article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
giant ly@xmath0nebulae , or `` blobs , '' are extended sources at @xmath1 @xmath2 26 with typical ly@xmath0sizes of @xmath65 ( @xmath650kpc ) and line luminosities of @xmath7ergs@xmath4 .
the low number counts , strong clustering , multiple embedded sources , and location in over - dense environments of the largest ly@xmath0blobs indicate that they lie in massive ( @xmath8 @xmath2 @xmath9@xmath10 ) dark matter halos , which will evolve into those typical of rich galaxy groups or clusters today @xcite .
therefore , ly@xmath0blobs are unique tracers of the formation of the most massive galaxies and their early interaction with the surrounding intergalactic medium ( igm ) .
this interaction is probably tied on some scale to the source of the blobs extended ly@xmath0emission , but that mechanism is poorly understood .
emission from ly@xmath0blobs could arise from several phenomena , which may even operate together , including shock - heating by galactic superwinds or gas photoionized by active galactic nuclei .
another possibility is smooth gas accretion , which is likely to play an important role in the formation of galaxies ( e.g. , * ? ? ?
* ; * ? ? ?
* ) and which should channel some of its gravitational cooling radiation into atomic emission lines such as ly@xmath0 .
another scenario is the resonant scattering of ly@xmath0photons produced by star formation or active galactic nuclei ( agn ) @xcite in the embedded galaxies . to resolve the debate about the nature of ly@xmath0blobs requires at the very least
that we discriminate between between outflowing and inflowing models .
because ly@xmath0is a resonant line and typically optically thick in the surrounding intergalactic medium , studies even of the same ly@xmath0blob s kinematics can disagree .
on one hand , @xcite argue that their integral field unit ( ifu ) spectra of a ly@xmath0blob are consistent with a simple model where the ly@xmath0emission is absorbed by a foreground slab of neutral gas swept out by a galactic scale outflow . on the other ,
@xcite explain the same data as arising from the infall of the surrounding intergalactic medium .
worse , @xcite claim that the same symmetric ly@xmath0profiles are most consistent with static surrounding gas . to distinguish among such possibilities
requires a comparison of the center of the ly@xmath0line profile with that of a non - resonant line like h@xmath0 @xmath116563or [ ] @xmath115007 .
these rest - frame optical nebular lines are better measure of the ly@xmath0blob s systemic velocity , i.e. , of the precise redshift , because it is not seriously altered by radiative transfer effects and is more concentrated about the galaxies in the ly@xmath0blob s core .
we illustrate this line offset technique in figure [ fig : cartoon ] . while gas accretion models predict different line profile shapes depending on various assumptions , e.g. , the location of ionizing sources , the detailed geometry , and the velocity field , they all predict that the overall ly@xmath0line profile , originating from the central source or the surrounding gas , will be blue - shifted with respect to the center of a non - resonant line such as h@xmath0 that is optically thin to the surrounding gas @xcite .
this is because the red - side of the ly@xmath0profile will see higher optical depth due to the infalling ( approaching ) gas . in other words ,
the h@xmath0 kinematics represent the true underlying velocity field if the ly@xmath0blob is accreting gas from the intergalactic medium .
the same is true if the gas is outflowing , except that the ly@xmath0line will be redshifted with respect to h@xmath0 .
thus , if we measure the direction of the ly@xmath0h@xmath0 and/or ly@xmath0[]line offset ( hereafter defined as v_ly@xmath3 ) , we can distinguish an inflow from an outflow .
the first such analysis for two ly@xmath0blobs shows that ly@xmath0is coincident with or redshifted by @xmath2200kms@xmath4 from the h@xmath0 line center ( * ? ? ?
* see also mclinden et al .
these offsets are much smaller than the @xmath21000kms@xmath4expected from superwind models and even smaller than those typical of lyman break galaxies ( lbgs ) , which are widely believed to have galactic outflows @xcite .
thus , if v_ly@xmath3 is a proxy for outflow velocities ( @xmath12 ) , our initial results suggest that star formation- or agn - produced winds may not be required for powering ly@xmath0blobs , making other interpretations of their emission more likely . however , we do not yet know if these results are representative of _ all _ ly@xmath0blobs or if we have failed to detect strong flows due to the projected orientations of these two sources .
for example , the gas flow may not be isotropic .
as in bipolar outflows in m82 , a galactic - scale outflow may occur in the direction of minimum pressure in the surrounding interstellar medium ( ism ) , often perpendicular to the stellar disks . or , if gas accretion is taking place in ly@xmath0blobs , numerical simulations suggest that the gas infall may occur preferentially along filamentary streams @xcite . thus , if the bulk motion of gas ( either infalling or outflowing ) happens to be mis - aligned with our line of sight ( los ) , then we may underestimate or even fail to detect the relative velocity shifts .
therefore , it is critical to measure v_ly@xmath3for a larger sample to average over any geometric effects and obtain better constraints on the incidence , direction , speed , and isotropy of bulk gas motions in blobs . in this paper , in order to overcome this geometry effects , we present new x - shooter optical and near - infrared ( nir ) spectroscopy of the six more ly@xmath0blobs at @xmath13 . note that the survey redshift of this ly@xmath0sample has been carefully selected to allow all important rest - frame optical diagnostic lines ( e.g. , [ ] @xmath113727 , [ ] @xmath115007 , h@xmath14@xmath114868 , h@xmath0@xmath116563 ) to fall in nir windows and to avoid bright oh sky lines and atmospheric absorption @xcite . with the resulting large sample of v_ly@xmath3measurements ( a total of eight ) ,
we determine the relative frequency of gas infall versus outflow .
benefiting from x - shooter s high spectral resolution and wide spectral coverage ( 3000 2.5 ) , we constrain the gas kinematics in ly@xmath0blobs using three different tracers : ( 1 ) the offset of the ly@xmath0profile with respect to a non - resonant nebular line , ( 2 ) the offset of an interstellar metal absorption line in the rest - frame uv with respect to the nebular emission line , and ( 3 ) a new indicator , the profile of the spectrally - resolved [ ] emission line . this paper is organized as follows . in section [ sec : observation ] , we review our sample selection and describe the x - shooter observations and data reduction . in section [ sec : result ] , we present the results from the x - shooter spectroscopy , confirming the ly@xmath0blobs redshift ( section [ sec : spec1d ] ) .
we present 1d and 2d spectra in section [ sec : spec1d ] and section [ sec : spec2d ] , respectively .
we briefly summarize the properties of individual systems in section [ sec : individual ] . in section [ sec :
kinematics ] , we constrain the gas kinematics using the three different techniques noted above . in section [ sec : shift ] , we compare the ly@xmath0profiles with the h@xmath0 or [ ] line centers to discriminate between simple infall and outflow scenarios and present the v_ly@xmath3 statistics for the sample . in section [ sec :
absorption ] , we describe the interstellar absorption lines detected in three galaxies . in section [ sec :
o3profile ] , we inspect the [ ] emission line profiles in detail to look for possible signatures of warm outflows . in section [ sec : column_density ] , we constrain the column density of a ly@xmath0blob by comparing its ly@xmath0profile with a simple radiative transfer ( rt ) model . in section [ sec : cdfs - lab10 ] , we focus on a ly@xmath0blob with a blue - shifted ly@xmath0component not directly associated with any detected galaxy , a possible marker of gas inflow .
section [ sec : conclusion ] summarizes our conclusions . throughout this paper ,
we adopt the cosmological parameters : @xmath15 = 70kms@xmath4mpc@xmath4 , @xmath16 , and @xmath17 .
we observe six ly@xmath0blobs from the @xcite sample .
these targets were chosen such that they are not x - ray detected [ @xmath18 @xmath5 ( 0.3 4.2)@xmath19@xmath20 ergs@xmath4 ; @xcite ] , and thus are not obvious agn , as our primary goal is to cleanly detect gas infall or outflow .
these six blobs lie in the extended chandra deep field south ( ecdfs ) and were discovered via deep narrowband imaging with the ctio-4 m mosaic ii camera and a custom narrowband filter ( _ nb_403 ) .
this filter has a central wavelength of @xmath21 , designed for selecting ly@xmath0-emitting sources at @xmath13 . in figure
[ fig : image ] , we show the images of these six ly@xmath0blobs ( cdfs - lab06 , 07 , 10 , 11 , 12 , 13 , 14 ) at various wavelengths ( _ ubk _ , ly@xmath0 , _ spitzer _ irac 3.6and _ hst _ f606w ; @xcite , @xcite , @xcite , @xcite ) . with x - shooter , we are targeting intermediate luminosity ( @xmath22@xmath2 10@xmath23 ergs@xmath4 ) ly@xmath0blobs : the higher ly@xmath0blob i d indicates the lower ly@xmath0luminosity in our sample .
our sample was obtained from a blind survey , so combined with the two brightest ly@xmath0blobs presented in @xcite , the full sample ( a total of eight ) spans a wide and more representative range of ly@xmath0luminosity and size .
furthermore , the transition from compact ly@xmath0emitters ( laes ; isophotal area of a few arcsec@xmath24 ) to extended ly@xmath0blobs ( @xmath2510arcsec@xmath24 ) is continuous @xcite , thus the gas kinematics for our faintest ly@xmath0blobs might share the properties with those of bright laes .
we refer readers to @xcite for details of the sample selection and to @xcite for the first results of our spectroscopic campaign .
we obtained high resolution optical
nir ( 3000 2.5 ) spectra of the six ly@xmath0blobs using x - shooter , a single object echelle spectrograph @xcite , on the vlt ut2 telescope in service mode between 2010 november 6 and 2011 january 28 . in table
[ tab : obslog ] , we summarize the x - shooter observations . in figure
[ fig : image ] , we show the location of the spectrograph slit on the sky , which was placed on uv - brightest galaxy or galaxies embedded at or near the ly@xmath0blob center .
these galaxies or galaxy fragments also lie in the region of brightest ly@xmath0emission .
later , we assume that their redshifts mark the systemic redshift of the ly@xmath0blob .
x - shooter consists of three arms ( uvb , vis , nir ) , which covers the spectral ranges of 30005500 , 55001 , and 12.5 , respectively .
this enormous spectral coverage allows us to obtain both ly@xmath0and h@xmath0lines with a _
single _ exposure in contrast to our previous approach @xcite involving both optical and nir spectrograph .
furthermore , x - shooter can detect at least one of the nebular lines ( [ ] , h@xmath14 , [ ] , h@xmath0 ) , which will provide the systemic velocity of the embedded galaxies .
because at the redshift of our targets ( @xmath26 ) most of the emission lines of interest ( ly@xmath0 , , , [ ] , [ ] , h@xmath0 ) are located in the uvb or nir arms , we focus only on the uvb ( @xmath27 = 9001660 ) and nir ( @xmath27 = 30207500 ) part of spectra in this paper .
the observations were carried out over eight nights and 14 observing blocks ( obs ) of one hour duration each . the sky condition was either clear or photometric , and the guide camera seeing ranged from 06 to 12 with a median of 08 depending on the ob .
we adopted 16 and 12-wide slits for uvb and nir , yielding a spectral resolution of @xmath28 @xmath2 3300 and @xmath29 , respectively .
the slit length is rather small ( 12 ) compared to typical long - slits . in each ob
, we placed the slit on a target using a blind offset from a nearby star .
using acquisition images taken after the blind - offset , we estimate that the telescope pointing and position angle of the slit are accurate within 0.2 and 0.5 on average , respectively .
the individual exposure times were 680s and 240s for uvb and nir , respectively , and the telescope was nodded along the slit by @xmath30 2 while keeping the science targets always on the slit but at different detector positions .
total exposure times were 0.8 3.2 hours depending on the targets . in general , we were always able to detect both ly@xmath0and at least one of the optical nebular lines within one ob .
for accurate wavelength calibration , we took thar lamp frames through pinhole mask right before the science exposures and at the same telescope pointing to compensate for the effect of instrument flexure .
for the uvb arm , we obtained another thar arc frame at the end of each ob in order to verify the wavelength solution where no bright sky lines are available .
telluric standard stars ( b type ) were taken after or before the science targets with similar airmass to correct for atmospheric absorption in the nir .
spectrophotometric standard stars were observed with 5-wide slits once during the night as a part of the observatory s baseline calibration plan .
we reduce the data using the eso x - shooter pipeline ( version 1.3.7 ) . in the uvb arm ,
the frames are overscan - corrected , bias - subtracted , and flat - fielded with halogen and deuterium lamps .
the sky background is then subtracted in the `` stare''-mode of the pipeline by modeling the sky in 2d as described in @xcite . in the nir arm ,
dark current and sky background are removed from each science frame by subtracting the dithered `` sky '' frame ( `` nodding''-mode of the pipeline ) .
then , we flat - field the data and correct for cosmic - ray hits and bad pixels . in both arms , these flat - fielded , sky - subtracted frames were corrected for the spatial distortions using multi - pinhole arc frames . because we will compare the velocity centers of the ly@xmath0and h@xmath0 lines , we carefully verify the wavelength calibration . in the nir
, we compare the wavelength solutions obtained from the oh sky lines in the science frames to those from the daytime arc lamps and flexure - compensation frames , i.e. , the pipeline solutions . in the uvb
, we also compare the wavelength solutions obtained from the attached thar arc frames with the pipeline solutions .
our wavelength calibration is accurate within @xmath24kms@xmath4 in both arms .
furthermore , in the cases where we visited sources multiple times , all spectra agree each other .
these frames are then rectified ( resampled ) and combined to create 2d spectra .
we collapse the 2d spectra in the wavelength direction to measure the spatial extent of each emission line .
then we extract 1d spectra from [ @xmath31 , @xmath32 apertures , where the @xmath33 is the gaussian width of the spatial profile .
the aperture sizes are 2 35 in the uvb and 15 25 in the nir depending on the target .
finally , the 1d spectra are corrected to heliocentric velocities and transformed to the vacuum wavelength .
various emission lines in the uvb and nir arms confirm that ly@xmath0blobs lie at the survey redshift , @xmath34 .
in addition to ly@xmath0and h@xmath0 , we cover other uv emission lines and the non - resonant [ ] @xmath353727,3729 , [ ] @xmath354959,5007 , and h@xmath14 @xmath114861lines . in figure [ fig : spec1d ] , we show 1d and 2d spectra of the six ly@xmath0blobs .
the first three columns show the rest - frame uv emission lines ( ly@xmath0 , , ) from the x - shooter uvb arm , and the remaining columns show the rest - frame optical nebular emission lines ( [ ] , h@xmath14 , [ ] , h@xmath0 ) from the nir arm . among the rest - frame optical nebular lines , the [ ]
line is the brightest and detected with highest signal - to - noise ( s / n ) ratio in all cases , partly due to the low sky background and thermal instrument background in _
we determine the systemic redshift using [ ] doublets . in one case ( cdfs - lab14 ) , the brighter [ ] line ( @xmath36 ) falls on top of an oh sky line , thus making it impossible to determine the line center . in this case
we used the fainter [ ] line ( @xmath37 ) .
the vertical dashed lines in figure [ fig : spec1d ] indicate the line centers determined by [ ] lines , which are then overlayed on other emission line profiles . as expected , we find that the line centers of all non - resonant emission lines agree well each other , to within @xmath210kms@xmath4 , showing that all of these lines are good indicators of systemic velocity .
thus , the brightest [ ] line can serve as best emission line to target for this survey redshift and instrument .
the ly@xmath0 profiles are significantly broad compared to the non - resonant lines ( h@xmath0and / or [ ] ) in six cases out of the sample of eight ly@xmath0blobs , including the two from our previous work @xcite .
these integrated 1d profile shapes range from an asymmetric single - peaked profile ( cdfs - lab06 , 07 , 10 ) , to a double - peaked profile with a stronger red peak ( cdfs - lab02 , 13 ) , to a double - peaked profile with two similar intensity peaks ( cdfs - lab14 ) .
the ly@xmath0profiles of even this small sample show extremely diverse morphologies consistent with simple radiative model predictions @xcite with varying geometry and outflow velocities ( see also * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) .
the remaining two ly@xmath0profiles are narrower _ relative to _ the [ ] lines and show slightly extended wings ( cdfs - lab01 and 11 ) .
note that the ly@xmath0line width of cdfs - lab01 is one of the largest among our sample , but the blue side of its ly@xmath0profile agrees well with its h@xmath0profile @xcite . while there is an underlying broad component in cdfs - lab11 ,
the width of the dominant narrow component is small , comparable to that of the [ ] line ( see [ sec : type2 ] ) .
these are the two ly@xmath0blobs where and emission lines are also detected indicating that they contain a hard ionizing source , possibly an agn . if photo - ionization by an agn is indeed responsible for the and emission , and possibly the extended ly@xmath0emission as well , the discovery of narrow ly@xmath0profiles suggests that the ly@xmath0blob gas is highly - ionized , i.e. , that the resonant scattering of ly@xmath0is not effective enough to alter the profile significantly
. we will further investigate the details of these , emission lines and the implications for agn in a future paper ( y.yang in preparation ) .
the fraction of double - peaked profiles is significant : @xmath238% ( 3/8 ) , which is roughly consistent with the findings for lbgs and laes at @xmath1 = 23 . among lbgs
, @xcite find that @xmath230% of lbgs with ly@xmath0emission show multiple - peaked profiles .
@xcite also find that @xmath250% of lae s profiles have multiple peaks .
+ + + + + + + + + + + using 2d spectra , we detect the extended ly@xmath0lines and identify the exact locations from which the ly@xmath0or [ ] line originates . in figure [ fig : spec2d ] , we show close - ups of the 2d [ ] and ly@xmath0profiles in the first and second columns , respectively . to aid the comparison between the [ ] and ly@xmath0profiles , we overlay the rough boundary of the [ ] spectrum on each ly@xmath0panel with ellipses .
the third column shows the spatial profiles along the slit , i.e. , collapsed in the wavelength direction .
the last column shows the 1d ly@xmath0and [ ] profiles that are extracted from the different parts of the slit , as indicated with vertical arrows in the third column . in four cases ( cdfs - lab06 , 10 , 13 , 14 )
, the ly@xmath0 spectrum is spatially extended relative to the [ ] line , confirming the narrowband imaging result that the ly@xmath0-emitting gas extends beyond the embedded galaxies that are probably responsible for the [ ] emission .
note that the x - shooter spectroscopy reaches much shallower ly@xmath0surface brightness limit ( @xmath21.53@xmath19@xmath38ergs@xmath4@xmath39arcsec@xmath40 ) than the narrowband imaging ( @xmath25@xmath19@xmath41ergs@xmath4@xmath39arcsec@xmath40 ; 3@xmath33limit ) . in two ly@xmath0blobs ( cdfs - lab06 and 10 ) , the [ ] lines
are spatially resolved . however , there is no evidence that the [ ] or h@xmath0lines are extended beyond the uv continuum emission arising from stars in the embedded galaxies seen in figure 2 .
deeper nir spectroscopic , preferentially ifu , observations are required to better define the spatial extent of these lines and to measure the spatially - resolved gas kinematics . throughout the paper , we assume that the [ ] and h@xmath0lines originate from the central embedded galaxies , not from the extended ly@xmath0-emitting gas , and thus that their line centers represent the systemic velocity of the ly@xmath0blob . in the next section ,
we briefly describe individual systems in detail .
cdfs - lab06 has two rest - frame uv sources with small separation ( 08 ; 6.5kpc ) in the _ hst _ image ( fig .
[ fig : image ] ) . both components ( or clumps ) were placed in the x - shooter slit and are detected in [ ] and ly@xmath0(fig .
[ fig : spec2d ] ) .
these two sources are separated by only @xmath250kms@xmath4 in velocity space , thus it is not clear whether they belong to one galaxy or are interacting with each other .
we adopt the redshift of the brighter uv source as the systemic velocity of cdfs - lab06 .
the ly@xmath0emission detected in the spectrum ( between @xmath42 = @xmath43 and @xmath44 ) is spatially extended due to the other galaxy , so it does not represent the igm or circum - galactic medium ( cgm ) .
the galaxy within cdfs - lab07 has a bar - like morphology in the _ hst _ image with which we align the slit .
this galaxy or galaxy fragments were marginally resolved in the [ ] emission line .
all optical nebular emission lines ( [ ] , h@xmath14 , [ ] , and h@xmath0 ) are detected .
faint uv continuum emission is marginally detected , allowing us to study the gas kinematics with metal absorption lines ( [ sec : absorption ] ) .
the [ ] and h@xmath0lines show asymmetric profiles extending toward the blue . this profile can be fitted with a narrow gaussian component at the velocity center superposed on the blueshifted broad component ( [ sec : o3profile ] ) .
cdfs - lab10 is the most puzzling and complex source in the x - shooter sample .
the ly@xmath0emission in the narrow - band image is elongated over 10 ( @xmath280kpc ) . in the nir spectrum ,
three sources are detected : two with [ ] emission lines ( galaxies b and c ) , and the other ( galaxy a ) with very faint continuum ( fig . [ fig : spec2d ] ) .
this nir continuum source ( galaxy a ) is located at the slit center , while the strongest [ ] -emitting source ( galaxy c ) is offset by @xmath21toward north - east from the center and barely detected in the _ hst _ image .
the 2d [ ] spectrum of galaxy c shows a velocity shear indicative of rotating disk . in the 2d ly@xmath0spectrum , there are also three distinct components .
although the 1d ly@xmath0spectrum of the entire blob looks single - peaked with a broad linewidth , it is in fact composed of these three components .
we will investigate these various emission line and continuum sources in [ sec : cdfs - lab10 ] .
there are one compact uv source at the slit center and a diffuse emission toward northeast in the _ hst _ image .
it appears that the [ ] emission originates from the central compact source .
unlike commonly observed broad ly@xmath0profiles found in ly@xmath0blobs and emitters , cdfs - lab11 has a peculiar ly@xmath0profile that is almost symmetric and narrow . as will be discussed in [ sec : type2 ] , both narrow and underlying broad components are required to explain the ly@xmath0profile .
the ly@xmath0is redshifted against h@xmath0by small amount : v_ly@xmath3= 84@xmath306kms@xmath4 . both [ ] and ly@xmath0are spatially compact ( fig .
[ fig : spec2d ] ) . both @xmath111546and
@xmath111640 emission lines are also detected ( fig .
[ fig : spec1d ] ) , implying the presence of hard ionizing source .
a total of two ly@xmath0blobs from our ecdfs sample ( cdfs - lab01 and 11 ; 2/8 ) show these emission lines .
note that and _ narrow _ line emission are commonly detected in bright ly@xmath0blobs @xcite , while the nature of the hard - ionizing source is unknown .
cdfs - lab13 has a double - peaked profile with a stronger red peak , and asymmetric [ ] profile ( [ sec : o3profile ] ) .
ly@xmath0 in cdfs - lab13 is likely more extended , but the low s / n of the [ ] line makes the comparison difficult .
there are multiple galaxy fragments in the _ hst _ uv continuum image , but they were not spatially resolved in the x - shooter observations . in the _ hst _ and narrowband images ,
a uv source is located at the upper boundary of the ly@xmath0emission contours . in the x - shooter
2d spectra ( figure [ fig : spec2d ] ) , the [ ] line is centered on this uv continuum source and ly@xmath0is more extended toward the south in agreement with narrowband imaging .
cdfs - lab14 is one of two cases where extended ly@xmath0emission is well detected in the spectroscopy .
the ly@xmath0profile has two peaks with similar intensities .
the peak separation narrows as the slit distance from the central galaxy increases .
faint uv continuum is also detected at the location of the uv source allowing us to measure outflow speed from metal absorption lines ( [ sec : absorption ] ) . combining the ly@xmath0and the absorption profiles
, we will put constraints on the gas kinematics and the neutral column density of this system ( [ sec : column_density ] ) .
[ ] has a broad wing on top of the narrow component ( [ sec : o3profile ] ) in this section , we constrain the gas kinematics in the ly@xmath0blobs using three different techniques and compare those results .
first , for the six blobs in the x - shooter survey , we compare the optically thick ly@xmath0and non - resonant [ ] line ( either @xmath36 or @xmath37 in the case of cdfs - lab14 ) to measure ly@xmath0velocity offsets from the systemic velocity of the ly@xmath0blobs .
for the two ly@xmath0blobs from our previous work ( cdfs - lab01a and 02 ; @xcite ) , where an [ ] line is unavailable , we use h@xmath0 . as mentioned previously , the line centers of both [ ] lines and h@xmath0are all consistent .
second , we constrain the outflow speed from the interstellar metal absorption lines in three galaxies where we are able to detect the rest - frame uv continuum in the spectrum .
lastly , we present a new tracer of kinematics in four ly@xmath0blobs : characterizing the breadth and asymmetry of the [ ] line profile . note that while we detect an asymmetric wing in some [ ] line profiles , it does not affect the line centroid , which is essential in determining the systemic velocity in the first technique above .
we compare the peak of each ly@xmath0profile with the center of a non - resonant nebular emission line , particularly [ ] . throughout the paper , the ly@xmath0h@xmath0and ly@xmath0[]offsets are used interchangeably to represent the ly@xmath0offset from the systemic velocity : v_ly@xmath3 .
figure [ fig : line_comparison ] shows eight ly@xmath0profiles from this work and from @xcite , plotted with increasing v_ly@xmath3 .
because ly@xmath0spectra are somewhat noisy because of the small spectral dispersion ( 2030kms@xmath4 per pixel ) , we measure v_ly@xmath3 using the following two methods .
first , we measure the velocity of the ly@xmath0peak after smoothing the spectra with a gaussian filter with fwhm = 90kms@xmath4 , corresponding to our velocity resolution .
second , we fit the red peaks with an asymmetric gaussian function , which consists of two gaussian functions with different fwhms being joined at the center .
the two measurements agree to within @xmath250kms@xmath4 , except for cdfs - lab06 where the second method gives v_ly@xmath3
= 120kms@xmath4compared to 320kms@xmath4from simple smoothing due to the very sharp edge at the blue side of ly@xmath0profile .
we adopt the second measurements in this paper and show these fits in figure [ fig : line_comparison ] and table [ tab : lyman ] .
none of the conclusions in this paper are affected by this choice .
figure [ fig : voffset ] shows the distribution of v_ly@xmath3 of the eight ly@xmath0blob galaxies in comparison with 41 lbgs that have both h@xmath0 and ly@xmath0spectra @xcite . because the spectral resolutions of our ly@xmath0spectra obtained from vlt / x - shooter and magellan / mage are higher than those of the ly@xmath0profiles of the lbgs ( fwhm @xmath45 370kms@xmath4 ) , we test if the different spectral resolutions affect the ly@xmath0blob lbg comparison .
we repeat the measurement of v_ly@xmath3 after convolving our ly@xmath0profiles to the spectral resolution of lbg sample .
the v_ly@xmath3 values change by a negligible amount ( only @xmath3050kms@xmath4 ) except for cdfs - lab02 , where v_ly@xmath3 increases by + 100kms@xmath4because of its sharp red peak .
therefore , we conclude that the v_ly@xmath3 distributions of ly@xmath0blobs and lbgs can be compared directly .
the v_ly@xmath3 distribution for galaxies within our ly@xmath0blobs reveals smaller velocity offsets than typical of lbgs , confirming previous claims ( * ? ? ?
* see also mclinden et al .
galaxies within blobs have v_ly@xmath3= @xmath4660 @xmath47 + 400kms@xmath4 with an average of @xmath48 = 160kms@xmath4 , while lbgs at similar redshifts have v_ly@xmath3= 250
900kms@xmath4 with @xmath48 = 445kms@xmath4 .
the remaining question is how to interpret these small v_ly@xmath3 values .
two possibilities are : ( 1 ) v_ly@xmath3 is a proxy for the gas outflow velocity ( @xmath12 ) , suggesting that the outflows here are weaker than in other star - forming galaxies at @xmath1 = 23 @xcite , or ( 2 ) there is less neutral gas close to the systemic velocity of the embedded galaxies @xcite , and v_ly@xmath3 is independent of the outflow speed .
we discussed the caveats associated with interpreting v_ly@xmath3 in @xcite and revisit this issue in [ sec : dvlya ] . in the direction of the embedded galaxies ,
we do not find any ly@xmath0profile that is blue - peak dominated , i.e. , v_ly@xmath3@xmath5 0kms@xmath4 .
therefore , there is no evidence along these lines - of - sight for infalling gas ( but see [ sec : cdfs - lab10 ] ) . while the statistics are still small , we place an upper limit on the covering factor of infalling gas ( if any ) detectable with the v_ly@xmath3 technique . here
the covering factor of any inflowing streams like those predicted by cold - mode accretion @xcite must be less than @xmath213% ( 1/8 ) .
we further discuss the covering factor of cold streams in [ sec : covering_factor ] . while strong emission lines such as ly@xmath0and [ ] are relatively easy to detect , interstellar metal absorption lines provide a less ambiguous way to measure the outflow velocity of neutral gas lying _ in front of _ the galaxies targeted with our spectroscopic slit .
however , due to the faint uv continuum ( @xmath49 = 23.826.5 mag ) of our targets , even 8-m telescope struggles to detect their continuum for absorption line studies .
therefore , we obtain higher - s / n absorption line profiles by stacking several lines in each galaxy .
our aim is to test the mild outflow interpretation of our v_ly@xmath3 results by comparing the velocity offset of the stacked ism absorption profile to the systemic velocity of the galaxy : @xmath50 . among the six ly@xmath0blobs ,
only the uv continua of the galaxies in cdfs - lab07 and cdfs - lab14 have s / n higher than 1.5 per spectral pixel ( @xmath22030kms@xmath4 ) , allowing us to marginally extract the absorption profiles .
figure [ fig : spec_abs](top ) shows the rest - frame uv spectra of these two galaxies in comparison with the lbg composite @xcite . for comparison , we also show the boxcar - smoothed spectra over 20 pixels , corresponding to a velocity width of 450600kms@xmath4 .
while it is difficult to identify the absorption lines in the un - binned spectra , the smoothed spectra show a good match to the composite spectrum , revealing several low- and high - ionization lines .
as suggested above , it is still difficult to measure individual absorption line profiles , so we stack the five low - ionization lines : @xmath111260 , @xmath111302 , @xmath111334 , @xmath111526 , @xmath111670 to increase the s / n .
we do not include the and lines in the stacking , as these high ionization lines are contaminated with broader absorption features arising from stellar winds from massive stars @xcite , which are hard to remove in our low s / n spectra .
furthermore , it is possible that these high ionization lines trace different state of gas , while the low ionization lines show similar profiles @xcite . in figure
[ fig : spec_abs](bottom ) , we show the stacked absorption profiles for three galaxies , including the re - analyzed cdfs - lab02 spectrum that we obtained earlier with magellan / mage @xcite .
the stacked absorption profiles here are consistent with weaker outflows than required by the super / hyperwind hypothesis for ly@xmath0 blob emission .
these profiles are also consistent with the interpretation of small v_ly@xmath3 indicating small outflow speed , as discussed in
[ sec : shift ] . in two galaxies
( within cdfs - lab02 and 07 ) , the absorption profiles have minima around @xmath46200kms@xmath4and might extend up to @xmath46500kms@xmath4 , although the exact end of the profile is uncertain due to the low s / n .
the shapes of these two absorption profiles are similar to those of typical star - forming galaxies ( i.e. , lbgs ) at the same epoch @xcite , and the implied outflow velocities are comparable to or slower than in the lbgs . by fitting a gaussian profile ,
we obtain @xmath50= @xmath46177@xmath3031kms@xmath4and @xmath46236@xmath3031kms@xmath4for cdfs - lab02 and cdfs - lab07 , respectively .
we do not find any redshifted absorption component , which is consistent with the absence of any blueshifted ly@xmath0emission line associated with these galaxies .
in contrast , the profile of cdfs - lab14 s galaxy has a minimum at @xmath51 @xmath45 0kms@xmath4 ( @xmath50= @xmath4659@xmath3032 kms@xmath4 ) without any significant blueshifted ( outflowing ) component .
while the other two profiles terminate roughly at @xmath51 @xmath45 0kms@xmath4 , cdfs - lab14 s profile extends to @xmath51 @xmath25 0kms@xmath4 .
thus , its ism absorption and ly@xmath0 profiles are both consistent with the simple rt model expectation that a symmetric , double - peaked ly@xmath0profile should emerge from static gas or the absence of of bulk motions .
furthermore , the absorption profile is almost saturated , indicating that the column density in this ly@xmath0blob could be higher than for the other two systems .
we will place a constraint on the column density of its ly@xmath0-emitting gas in [ sec : column_density ] . for cdfs - lab14 ,
the location of the dip between the blue and red ly@xmath0velocity peaks changes little for the two extraction apertures shown in figure [ fig : spec2d ] : one along the line of sight toward the embedded galaxy ( the [ ] source ) and the other encompassing the spatially extended gas around @xmath52 = @xmath460.5 .
note that such a trough between ly@xmath0peaks is often interpreted as arising from absorption by the neutral media between the ly@xmath0source and observers .
for example , @xcite claim that their ifu spectra of a ly@xmath0blob ( ssa22-lab02 ; steidel blob 2 ) suggest that the ly@xmath0emission is absorbed by a foreground slab of neutral gas swept out by a galactic scale outflow .
more recently , @xcite show instead that the absorption troughs in the ly@xmath0emission are actually located at the systemic velocity determined by the [ ] emission line , i.e. , there are negligible velocity offsets between any foreground screen and the systemic velocity .
cdfs - lab14 also demonstrates that a coherent velocity trough , at least over a @xmath210kpc scale , can arise entirely from complicated radiative transfer effects even if there are no significant bulk motions in the ly@xmath0-emitting gas .
the central galaxies in four ly@xmath0blobs ( cdfs - lab06 , 07 , 13 , 14 ) show a broad and/or shifted underlying component to the [ ] profile , which provides additional constraints on the kinematics of warm ionized gas in the vicinity of the galaxies . to the authors knowledge , this is the first detection of broadened , asymmetric [ ] profiles from narrowband - selected ly@xmath0-emitting galaxies at high redshifts , demonstrating that high spectral resolution is required to fully exploit the nir spectroscopy of ly@xmath0galaxies .
we emphasize again that these broad underlying components , even when shifted in velocity , do not affect our measurements of [ ] line centers , because the [ ] flux density at the core is dominated by the narrower component , which is likely to arise from nebular regions in the embedded galaxy . as discussed previously , the line centers determined from other non - resonant emission lines ( e.g. , h@xmath0and h@xmath14 ) agree to within @xmath210kms@xmath4 .
therefore , our measurements of systemic velocity , critical for determining v_ly@xmath3 and @xmath50 , remain unchanged . in figure [
fig : o3profile ] , we show close - ups of the [ ] profiles . in cdfs - lab11 , which has the narrowest and the most symmetric ly@xmath0profile , the [ ] line is also symmetric .
cdfs - lab10 is excluded from the following analysis because its [ ] line is elongated in position - velocity space , suggesting some velocity shear , and the neighboring galaxy ( cdfs - lab10a ) makes it difficult to reliably extract its profile ( see [ sec : cdfs - lab10 ] for more details of this system ) . in the remaining four ly@xmath0blobs ,
the [ ] profile either has an asymmetric wing ( cdfs - lab06 , 07 , 13 ) or a symmetric broad component or components ( cdfs - lab14 ) .
note that the cdfs - lab06 profile in figure [ fig : o3profile ] also includes the light from a faint neighbor or clump to the northwest that is slightly blueshifted ( @xmath250kms@xmath4 ) .
therefore , the somewhat redshifted wing of cdfs - lab06 is not due to this contamination . to extract the underlying broad components
, we fit the line profiles with two gaussian functions ( the two dot - dashed lines in figure [ fig : o3profile ] ) and list the results in table [ tab : o3profile ] . in the appendix
, we describe our fitting procedures in detail .
we adopt a two - component fit for simplicity and for comparison with previous studies ( although we can not rule out the possibility that the velocity wings consist of multiple small narrow components ) .
the line center of the broad component ( @xmath53 ) agree with that of the narrow component in cdfs - lab14 , is blueshifted in cdfs - lab07 , and in cdfs - lab13 , and is marginally redshifted in cdfs - lab06 .
the widths of the broad components , corrected for the instrumental resolution , are relatively small : @xmath54 = 45 120kms@xmath4(fwhm = 100 280kms@xmath4 ) , which would not be detected with lower resolution or lower s / n spectra . to quantify the contribution of the broad components , we measure the `` broad - to - narrow ratio '' ( @xmath55 ) , which is defined as the ratio of the flux in the broad [ ] emission line to the flux in the narrow component . because the fluxes in the two gaussian components are anti - correlated with each other , this ratio has large uncertainties .
we also list the `` broad flux fraction '' ( @xmath56 ) , which is the ratio of the flux in broad emission to the total [ ] flux ( @xmath57/@xmath58 ) . the broad emission component is significant with @xmath55 = 0.4 0.8 constituting 30%45% of the total [ ] line flux ( although the uncertainties are fairly large ) .
the measurements of the underlying components are highly dependent on the s / n of spectra . due to the low s / n in the @xmath59-band , we are not able to reliably measure the broad line components from permitted lines such as h@xmath0or h@xmath14 , but similar broad wings are also present in the h@xmath0profiles of at least two galaxies ( cdfs - lab07 and 14 ; see figure [ fig : spec1d ] ) .
what is the mechanism responsible for the broad component in the [ ] profiles ?
a broad component with a larger line - width ( @xmath60 ) of a few hundred kms@xmath4is generally interpreted as a signature of starburst - driven galactic winds , and often observed in local dwarf starbursts ( e.g. , * ? ? ? * ) and in local ultraluminous infrared galaxies ( ulirgs ) ( e.g. , * ? ? ? * ) . while broad components in forbidden lines such as [ ] are not related to the broad line region ( blr ) of type 1 agn , they might still arise from the shocked narrow line region ( nlr ) in agn
. however , the composite spectrum of all of our x - shooter sample has a line ratio @xmath61([]/h@xmath0)@xmath5 @xmath62 , excluding any significant contribution from agn ( y. yang et al . in preparation ) .
furthermore , in cdfs - lab11 , whose and emission lines hint at the presence of an agn , no broad component is detected . at high redshift ( @xmath63 ) , broadened emission lines were first reported by @xcite .
they found that the stacked spectrum of @xmath64 star - forming galaxies ( sfgs ) shows broad ( fwhm @xmath2 550kms@xmath4 ) emission underneath the h@xmath0+[]line complex .
more recently , stacked spectra of higher s / n data have revealed that the broad emission is spatially extended over a half - light radius @xcite .
broad emission lines are now detected from individual sfgs and even from giant star - forming clumps within them @xcite .
