sense that we denote as simply biseparable ) or fully inseparable states was given ;
generalized biseparable states were not considered as a class different from fully inseparable ones , and the existence or not of reduced binary entanglements was not considered .
an ideal measure of the full tripartite entanglement of three qubits should have at least the following characteristics : i ) to be zero for any fully separable or biseparable state and non - zero for any fully entangled state , ii ) to be invariant under lu , iii ) to be non increasing under locc , that is , to be an entanglement monotone @xcite .
condition i ) seems self evident , but some proposals for tripartite entanglement measures do not fulfill it , even for pure states , as we will see below ; the proposal that we will make ( tripartite negativity ) satisfies both parts of i ) for pure states .
conditions ii ) and iii ) are the mathematical expression of the non - local character of entanglement .
some authors include other desirable conditions @xcite but we will restrict ourselves to the three conditions listed above .
there are in the literature proposals for measures of tripartite entanglement of pure states , for instance those in @xcite , @xcite , @xcite , @xcite , @xcite ; besides specific objections that we will comment below , they can not be generally extended to non - pure states . in @xcite , @xcite and @xcite ,
some particular results were given for non - pure states .
in @xcite a generalization of the bipartite concurrence , called the _ 3-tangle _ , is proposed .
the original @xmath85 states have 3-tangle equal to zero @xcite ; in fact , we found that the 3-tangle is zero for a large number of pure states in subtype @xmath82 ( @xmath85-like ) , although none of their qubits is separable ( the three reduced states @xmath90 for these @xmath85-like states are all of them non - pure ) . therefore , the 3-tangle is not a good measure of full tripartite entanglement even for pure states ; it is a measure of something that has been called _ residual entanglement _ by some authors @xcite ; in @xcite it was used to characterize one of the slocc classes ( called @xmath91 by the authors ) .
yu and song showed @xcite that any good measure @xmath92 of two - particle entanglement could be extended to multiparticle systems , by taking bipartite partitions of them ; they would define the following measure of tripartite entanglement : @xmath93 this is the idea underlying @xcite and @xcite , which generalize concurrence , and @xcite which generalize von neumman s entropy of reduced states .
since the three terms in ( 2 ) verify separately conditions ii ) and iii ) , so does also the tripartite additive measure @xmath94 .
this was used in @xcite , @xcite , @xcite and @xcite to prove that generalizations of concurrence and von neumann s entropy of reduced states verify conditions ii ) and iii ) .
but , with any of these elections , @xmath94 would be non - zero for pure biseparable states of subtype @xmath56 , violating the first part of condition i ) .
the same would also happen if a probability density function were used , as in @xcite .
it is possible to avoid this objection by using the geometric mean instead of the arithmetic one : @xmath95 this idea was proposed , in a more general context , in @xcite .
so , if @xmath96 is the tangle ( the square of the concurrence ) , we have a multiplicative redefinition of the global entanglement @xmath97 ( whose additive version was introduced in @xcite and @xcite ) , and if @xmath96 is von neumann s entropy of reduced states , we have a redefinition of the additive measure @xmath98 that appears in @xcite .
the same argument of the previous paragraph proves that these redefined product versions of @xmath97 and @xmath98 verify conditions ii ) and iii ) .
von neumann s entropy of reduced states is an unambiguous measure of entanglement only for pure states , and the concurrence , although well defined for non - pure states of two qubits , has been extended in a practical way to higher dimensions only for pure states .
therefore , in order to have a measure of tripartite entanglement valid also for non - pure states we will use the negativity . we will define the tripartite negativity of a state @xmath12 as @xmath99 where the bipartite negativities are defined as in ( 1 ) , + @xmath100 , @xmath101 being the negative eigenvalues of @xmath102 , the partial transpose of @xmath12 with respect to subsystem @xmath103 , @xmath104 , with @xmath105 , and @xmath106 , respectively .
the bipartite negativities @xmath107 verify conditions ii ) and iii ) @xcite , and so our tripartite negativity @xmath108 verifies them . for pure states ,
our multiplicative tripartite negativity fulfills the three conditions at the beginning of this section .
although the qualitative classification of pure three - qubit states was easily done in [ sec:2 ] , tripartite negativity adds a quantitative appraisal of the full tripartite entanglement of these states . for non - pure two - party states in dimensions
@xmath109 there are entangled states with zero negativity @xcite , @xcite .
therefore , @xmath108 violates the second part of i ) . on the other hand
, @xmath108 could be non - zero for generalized biseparable states , violating also the first part of i ) .
we could fulfill desideratum i ) if we were able to replace negativity for some other measure that will be non - zero for any entangled two - party state in dimension @xmath109 , and also found a way to discriminate unambiguously between generalized biseparable and fully entangled states of three - qubits : these are difficult and still open problems .
the set of three bipartite measures @xmath110 , @xmath111 , @xmath112 contains more information than the geometric mean @xmath108 .
but even this set can not completely discriminate between fully separable , biseparable and fully entangled general mixed states , and therefore does not improve essentially on the unique tripartite negativity @xmath108 , at least for classification purposes . from the results in @xcite it can be proved that @xmath113 is a sufficient condition for distillability to a ghz state ( _ ghz - distillability _ ) , a property of central importance in quantum computation @xcite ; therefore , tripartite negativity is useful also for non - pure states , even if it does not solve the separability vs. entanglement problem .