@xcite show that @xmath64 sfgs show broad wings of h@xmath0emission with fwhm @xmath2 300
1000kms@xmath4(@xmath54 @xmath2 125 425 ) and employ the maximum blueshifted velocity , @xmath65= @xmath66 @xmath46 @xmath67 as a proxy of the outflow speed , finding that sfgs have @xmath65= 380 1000kms@xmath4
. thus , the broad component in these galaxies is attributed to powerful galactic outflows . for compact ly@xmath0emitters , although there are increasing number of detections of [ ] and h@xmath0(e.g .
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , broad emission in [ ] has not been reported so far , perhaps due to the lower spectral resolution or lower s / n of these studies .
therefore , at the moment , it is not clear whether the broad component and sometime line asymmetry that we observe in ly@xmath0blobs is a general property of ly@xmath0-selected galaxies . as in local ulirgs and in high redshift sfgs
, the detection of a broad component in the [ ] profile here suggests warm ionized outflows from the galaxies within ly@xmath0blobs , presumably driven by supernovae and stellar winds .
however , while the flux fraction of our broad emission is comparable to those of sfgs , the broad [ ] wings are narrower ( @xmath54 = 45 120kms@xmath4 ) and the inferred velocities much smaller @xmath65= 150 260kms@xmath4 .
therefore , at face value , the warm ionized outflows from the ly@xmath0blob galaxies are not as strong as those in sfgs at similar redshift .
furthermore , these estimates for outflow velocity are roughly consistent with the values obtained from our two other kinematic measures , v_ly@xmath3and @xmath50([sec : shift ] and [ sec : absorption ] ) , independently discounting shock - heating via super / hyperwinds as a viable powering mechanism .
constraining the physical state of the ly@xmath0-emitting gas in a ly@xmath0blob is a critical step to understand its emission mechanism and to directly compare the observations with the numerical simulations , which predict the ly@xmath0emissivity maps from the gas density , temperature , and uv radiation fields . in this section , as a first step , we estimate the column density of a ly@xmath0blob , cdfs - lab14 .
the v_ly@xmath3and @xmath50analyses of cdfs - lab14 are consistent with a simple rt model in which the surrounding gas is static or without bulk motions .
the third kinematic indicator , the [ ] profile , which suggests a mild outflow , either contradicts the other indicators or is sensitive to gas in a different state ( e.g. , warm , ionized instead of cold , neutral ) or distribution ( e.g. , around galaxy instead of in galaxy ) .
here we use the ly@xmath0emission and ism absorption line profiles to place a constraint on the amount of neutral hydrogen in this ly@xmath0blob .
ultimately , by comparing the ly@xmath0profile with detailed rt models , one could extract a wealth of information , including outflow speed , optical depth , and column density ( e.g. , * ? ? ?
we defer such detailed analysis to the future papers and consider the most simplistic case in this section .
ly@xmath0line transfer in an extremely thick medium of neutral gas has been studied over many decades , and the analytic solutions for simple geometries ( like a static homogeneous slab or uniform sphere ) are known @xcite .
we consider a simple geometry in which a ly@xmath0source is located at the center of a static homogeneous slab with optical depth of @xmath68 from the center to the edge . because ly@xmath0photons generated at the center escape the system through random scattering in the frequency space , the emergent line spectrum is double - peaked profile with its maxima at @xmath69 , where @xmath70 represents the line frequency in units of doppler width @xmath71 . @xmath72 and @xmath68 are the voigt parameter and ly@xmath0optical depth at the line center , respectively @xcite . in terms of velocity ,
the separation between blue and red peaks with same the intensity is @xmath73 where @xmath74 and @xmath75 represent the doppler parameter ( @xmath76 ) and the column density of neutral hydrogen , respectively . if we adopt a uniform sphere geometry instead of a slab , the coefficient of the above equation will decrease to 336kms@xmath4 , and the product ( @xmath74 @xmath75 ) will increase by a factor of two for a fixed @xmath77 value .
note that the blue - to - red peak separation is degenerate between two parameters , @xmath74 and @xmath75 . from the ly@xmath0profile of cdfs - lab14 ( figure [ fig : spec1d ] and [ fig : spec2d ] ) , we measure the separation between the blue and red peaks , @xmath77 = 790@xmath3039kms@xmath4 . in the analytic solution ,
the peaks on the blue and red sides of the systemic velocity are each symmetric , whereas for cdfs - lab14 they are slightly asymmetric . as a result ,
the rt in this system may require ultimately a more complicated model than the assumed simple geometry . to constrain @xmath75 ,
we consider two extreme cases for the doppler parameter @xmath74 : first , as a lower limit , @xmath74 @xmath25 12.85kms@xmath4 , obtained by assuming a temperature @xmath78 = 10@xmath79k and ignoring the turbulence term .
second , @xmath74
@xmath5 @xmath80 , the width of the metal absorption lines .
in other words , we assume that the observed velocity width of absorbing material is purely due to the turbulence or random motions inside the slab . the width of the absorption line is @xmath80 = 152 @xmath30 35kms@xmath4from a single gaussian fit ( [ sec : absorption ] and figure [ fig : spec_abs ] ) and after being corrected for the instrumental line width . for this range of @xmath74
, we obtain 19.7 @xmath5 @xmath81 @xmath5 20.8 .
because the ly@xmath0emission is spatially resolved in cdfs - lab14 , we further apply this technique to the extended ly@xmath0-emitting gas in a direction other than toward the embedded galaxy .
as shown in figure [ fig : spec2d ] , the ly@xmath0profile remains double - peaked as we move away from the sight - line directly toward the embedded galaxy .
the separation between the blue and red peaks decreases , indicating that @xmath74 @xmath82 also decreases . for @xmath77 @xmath45 650 kms@xmath4 ,
we obtain slightly smaller estimates for the column density : 19.5 @xmath5 @xmath81 @xmath5 20.6 . using the accurate measurement of systemic velocity , the metal absorption lines , and the symmetric double - peaked ly@xmath0profile , we are able to place constraints on the @xmath75 toward the galaxy and the extended ly@xmath0-emitting gas , albeit with large uncertainties . it is intriguing that this rough estimate of the column density is similar to those of damped ly@xmath0absorption systems ( dla ; @xmath75 @xmath25 2@xmath19@xmath83@xmath39 )
. while we have identified this ly@xmath0blob by searching for extended ly@xmath0emission , it would be interpreted as a dla if there were a background qso whose continuum spectrum showed absorption at the blob redshift .
there are similar systems where extended ly@xmath0emission is identified from dlas very close to the qso redshift ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) .
+ up to this point , we have examined only ly@xmath0emission along the line - of - sight to galaxies embedded in our blobs . in the 2d spectrum of cdfs - lab10 , however , we identify a region of blueshifted ly@xmath0emission that does not spatially coincide with any of the blob galaxies identified with _
this is the first case in our sample where we detect the ly@xmath0emission from the extended gas itself .
given that a blueshifted ly@xmath0line is a long - sought signature of gas inflows as described in
[ sec : intro ] and figure [ fig : cartoon ] , we present a detailed analysis of cdfs - lab10 here . in figure [
fig : slitprof](bottom ) , we show the _ hst _ f606w image of cdfs - lab10 now rotated and smoothed with a gaussian kernel to increase the contrast of faint sources .
there are three broadband sources labeled cdfs - lab10a , b , and c within the ly@xmath0contour , which is elongated over 10 ( @xmath282kpc ) .
figure [ fig : slitprof](top ) shows the spatial profiles of the stellar continuum and the emission lines ( ly@xmath0and [ ] ) .
the latter are the same as in figure [ fig : spec2d ] . to obtain the spatial profiles for the continuum
, we collapse the 2d spectra for the rest - frame wavelength range , [ 1260 , 1600 ] and [ 4340 , 5400 ] , i.e. , redward of ly@xmath0and both sides of the [ ] lines , respectively .
because of different data reduction modes , the spatial profiles from the nir arms ( [ ] ) cover only the central @xmath26 , while the profiles from the uvb arm ( ly@xmath0 ) cover @xmath214 .
the brightest uv source in the _ hst _ image , cdfs - lab10a at the slit center ( @xmath52 = 0 ) , is detected in both the rest - frame uv and optical continua , but not in ly@xmath0or any strong emission lines ( [ ] , [ ] , h@xmath14 , h@xmath0 ) .
it is still possible that there is a faint [ ] @xmath115007line under a sky line at @xmath84 @xmath2 @xmath85kms@xmath4(see figure [ fig : spec2d ] ) .
therefore , it is not clear whether galaxy a is at the same redshift as the ly@xmath0-emitting gas or is a fore- or background galaxy .
the galaxy b at @xmath52 @xmath2@xmath463 is detected in ly@xmath0 , [ ] , and uv continuum and is thus a member of the cdfs - lab10 blob .
galaxy b s ly@xmath0emission is slightly offset from its uv continuum .
cdfs - lab10c , which has a filamentary or elongated morphology in the _ hst _ image , is the faintest ( @xmath86 = @xmath87 ) of the three sources , but the brightest in ly@xmath0and [ ] .
thus , galaxy c is the dominant source of ly@xmath0emission from the known galaxies in this ly@xmath0blob .
it appears to be a classic example of an object heavily extincted in the uv whose ly@xmath0photons can still escape .
its 2d [ ] spectrum shows a velocity shear , suggesting a disk .
there is no obvious counterpart of the ly@xmath0emission at @xmath52 = @xmath43 and @xmath84 @xmath2 @xmath46500kms@xmath4 in either the _ hst _ image or in the x - shooter continuum spectrum .
therefore , this isolated ly@xmath0emission probably arises from the extended gas .
its line profile is broad , double - peaked , and blueshifted from galaxy b and galaxy c ( figure [ fig : spec2d ] ) .
it is not possible to distinguish whether the bulk motion is inflowing or outflowing ( at @xmath2500kms@xmath4 ) with respect to galaxy c , the brightest [ ] source and marker of systemic velocity ( figure [ fig : spec2d ] ) , because we do not know whether the gas lies in front of or behind galaxy c. this is the first unambiguous detection of a blueshifted ly@xmath0line with respect to galaxies embedded within ly@xmath0blobs .
it is likely that the blueshift is relative to the systemic velocity of the ly@xmath0blob as well .
it is not clear what this blueshifted ly@xmath0emission represents .
we consider three possibilities .
first , although unlikely , this ly@xmath0component could be associated with a heavily - extincted galaxy that lies below the detection limit of the _ hst _ image ( @xmath86 @xmath88 28 mag ) or with galaxy a , whose redshift is unknown .
second , this gas might be tidally - stripped material arising from a galaxy - galaxy interaction : the two galaxies ( b and c ) are separated by @xmath230kpc in projected distance and @xmath2200kms@xmath4 in velocity space .
lastly , but most interestingly , this blueshifted ly@xmath0emission could be the long - sought , but elusive cold gas accretion along filamentary streams @xcite .
note that similarly blueshifted ly@xmath0emission has been reported in a faint ly@xmath0emitter at @xmath1 = 3.344 , which was discovered in an extremely deep , blind spectroscopic search @xcite .
the spectrum of this peculiar system also has very complex structure ( e.g. , diffuse fan - like blueshifted ly@xmath0emission and a dla system ) .
@xcite suggest that this blueshifted ly@xmath0emission can be explained if the gas is inflowing along a filament behind the galaxy and emits fluorescent ly@xmath0photons induced by the ionizing flux escaping from the galaxy .
discriminating among the above possibilities will require us to further constrain the ly@xmath0line profile of this blueshifted component and to more fully survey possible member galaxies ( or energy sources ) within cdfs - lab10 .
still , we were able to successfully link the ly@xmath0and [ ] sources with the embedded galaxies in this complex system and to find blueshifted ly@xmath0emission that might point to gas inflow .
this example raises concerns about how to identify the sources of ly@xmath0emission and to interpret the gas morphologies and kinematics solely from ly@xmath0lines .
elongated or filamentary ly@xmath0morphologies may be a sign of bipolar outflows @xcite or related to filamentary cold streams . in cdfs - lab10 ,
the ly@xmath0spectrum shows two kinematically distinct components around the brightest galaxy a : the upper ( @xmath42 @xmath25 0 ) and lower parts are blue- and redshifted , respectively . in the absence of [ ] spectroscopy , the ly@xmath0data appear to be consistent with either of the above outflow or infalling stream scenarios .
however , our detailed analysis including the [ ] line clearly shows that one of the ly@xmath0components arises from the faint galaxy c , which would be difficult to detect without _ hst _ imaging and is likely the dominant source of ly@xmath0emission .
therefore , we stress that interpreting the ly@xmath0morphology and spectra requires deep high - resolution imaging and the determination of the systemic velocity through nir spectroscopy .
under the assumption that inflowing gas streams ( if any ) are randomly distributed , and from the non - detection of any blueshifted ly@xmath0h@xmath0or ly@xmath0[]offset in the direction of the eight embedded galaxies tested here ( although see section [ sec : cdfs - lab10 ] ) , we constrain the covering fraction of inflows to be @xmath51/8 ( 13% ) . likewise ,
if all of our v_ly@xmath3offsets are different projections of the same collimated outflow from the galaxies and are a proxy for outflow speed , the covering fraction of _ strong _ outflows , i.e. , super / hyper - winds , is less than @xmath213% .
the covering fraction of gas flows to which we are referring here has a different meaning than often discussed in the literature . in these theoretical papers ,
the covering factor is defined as how many sight - lines will be detected in metal or _ absorption _ when bright background sources close to the galaxies are targeted . what we measure in this paper is how often the inflowing gas is aligned with observer s sight - lines so that the column of neutral gas becomes optically thick and blue - shifts ly@xmath0lines against systemic velocity . nonetheless , predict that the covering factor within the virial radius ( @xmath89 @xmath45 75kpc ) from their simulated galaxies with a halo mass of @xmath90 = 3@xmath19@xmath91@xmath10 at @xmath92 is relatively small : @xmath23% and 10% for dlas and lyman limit systems ( lls ) , respectively . within 0.5@xmath19@xmath89 ,
this factor increases to @xmath210% and 30% , respectively , and presumably will be much higher directly towards the galaxies ( i.e. , for a pencil beam or looking down - the - barrel ) . note that ignore galactic winds in order to isolate the inflowing streams , so we expect that there will be enough dense material arising from gas accretion to affect the transfer of ly@xmath0photons in our experiments .
what is uncertain is that how the velocity field of this dense material near the galaxy will shift the emerging ly@xmath0profiles and the statistics of v_ly@xmath3 , because answering this requires full ly@xmath0radiative transfer treatment .
unlike rt models with simple geometry ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , cosmological simulations @xcite predict that the gas inflow can only slightly enhance the blue ly@xmath0peak , implying that the overall profile will be still red - peak dominated if the effects of outflows from the galaxies and igm absorption are fully considered . therefore , according to these simulations , our non - detection of infall signatures does not contradict the cold stream model .
the ly@xmath0profiles calculated from cold - stream models are in general the integrated profiles of ly@xmath0-emitting gas around the galaxies , which are not detectable by our study .
what we need are the predictions for how ly@xmath0profiles are modified and/or shifted against optically - thin lines along the line - of - sight to the galaxies embedded in the extended gas . in this way
, the distribution of predicted v_ly@xmath3can be compared directly with the observations .
it is also critical to take into account the effect of simultaneous outflows and inflows .
for example , the h@xmath0 and [ ] detections suggest that the galaxies embedded in the blobs are forming stars , which could generate mechanical feedback into the surrounding gas cloud .
thus , one might have expected that the innermost part of the gas cloud , close to the galaxies , has galactic scale outflows similar to those of other star - forming galaxies at @xmath93 ( e.g. , * ? ? ?
gas infall ( if any ) may dominate at larger radii ( up to @xmath250 kpc , the typical blob size ) . in this case
, the emerging ly@xmath0profile will be more sensitive to the core of the ly@xmath0blob , presumably the densest part of the cgm , than to the gas infall . as a result
, it might be difficult to detect the infalling gas by measuring v_ly@xmath3 . this kind of more realistic , infall+outflow model has not been considered yet in rt calculations , so its spectral signatures are unknown . in our sample of eight ly@xmath0blobs , the embedded galaxies have smaller velocity offsets ( @xmath48 = 160kms@xmath4 ) than those of lbgs ( 445kms@xmath4 ) . the remaining question is how to interpret these small v_ly@xmath3values .
we consider two possibilities here . first , if we assume a simple geometry where the outflowing material forms a spherical shell that consists of continuous media , v_ly@xmath3originates from the resonant scattering of the ly@xmath0photons at the shell as discussed in @xcite .
note that the emerging ly@xmath0profile is independent of the physical size of the shell as long as the shell have the same outflow velocity .
therefore , this model is applicable to ly@xmath0blobs . in this shell model , a central monochromatic point source is surrounded by an expanding shell of neutral gas with varying column density ( @xmath94 ) and doppler @xmath74 parameter ( e.g. , * ? ? ?
* ; * ? ? ?
a generic prediction is that the ly@xmath0emission is asymmetric , with the details of the line shape depending on the shell velocity , doppler parameter @xmath74 , and optical depth of column in the shell .
in this simple geometry , the v_ly@xmath3 values can be used as a proxy for outflow velocity of expanding shell , i.e. , @xmath12 @xmath45 0.5v_ly@xmath3 .
if this is the case , outflow velocities from blob galaxies are much smaller than the values expected from models of strong galactic winds .
furthermore , these offsets are even smaller than the typical ly@xmath0h@xmath0 offsets ( 250900kms@xmath4 ) of lbgs @xcite .
in particular , while cdfs - lab14 has the largest v_ly@xmath3 for its red peak , its blue and red peaks have similar intensity , which is a characteristic of a static medium or no bulk motions @xcite .
second , in an `` expanding bullet '' or `` clumpy cgm '' model @xcite , the outflowing material , which consists of small individual clumps , has a wide velocity range rather than a single value , and the ly@xmath0profiles are determined by the doppler shift that photons acquire when they are last scattered by the clumps just before escaping the system . in this case , the ly@xmath0h@xmath0 offsets are primarily modulated by the amount of gas that has @xmath95 component ( though it is not clear where this material is spatially located ) ; thus v_ly@xmath3 is not directly correlated with outflow velocity .
if this is the case , the small v_ly@xmath3 value of the galaxies in our ly@xmath0blobs simply indicates that there is less neutral gas at the galaxies systemic velocity ( presumably near the galaxy ) compared to lbgs .
testing these two hypotheses for ly@xmath0blobs or lbgs first requires a detailed comparison between high resolution ly@xmath0profiles and rt predictions .
this test is beyond the scope of this paper and will be discussed in the future . as emphasized by @xcite ,
one also needs to check the consistency of the ly@xmath0profiles with ism metal absorption profiles , which depends on high s / n continuum spectra that are not available here . nonetheless , we attempted such an analysis by stacking many low - ionization absorption lines in three galaxies ( section [ sec : absorption ] ) . for those three embedded galaxies with measured v_ly@xmath3 and @xmath50
, the outflow velocity estimates from both methods roughly agree within the uncertainties arising from low s / n and from rt complications .
however , closer inspection reveals a discrepancy between the observations and the simplest rt models in that we obtain @xmath50@xmath2 @xmath46200kms@xmath4and also v_ly@xmath3@xmath2 200kms@xmath4while the rt model with an expanding shell geometry predicts @xmath50= @xmath46@xmath12= @xmath96 v_ly@xmath3 .
this discrepancy has also appeared in lbgs @xcite , and @xcite show that the stacked ly@xmath0profile of lbgs with double - peaked profiles can not be reproduced accurately by the shell model . for a wide range of parameters ( @xmath12 , @xmath74 , @xmath97 ) , @xcite can not reproduce the location of the red peak , the width of the ly@xmath0 profile , and the metal absorption profile at the same time .
we attribute this discrepancy to the very simple nature of the shell model . by construction or by definition of the expanding `` shell ''
, the internal velocity dispersion of the media that constitute the shell should be much smaller than the expansion velocity of the shell itself , i.e. , @xmath12/@xmath74 @xmath88 1 .
therefore , one expects that the metal absorption lines arising from this thin shell have very narrow line widths @xmath98 similar to @xmath74 , typically @xmath2tens of kms@xmath4 .
however , such narrow absorption lines are not observed in either lbgs or ly@xmath0blobs . clearly , rt calculations with more realistic geometries and allowing a more thorough comparison with the observed ly@xmath0profiles are required .
given that the galaxies within our ly@xmath0blobs were selected , by definition , as ly@xmath0emitters with high equivalent width ( ew ) , the comparison with laes suggests that small v_ly@xmath3 values are a general characteristic of all high ew ly@xmath0-selected populations , be they compact or extended .
in addition to this work and @xcite , there are recent studies that measure v_ly@xmath3 for bright compact ly@xmath0-emitters @xcite and ly@xmath0blobs @xcite , which also find small ly@xmath0[]offsets for laes ranging from 35 340kms@xmath4 . a kolmogorov smirnov test fails to distinguish the v_ly@xmath3distribution of our eight ly@xmath0blobs from that of ten compact laes compiled from the four studies mentioned above .
thus , we now have a fairly large sample of high
ew ly@xmath0emitters ( compact or extended ) with v_ly@xmath3 = @xmath4660 @xmath47 + 400kms@xmath4 , which is clearly different from the distribution for lbgs ( see section [ sec : shift ] ) .
the smaller v_ly@xmath3of laes and labs might be related to their higher ly@xmath0equivalent widths .
an anti - correlation between ew(ly@xmath0 ) and v_ly@xmath3 in a compilation of lae and lbg samples has been suggested @xcite . even within the lbg population itself
there is an indication that v_ly@xmath3decreases as ew(ly@xmath0 ) increases .
@xcite find an anti - correlation between the ew(ly@xmath0 ) and the velocity offsets between interstellar absorption and ly@xmath0emission lines , @xmath99 .
is used in @xcite . ] from the stacked spectra of lbgs binned at different ew(ly@xmath0 ) , they show that with increasing ly@xmath0line strength from @xmath4615to 53 , @xmath100 decreases from 800kms@xmath4to 480kms@xmath4 .
because @xmath100 = v_ly@xmath3 + @xmath101 , we can infer that v_ly@xmath3is likely to decrease as ew(ly@xmath0 ) increases unless the observed anti - correlation is entirely due to @xmath101 . the origin of the apparent relationship between larger ew(ly@xmath0 ) and smaller v_ly@xmath3 is not understood .
it is possible that other physical properties drive this anti - correlation : for example , lbgs tend to have brighter continuum magnitudes , thus likely higher star formation rates and stellar masses , than are of typical laes . @xcite
propose that low column density , and thus a small number of resonant scatterings of ly@xmath0photons , might be responsible for the strong ly@xmath0emission and small v_ly@xmath3of ly@xmath0emitters .
what we do know from our v_ly@xmath3 measurements is that laes and ly@xmath0blobs are kinematically similar .
therefore , the mystery remains as to what powers ly@xmath0 nebulae , or , in other words , why certain galaxies have more spatially - extended ly@xmath0 gas ( i.e. , blobs ) than others ( i.e. , compact laes ) with similar v_ly@xmath3 and ew(ly@xmath0 ) . while the answer may be related to photo - ionization from ( buried ) agn ( e.g. , see section [ sec : type2 ] or @xcite ) , an extended proto - intracluster medium that can scatter or transport the ly@xmath0photons out to larger distances , or less dust to destroy ly@xmath0 photons ( cf .
* ) , discriminating among these possibilities will require a multi - wavelength analysis of a large sample of ly@xmath0blobs .
for now , the similarity here of labs and laes suggests that differences in gas kinematics are not responsible for the extended ly@xmath0halos . from the nearly symmetric ly@xmath0profile ( fig .
[ fig : cdfs - lab11 ] ) , and detection , and rest - frame optical line ratios , we hypothesize that an agn in cdfs - lab11 ionizes the gas surrounding the galaxies , making ly@xmath0 _ relatively _ optically thin and preventing the resonant scattering of ly@xmath0photons from dominating the shape of the profile .
while fitting the profile requires an underlying broad component ( fwhm @xmath45 480kms@xmath4 ) , the narrow component ( redshifted by @xmath45 80kms@xmath4 ) has a small velocity width ( fwhm @xmath45 110kms@xmath4 ) , comparable to that of the [ ] line ( @xmath4580kms@xmath4 ) .
therefore , if we assume that ly@xmath0and [ ] photons originate from the same region , resonant scattering by the igm / cgm does not broaden the intrinsic profile significantly . in other words , at least along our los , the optical depth of cdfs - lab11 s igm / cgm is smaller than for the other ly@xmath0blobs in our sample and for typical ly@xmath0emitters at high redshifts .
a nearly symmetric and narrow ly@xmath0profile such as cdfs - lab11 s is rarely observed among high-@xmath1 ly@xmath0-emitting galaxies .
however , ly@xmath0radiative transfer calculations show that such ly@xmath0profiles can be generated in the presence of a central ionizing source . for example , @xcite investigate the emerging ly@xmath0profiles from an collapsing / expanding sphere centered on an ionizing source ( an agn ) for different luminosities ( see their figures 11 and 12 ) .
they find that if the ionizing source is strong and the ly@xmath0is _ relatively _ optically thin ( line - center optical depth @xmath68 @xmath5 @xmath102 ) , the separation of the characteristic double - peaks become smaller ( @xmath77 @xmath103 100kms@xmath4 ) as the ionizing source gets stronger .
the overall profile becomes symmetric , and is blue or redshifted depending on the flow direction , but again by small amount ( v_ly@xmath3@xmath103 100kms@xmath4 ) .
these authors were skeptical that the sign of the ly@xmath0shift and the small ly@xmath0offset ( v_ly@xmath3@xmath2 a few tens kms@xmath4 ) from the systemic velocity could be measured .
however , with the right strategy ( i.e. , high spectral resolution and a careful choice of survey redshift ) , we are able to reliably measure the predicted small offset .
clearly , more detailed comparisons with the rt models are required . a broad , asymmetric ly@xmath0profile with a sharp blue edge ( e.g. , cdfs - lab06 and 07 ) is characteristic of high-@xmath1 ly@xmath0-emitters . in spectroscopic
follow - up observations of ly@xmath0-emitter candidates , where spectral coverage is limited , these characteristics alone are often used to discriminate high-@xmath1 ly@xmath0emitters from possible low-@xmath1 interlopers .
the example of cdfs - lab11 demonstrates that caution is required because a narrow and symmetric ly@xmath0line can also arise when the ism or cgm of a candidate galaxy is significantly photo - ionized , e.g. , by agn . on the bright side ,
our v_ly@xmath3 velocity offset technique could be used for studying gas infall or outflow in high-@xmath1 qsos instead of relying on only the ly@xmath0line profile ( e.g. , * ? ? ?
exploring the origin of ly@xmath0nebulae ( blobs " ) at high redshift requires measurements of their gas kinematics that are difficult with only the resonant , optically - thick ly@xmath0line . to define gas motions relative to the systemic velocity of the nebula
, the ly@xmath0line must be compared with non - resonant lines , which are not much altered by radiative transfer effects .
we made first comparison of non - resonant h@xmath0 @xmath116563to extended ly@xmath0emission for two bright ly@xmath0blobs in @xcite , concluding that , within the context of a simple radiative transfer model , the gas was static or mildly outflowing at @xmath104250kms@xmath4 .
however , it was unclear if these two ly@xmath0blobs , which are the brightest in the sample , are representative of the general ly@xmath0blob population .
furthermore , geometric effects infall along filaments or bi - polar outflows might hide bulk motions of the gas when only viewed from the two directions toward these two ly@xmath0blobs .
with vlt x - shooter , we obtain optical and near - infrared spectra of six additional ly@xmath0blobs from the @xcite sample . with a total of eight ly@xmath0blobs , we investigate the gas kinematics within ly@xmath0blobs using three techniques : the ly@xmath0offset from the systemic velocity ( v_ly@xmath3 ) , the shape and shift ( @xmath50 ) of the ism metal absorption line profiles , and the breadth of the [ ] line profile .
our findings are : 1 .
both ly@xmath0and non - resonant lines confirm that these blobs lie at the survey redshift ( @xmath34 ) .
we also detect the [ ] @xmath353727,3729 , [ ] @xmath114959 , [ ] @xmath115007 , and h@xmath14 @xmath114861lines .
all non - resonant line velocities are consistent with each other and with arising from the galaxy or galaxies embedded in the ly@xmath0blob .
[ ] , which is observed at high signal - to - noise in all cases and whose profile is an rt - independent constraint on the gas kinematics , is a particularly good diagnostic line for this redshift and instrument .
the majority of the blobs ( 6/8 ) have broadened ly@xmath0profiles indicating radiative transfer effects .
these ly@xmath0profiles are consistent with being in the same family of objects as predicted by rt , with profile shapes ranging from symmetric double - peaked , to asymmetric red peak dominated , to a single red peak .
the fraction of double - peaked profiles is @xmath238% ( 3/8 ) .
3 . the narrow ly@xmath0profile systems ( cdfs - lab01a , cdfs - lab11 )
, whose ly@xmath0profile is not significantly broader than the [ ] or h@xmath0lines , have the smallest v_ly@xmath3offsets , the most spatially compact ly@xmath0emission , and the only and lines detected , implying that a hard ionizing source , possibly an agn , is responsible for the lower optical depth toward the central embedded galaxies .
4 . with a combination of v_ly@xmath3 , the interstellar metal absorption line profile , and a new indicator , the spectrally - resolved [ ] line profile , we detect gas moving along the line of sight to galaxies embedded in the ly@xmath0blob center .
although not all three indicators are available for all ly@xmath0blobs , the implied speeds and direction are roughly consistent for the sample , suggesting a simple picture in which the gas is stationary or slowly outflowing at a few hundred kms@xmath4from the embedded galaxies .
these outflow speeds are similar to those of laes , suggesting that outflow speed is not the dominant driver of extended ly@xmath0emission .
furthermore , these outflow speeds exclude models in which star formation or agn produce `` super '' or `` hyper '' winds of up to @xmath21000kms@xmath4 . + more specifically : 1 .
we compare the non - resonant emission lines [ ] and h@xmath0to the ly@xmath0profile to obtain the velocity offset v_ly@xmath3 .
the galaxies embedded within our ly@xmath0blobs have smaller v_ly@xmath3than those of lbgs , confirming the previous claims @xcite .
the galaxies within ly@xmath0blobs have v_ly@xmath3= @xmath4660 @xmath47 + 400kms@xmath4 with an average of @xmath48 = 160kms@xmath4 , while lbgs at similar redshifts have v_ly@xmath3= 250 900kms@xmath4 .
the small v_ly@xmath3 in the ly@xmath0blobs are consistent with those measured for compact laes .
2 . by stacking low - ionization metal absorption lines
, we measure the outflow velocity of neutral gas in front of the galaxies in the ly@xmath0blobs .
galaxies in two ly@xmath0blobs show an outflow speed of @xmath2250kms@xmath4 , while another has an almost symmetric absorption line profile centered at @xmath50= 0kms@xmath4 , consistent with no significant bulk motion .
the ism absorption line profiles here have low s / n , but are very roughly consistent with those of some lbgs ( at the level of several hundred kms@xmath4outflows ) .
the high spectral resolution of our data reveals broad wings in the [ ] profiles of four ly@xmath0blobs .
this new kinematic diagnostic suggests warm ionized outflows driven by supernovae and stellar winds .
these broad line components are narrower ( @xmath54 = 45 120kms@xmath4 ) and have a maximum blueshifted velocity ( @xmath65= 150 260kms@xmath4 ) smaller than those of @xmath64 star - forming galaxies ( sfgs ) , implying weaker outflows here than for lbgs and sfgs at similar redshifts .
4 . if we assume that the detected outflows are different projections of the same outflow from the ly@xmath0blob center , we can estimate the effects of flow geometry on our measurements given that our large sample size allows averaging over many lines - of - sight .
the absence of any strong ( @xmath21000kms@xmath4 ) outflows among the eight galaxies tested is not a projection effect : their covering fraction is @xmath51/8 ( 13% ) .
likewise , the lack of a blue - peak dominated ly@xmath0profile , at least in the direction of the embedded galaxies ( see point 6 .
below ) , implies that the covering factor of any cold streams @xcite is less than 13% .
the channeling of gravitational cooling radiation into ly@xmath0may not be significant over the radii probed by our techniques here .
constraining the physical state of ly@xmath0-emitting gas in a ly@xmath0blob is a critical step in understanding its emission mechanism and in comparing to simulations . for one ly@xmath0blob whose ly@xmath0profile and ism metal absorption lines suggest no significant bulk motion ( cdfs - lab14 ) , at least in its cool and neutral gas , we assume a simple rt model and make the first column density measurement of gas in a embedded galaxy , finding that it is consistent with a dla .
for one peculiar system ( cdfs - lab10 ) , we discover blueshifted ly@xmath0emission that is _ not _ directly associated with any embedded galaxy .
this ly@xmath0emitting gas is blueshifted relative to two embedded galaxies , suggesting that it arose from a tidal interaction between the galaxies or is actually flowing into the blob center .
the former is expected in these overdense regions , where _ hst _
images resolve many galaxies .
the latter might signify the predicted but elusive cold gas accretion along filaments .
we thank the anonymous referee for her or his thorough reading of the manuscript and helpful comments .
the authors thank daniel eisenstein for his contributions at the start of this project .
we thank jason x. prochaska for helpful discussions .
yy thanks the mpia enigma group for the helpful discussions .
yy also thanks the theoretical astrophysics center at the university of california , berkeley for the travel support , as well as claude - andr faucher - gigure , duan kere and daniel kasen for the helpful discussions during that stay .
yy also thanks sangeeta malhotra for a helpful discussion regarding the lae and lbg connection .
yy acknowledges support from the bmbf / dlr grant nr .
50 or 1306 .
thanks the max - planck - institut fr astronomie and the center for cosmology and particle physics at new york university for their hospitality and support during her stays there . a.i.z . acknowledges support from the nsf astronomy and astrophysics research program through grant ast-0908280 and from the nasa astrophysics data analysis program through grant nnx10ad47 g .
she also thank the generosity of the john simon guggenheim memorial foundation .
facilities : bland , j. , & tully , b. 1988 , , 334 , 43 cen , r. , & zheng , z. 2013 , , 775 , 112 colbert , j. w. , teplitz , h. , francis , p. , palunas , p. , williger , g. m. , & woodgate , b. 2006 , , 637 , l89 damen , m. , labb , i. , van dokkum , p. g. , et al .
2011 , , 727 , 1 dekel , a. , et al .
2009 , , 457 , 451 dey , a. , et al .
2005 , , 629 , 654 dijkstra , m. , haiman , z. , & spaans , m. 2006 , , 649 , 14 dijkstra , m. , haiman , z. , & spaans , m. 2006 , , 649 , 37 dijkstra , m. , & loeb , a. 2009 , , 400 , 1109 erb , d. k. , bogosavljevi , m. , & steidel , c. c. 2011 , , 740 , l31 fardal , m. a. , katz , n. , gardner , j. p. , hernquist , l. , weinberg , d. h. , & dav ' e , r. 2001 , , 562 , 605 faucher - gigure , c .- a .
, kere , d. , dijkstra , m. , hernquist , l. , & zaldarriaga , m. 2010 , , 725 , 633 faucher - gigure , c .- a .
, & kere , d. 2011 , , 412 , l118 finkelstein , s. l. , hill , g. j. , gebhardt , k. , et al .
2011 , , 729 , 140 foreman - mackey , d. , hogg , d. w. , lang , d. , & goodman , j. 2013 , , 125 , 306 francis , p. j. , et al . 2001 , , 554 , 1001 fumagalli ,
m. , prochaska , j. x. , kasen , d. , et al .
2011 , , 418 , 1796 fynbo , j. u. , mller , p. , & warren , s. j. 1999 , , 305 , 849 gawiser , e. , et al .
2006 , , 162 , 1 geach , j. e. , et al .
2009 , , 700 , 1 genzel , r. , newman , s. , jones , t. , et al .
2011 , , 733 , 101 goerdt , t. , dekel , a. , sternberg , a. , ceverino , d. , teyssier , r. , & primack , j. r. 2010 , , 933 guaita , l. , francke , h. , gawiser , e. , et al .
2013 , , 551 , a93 haiman , z. , & rees , m. j. 2001 , , 556 , 87 haiman , z. , spaans , m. , & quataert , e. 2000 , , 537 , l5 harrington , j. p. 1973
, , 162 , 43 hashimoto , t. , ouchi , m. , shimasaku , k. , et al .
2013 , , 765 , 70 hayes , m. , scarlata , c. , & siana , b. 2011 , , 476 , 304 hayes , m. , stlin , g. , schaerer , d. , et al .
2013 , , 765 , l27 hennawi , j. f. , prochaska , j. x. , kollmeier , j. , & zheng , z. 2009 , , 693 , l49 keel , w. c. , cohen , s. h. , windhorst , r. a. , & waddington , i. 1999 , , 118 , 2547 kelson , d. d. 2003 , , 115 , 688 kere , d. , katz , n. , weinberg , d. h. , & dav , r. 2005 , , 363 , 2 kere , d. , katz , n. , fardal , m. , dav , r. , & weinberg , d. h. 2009 , , 395 , 160 kimm , t. , slyz , a. , devriendt , j. , & pichon , c. 2011 , , 413 , l51 kollmeier , j. a. , zheng , z. , dav , r. , gould , a. , katz , n. , miralda - escud , j. , & weinberg , d. h. 2010 , , 708 , 1048 kulas , k. r. , shapley , a. e. , kollmeier , j. a. , et al .
2012 , , 745 , 33 latif , m. a. , schleicher , d. r. g. , spaans , m. , & zaroubi , s. 2011 , , 413 , l33 lehmer , b. d. , brandt , w. n. , alexander , d. m. , et al .
2005 , , 161 , 21 luo , b. , bauer , f. e. , brandt , w. n. , et al .
2008 , , 179 , 19 martin , d. c. , chang , d. , matuszewski , m. , et al .
2014 , arxiv:1402.4809 matsuda , y. , et al .
2004 , , 128 , 569 matsuda , y. , yamada , t. , hayashino , t. , yamauchi , r. , & nakamura , y. 2006 , , 640 , l123 matsuda , y. , yamada , t. , hayashino , t. , et al .
2011 , , 410 , l13 mclinden , e. m. , finkelstein , s. l. , rhoads , j. e. , et al .
2011 , , 730 , 136 mclinden , e. m. , malhotra , s. , rhoads , j. e. , et al .
2013 , , 767 , 48 mller , p. , warren , s. j. , & fynbo , j. u. 1998 , , 330 , 19 nakajima , k. , ouchi , m. , shimasaku , k. , et al .
2012 , , 745 , 12 neufeld , d. a. 1990 , , 350 , 216 newman , s. f. , genzel , r. , frster - schreiber , n. m. , et al .
2012 , , 761 , 43 nilsson , k. k. , fynbo , j. p. u. , mller , p. , sommer - larsen , j. , & ledoux , c. 2006 , , 452 , l23 ohyama , y. , et al .
2003 , , 591 , l9 ouchi , m. , et al .
2009 , , 696 , 1164 prescott , m. k. m. , kashikawa , n. , dey , a. , & matsuda , y. 2008 , , 678 , l77 prescott , m. k. m. , dey , a. , & jannuzi , b. t. 2009 , , 702 , 554 prescott , m. k. m. , dey , a. , & jannuzi , b. t. 2012 , , 748 , 125 prescott , m. k. m. , dey , a. , brodwin , m. , et al .
2012 , , 752 , 86 rauch , m. , becker , g. d. , haehnelt , m. g. , et al .
2011 , , 418 , 1115 rix , h .- w . , et al .
2004 , , 152 , 163 rosdahl , j. , & blaizot , j. 2012 , , 423 , 344 saito , t. , shimasaku , k. , okamura , s. , ouchi , m. , akiyama , m. , & yoshida , m. 2006 , , 648 , 54 saito , t. , shimasaku , k. , okamura , s. , ouchi , m. , akiyama , m. , yoshida , m. , & ueda , y. 2008 , , 675 , 1076 scarlata , c. , et al .
2009 , , 706 , 1241 schaerer , d. , & verhamme , a. 2008 , , 480 , 369 shapiro , k. l. , genzel , r. , quataert , e. , et al .
2009 , , 701 , 955 shapley , a. e. , steidel , c. c. , pettini , m. , & adelberger , k. l. 2003 , , 588 , 65 smith , d. j. b. , & jarvis , m. j. 2007 , , 378 , l49 soto , k. t. , martin , c. l. , prescott , m. k. m. , & armus , l. 2012 , , 757 , 86 steidel , c. c. , adelberger , k. l. , shapley , a. e. , pettini , m. , dickinson , m. , & giavalisco , m. 2000 , , 532 , 170 steidel , c. c. , shapley , a. e. , pettini , m. , adelberger , k. l. , erb , d. k. , reddy , n. a. , & hunt , m. p. 2004
, , 604 , 534 steidel , c. c. , erb , d. k. , shapley , a. e. , et al .
2010 , , 717 , 289 taniguchi , y. , shioya , y. , & kakazu , y. 2001 , , 562 , l15 taniguchi , y. & shioya , y. 2000 , , 532 , l13 verhamme , a. , schaerer , d. , & maselli , a. 2006 , , 460 , 397 verhamme , a. , schaerer , d. , atek , h. , & tapken , c. 2008 , , 491 , 89 vernet , j. , dekker , h. , dodorico , s. , et al . 2011 , , 536 , a105 weidinger , m. , mller , p. , & fynbo , j. p. u. 2004 , , 430 , 999 weijmans , a .- m . , bower , r. g. , geach , j. e. , et al .
2010 , , 402 , 2245 westmoquette , m. s. , exter , k. m. , smith , l. j. , & gallagher , j. s. 2007 , , 381 , 894 wilman , r. j. , gerssen , j. , bower , r. g. , morris , s. l. , bacon , r. , de zeeuw , p. t. , & davies , r. l. 2005 , , 436 , 227 yamada , t. , matsuda , y. , kousai , k. , et al .
2012 , , 751 , 29 yang , y. , zabludoff , a. , tremonti , c. , eisenstein , d. , & dav , r. 2009 , , 693 , 1579 yang , y. , zabludoff , a. , eisenstein , d. , & dav , r. 2010 , , 719 , 1654 yang , y. , zabludoff , a. , jahnke , k. , et al .
2011 , , 735 , 87 yang , y. , walter , f. , decarli , r. , et al .
2014 , , 784 , 171
to test if two velocity components are needed to fit our spectrally - resolved [ ] profiles and to extract the line parameters , we employ the [ ] profiles with two gaussian functions using a markov chain monte carlo ( mcmc ) technique .
we use the emcee software @xcite to sample the distributions of six parameters : line center , width , and fluxes for the two gaussian profiles .
we require that ( 1 ) the line widths of both components are larger than the instrumental line width ( 31kms@xmath4 ; the vertical dotted lines in figure [ fig : o3mcmc ] ) , ( 2 ) the peak of each component should be at least 5% of the observed peak intensity , and ( 3 ) the peak of the narrow component is higher than that of the broad component . the latter two priors are imposed to prevent the mcmc chains from getting stuck in parameter spaces with extremely broad lines but with negligible fluxes . in figure
[ fig : o3mcmc ] , we show the likelihood distributions of the line widths of the two components : @xmath105 and @xmath106 .
note that we exclude cdfs - lab10 in this analysis because its neighboring galaxy ( cdfs - lab10a ) makes it difficult to reliably extract its profile ( see sections [ sec : o3profile ] and [ sec : cdfs - lab10 ] ) .
except for cdfs - lab11 , the likelihood distributions of the two line - widths do not overlap significantly , and the peaks of the joint 2d distributions are not located near the @xmath105 = @xmath54 line ( the dashed line ) .
therefore , we conclude that the remaining four systems ( cdfs - lab06 , 07 , 13 , 14 ) are likely to consist of two components . the uncertainties of the parameters are determined from the 68.2% confidence interval of the marginalized distribution .
c cc cc ccccc cdfs - lab06 & 1.57 @xmath30 0.07 & 56 @xmath30 3 & 03:32:19.78 & -27:47:31.5 & 2011 - 01 - 07 & 1.07 & 0.80 & 0.76 & 0.80 + cdfs - lab07 & 1.14 @xmath30 0.05 & 58 @xmath30 3 & 03:32:03.36 & -27:45:24.4 & 2011 - 01 - 06 & 1.05 & 0.77 & 0.76 & 0.80 + & & & & & 2011 - 01 - 06 & 1.18 & 0.98 & 0.76 & 0.80 + & & & & & 2011 - 01 - 28 & 1.12 & 0.81 & 0.76 & 0.80 + cdfs - lab10 & 0.71 @xmath30 0.04 & 318 @xmath30 36 & 03:32:37.45 & -28:02:05.7 & 2010 - 11 - 06 & 1.00 & 0.71 & 0.76 & 0.80 + & & & & & 2010 - 11 - 07 & 1.00 & 0.68 & 0.76 & 0.80 + & & & & & 2010 - 11 - 07 & 1.08 & 1.13 & 0.76 & 0.80 + & & & & & 2010 - 11 - 08 & 1.08 & 1.23 & 0.76 & 0.80 + cdfs - lab11 & 1.02 @xmath30 0.04 & 120 @xmath30 10 & 03:32:43.25 & -27:42:58.3 & 2011 - 01 - 08 & 1.07 & 0.75 & 0.76 & 0.80 + & & & & & 2011 - 01 - 08 & 1.23 & 0.85 & 0.76 & 0.80 + cdfs - lab13 & 0.94 @xmath30 0.04 & 184 @xmath30 25 & 03:32:32.75 & -27:39:06.4 & 2011 - 01 - 07 & 1.22 & 0.79 & 0.76 & 0.80 + & & & & & 2011 - 01 - 08 & 1.01 & 0.70 & 0.76 & 0.80 + cdfs - lab14 & 0.93 @xmath30 0.05 & 67 @xmath30 5 & 03:32:32.29 & -27:41:26.4 & 2011 - 01 - 09 & 1.07 & 0.80 & 0.76 & 0.80 + & & & & & 2011 - 01 - 09 & 1.24 & 0.73 & 0.76 & 0.80 + + cdfs - lab01 & 8.02 @xmath30 0.24 & 512 @xmath30 50 & & & & & & & + cdfs - lab02 & 2.88 @xmath30 0.12 & 43 @xmath30 2 & & & & & & & [ tab : obslog ] c ccc r ccc l cdfs - lab01 & @xmath4665 @xmath30 20 & 228 @xmath30 14 & 45.2 @xmath30 1.6 & & & & & ( 1 ) + cdfs - lab11 & 84 @xmath30 6 & 79 @xmath30 3 & 26.5 @xmath30 0.7 & & & & & ( 1 ) + cdfs - lab10 & 247 @xmath30 147 & 425 @xmath30 97 & 18.7 @xmath30 2.4 & & & & & ( 1 ) + cdfs - lab13 & 152 @xmath30 14 & 100 @xmath30 9 & 13.7 @xmath30 0.7 & & @xmath46277 @xmath30 8 & 43 @xmath30 8 & 4.0 @xmath30 0.5 & ( 2 ) +
cdfs - lab06 & 123 @xmath30 15 & 150 @xmath30 13 & 25.7 @xmath30 1.3 & & & & & ( 1 ) + cdfs - lab02 & 211 @xmath30 43 & 192 @xmath30 35 & 8.2 @xmath30 0.9 & & @xmath46342 @xmath30 117 & 273 @xmath30 137 & 3.3 @xmath30 1.3 & ( 2 ) + cdfs - lab07 & 181 @xmath30 15 & 168 @xmath30 10 & 22.2 @xmath30 0.8 & & & & & ( 1 ) + cdfs - lab14 & 371 @xmath30 34 & 202 @xmath30 24 & 18.0 @xmath30 1.2 & & @xmath46404 @xmath30 17 & 176 @xmath30 18 & 12.4 @xmath30 1.1 & ( 3 ) [ tab : lyman ] c cc c cc c cdfs - lab06 & 38 @xmath30 3 & @xmath10739 @xmath30 27 & 55 @xmath30 10 & 0.51@xmath108 & 0.35 @xmath30 0.17 & 152 @xmath30 15 + cdfs - lab07 & 44 @xmath30 2 & @xmath4697 @xmath30 23 & 64 @xmath30 10 & 0.44@xmath109 & 0.32 @xmath30 0.12 & 222 @xmath30 11 + cdfs - lab13 & 35 @xmath30 5 & @xmath4656 @xmath30 21 & 60 @xmath30 11 & 0.84@xmath110 & 0.46 @xmath30 0.13 & 174 @xmath30 18 + cdfs - lab14 & 67 @xmath30 16 & @xmath4612 @xmath30 71 & 125 @xmath30 40 & 0.47@xmath111 & 0.33 @xmath30 0.20 & 262 @xmath30 75 + cdfs - lab10 & 71 @xmath30 1 & & & & & + cdfs - lab11 & 47 @xmath30 1 & & & & & [ tab : o3profile ] | with vlt / x - shooter , we obtain optical and near - infrared spectra of six ly@xmath0blobs at @xmath1 @xmath2 2.3 . for a total sample of eight ly@xmath0blobs ( including two that we have previously studied ) , the majority ( 6/8 )
have broadened ly@xmath0profiles with shapes ranging from a single peak to symmetric or asymmetric double - peaked .
the remaining two systems , in which the ly@xmath0profile is not significantly broader than the [ ] or h@xmath0emission lines , have the most spatially compact ly@xmath0emission , the smallest offset between the ly@xmath0and the [ ] or h@xmath0line velocities , and the only detected and lines in the sample , implying that a hard ionizing source , possibly an agn , is responsible for their lower optical depth . using three measures
the velocity offset between the ly@xmath0line and the non - resonant [ ] or h@xmath0line ( v_ly@xmath3 ) , the offset of stacked interstellar metal absorption lines , and a new indicator , the spectrally - resolved [ ] line profile we study the kinematics of gas along the line of sight to galaxies within each blob center .
these three indicators generally agree in velocity and direction , and are consistent with a simple picture in which the gas is stationary or slowly outflowing at a few hundred kms@xmath4from the embedded galaxies .
the absence of stronger outflows is not a projection effect : the covering fraction for our sample is limited to @xmath51/8 ( 13% ) .
the outflow velocities exclude models in which star formation or agn produce `` super '' or `` hyper '' winds of up to @xmath21000kms@xmath4 .
the v_ly@xmath3offsets here are smaller than typical of lyman break galaxies ( lbgs ) , but similar to those of compact ly@xmath0emitters .
the latter suggests a connection between blob galaxies and ly@xmath0emitters and that outflow speed can not be a dominant factor in driving extended ly@xmath0emission .
for one ly@xmath0blob ( cdfs - lab14 ) , whose ly@xmath0profile and metal absorption line offsets suggest no significant bulk motion , we use a simple radiative transfer model to make the first column density measurement of gas in an embedded galaxy , finding it consistent with a damped ly@xmath0absorption system .
overall , the absence of clear inflow signatures suggests that the channeling of gravitational cooling radiation into ly@xmath0is not significant over the radii probed here .
however , one peculiar system ( cdfs - lab10 ) has a blueshifted ly@xmath0component that is not obviously associated with any galaxy , suggesting either displaced gas arising from tidal interactions among blob galaxies or gas flowing into the blob center .
the former is expected in these overdense regions , where _ hst _
images resolve many galaxies .
the latter might signify the predicted but elusive cold gas accretion along filaments . |
we acknowledge partial financial support by dgu - mec ( spanish ministry of education ) through project phb2007 - 0095-pc , and by cnpq , faperj and capes ( brazilian agencies ) .
00 j.w .
gibbs , elementary principles in statistical mechanics developed with especial reference to the rational foundation of thermodynamics , c. scribner s sons , new york , 1902 ; yale university press , new haven , 1948 ; ox bow press , woodbridge , connecticut , 1981 ( page 35 ) .
p. jund , s.g .
kim and c. tsallis , phys .
b * 52 * , 50 ( 1995 ) ; j.r .
grigera , phys .
a * 217 * , 47 ( 1996 ) ; s.a .
cannas and f.a .
tamarit , phys .
b * 54 * , r12661 ( 1996 ) ; l.c .
sampaio , m.p .
de albuquerque and f.s .
de menezes , phys .
b * 55 * , 5611 ( 1997 ) ; c. anteneodo and c. tsallis , phys . rev
. lett . * 80 * , 5313 ( 1998 ) ; curilef and c. tsallis , phys .
lett . a * 264 * , 270 ( 1999 ) ; r.f.s . andrade and s.t.r .
pinho , phys .
e * 71 * , 026126 ( 2005 ) .
let us remind that the @xmath1-gaussians for @xmath64 are also called _
@xmath99-distributions _ in the plasma physics community , and also occasionally _ generalized lorentzians _ ; for special rational values of @xmath1 they correspond to the student s @xmath116-distributions for an integer number of degrees of freedom . for special rational values of @xmath117 ,
@xmath1-gaussians correspond to the @xmath16-distributions ( which have a compact support ) .
u. tirnakli , c. tsallis and m.l .
lyra , eur .
j. b * 11 * , 309 ( 1999 ) ; f. baldovin and a. robledo , europhys .
60 * , 518 ( 2002 ) ; f. baldovin and a. robledo , phys .
e * 66 * , r045104 ( 2002 ) ; e.p .
borges , c. tsallis , g.f.j .
ananos and p.m.c .
oliveira , phys .
* 89 * , 254103 ( 2002 ) ; f. baldovin and a. robledo , phys .
e * 69 * , 045202(r ) ( 2004 ) ; e. mayoral and a. robledo , physica a * 340 * , 219 ( 2004 ) ; g.f.j .
ananos and c. tsallis , phys . rev
. lett . * 93 * , 020601 ( 2004 ) .
g. casati , c. tsallis and f. baldovin , europhys . lett . * 72 * , 355 ( 2005 ) ; e. mayoral and a. robledo , phys . rev .
e * 72 * , 026209 ( 2005 ) ; g. ruiz and c. tsallis , eur .
j. b * 67 * , 577 ( 2009 ) .
j. s. andrade jr . , g.f.t .
da silva , a.a .
moreira , f.d .
nobre and e.m.f .
curado , phys .
lett . * 105 * , 260601 ( 2010 ) ; y. levin and r. pakter , phys .
lett . * 107 * , 088901 ( 2011 ) ; j. s. andrade jr .
da silva , a.a .
moreira , f.d .
nobre and e.m.f .
curado , phys .
lett . * 107 * , 088902 ( 2011 ) ; m.s .
ribeiro , f.d .
nobre and e.m.f .
curado , in special issue _ tsallis entropy _ , ed .
a. anastasiadis , entropy * 13 * , 1928 ( 2011 ) ; m.s .
ribeiro , f.d .
nobre and e.m.f .
curado , phys .
e * 85 * , 021146 ( 2012 ) ; m.s .
ribeiro , f.d .
nobre and e.m.f .
curado , _ overdamped motion of interacting particles in general confining potentials : time - dependent and stationary - state analyses _ , eur . phys .
j. b ( 2012 ) , in press .
plastino and a. plastino , physica a * 222 * , 347 ( 1995 ) ; c. tsallis and d.j .
bukman , phys .
e * 54 * , r2197 ( 1996 ) ; l.c .
malacarne , r.s .
mendes , i.t .
pedron and e.k .
lenzi , phys .
e * 65 * , 052101 ( 2002 ) .
l. borland , phys .
lett . a * 245 * , 67 ( 1998 )
; l. borland , f. pennini , a. r. plastino and a. plastino eur .
j. b * 12 * , 285 ( 1999 ) ; c. anteneodo and c. tsallis , j. math .
phys . * 44 * , 5194 ( 2003 ) ; v. schwammle , e.m.f .
curado and f.d .
nobre , eur .
j. b * 58 * , 159 - 165 ( 2007 ) ; v. schwammle , f.d . nobre and e.m.f .
curado , phys .
e * 76 * , 041123 ( 2007 ) ; m.a .
fuentes and m.o .
caceres , phys .
lett . a 372 , 1236 ( 2008 ) [ to compare with the present paper
, it must be done @xmath118 ; b.c.c . dos santos and c. tsallis , phys .
e * 82 * , 061119 ( 2010 ) ; a. mariz and c. tsallis , phys .
a * 376 * , 3088 ( 2012 ) .
burlaga and a.f.vinas , physica a * 356 * , 375 ( 2005 ) ; l.f .
burlaga , a.f .
vinas , n.f .
ness and m.h .
acuna , astrophys .
j. * 644 * , l83 ( 2006 ) ; l.f .
burlaga and n.f .
ness , astrophys .
j. * 703 * , 311 ( 2009 ) ; l.f .
burlaga , n.f .
ness , m.h .
acuna , astrophys .
j. * 691 * , l82 ( 2009 ) ; j. cho , a. lazarian , astrophys .
j. * 701 * , 236 ( 2009 ) ; l.f .
burlaga and n.f .
ness , astrophys .
j. * 725 * , 1306 ( 2010 ) 1306 ; a. esquivel , a. lazarian , astrophys
. j. * 710 * , 125 ( 2010 ) ; l.f .
burlaga and n.f .
ness , astrophys .
j. * 737 * , 35 ( 2011 ) .
p. douglas , s. bergamini and f. renzoni , phys .
* 96 * , 110601 ( 2006 ) .
b. liu and j. goree , phys .
lett . * 100 * , 055003 ( 2008 ) ; r.m .
pickup , r. cywinski , c. pappas , b. farago , p. fouquet , phys .
lett . * 102 * , 097202 ( 2009 ) ; r.g .
devoe , phys .
102 * , 063001 ( 2009 ) .
cms collaboration , j. high energy phys . *
02 * , 041 ( 2010 ) ; cms collaboration , phys .
lett . * 105 * , 022002 ( 2010 ) ; cms collaboration , j. high energy phys . *
09 * , 091 ( 2010 ) ; cms collaboration , j. high energy phys . * 08 * , 086 ( 2011 ) ; cms collaboration , j. high energy phys .
* 05 * , 064 ( 2011 ) ; alice collaboration , phys .
b * 693 * , 53 ( 2010 ) ; alice collaboration , eur .
j. c * 71 * , 1594 ( 2011 ) ; alice collaboration , eur .
j. c * 71 * , 1655 ( 2011 ) ; atlas collaboration , new j. physics * 13 * , 053033 ( 2011 ) ; phenix collaboration , phys . rev .
d * 83 * , 052004 ( 2011 ) ; phenix collaboration , phys .
c * 83 * , 024909 ( 2011 ) ; phenix collaboration , phys .
c * 83 * , 064903 ( 2011 ) ; phenix collaboration , phys .
c * 84 * , 044902 ( 2011 ) ; m. shao , l. yi , z.b .
tang , h.f .
chen , c. li , z.b .
xu , j. phys .
g * 37 * , 084104 ( 2010 ) ; a. tawfik , nuclear phys .
a * 859 * , 63 ( 2011 ) .
u. tirnakli , c. beck and c. tsallis , phys . rev .
e * 75 * , 040106(r ) ( 2007 ) ; u. tirnakli , c. tsallis and c. beck , phys .
e * 79 * , 056209 ( 2009 ) ; c. tsallis and u. tirnakli , j. phys .
series * 201 * , 012001 ( 2010 ) . c. tsallis , m. gell - mann and y. sato , proc .
sci . * 102 * , 15377 ( 2005 ) ; f. caruso and c. tsallis , phys .
e * 78 * , 021101 ( 2008);a .
saguia and m.s .
sarandy , phys .
a * 374 * , 3384 ( 2010 ) .
s. umarov , c. tsallis and s. steinberg , milan j. math . * 76 * , 307 ( 2008 ) ; for a simplified version , see s.m.d .
queiros and c. tsallis , aip conference proceedings * 965 * , 21 ( new york , 2007 ) .
see also s. umarov , c. tsallis , m. gell - mann and s. steinberg , j. math . phys . * 51 * , 033502 ( 2010 ) .
a gap has been recently detected , by h.j .
hilhorst , j. stat .
mech . , p10023 ( 2010 ) , in the proofs of @xcite , namely that the @xmath1-fourier transform has not , for @xmath119 , an inverse in the simple sense that the standard fourier transform has .
this point certainly is interesting on its own , even if the thesis of the theorems are consistent with alternative proofs that do not use the @xmath1-fourier transform .
very recently it has been shown by m. jauregui and c. tsallis , phys .
a * 375 * , 2085 ( 2011 ) , m. jauregui , c. tsallis and e.m.f .
curado , j. stat .
mech . , p10016 ( 2011 ) , a. plastino and m.c .
rocca , physica a * 391 * , 4740 ( 2012 ) , and a. plastino and m.c .
rocca , milan j. math .
( 2012 ) , doi 10.1007/s00032 - 012 - 0179 - 6 , that an _ unique _ inverse of the @xmath1-fourier transform does exist if some supplementary information is given ( something which is unnecessary only for @xmath120 ) .
the use of these results in order to fill the mentioned gap in the proofs of the theorems @xcite remains now to be done . | the paper that is commented by touchette contains a computational study which opens the door to a desirable generalization of the standard large deviation theory ( applicable to a set of @xmath0 nearly independent random variables ) to systems belonging to a special , though ubiquitous , class of strong correlations .
it focuses on three inter - related aspects , namely ( i ) we exhibit strong numerical indications which suggest that the standard exponential probability law is asymptotically replaced by a power - law as its dominant term for large @xmath0 ; ( ii ) the subdominant term appears to be consistent with the @xmath1-exponential behavior typical of systems following @xmath1-statistics , thus reinforcing the thermodynamically extensive entropic nature of the exponent of the @xmath1-exponential , basically @xmath0 times the @xmath1-generalized rate function ; ( iii ) the class of strong correlations that we have focused on corresponds to attractors in the sense of the central limit theorem which are @xmath2-gaussian distributions ( in principle @xmath3 ) , which relevantly differ from ( symmetric ) lvy distributions , with the unique exception of cauchy - lorentz distributions ( which correspond to @xmath4 ) , where they coincide , as well known . in his comment , touchette has agreeably discussed point ( i ) , but , unfortunately , points ( ii ) and ( iii ) have , as we detail here , visibly escaped to his analysis . consequently
, his conclusion claiming the absence of special connection with @xmath1-exponentials is unjustified . before addressing in detail the comment by touchette @xcite on our paper @xcite ,
let us describe the physical scenario within which we have undertaken a possible generalization of the standard large deviation theory ( ldt ) .
a standard many - body hamiltonian system in thermal equilibrium with a ther- mostat at temperature @xmath5 is described by the boltzmann - gibbs ( bg ) weight , proportional to @xmath6n}$ ] , where @xmath7 is the @xmath0-particle hamiltonian , and @xmath8 . for standard hamiltonian systems
( typically involving short - range interactions and an ergodic behavior ) , the total energy is _
extensive_. consequently , the quantity @xmath9 $ ] scales with @xmath0 , analogously to a ( thermodynamically ) _ intensive _ variable .
this is to be compared with the ldt probability @xmath10 , where the rate function @xmath11 ( the meaning of the subindex @xmath12 will soon become clear ) is related to a bg entropic quantity _ per particle _ , and plays a role analogous to @xmath13 $ ] ( we remind that , for such standard systems , @xmath14 is an intensive variable ) .
if now we focus on say a @xmath15-dimensional classical system involving two- body interactions whose potential asymptotically decays at long distance @xmath16 like @xmath17 @xmath18 ,
the canonical bg partition function converges whenever the potential is integrable , i.e. for @xmath19 ( short - range interactions ) , and diverges whenever it is nonintegrable , i.e. for @xmath20 ( long - range interactions ) .
the use of the bg weight becomes unjustified [ illusory " in gibbs words @xcite for say newtonian gravitation , which in the present notation corresponds to @xmath21 , hence @xmath22 in the later case because of the divergence of the bg partition function .
we might therefore expect the emergence of some function @xmath23 different from the exponential one , in order to describe some specific stationary ( or quasi - stationary ) states differing from thermal equilibrium .
the hamiltonian @xmath7 generically scales like @xmath24 with @xmath25 ( with the _
@xmath1-logarithmic function _ defined as @xmath26 ) .
notice that ( @xmath27 ) @xmath28 for @xmath29 , @xmath30 for @xmath31 , and @xmath32 for @xmath19 . the particular case @xmath33 yields @xmath34 , thus recovering the usual prefactor of mean field theories .
the quantity @xmath35 can be rewritten as @xmath36n = [ { \tilde \beta}{\cal h}_n /(n{\tilde n})]n$ ] , where @xmath37 plays the role of an intensive variable . the correctness of all these scalings has been profusely verified in various kinds of thermal @xcite , diffusive @xcite and geometrical ( percolation ) @xcite systems ( see also @xcite ) .
we see that , not only for the usual case of short - range interactions but also for long - range ones , @xmath38 $ ] plays a role analogous to an intensive variable .
the _
@xmath1-exponential function _
@xmath39^{\frac{1}{1-q}}$ ] ( @xmath40 ) ( and its associated @xmath1-gaussian @xcite ) has already emerged , in a considerable amount of nonextensive and similar systems ( see @xcite among others ) , as the appropriate generalization of the exponential one ( and its associated gaussian ) .
therefore , it appears as rather natural to conjecture that , in some sense that remains to be precisely defined , the ldt expression @xmath41 becomes generalized into something close to @xmath42 ( @xmath43 ) , where the generalized function rate @xmath44 should be some generalized entropic quantity _ per particle_.
let us stress a crucial point : we are _ not _ proposing for long - range interactions , and other nonstandard systems , something like @xmath45 with @xmath46 , but we are expecting instead @xmath47 , i.e. , the _ extensivity _ of the total @xmath1-generalized entropic form to still hold @xcite , in order to be consistent with many other related results ( e.g. , @xcite ) .
we shall soon see that this important assumption indeed appears to be verified in the model , characterized by @xmath48 , that we numerically studied in @xcite .
let us start by exhibiting that its @xmath49 ldt asymptotic behavior numerically satisfies @xmath50 \;\;\;(b(x ) > 0 ; \ ,
c(x ) > 0)\ , , \label{asymptotic}\ ] ] with @xmath51 @xcite .
this implies the existence of a generically positive _ finite _ @xmath52 such that @xmath53n = c(x ) , \label{generalization}\ ] ] @xmath54 being a generically positive _ finite _ number for all values of @xmath55 different from 1/2 .
this is indeed verified , as exhibited in fig .
[ method ] and [ method2 ] .
more precisely , we verify for fixed @xmath48 that @xmath52 is _ unique _ for any given @xmath55 , whereas @xmath54 is in fact a set of values , noted @xmath56 , with @xmath57 ( the value of @xmath58 depends on @xmath55 ; for example , we can see that , for the illustration exhibited in fig .
[ method2 ] , @xmath59 for @xmath60 ) .
let us emphasize that the @xmath61 correction to the power law @xmath62 in ( [ asymptotic ] ) is consistent with the total entropy of the system always being _ extensive _ in the thermodynamical sense .
let us next check the conjecture made in @xcite , namely that @xmath63 is , for @xmath64 , well approached by @xmath65^{\frac{1}{q-1 } } } e_q^{-\frac{r_q(x)}{1+(q-1)r_q(x)}(n-1 ) } \nonumber \\ & = & e_q^{-\bigl\{\frac{r_q(x)}{[a(x)]^{q-1}}n+ \frac{1-[a(x)]^{q-1}}{(q-1)[a(x)]^{q-1}}\bigr\ } } \nonumber \\ & = & \frac{a(x)}{[1 + ( q-1 ) r_q(x ) n]^{\frac{1}{q-1 } } } \nonumber \\ & = & \frac{a(x)}{[(q-1)r_q(x ) n]^{\frac{1}{q-1 } } } \nonumber \\ & \times & \bigl[1-\frac{1}{(q-1)^2 r_q(x ) n}+ \frac{q}{2(q-1)^4 ( r_q(x ) n)^2 } - \frac{q(2q-1)}{6(q-1)^6 ( r_q(x ) n)^3 } + \dots \bigr ] \nonumber \\ & = & \frac{a(x)}{[(q-1)r_q(x ) n]^{\frac{1}{q-1 } } } \nonumber \\ & \times & \bigl[1-\frac{1}{(q-1)^2 r_q(x ) n}+ \sum_{m=2}^{\infty}(-1)^m \frac { q(2q-1)\dots [ ( m-1)q-(m-2)]}{m!(q-1)^{2m}(r_q(x)n)^m } \bigr ] \ , . \label{eq1}\end{aligned}\
] ] by identifying this expansion with eq .
( [ asymptotic ] ) we obtain @xmath66^{\frac{1}{q-1 } } \;\;\;(j=1,2 , \dots , j_{max } ) \ , , \label{identity1}\ ] ] and @xmath67 since @xmath52 and @xmath56 are numerically known , we can easily calculate @xmath68 and @xmath69 by using eqs .
( [ identity1 ] ) and ( [ identity2 ] ) . knowing these
, we calculate @xmath70 ( @xmath71 ) and compare with our numerical data .
we then bound our numerical results from both below and above ( see figs .
[ comparison ] and [ comparisonqlog ] for illustrations ) .
more precisely , for each value of @xmath55 , we have adopted two values , noted @xmath72 and @xmath73 , such that @xmath1-exponential upper and lower bounds for the entire set of numerical values for @xmath74 are obtained .
these @xmath75 and @xmath76 values turn out to be comparable to the corresponding set @xmath77 ( see fig . [ method2 ] ) . in other words
, we obtain the values of @xmath72 , @xmath73 , @xmath78 and @xmath79 , consistent with eq .
( [ identity2 ] ) . by introducing these values in eq .
( [ identity1 ] ) we obtain @xmath80 and @xmath81 , and verify that , in the model studied in @xcite , @xmath82 , @xmath83 , @xmath84^{\frac{1}{q-1 } } e_q^{-r_q^{(lower\,bound)}(x ) \,n } \nonumber \\ & = & \frac{b(x)}{n^{\frac{1}{q-1 } } } \bigl[1-\frac{1}{(q-1)^2 \ , r_q^{(lower\,bound)}(x)}\frac{1}{n } + o(1/n^2 ) \bigr ] \nonumber \\ & & \le p(n ; n / n < x ) \le \nonumber \\ & & b(x)[(q-1)r_q^{(upper\,bound)}(x)]^{\frac{1}{q-1 } } e_q^{-r_q^{(upper\,bound)}(x ) \,n } \nonumber \\ & = & \frac{b(x)}{n^{\frac{1}{q-1 } } } \bigl[1-\frac{1}{(q-1)^2 \ , r_q^{(upper\,bound)}(x)}\frac{1}{n } + o(1/n^2 ) \bigr ] \ , .
\label{sandwich}\end{aligned}\ ] ] we may summarize the above considerations by conjecturing that , for all strongly correlated systems which have @xmath2-gaussians ( @xmath85 ) as attractors in the sense of the central limit theorem ( see @xcite ) , a model - dependent set @xmath86 $ ] might exist such that @xmath74 generically satisfies inequalities ( [ sandwich ] ) . in our present example , this set depends on @xmath48 .
typical values of @xmath87 $ ] are illustrated in fig .
[ fig3 ] and compared with @xmath1-generalized entropic quantities . ) in _ one _ of the 10 lines " ( the bottom one , to be more precise ) observed in fig .
[ generalization ] for @xmath60 .
this procedure enables a high precision numerical determination of @xmath52 for any chosen value of @xmath55 .
for a given @xmath48 model , the value of @xmath52 is one and the same for all the lines " associated with a given value of @xmath55 . not so for @xmath54 : indeed , for fixed @xmath55 , we observe the existence of a set of values for @xmath54 which we note @xmath56 , with @xmath88 ( in the present illustration @xmath59 ) .
the finiteness of the set @xmath56 here and in fig .
[ method2 ] means that the corrections to the @xmath89 power in eq .
( [ asymptotic ] ) are of the @xmath61 order .
the finiteness that we observe ( here and in fig .
[ method2 ] ) in the slopes at the origin means that the next corrections are of the @xmath90 order . in this example
, we have run @xmath0 up to @xmath91 .
, width=529 ] ) , @xmath0 running up to @xmath91 . for fixed
@xmath55 , @xmath52 is unique , whereas @xmath54 corresponds to a set of values @xmath92 ( @xmath71 ) , where @xmath58 depends on @xmath55 .
we also see that the next correction is of the type @xmath90 . the upper and lower values @xmath93 and @xmath94 that are indicated by the arrows precisely correspond to the upper and lower bounds @xmath79 and @xmath78 such that all present numerical results are within two @xmath1-exponentials , as indicated in figs . [ comparison ] and [ comparisonqlog ] . , width=415 ] , where @xmath95 have been calculated from @xmath96 by using eqs .
( [ identity1 ] ) and ( [ identity2 ] ) .
the values for @xmath54 that have been used are those indicated by arrows in fig .
[ method2 ] .
two values for @xmath55 , namely @xmath97 and @xmath98 , have been illustrated here .
, width=453 ] in ( @xmath1-log)-linear representation .
let us stress that the _ unique _ asymptotically - power - law function which provides straight lines _ at all scales _ of a ( @xmath1-log)-linear representation is the @xmath1-exponential function .
the inset shows the results corresponding to @xmath0 up to 50 .
, width=491 ] touchette mentions kaniadakis @xmath99-logarithm and @xmath99-exponential @xcite as an alternative to the @xmath1-exponential and @xmath1-logarithm herein conjectured .
let us address this point through the definition @xmath100 ( notice a misprint in the definition of the @xmath99-logarithm appearing in touchette s comment ) .
it straightforwardly follows the asymptotic series @xmath101^{1/\kappa } } \nonumber \\ & = & \frac{1}{[2\kappa \,r_\kappa n]^{1/\kappa } } \bigl[1-\frac{1}{4 \kappa^3(r_{\kappa } n ) ^2 } \nonumber \\ & & \qquad + \sum_{m=2}^{\infty}(-1)^{m}\frac{[(m+1)\kappa+1]\cdot[(m+2)\kappa+1 ] \cdots[(2m-1)\kappa+1]}{2^{2m}m!\kappa^{3m}(r_{\kappa }
n ) ^{2 m } } \bigr]\nonumber\\ \label{eq4}\end{aligned}\ ] ] the dominant term is a power - law , and at this approximation it is trivially as admissible as virtually any other power - law .
however , we verify a highly meaningful discrepancy with the @xmath1-exponential function , namely that its subdominant correction is in @xmath90 , instead of @xmath61 .
this fact excludes the @xmath99-exponential function as an adequate one for the present purpose ; indeed , it can not satisfactorily reproduce the results exhibited in figs .