our objective in this section is to divide the infinite set of lu equivalence classes of three - qubit pure states in six big subsets corresponding to the six subtypes of fig.1 .
since a lu equivalence class is defined by a set of gsd coefficients , we exhaustively analyzed the type of entanglement of all possible sets .
the full entanglement of the three - qubit state can be checked by studying its factorizability , as explained in [ sec:2 ] .
the entanglement of the reduced states ( non - pure in general ) will be determined calculating their negativity ( for these mixed two - qubit states , we could have used also the concurrence , but not von neumann s entropy ) .
the results are listed below . in the following
, we will denote the five coefficients of the gsd decomposition by @xmath114 ; an arbitrary three - qubit pure state can always be transformed to : @xmath115 this decomposition is symmetric in the interchange of the last two qubits , but not in the exchange of any of them with the first .
a more elegant totally symmetric gsd with five coefficients is also possible @xcite , but the algorithm to obtain the coefficients from an arbitrary initial state is much more complicated ; therefore it is not very useful for practical purposes .
although the canonical form ( 5 ) contains five complex coefficients , it depends on only five independent real parameters ; by suitably choosing the relative phases for each qubit and the global phase of the state @xmath116 one can express it in terms of five positive coefficients , related by the normalization condition , and a unique relevant phase , as it was done in @xcite . but starting from several generic vectors , different choices of phases will in general be needed to obtain a simplified form of this kind ; therefore , in order to compare the canonical forms for different initial vectors , we will postpone any phase choices ( and the normalization condition ) until they are convenient ; we will come back to this below when discussing ghz states .
now we will show the canonical forms for our six entanglement subtypes : * type @xmath50 : the conditions for fully separability are listed for instance in @xcite , in terms of the eight coefficients of the hilbert space .
the gsd forms can be obtained from those conditions , taking into account that three coefficients ( @xmath117 in the notation of @xcite ) are always zero .
* subtype @xmath56 : they have one of the following gsd forms : + @xmath118 ( where @xmath119 and @xmath120 or @xmath121 can be zero ; if @xmath122 the state is of type @xmath50 ) .
+ @xmath123 ( where @xmath124 can be zero ) + @xmath125 ( where @xmath124 can be zero ) .
+ for instance , for state @xmath126 the bipartite reduced states @xmath127 , @xmath128 are separable , while @xmath129is entangled , with negativity @xmath130 .
* subtype @xmath81 ( ghz - like states ) : + the states with these properties are of the canonical form @xmath131 .
+ the three separable reduced states are of the form : @xmath132 .
+ in particular , if the two coefficients are equal in modulus we will say that we have a ghz state , and denote it by @xmath133 .
all the vectors of the form @xmath134 , with @xmath135 real , belong to the bidimensional space of ghz states . with the election of canonical form of @xcite only the final vector @xmath136
would have been obtained ; then , two orthogonal initial vectors in the ghz class would have been reduced to the same canonical vector , while by postponing the choice of relative phases we have preserved the dimensionality of the ghz class and in particular the orthogonality relations in it .
+ fig.2 shows several multiplicative measures of the full tripartite entanglement of ghz - like states , as functions of the coefficient @xmath137 ( now we impose the normalization condition @xmath138 ) .
+ ( solid line ) , @xmath97 ( dash line ) and @xmath98 ( dot line ) , as a function of @xmath137 for ghz - like states.(@xmath131 , @xmath139 $ ] ; for @xmath140 $ ] the graphic is symmetric and can be thought off as representing the simultaneous interchange of values 0 and 1 for the three qubits ) .
the three measures induce the same order for the full tripartite entanglement .
ghz states ( @xmath141 ) are the ones with maximum full tripartite entanglement ( @xmath142 ) .
@xmath108 is more sensitive for small values of @xmath137.,scaledwidth=90.0% ] + [ fig:2 ] + the three measures induce the same order for the full tripartite entanglement ; if a state has larger entanglement than other with one measure , it has larger entanglement with the others .
the negativity @xmath108 is more sensitive for small values of @xmath137 .
+ we have performed a similar analysis for all the other classes , but the results depend on more than one parameter , and are more difficult to represent : therefore we will not show them here .
the results show in all cases that the three measures above considered give qualitatively similar behaviours , and that the maximum full tripartite entanglement of any three - qubit state corresponds to the lu equivalence class of ghz states .
another explicit example will be given below when discussing subtype @xmath82 .
* subtype @xmath83 : states of this class are those with one of the three following gsd forms : + @xmath143 , + @xmath144 , + @xmath145 , + with the three coefficients non - zero in each case . + for instance , for state @xmath146 the separable reduced states are latexmath:[$\rho^{(ab ) } = \rho^{(ac)}=(\alpha|0\rangle+\beta|1\rangle)(\alpha^{\star}\langle0|+\beta^{\star}\langle1|)\otimes while @xmath129 is entangled , with negativity @xmath148 .
note that if one of the three coefficients goes to zero , the state @xmath149 changes its classification ; in the limit @xmath150 we obtain a state of subtype @xmath81 , in the limit @xmath151 a state @xmath152 of subtype @xmath56 , and in the limit @xmath153 a state of subtype @xmath50 .
* subtype @xmath84 or star - shaped states : the gsd forms that have these properties are + @xmath154 , + @xmath155 , + with all the coefficients non - zero .