[ comparison ] and [ comparisonqlog ] .
-dependences of the lower and upper bounds for the rate function @xmath102 obtained from @xmath1-exponential fittings of the numerical data ( see figs . [ comparison ] and [ comparisonqlog ] ) . the analytical curves @xmath103 and @xmath104 are included for comparison .
the inset exhibits the quasi - parabolic behavior at both sides of @xmath105 .
, width=453 ] a point remains to be discussed .
the lvy - gnedenko theorem concerns sums of infinitely many _
independent _ ( or nearly independent , in a specific sense ) random variables , whereas the 2008 @xmath2-central limit theorem @xcite concerns sums of infinitely many _ strongly correlated _ variables within a specific class .
the first case corresponds to divergent standard variance , whereas the second one concerns finite @xmath106-variance ( @xmath107 ; see details in @xcite ) .
the attractors for the former case are lvy distributions , whereas those for the latter are @xmath2-gaussians .
both classes have long tails .
for the lvy distributions , the decay is slower than @xmath108 and faster than @xmath109 ; for the @xmath1-gaussians , any power - law faster than @xmath109 is admissible .
it is known that they have this and other relevant differences .
they always differ excepting for an unique case , which happens to be precisely the case focused on by touchette , i.e. @xmath110 , namely the cauchy - lorentz distribution ( named after cauchy by mathematicians , and after lorentz by physicists ) .
they can be simply thought as having @xmath111 , which , as acknowledged by touchette , is not particularly enlightening .
but they can be also thought in a much more interesting way , namely as having @xmath112 different from zero , which neatly illustrates the usefulness of the approach adopted in @xcite .
in fact , it is well known that @xmath4 is a highly peculiar case within the interval @xmath3 .
for example , the anomalous diffusion coefficient in the nonlinear fokker - planck equation known as the porous medium equation and discussed in @xcite changes its sign precisely at @xmath110 ( see also @xcite ) .
the fact that , for @xmath4 , @xmath113 whereas @xmath114 is totally analogous to a variety of dissipative one - dimensional maps whose lyapunov exponent vanishes at the edge of chaos . in such cases ,
the use of the nonadditive entropy @xmath115 instead of the bg one makes the discussion much richer since it enables a simple quantitative characterization of the nonlinear dynamical behavior ( by generalizing the standard exponential sensitivity to the initial conditions when the maximal lyapunov exponent is strictly positive to the @xmath1-exponential form at the edge of chaos , when the maximal lyapunov exponent vanishes ) .
this has been verified both analytically and numerically in very many cases @xcite .
let us conclude by saying that point ( i ) of the present abstract is agreeably discussed in touchette s comment , but a neat analysis of the important points ( ii ) and ( iii ) is notoriously absent in his paper . in other words , the @xmath1-exponential ansatz proposed in @xcite for ( asymptotically ) generalizing the standard ldt remains ( either exactly or approximatively : see the quantity ( [ generalization ] ) , expected to be finite , and the inequalities ( [ sandwich ] ) ) as a very strong candidate for a wide class of systems whose elements are strongly correlated .
this fact may be seen as a strong indication that , consistently with other results available in the literature ( see @xcite ) , the total entropy remains _ extensive _
( i.e. , thermodynamically admissible ) even in nonstandard cases where the bg entropy fails to be extensive .
any analytical results along these or similar lines would obviously be highly interesting and welcome . |
modern theoretical studies of the influence of dissipation on the propagation of sound on the basis of the navier - stokes equations may be said to have begun with the work of kirchhoff @xcite .
a principal aim of that and subsequent studies is to determine how the propagation speed and the rate of dissipation of the waves depend on their frequencies . for this problem
, the predictions from the standard navier - stokes equations of fluid dynamics do not agree well with experiments when the periods of the sound waves become as short as the mean flight times of the particles of the gas , that is , when we enter the ultrasound regime .
there are two directions from which to enter that regime .
we can begin with a gas of freely steaming particles and introduce weak interactions among them . in that case
, we may with uhlenbeck @xcite ask , `` how is it possible to impose on the random motion of the molecules the ordered motion ... which a sound wave represents ? '' in the modern language of dynamical systems theory , this could be seen as a problem of synchronization in which we witness increasing numbers of particles going into cooperative motion until all are engulfed . on the other hand , we may start from the case of continuum mechanics and attempt to extend the validity of that description to the case of longer and longer mean free paths .
it is unlikely that in either case we can successfully traverse the full range of possible conditions , but we may expect to encounter an interesting transition between the two regimes . in this paper
, we examine how well the fluid dynamical description of paper i of this series @xcite extends into the domain where the particle mean free paths are comparable to the characteristic macroscopic length scale of the medium . in paper
i , we derived an extension of the fluid dynamical equations that we hope may offer an improvement of this kind and , in the present paper , we study their linear form and the resultant dispersion relation for sound waves . in this first section ,
we restate the equations given in i before going on to the straightforward determination of the dispersion relation they imply for the linear theory of sound waves .
the basic form of the macroscopic equations derived from kinetic theory , are @xcite @xmath0 where @xmath1 is mass density , @xmath2 is the velocity field , @xmath3 is the temperature , @xmath4 is the pressure tensor , @xmath5 is the heat flux vector and the colon stands for a double dot product .
we have not included an external force .
these equations express newton s laws of motion for a continuum in phenomenological theories and they are a formal consequence of most kinetic theories . where approaches to the derivation of these equations from kinetic theory may differ is in the expressions for the higher moments , @xmath4 and @xmath5 .
the derivations from kinetic theory are important since they provide formulas for the transport coefficients that appear in the specific expressions for the pressure tensor and the heat flux . however , not all treatments of the kinetic theory give the same explicit formulas for @xmath4 and @xmath5 , there being differences of degree and style of the approximations used . of course , when the mean free path of the constituent particles is sufficiently short
compared to all macroscopic lengths in the problem , there is no real disagreement , since the standard navier - stokes forms work well enough for most purposes .
but when the macroscopic lengths become short and are comparable to the mean free paths of the particles , those standard results do not agree with experiment , as we shall see .
therefore we must ask whether there is a continuum approximation that may provide improved treatments of such problems . to test whether the expressions for @xmath4 and @xmath5 derived in paper i from the relaxation model of kinetic theory @xcite may fulfill this need , we here apply them to study of the propagation of ultrasound . in the relaxation model , the relaxation time , @xmath6 , may be taken to be of order of the mean flight time of particles , where the mean speed is of the order of the speed of sound . then , we have @xmath7 where @xmath8 is a constant that depends on the collision cross section and the gas constant and we have ignored a possible dependence of the particle cross - section time on particle speed .
the results in i are based on an expansion in @xmath6 , up to first order .
those expansions led to a pressure tensor , @xmath9\ , \mathbb{i } - \mu \ , \mathbb{e } \label{x39}\ ] ] where @xmath10 @xmath11 is the gas constant , @xmath12 @xmath13 is the viscosity and @xmath14 .
the result for @xmath15 , together with ( [ tau ] ) , implies maxwell s conclusion that viscosity does not depend on density for simple gases . for the heat flux
, we obtained @xmath16 where @xmath17 is the conductivity .
both ( [ x39 ] ) and ( [ x40 ] ) carry errors of order @xmath18 that are not indicated explicitly .
these formulae for @xmath4 and @xmath5 are not expressed explicitly in terms of the fluid fields .
rather , their expressions involve some of the same time derivatives of these fields that appear in the fluid equations . here is the central difference between our results and those obtained in the chapman - enskog approach . in the latter
, partial derivatives with respect to time are eliminated by the use of lower order results .
though we do not use that elimination procedure in deriving the closure relations , we can nevertheless readily recover the navier - stokes results from ours , when @xmath19 , as we described in i. however , even though both theories formally have first - order accuracy in @xmath6 , the results from them are significantly different at knudsen numbers of order unity , as we shall see in what follows .
we consider the evolution of perturbations on a uniform medium and define perturbation variables @xmath20 and @xmath21 through the relations @xmath22 where @xmath23 , @xmath24 and @xmath25 are the constant background values of the thermodynamic fields .
the perturbation quantities all have small amplitudes with , for example , @xmath26 . from the linearization of ( [ press ] ) we obtain
we further assume that there is no background motion so that @xmath2 is small and needs no subscripts . in analogy with ( [ 1.1 ] )
we write for the pressure tensor , the linearized form @xmath28 where @xmath29 where @xmath6 is evaluated for the state variables of the background medium and @xmath30 is small . similarly ,
since there is no zeroth order heat flux , we get for the linearized heat flux , @xmath31 the quantities @xmath32 and @xmath33 are the viscosity and conductivity evaluated in terms of the state variables for the constant background medium .
the linearization of ( [ cont ] ) is @xmath34 a compact form of the linearized ( [ mom ] ) is @xmath35 when we take the divergence of ( [ lmom ] ) and use ( [ lcont ] ) , find that @xmath36 from ( [ ip1 ] ) we find that @xmath37 we see from ( [ e ] ) that @xmath38 then , with the help of ( [ lcont ] ) , we find @xmath39 on using ( [ lcont ] ) again , we find that @xmath40 which we may introduce into ( [ 2mom ] ) .
next we define the laplacian speed of sound , @xmath41 , and the kinematic viscosity , @xmath42 , as in @xmath43 we then obtain the dissipative wave equation @xmath44 to complete this discussion , it is useful to introduce the thermal diffusivity @xmath45 where @xmath46 . thus , @xmath47 .
this is used in the linearized heat equation , @xmath48 where we may write @xmath49 since @xmath50 , we then find @xmath51 where @xmath52 is the prandtl number of the undisturbed medium .
finally , to further simplify the appearance of these formulae , we let the unit of time be @xmath6 and the unit of length be @xmath53 .
then our linearized equations for sound waves are @xmath54 @xmath55 for comparison we note that the analogous linear equations for the navier - stokes case ( with zero bulk viscosity ) are these : @xmath56 @xmath57
we may seek solutions to the linear equations ( [ 1])-([2 ] ) in which @xmath58 and @xmath59 vary like @xmath60 . since the mean free path is the unit of length , the wave number @xmath61 , which is nondimensional , is effectively the knudsen number for this problem .
the dispersion relation is @xmath62 for comparison , we report that the dispersion relation for the navier - stokes equations is @xmath63 to get a feeling for what these results mean , we look at free modes for which @xmath61 is real
. then we set @xmath64 where @xmath8 and @xmath65 are also real .
when we introduce this into ( [ disp ] ) we find that there is a ( thermal ) mode with @xmath66 and a pair of ( acoustic ) modes whose frequencies satisfy @xmath67 which gives the frequencies of sound waves .
as we may confirm , @xmath8 is of order unity for large @xmath61 and it grows in proportion to @xmath68 for small @xmath61 .
hence , for both very large and very small @xmath61 , the last term on the right of ( [ freq ] ) is the largest one on that side .
so we may write the uniform approximation @xmath69 for small @xmath61 , this gives the phase speed @xmath70 , which is the usual speed of sound for an adiabatic sound wave , as is to be expected for very long wave lengths .
for large @xmath61 , we obtain the phase speed @xmath71 . for the n - s equations with zero bulk viscosity
there is the same number of modes : a thermal mode with zero frequency and sound waves with @xmath72 as expected , the two sets of equations agree in the limit of very small @xmath61 , where the n - s equations return the phase speed @xmath73 .
but for large @xmath61 , the differences between the two theories become qualitative . with the navier - stokes equations
, we find that at large @xmath61 , instead of reaching a finite limit , the phase speed is proportional to @xmath61 for large @xmath61 .
as we shall see when we look at the experimental results , the n - s prediction is qualitatively wrong ; the phase speed goes to a finite value at large @xmath61 .
the equation for the damping rate is @xmath74\alpha + ( { 10\over \sigma}-1)k^2 \omega^2 - { 5\over \sigma } k^4 = 0 . \label{damp}\ ] ] for the thermal mode , for which @xmath66 , we find the damping rates @xmath75 for small @xmath61 and @xmath76 for large @xmath61 .
thus , there is very little damping for long waves while short waves are damped on the collisional time scale .
moreover , on examination of these two limits , we see that they each emerge from the balance of the same two terms in ( [ damp ] ) .
hence we may write the approximate formula @xmath77 as a reasonable approximation to the damping rate for all @xmath61 , in the thermal mode .
similarly , in the case of sound waves , we see that @xmath8 is also the result of the balance between the same two terms in ( [ damp ] ) in the limits of for large and small @xmath61 .
hence , we find that for sound waves , to good approximation , the damping rate is given by @xmath78 where @xmath65 is given in ( [ freaq ] ) .
for long wave lengths , the damping is again slight since it goes to zero like @xmath79k^2 $ ] . for large @xmath61
, we obtain the finite limit @xmath80 . for the navier - stokes equations ,
the damping rate is given by @xmath81\alpha + ( { 5\over \sigma}+4)k^2\omega^2 - { 5\over \sigma}k^4 = 0 \ .
\label{n - s}\ ] ] the damping rate for the acoustic modes also goes like @xmath68 for small @xmath61 for both the thermal mode and the acoustic modes . for the acoustic modes ,
the damping rate is @xmath82k^2 $ ] , so that we get the same wave number dependence , but with a different coefficient than is obtained from our equations in the small @xmath61 limit .
however , for increasing @xmath61 , the n - s damping rates _
grow _ like @xmath68 for sound waves , which is in disagreement with experiment @xcite .
though the study of free modes in the previous section is intuitively clear , it does not directly represent the way experiments on sound propagation are usually carried out . in the experiments , it is more typical that one drives the fluid at a real , fixed frequency and then studies the propagation of waves in space .
the forcing may be accomplished by vibrating the end wall of a tube containing gas at a fixed ( real ) frequency @xmath65 and observing the propagation down the tube . to model this procedure in full detail
would involve a careful treatment of the forcing procedure , which usually requires attention to boundary conditions .
however , in this first reconnaissance of the way our equations describe sound waves , we shall adopt a standard theoretical practice @xcite and simply fix the wave frequency , @xmath65 , in the dispersion relation and compute the resulting @xmath61 , which will typically be complex .
thus , in ( [ disp ] ) we let @xmath83 and we find that the equation for @xmath61 becomes @xmath84k^2 + 3i\omega^3 = 0 .
\label{spatial}\ ] ] we may similarly obtain such an equation for the n - s case , ( [ dispa ] ) . in order to emphasize the results for large knudsen number we plot the results in the manner used , for example , by cercignani @xcite .
that is , we introduce the quantity @xmath85 where @xmath86 is a normalized inverse propagation speed .
the factor @xmath87 is included so that the phase speed is nondimensionalized on the laplacian ( or adiabatic ) speed of sound , rather than the newtonian ( or isothermal ) speed of sound as above
. then we find that @xmath88k^2 + { 5 \sigma i\over 3\omega } = 0 .
\label{k4}\ ] ] to see how this representation contains the results for free modes , we note that , in the limit @xmath89 , ( [ k4 ] ) reduces to @xmath90 .
that is , for low frequencies , the usual adiabatic sound speed is recovered . in the ( more interesting ) opposite limit , @xmath91 , we obtain @xmath92 for the propagative modes .
thus we see that , in the limit of forcing at high frequency , sound waves propagate with phase speeds @xmath93 , independently of frequency .
the data shown in the accompanying figure ( fig . [ thefig ] ) confirm this independence of frequency ( or wave number ) of the speed of propagation of ultrasound . the observed nondimensional phase speed is @xmath94 .
the prandtl number found from the relaxation model of kinetic theory , either by the methods of chapman and enskog or those described in paper i , is unity . with this value ,
we obtain @xmath95 for the limiting phase speed , so this represents a small quantitative error . however , the value of @xmath96 found in kinetic theory depends on the atomic model used , that is , on the nature of the collision term .
though the relaxation model gives the explicit value unity for @xmath96 , the value found with the traditional boltzmann equation for hard spheres is @xmath97 .
this difference has nothing to do with the approximation method ( our procedure gives @xmath98 when applied to the boltzmann equation ) but is a consequence of the nature of the form of the atomic interactions that is adopted .
we therefore follow a common practice put the empirical prandtl number into the theoretical results when comparing with experiments .
since the experimental data we shall refer to are for noble gases whose values of @xmath96 are @xmath99 or @xmath100 we shall here adopt the value @xmath98 suggested by the boltzmann equation .
when we use that value of the prandtl number in evaluating the phase speed , we obtain @xmath101 . even without this adjustment ,
the results for the propagation of ultrasound are good , but we would propose to anyone thinking of using our equations from this first - order development from the relaxation equation to introduce this phenomenological improvement of the theory . in the accompanying figure ( fig .
[ thefig ] ) , we show the variation of @xmath102 as a function of @xmath103 from a number of sources .
the experimental values ( @xcite ) are indicated as individual points ( the diamonds ) and they appear to be tending toward a nonzero constant value at high frequency .
this is qualitatively in accord with our results , here shown as a solid line for the case of @xmath98 , and it is in stark disagreement with the prediction from the navier - stokes equations ( long dashes with double dots ) , which predict that @xmath104 goes to zero like @xmath103 . since our limiting value for @xmath104
was found to be @xmath105 , the remarkable agreement of our results with experiment owes something to our using the experimental value of @xmath97 for @xmath96 for limiting value @xmath106 .
nevertheless , even without this choice , the results would be adequate and comparable to those shown for the moment method ( @xcite ) with @xmath107 moments ( short dashes ) .
other theoretical studies of ultrasound are based on direct solution of the boltzmann equation @xcite and we show the results of sirovich and thurber @xcite obtained in this way ( medium dashes ) , for which the prandtl number automatically has the value @xmath97 .
in the study of the thermal damping of sound waves by electromagnetic radiation @xcite , one finds that , for thermal times much less than the acoustic period , sound propagates at the isothermal speed of sound with negligible dissipation . in the opposite limit of long thermal times , there is also little dissipation , but propagation is at the adiabatic speed of sound .
the experiments , and the solution of the boltzmann equation show similar behavior when the relevant parameter is the ratio of the collisional relaxation time to the acoustic period .
our equations , as well as those of the moment method ( with tens of thousands of moments ) , reproduce this behavior but the navier - stokes equations do not .
moreover , when the prandtl number is chosen to be that of the experimental gas , the quantitative agreement becomes very good . in the next installation of this series , we shall compute the profile of a stationary shock wave . as we shall see , the agreement with the experiments is good in that case too . | the equations of fluid dynamics developed in paper i are applied to the study of the propagation of ultrasound waves .
there is good agreement between the predicted propagation speed and experimental results for a wide range of knudsen numbers . |
two decades of _ hubble space telescope ( hst ) _ optical / near - infrared ( nir ) observations and more recently also ionized or molecular gas data have unveiled a complexity of morphologies in high redshift star - forming galaxies .
many @xmath3@xmath141 galaxies do not display the disk or spheroidal morphology that is observed in the majority of local galaxies but are instead characterized by giant , star - forming clumps which dominate the light profiles and result in largely asymmetric appearances @xcite .
although such irregular structure has often been associated with mergers events , the use of nir , integral - field ( ifu ) spectroscopy has enabled substantial progress in the classification of high redshift galaxies and detailed kinematic analysis have revealed ordered rotational motion in a large number of these visually disturbed galaxies @xcite . the formation of giant clumps in normal " disk galaxies is thought to be the outcome of violent disk instability and fragmentation @xcite , developing as a consequence of the high gas fractions that are typical for distant galaxies @xcite . in the lack of resolved kinematic data ,
the distinction between merging galaxies and clumpy disks remains however an observational challenge . in this paper
we show how a quantitative classification performed on resolved stellar mass maps , rather than optical or nir single - band images , can help disentangling the population of truly merging galaxies from that of clumpy disks even without available ifu observations .
our approach is motivated by a number of reasons .
even with the currently available second - generation instruments , obtaining reliable ifu kinematic measurements still requires major telescope time investments ( see e.g. , the kmos-3d campaign , * ? ? ?
therefore , techniques which can provide robust proxies for the full kinematic informations are necessary .
canonical merger classification methods , however , suffer from limitations which can hamper the distinction between mergers and normal - but - clumpy " galaxies .
the selection of close pairs in the spatial and velocity domain ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* among the others ) , for example , identifies by definition physically associated systems but it can be biased against very close galaxies ( fiber collision " ) and thus late interaction stages , namely the merger phases in which the distinction between mergers and clumpy disks becomes more hazy .
close kinematic pairs would typically have a large enough separation to allow the individual morphological classification of each galaxy .
another common way of identifying mergers relies on the degree of irregularity in the light distribution either through non parametric measures such as the concentration , asymmetry , clumpiness ( cas ) and gini-@xmath15 indices @xcite or through other indicators of the presence of multiple components / tidal interactions @xcite . using hydrodynamical numerical simulations ,
@xcite have shown that combinations of non - parametric structural estimators ( g - a-@xmath15 ) are sensitive to the coalescence phase and thus can be used also for evolved mergers . as mentioned above , however , at @xmath16
the light profiles are dominated by giant star - forming clumps even in regular disk galaxies and cas - like classification schemes applied to single - band optical / nir images typically fail in distinguishing mergers from non - interacting galaxies .
while the giant @xmath17 clumps contribute to 20%-50% of the flux in resolved optical / uv or star formation rate ( sfr ) maps of @xmath14@xmath18 star - forming galaxies , they show a lower contrast with respect to the underlying disks on stellar mass maps , contributing to @xmath19 of the total mass budget @xcite .
this suggests a potentially lower contamination of falsely identified mergers if measuring the aforementioned non - parametric structural indicators directly on the stellar mass maps instead of the single band images : for clumpy galaxies we expect the stellar mass maps to display a regular , centrally concentrated profile , whereas for merging galaxies multiple components will be present , with no clear central mass concentration .
the exploitation of resolved mass ( and also sfr or age ) maps has become a common method of investigating the physical properties of low and high - redshift galaxies ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) but have not been used so far for a quantitative merger classification .
the goal of this paper is to assess the performance of such a classification .
we mainly focus on the testing and calibrating the proposed mass - based selection of mergers and defer a more in depth discussion on the properties of mass - identified mergers to a forthcoming paper ( a. cibinel et al . in preparation ) .
we use a set of mergers and isolated galaxies from the mirage simulations @xcite to quantitatively determine the efficiency and time scales probed when selecting mergers with mass - based structural parameters .
although other studies have investigated the morphology of interacting galaxies in numerical simulations , the use of the mirage sample enables us to make steps forward with respect to these previous analysis .
the mirage simulations suite includes in fact key physical processes that are paramount for the formation and regulation of the giant star - forming clumps and thus naturally reproduces the complexity of clump - dominated morphologies of high redshift galaxies .
we then apply our classification scheme on a fiducial sample of 1.5@xmath203 galaxies in the hubble ultra deep field ( hudf , * ? ? ?
* ) and compare the results of our new method with the @xmath13band classification and also with previously published kinematic analyses . specifically , the paper is organized as follows .
we present in section [ sec : data ] the observational data utilized in our analysis , the sample basic properties and the generation of the resolved mass maps for the hudf galaxies .
section [ sec : sims ] presents the mirage simulations and post - processing of the simulation output .
we discuss our definition of a merging galaxy and some caveats regarding the simulations in section [ sec : mergercons ] .
we provide a summary of the structural measurements performed on both real and simulated galaxies in section [ sec : strcmeasurements ] . in section [ sec : simulationinsight ] we calibrate the classification performed on the mass maps using the mirage simulations and ancillary data with kinematic information .
we then compare in section [ sec : masslightrealgal ] the mass - based classification and the standard @xmath13band classification for the real hudf galaxies .
finally , section [ sec : conclusions ] summarizes our findings and conclusions .
considerations about signal - to - noise ( s / n ) effects and possible systematic biases are presented in section [ sec : biases ] and appendix [ appendix : massmapsreliability ] .
all magnitudes are in the ab system and corrected for galactic absorption using the dust maps of @xcite when necessary . throughout the paper we use interchangeably the notation @xmath21-band " to refer to the @xmath22/wide field camera 3 ( wfc3 ) f160w filter .
if needed , we also use the abbreviations @xmath23 , @xmath3 and @xmath24 when referring to the @xmath22/advanced camera for surveys ( acs ) f435w , @xmath22/acs f850lp and @xmath22/wfc3 f105w filters , respectively .
quoted masses assume a chabrier initial mass function ( imf ) .
we finally note that we will sometimes use the notation clumpy disks " to refer those galaxies that have a clumpy appearance in the @xmath13band / optical images but are not classified as mergers with our method .
we stress however that this is not meant to be a quantification of the intrinsic strength of the bulge component in these galaxies .
location of the galaxy sample on the mass versus sfr plane .
the dashed line shows the locus of the @xmath25 main sequence of star - forming galaxies ( based on literature compilation in * ? ? ?
* ) and the small gray points are all galaxies in the @xcite goods - s photometric catalog with a photometric redshift 1.5@xmath6@xmath3@xmath63 .
the large green symbols correspond to the initial sample of 132 galaxies studied in the present work , prior to applying the size and magnitude selection of section [ sec : biases ] . the final sample of 87 galaxies on which reliable mass maps could be derived after applying this selection is shown with the red symbols .
triangles indicate ir - based sfr ( either from a combination of nuv+monochromatic 24@xmath26 m flux density or from a fit to the mid - to - far ir sed ) .
circles are instead galaxies without reliable ir photometry and for which the sfr is thus estimated from the dust - corrected uv luminosity .
galaxies with very low sfr are identified as quiescent based on their @xmath27 or @xmath28 colors and the upper limits on their ir fluxes .
for these galaxies no dust extinction is applied when computing the uv - based sfr and are hence imposed to lie below the main sequence . the sample considered here
is mostly composed of normal ( main sequence ) , star - forming galaxies.,scaledwidth=45.0% ]
we apply the new classification scheme on a sample of galaxies in the hudf field which , thanks to the availability of extremely deep observations but also medium and shallow coverage over the same area , enables us to generate high accuracy mass maps as well as to assess the impact of s / n on the classification .
we briefly summarize here the sample selection criteria and the relevant information for the public data sets utilized in this study , referring the reader to the original works for further details . to build the pixel - by - pixel mass maps and perform the analysis described in section [ sec : masslightrealgal ] , we exploited the data from several public campaigns covering the hudf area with a multi - tiered approach .
for galaxies in the original hudf acs field ( 3@xmath293@xmath30 ) we generated two versions of the mass maps using either ( 1 ) the deep @xmath22/acs f435w , f606w , f775w and f850lp images from hudf @xcite combined with the @xmath22/wfc3 f105w , f125w and f160w images from the candels - deep survey @xcite , or ( 2 ) the f435w , f606w , f775w and f850lp observations from the goods survey @xcite plus the candels - deep nir photometry . for the sub - set of galaxies in the central 2@xmath292@xmath30 region of hudf ( corresponding to the wfc3 field of view ) , we also generated a third mass map making use of the extremely deep observations available at all wavelengths : the optical hudf images and the @xmath22/wfc3 f105w , f125w , f140w and f160w coverage from the hudf09 and hudf12 surveys @xcite .
we specifically employed the mosaics provided by the hudf12 team which are combined with the hudf09 datasets . in all cases ,
candels observations are also used for the f814w filter .
the identification of the galaxy sample and the calculation of photometric redshifts ( photo-@xmath3 ) are based on the @xmath21-band selected , multi - wavelength catalog published by the candels team in the goods - s field ( * ? ? ? * and references therein ) .
the photometric data available in this compilation consists of imaging in 17 medium and broad - band filters ranging from the @xmath31-band to the irac @xmath32 channel .
we derived photo-@xmath3 and integrated galaxy stellar masses for all galaxies in the @xcite sample as described in full details in @xcite .
briefly , photo-@xmath3 were estimated from the @xcite photometric catalog using the public code eazy @xcite and a combination of the standard set of templates from @xcite .
these photo-@xmath3 reach an accuracy of @xmath33 .
published spectroscopic redshifts ( spec-@xmath3 ) are available for about 30% of our final hudf sample . from several spectroscopic surveys which have covered ( also ) the hudf field . ] for galaxies with a _ secure _ spectroscopic measurement we considered the spec-@xmath3 as the final redshift , whereas the photo-@xmath3 was preferred in those cases in which only a tentative ( or no spec-@xmath3 ) is given .
we then obtained galaxies stellar masses through fitting of the spectral energy distribution ( sed ) with fast @xcite , keeping the redshift fixed and using a set of @xcite stellar population models with delayed exponentially declining star formation histories .
dust absorption with a maximum of @xmath34 was allowed in the fitting @xcite .
the initial galaxy sample was extracted from the @xcite catalog by selecting galaxies over the hudf area having a spectroscopic or photometric redshift between 1.5@xmath6@xmath3@xmath63 and an @xmath21-band magnitude brighter than @xmath21@xmath626 mag .
the redshift selection allows us to probe the rest - frame fuv to optical for all galaxies ; the luminosity cut is instead applied to ensure that a minimum signal - to - noise is reached in most pass - bands and thus a reliable photo-@xmath3 estimate can be obtained .
the exact value of @xmath21@xmath626 mag was chosen empirically by requiring that 80% of all goods - s galaxies , within the same redshift bin as the one here considered , have a s / n@xmath83 in at least 10 of the pass - bands which are used in the photo-@xmath3 calculation .
this magnitude threshold also ensures that the sample lies comfortably above the 50% completeness limit of the parent @xcite photometric catalog ( @xmath21=26.6 ) .
the @xmath13band magnitude limit translates into a redshift dependent mass completeness threshold . for a clear selection of the sample
, we hence apply a further cut in stellar mass to include only galaxies above the completeness value .
given that we are mostly interested in studying the properties of clump - dominated , star - forming galaxies , we consider in the following the mass completeness limit for star - forming galaxies .
to derive this threshold we followed the procedure described in e.g. , @xcite .
for each galaxy we estimated the mass @xmath35 that it would have , keeping its mass - to - light ratio constant , if faded to the limiting magnitude @xmath3626 .
we then calculated , at each redshift , the mass below which lie 90% of @xmath35 in the 30% faintest galaxies considered to be representative of the typical @xmath37 of a galaxy close to the magnitude limit .
the final completeness limit is set by the highest redshift here considered ( @xmath3=3 ) , corresponding to a value of @xmath38 for star - forming galaxies .
the equivalent number for quiescent galaxies would be @xmath39 .
after also rejecting galaxies which fall too close to the hudf edges for reliable measurements , our initial sample includes 132 galaxies with 1.5@xmath6@xmath3@xmath63 , h@xmath626 and m@xmath40 .
we derived structural parameters for all these galaxies and we provide them in table [ tab : sampleprop ] , but the sample is further restricted for our final analysis as a result of the reliability assessment of the stellar mass maps that we present in section [ sec : biases ] and which is based on tests performed on this initial , larger sample . while deferring to a forthcoming paper the detailed analysis of the star formation properties of mass - selected mergers , we show in figure [ fig : f1 ] the position of this initial sample of 132 galaxies on the mass versus sfr plane for illustration purposes . even above the mass completeness limit for quenched galaxies ,
the majority of the galaxies here considered lie on the locus of the so - called main sequence of star formation . the few galaxies with very low sfr ( sfr@xmath411@xmath42yr@xmath43 )
have been identified as quiescent from a combination of their @xmath27 colors or @xmath44 colors ( see * ? ? ?
* ; * ? ? ?
* ) and the constraints coming from their ir flux upper limits .
these galaxies are by definition forced to lie below the main sequence by imposing no dust extinction in the calculation of the uv - based sfr .
our sample is thus representative of the typical population of @xmath45 star - forming galaxies .
a summary of the properties for the full sample of 132 galaxies and the classification into mergers and non - interacting galaxies from the stellar mass map analysis is given in table [ tab : sampleprop ] .
as a first step for the generation of the resolved mass maps , we registered all the acs and wfc3 tiles to the same resolution and pixel scale of the @xmath21-band which has the worst point - spread function ( psf ) among the other available filters ( @xmath50.@xmath46 ) .
to do so , we created an individual psf for each filter by stacking several unsaturated stars in the hudf field and computed the convolution kernels to match the psf of the @xmath21-band psf varies slightly across the field of view and this effect could be taken into account by selecting for each object a nearby star instead of using a common psf for the entire field
. however , we estimate that the error introduced by using a single psf is comparable with the uncertainty associated to noise effects when using individual stars . ] . from the matched images
, we then extracted postage stamps for each galaxy in our sample with a size equal to 3 times the @xmath21-band kron radius and , as further justified in section [ sec : mergerdefinition ] , we cleaned from the stamps any nearby companion galaxy with a known spec- or photo-@xmath3 . to derive pixel - based stellar masses and the actual mass maps , we extracted pixel - by - pixel seds within an elliptical aperture equal to the galaxy @xmath13band kron semi - major axis and fitted them with stellar population models . given the relatively low flux in each individual pixel , some degree of smoothing or binning is required to ensure a minimum s / n on most of the filters and thus obtain reliable parameters from the pixel sed fitting .
several approaches have been used in the literature to deal with s / n homogenization problems @xcite .
we opted here for the publicly available code adaptsmoothzibetti / software / adaptsmooth.html ] developed in @xcite and @xcite .
whenever the s / n falls below a given threshold , this algorithm performs an adaptive smoothing of the images by replacing the original pixel values with an average of the galaxy flux over larger and larger circular areas as the s / n decreases .
adaptsmooth features two useful options : ( a ) the smoothing of several images on the same scale lengths necessary to derive self - consistent seds
can be easily performed and ( b ) the pixel identity is maintained , as opposed to binning schemes in which neighboring pixels are assigned a common value , effectively grouping them together into a final macro pixel " .
results of the tests performed on the toy - model mass profiles to assess the ability of reconstructing the stellar mass distribution . for all models falling in any region of the the size ( kron radius normalized to psf ) versus @xmath13band magnitude plane marked by the horizontal and vertical dashed lines , we calculated the median absolute residual ( @xmath47 ) and the dispersion ( @xmath48 ) between input and reconstructed mass maps .
the region boundaries are selected such to have at least 5 models in each bin .
areas where @xmath47 ( solid , 45 degrees cw shading ) or @xmath48 ( dashed , 45 degrees ccw shading ) are low are colored in green , whereas a red color indicates systematic shifts and large scatter around the input models , i.e. , a high uncertainty in the derivation of the mass maps .
the colorbars at the top of the figure provide the mapping between colors and the absolute values of @xmath47 and @xmath48 , in dex .
the figure refers to models created to reproduce the typical depth of the optical hudf and candels - deep nir data , which is in between the three combinations of photometry here explored ( see section [ sec : archivaldata]).,scaledwidth=50.0% ] the choice of the reference band(s ) defining the smoothing kernels applied by adaptsmooth is a trade - off between attaining the best s / n and maintaining spatial resolution : using the band with the lowest s / n will obviously degrades the resolution , while referring to the one with the highest s / n will likely result in noisy seds .
furthermore , also in the light of forthcoming analyses on the comparison between the resolved mass and sfr distribution , we are interested in detecting features such as giant star - forming clumps which may be intrinsically bright in one band but have a smaller flux contrast at longer wavelengths .
a too broad smoothing on the red band may completely erase these structures in the blue filters .
after testing single or multiple bands smoothing , we found the optimal configuration by running adaptsmooth on stacked images of all acs and wfc3 stamps : this ensures that the smoothing is applied on those pixels where the majority of the bands reach a low s / n while preserving the structural variations in the different filters .
specifically , we run adaptsmooth by requiring a minimum s / n=5 on the stacked images and halting the adaptive smoothing when the averaging area reaches the maximum radius of 5 pixels .
the smoothing pattern thus obtained was then applied to all available bands giving a median s / n@xmath55 also on the individual pixels for most filters .
we note that we have tested that our results are not substantially affected by a different choice of the smoothing kernel ( e.g. , by applying the smoothing on the @xmath13band only ) .
we fitted the adaptively smoothed pixel seds with lephare @xcite using the @xcite synthetic spectral library with a @xcite imf and a delayed exponential star formation history , @xmath49 .
the characteristic time scale @xmath50 was let vary between 0.01 and 10gyr in 22 steps and template ages were chosen between 100myr and the age of the universe at the given redshift .
we allowed three metallicity values in the fitting ( @xmath51=0.2@xmath52 , @xmath51=0.4@xmath52 and @xmath51=@xmath52 ) and furthermore applied internal dust extinction by assuming a calzetti law and @xmath53 ranging between 0 and 0.9mag .
we defined our fiducial pixel mass estimate as the median mass from the full probability distribution function from all templates , but our results would remain unchanged if we had used the mass from the best - fit template ( i.e. , minimum @xmath54 ) instead . as a validation of the derived mass maps , we verified the consistency between the sum of the pixel - based masses and the integrated galaxy mass in appendix [ sec : resolved_vs_integrated ] .
we find an agreement at the level of @xmath120.1 dex between the two estimates .
an example of the derived mass maps is given in figure [ fig : f2 ] , where we compare the @xmath27 composite image , the @xmath13band image and the mass map for a galaxy in our sample ( i d 12451 ) .
we have intentionally chosen a galaxy which displays a different structure in the @xmath13band than in the mass map to clearly illustrate how at z@xmath81 @xmath13band light and mass are not equivalent tracers of morphology .
this was already pointed out by @xcite ( see for example their figure 2 which has two galaxies overlapping with our sample in figure [ fig : f11 ] ) and we will further demonstrate it in section [ sec : masslightrealgal ] .
numerous studied have shown that the ability of recovering intrinsic galaxy properties from observed flux distributions depends strongly on the image s / n , resolution and even on the intrinsic properties of the galaxies themselves @xcite .
our mass - based measurements will also be affected by similar limitations . for structural measurements performed on optical images , tests on artificial galaxies
have demonstrated that it is possible to derive correction schemes that can largely account for the systematic biases in the estimates of galaxy structure e.g. , galaxy radii and concentrations for both local ( see @xcite for an application to @xmath55 galaxies from the zens sample in @xcite ) and high redshift galaxies ( see @xcite for an application to the cosmos survey , @xcite ) .
we do not attempt here a derivation of similar corrections in the mass domain as this would require large suites of artificial mass maps and hence significant assumptions on the mass ( and dust ) distribution in merging and non - merging galaxies .
we can nonetheless perform some tests which enable us to define the boundaries of applicability " of our method , i.e. the regimes where we can obtain reliable measurements .
it is clear that for unresolved objects no meaningful mass reconstruction can be performed nor the individual galaxies participating in the merger can be identified .
likewise , the reconstruction of the mass distribution becomes more and more difficult as the flux in the pixels reaches the surface brightness limit of the observations . to derive a global magnitude and size limit below which we can not reliably derive mass maps , we tested our sed - fitting and mass reconstruction technique on a set of artificial galaxies with known mass distribution drawn from our initial sample of 1.5@xmath6@xmath3@xmath63 .
we describe these models in detail in appendix [ appendix : resolutionlimit ] . for each of the toy galaxies , we compared the mass profile reconstructed following the procedure in section [ sec : massmapsreal ] with the input model and calculated the median of the residuals as well as the typical dispersion around the input model .
the results of the comparison are presented in figure [ fig : f3 ] where we color code each region of the size versus magnitude plane according to the mean absolute residual value and the dispersion of all models falling in that specific area : a red / orange color indicates high residuals / scatter in the reconstructed maps , green corresponds to low residuals / scatter .
the figure illustrates how the ability of measuring reliable mass maps degrades as galaxies approach the resolution limit or reach low surface brightnesses : at sizes @xmath56psf ( roughly 15 pixels ) and magnitudes @xmath57 systematic shifts and/or large deviations from the input model affect the measured mass maps .. for observations at a different depth this limit will scale accordingly . ]
we thus used the thresholds @[email protected] and @xmath58@xmath85@xmath59psf to select galaxies with reliable mass maps , reducing the sample of @xmath60 galaxies to 89 objects .
two of these galaxies , i d 13508 and i d 11800 , are strongly contaminated by a bright neighbor and have for this reason less robust mass maps .
these cases are flagged in table [ tab : sampleprop ] and excluded from the sample used for the analysis in section [ sec : masslightrealgal ] .
we show this final sample of 87 galaxies with red points in figure [ fig : f1 ] .
note that the @[email protected] selection is incidentally the same cut that has be applied for a reliable visual morphological classifications on the candels fields @xcite .
the details on the technical aspects of the mirage simulations are presented in @xcite and further discussions on the physics implemented in these simulations can also be found in @xcite and @xcite .
briefly , three closed - box disk models with a bulge - to - total fraction of 8% in mass and stellar masses of @xmath61 ( simulation g3 in @xcite ) , @xmath62 ( g2 ) and @xmath2 ( g1 ) were generated using an adaptive mesh refinement technique with the ramses code @xcite .
these simulations reach a resolution of 7.3pc at the highest level of refinement and have initial stellar mass particles of 1.2@xmath63 and 1.7@xmath63 in the bulge and disk components , respectively .
the disk galaxies were evolved in isolation or merged with each other , effectively probing merger ratios of 1:1 , 1:2.5 and 1:6.3 . in order to construct a representative sample of galaxy mergers ,
four different orbital parameters were explored for each merger ratio combination , resulting in a total of 20 mergers simulations ( plus the 3 isolated disk models ) . for each simulation configuration
we utilize in the following a set of 16 snapshots separated by 40myr each , covering an epoch from 200myr to 800myr from the simulations initial conditions ; for the merger models this corresponds to follow the pre- and post - coalescence phases for roughly 300myr each ( the coalescence time is visually determined in @xcite ) .
the mirage simulations feature several aspects which are paramount for a correct description of the ism physics and thus for a robust comparison with the real data .
first , the high level of grid refinement enables us to resolve jeans - unstable regions within the disks , consequently we can not only model the disk fragmentation and the formation of large clumps but also properly describe outflows and heating within the giant clumps themselves .
second , a physically motivated feedback model is implemented in the simulations by coupling standard supernovae feedback @xcite with the novel recipe for photoionization and radiation pressure feedback from ob - stars developed in @xcite .
this feedback model reproduces the typical outflows , star formation rates and the stellar population ages ( @xmath12200myr ) observed in real clumps @xcite .
third , the simulated disks have an initial total gas fraction of @xmath64 which is well representative of the observed high molecular fractions in typical @xmath16 , star - forming galaxies and is essential for the onset of gravitational instability and the generation of the giant clumps .
finally , the mirage simulations were originally designed as a numerical counterpart " for the massiv galaxy sample @xcite , and for this reason they are tailored to high-@xmath3 galaxies in terms of global properties ( e.g. , their sizes and sfr , see figures 6 and 9 of @xcite ) . [ cols="^ " , ] there is a clear trend of decreasing contamination from clumpy disks going from the @xmath3- to @xmath21-band and mass maps , supporting our approach .
although in the simulations the difference between @xmath13band and mass is not statistically significant , this trend is reinforced by the results in figure [ fig : ap2 ] , showing that the contamination is minimized if the classification is performed on the mass maps also for other combinations of structural indicators .
we emphasize here that as a consequence of all limitations listed in section [ sec : simscaveats ] , the simulated @xmath21-band images likely trace the mass distribution more closely than in real galaxies in which , e.g. , patchy dust obscuration or higher gas fractions , will increase the galaxy clumpiness and worsen the ability of separating mergers from disks .
we hence expect that the differences between the mass - based and flux - based classification should be more pronounced in real data .
this hypothesis is supported by the observational results discussed in section [ sec : masslightrealgal ] .
we also note that the mass - based selection of mergers reaches a completeness level that is comparable to the @xmath13band for simulated major mergers , whereas it is somewhat lower for the minor mergers .
this is not surprising since , by definition , the mass - based classification identifies galaxies with large mass contrasts / asymmetries and hence is less sensitive to minor mergers . in the simulations , some minor mergers can still be detected in the optical / nir bands as a consequence of sfr enhancements .
again , in real observations the classification of these minor mergers will be further complicated by the effects discussed above .
we use the results of the svm partition to define a quantitative criteria for selecting mergers based on the position in the @xmath11 and @xmath10 plane .
we concluded in section [ sec : simscaveats ] that an exact one - to - one match between the relative numbers of disks and mergers in the simulations and the real data is not expected .
for this reason we have refrained from performing any fine - tuning of the coefficients in the relation between @xmath11 and @xmath10 such as to optimize the completeness versus contamination level .
this would in fact depend on the input merger fraction in the simulations .
we rather follow the conservative approach of identifying a locus in the mass - derived @xmath11 versus @xmath10 plane which unless real galaxies have extreme mass distributions that are not reproduced by current models should be populated by mergers only .
this is of course still model dependent , but less affected by the specific choice of the merger fraction in the simulations .
our simulation - justified dividing line between mergers and normal galaxies is hence given by the following relation : @xmath65 and it is obtained by simply applying a shift to the best svm line in the @xmath11 versus @xmath10 plane of figure [ fig : f7 ] such to exclude the most extreme disks .
this relation is shown with a solid line in figure [ fig : f7 ] .
the completeness level over all pre - coalescence merger snapshots ( including the unperturbed light - blue points in figure [ fig : f7 ] ) reached with the selection of equation [ eq : eq1 ] is shown with a star symbol in the lower - right panel in figure [ fig : f7 ] . as a consequence of our rather conservative choice of the dividing line in equation [ eq : eq1 ] , the mass - based classification results in a very high purity of the selected mergers at the expense of the completeness of the merger sample . on the mirage simulated data , we estimate a completeness of up to @xmath540% for major mergers ( ratios 1:1 to 1:2.5 ) and up to 10% completeness for the minor mergers when applying equation [ eq : eq1 ] .
when dealing with real data , it is useful to introduce , together with the binomial classification of equation [ eq : eq1 ] , also a more probabilistic description of the @xmath11 versus @xmath10 plane which could be used to assign a merger likelihood on individual galaxies . for this reason using the results from the simulations
, we also derived a continuous parametrization of the contamination from clumpy disks over that plane .
this was obtained by modeling the number densities of the mergers and disk snapshots with a gaussian mixture approach and calculating the fraction of isolated disks over all snapshots in any given region . as we discussed in detail in section
[ sec : simscaveats ] , the relative number of mergers and disks in the simulation is almost 1:1 , whereas the major merger fractions reported in the literature at the redshift considered here are of order of 10% . to derive a realistic value of the relative abundance of disk and merger snapshots in @xmath11 versus @xmath10 space
, we hence rescaled the number counts of simulated mergers such as to reach a total merger fraction of 10% .
such that @xmath66 . in the above , @xmath67 and @xmath68
are the original number of simulated disks and mergers and @xmath69 is our target fraction of 10% . ]
we show this probabilistic mapping of disk contamination in the @xmath11 versus @xmath10 plane in the right - hand panel of figure [ fig : f8 ] .
this information can be used to refine the classification on a case by case analysis .
a kinematic classifications into rotationally supported disk or merger is currently available for some @xmath16 galaxies which can thus be used to further validate the calibration of the classification in the @xmath11 versus @xmath10 plane obtained from the mirage simulations . for this test
we consider galaxies with a similar optical+nir wavelength coverage as our hudf sample . in particular , a subset of the galaxies in the sins survey @xcite are located in the goods - s field and thus stellar mass maps
can be built using the goods+candels data .
specifically , we refer to those sins galaxies which have a kinemetry classification in table 9 of @xcite
. there are three such galaxies falling in the candels / goods field : k20-id6 , k20-id7 , k20-id8 which were originally extracted from the sample of @xcite .
k20-id6 and k20-id8 are classified as disks , while the nature of k20-id7 is less clear as it has been classified as merger by the kinemetry analysis in @xcite or as a disk displaying disturbed rotation in @xcite .
we label it as perturbed / merger " to indicate that it is not a regular disk .
for another galaxy that is included in our sample of galaxies with hudf+hudf12 coverage ( candels i d 15011 or udf i d 6462 ) a kinematic analysis based on sinfoni data has been published in the study of @xcite .
as discussed in that work , in spite of its disturbed appearance this galaxy clearly displays a rotational motion in the @xmath70 velocity field .
we hence include this object in the sample of kinematically classified disks .
we performed our mass map analysis on these galaxies and compare in figure [ fig : f8 ] their location with the expectations from the simulations .
inspecting figure [ fig : f8 ] , we find a very good agreement between our simulation - based partition of the @xmath11 versus @xmath10 plane and the location of the observed galaxies : all kinematically confirmed disks lie below the line defined in equation [ eq : eq1 ] in the disk - dominated locus and the kinematic merger / perturbed galaxy is on top of the merger locus , on a region where the disk contamination is expected to be @xmath1220% at the hudf depth .
we note that the real kinematically confirmed disks display somewhat higher @xmath10 and @xmath11 values than the bulk of simulated mirage disks . as mentioned in section [ sec : simscaveats ] , a lower gas fraction in the simulations with respect to real galaxies or the lack of dust could be responsible for the difference .
furthermore , we discuss in section [ sec : hvsmasshudf ] that a lower s / n causes a shift to higher @xmath15 .
small modifications to equation [ eq : eq1 ] should be applied when using data that is shallower than hudf , as for the sins galaxies .
when accounting for this effect , we still find a good agreement between the kinematic and mass classification , with galaxies k20-id6 , k20-id8 and i d 15011 falling in the disk - dominated region and the perturbed galaxy k20-id7 on the transition region between disks and mergers , consistently with its ambiguous kinematic classification .
we can also use the simulations to convert the structural selection into a typical observability window during which mergers are identified as such by the mass - based classification .
figure [ fig : f9 ] shows the evolution in time of the structural parameters in the mirage simulated disks and mergers . as a further illustration of the results of section [ sec : mass_vs_lightsims ] , we present the time variations for indices measured on the stellar mass maps as well as for those derived on the three optical / nir bands .
the different shades of blue illustrate the evolution of mergers with different mass ratios ( increasing merger ratio for darker shades ) , whereas the magenta lines are for the isolated disks .
for interacting galaxies we trace the parameters from an early merger phase at 50myr before pericenter when the two galaxies are still separated and hence the structural indices are measured on the primary only to after the complete fusion between the two galaxies .
the peak observed in most of the indices coincides with the pre - coalescence phase that we are targeting with our selection ; clearly it becomes more distinguishable from other phases when moving from the optical to the nir and the mass maps . the shaded gray areas in the right panels of the figure highlight the time interval over which 90% of the mergers identified by the selection in equation [ eq : eq1 ] are found : following our classification criteria we are able to identify mergers for about 200myr , i.e. between 300myr and 50myr before coalescence .
this time scale is consistent with those reported by , e.g. , @xcite for optically selected mergers , although these authors find a wide range of observability time - scales ( between @xmath50.1 and 1gyr ) depending on the orientation , type of mergers as well as on the structural indicators used to detect the merger features .
finally , in the top panel of figure [ fig : f9 ] we show for comparison the evolution of the sfr in the simulated galaxies ( see also figure 7 of @xcite ) .
interestingly , the variation in the structural parameters occurring during the merger event is not associated to a clear sfr enhancement in the simulations .
@xcite list several plausible causes for the lack of the starburst phase , including the interplay between the high gas fractions and the feedback treatment in the simulations .
a detailed study of the star formation histories in the mirage simulations is beyond the scope of this paper ( and is also affected by uncertainty in the models ) , the information is presented for completeness here but we will not further speculate on the links ( or absence thereof ) between the structural properties and star formation activity of the simulations .
building on the results from the previous section , we now apply the mass - based classification and compare the outcome with the results from the @xmath13band images for the 87 galaxies with reliable mass maps in the hudf area ( i.e. , the 89 galaxies with sextractor kron radii larger than 5@xmath59 the psf fwhm and magnitudes @xmath7124.5 minus the two galaxies with mass maps affected by strong contamination from neighboring galaxies , see section [ sec : biases ] ) . for the galaxies in the hudf area ,
the availability of multi - depth data allows us to verify the reliability of the mass - based classification at different s / n .
specifically , we consider mass maps derived from the three combinations of optical plus nir imaging described in section [ sec : massmapsreal ] : ( 1 ) goods and candels - deep data , ( 2 ) hudf plus candels - deep , and ( 3 ) the deepest maps obtained from the hudf12+hudf observations . of the 87 galaxies here considered , 50 galaxies have coverage at all depths , while for the other 37 only the first two versions of the mass maps could be derived . for comparison with the mass maps results ,
we also consider structural parameters measured on the @xmath13band images at the hudf12 or candels - deep depths . in figure
[ fig : f10 ] we present the location of the hudf galaxy sample in the @xmath15 versus asymmetry plane , either derived from the @xmath13band images ( left panels ) or the stellar mass maps ( right panels ) . from top to bottom
we show measurements performed on images of increasingly shallower depths .
the dotted lines in the figure indicate the dividing locus of equation [ eq : eq1 ] and the different symbols highlight the visual types of section [ sec : visualclass ] . in the following
we will regard galaxies with a visual class for the mass maps that is inconsistent with the quantitative criterion on @xmath10 and @xmath11 as failures of the mass - based classification , i.e. , as contaminants in the merger sample . in some cases such galaxies display genuine color gradients which could justify the mass asymmetries , but we will take the conservative approach of assuming they are the result of noise in the mass maps .
we start the analysis on the real galaxies by identifying in figure [ fig : f10 ] those galaxies which would be classified as mergers based on the irregularity of their @xmath13band images . for a consistent comparison with the selection based on the mass map , we consider as mergers candidates those galaxies that satisfy equation [ eq : eq1 ] also for the @xmath13band measurements , i.e. , galaxies above the dotted lines in figure [ fig : f10 ] .
other criteria are often used in the literature when using @xmath15 and asymmetries derived from optical images . at low redshift ( z@xmath411 ) a cut at @[email protected] - 0.35 is typically applied to select major mergers @xcite but for [email protected] galaxies a value of @[email protected] is preferred to account for the effects of the morphological @xmath73-correction , the decrease in the image resolution and surface brightness dimming @xcite . the distribution of irregular galaxies in figure 10 of @xcite would also suggest a division at @xmath74@xmath8 - 1.7 to identify mergers candidates .
we have checked ( and it can be verified by inspecting figure [ fig : f10 ] ) that the result here presented would not change if these other criteria are applied .
more than 90% of the galaxies that satisfy equation [ eq : eq1 ] in the @xmath13band were assigned to the class of multi - component / disturbed " galaxies also by the visual @xmath27 morphological analysis .
this means that these galaxies would be classified as mergers by both a multi - wavelength visual inspection and a quantitative structural analysis on the @xmath13band .
conversely , it can be seen that several of the merger candidates lying above the dotted line in the @xmath13band planes are visually classified as normal disk galaxies based on their mass map appearance ( orange filled triangles in the left panels of figure [ fig : f10 ] ) .
these galaxies have accordingly low asymmetry and @xmath15 indices measured on the mass map ( see right panels of figure [ fig : f10 ] ) . at both the hudf12 and candels - deep depth , about @xmath5 50% of those galaxies with large enough @xmath75 and @xmath74 to be classified as mergers in the @xmath13band have a mass profile consistent with that of a normal disk galaxy and fall below equation [ eq : eq1 ] when the mass - based indices are used . to illustrate these differences between mass and @xmath13band classification , we present in figure [ fig : f11 ] the stamp images for those galaxies with @xmath75 and @xmath74 as high as those of mergers ( using either the hudf12 or candels photometry ) but a disk - like mass profile . in spite of displaying composite @xmath27 and @xmath13band morphologies dominated by multiple clumps and asymmetric light distribution which could be suggestive of a merger event ,
all such galaxies have a regular , centrally concentrated distribution of mass , typically associated with a red nucleus . focusing on the right panels of figure [ fig : f10 ]
, we can instead study the results of the measurements performed on the mass maps . at the hudf+hudf12 depth
top most panel we find an excellent agreement between the locus of mergers expected from the analysis on the simulations ( see light shaded area and dotted line ) and the visual classes for the real data .
a comparison with the other right - hand panels , however , reveals some important facts : not surprisingly , the separation between galaxies visually classified as disks / compact and galaxies with visually disturbed morphologies becomes less clear as the s / n of the mass maps decreases . at the goods+candels - deep depth ,
the distribution of galaxies in the @xmath10 and @xmath11 plane is noticeably more clustered than for the structural parameters derived on the hudf+hudf12 mass maps .
it also appears rather clearly from figure [ fig : f10 ] that the bulk of real galaxies shifts toward higher @xmath11 values as a consequence of the decrease in s / n .
this is particularly evident when comparing the data points with the shaded locus defined by the simulations .
we showed in appendix [ appendix : noiseeffectsparameters ] that , while the asymmetry is almost insensitive to the image noise , some dependence of the @xmath15 on the s / n is expected and variations of the order of 10% are measured for @xmath15 indices derived at different depths .
this suggests that the criterion in equation [ eq : eq1 ] , which was derived for simulations reproducing the hudf+hudf12 depth , should be revised as an adaptive " threshold varying with the images s / n .
furthermore , we note that at the shallower depth a few merger candidates appear in a region of very low asymmetry ( @xmath76 ) but very high @xmath11 ( @xmath77 ) where virtually no simulated mergers nor galaxy visually classified as merger are found .
we suspect these galaxies to be affected by measurement errors .
we use the above empirical findings to provide refined criteria to robustly select mergers at the various depths : [ eq : eq2 ] m_20 , mass & > -1.1a_mass+ -1.12 h+h12 + -1.00 h+c + -0.98 g+c + + a_mass&>0.1 , where h+h12 " , h+c " and g+c " stand for the mass maps derived from the hudf+hudf12 , hudf+candels - deep and goods+candels - deep photometry , respectively .
these new relations where obtained by adding to the locus of equation [ eq : eq1 ] the average offsets in @xmath11 measured in figure [ fig : ap2 ] .
the new selection limits are shown with dashed lines in figure [ fig : f10 ] . by selecting as merger candidates those galaxies that satisfy the generic equations [ eq : eq1 ] we find a total of 9 , 21 and 26 mergers for the hudf+hudf12 , hudf+candels - deep and goods+candels - deep combinations , respectively .
of these 2 , 6 and 10 have a visual class that is inconsistent with the quantitative analysis on the mass maps , i.e they have been assigned an either compact " or disk " class in the visual inspection of the mass maps .
the use of the fixed selection of equation [ eq : eq1 ] would hence result in an increase of the contamination from @xmath78 for the hudf+hudf12 maps to @xmath79 for the goods+candels - deep combination .
conversely , using the relations in equation [ eq : eq2 ] we find 9 , 15 and 15 mergers candidates with a total of 2 , 3 and 4 galaxies visually classified as non interacting , namely a roughly constant contamination of @xmath78 , regardless of the image depth . what should be stressed here
is that applying the ` adaptive ' criteria of equation [ eq : eq2 ] also to the h - band would not reduce the contamination of clumpy disks : as it is clear from the bottom left panel of figure [ fig : f10 ] , these clumpy disks are distributed everywhere in the @xmath74 and @xmath75 plane and there is no selection that would substantially reduce the contamination . as a validation of the latter statement we considered the 50 galaxies for which all three versions of the mass maps are available and , for each combination of the photometric data
, we independently identified the samples of non - interacting galaxies and mergers , even when accounting for errors ( i.e. we selected as merger candidates those galaxies that have error bars above the relations in equation [ eq : eq2 ] ) .
we did so for both the @xmath21-band and mass structural indices .
we then assumed that the classification performed using the deep hudf+hudf12 mass maps is the
correct " one and we calculated for the mergers and disks identified at the other depths or in the @xmath13band the fraction that have a consistent classification
i.e. mergers or non - interacting galaxies classified as such also on the deep mass maps and the fraction of galaxies which are instead misclassified as a consequence of the lower s / n i.e. galaxies classified as mergers that are identified as not - merging on the deep hudf+hudf12 maps or @xmath13band images , and vice versa .
we show the results of this calculation in figure [ fig : f13 ] .
the number of identified mergers is small and thus the error bars relatively large , however the figure shows that by applying the relations in equation [ eq : eq2 ] the fraction of misclassified mergers " in the mass selected sample is consistent with being @xmath520% for mass maps obtained from both the hudf+candels - deep and goods+candels - deep combinations .
conversely , for the structural indices @xmath75 and @xmath74 the contamination of false mergers remains as high as @xmath550% for both the hudf and candels observations , even when imposing the condition in equation [ eq : eq2 ] . as witnessed by the increase in the fraction of galaxies that are misclassified as disks in figure [ fig : f13 ] , the drawback of the refined selection is that a larger number of mergers are missed from the selection at the shallower depths . summarizing the above findings , the availability of deep photometry is certainly necessary for maximizing the accuracy and completeness of the mass based classification
nonetheless , applying small corrections to our simulation - motivated selection of mergers reliable mergers samples can be identified also on medium depth images , albeit reaching a lower completeness .
most notably , independently of whether equations [ eq : eq1 ] or the refined relation in [ eq : eq2 ] are applied to the @xmath75 and @xmath74 indices , we find the @xmath13band classification results in a twice as large contamination of clumpy disks in the merger sample at any depth here considered .
we identify our final sample of merger candidates among the hudf sample as those galaxies that satisfy the selection criteria in equation [ eq : eq2 ] even when accounting for the error in the measurements ( i.e. , have error bars above those relations ) .
the classification is performed on the deep hudf+hudf12 mass maps when available and on the hudf+candels - deep maps for those galaxies with no hudf12 coverage .
we find a total of 11 such candidates over the 87 galaxies with reliable mass maps .
for the latter , we show in figure [ fig : f12 ] the @xmath27 , @xmath13band and mass stamps . among this sample of mergers
are included one galaxy with an inconsistent visual class and also one object that has been flagged as possible chance projection in section [ sec : projectioneffects ] ( i d 15844 , but note that even excluding the secondary blue clump this galaxy would still be classified as a merger ) , suggesting again a 20% contamination level .
thus far we have focused on those galaxies which are misclassified as mergers in the @xmath13band because of the presence of bright clumps .
it is however interesting to test whether also the opposite occurs , i.e. , whether there are galaxies classified as mergers in the structural analysis performed on the mass maps that instead appear smooth(er ) in the @xmath13band , for example as a result of dust lanes or intrinsically high mass - to - light ratios .
figure [ fig : f13 ] suggests that this happens for roughly 10% of the
@xmath21-band smooth " galaxies . in our sample of 11 mass - selected merger candidates we find five galaxies ( ids 9704 , 11388 , 12624 , 14533 and 15432 ) that fall below the dotted lines in figure [ fig : f8 ] when using @xmath74 and @xmath75 and hence would not be classified as mergers in the @xmath13band .
to understand the origin of this discrepancy , we generated new versions of the mass maps for these galaxies obtained by applying no dust extinction during the pixel - by - pixel sed fitting .
this gives us an indication of whether the mass asymmetry is driven by the dust correction applied to the templates .
ids 9704 and 14533 display significant asymmetries also in these mass maps , indicating that the classification is independent of the extinction law assumed .
ids 12624 , 11388 and 15432 instead would not satisfy our selection criteria with dust - free mass maps .
we note that both ids 12624 and 15432 are detected in the fir with total ir luminosities @xmath80 , implying a substantial amount of dust in these galaxies leading us to favor the results obtained including dust corrections . while it is clear that additional , e.g. , kinematic , information is required for a complete characterization of the population of mass - smooth and h - clumpy " galaxies
, the available data suggest that most of these galaxies have distinct substructure and that dust obscuration can explain some of the differences between the @xmath21-band and mass morphology .
using a sample of about 100 hudf galaxies with @xmath81 , we have compared a morphological classification of merging galaxies based on non parametric structural indices derived on resolved stellar mass maps with a canonical classification obtained from @xmath13band measurements .
we tested this approach using photometry at the different depths available on the hudf area ( goods , candels and hudf / hudf12 ) and , performing tests on artificial stellar mass distributions , we derived luminosity and size limits for which reliable mass maps can be obtained .
the selection of mergers in the mass domain is calibrated using a sample of isolated and merging galaxies from the mirage hydrodynamical simulations which have been post - processed and analyzed to closely reproduce the observational data .
we summarize our findings as follows : 1 .
as also discussed in previous works @xcite , the stellar mass maps morphologies can not be reproduced by @xmath13band data alone which often display merger - like features even for galaxies with perfectly smooth , disk - like mass profiles .
although some of these galaxies display a red nucleus indicative of a central mass concentration , even a combined visual inspection of multiple bands is still affected by the presence of star - forming clumps and does not provide a full proxy for the mass distribution . as a consequence of the above , we quantify that merger samples identified on the basis of asymmetry / irregularity in the @xmath13band images can have a contamination from clumpy galaxies as high as 50% .
our analysis also suggests that differences between the mass- and @xmath21-band selection may originate from a population of galaxies with smoother @xmath21-band images than mass distribution , possibly as a result of dust extinction .
2 . on both the data and the simulations
, we find that a combination of @xmath10 and @xmath11 measured on the stellar mass maps is instead most cleanly separating major mergers from isolated , clumpy galaxies .
we use the results from the simulations and the analysis on the real hudf galaxies to provide in equations [ eq : eq1 ] and [ eq : eq2 ] quantitative criteria to separate the galaxy population into mergers and disks using mass maps derived at the hudf+hudf12 , hudf+candels - deep and goods+candels - deep depths . applying these criteria
, we estimate that chance projections or the scattering of clumpy disks in the merger sample by noise in the mass maps result in a contamination of roughly 20% . when applied to galaxies with available kinematic data ,
our @xmath10 and @xmath11 based classification results in morphologies that are consistent with the kinematic classes .
the ability of the mass - based @xmath10 and @xmath11 indices to select true mergers is not significantly affected by a moderate decrease in the s / n of the parent images if the conditions in equation [ eq : eq2 ] are used .
conversely , the @xmath21-band classification result in a twice as large contamination from clumpy disks , independently of the criteria that are applied .
4 . from the analysis of the mirage simulated mergers we estimate the proposed @xmath10 , @xmath11 selection should be sensitive to major mergers between 300 myr and 50 myr before coalescence .
based on the above results , we thus suggest that to identify major mergers a classification in the mass domain rather than from optical / nir images should be preferred and performed whenever possible .
obviously , our technique is only sensitive to merger phases in which the perturbation in the mass profiles is measurable ( roughly up to 300myr before coalescence as specified above ) and , as also a consequence of imposing a high purity in the resulting merger sample , is strongly biased against earlier merger stages .
a combination with other techniques ( e.g. , close kinematic pair selection ) would be hence necessary for a full census of merging systems .
finally , it is possible that variations in the gas fraction of galaxies and the actual masses of giant clumps could introduce some redshift dependent scaling of the threshold here derived , moving the locus of clumpy disks toward slightly higher @xmath10 and/or @xmath11 values at higher redshift .
however , we expect that these variations would be comparable to the uncertainties in the mass maps and the derived structural parameters and hence would not strongly affect the proposed classification for sufficiently large samples .
a.c . thanks p. hurley , m. t. sargent and v. strazzullo for useful discussions and suggestions .
we also thank the anonymous referee for valuable comments which improved the manuscript .
we acknowledge financial support from the swiss national science fundation ( a.c . , project pbezp2_137312 ) , from the agence nationale de la recherche ( a.c . ,
e. l. f. , contract # anr- 12-js05 - 0008 - 01 ) and from the e. c. through an erc grant ( a.c .
, f. b. , stg-257720 ) .
the simulations used in this work were performed on genci resources at the tres grand centre de calcul ( project 04 - 2192 ) and at the lrz supermuc facility under prace allocation number 50816 .
this work is based on observations taken by the candels multi - cycle treasury program with the nasa / esa hst , which is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas5 - 26555 .
this work has made use of the adaptive smoothing code adaptsmooth , developed by stefano zibetti and available at the url http://www.arcetri.astro.it/ zibetti / software / adaptsmooth.html + abraham , r. g. , van den bergh , s. , & nair , p. 2003
, , 588 , 218 agertz , o. , teyssier , r. , & moore , b. 2009 , , 397 , l64 arnouts , s. , cristiani , s. , moscardini , l. , et al . 1999 , , 310 , 540 bailin , j. , & harris , w. e. 2008 , , 681 , 225 barton , e. j. , geller , m. j. , & kenyon , s. j. 2000 , , 530 , 660 beckwith , s. v. w. , stiavelli , m. , koekemoer , a. m. , et al .
2006 , , 132 , 1729 bershady , m. a. , jangren , a. , & conselice , c. j. 2000 , , 119 , 2645 bluck , a. f. l. , conselice , c. j. , bouwens , r. j. , et al .
2009 , , 394 , l51 bohlin , r. c. , cornett , r. h. , hill , j. k. , et al .
1991 , , 368 , 12 bournaud , f. , daddi , e. , elmegreen , b. g. , et al . 2008 , , 486 , 741 bournaud , f. , elmegreen , b. g. , & elmegreen , d. m. 2007 , , 670 , 237 bournaud , f. , perret , v. , renaud , f. , et al .
2014 , , 780 , 57 bournaud , f. , daddi , e. , wei , a. , et al .
2015 , , 575 , a56 bouwens , r. j. , illingworth , g. d. , oesch , p. a. , et al .
2011 , , 737 , 90 brammer , g. b. , van dokkum , p. g. , & coppi , p. 2008
, , 686 , 1503 bridge , c. r. , carlberg , r. g. , & sullivan , m. 2010 , , 709 , 1067 brinchmann , j. , charlot , s. , white , s. d. m. , et al .
2004 , , 351 , 1151 bruzual , g. , & charlot , s. 2003 , , 344 , 1000 carlberg , r. g. , cohen , j. g. , patton , d. r. , et al .
2000 , , 532 , l1 carollo , c. m. , bschorr , t. j. , renzini , a. , et al . 2013a , , 773 , 112 carollo , c. m. , cibinel , a. , lilly , s. j. , et al .
2013b , , 776 , 71 calzetti , d. , armus , l. , bohlin , r. c. , et al .
2000 , , 533 , 682 cameron , e. , & driver , s. p. 2007
, , 377 , 523 cameron , e. , carollo , c. m. , oesch , p. a. , et al .
2011 , , 743 , 146 cappellari , m. , & copin , y. 2003 , , 342 , 345 ceverino , d. , dekel , a. , & bournaud , f. 2010 , , 404 , 2151 chabrier , g. 2003 , , 115 , 763 cibinel , a.,carollo , c. m. , lilly , s. j. , et al . 2013a , , 776 , 72 cibinel , a. , carollo , c. m. , lilly , s. j. , et al .
2013b , , 777 , 116 coe , d. , bentez , n. , snchez , s. f. , et al .
2006 , , 132 , 926 conselice , c. j. 2003 , , 147 , 1 conselice , c. j. , blackburne , j. a. , & papovich , c. 2005 , , 620 , 564 conselice , c. j. , rajgor , s. , & myers , r. 2008 , , 386 , 909 contini , t. , garilli , b. , le fvre , o. , et al .
2012 , , 539 , a91 cowie , l. l. , hu , e. m. , & songaila , a. 1995 , , 110 , 1576 daddi , e. , cimatti , a. , renzini , a. , et al . 2004a , , 600 , l127 daddi , e. , cimatti , a. , renzini , a. , et al .
2004b , , 617 , 746 daddi , e. , dickinson , m. , morrison , g. , et al .
2007 , , 670 , 156 daddi , e. , bournaud , f. , walter , f. , et al .
2010 , , 713 , 686 daddi , e. , dannerbauer , h. , liu , d. , et al .
2014 , arxiv:1409.8158 dekel , a. , sari , r. , & ceverino , d. 2009 , , 703 , 785 de ravel , l. , le fvre , o. , tresse , l. , et al .
2009 , , 498 , 379 disney , m. j. 1976 , , 263 , 573 dubois , y. , & teyssier , r. 2008 , , 477 , 79 ebeling , h. , white , d. a. , & rangarajan , f. v. n. 2006 , , 368 , 65 elbaz , d. , daddi , e. , le borgne , d. , et al .
2007 , , 468 , 33 ellis , r. s. , mclure , r. j. , dunlop , j. s. , et al .
2013 , , 763 , l7 ellison , s. l. , patton , d. r. , simard , l. , & mcconnachie , a. w. 2008 , , 135 , 1877 elmegreen , d. m. , elmegreen , b. g. , ravindranath , s. , & coe , d. a. 2007 , , 658 , 763 epinat , b. , contini , t. , le fvre , o. , et al .
2009 , , 504 , 789 epinat , b. , tasca , l. , amram , p. , et al .
2012 , , 539 , a92 erb , d. k. , steidel , c. c. , shapley , a. e. , pettini , m. , & adelberger , k. l. 2004 , , 612 , 122 erb , d. k. , shapley , a. e. , pettini , m. , et al .
2006 , , 644 , 813 frster schreiber , n. m. , genzel , r. , bouch , n. , et al .
2009 , , 706 , 1364 frster schreiber , n. m. , shapley , a. e. , genzel , r. , et al .
2011 , , 739 , 45 franx , m. , illingworth , g. , & heckman , t. 1989 , , 98 , 538 giavalisco , m. , livio , m. , bohlin , r. c. , macchetto , f. d. , & stecher , t. p. 1996
, , 112 , 369 giavalisco , m. , ferguson , h. c. , koekemoer , a. m. , et al .
2004 , , 600 , l93 genel , s. , naab , t. , genzel , r. , et al .
2012 , , 745 , 11 genzel , r. , tacconi , l. j. , eisenhauer , f. , et al . 2006 , , 442 , 786 graham , a. w. , driver , s. p. , petrosian , v. , conselice , c. j. , bershady , m. a. , crawford , s. m. , & goto , t. 2005 , , 130 , 1535 graham , a. w. , & worley , c. c. 2008 , , 388 , 1708 grogin , n. a. , kocevski , d. d. , faber , s. m. , et al .
2011 , , 197 , 35 guo , y. , giavalisco , m. , ferguson , h. c. , cassata , p. , & koekemoer , a. m. 2012 , , 757 , 120 guo , y. , ferguson , h. c. , giavalisco , m. , et al .
2013 , , 207 , 24 huertas - company , m. , kaviraj , s. , mei , s. , et al .
2014 , arxiv:1406.1175 ilbert , o. , arnouts , s. , mccracken , h. j. , et al .
2006 , , 457 , 841 impey , c. , & bothun , g. 1997 , , 35 , 267 kampczyk , p. , lilly , s. j. , carollo , c. m. , et al .
2007 , , 172 , 329 kampczyk , p. , lilly , s. j. , de ravel , l. , et al .
2013 , , 762 , 43 kartaltepe , j. s. , dickinson , m. , alexander , d. m. , et al .
2012 , , 757 , 23 kartaltepe , j. s. , mozena , m. , kocevski , d. , et al . 2014 ,
arxiv:1401.2455 kennicutt , r. c. , jr .