+ for instance , for state @xmath156 the separable reduced state is @xmath157 , while @xmath129 and @xmath128 are entangled . in this case , although the negativities are perfectly computable for any value of the coefficients , their analytic expressions in terms of generic coefficients are not very comfortable .
thus , we give here the concurrences : @xmath158 , and @xmath159 ( for two - qubits , concurrence and negativity coincide for pure states , not for mixed states like @xmath129 , @xmath128 ) .
* subtype @xmath82 or @xmath85-like class : states with the most general gsd form ( five coefficients @xmath160 , @xmath124 , @xmath161 , @xmath162 , @xmath163 different from zero ) belong to this class , and also those who have @xmath124 and/or @xmath163 equal to zero . + states with the canonical form @xmath164 , with
the three coefficients non - zero , are examples of this class . for these states ,
the bipartite reduced states are : + @xmath165 , @xmath166 , and @xmath167 , with negativities : @xmath168 ; + @xmath169 ; + @xmath170 .
+ in particular , if the three coefficients in @xmath171 are equal in modulus we will denote the state by @xmath172 and refer to it as a @xmath85 state ; the original symmetric @xmath85 state @xcite , @xmath173 has a gsd of this form .
+ the maximum values of the tripartite entanglement measures in the @xmath85-like class , are reached for the @xmath172 states : @xmath174 , @xmath175 , @xmath176 .
+ the negativities of the reduced bipartite states of the states @xmath172 ( which are the same as the negativities for the symmetric @xmath177 state ) are @xmath178 . our classification of pure three - qubit states in terms of full tripartite entanglement and reduced binary entanglements could be used to give a physical interpretation to the abstract gsd classes of @xcite and @xcite .
we do not have a canonical form like the gsd of the pure case to simplify the general form of mixed states ; therefore we will restrict ourselves to consider some concrete uniparametric families of non - pure states , to show some of the improvements allowed by the use of tripartite negativity , the problems remaining and relations with previous works . as a first example we will consider a family of mixed ghz and @xmath85 states : @xmath179 where @xmath177 has been defined in ( 6 ) and @xmath180 .
its tripartite negativity is + @xmath181 + ( @xmath182 ) , that is different from zero for any value of the parameter @xmath183 ( see fig .
3 ) , and has an absolute maximum for pure ghz states and a secondary maximum for pure @xmath85 states ( fig.3 ) .
this excludes full separability or simple biseparability for any state in the family , but can not discriminate between distributed binary entanglement and full tripartite entanglement .
nevertheless this result improves on @xcite , where the 3-tangle for this family was found to be zero for a value of the parameter @xmath183 ( generalized biseparability was not considered as distinct from full entanglement in @xcite ) .
+ states , @xmath184 , as a function of p.,scaledwidth=90.0% ] our second example is a family of mixed states with non - zero tripartite negativity @xmath108 only for some values of a parameter : @xmath185 , where * 1 * is the unity matrix , and @xmath186 . we find that @xmath187 if @xmath188 , and @xmath189 if @xmath190 , reaching a maximum @xmath191 for @xmath192 ( a pure ghz state ) .
dr et al .
@xcite showed that this state is ghz - distillable if @xmath190 .
therefore , for this family of mixed states our tripartite negativity quantifies ghz - distillability ; it starts at @xmath193 and increases with @xmath183 , with an obvious maximum of 1 if the state is a ghz state .
+ finally , we will consider a family of three - qubit states that can be fully entangled or generalized biseparable and nevertheless have zero tripartite negativity . in @xcite , a family of mixed @xmath109 states of the form + @xmath194 + with @xmath195 was considered .
these states have zero negativity for any value of the parameter b , although they were proved to be non - separable .
we do not know of any practical measure of bipartite entanglement that will discriminate these bound entangled states from separable ones .
biconcurrence @xcite , could do this in principle ; unfortunately its determination needs the calculus of the minimum of a certain function for any unitary operator in a hilbert space of dimension 64 , and no efficient way to do this is known @xcite .
we can convert these states in dimension @xmath109 to three - qubit states simply by taking the following basis in the 4 dimensional space:@xmath196 .
the bipartite negativities are then @xmath197 @xmath198 , @xmath199 @xmath200 , @xmath201 , and therefore @xmath202 . according to @xcite ,
the states are not separable into a and bc subsystems .
since the other two negativities are non - zero , we also know that they are not fully separable nor simply biseparable .
thus , this family of states can be fully entangled or generalized biseparable .
we have proposed a classification of three - qubit states based on the existence of bipartite and tripartite entanglements and the diverse possibilities for the reduced binary entanglements , including a graphic representation for pure states that can be extended to non - pure ones , although we have not done it here for reasons of space .
we have considered a measure ( tripartite negativity ) of the full tripartite entanglement avoiding some of the problems of previous proposals ; for pure states this measure quantifies full tripartite entanglement and confirms the distinction between fully entangled states and biseparable or fully separable ones that was obtained in [ sec:2 ] in a qualitative way .
we have given also the explicit form of the pure states in each subtype of our classification , after performing an easily computable gsd to a simplified canonical form ; in the simplest cases we have compared their tripartite negativity with other multiplicative generalizations of bipartite entanglement measures , concluding that they induce the same ordering of full tripartite entanglement .
we have analyzed some non - pure states that have non - zero tripartite negativity ( a sufficient condition for ghz - distillability ) or that have zero tripartite negativity although they are known to be entangled , to show the problems that remain in the practical classification of mixed three - qubit states .