1998 , , 36 , 189 koekemoer , a. m. , faber , s. m. , ferguson , h. c. , et al .
2011 , , 197 , 36 kriek , m. , van dokkum , p. g. , labb , i. , et al .
2009 , , 700 , 221 lackner , c. n. , silverman , j. d. , salvato , m. , et al .
2014 , , 148 , 137 lang , p. , wuyts , s. , somerville , r. s. , et al .
2014 , , 788 , 11 law , d. r. , steidel , c. c. , erb , d. k. , et al .
2007 , , 656 , 1 leitherer , c. , schaerer , d. , goldader , j. d. , et al . 1999 , , 123 , 3 lisker , t. 2008 , , 179 , 319 lotz , j. m. , primack , j. , & madau , p. 2004
, , 128 , 163 lotz , j. m. , davis , m. , faber , s. m. , et al .
2008a , , 672 , 177 lotz , j. m. , jonsson , p. , cox , t. j. , & primack , j. r. 2008b , , 391 , 1137 lotz , j. m. , jonsson , p. , cox , t. j. , & primack , j. r. 2010 , , 404 , 575 magdis , g. e. , daddi , e. , bthermin , m. , et al .
2012 , , 760 , 6 maller , a. h. , berlind , a. a. , blanton , m. r. , & hogg , d. w. 2009 , , 691 , 394 man , a. w. s. , zirm , a. w. , & toft , s. 2014 , arxiv:1410.3479 mandelker , n. , dekel , a. , ceverino , d. , et al .
2014 , , 443 , 3675 moody , c. e. , guo , y. , mandelker , n. , et al .
2014 , , 444 , 1389 newman , a. b. , ellis , r. s. , bundy , k. , & treu , t. 2012 , , 746 , 162 noguchi , m. 1999 , , 514 , 77 pannella , m. , elbaz , d. , daddi , e. , et al .
2014 , arxiv:1407.5072 papovich , c. , giavalisco , m. , dickinson , m. , conselice , c. j. , & ferguson , h. c. 2003 , , 598 , 827 papovich , c. , dickinson , m. , giavalisco , m. , conselice , c. j. , & ferguson , h. c. 2005 , , 631 , 101 peng , c. y. , ho , l. c. , impey , c. d. , & rix , h .- w .
2010 , , 139 , 2097 perret , v. , renaud , f. , epinat , b. , et al .
2014 , , 562 , a1 pozzetti , l. , bolzonella , m. , zucca , e. , et al .
2010 , , 523 , a13 pipino , a. , cibinel , a. , tacchella , s. , et al .
2014 , , 797 , 127 renaud , f. , bournaud , f. , emsellem , e. , et al .
2013 , , 436 , 1836 saintonge , a. , lutz , d. , genzel , r. , et al .
2013 , , 778 , 2 sanders , j. s. , & fabian , a. c. 2001 , , 325 , 178 sargent , m. t. , daddi , e. , bthermin , m. , et al .
2014 , , 793 , 19 scarlata , c. , carollo , c. m. , lilly , s. , et al .
2007 , , 172 , 406 scoville , n. , aussel , h. , brusa , m. , et al .
2007 , , 172 , 1 shapiro , k. l. , genzel , r. , frster schreiber , n. m. , et al .
2008 , , 682 , 231 tacchella , s. , lang , p. , carollo , c. m. , et al .
2015 , , 802 , 101 schlegel , d. j. , finkbeiner , d. p. , & davis , m. 1998 , , 500 , 525 schweizer , f. 1979 , , 233 , 23 swinbank , a. m. , smail , i. , longmore , s. , et al .
2010 , , 464 , 733 tacconi , l. j. , genzel , r. , neri , r. , et al .
2010 , , 463 , 781 tacconi , l. j. , neri , r. , genzel , r. , et al .
2013 , , 768 , 74 trujillo , i. , aguerri , j. a. l. , cepa , j. , & gutirrez , c. m. 2001 , , 328 , 977 teyssier , r. 2002 , , 385 , 337 van starkenburg , l. , van der werf , p. p. , franx , m. , et al .
2008 , , 488 , 99 welikala , n. , connolly , a. j. , hopkins , a. m. , scranton , r. , & conti , a. 2008 , , 677 , 970 whitaker , k. e. , labb , i. , van dokkum , p. g. , et al .
2011 , , 735 , 86 wijesinghe , d. b. , hopkins , a. m. , kelly , b. c. , welikala , n. , & connolly , a. j. 2010 , , 404 , 2077 williams , r. j. , quadri , r. f. , & franx , m. 2011 , , 738 , ll25 windhorst , r. a. , taylor , v. a. , jansen , r. a. , et al .
2002 , , 143 , 113 wisnioski , e. , frster schreiber , n. m. , wuyts , s. , et al .
2015 , , 799 , 209 wuyts , s. , frster schreiber , n. m. , genzel , r. , et al .
2012 , , 753 , 114 zamojski , m. a. , schiminovich , d. , rich , r. m. , et al .
2007 , , 172 , 468 zibetti , s. 2009 , arxiv:0911.4956 zibetti , s. , charlot , s. , & rix , h .- w .
2009 , , 400 , 1181
as a basic consistency check of the mass maps for the sample described in section [ sec : sampleselection ] , we show in figure [ fig : ap1 ] the comparison between the sum of the masses in each individual pixel of the galaxy and the total galaxy mass that is obtained from the integrated photometry . although for highly obscured or strongly star - forming galaxies differences between the two estimates can be expected on an object by object basis due to the patchy distribution of dust , strong disagreement between the resolved and integrated estimates would be an indication of biases in the pixel - by - pixel sed fits .
we instead find a very good agreement between the pixel - based total mass and the whole galaxy mass , with a median difference which is less than 0.1dex . as mentioned in section [ sec : biases ] , the resolution and noise of the parent observations set major limitations for the estimate of the stellar mass maps . in order to derive the size and magnitude limits for which a reliable mass map can be obtained , we tested the pixel - based sed fitting on a sample of artificial galaxies with known mass distribution . to perform the test on model galaxies which are representative of the typical real galaxies
, the models were drawn from the observed sample as follows .
for each real hudf galaxy in the initial sample presented in section [ sec : sampleselection ] , we generated a toy mass distribution characterized by a srsic profile with a total mass equal to the galaxy integrated mass and structural parameters determined by the best - fit @xmath13band galfit @xcite model . to densely populate the magnitude - size plane
, we also generated extra models by adding random perturbations on the original galaxies parameters .
we then imposed each pixel in the artificial galaxy to have an sed equal to the observed best - fit sed for the entire galaxy ( including a uniform extinction for all pixels ) and used this sed to predict the pixel fluxes . from these flux maps we generated artificial @xmath22 images which we pasted into blank sky regions to mimic the typical s / n properties reached by the real galaxies with hudf optical plus candels - deep nir coverage .
this depth is in between the two extreme combination of imaging ( deep hudf+hudf12 or shallow goods+candels - deep ) used for creating the mass maps ( see section [ sec : archivaldata ] ) . specifically , we created stamps at the hudf depth for the f435w , f606w , f775w and f850lp filters , while the f814w , f105w , f125w and f160w images were matched to the candels - deep depth .
all images were degraded to the resolution of the f160w filter as in the real sample .
we finally computed observed " mass maps form these artificial images as described in section [ sec : massmapsreal ] , i.e. , by running adaptsmooth , extracting pixel - based seds and fitting them with lephare and used these observed " mass maps for our tests in section [ sec : biases ] .
although these simulations are undoubtedly a simplification of reality which is complicated by pixel - by - pixel variations of the sed , dust content , etc .
they provide us with a measure of systematic biases in the mass estimation . 1 . the concentration @xmath82 defined as the logarithmic ratio of the radii containing 20% and 80% of the total flux @xmath83 @xcite .
2 . the gini coefficient @xcite , which describes the uniformity of the flux distribution on a scale between 0 ( all pixels with equal flux ) and 1 ( all flux in just one pixel ) .
3 . and finally the multiplicity @xmath84 that quantifies the presence of multiple components through a comparison between the original image and a resampled version in which the pixels are re - arranged in decreasing flux order from the brightest pixel @xcite . for this latter parameter
we slightly modified the original definition proposed in @xcite to account for ellipsoidal light / mass distribution .
we present in figure [ fig : f7_appendix ] the analogous of figure [ fig : f7 ] but this time considering planes that include also @xmath82 , gini and @xmath84 . on one hand
, the figure reinforces the results of sections [ sec : simulationinsight ] and [ sec : masslightrealgal ] : even for these different combinations of structural indicators , the contamination in a mass - based selection of mergers is smaller than that obtained with measurements performed on the f160w and f850lp images . on the other ,
it justifies the choice of @xmath11@xmath10 as our fiducial combination : it can be noticed in fact that these other sets of indices result in either a higher contamination or lower completeness with respect to a classification based on @xmath11@xmath10 ( see lower panels of figure [ fig : f7_appendix ] and figure [ fig : f7 ] ) .
an exception is the gini@xmath85@xmath10 combination which produces comparable results to those obtained for @xmath11@xmath10 .
we show in the following that the gini coefficient is however more sensitive to the choice of the aperture used in the calculation and the noise in the image than the @xmath15 or @xmath72 indices and hence a less stable structural measurement .
this is illustrated in figure [ fig : ap2 ] where study the variation of the structural indices with image depth and aperture size used in the calculation .
we only discuss the comparison for the structural indices measured on the mass maps obtained from the combinations of hudf+hudf12 , hudf+candels - deep and goods+candels - deep photometry .
nonetheless , we also present the results for the indices derived on the @xmath13band at hudf or candels depth for completeness of information
. specifically , we considered those galaxies ( 50 ) with @xmath86psf and @xmath7124.5 that have coverage in the hudf , candels as well as goods fields and calculated the structural indices at all depths , within either a kron or petrosian aperture .
the use of two different apertures allows us to further test the impact of the noise in the images : the kron aperture is usually larger than the petrosian radius . ]
( see table [ tab : sampleprop ] ) and hence includes a higher number of low flux pixels ; differences between the two estimates can be used to assess the stability of the measurement .
this is also the reason why we opted for a petrosian aperture in the calculation of the parameters in section [ sec : nonparam_indices ] .
the points in figure [ fig : ap2 ] show the median values of the structural indices over all the 50 galaxies here considered .
a reliable measure of structure should show as little variation as possible with both imaging depth and aperture size in figure [ fig : ap2 ] .
the asymmetry @xmath10 is in this sense very robust with median values that remain almost identical at all depths and for both apertures .
this should be compared , for example , with the results for the gini coefficient in the right - most column which clearly shows a strong dependence on the aperture size with differences up to about 30% . as already pointed out in the work of @xcite ,
this is a consequence of the fact that gini becomes a tracer of the noise in the images rather than galactic structure for large apertures .
the other parameters , @xmath11 , @xmath87 and @xmath88 , all show some dependence on the depth of the images used for constructing the maps , shifting towards lower concentrations / higher clumpiness when using the goods+candels - deep photometry instead of the deep hudf+hudf12 imaging .
the largest variations are measured for the multiplicity @xmath87 ( more than 50% change among the various cases ) , whereas the @xmath11 coefficient varies by @xmath89 . in combination with figure [ fig : f7 ] ,
these findings justify our choice of @xmath10- @xmath11 in section [ sec : testonsims1 ] as the most powerful and less noise- or aperture - dependent parameter set for the discrimination of merging and clumpy galaxies .
cccccccccccc 8740 & & & & 53.16764 & -27.83037 & 1.88 & -1.670 & 0.065 & 0.114 & 0.010 & not merging + 8750 & & & & 53.16287 & -27.82947 & 2.04 & -1.220 & 0.125 & 0.125 & 0.040 & unres./faint + 9295 & & & & 53.17170 & -27.82566 & 1.73 & -1.850 & 0.070 & 0.096 & 0.010 & not merging + 9343 & & & & 53.16714 & -27.82450 & 1.84 & -1.210 & 0.095 & 0.160 & 0.014 & not merging + 9407 & & & & 53.17061 & -27.82379 & 2.69@xmath90 & -0.610 & 0.070 & 0.460 & 0.186 & merger ( 2 ) + 9474 & & & & 53.16978 & -27.82394 & 2.24 & -1.260 & 0.055 & 0.112 & 0.021 & not merging + 9527 & & & & 53.15673 & -27.82306 & 1.72 & -1.950 & 0.085 & 0.036 & 0.013 & not merging + 9835 & & & & 53.17343 & -27.82028 & 2.28 & -1.320 & 0.105 & 0.036 & 0.078 & unres./faint + 9987 & & & & 53.14894 & -27.81928 & 2.23 & -1.180 & 0.085 & 0.183 & 0.049 & not merging + | at @xmath0 , the distinction between merging and normal " star - forming galaxies based on single band morphology is often hampered by the presence of large clumps which result in a disturbed , merger - like appearance even in rotationally supported disks . in this paper
we discuss how a classification based on canonical , non - parametric structural indices measured on resolved stellar mass maps , rather than on single - band images , reduces the misclassification of clumpy but not merging galaxies .
we calibrate the mass - based selection of mergers using the mirage hydrodynamical numerical simulations of isolated and merging galaxies which span a stellar mass range of @xmath1@xmath2 and merger ratios between 1:11:6.3 .
these simulations are processed to reproduce the typical depth and spatial resolution of observed hubble ultra deep field ( hudf ) data .
we test our approach on a sample of real @xmath3@xmath42 galaxies with kinematic classification into disks or mergers and on @xmath5100 galaxies in the hudf field with photometric / spectroscopic redshift between 1.5@xmath6@xmath3@xmath63 and @xmath7@xmath8@xmath9 .
we find that a combination of the asymmetry @xmath10 and @xmath11 indices measured on the stellar mass maps can efficiently identify real ( major ) mergers with @xmath1220% contamination from clumpy disks in the merger sample .
this mass - based classification can not be reproduced in star - forming galaxies by @xmath13band measurements alone , which instead result in a contamination from clumpy galaxies that can be as high as 50% .
moreover , we find that the mass - based classification always results in a lower contamination from clumpy galaxies than an @xmath13band classification , regardless of the depth of the imaging used ( e.g. , candels versus hudf ) . |
if @xmath0 and @xmath1 are both primes , then this prime pair is called cousin primes .
hardy and littlewood conjectured that cousin primes have the same asymptotic density as twin primes@xcite .
is there infinitely cousin primes ?
it is an old unsolved conjecture@xcite
. we will prove it as theorem [ theorem ] .
[ theorem ] there are infinitely many cousin primes .
let p=@xmath2 be the primes not exceeding @xmath3 , then the number of primes not exceeding @xmath4 @xcite is , @xmath5 for simplicity , we can write it as , @xmath6 \left[1- \frac{1}{p_2}\right ] \cdots \left[1- \frac{1}{p_v}\right]\\ & = ( \pi(\sqrt{n})-1)+n\prod_{i=1}^{v}\left[1- \frac{1}{p_i}\right ] , \\ \end{array}\ ] ] where @xmath7=n-\left\lfloor\frac{n}{p_i}\right\rfloor$ ] , @xmath8\left[1- \frac{1}{p_j}\right]=n-\left\lfloor\frac{n}{p_i}\right\rfloor -\left\lfloor\frac{n}{p_j}\right\rfloor+\left\lfloor\frac{n}{p_ip_j}\right\rfloor$ ] .
the operator @xmath9 $ ] will leave the items which are not multiples of @xmath10 . in this paper , @xmath11 is the floor function , @xmath12 the ceiling function of @xmath13 .
the integral operator ` @xmath14 $ ] ' , which is not the floor function in this paper , has meanings only operating on ( real ) number : @xmath15=[m]=\lfloor m\rfloor$ ] .
let @xmath17 be a natural arithmetic progression , @xmath18 be its accompanying arithmetic progression , so that @xmath19 .
there are @xmath4 such pairs .
@xmath20 if we delete all the pairs in which one or both items @xmath21 and @xmath22 are composite integers of the primes @xmath23 , then the pairs left are all cousin primes . because after deleted the composite integers in sets @xmath24 and @xmath25 , the items in the middle set @xmath26 are all composite integers : if @xmath27 is a prime , then @xmath28 or @xmath29 , we have either @xmath30 or @xmath31 , so this item has already been deleted . for a given @xmath10 ,
first we delete the multiples of @xmath10 in set @xmath24 , or the items of @xmath32 , @xmath33=\left\lfloor\frac{n}{p_i}\right\rfloor.\ ] ] secondly we delete the multiples of @xmath10 in @xmath25 , i.e. the items of @xmath34 , or @xmath35 ( or @xmath36 for @xmath37 , hereafter ) , @xmath38=\left\lfloor\frac{n+4{{\bf mod \/}}p_i}{p_i}\right\rfloor.\ ] ] because @xmath39 when @xmath40 , the items deleted by @xmath41 and @xmath42 are not the same . for @xmath43 , @xmath44 , we need only delete the items in set @xmath24 : @xmath45 eq . ( [ equ : y1 ] ) will delete all pairs with @xmath32 in set @xmath24 , and eq .
( [ equ : y2 ] ) will delete all pairs with @xmath46 in set @xmath24 .
@xmath47=n\left\lfloor\frac{1}{p_i}\right\rfloor+\delta,\ ] ] where @xmath48 with , @xmath49 the pairs left , after deleted the multiples of @xmath10 in both @xmath24 and @xmath25 , have , @xmath50.\ ] ] the operator @xmath51 $ ] , when operating on @xmath4 , will leaves the items having not those of @xmath52 . when the multiples of the primes @xmath53 in both @xmath24 and @xmath25 have been deleted , the pairs left will be prime pairs ( cousin primes ) and have , @xmath54 \left[1-\frac{1}{p_2}-\frac{\widetilde{1}}{p_2}\right ] \cdots\left[1-\frac{1}{p_{v}}-\frac{\widetilde{1}}{p_{v}}\right]\\ & = n\left[1-\frac{1}{p_1}\right ] \prod_{i=2}^v\limits\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right],\\ \end{array}\ ] ] the meaning is as follows , @xmath55=\left\lfloor\frac{n}{p_i}\right\rfloor\\ \\ n\left[\frac{\widetilde{1}}{p_i}\right]=\left\lfloor\frac{n+4{{\bf mod \/}}p_i}{p_i}\right\rfloor\\ \\ n\left[\frac{1}{p_i}\right]\left[\frac{1}{p_j}\right ] = n\left[\frac{1}{p_ip_j}\right ] = \left\lfloor\frac{n}{p_ip_j}\right\rfloor\\ \\ n\left[\frac{\widetilde{1}}{p_i}\right]\left[\frac{\widetilde{1}}{p_j}\right ] = n\left[\frac{\widetilde{1}}{p_ip_j}\right ] = \left\lfloor\frac{n+4}{p_ip_j}\right\rfloor\\ \\ n\left[\frac{1}{p_i}\right]\left[\frac{\widetilde{1}}{p_j}\right ] = n\left[\frac{1}{p_i}\frac{\widetilde{1}}{p_j}\right ] = \left\lfloor\frac{n+\theta_{i , j}}{p_ip_j}\right\rfloor = \left\lfloor\frac{n+p_ip_j-\lambda_{i , j}}{p_ip_j}\right\rfloor,\\ \end{array } \right.\ ] ] where @xmath56 is the position of the first item with @xmath57 and @xmath58 , @xmath59 let @xmath60 , @xmath61 be the position of the first item with @xmath62 , then @xmath63 , @xmath64\left[\frac{\widetilde{1}}{p_j}\right ] = \left\lfloor\frac{n+p_ip_j-\lambda_{i , j}}{p_ip_j}\right\rfloor = \left\lfloor\frac { \left\lfloor\frac{n}{p_i}\right\rfloor + p_j-\lambda_{j } } { p_j}\right\rfloor.\\ \end{array}\ ] ] if there is no such @xmath65 in @xmath24 , i.e. , @xmath66 , then the last item in eq .
( [ equ : expandp ] ) will equal zero . because @xmath67 , so @xmath68 .
the total number of cousin primes in @xmath4 is , @xmath69 \prod_{i=2}^v\limits\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right ] + d(\sqrt{n})-d_1,\\ \end{array}\ ] ] where @xmath70 is the number of cousin primes @xmath71 $ ] when @xmath72 , where @xmath73 and , @xmath74 @xmath75 is the number of pairs which are not cousin primes and have not been deleted by the process . when @xmath76 , some pairs which are not cousin primes may not be deleted . if @xmath77 then @xmath78 . because if @xmath79 , then @xmath80 , if @xmath81 , as any prime must has the form @xmath82 , where @xmath83 is an integer , so @xmath84 , this item has already been deleted .
the only item which may not be deleted is the one @xmath85 .
but after deleted the items of @xmath86 in set @xmath24 , the items left in set @xmath24 must have the form @xmath87 and set @xmath25 have the form @xmath88 correspondingly .
but @xmath89 , @xmath90 ( not 1 ) , so the pair including this item has also been deleted .
certainly , the pair ( 1,5 ) has been deleted by @xmath91 . if @xmath92 then @xmath93 . because the pair of ( 1,5 ) is not one of cousin primes and has not been deleted .
besides , we should note that if @xmath94 , then one number ( 11 ) in the cousin primes ( 7,11 ) is not less than @xmath95 , and if we delete this pair , then @xmath96 for @xmath43 . later , we will suppose that @xmath77 , and so @xmath78 for ` large ' number @xmath4 .
let @xmath97 , then @xmath98 , @xmath99 $ ] , @xmath100 $ ] . from eq .
( [ equ : d0n ] ) , ( [ equ : lambda2 ] ) , @xmath101 \left[1-\frac{1}{3}-\frac{\widetilde{1}}{3}\right ] \left[1-\frac{1}{5}-\frac{\widetilde{1}}{5}\right]\\ & = 48\left[1-\frac{1}{2}-\frac{1}{3}+\frac{1}{6 } -\frac{\widetilde{1}}{3}+\frac{1}{2}\frac{\widetilde{1}}{3 } -\frac{1}{5}+\frac{1}{10}+\frac{1}{15}-\frac{1}{30 } + \frac{1}{5}\frac{\widetilde{1}}{3}-\frac{1}{10}\frac{\widetilde{1}}{3 } \right.\\ & { \ \ } \left .
-\frac{\widetilde{1}}{5}+\frac{1}{2}\frac{\widetilde{1}}{5 } + \frac{1}{3}\frac{\widetilde{1}}{5 } -\frac{1}{6}\frac{\widetilde{1}}{5 } + \frac{\widetilde{1}}{15}-\frac{1}{2}\frac{\widetilde{1}}{15}\right]\\ & = 48-\left\lfloor\frac{48}{2 } \right\rfloor -\left\lfloor\frac{48}{3 } \right\rfloor+\left\lfloor\frac{48}{6 } \right\rfloor -\left[\frac{\widetilde{48}}{3 } \right ] + \left[\frac{48}{2}\frac{\widetilde{1}}{3}\right]\\ & { \ \ } -\left\lfloor\frac{48}{5 } \right\rfloor+\left\lfloor\frac{48}{10 } \right\rfloor + \left\lfloor\frac{48}{15 } \right\rfloor -\left\lfloor\frac{48}{30 } \right\rfloor + \left[\frac{48}{5}\frac{\widetilde{1}}{3}\right ] -\left [ \frac{48}{10}\frac{\widetilde{1}}{3}\right]\\ & { \ \ } -\left[\frac{\widetilde{48}}{5 } \right ] + \left[\frac{48}{2}\frac{\widetilde{1}}{5 } \right ] + \left[\frac{48}{3}\frac{\widetilde{1}}{5 } \right ] -\left[\frac{48}{6}\frac{\widetilde{1}}{5 } \right ] + \left[\frac{\widetilde{48}}{15 } \right ] -\left[\frac{48}{2}\frac{\widetilde{1}}{15 } \right]\\ & = 48 - 24 -16 + 8 -\left\lfloor\frac{48 + 4{{\bf mod \/}}3}{3 } \right\rfloor + \left\lfloor\frac { \left\lfloor\frac{48}{2}\right\rfloor + 3 - 1 } { 3}\right\rfloor\\ & { \ \ } -9 + 4 + 3 -1 + \left\lfloor\frac { \left\lfloor\frac{48}{5}\right\rfloor + 3 - 1 } { 3}\right\rfloor -\left\lfloor\frac { \left\lfloor\frac{48}{10}\right\rfloor + 3 - 2 } { 3}\right\rfloor\\ & { \ \ } -\left\lfloor\frac{48 + 4}{5 } \right\rfloor + \left\lfloor\frac { \left\lfloor\frac{48}{2}\right\rfloor + 5 - 3 } { 5}\right\rfloor + \left\lfloor\frac { \left\lfloor\frac{48}{3}\right\rfloor + 5 - 2 } { 5}\right\rfloor -\left\lfloor\frac { \left\lfloor\frac{48}{6}\right\rfloor + 5 - 1 } { 5}\right\rfloor + \left\lfloor\frac{48 + 4}{15 } \right\rfloor -\left\lfloor\frac { \left\lfloor\frac{48}{2}\right\rfloor + 15 - 13 } { 15}\right\rfloor\\ & = 16 -16 + 8\\ & { \ \ } -3 + 3 - 1\\ & { \ \ } -10 + 5 + 3 - 2 + 3 - 1\\ & = 5 . \end{array}\ ] ] we can check it directly , @xmath102 $ ] @xmath103 $ ] @xmath104 $ ] @xmath105 $ ] @xmath106 $ ] .
so @xmath107 .
it is the same as before . from @xmath108 for @xmath109 , @xmath110 for @xmath111 .
so @xmath112 .
the set : @xmath113 .
the five cousin primes are @xmath114 and @xmath115 .
[ def:3 ] let @xmath116 be an ( any ) integer set , then @xmath117 : = \sum\limits _ { x_k{{\bf mod \/}}p_i\ne0,p_i-4{{\bf mod \/}}p_i}1 = t-\left[\frac{t+\phi'_i}{p_i}\right]-\left[\frac{t+\phi''_i}{p_i}\right ] \geq0 \end{array}\ ] ] is the number of items left after deleted the items of @xmath118 .
usually @xmath119\left[1-\frac{1}{p_j}-\frac{\widetilde{1}}{p_j}\right ] \ne \left(m\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right]\right ) \left[1-\frac{1}{p_j}-\frac{\widetilde{1}}{p_j}\right]$ ] , we can express it as @xmath120\left[1-\frac{1}{p_j}-\frac{\widetilde{1}}{p_j}\right ] = \left(m\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right]\right)\left[1-\frac{1'}{p_j}-\frac{1''}{p_j}\right ] , \end{array}\ ] ] is the number of items left when we first delete those @xmath121 from set @xmath24 , and then delete those @xmath122 from set @xmath123\}$ ] , where set @xmath124 is no longer an arithmetic sequence .
in general , @xmath125 \left[1-\frac{1}{p_j}-\frac{\widetilde{1}}{p_j}\right ] = \left(m\prod_{i=1}^{i_m}\limits\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right]\right ) \left[1-\frac{1'}{p_j}-\frac{1''}{p_j}\right ] .
\end{array}\ ] ]
@xmath126\left[1-\frac{1}{p_j}-\frac{\widetilde{1}}{p_j}\right ] = \left[1-\frac{1}{p_j}-\frac{\widetilde{1}}{p_j}\right]\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right ] .
\end{array } \ ] ] @xmath127 & = \left[\frac{m_1+m_2}{p_i}\right ] = \left[\frac{m_1}{p_i}\right]+\left[\frac{m_2}{p_i}\right]+\left[\frac{m_1 { { \bf mod \/}}p_i+m_2 { { \bf mod \/}}p_i}{p_i}\right ] .
\end{array}\ ] ] @xmath128 = ap_i\left(1-\frac{2}{p_i}\right)+b\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right ] \mbox{\quad for $ m = ap_i+b$}. \end{array}\ ] ] @xmath129 -m_1\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right ] -m_2\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right]\leq 1.\\ \end{array}\ ] ] let @xmath130 @xmath131 -m_1\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right ] -m_2\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right]\nonumber\\ = m_1+m_2-\left\lfloor\frac{m_1+m_2}{p_i}\right\rfloor-\left\lfloor\frac{m_1+m_2 + 4{{\bf mod \/}}p_i}{p_i}\right\rfloor -m_1+\left\lfloor\frac{m_1}{p_i}\right\rfloor+\left\lfloor\frac{m_1 + 4{{\bf mod \/}}p_i}{p_i}\right\rfloor\\ \
\ -m_2+\left\lfloor\frac{m_2}{p_i}\right\rfloor+\left\lfloor\frac{m_2 + 4{{\bf mod \/}}p_i}{p_i}\right\rfloor\nonumber\\ \\
= -\left\lfloor\frac{\alpha+\beta}{p_i}\right\rfloor-\left\lfloor\frac{\alpha+\beta+4{{\bf mod \/}}p_i}{p_i}\right\rfloor + \left\lfloor\frac{\alpha+4{{\bf mod \/}}p_i}{p_i}\right\rfloor + \left\lfloor\frac{\beta+4{{\bf mod \/}}p_i}{p_i}\right\rfloor\nonumber.\\ \end{array}\ ] ] the minimum : @xmath132 , when @xmath133 , @xmath134 , @xmath135 , and @xmath136 .
the maximum : @xmath137 , when @xmath138 , @xmath139 , @xmath140 ( it is true for @xmath37 ) . we can represent it as @xmath141 = m_1+m_2-\left[\frac{m_1+m_2}{p_i}\right]-\left[\frac{m_1+m_2 + 4{{\bf mod \/}}p_i}{p_i}\right]\\ & = & m_1-\left[\frac{m_1}{p_i}\right]-\left[\frac{m_1 + 4{{\bf mod \/}}p_i}{p_i}\right ] + m_2-\left[\frac{m_1 { { \bf mod \/}}p_i+m_2}{p_i}\right]-\left[\frac{m_2+(m_1 + 4{{\bf mod \/}}p_i){{\bf
mod \/}}p_i}{p_i}\right]\\ & = & m_1\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right ] + m_2\left[1-\frac{1'}{p_i}-\frac{1''}{p_i}\right].\\ \end{array}\ ] ] @xmath142 $ ] will delete the items of @xmath143 in the sequence of @xmath144 , and @xmath145 $ ] will delete the items of @xmath146 in the sequence of @xmath147 or the items of @xmath148 in set @xmath124 . for any @xmath149 , @xmath150 \leq m_2 $ ] , so , @xmath151 \geq m\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right ] \mbox{\quad for \ $ m'\geq m$}. \end{array}\ ] ]
[ lem:1 ] for @xmath152 , @xmath153 \geq \left\lceil m\left(1-\frac{3}{p_i}\right)\right\rceil.\\ \end{array}\ ] ] @xmath154 [ lem:3 ] for @xmath155 , @xmath156 , @xmath157 \left[1-\frac{1}{p_j}-\frac{\widetilde{1}}{p_j}\right ] \geq \left\lceil m\left(1-\frac{3}{p_j}\right)\right\rceil \left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right].\\ \end{array}\ ] ] @xmath158 \left[1-\frac{1}{p_j}-\frac{\widetilde{1}}{p_j}\right ] \geq \left\lceil m\left(1-\frac{3}{p_j}\right)\right\rceil \left[1-\frac{1}{p_i}\right]\mbox{\quad for $ p_i=2$}. \end{array}\ ] ] let @xmath159 , @xmath160 , @xmath161 , because @xmath162 , so @xmath163 . for simplicity ,
we write @xmath164 as 4 except for @xmath37 , or @xmath165 . from eq .
( [ equ : p2 ] ) , ( [ equ : p3 ] ) , @xmath166 \left[1-\frac{1}{p_j}-\frac{\widetilde{1}}{p_j}\right ] -\left\lceil m\left(1-\frac{3}{p_j}\right)\right\rceil \left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right]\\ & = sp_ip_j\left(1-\frac{2}{p_i}\right)\left(1-\frac{2}{p_j}\right ) -sp_ip_j\left(1-\frac{2}{p_i}\right)\left(1-\frac{3}{p_j}\right)\\ & \ \ + t-\left\lfloor\frac{t}{p_i}\right\rfloor-\left\lfloor\frac{t+4}{p_i}\right\rfloor -\left\lfloor\frac{t}{p_j}\right\rfloor+\left\lfloor\frac{t}{p_ip_j}\right\rfloor + \left\lfloor\frac{t+\theta_{j , i}}{p_ip_j}\right\rfloor -\left\lfloor\frac{t+4}{p_j}\right\rfloor+\left\lfloor\frac{t+\theta_{i , j}}{p_ip_j}\right\rfloor + \left\lfloor\frac{t+4}{p_ip_j}\right\rfloor\\ \\ & \ \
-\left(t-\left\lfloor\frac{3t}{p_j}\right\rfloor\right ) + \left\lfloor\frac{t-\left\lfloor\frac{3t}{p_j}\right\rfloor}{p_i}\right\rfloor + \left\lfloor\frac{t+4-\left\lfloor\frac{3t}{p_j}\right\rfloor}{p_i}\right\rfloor\\ & = s(p_i-2)\\ & \ \
+ \left\lfloor\frac{3t}{p_j}\right\rfloor -\left\lfloor\frac{t}{p_j}\right\rfloor-\left\lfloor\frac{t+4}{p_j}\right\rfloor + \left\lfloor\frac{t}{p_ip_j}\right\rfloor + \left\lfloor\frac{t+\theta_{j , i}}{p_ip_j}\right\rfloor + \left\lfloor\frac{t+\theta_{i , j}}{p_ip_j}\right\rfloor + \left\lfloor\frac{t+4}{p_ip_j}\right\rfloor\\ \\ & \ \ -\left\lfloor\frac{\left\lfloor\frac{3t}{p_j}\right\rfloor}{p_i}\right\rfloor -\varepsilon_1 -\left\lfloor\frac{\left\lfloor\frac{3t}{p_j}\right\rfloor}{p_i}\right\rfloor -\varepsilon_2\\ & = s(p_i-2)\\ & \ \ + 3a+\left\lfloor\frac{3b}{p_j}\right\rfloor -a - a-\left\lfloor\frac{b+4}{p_j}\right\rfloor + 0 + \left\lfloor\frac{ap_j+b+\theta_{j , i}}{p_ip_j}\right\rfloor + \left\lfloor\frac{ap_j+b+\theta_{i , j}}{p_ip_j}\right\rfloor + \left\lfloor\frac{ap_j+b+4}{p_ip_j}\right\rfloor\\ & \ \ -2\left\lfloor\frac{3a+\left\lfloor\frac{3b}{p_j}\right\rfloor}{p_i}\right\rfloor -\varepsilon_1 -\varepsilon_2\\ & = s(p_i-2)+a+\left\lfloor\frac{3b}{p_j}\right\rfloor + \left\lfloor\frac{ap_j+b+\theta_{j , i}}{p_ip_j}\right\rfloor + \left\lfloor\frac{ap_j+b+\theta_{i , j}}{p_ip_j}\right\rfloor + \left\lfloor\frac{ap_j+b+4}{p_ip_j}\right\rfloor\\ & \
\ -\varepsilon_1 -\varepsilon_2 - 2\varepsilon_3-\varepsilon_4,\\ \end{array}\ ] ] where @xmath167 @xmath168 1 . if @xmath169 , then @xmath170 2 . if @xmath171 , then 1 . if @xmath172 , then @xmath173 2 . if @xmath174 , then @xmath175 , @xmath176 , so @xmath177 3 . if @xmath178 , then 1
if @xmath179 , then @xmath180 2 .
if @xmath181 , then @xmath182 , @xmath183 , so @xmath184 3 .
if @xmath185 : if @xmath186\geq1 $ ] then @xmath187 , @xmath188 , so @xmath189-\delta \varepsilon\geq 3 + 1 + 1 - 5=0 $ ] ; else for @xmath186=0 $ ] then @xmath190 , @xmath176 , so @xmath191 .
4 . if @xmath192 , then @xmath193 , @xmath194 , so @xmath195 5 . if @xmath196 : if @xmath186\geq1 $ ] then @xmath189-\delta \varepsilon\geq 3 + 3 + 1 - 7=0 $ ] ; else for @xmath186=0 $ ] then @xmath197 , @xmath188 , so @xmath198 .
4 . if @xmath37 , then @xmath199 , + @xmath200 + @xmath201 1 . if @xmath202 , then @xmath203 , 1 . if @xmath204 , then @xmath205 , so @xmath206 . if @xmath174 then @xmath207 . else for @xmath208 , because @xmath209 , @xmath210 .
so @xmath211 .
else for @xmath212 , from @xmath213 , we have @xmath214 . if b=0 we follow the process above . for b=1 , @xmath215 .
@xmath216 * for a=0 , @xmath217 ; * for a=2 , @xmath218 ( it comes from one of 5,10,15 ) , @xmath219 ; * for a=1 , if @xmath220 then @xmath221 ; else for s=1 , from @xmath222 , and @xmath223 , we have @xmath192 .
so this case is impossible .
2 . if @xmath224 ( note @xmath225 ) , then @xmath226 1 . if @xmath220 then @xmath227 .
2 . if @xmath228 , i.e. , @xmath222 , so @xmath212 . from @xmath223
, we have @xmath192 .
let us consider @xmath229 , @xmath230 , with the condition @xmath231 i.e. @xmath232 , @xmath233 .
@xmath234 , therefore , @xmath235 + besides , @xmath236 1 . if @xmath237 , then @xmath238 .
2 . if @xmath239 , because , + @xmath240 + so @xmath241 and @xmath242 + therefore , @xmath243 + if @xmath244 then @xmath245 .
+ else for @xmath208 , because @xmath246 , @xmath247 .
so @xmath248 . in summary , @xmath249 for all @xmath250 .
when @xmath43 , @xmath251\equiv0 $ ] , @xmath252 , @xmath253 , @xmath254 , @xmath163 .