+ we thank g. lvarez , d. salgado , l. lamata and j. len for valuable discussion and help .
acknowledge financial support from csic i3 program .
99 l. gurvits , journal of computer and system sciences * 69 * , 448 ( 2004 ) .
a. acn , a. andrianov , l.costa , e. jan , j.i .
latorre , r. tarrach , phys .
* 85 * , 7 ( 2000 ) .
a. acn , a. andrianov , e. jan , r. tarrach , j. phys .
gen . * 34 * , 6725 ( 2001 ) .
w. dr , g. vidal and j. cirac , phys . rev . a * 62 * , 062314 ( 2000 ) .
l. lamata , j. len , d. salgado and e. solano , phys .
a 74 , 052336 ( 2006 ) .
m. plesch , v. buek , phys .
a * 67 * , 012322 ( 2003 ) .
m. plesch , v. buek , phys .
a * 68 * , 012313 ( 2003 ) . v. coffman , j.kundu , w.k .
wootters , phys .
rev a * 61 * , 0532306 ( 2000 ) .
chang - shui yu , he - shan song , phys . lett a 330 , 377 ( 2004 ) .
d. meyer , n.r .
wallach , j. of math .
phys . , * 43 * , pp.4273 ( 2002 ) .
brennen , quantum information and computation , vol . 3 ( 6 ) , 619 - 626 ( 2003 )
. f. pan , d. liu , g. lu , j.p .
draayer , int.j.theor.phys .
* 43 * , 1241 ( 2004 ) .
p.facchi , g. florio and s. pascazio , phys .
a * 74 * , 042331 ( 2006 ) .
werner , phys .
a * 40*,4277 ( 1989 ) .
t. radtke , s. fritzsche , comput .
. comm . * 175 * , 145 ( 2006 ) .
sakurai , _ modern quantum mechanics _ , addison - wesley publishing company ( 1994 ) , 183 - 184 .
wootters , s. hill , phys .
lett * 78 * , 5022 - 5025 ( 1997 ) .
g. vidal , r.f .
werner , phys . rev . a * 65 * , 032314 ( 2002 ) . s. barn , s.j.d .
phoenix , phys .
a * 44 * , 535 ( 1991 ) .
p. rungta , v. buzek , c. m. caves , m. hillery , g. j. millburn , phys . rev .
a , * 64 * , 042315 ( 2001 ) .
r. horodecki , m. horodecki , p. horodecki , k. horodecki , quant - ph/070225 .
p.badziag , p. deuar , m. horodecki , p. horodecki , and r. horodecki , j. mod . opt . * 49 * , 1289 ( 2002 ) .
a. miranowicz , a. grudka , j. opt b : quantum semiclass .
optics 6 542 - 548 ( 2004 ) .
a. miranowicz , a. grudka , phys .
a * 70 * , 032326 ( 2004 ) .
plenio , s. virmani , quant .
inf . comp .
* 7 * , 1 ( 2007 ) .
plenio , phys .
lett . * 95 * , 090503 ( 2005 ) .
m. horodecki , p. horodecki , r. horodecki , phys .
lett . * 80 * 5239 - 5242 ( 1998 ) .
p. horodecki , phys .
a , * 232 * , 333 ( 1997 ) .
a. acn , d. bru , m.lewnstein , a. sanpera , phys .
lett . * 87 * , 040401 ( 2001 ) .
d.bru et al . phys .
a * 72 * , 014301 ( 2005 ) .
w. dr , j.i .
cirac , r. tarrach , phys .
lett . * 83 * 3562 - 3565 ( 1999 ) .
t. eggeling and r. f. werner , phys .
a * 63 * , 042111 ( 2001 ) .
w. laskowki , m. zukowski , phys .
a , * 72 * , 062112 ( 2005 ) .
ballentine , _ quantum mechanics : a modern development _ , world scientific publishing , 1998 .
g. vidal , j. mod . opt . * * 47**355 ( 2000 ) .
chang - shui yu , he - shan song , phys .
a * 73 * , 022325 ( 2006 ) .
chang - shui yu , he - shan song , phys .
a * 73 * , 032322 ( 2006 ) .
tzu - chieh wei , paul m. goldbart , phys .
a * 68 * , 024307 ( 2003 ) .
r. lohmayer , a. osterloh , j. siewert , a. uhlmann , phys .
lett . * 97 * , 260502 ( 2006 ) .
love et al . , quantum information processing * 6 * , 187 ( 2007 ) .
akhtarshenas , j. phys .
a * 38 * , 6777 ( 2005 ) .
h. a. carteret , a. higuchi , a. sudbery , j. math .
phys . * 41 * , ( 2000 ) 7932 - 7939 .
chang - shui yu , he - shan song , phys .
a * 72 * , 022333 ( 2005 ) . | we present a classification of three - qubit states based in their three - qubit and reduced two - qubit entanglements .
for pure states these criteria can be easily implemented , and the different types can be related with sets of equivalence classes under local unitary operations . for mixed states characterization of full tripartite entanglement
is not yet solved in general ; some partial results will be presented here . |
let us consider a portfolio of @xmath0 loans .
let the notional of loan @xmath1 be equal to @xmath2 .
then the @xmath1th loan represents fraction @xmath3 of the notional of the whole portfolio .
this means that if loan @xmath1 defaults and the entire notional of the loan is lost the portfolio loses fraction @xmath4 or @xmath5 of its value . in practice
when a loan @xmath1 defaults a fraction @xmath6 of its notional will be recovered by the creditors . thus the actual loss given default ( lgd ) of loan @xmath1 is @xmath7 fraction or @xmath8 of the notional of the entire portfolio .
we now describe the gaussian m - factor model of portfolio losses from default .
the model requires a number of input parameters . for each loan
@xmath1 we are give a probability @xmath9 of its default .
also for each @xmath1 and each @xmath10 we are given a number @xmath11 such that @xmath12 .
the number @xmath11 is the loading factor of the loan @xmath1 with respect to factor @xmath13 .
let @xmath14 and @xmath15 be independent standard normal random variables .
let @xmath16 be the cdf of the standard normal distribution . in our model loan @xmath1 defaults if @xmath17 this indeed happens with probability @xmath9 .
the factors @xmath18 are usually interpreted as the state of the global economy , the state of the regional economy , the state of a particular industry and so on .
thus they are the factors that affect the default behavior of all or at least a large group of loans in the portfolio .
the factors @xmath19 are interpreted as the idiosyncratic risks of the loans in the portfolio .
let @xmath20 be defined by @xmath21 we define the random loss caused by the default of loan @xmath1 as @xmath22 where @xmath6 is the recovery rate of loan @xmath1 .
the total loss of the portfolio is @xmath23 an important property of the gaussian factor model is that the @xmath24 s are not independent of each other .
their mutual dependence is induced by the dependence of each @xmath24 on the common factors @xmath18 .
historical data supports the conclusion that losses due to defaults on different loans are correlated with each other .
historical data can also be used to calibrate the loadings @xmath11.the @xmath24 s are not independent of each other .
their mutual dependence is induced by the dependence of each @xmath24 on the common factors @xmath18 .
historical data supports the conclusion that losses due to defaults on different loans are correlated with each other .
historical data can also be used to calibrate the loadings @xmath11 .
when the values of the factors @xmath18 are fixed , the probability of the default of loan @xmath1 becomes @xmath26 the random losses @xmath24 become conditionally independent bernoulli variables with the mean given by @xmath27 and the variance given by @xmath28 by the central limit theorem @xcite the conditional distribution of the portfolio loss @xmath25 , given the values of the factors @xmath18 , can be approximated by the normal distribution with the mean @xmath29 and the variance @xmath30 we define the cdf of the unconditional portfolio loss @xmath25 by @xmath31 for all real numbers @xmath32 .
since the conditional loss distribution is approximately normal , the cdf can be approximated by @xmath33 where @xmath34 is the cdf of the normal distribution with mean @xmath35 and variance @xmath36 , while @xmath37 is the density of the standard normal distribution .
in this section we apply the proposed algorithm to the single factor gaussian model of a portfolio with @xmath38 names . we choose a 125 name portfolio because it is the size of the standard djcdx.na.ig portfolio .
we choose a single factor model because it is the one most frequently used in practice .
the parameters of the portfolio are @xmath39 where @xmath40 . in figure [ resultsfig ]
we compare the cdf computed using @xmath41 monte carlo samples with the cdf computed using formula ( [ theformula ] ) .
the agreement between the two is good .
we add that the quality of approximation will improve even further for a bigger portfolio and many bank portfolios have much more than 125 names .
[ resultsfig ]
we define the @xmath42 ( value at risk ) of a given portfolio as the level of loss ( expressed as fraction of portfolio notional ) , such that the probability of the portfolio loss being less or equal to @xmath42 is equal to a predefined confidence level @xmath43 .
typically , @xmath43 is chosen to be between @xmath44 and @xmath45 .
thus we have @xmath46 from the results of the previous section it follows that an accurate approximation to @xmath42 can be found by solving the equation @xmath47 this equation can be solved , for example , by bisection . to find the solution to ( [ equv ] ) with accuracy of 1 basis point ( 0.01% ) we would need to evaluate the left hand side of ( [ equv ] ) no more than 14 times .
* example . *
_ we want to calculate the @xmath42 of the portfolio ( [ port ] ) with the confidence level @xmath48 .
we solve the equation ( [ equv ] ) and round the solution to the nearest basis point to arrive at @xmath49 .
we now run monte carlo simulation with @xmath50 samples and compute the probability that the portfolio loss is less or equal to @xmath51 . rounded to the nearest basis point it turns out to be @xmath52 .
thus the @xmath42 of the portfolio is indeed @xmath51 .
_ if desired , the convergence can be sped - up by using newton s method .
after the @xmath42 is calculated we can calculate the economic capital by subtracting the average portfolio loss from @xmath42 .
we now compare the algorithm proposed here with other proposed alternatives .
the fft based methods @xcite require the computation of a large number of fourier transforms . to determine the @xmath42 with an error of less than 1 basis point ( 0.01% ) it is necessary to compute approximately 10,000 fourier transforms .
each fourier transform is as expensive to evaluate as the left hand side of ( [ equv ] ) .
thus our algorithm is significantly faster than the fft based methods .
it is well known that the fft methods are much faster than the direct monte carlo simulation .
thus our algorithm is much faster than the monte carlo approach .
finally , the recursive approach of hull - white @xcite is comparable in speed to the algorithm proposed here .