@xmath255 \left[1-\frac{1}{p_j}-\frac{\widetilde{1}}{p_j}\right ] -\left\lceil m\left(1-\frac{3}{p_j}\right)\right\rceil \left[1-\frac{1}{2}\right]\\ & = 2sp_j\left(1-\frac{1}{2}\right)\left(1-\frac{2}{p_j}\right ) -2sp_j\left(1-\frac{1}{2}\right)\left(1-\frac{3}{p_j}\right)\\ & \
\ + t-\left\lfloor\frac{t}{2}\right\rfloor -\left\lfloor\frac{t}{p_j}\right\rfloor+\left\lfloor\frac{t}{2p_j}\right\rfloor -\left\lfloor\frac{t+4{{\bf mod \/}}p_j}{p_j}\right\rfloor+\left\lfloor\frac{t+\theta_{i , j}}{2p_j}\right\rfloor \\ & \ \
-\left(t-\left\lfloor\frac{3t}{p_j}\right\rfloor\right ) + \left\lfloor\frac{t-\left\lfloor\frac{3t}{p_j}\right\rfloor}{2}\right\rfloor\\ & = \ \
s+\left\lfloor\frac{3t}{p_j}\right\rfloor -\left\lfloor\frac{t}{p_j}\right\rfloor + 0-\left\lfloor\frac{t+4{{\bf mod \/}}p_j}{p_j}\right\rfloor + \left\lfloor\frac{t+\theta_{i , j}}{2p_j}\right\rfloor -\left\lfloor\frac{\left\lfloor\frac{3t}{p_j}\right\rfloor}{2}\right\rfloor -\varepsilon_1\\ & = s+3a+\left\lfloor\frac{3b}{p_j}\right\rfloor -a - a-\left\lfloor\frac{b+4{{\bf mod \/}}p_j}{p_j}\right\rfloor + \left\lfloor\frac{ap_j+b+\theta_{i , j}}{2p_j}\right\rfloor -\left\lfloor\frac{3a+\left\lfloor\frac{3b}{p_j}\right\rfloor}{2}\right\rfloor -\varepsilon_1\\ & = s+\left\lfloor\frac{3b}{p_j}\right\rfloor + \left\lfloor\frac{ap_j+b+\theta_{i , j}}{2p_j}\right\rfloor -\varepsilon_1-\varepsilon_3-\varepsilon_4,\\ \end{array}\ ] ] where @xmath256 @xmath257 1 .
if @xmath258 , then 1 . if @xmath213 , then @xmath259 , @xmath260 + if @xmath181 , then @xmath261 , @xmath262 , @xmath263 ; + else for @xmath185 , @xmath264 .
@xmath265 , @xmath266 .
+ @xmath267 + @xmath268 + so @xmath269 2 . if @xmath270 , then @xmath271 : + if @xmath181 , then @xmath259 , @xmath272 ; + else for @xmath185 , then @xmath261 , @xmath273 .
3 . if @xmath274 , then @xmath275 .
if @xmath165 , we can follow the process of @xmath225 , only note that ` 4 ' should be changed to @xmath276 and @xmath277 , i.e. @xmath278 , @xmath279 . in summary , @xmath249 for all @xmath280 .
this lemma means that , from equation ( [ equ : l1 ] ) and ( [ equ : p5 ] ) , we can let @xmath281= \left\lceil m\left(1-\frac{3}{p_j}\right)\right\rceil+t , t\geq 0 $ ] , and operated by @xmath51 $ ] to get the items which have no multiples of @xmath10 and @xmath282 in @xmath24 and@xmath25 .
@xmath283 \left[1-\frac{1}{p_j}-\frac{\widetilde{1}}{p_j}\right ] = \left ( \left\lceil m\left(1-\frac{3}{p_j}\right)\right\rceil+t\right ) \left[1-\frac{1'}{p_i}-\frac{1''}{p_i}\right]\\ \\
= \left\lceil m\left(1-\frac{3}{p_j}\right)\right\rceil \left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right ] + t\left[1-\frac{1'}{p_i}-\frac{1''}{p_i}\right]\\ \\
\geq \left\lceil m\left(1-\frac{3}{p_j}\right)\right\rceil \left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right ] .
\end{array}\ ] ] [ lem:4 ] @xmath284 \left[1-\frac{1}{3}-\frac{\widetilde{1}}{3}\right ] \geq\left\lfloor\frac{m}{6}\right\rfloor.\\ \end{array}\ ] ] for @xmath285 , let @xmath286 , @xmath287 \left[1-\frac{1}{3}-\frac{\widetilde{1}}{3}\right ] = 6s\left[1-\frac{1}{2}\right ] \left[1-\frac{1}{3}-\frac{\widetilde{1}}{3}\right ] + t\left[1-\frac{1}{2}\right ] \left[1-\frac{1}{3}-\frac{\widetilde{1}}{3}\right]\\ \\
\geq 6s\left(1-\frac{1}{2}\right)\left(1-\frac{2}{3}\right ) = s=\left\lfloor\frac{m}{6}\right\rfloor .
\end{array}\ ] ] for the residual class of modulo 6 , @xmath288 , @xmath289 , there are 4 elements @xmath290 of multiples of 2 or 3 . for the other elements @xmath291 ,
there is at least one with @xmath292 , i.e. , @xmath293 , and @xmath294 .
the cousin primes have at least @xmath295 items , or @xmath296\left[1-\frac{1}{3}-\frac{\widetilde{1}}{3}\right ] \geq\left\lfloor\frac{m}{6}\right\rfloor$ ] . [ lem:5 ] for @xmath297 , @xmath298 \left[1-\frac{1}{p_j}-\frac{\widetilde{1}}{p_j}\right ] \geq \left\lceil m\left(1-\frac{3}{p_j}\right)\right\rceil \prod_{i=1}^{i_m}\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right],\\ \end{array}\ ] ] where @xmath299 for @xmath43 .
suppose that for @xmath300 , @xmath301 \left[1-\frac{1}{p_j}-\frac{\widetilde{1}}{p_j}\right ] = \left\lceil m\left(1-\frac{3}{p_j}\right)\right\rceil \prod_{i = r}^{i_m}\limits\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right]+t,\\ \end{array}\ ] ] where @xmath302 .
it means that the effect of the operator @xmath303 $ ] when operating on @xmath304 is that dividing @xmath305 into two effective sets @xmath306 and @xmath307 . from equation ( [ equ : d32 ] ) , ( [ equ : d33 ] ) , ( [ equ : l1 ] ) , ( [ equ : l3 ] ) , and ( [ equ : l5 t ] ) , we have @xmath308 \left[1-\frac{1}{p_j}-\frac{\widetilde{1}}{p_j}\right]\\ = m\prod_{i = r}^{i_m}\limits\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right ] \left[1-\frac{1}{p_j}-\frac{\widetilde{1}}{p_j}\right ] \left[1-\frac{1}{p_{r-1}}-\frac{\widetilde{1}}{p_{r-1}}\right]\\ = \left ( \left\lceil m\left(1-\frac{3}{p_j}\right)\right\rceil \prod_{i = r}^{i_m}\limits\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right]+t \right ) \left[1-\frac{1'}{p_{r-1}}-\frac{1''}{p_{r-1}}\right]\\ \geq \left(\left\lceil m\left(1-\frac{3}{p_j}\right)\right\rceil \prod_{i = r}^{i_m}\limits\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right]\right ) \left[1-\frac{1'}{p_{r-1}}-\frac{1''}{p_{r-1}}\right]\\ = \left\lceil m\left(1-\frac{3}{p_j}\right)\right\rceil \prod_{i = r-1}^{i_m}\limits\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right].\\ \end{array}\ ] ] or @xmath309
\left[1-\frac{1}{p_j}-\frac{\widetilde{1}}{p_j}\right ] \left[1-\frac{1}{p_{r-1}}-\frac{\widetilde{1}}{p_{r-1}}\right]\\ = m\prod_{i = r}^{i_m}\limits\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right ] \left[1-\frac{1}{p_j}-\frac{\widetilde{1}}{p_j}\right ] -m\prod_{i = r}^{i_m}\limits\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right ] \left[1-\frac{1}{p_j}-\frac{\widetilde{1}}{p_j}\right ] \left[\frac{1}{p_{r-1}}\right]\\ { \ \ } -m\prod_{i = r}^{i_m}\limits\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right ] \left[1-\frac{1}{p_j}-\frac{\widetilde{1}}{p_j}\right ] \left[\frac{\widetilde{1}}{p_{r-1}}\right]\\ = \left\lceil m\left(1-\frac{3}{p_j}\right)\right\rceil \prod_{i = r}^{i_m}\limits\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right ] + t\\ { \ \ } -\left\lceil m\left(1-\frac{3}{p_j}\right)\right\rceil \prod_{i = r}^{i_m}\limits\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right ] \left[\frac{1}{p_{r-1}}\right ] -t\left[\frac{1'}{p_{r-1}}\right]\\ { \ \ } -\left\lceil m\left(1-\frac{3}{p_j}\right)\right\rceil \prod_{i = r}^{i_m}\limits\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right ] \left[\frac{\widetilde{1}}{p_{r-1}}\right ] -t\left[\frac{1''}{p_{r-1}}\right]\\ = \left\lceil m\left(1-\frac{3}{p_j}\right)\right\rceil \prod_{i = r}^{i_m}\limits\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right ] \left[1-\frac{1}{p_{r-1}}-\frac{\widetilde{1}}{p_{r-1}}\right ] + t\left[1-\frac{1'}{p_{r-1}}-\frac{1''}{p_{r-1}}\right]\\
\geq \left\lceil m\left(1-\frac{3}{p_j}\right)\right\rceil \prod_{i = r-1}^{i_m}\limits\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right ] .
\end{array}\ ] ] in fact , if @xmath310 \left[1-\frac{1}{p_j}-\frac{\widetilde{1}}{p_j}\right ] < \left\lceil m\left(1-\frac{3}{p_j}\right)\right\rceil \prod_{i=1}^{i_m}\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right ] , \end{array}\ ] ] then for any @xmath311 , before deleting the multiples of other primes , it must have + @xmath312 \left[1-\frac{1}{p_j}-\frac{\widetilde{1}}{p_j}\right ] < \left\lceil m\left(1-\frac{3}{p_j}\right)\right\rceil \left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right ] .
\end{array}\ ] ] which contradicts lemma [ lem:3 ] .
so this lemma is true .
if @xmath313 , then @xmath314\prod_{i = r}^{i_m}\limits\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right ] \left[1-\frac{1}{p_j}-\frac{\widetilde{1}}{p_j}\right]\\ \geq \left\lceil m\left(1-\frac{3}{p_j}\right)\right\rceil \left[1-\frac{1}{p_{r-1}}\right ] \prod_{i = r}^{i_m}\limits\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right].\\ \end{array}\ ] ] with lemma [ lem:3 ] , the operator @xmath303 $ ] can be represented by @xmath315 , and other operator @xmath51 $ ] can operate on this inequality unchanged .
let @xmath316 , then for all @xmath317 , @xmath318 $ ] is effective to a natural sequence @xmath319 whose number is not less than @xmath320 .
the reason is as follows .
when a natural sequence is deleted by the items of @xmath321 , the sequence is subtracted by @xmath322+\left[\frac{m+4{{\bf mod \/}}p_j}{p_j}\right]$ ] .
we can arrange the m items in a table of @xmath282 rows ( table 1 ) .
@xmath322 $ ] will delete the @xmath323 row , and @xmath324 $ ] will delete the @xmath325 row .
thus there are @xmath326 rows left in which each item @xmath327 .
[ table1 ] @xmath328 but every @xmath10 items ( @xmath329 ) in any row of the first @xmath330 columns consist in a complete systems of residues modulo @xmath10 , because @xmath331 and @xmath332 are both complete system of residues modulo @xmath10 , where @xmath333 is an any ( row or column ) constant .
there are @xmath326 such rows or @xmath334 items left .
these items are effective to a nature sequence when deleting the items of @xmath335 .
@xmath336 \left[1-\frac{1}{p_j}-\frac{\widetilde{1}}{p_j}\right ] = sp_i(p_j-2)\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right]=s(p_i-2)(p_j-2).$ ] let @xmath160 , @xmath337 , @xmath338 .
@xmath339 $ ] will delete at most @xmath340 items .
if we add these items by removing those from the end of sequence then the sequence is again effective to a nature sequence , the sequence left has at least , @xmath341-[a+(a+1)]\\ & \geq sp_i(p_j-3)+t+sp_i-4a-2 \\ & = sp_i(p_j-3)-3\left\lfloor\frac{t}{p_j}\right\rfloor+t+(sp_i - a-2)\\ & \geq sp_ip_j(1-\frac{3}{p_j})+\left\lceil t-\frac{3t}{p_j}\right\rceil+(sp_i - a-2)\\ & = \left\lceil(sp_ip_j+t)(1-\frac{3}{p_j})\right\rceil+(sp_i - a-2)\\ \\ & = \left\lceil m(1-\frac{3}{p_j})\right\rceil+(sp_i - a-2 ) .
\end{array}\ ] ] for @xmath342 or @xmath343 , we have @xmath344 , @xmath345 . for @xmath346 and @xmath347 ,
the items of t have @xmath282 rows , @xmath348 columns and some @xmath349 items . in each of the first @xmath349 rows , there are exact @xmath10 items which consist in a complete system of residues modulo @xmath10 , and these items can be considered as an effective nature sequence when deleting the multiples of @xmath10 ( @xmath335 ) .
the other items have at most @xmath282 rows and @xmath348 columns where the multiples of @xmath282 have at most @xmath350 . as
before , we can add these items to make the @xmath351 as an effective nature sequence , therefore , @xmath352 - 2(p_i-1)\\ & \geq \left\lceil m(1-\frac{3}{p_j})\right\rceil+(sp_i - a-1)\geq \left\lceil m(1-\frac{3}{p_j})\right\rceil . \end{array}\ ] ] thus for any @xmath353 , the original sequence of @xmath155 , after deleted the items of @xmath321 from @xmath24 , is effective to reconstruct a new nature sequence having at least @xmath354 items .
the analysis above is for @xmath204 . for @xmath212
, we can reach at the same conclusion .
@xmath355 , @xmath356 , @xmath357 . for @xmath212 , after deleted the item of @xmath358 , it becomes + @xmath359 .
we can rearrange these items as @xmath360 .
the first @xmath361 items can be taken as an effective nature sequence ( @xmath362 ) from the original one when deleting the items of @xmath363 .
the other sequence @xmath364 , having at least zero item when deleting the multiples of all primes , will be neglected in further process .
for a given @xmath4 , consider the possible cousin primes of @xmath365 $ ] in set @xmath366 , @xmath367 .
@xmath376\prod_{i=2}^v\limits\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right]\\ & \geq \left\lceil n\left(1-\frac{3}{p_v}\right)\right\rceil\left[1-\frac{1}{2}\right ] \prod_{i=2}^{v-1}\limits\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right]\\ & \geq \left\lceil\left\lceil n\left(1-\frac{3}{p_v}\right)\right\rceil \left(1-\frac{3}{p_{v-1}}\right)\right\rceil \left[1-\frac{1}{2}\right ] \prod_{i=2}^{v-2}\limits\left[1-\frac{1}{p_i}-\frac{\widetilde{1}}{p_i}\right]\\ & \geq \cdots\\ & \geq \left\lceil n\prod_{i=3}^{v}\limits\left(1-\frac{3}{p_i}\right)\right\rceil \left[1-\frac{1}{2}\right ] \left[1-\frac{1}{p_2}-\frac{\widetilde{1}}{p_2}\right]\\ & \geq \
\left\lfloor\frac{n}{6}\prod\limits_{i=3}^{v}\left(1-\frac{3}{p_i}\right)\right\rfloor , \end{array}\ ] ] therefore , @xmath398 , and @xmath399 so that @xmath400 .
it contradicts the supposition of eq .
( [ equ : tp1 ] ) .
therefore ` there are infinitely many pairs of cousin prime ' . from eq .
( [ equ : tp3 ] ) and ( [ equ : tp4 ] ) , @xmath401 approaches infinity as n grows without bound .
the proof is completed .
figure [ fig : fig1 ] shows the actual pairs of cousin prime @xmath404 ( solid line ) in the range of @xmath405 $ ] , and its simplified formula @xmath406(dashed line ) from eq .
( [ equ : f1 ] ) .
it clearly shows that @xmath403 , and @xmath407 has no up bound for enough large @xmath4 . | we proved that there are infinitely many cousin primes . |
the search for the sm higgs at lep2 is well understood .
the reach depends on the centre of mass energy which can be achieved , and the integrated luminosity obtained . for 150
pb@xmath0 per experiment at @xmath3 gev , higgs masses @xmath4 gev will be ruled out@xcite .
the lep energy may be pushed a few gev higher , but to go much beyond this will require the use of the tevatron .
much interest in the tevatron s potential for higgs searches was sparked by work carried out during the run ii susy / higgs workshop in 1998@xcite .
the higgs cross section at the tevatron is large ( 1 pb for @xmath5 gev ) but since the @xmath6 process dominates , there is a huge qcd background . for higgs masses below 130140
gev , the best potential seems to come from the processes where a @xmath7 or @xmath8 is produced in association with the higgs : * @xmath9 : backgrounds from @xmath10 , @xmath11 , @xmath12 , single top .
a factor of 1.3 improvement in signal to background has been demonstrated in this channel by using a neural network compared to standard cuts .
it is also possible that additional gains may be had if the angular distributions ( @xmath13 , spin zero , vs. @xmath14 , spin one ) can be exploited .
@xmath15 is overwhelmed by the qcd background .
* @xmath16 : backgrounds from @xmath17 , @xmath18 , @xmath12 .
* @xmath19 : backgrounds from qcd , @xmath17 , @xmath18 , @xmath12 .
this requires a relatively soft missing @xmath20 trigger ( 35 gev ) but is powerful because of the large @xmath21 branching ratio . for higgs masses above 130140 gev :
* @xmath22 : backgrounds from drell - yan , @xmath23 , @xmath11 , @xmath18 , @xmath12 , @xmath24 .
the initial signal to background ratio is 1:140 , so many , rather finely tuned , selections are required , culminating in angular cuts to separate the signal from the `` irreducible '' @xmath23 background . combining all these search channels , and assuming a 10% systematic error on the backgrounds , one finds the sensitivity shown in fig .
[ fig : higgs ] . with 2 fb@xmath0 of data , only a modest extension of the lep reach
is possible . with 15 fb@xmath0 ,
on the other hand , it should be possible to either exclude , or observe at the 35@xmath25 level , a sm higgs up to @xmath26 gev .
this is an exciting prospect .
is it credible ?
in my view , yes : it is an exercise similar in scale to the top discovery , with a similar number of backgrounds , and requiring a similar level of detector understanding
. it will be harder the irredicible signal to background ratio is worse but it has caught the imagination of the experimenters .
one problem with the studies so far ( in my opinion ) is the @xmath27 mass resolution .
can the assumed resolution really be attained in a high luminosity environment ?
the resolution not only affects the mass bins over which a higgs signal would be smeared , but a detailed understanding of the shape of the @xmath27 mass spectrum will be needed to separate a putative higgs signal from the nearby mass peak due to @xmath28 decays ( see fig . [
fig : higgs ] ) . for this reason
, calibration samples of @xmath28 events will be very important .
tens of thousands of events ( after cuts ) are expected for 2 fb@xmath0 , using a displaced vertex trigger , in both cdf and d . [ cols="^,^ " , ]
even the minimal spectrum of supersymmetry ( the sm particles plus two higgs doublets and their superpartners ) can have many faces .
firstly , whether or not @xmath29 parity is conserved determines whether the lightest supersymmetric particle ( lsp ) will or will not be stable .
secondly , how supersymmetry is broken determines the mass heirarchy of states and hence their decays .
the typical benchmark , the supergravity - inspired or `` minimal sugra '' model , is determined by five parameters @xmath30 , @xmath31 , @xmath32 , @xmath33 and @xmath34 . in this picture ,
radiative electroweak symmetry breaking occurs naturally from the large top mass .
the lightest neutralino is the lsp ; the two lightest neutralinos , the lighest chargino and the lightest higgs @xmath35 are all `` light '' , while the other charginos and neutralinos , squarks , gluinos and the other higgs states are all `` heavy '' ( above a few hundred gev , perhaps ) . in gauge - mediated models , by contrast , the lsp is a gravitino , and there are signatures with photons and/or slow moving particles which may decay within or outside the detector .
recent anomaly - mediated models suggest that the lightest chargino and neutralino may be almost degenerate . at lep2
, the increased centre of mass energy will allow the present susy limits to be raised by 510 gev ( my guess ) with full use of data at and above 200 gev .
at the tevatron , the highest cross sections are for the pair production of colored particles like squarks and gluinos .
as long as @xmath29-parity is conserved , the signature will be jets plus missing transverse energy ( @xmath36 ) .
the run i d analysis@xcite required three jets ( one with @xmath37 gev ) and @xmath38 gev ; it excludes squark masses below about 250 gev and gluino masses below about 200 gev .
this search may be combined with complementary channels where one or more leptons are required , together with jets and @xmath36@xcite . with 2 fb@xmath0 of data
the reach will approximately be doubled , to gluino masses of order 400 gev@xcite .
for masses much larger than this , falling parton distributions kill the production rate very quickly . to extend the reach in parameter space , chargino /
neutralino production will become increasingly important .
present searches in the `` golden '' trilepton mode do not really constrain models , but with 2 fb@xmath0 of data , chargino masses up to 180 gev should be probed ( 150 gev if @xmath33 is large)@xcite .
this places an emphasis on low-@xmath39 lepton triggering , since one or more of the leptons tends to be soft .
it would also help a lot if @xmath40 modes could be included .
in many models , the stop and sbottom squarks are significantly lighter than the others .
it is therefore interesting to search for them separately .
the decay channels involve @xmath41 or @xmath42 jets , and for stop there is also the possibility of @xmath43 decays or vice versa ( depending on the masses ) .
present cdf limits@xcite explore @xmath44 gev and @xmath45 gev ; with 2 fb@xmath0 , the sensitivity should improve to about 200 gev@xcite .
much interest in gauge - mediated supersymmetry was sparked a few years ago by the observation of a single @xmath46 event at cdf@xcite .
this final state is consistent with selectron production in a gauge mediated scenario .
all we can now say is that searches for the expected related signatures have all proved negative : @xmath47 and @xmath48 at the tevatron@xcite , and one or more photons plus missing energy at lep@xcite .
this type of signature is just one example of what may be expected in gauge mediated models .
depending on the details of the model , the next - to - lightest superpartner or nlsp ( which then decays into the gravitino lsp ) may be a neutralino , a stau , or effectively more than one slepton state ( if they are almost degenerate in mass ) .
prompt decays of the nlsp will then give final states containing photons@xmath49 , taus , and multileptons , respectively .
searching for such signatures is relatively straightforward .
a more challenging possibility is that the nlsp has a finite decay length since this is determined by the susy breaking interactions it is totally unknown . for neutralino nlsp s , decay lengths @xmath50 more than a few metres will give standard @xmath36 signatures since the nlsp will escape the detector .
decay lengths of a few centimetres to a metre result in photons which do not point back to the primary vertex : searches at lep@xcite for such photons have excluded neutralino nlsp masses less than 85 gev , for @xmath51 m. in run ii at the tevatron , d will be able to make use of new preshower detectors upstream of the electromagnetic calorimeter to detect non - pointing photons .
the resolution in the distance of closest approach to the vertex will be 2.2 cm along the beamline and 1.4 cm radially@xcite .
charged nlsp s with long decay lengths appear as massive , slow - moving particles which exit the detector ( so called `` cannonballs '' ) .
lep limits exclude stau nlsp masses less than 76 gev , or slepton co - nlsp masses less than 85 gev@xcite . in run ii at the tevatron , @xmath52 and
timing information will be available ( using time of flight counters in cdf , and the muon system in d ) .
cdf expect sensitivity up to stau masses of 180 gev using the time of flight system@xcite .
short decay lengths @xmath53 cm will give reconstructable impact parameters in the vertex detectors ; 1 cm @xmath54 m is harder , especially to trigger .
( with enough data , a combination of impact parameter and cannonball searches may be sensitive enough to exclude this intermediate region as well . )
a general problem with understanding delayed decays is that event generators are not widely available , and the interface to the detector simulation is non - trivial .
recently there has been considerable interest in anomaly - mediated susy models .
the phenomenology of such models contains a light chargino which is almost degenerate with the lsp ( a neutralino ) .
such small mass differences may result in delayed chargino decays , with cannonball type signatures , or decays inside the detector @xmath55 : very challenging indeed .
there is also considerable interest in scenarios with large extra dimensions .
gravitons may propagate into the higher dimensional space .
searches have been carried out@xcite for @xmath56nothing , @xmath57nothing , and @xmath58jet@xmath59nothing .
indirect effects have been searched for in @xmath60 .
turning now to susy with @xmath29-parity violating decays , the usual assumption is that the @xmath29 violating couplings are small compared with the @xmath29 conserving ones .
the production processes and subsequent decay chain then proceed as in minimal sugra , but the final lsp s decay via baryon or lepton number violating interactions .
hence there is no @xmath36 from the escaping lsp , but there are usually multi - leptons from cascade decays and @xmath36 from neutrinos . both at lep@xcite and the tevatron@xcite ,
it is found ( perhaps surprisingly ) that the excluded regions end up very comparable to those for msugra with @xmath29 conserved .
@xmath29-violating couplings could also enter in the production of susy particles . at hera
, `` leptoquark '' searches have been interpreted in terms of @xmath61@xcite , while at lep , limits have been placed on processes like @xmath62@xcite . both at lep2 and the tevatron , we therefore have a basic menu of supersymmetry searches which is well - defined and we should have no trouble in exploring : * minimal sugra ; * gauge mediated supersymmetry with prompt photon signatures ; * some subset of @xmath29 violation .
our concerns are what we may have forgotten , especially at the tevatron where triggering is a crucial issue .
can we cover : * slow moving massive particles ; * gauge mediated supersymmetry with detached photons or taus ; * anomaly - mediated supersymmetry ( _ e.g. _ @xmath63 ) ; * general extra dimension signatures , _ etc ? _ a critical look at the proposed trigger lists is probably called for .
at lep2 , two complementary processes allow the neutral higgs states of supersymmetric models to be explored : @xmath64 and @xmath65 . combining these channels and the four experiments , one can exclude ( summer 1999@xcite ) @xmath66 gev and @xmath67 gev .
moreover , @xmath68 ( @xmath69 ) is excluded for maximal ( and for minimal ) stop mixing .
the @xmath33 exclusion is very sensitive to the top mass , and it should be noted that there is no excluded range in @xmath33 if @xmath70 gev .
general scans of minimal susy parameter space find some points with strange decay patterns that can evade the limits , but these are rather few .
invisible higgs decays have been included in the searches . in run ii of the tevatron
, rather stringent limits will be set on the susy higgs sector , given sufficient luminosity , since the whole allowed mass range for the lightest higgs @xmath35 ( @xmath71 gev ) is covered .
one may also exploit the enhanced couplings between the higgs states and @xmath41 quarks at large @xmath33 to search for @xmath72 . using present data and requiring three @xmath41-tagged jets ,
cdf have been able to explore the region of @xmath73 and above@xcite . with 10
fb@xmath0 the sensitivity will extend to @xmath74 for @xmath75 up to 150 gev .
supersymmetry also predicts the existence of charged higgs states .
searches for pair production at lep have excluded masses @xmath76 gev@xcite . in minimal susy models , it is expected that @xmath77 , and lep is not really sensitive to this region .
searches at the tevatron have concentrated on the production of @xmath78 in top decays@xcite in competition with the standard @xmath79 mode .
this has been carried out both as an `` appearance '' experiment ( looking for @xmath80 ) and as a `` disappearance '' experiment ( looking for fewer than expected @xmath81 ) .
present limits are sensitive only for @xmath82 and @xmath83 , but with 2 fb@xmath0 the sensitivity will extend to @xmath84 and 20 .
there is no fully worked - out scenario for electroweak symmetry breaking through a new strong interaction , but this does not mean that it is not possible .
indeed , many schemes have been outlined in some detail , and a straw - man technicolor model is now implemented in pythia@xcite . in general , dynamical symmetry breaking schemes like technicolor and topcolor predict : * new particles in the mass range 100 gev 1 tev , * with strong couplings and large cross sections , * decaying to vector bosons and ( preferentially third generation ? ) fermions .
recently l3 has reported searches for technicolor resonances at lep2@xcite , and several technicolor and topcolor searches have been made on present fermilab data .
all are negative so far . with run ii ,
the reach of the tevatron will be greatly extended . besides all of the above scenarios , one should always be looking ( at lep , hera and the tevatron ) for anything unexpected : * leptoquarks * fourth generation fermions , or isosinglet fermions * @xmath85 and @xmath86 * contact interactions or compositeness , _ etc . _
there are plenty of opportunities for us to find `` something new '' before the lhc starts operation .
the standard scenarios have been explored in some detail , but different supersymmetry breaking schemes can produce radically different signatures .
theoretical fashion moves fast , and event generators are not always available . we need to keep an open mind in our searches .
this is easy to say once one has the data in hand ; the challenge right now is to ensure that we do our best to trigger on all possible interesting things , especially in the hadron collider environment .
i would like to note my appreciation for the hospitality of st .
john s college , durham , and thank m. whalley , j. forshaw and e.w.n .
glover for all their organisational work .
99 g. quast [ for the lep electroweak working group ] , `` electroweak review at lep i , '' _ to be published in the proceedings of international europhysics conference on high - energy physics ( eps - hep 99 ) , tampere , finland , 1521 july 1999_. a.
savoy - navarro , supersymmetry searches at the tevatron in run i and run ii, fermilab - conf-99 - 281-e , _ to be published in the proceedings of international europhysics conference on high - energy physics ( eps - hep 99 ) , tampere , finland , 1521 july 1999 _ , and available at http://fnalpubs.fnal.gov/1999/conf/conf-99-281-e.html .
j. a. valls [ cdf collaboration ] , `` higgs searches at the tevatron in run i , '' fermilab - conf-99 - 263-e , _ to be published in the proceedings of international europhysics conference on high - energy physics ( eps - hep 99 ) , tampere , finland , 1521 july 1999_. c. rembser , `` search for susy with gravitino lsp at lep , '' _ to be published in the proceedings of international europhysics conference on high - energy physics ( eps - hep 99 ) , tampere , finland , 1521 july 1999_. f. ledroit , `` search for susy with r - parity violation at lep , '' _ to be published in the proceedings of international europhysics conference on high - energy physics ( eps - hep 99 ) , tampere , finland , 1521 july 1999_. b. abbott _ et .
[ d0 collaboration ] , phys.rev.lett .
* 82 * ( 1999 ) 4975 ; t. affolder _ et al . _
[ cdf collaboration ] , `` search for the charged higgs boson in the decays of top quark pairs in the @xmath90 and @xmath91 channels at @xmath92=1.8 tev , '' hep - ex/9912013 , _ submitted to phys .
l3 collaboration , `` search for technicolour production at lep , '' l3 note 2428 , _ submitted to the international europhysics conference on high energy physics ( eps - hep 99 ) , tampere , finland , 1521 july 1999 . | i shall briefly summarize the prospects for extending our understanding of physics beyond the standard model within the next five years .
fermilab - conf-00/001 in this necessarily brief presentation , i shall attempt to outline how we may be able to discover physics beyond the standard model within the next five years .
i shall interpret `` beyond the standard model '' to mean the physics of electroweak symmetry breaking , including the standard model higgs boson .
the nature of this tev - scale new physics is perhaps the most crucial question facing high - energy physics , but we should recall ( neutrino oscillations ! ) that there is ample evidence for interesting physics in the flavour sector too . in the next five years ,
before the lhc starts operations , the facilities available will be lep2 , hera and the fermilab tevatron .
i shall devote a bit more time to the tevatron as it is a new initiative for united kingdom institutions .
the tevatron schedule now calls for data taking in run ii , using two upgraded detectors , to begin on march 1 , 2001 , with 2 fb@xmath0 accumulated in the first two years .
a nine - month shutdown will follow , to allow new silicon detector layers to be installed , and then running will resume with a goal of accumulating 15 fb@xmath0 ( or more ) by 2006 .
where does the standard model stand , circa 2000 ?
we know that it works very well , indeed some precision observables test it at the 10@xmath1 level@xcite .
all observations are consistent with a single , light standard model ( sm ) higgs boson , though no such beast has yet been observed . as of autumn 1999@xcite ,
higgs masses less than 106 gev are excluded by direct searches and greater than 245 gev are excluded by fits to precision sm observables@xcite . despite this consistency with the sm , there are strong general arguments for the existence of new physics at the electroweak scale ( 250 gev1 tev ) .
the goodness of the sm fits may perhaps suggest that this new physics is weakly coupled .
in addition there are indirect pointers that may suggest supersymmetry ( susy ) is a prime candidate for the new physics , though once again , all direct searches for supersymmetric particles have so far proved negative .
data from lep2 have been used to rule out@xcite squarks ( stop and sbottom ) below 8090 gev , sleptons ( selectron , smuon and stau ) below 7090 gev , charginos below 7090 gev , and the lightest neutralino below 36 gev .
searches for squarks and gluinos , stop , sbottom , charginos and neutralinos at the tevatron have also all proved negative and yield comparable limits@xcite . your mission , should you choose to accept it , is then clear . at your earliest convenience
, please carry out one ( or more ! ) of the following challenges : * discover the standard model higgs ; * discover or exclude the lightest higgs of minimal susy ( with masses up to @xmath2 gev ) ; * discover one or more superpartners ; * exclude ( or at least disfavour ) supersymmetry at the tev scale by discovering some other new physics .
we know the lhc can address all of these questions given sufficient luminosity .
but can any of this be done sooner , _
i.e. _ in the next five years ? |
the ferromagnetic resonance ( fmr ) spectrum of thin films is known to evolve with the angular configuration of the applied magnetic field with respect to the film surface . in this evolution
the multi - peak fmr spectrum often becomes a single - peak one in a certain angular configuration , to regain its multi - peak character beyond it . a careful analysis of this effect , in which a multi - peak fmr spectrum collapses to a single - peak one , leads to the observation of two types of behavior of the collapsing spectrum .
in one type the intensity of all the peaks except the first one diminishes progressively to vanish completely in the collapse configuration . beyond this configuration angle
the satellite resonance lines emerge again in an unchanged order .
the other type of collapse involves resonance positions rather than intensities : in the collapse configuration the lines shift to the position of the main peak , to reemerge in a reversed order beyond this configuration .
the first type of collapse has been long known in the literature to be a surface effect due to the changes of the surface magnetic anisotropy ( responsible for the surface spin pinning ) with the configuration of the external field with respect to the surface of the thin film . in this pattern
a complete collapse of the resonance spectrum occurs in a specific angular configuration ( referred to as critical angle of the surface anisotropy ) in which the surface anisotropy has no effect on the surface spin pinning , and the resonance precession of the all spins is homogeneous throughout the sample .
in contrast , the other type of collapse is a bulk effect that occurs as a result of the ferromagnetic resonance in separate regions of slightly different magnetic character in a heterogeneous thin - film sample .
a heterogeneous magnetic structure is known to result in heterogeneous conditions of ferromagnetic resonance in the sample .
thus , in this interpretation each line in the multi - peak resonance spectrum corresponds to a resonance in a different region of the sample , and the collapse configuration of the external field is a particular configuration in which all the magnetically different regions participate in the resonance for resonance fields that differ only negligibly .
the collapse of the fmr spectrum in this specific angular configuration of the external field in which the resonance heterogeneity of the sample is eliminated has not been thoroughly analyzed in the literature so far .
the aim of this paper is to elucidate in detail the theoretical grounds of this effect .
the paper is organized as follows . in section
ii we recall the derivation of the universal smit - beljers - suhl formula expressing the condition for ferromagnetic resonance to occur in a homogeneous bulk ferromagnetic sample . on the basis of this formula , in section
iii we derive the configuration condition of resonance in thin films . in section
iv we use this condition for analyzing the configuration evolution of the resonance field in thin - film samples with a perpendicular uniaxial anisotropy . we formulate a universal condition for a multi - peak fmr spectrum to collapse into a single - peak one in such samples , and ( in section v )
demonstrate the bulk character of this effect .
let us consider the dynamics of a body with an angular momentum @xmath1 and a magnetic moment @xmath2 collinear to it ( @xmath3 is the gyromagnetic ratio ) in a magnetic field @xmath4 .
the field acts on the dipole to produce a torque @xmath5 that sets the body in motion according to the equation : @xmath6 let us represent the vectors in the above equation in the local basis of orthonormal spherical vectors @xmath7 determined by the vector @xmath8 ( obviously , @xmath9 , see fig.[fig : fig1 ] and comments in @xcite ) .
related to vector @xmath10.,width=151 ] thus : @xmath11 now , let us determine the components @xmath12 and @xmath13 of the magnetic field @xmath4 by considering the change in the energy of the dipole with an infinitesimal rotation ( we only consider rotation by the polar angle @xmath14 ) .
the energy of a magnetic moment @xmath10 in a magnetic field @xmath4 is expressed by the equation : @xmath15 which , represented in the basis @xmath7 , becomes : @xmath16 as a result of the rotation of @xmath10 by a small angle @xmath17 ( @xmath18 ) the energy changes to : @xmath19 the vector @xmath20 can be expressed as @xmath21 , so the above equation will become : @xmath22 the following relations are seen to occur : @xmath23 @xmath24 now , we can get back to the equation of motion and write the cross product as the determinant : @xmath25 = \gamma \left| \begin{matrix } \hat{\bm{r}}&\hat{\bm{\theta}}&\hat{\bm{\phi}}\\ m&0&0\\ h_r&\left.\frac{-1}{m}{\frac{\partiale}{\partial\theta}}\right|_{\theta,\phi}&\left.\frac{-1}{m\sin\theta}{\frac{\partiale}{\partial\phi}}\right|_{\theta,\phi}\\ \end{matrix } \right| .\ ] ] hence we obtain the system of equations : @xmath26 this system of equations is to be solved in four steps .
we ( i ) determine the equilibrium angles @xmath27 and @xmath28 as the solution of the system of equations @xmath29 ; ( ii ) expand @xmath30 and @xmath31 into a taylor series at @xmath27 and @xmath28 , respectively ; ( iii ) assume that the time dependence of the angles has the form @xmath32 , and ( iv ) exclude all the terms except the lowest - order ones ( see @xcite ) .
this procedure leads to the smit - beljers - suhl formula @xcite that represents the ferromagnetic resonance condition : @xmath33 where the values of the derivatives correspond to the equilibrium angles @xmath27 and @xmath28 .