however , it assumes that all the loans in the portfolio have equal notionals and recovery rates .
this is a very restrictive assumption which is unrealistic for many portfolios encountered in practice .
our algorithm makes no assumptions about homogeneity of the portfolio .
additionally , it is easier to implement than the algorithm of hull - white .
since @xmath42 satisfies equation ( [ equv ] ) we can use the implicit function theorem @xcite to find its partial derivatives with respect to the parameters of the model .
these partial derivatives are traditionally called the greeks in finance .
we arrive at the following expressions @xmath53 @xmath54 @xmath55 @xmath56 and @xmath57 these expressions can be easily evaluated by numerical integration using hermite - gauss quadrature .
we proposed an algorithm for computing the @xmath42 and the economic capital of a loan portfolio in the gaussian factor model .
the proposed method was tested on a portfolio of 125 names and gave high accuracy results
. the accuracy will be even higher for portfolios with more names .
many of the bank portfolios are much larger than 125 names .
the proposed algorithm is much faster than the fft based methods @xcite and the brute force monte carlo approach .
the speed of the hull - white algorithm @xcite is comparable to that of the the algorithm proposed here , but the hull - white algorithm requires that all the loans in the portfolio have equal notionals and recovery rates .
this is a very restrictive assumption which is unrealistic for many portfolios encountered in practice .
our algorithm makes no assumptions about the homogeneity of the portfolio .
also , it is easier to implement than the algorithm of hull - white . additionally , we obtained analytical expressions for the greeks using the implicit function theorem .
we also comment that the algorithm can be extended trivially to the case of stochastic recovery rates and recovery rates correlated with the state of the factor variables .
some of the ideas used in this paper were previously explored in @xcite and @xcite .
i thank my adviser a. chorin for his help and guidance during my time in uc berkeley .
i thank mathilda regan and valerie heatlie for their help in preparing this article .
last , but not least , i thank my family for their constant support .
p. okunev .
using hermite expansions for fast and arbitrarily accurate computation of the expected loss of a loan portfolio tranche in the gaussian factor model .
report - lbnl-57835 , lawrence berkeley national laboratory , berkeley , ca , 2005 . | we propose a fast algorithm for computing the economic capital , value at risk and greeks in the gaussian factor model .
the algorithm proposed here is much faster than brute force monte carlo simulations or fourier transform based methods @xcite .
while the algorithm of hull - white @xcite is comparably fast , it assumes that all the loans in the portfolio have equal notionals and recovery rates .
this is a very restrictive assumption which is unrealistic for many portfolios encountered in practice .
our algorithm makes no assumptions about the homogeneity of the portfolio .
additionally , it is easier to implement than the algorithm of hull - white .
we use the implicit function theorem to derive analytic expressions for the greeks . |
in this paper we present an update of the analysis of double - tag two - photon events : @xmath3 already published for @xmath4 and @xmath5 @xcite , obtained at lep with the l3 detector .
the data , collected at centre - of - mass energy @xmath6 , correspond to an integrated luminosity of 176 pb@xmath7 .
both scattered electrons are detected in the small angle electromagnetic calorimeters ( fig .
[ fig : kin ] ) .
the virtuality of the two photons , @xmath8 and @xmath9 , is in the range of @xmath10 .
the centre - of - mass energy of the two virtual photons , @xmath11 , ranges from @xmath12 to @xmath13 . _ ] for @xmath14 ( untagged events ) @xcite , the two - photon cross - section , @xmath15 , is dominated by vector - vector interactions , vdm ( fig .
[ fig : feyn]a ) . interactions .
[ fig : feyn ] _ ] with increasing @xmath16 , the vdm process is suppressed by the vector meson form factor and the quark parton model process ( qpm ) , shown in fig .
[ fig : feyn]b , becomes important .
single - tag two - photon events , where @xmath17 , are usually analysed within the dis formalism @xcite and a photon structure function is introduced ( _ resolved photon _ ) . since the highly virtual photon , unlike the proton , does not contain constituent quarks with an unknown density distribution , one may hope to have a complete qcd calculation under particular kinematical conditions . in this formalism , used today by the monte carlo generators , one considers one or two resolved photons ( _ single or double resolved processes _ ) to calculate the qcd leading order diagrams .
an alternative qcd approach is based on the bfkl equation @xcite . here the highly virtual two - photon process , with @xmath18 , is considered as the `` golden '' process where the calculation can be verified without phenomenological inputs @xcite .
the @xmath19 interaction can be seen as the interaction of two @xmath20 pairs scattering off each other via multiple gluon exchange ( fig .
2c ) . for @xmath21
a diagram with one - gluon exchange could be sufficient and the cross - section would be constant ( fig .
2d ) . in the limit of high energy , @xmath22 , the diagram of fig .
2c is calculable by the resummation of the large logarithms . in this scheme
the cross - section for the collision of two virtual photons is @xcite : @xmath23 here @xmath24 where @xmath25 is the beam energy , @xmath26 and @xmath27 are the energy and polar angle of the scattered electrons and @xmath28 is the `` hard pomeron '' intercept . the centre - of - mass energy of the two - photon system is related to the @xmath29 centre - of - mass energy @xmath30 by @xmath31 . in leading order
one has @xmath32 where @xmath33 is the number of colours . using @xmath34 and @xmath35 ,
one obtains @xmath36 @xcite ; in the next - to - leading order the bfkl contribution is calculated to be smaller , @xmath37 @xcite .
double - tag interactions have been measured in previous experiments @xcite at lower values of @xmath16 and @xmath38 . for comparison with the prediction of the bfkl models
, the cross - sections will be given as a function of the variable @xmath39 instead of @xmath38 as used in ref .
the advantage in using this variable is that @xmath40 is independent of the beam energy : @xmath41 with @xmath42 . + as @xmath43 and @xmath44 : + @xmath45 where @xmath46 is the angle between the two photons and @xmath47 are the tagging angles .
the monte carlo generators used in this analysis are jamvg @xcite which generates events with the matrix element of figure [ fig : feyn]b and phojet @xcite which gives a good description of the single - tag events @xcite and uses the grv - lo @xcite parton density in the photon to initiate qcd processes .
the dominant backgrounds are @xmath48 , simulated by jamvg @xcite , and single - tag two - photon hadronic events , where a hadron is misidentified as a scattered electron .
the contamination by annihilation processes is simulated by pythia @xcite ( @xmath49 ) , koralz @xcite ( @xmath50 ) and koralw @xcite ( @xmath51 ) .
all monte carlo events are passed through a full detector simulation using the geant @xcite and the geisha @xcite programs and are reconstructed in the same way as the data .
a detailed description of the l3 detector is given in ref .
the two - photon hadronic events are mainly triggered by two independent triggers : the central track @xcite and the single and double tag energy @xcite triggers .
the total trigger inefficiency of the selected events is less than 1% . and energy of the scattered electrons .
the data is compared to the monte carlo predictions , normalised to the number of data events there are two entries per event .
[ fig : q2 ] _ , title="fig : " ] and energy of the scattered electrons .
the data is compared to the monte carlo predictions , normalised to the number of data events there are two entries per event .
[ fig : q2 ] _ , title="fig : " ] two - photon hadronic event candidates , @xmath3 , are selected using the following cuts : * there must be two identified electrons , forward and backward ( double - tag ) , in the small angle electromagnetic calorimeters .
each electron is identified as the highest energy cluster in one of the calorimeters , with energy greater than @xmath52 .
the polar angle of the two tagged electrons has to be in the range @xmath53 mrad and @xmath54 mrad .
* the number of tracks , measured in the polar angle region @xmath55 , must be greater than two .
the tracks are required to have a transverse momentum , @xmath56 , greater than @xmath57 and a distance of closest approach in the transverse plane to the interaction vertex smaller than 10 mm . *
the value of @xmath39 is required to be in the range @xmath58 .
after these cuts , 108 events are selected with an estimated background of 14 events from @xmath48 and misidentified single - tag events .
the contamination from annihilation processes is negligible . , and the variable @xmath40 .
the data are compared to the monte carlo predictions , normalised to the number of data events .
[ fig : wtru ] _ , title="fig : " ] , and the variable @xmath40 .
the data are compared to the monte carlo predictions , normalised to the number of data events .
[ fig : wtru ] _ , title="fig : " ] [ cols="^,^,^,^,^ " , ] in fig .
[ fig : ggcro ] we show @xmath59 , after subtraction of the qpm contribution given in table [ tab : ggcro ] , as a function of @xmath40 . using an average value of @xmath16 , @xmath60 , we calculate the one - gluon exchange contribution with the asymptotic formula ( eq . 10.2 of ref .
the expectations are below the data .
the leading order expectations of the bfkl model ( eq . 4.19 of ref .
@xcite ) , shown as a dotted line in fig .
[ fig : ggcro ] , are too high . by leaving @xmath28 as a free parameter , a fit to the data , taking into account the statistical errors , yields : @xmath61 the results
are shown in fig .
[ fig : ggcro ] as a continuous line . in the same figure
are also shown our previous results @xcite obtained at @xmath62 and @xmath5
. a combined fit of the three sets of data leaving @xmath28 as a free parameter gives : @xmath63 with @xmath64 .
, after subtraction of the qpm contribution at @xmath4 ( @xmath65 ) , @xmath5 ( @xmath66)and @xmath6 ( @xmath60 ) .
the data are compared to the predictions of the lo bfkl calculation and of the one - gluon exchange diagram .
the continuous line is a fit to the data with eq . 1 by leaving @xmath28 as a free parameter .
[ fig : ggcro ] _ ] but if one fixes @xmath28 to its lo and nlo calculated values , one finds respectively : @xmath67 @xmath68
the cross - sections of double - tag @xmath3 events is measured at @xmath6 .
the events are well descibed by the phojet monte carlo model which uses the grv - lo parton density in the photon and leading order perturbative qcd .
a combined fit of our results obtained at @xmath69 gives a value of @xmath70 smaller than expected from the lo bfkl calculations and higher than the value predicted by nlo bfkl calculations .
i would really like to thank maneesh wadhwa for providing me the results and some help , and all the l3 @xmath71 group for discussions and support .
+ thanks also to the organizers of this interesting conference .
twogam version 1.71 is used .
+ l. l@xmath73nnblad , `` @xmath74 event generators '' , in physics at lep2 , ed .
g. altarelli , t. sj@xmath73strand and f. zwirner , cern 96 - 01 ( 1996 ) , volume 2 , 224 .
+ s. nova , delphi note 90 - 35 ( 1990 ) .
+ we thank our colleagues from delphi to make their program available to us | a preliminary study of double tag events using the l3 detector at center of mass energy @xmath0 has been performed .
the cross - section of @xmath1 collisions is measured at average @xmath2 .
the results are in agreement with predictions based on perturbative qcd , while the quark parton model alone is insufficient to describe the data .
the measurements lie below the lo and above the nlo bfkl calculations . |
it is a common notion that the laughlin state is incompressible .
this statement however relates to a thermodynamical property and does not contradict the fact that even an arbitrarily small impurity potential locally changes the electron density , fig .
[ fig-14 ] . as a function of the impurity strength @xmath9 , the local density at the position of the impurity @xmath10 shows three distinct types of behaviour marked by i ,
ii and iii in fig .
[ fig-21 ] .
the boundary between i and ii can roughly be identified with the incompressibility gap @xmath11 . while the density change @xmath12 is roughly proportional to @xmath9 for @xmath13 ( region ii ) , reminiscent of a standard compressible behaviour as of a fermi gas or liquid ,
the density change is proportional to @xmath14 for weak impurities ( region i ) .
the latter non - linear region does not appear in earlier data@xcite on a sphere and we attribute it to the center - of - mass part of the wavefunction as we explain below .
finally , the response diverts again from the linear regime for very strong impurities ( region iii ) .
we will not investigate this regime here and focus only on the regions of @xmath15 and @xmath16 .
[ cols="^,^ " , ]
we find that the polarized @xmath1 , @xmath5 and @xmath3 states all respond similarly to isolated impurities , represented in this paper by a weak @xmath4-line impurity . since @xmath3 gives slightly different results than the other two systems we conclude that the particle - hole conjugation ( @xmath7 and @xmath5 ) is a stronger link than the reversal of the effective field in the composite fermion picture ( @xmath5 and @xmath3 ) .
the singlet states react differently : both @xmath5 and @xmath3 respond quite unequally and more weakly to an electric potential impurity than @xmath7 .
in particular , @xmath5 gives a much weaker response than @xmath7 .
this was unexpected because the laughlin state has the largest incompressibility gap .
the spin - resolved _ and _ spin - unresolved density - density correlation functions of the @xmath5 singlet state suggest that electrons in it appear in zero - spin pairs with characteristic size of @xmath17 magnetic lengths and these form a full - landau - level - like state .
high polarizability by magnetic impurities and a relatively small effect of impurities affecting only one spin species are compatible with this interpretation .
the authors acknowledge support from the following grants : av0z10100521 of the academy of sciences of the czech republic ( kv ) , lc510 of the ministry of education of the czech republic ( kv ) , and sfb 508 quantenmaterialien . | using exact diagonalization we examine the response of several most prominent fractional quantum hall states to a single local impurity .
the @xmath0 singlet state is found to be more inert than the polarized one in spite of its smaller incompressibility gap . based on its spin - spin correlation functions
we interpret it as a liquid of electron pairs with opposite spin .
a comparison of different types of impurities , non magnetic and magnetic , is presented .
the effect of impurities on the incompressible ground states was in the focus of many theoretical studies since the early days of the fractional quantum hall effect ( fqhe ) .
the principal interest is based on the knowledge that the fqhe occurs only in high mobility samples implying that random electric potential ubiquitous in experiments leads to a decrease of the incompressibility gap eventually destroying the incompressibility completely .
a large part of the theoretical work therefore deals with random ensembles of impurities@xcite and this , due to the complexity of such calculations , in context of the most robust incompressible fqh state , bearing the name of r. laughlin , which occurs at filling factor@xcite @xmath1 .
a single impurity can , however , be also viewed as a probe into the correlated state revealing additional information on its properties and nature@xcite . beyond this , as we believe fundamental interest , this information may help to understand new correlated states for which only numerical many - body wavefunctions are available .
the particular example we bear in mind are the half - polarized states at @xmath2 and @xmath3 discovered first in optical experiments@xcite .
numerically , the wavefunctions of suitable candidates can be obtained by exact diagonalization but it was not possible to link them conclusively with any of the physical states proposed subsequently@xcite@xcite .
the current study of the parent polarized and unpolarized states may provide a guideline to achieve this . first , the laughlin wavefunction is investigated and different regimes regarding tis response to a point impurity are identified . here
, we build on the seminal articles of zhang@xcite and rezayi@xcite . a comparison between point and @xmath4-line impurities ,
is presented .
proceeding to filling factors @xmath5 and @xmath3 , it proves advantageous to apply different types of impurities , magnetic and non - magnetic .
this investigation reveals basic structural properties of the various fqh states like local density and polarization responses to impurities .
based on these , we propose that the @xmath2 singlet state is a @xmath6 liquid of electron pairs with opposite spin .
we recall that the three filling factors studied here , @xmath7 , @xmath5 and @xmath3 are all closely related . in the composite fermion ( cf )
picture@xcite , they correspond to filling factors @xmath8 equal to 1 , 2 and 2 respectively and the last two differ only in the direction of the effective field .
systems at @xmath2 and @xmath3 can therefore be expected to behave very similarly , at least within the mean field of the cf . on the other hand , with the spin degree of freedom frozen out ( e.g. by large zeeman energy ) ,
the ground states at @xmath1 and @xmath5 are identical up to a particle - hole conjugation@xcite .
here we show how far these statements of close relationship apply when the excitations become involved in addition to the ground states as it is the case with inhomogeneous systems . |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.