( note by the way that the radicand is the hessian of the function @xmath30 at the equilibrium point ; its positive value is indicative of the existence of a minimum of the function at the point at which the first partial derivatives vanish . )
since we shall henceforth consider samples with a magnetization @xmath34 , the formulas derived above , with @xmath34 in place of @xmath10 , will apply to the description of the magnetization dynamics in the studied case .
let us consider a sample in which the free energy density @xmath30 is the sum of the zeeman energy , the demagnetization energy and the uniaxial anisotropy energy : @xmath35 where @xmath36 is a unit vector normal to the surface of the sample , and @xmath37 is a unit vector oriented along the easy magnetization axis .
the applied magnetic field and the magnetization of the sample will be henceforth represented as ( see fig.[fig : fig2]a ) : and the applied magnetic field @xmath4 with respect to the film surface.,width=377 ] @xmath38 we shall consider three special cases corresponding to three different choices of the easy axis ( @xmath39,[010]$ ] or @xmath40 $ ] , see fig.[fig : fig2]b ) . in each case
we shall investigate the angular configuration dependence @xmath41 implied by the assumed form of the free energy density .
in this case the unit vector @xmath36 normal to the surface of the sample and the unit vector @xmath37 oriented along the easy axis are identical : @xmath43.\ ] ] thus , the free energy density reads : @xmath44 let us determine the equilibrium conditions of the system . the first partial derivatives with respect to the angles @xmath14 and @xmath45 have the form : @xmath46 from the condition of vanishing of these derivatives we obtain the equations that determine the coordinates @xmath47 and @xmath48 of the equilibrium point ( fig.[fig : fig2]b ) : [ equi1 ] @xmath49 where @xmath50 is the effective magnetization , defined as : @xmath51 in this case the smit - beljers - suhl formula becomes : @xmath52 \left[h\cos(\theta_m-\theta_h)-4\pi m_{\text{eff}}\cos2\theta_m\right].\ ] ] in order to plot the corresponding configuration dependence @xmath41 of the resonance field we must solve the system of two nonlinear equations and .
in this case the unit vector @xmath36 normal to the surface and the unit vector @xmath37 oriented along the anisotropy direction are : @xmath54 \quad\text{and}\quad \bm{u } = [ 010].\ ] ] the free energy density is : @xmath55 the first partial derivatives with respect to the angles @xmath14 and @xmath45 read : @xmath56 thus , the equilibrium direction of magnetization is determined by the equations : @xmath57 the smit - beljers - suhl formula for this case is : @xmath58 \left[h\frac{\sin\theta_h}{\sin\theta_m}+\frac{2k}{m}\right],\ ] ] or , equivalently : @xmath59\\\times \left[h\cos(\theta_m-\theta_h)-\left(4\pi m + \frac{2k}{m}\right)\cos^2\theta_m+\frac{2k}{m}\right].\end{gathered}\ ] ] note that this condition does not involve the azimuth angle @xmath45 , since the equilibrium condition requires again that @xmath60 . in this case
the unit vector @xmath36 normal to the sample surface and the unit vector @xmath37 parallel to the anisotropy direction are : @xmath54 \quad\text{and}\quad \bm{u } = [ 100].\ ] ] the free energy density reads : @xmath62 the first partial derivatives of the energy density with respect to the angles @xmath14 and @xmath45 have the form : @xmath63 thus , the equilibrium direction of magnetization is determined by the angles @xmath47 and @xmath48 that fulfill the equations : @xmath64 the smit - beljers - suhl formula for this case has the form : @xmath65\\ \times \left[\frac{h\sin\phi_m\sin\theta_h}{\sin\theta_m}+\frac{2k}{m}\cos2\phi_m\right ] -\left(\frac{h\cos\phi_m\cos\theta_m\sin\theta_h}{\sin\theta_m}-\frac{2k}{m}\sin2\phi_m\cos\theta_m\right)^2\end{gathered}\ ] ] or , equivalently : @xmath66 \\\times \left[h(\sin\phi_m\sin\theta_m\sin\theta_h+\cos\theta_m\cos\theta_h)- \left(4\pi m + \frac{2k}{m}\cos^2\phi_m\right)\cos^2\theta_m + \frac{2k}{m}\cos2\phi_m\right ] \\-\left(\frac{2k}{m}\right)^2\sin^2\phi_m\cos^2\phi_m\cos^2\theta_m.\end{gathered}\ ] ]
in this section we shall only consider the simplest of the three cases mentioned above , i.e. a thin film with perpendicular uniaxial anisotropy ( case [ 001 ] ) ; we leave the remaining two cases to be similarly discussed in a separate paper .
the condition allows to determine the equilibrium angle @xmath47 of the magnetization vector corresponding to a specific configuration of the external field @xmath67 .
figure [ fig : ferromagnetic_resonance_a_angle ] shows the consequent configuration dependence @xmath68 for two values of @xmath69 . the deviation of the magnetization angle @xmath47 from the field angle @xmath67 is seen to be the largest in the middle of the interval @xmath70 $ ] , and to grow with increasing value of @xmath69 ; in this range the deviation is of ca .
@xmath71 to @xmath72 , far from being negligible !
now , let us assume that the thin film under consideration is heterogeneous and planarly stratified into regions with two different values of effective magnetization @xmath69 .
the resonance condition with the equilibrium condition used leads to two respective configuration resonance curves @xmath41 , see fig.[fig : ferromagnetic_resonance_a_field+dhdm]a .
this means that for a configuration @xmath67 each of the ferromagnetic strata resonates for a different value of the external field .
this applies to the whole range of @xmath67 with the exception of one particular configuration corresponding to the point of intersection of the two curves . in this particular configuration both regions
in spite of their significantly different magnetic parameters !
produce a resonance simultaneously for the same magnitude of the external field .
this implies homogeneous precession throughout the sample , as if its heterogeneity were completely eliminated in this configuration !
thus , for this particular angle @xmath67 the sample can be regarded as _ dynamically homogeneous_. the dynamic homogeneity ( dh ) is evidenced by the collapse of the two - peak resonance spectrum to a single resonance line .
note that past this particular angle the two peaks reemerge , but the resonance sequence is reversed .
the occurrence of a _ dynamic homogeneity angle _ ( dha ) is an intrinsic characteristic of the smit - beljers - suhl ( sbs ) resonance formula .
thus , we have grounds to ask whether the sbs equation can provide the basis for an analytical condition that would allow to predict the dha angle for a ferromagnetic thin - film sample .
we think we have managed to formulate such a condition a posteriori on the basis of the numerical results shown in fig.[fig : ferromagnetic_resonance_a_field+dhdm]b , presenting the configuration dependence of the derivative @xmath73 determined for each of the ferromagnetic regions shown in fig.[fig : ferromagnetic_resonance_a_field+dhdm]a . the dha angle is seen to be in the range delimited by the extreme configurations for which the considered derivative vanishes in each stratum .
the range shrinks with decreasing difference between the extreme values of @xmath69 in the two regions .
thus , we can expect that the dha angle will be determined with a good approximation by the condition of vanishing of the derivative : @xmath74 in the region the magnetic properties of which predominate in the sample .
it is worthy of notice that the above - discussed collapse of the fmr spectrum in a particular field configuration defined by the dha angle is a dynamic * volume * effect , since only bulk quantities figure in the adopted expression for free energy density , and the presence of the surface is only manifested in the shape anisotropy of the sample .
thus , the energy of surface anisotropy was not taken into account in our considerations .
this is an important statement , since also the surface anisotropy by itself can cause a multipeak swr spectrum to collapse to a single resonance line in a certain configuration , referred to as _ critical angle _
@xcite , of the applied field with respect to the film surface .
however , there is a substantial difference between the collapse of an fmr spectrum and that of an swr spectrum . in the former
the resonance line positions merge to form a single line at the dha angle , while the collapse of an swr spectrum consists in the suppression of the intensity of all the lines except the main one at the critical angle .
this is due to the different nature of the two effects : the essence of the fmr collapse is the attainment of homogeneous resonance dynamics throughout the _ volume _ of the sample , whereas the swr effect consists in the elimination , in the _ critical _ field configuration , of the impact of the surface anisotropy on the spin dynamics in the thin film .
therefore , we shall refer to the latter configuration as _ surface critical angle _
the sca is sensitive to the conditions on the surface of the sample and can vary with them , while the dha , only determined by the bulk characteristics that enter the condition , is a fixed parameter of the sample .
it should be emphasized that the above - mentioned differences between the two effects can be used in practice as a criterion that allows to determine whether a collapse of a multipeak fmr spectrum observed in a thin film is of surface or bulk character .
this study was supported by the polish ministry of science and higher education , grant no .
n n202 1945 33 . vs. external field angle @xmath67 determined for a thin film with perpendicular uniaxial anisotropy ( f=33.5 ghz ) ; two curves corresponding to two effective magnetization values are plotted for comparison.,width=377 ] and ( b ) its derivative @xmath75 in a thin film with perpendicular uniaxial anisotropy .
( a ) note the intersection of the resonance curves plotted for the two regions of different effective magnetization magnitude .
the intersection point corresponds to the particular field configuration in which the sample produces a homogeneous resonance for the same field magnitude throughout its volume .
( b ) note that the homogeneous resonance configuration lies in the range determined by the vanishing of the derivative @xmath75 in each resonating region.,width=302 ] 90 david j. griffiths , _ introduction to electrodynamics _
( prentice - hall , new jersey , 1999 ) pp .
38 - 43 allan h. morrish , _ physical principles of magnetism _ ( john wiley and sons , new york , 1965 ) j. smit , h. g. beljers , philips res . repts . *
10 * ( 1955 ) 113 h. suhl , physical review * 97 * ( 1955 ) 555 h. puszkarski , progress in surface science * 9 * ( 1979 ) 191 | we demonstrate that a multi - peak fmr spectrum , with lines corresponding to resonance in different ferromagnetic regions of a heterogeneous thin - film sample , can collapse to a single - peak spectrum if there exists a particular field configuration , or the configuration of the external magnetic field with respect to the film surface , in which @xmath0 within the region magnetically dominating in the sample . |
scanning tunneling microscopy ( stm ) has become one of the most basic tools for the manipulation of matter at the atomic scale .
although this experimental technique has reached maturity , the detailed theoretical understanding of experimental data is still incomplete and/or contradictory .
one of the most famous examples of atomic manipulation is associated with the surface kondo effect observed when transition metal ions ( like co ) are placed on a metallic surface ( such as cu @xmath0 ) @xcite .
the surface kondo effect is the basis for the observation of surprising phenomena like quantum mirages @xcite , and has attracted a lot of attention and interest in the last few years .
early interpretations of these observations were based on the assumption that only surface states of cu @xmath0 are involved in the scattering of electron waves by the co adatoms @xcite .
however , later experiments with co atoms on the cu ( 100 ) surface ( that does not have any surface state ) @xcite , or in cu ( 111 ) but close to atomic surface steps ( that affect the surface states ) @xcite have indicated that bulk rather than surface states are responsible for the kondo effect in these situations .
the latter can be important for fine tuning of surface electronic structure , with potential applications to nanotechnology .
a recent study of cocu@xmath1 clusters on cu ( 111 ) demonstrated this tunablilty by atomic manipulation and showed that each atom in the vicinity of the magnetic impurity matters for determining the kondo effect @xcite .
moreover , the relevance of the kondo effect for the electronic structure of metal surfaces themselves was demonstrated by the discovery of a sharp density of states peak on the cr @xmath2 surface and its possible interpretation as an orbital kondo resonance @xcite . at the same time , when calculating the kondo temperatures for real electronic structures a mapping onto one - orbital anderson impurity model @xcite was used .
the realistic atomic geometry of kondo systems plays a crucial role in complex electronic properies @xcite and it is , a priori , not obvious that a one - orbital anderson impurity approach is sufficient : even the two - orbital anderson model demonstrates kondo physics essentially different from the single - orbital one @xcite .
a recent theoretical investigation of fe impurities in gold and silver showed that the proper kondo model corresponds to a s=3/2 spin state@xcite .
a realistic , multi - band consideration of correlation effects in specific solids is possible in the framework of the local density approximation + dynamical mean - field theory ( lda+dmft ) approach ( for review , see ref . ) .
however , formally accurate quantum monte carlo ( qmc ) calculations are always done with taking into account only the diagonal part of coulomb interaction @xcite , even with realistic hybridization functions obtained in the lda .
this approximation is , strictly speaking , uncontrollable . at the same time , approximate schemes working with the complete coulomb interaction matrix , such as the perturbative scheme @xcite which is frequently used to calculate electronic structure of transition metals and alloys @xcite are not sufficient to reproduce so subtle correlation features like the kondo effect , properly . as for the exact diagonalization @xcite or numerical renormalization group @xcite methods they are hardly applicable , due to computational problems , for more than two orbitals per impurity .
the recent progress in continuous time qmc scheme ( ct - qmc ) @xcite makes it perspective to treat the complicated kondo systems @xcite . here
we will apply this method to calculate kondo temperatures as well as spectral functions for the case of a co impurity in bulk cu , in a cu @xmath0 surface and on top of a cu @xmath0 surface .
in contrast with all previous calculations we will work with an accurate complete coulomb interaction @xmath3-matrix for correlated @xmath4 orbitals .
the latter can be calculated from first principles in a parameter - free way by the gw technique @xcite so this approach is completely _
ab initio_. moreover , the ct - qmc method allows to work , without any essential difficulties , even with the rigorous frequency - dependent @xmath3-matrix . as the first step , we present calculations for the static @xmath3-matrix , but this restriction is purely technical and can be relatively easily removed in the future , with a growth of available computer resources .
the multi - orbital impurity problem with a general @xmath3-matrix is described by the effective action @xmath5 where @xmath6 are orbital indices , and @xmath7 are spin indices , @xmath8 is the local non - interacting green function for correlated orbitals obtained from the density functional theory ( dft ) with the help of optimal projection operator to the impurity d - states : @xmath9 here @xmath10 is the energy spectrum and @xmath11 is the corresponding wave function of our system ( metal host with magnetic impurity ) , described by @xmath12 localized orbitals , and @xmath13 is the coulomb interaction matrix element : @xmath14 here @xmath15 is local orthogonal wave function for correlated orbitals and @xmath16 is screened spin - independent coulomb interaction between electrons at the coordinates @xmath17 and @xmath18 .
we used standard quasiatomic lda+u parametrization of coulomb matrix for d - electron via effective slater integarls or average coulomb parameter u and exchange parameter j as described in ref .. we choose the orbital basis related to spherical harmonics to be sure that magnetic orbital quantum numbers in @xmath13 matrix are satisfied the following sum rule : @xmath19 . in this case
we will get rid of so - called three - site terms like @xmath20 with @xmath21 which turns out to result in a strong sign problem in qmc calculations with real spherical harmonics . following the general ct - qmc scheme @xcite we expand the partition function around the gaussian part of our multiorbital action eq.([sinit ] ) which gives the fermionic determinant over the non - interacting green functions with the rank @xmath22 : @xmath23 in order to minimize the number of different interaction vertices we group different matrix elements of the multiorbital coulomb interactions which have a similar structure of fermionic operators .
since the @xmath13 matrix elements are spin independent , one should look over all possible combinations of orbital and spin indices , to generate all terms for the interaction in the action eq.([sinit ] ) .
some combinations can violate the pauli principle and should be removed .
for ct - qmc algorithm it is useful to represent the interaction hamiltonian in the following form : @xmath24 .
the interaction terms can be transformed to the desired form , depending on relations between spin and orbital indices : \(i ) if @xmath25 , we can just commute @xmath26 and @xmath27 and then @xmath26 and @xmath28 .
another combination of indices , that allows the same commutation , is the following : @xmath29 , @xmath30 and @xmath31 ( the latter two are following from the pauli principle ) , and also @xmath32 .
these terms we can transform to the following desirable representation : @xmath33 ( ii ) in the case when @xmath29 and @xmath34 we can commute @xmath27 and @xmath35 , since in this case @xmath30 and @xmath31 due to the pauli principle : @xmath36 after generating all this terms it is useful to collect and _ symmetrize _ all the terms with identical and equivalent ( i.e. @xmath37 and @xmath38 ) quantum numbers . in order to reduce the fermionic sign problem we introduce additional parameters , @xmath39 , to optimize the splitting of the gaussian and interaction parts of the action eq.([sinit ] ) @xmath40 one can see , that the first item in ( [ s0 m ] ) on matsubara frequences corresponds to bare green s function @xmath41 where @xmath42 is the hybridization matrix .
the second term is just a constant which we can absorb to the new chemical potential @xmath43 .
therefore we can rewrite the bare green function in the following matrix form : @xmath44 the optimal choice of parameters @xmath45 would lead to effective reduction of interaction terms in the action eq .
( [ s0 m ] ) and therefore minimization of average perturbation order in eq . ( [ zexp ] ) .
note that relation between @xmath46 and @xmath47 can be represented from eq .
( [ s0 m ] ) in the following spin and orbital matrix form : @xmath48 here we used the fact that @xmath49 following from the definition of the coulomb matrix elements ( [ umat ] ) .
we also need to minimize the fermionic sign problem which finally leads us to such expression for diagonal alpha parameters @xmath50 corresponding to to the following interaction fields @xmath51 .
the @xmath52 has to be found iteratively in order to get a proper occupation number of correlated electrons . in the case of half - filled one - band hubbard model
@xmath53 leads to the correct chemical potential shift of the @xmath54 and average @xmath55 which corresponds to the hartree - fock substraction . for non - diagonal alpha s which correspond to the fields of general form @xmath56 , where @xmath21 and @xmath57 we find the following optimal condition : @xmath58 since we symmetrize the interaction @xmath3 matrix it is necessary to extend the definition of the @xmath59 matrix to keep all the terms in the interaction part of initial action ( the last item in eq .
( [ s0 m ] ) ) .
it can be done in the following way @xcite : for every @xmath13 field in @xmath60 of updates we deliver the @xmath39 parameters as @xmath61 , @xmath62 , and in another @xmath60 as @xmath63 , @xmath64 for the case of @xmath65 and @xmath66 . for non - diagonal fields ,
i.e. @xmath21 and @xmath57 @xmath67 , @xmath68 , with @xmath60 probablility and @xmath69 , @xmath70 otherwise .
it was found that the sign problem is eliminated in the case when @xmath71 and @xmath72 for occupancy @xmath73 per state and @xmath74 , @xmath75 otherwise .
the optimal choice of @xmath76 parameter is few percent of @xmath77 to keep minimal average perturbation order .
another problem is a proper choice of non - diagonal @xmath78 parameter .
it is easy to see that @xmath78 is proportional to acceptance probability of non - diagonal field in the case where corresponding bare green function @xmath79 .
since these processes are unphysical , the natural choice is @xmath80 .
but it leads to division by zero in the updating the inverse green function matrix @xcite . on the other hand increasing the @xmath78 parameter causes a sign problem .
we find a reasonable choice of @xmath78 to be on the order of @xmath81 .
moreover for some special cases like the atomic limit , where @xmath82 is constant , a small noise should be added to all the @xmath39 parameters to avoid numerical divergency .
the co - cu system is treated as five - orbital impurity model representing @xmath83 electronic shell of the cobalt atom hybridized with a bath of a conduction cu electrons .
the bath green function was obtained using the first - principle density - functional theory within the supercell approach . for co impurity atoms in the bulk as well as in / on the cu ( 111 )
surface the bath green functions were obtained using the vienna - ab - initio simulation package ( vasp)@xcite using the projector augmented wave ( paw ) basis sets@xcite . the density functional calculation for cobalt impurity in the bulk was carried out using cocu@xmath84 supercell structure with lattice constant corresponding to pure copper .
the surfaces were modelled by supercells of cu ( 111 ) slabs containing 5 cu layers with @xmath85 and @xmath86 lateral extension for co in and on the surface , respectively .
the paw basis naturally provides the projectors @xmath87 required in eq .
( [ eqn : g0 ] ) . in using these paw projectors ,
directly , we employ here the same representation of localized orbitals as used within the lda+u - scheme implemented in the vasp - code itself or as discussed in the context of lda+dmft in ref . .
for the problem of a single co impurity in a bulk copper matrix the basis set of spherical harmonics @xmath88 is used . in this basis
the interaction part of the hamiltonian contains only terms of the following form : diagonal density - density like @xmath89 , where @xmath90 and non - diagonal @xmath91 where @xmath30 and @xmath31 .
the coulomb matrix for the @xmath4-electron shell in the basis of complex harmonics contains 45 non - equivalent diagonal terms .
non - diagonal terms can be further classified to a spin - flips , where @xmath65 , @xmath66 and the most general four - orbitals interactions , where this condition is not fulfilled .
notice , that pair - hopping terms ( @xmath92 , @xmath34 ) are restricted by symmetry in this basis . in description of @xmath4-electron shell we have to involve 20 non - equivalent spin - flips and 64 terms of the most general form . to find the effects ,
caused by non - diagonal terms , we used two different interaction hamiltonian .
first , interaction with only diagonal terms was used . in this case
there is no sign problem .
then , the complete coulomb interaction matrix of the @xmath83-electron shell of the cobalt atom with 129 terms was included .
ev , @xmath93 ev @xmath94 ev@xmath95 , for 5-orbital impurity at half - filling ; inset : @xmath96 ev , @xmath97 ev , @xmath98 ev@xmath95 , for 5-orbital impurity with 8 electrons.,scaledwidth=50.0% ] as a benchmark we use impurity problem in the atomic limit , since it can be compared with the result of exact diagonalization ( ed ) method .
the results imaginary time green function for 5-orbitals model with different chemical potentials corresponding to the @xmath99 and @xmath100 configurations are shown in the fig .
[ tbvsed ] in comparison with ed results .
the significant difference between density - density ( diagonal ) interaction and the full vertex can be found both at half - field case with relatively high temperature with the @xmath101 @xmath102 , @xmath93 @xmath102 , @xmath103 @xmath104 and at non - symmetric case even for lower temperature .
note that in the @xmath100 and @xmath105 cases the many - body ground - state have different symmetry for diagonal interactions and non - diagonal full vertex .
the results for @xmath100 configuration with the interaction parameters @xmath96 @xmath102 , @xmath97 @xmath102 , @xmath106 @xmath104 are shown in the insert to the fig .
[ tbvsed ] .
the difference between green function of the interacting system with full coulomb interaction and density - density one is visible on the @xmath107 .
we find a very good agreement between ct - qmc results and ed solution . in the inset of fig .
[ hist ] we show the distribution of non - diagonal terms , i.e. the contribution of coulomb fields of the form ( [ sym1 ] ) to the resulting green function .
the zero entry of this histogram counts the number of steps when all the fields contributing to the fermionic determinant ( [ zexp ] ) were of density - density type .
the entry with index @xmath108 show us the number of steps where the average ( [ zexp ] ) was containing two spin - flip type fields ( [ sym1 ] ) .
such situation takes place , for example , when one coulomb field representing spin - flip @xmath109 process was used to construct the determinant .
one can see in the insert of the fig .
[ hist ] , that only even orders of interaction histogram have large acceptance probability at high temperature and even the tenth order in non - diagonal interactions has non - zero contribution .
the 3-rd and 5-th order contributions exists due to the finite @xmath110 parameter .
ev , @xmath97 ev , @xmath111 ev@xmath95 for 5-orbital impurity coupled to realistic cu - bath with 7 electrons ; inset : @xmath112 ev , @xmath97 ev , @xmath113 ev@xmath95 in the case of 5-orbital impurity model , coupled to semi - elliptical bath with bandwidth @xmath114 ev at the half - filling , scaledwidth=50.0% ] typical distribution of the perturbation order ( 5-orbital aim with @xmath115 electrons , @xmath112 @xmath102 , @xmath97 @xmath102 , @xmath111 @xmath104 ) is shown in fig .
[ hist ] , main plot .
dash line denotes the perturbation order during accepted steps that involved non - diagonal fields .
the coincidence of distributions maxima of both histograms demonstrate that the acceptance rate mostly depends on diagonal interactions . for many - body calculations of the co impurity in the cu matrix
we need to find the effective @xmath4-orbital chemical potential which defines the number of 3d - electrons of cobalt .
the particular electronic configuration of a co atom in a copper matrix is unknown , but the dft results ( @xmath116 ) give us an evidence that it is close to @xmath105 configuration .
therefore we performed all impurity calculations for cobalt @xmath105 configuration .
ev , @xmath97 ev , @xmath117 ev@xmath95 for 5-orbital impurity with 7 electrons.,scaledwidth=50.0% ] the results of the ct - qmc calculations for @xmath112 ev and @xmath97 ev are presented in fig.[doses5b ] compared to the bare impurity density of states for cobalt impurity in the bulk .
there is a pronounced difference between kondo - like resonance near the fermi level.in the case of full @xmath3-vertex it becomes more narrow and located much closer to the fermi level .
the sign problem for realistic five - band model depends crucially on the symmetry of coulomb vertex @xmath13 and magnitude of non - diagonal terms in the bath green functions @xmath8 .
the most serious problem is related with non - diagonal terms of u - matrix , therefore we use a basis of complex spherical harmonics . in this case
there is no so - called three - cite terms or correlated hopping , e.g. @xmath20 . on the other hand , in this basis , the bath green - function matrix @xmath8 for d - electrons has two non - diagonal elements in the bulk of cubic crystals and much more on the surface and in the first layer
. moreover there are lot of small four - site terms @xmath13 which result in a large sign problem for surface - adatom calculations .
the sign problem for a co impurity in the bulk is not large and average sign is between 0.90 and 0.97 depends on the simulation temperature . in the case of non - diagonal interaction
we used so - called cluster steps which correspond to complex monte - carlo updates with more than one additional interaction field .
this scheme became essential for spin - flip like interaction or more general u - vertex which can contribute to the green - function only in the second or higher order
diagammatic expansion and this can let the monte - carlo process to explore all the phase space .
we note that probability of non - diagonal terms drastically decrease with increasing the hybridization to the bath . nevertheless , at least for three - band benchmarks we found remarkable effect of the spin - flip terms if the bath green function has peaks in the vicinity of the fermi level on the distance of the order of @xmath118 .
ev , @xmath119 ev , @xmath120 ev@xmath95 for 5-orbital impurity with 7 electrons.,scaledwidth=50.0% ] we estimated the renormalization factor @xmath121 for @xmath122 ev , @xmath97 ev and @xmath117 ev@xmath95 and find @xmath123 and @xmath124 which shows the reasonable strong interaction of co d - electrons .
we estimate the kondo temperature ( t@xmath125 ) using the temperature dependence of fwhm for resonance near fermi level .
since our simulation temperature is very high compared to t@xmath125 we can get only order of magnitude of t@xmath126 ev , which is reasonable for co - impurity systems .
we also performed the ct - qmc calculation of cobalt impurity on the surface of cu(111 ) and embedded into the first copper layer .
in contrast to the bulk system the surface one has a large sign problem , related with the relativelly large non - diagonal elements of the bath green functions .
although changing of the sign is a very rare event ( less then 0.03% of the accepted steps ) and we used a simple constrained sign calculations .
comparison of the different spectral functions for the bulk , surface and fist - layer cobalt impurity is presented in fig . [ 3dos ] .
one can see clearly the change of the kondo resonance width as a function of reduced dimensionality .
in conclusion , we perform the continuous time qmc calculation of realistic 5-orbital co impurity in copper and discuss the relevance of non - diagonal part of coulomb matrix in the kondo problem . comparing figs .
[ doses5b ] and [ 3dos ] we find that non - density - density terms in the coulomb vertex are required to obtain quantitative predictions of spectral functions and related properties .
the position of the hubbard peaks and the kondo peak is markedly changed by spin - flips and other non - diagonal terms of the coulomb vertex . thus ,
obtaining sensitive observables like kondo temperatures quantitatively requires accounting for these terms . on the other hand hybridization effects like bringing the co impurity from bulk to the surface and having it on top of the surface can be quite drastic : as fig [ 3dos ] shows , the sharpening of the kondo resonance and the shifting of the hubbard bands is much stronger when going from bulk to the surface then on switching on the non - diagonal part of coulomb matrix . only the qualitative overall shape the dos and its response to strong hybridization changes
are well described by density - density type terms of the coulomb vertex .
the authors acknowledge a financial support from dfg sfb-668 ( germany ) rffi ( russia ) and fom ( the netherlands ) . | the electronic structure of a prototype kondo system , a cobalt impurity in a copper host is calculated with accurate taking into account of correlation effects on the co atom .
using the recently developed continuous - time qmc technique , it is possible to describe the kondo resonance with a complete four - index coulomb interaction matrix .
this opens a way for completely first - principle calculations of the kondo temperature .
we have demonstrated that a standard practice of using a truncated hubbard hamiltonian to consider the kondo physics can be quantitatively inadequate . |
since the discovery of the @xmath15 and @xmath1 mesons in the seventies by the mark i collaboration @xcite , the interest in the charmed meson spectroscopy has grown exponentially and the newly developed experimental facilities like cleo , belle , babar , etc . , are continuously bringing more relevant information about the excited charmed meson states @xcite .
this information is very useful to elucidate the way the strong interaction works when a heavy quark is involved . in this aspect
, many theoretical efforts have been dedicated to the understanding of the properties of the excited meson states in the open as well as in the hidden charm sectors using different methods based on quark models , qcd sum rules , lattice qcd , effective field theories , etc .
, @xcite .
however , in spite of all the experimental and theoretical works , a rather scarce information is available on the properties of the charmed meson resonances like spin , isospin , parity , etc .
the most clear evidence of this fact is the summary of the knowledge on these particles given by the particle data group ( pdg ) @xcite , which lists only six @xmath15 and five @xmath1 meson excited states ( cataloging separately the neutral and the charged ones ) spread over the energy region of 2400 - 2750 mev . out of them ,
five @xmath15 and two @xmath1 states are omitted from the summary table due to controversy on their properties .
it is also interesting to notice that the energy region studied experimentally covers a range of only 350 mev from the first to the last excited state , while the interaction of a @xmath15 or a @xmath1 meson , due to their heavy mass ( around 2000 mev ) , with a vector or few pseudoscalar mesons could give rise to a resonance with a mass close to 3000 mev .
recently , a theoretical work along this line was made in ref .
@xcite , where a system formed of a @xmath15 meson and the @xmath16 resonance was investigated using two different techniques : one based on effective field theories and other based on qcd sum rules , and the agreement of both methods was remarkably good . as a result , a new @xmath15 meson state with mass around 2900 mev was predicted .
similarly , with another recent study of the @xmath0 system in s - wave using effective field theories based on hidden local symmetry @xcite an effort has been made to shed some more light on the nature of two of the @xmath1 resonances listed by the pdg : @xmath9 and @xmath10 . in ref .
@xcite the spin - parity of the former state was confirmed to be @xmath17 , and the spin and parity of the latter one was predicted to be @xmath18 .
in addition , a new state with mass close to 2600 mev and spin 0 , thus @xmath19 , was predicted . in this manuscript , with the objective of understanding the properties and nature of the resonances @xmath9 and @xmath10 , we investigate the @xmath0 system using qcd sum rules to determine the mass of its possible resonances and compare our results with the ones found in ref .
we would like to investigate the generation of @xmath0 molecular states using a formalism based on qcd sum rules , but before we enter into details about how to do this , let us first summarize the main aspects of the previous study of the @xmath0 system made by the authors of ref .
this discussion will be useful while comparing the results obtained in our work with those reported in ref .
@xcite . in ref .
@xcite , the scattering amplitudes of the @xmath0 and @xmath20 coupled channel system are calculated by solving the bethe - salpeter equation in its on - shell factorization form @xcite : @xmath21 in eq .
( [ bs ] ) , the kernel @xmath22 corresponds to a @xmath23 matrix whose elements are the lowest order amplitudes describing the transitions between the channels . these amplitudes are obtained from a lagrangian built on the basis of the hidden local symmetry @xcite .
the @xmath23 matrix @xmath24 in eq .
( [ bs ] ) is a diagonal matrix whose elements correspond to the loop function of the two hadrons present in each of the channels . as shown in ref .
@xcite , the kernel @xmath22 is a function of the polarization vectors of the vector mesons , @xmath25 , where @xmath26 is a time - spatial index .
therefore , to identify resonances arising from the dynamics involved in the interaction of the vector mesons considered , @xmath22 needs to be projected on the spin as well as on the isospin bases .
we can have two different isospins , @xmath2 or @xmath27 , and three different spins , @xmath3 , @xmath4 , and @xmath5 , for the @xmath0 system . using the phase convention @xmath28 and @xmath29
, the isospin @xmath30 and @xmath27 states can be written in terms of the charged particles as @xmath31 where @xmath32 and @xmath33 are spatial indices , project the kernel @xmath22 and , thus , the scattering matrix @xmath34 , on spin @xmath35 , @xmath4 and @xmath5 , respectively . using the amplitudes which result from the projection of the kernel @xmath22 on isospin and spin to solve eq .
( [ bs ] ) , it was found in ref .
@xcite that the scattering matrix @xmath34 shows poles in the second riemann sheet with masses and widths compatible with the ones of the states @xmath9 and @xmath10 , predicting the quantum numbers of the latter one to be @xmath18 .
the coupling of these poles to the @xmath0 channel was found to be much larger than the corresponding to the @xmath20 channel , thus , leading the authors of ref .
@xcite to conclude that the resonances @xmath9 and @xmath10 can be interpreted as @xmath0 molecular states .
in addition , the existence of a resonance with a mass around 2600 mev and quantum numbers @xmath19 was also predicted as a consequence of the interaction of a @xmath1 with a @xmath36 . in this manuscript
, we investigate the interaction of a @xmath1 and a @xmath36 considering a different point of view than the one in ref .
@xcite : the one of the qcd sum rules . in this case , the starting point to determine the mass of the possible @xmath0 states is the evaluation of the two - point correlation function @xcite @xmath37|0\rangle\label{corr},\end{aligned}\ ] ] where @xmath38 is the momentum flowing from @xmath35 to @xmath39 , @xmath40 $ ] represents the @xmath41-ordered product and @xmath33 is the current associated to the @xmath0 system .
the key idea of the qcd sum rule method is to consider that this correlation function is of dual nature and it depends on the value of the momentum @xmath38 . for large momentum , i.e. , short distances
, the correlation function represents a quark - antiquark fluctuation and can be calculated using perturbative qcd . in this case , the current @xmath33 is written in terms of the quark content of the @xmath1 and @xmath36 mesons .
however , since we are ultimately interested in studying the properties of hadrons , the relevant energies are lower and contributions from quark condensates , gluon condensates , etc . , need to be included in the evaluation of eq .
( [ corr ] ) .
this can be conveniently accomplished by using the wilson operator product expansion ( ope ) of the correlation function @xcite . in this case , eq .
( [ corr ] ) is expanded in terms of all local gauge invariant operators expressible in terms of quark and gluon fields in the form of condensates and a series of coefficients .
the local operators incorporate nonperturbative long - distance effects , while the coefficients , by construction , include only the short - distance domain and can be determined perturbatively .
this way of evaluating the correlation function is called the ope description " . in table
[ table ] we show the values used for the different condensates and quark masses involved in the ope description of eq .
( [ corr ] ) .
since the quark @xmath42 is much heavier than the quarks @xmath43 and @xmath44 , we can work in the limit of massless @xmath43 and @xmath44 quarks .
.values of the different parameters required for numerical calculations of the correlation function given by eq .
( [ corr ] ) ( see refs .
@xcite ) . [ cols="^,^",options="header " , ] at large distances , or , equivalently , small momentum , the currents @xmath45 and @xmath33 of eq .
( [ corr ] ) can be interpreted as operators of creation and annihilation of hadrons with the same quantum numbers as the ones of the current @xmath33 . in this case , the correlation function is obtained by inserting a complete set of states with the same quantum numbers as those of the current under consideration .
this manner of determining the correlation function is often referred to as the phenomenological description " .
the assumption made in the qcd sum rule method is that there must be a range of @xmath46 values in which both descriptions must be equivalent . calculating the correlation function of eq .
( [ corr ] ) using these two approaches and equating them , it is possible to obtain information about the properties of the hadronic states generated in the system . to determine the correlation function associated to the ope description we need to construct interpolating molecular currents for the @xmath0 system .
since both @xmath1 and @xmath36 are vector mesons , the simplest current we can use for the @xmath0 system is of the form @xmath47\left[\bar{q}^2_b(x)\gamma_\nu q^3_b(x)\right],\end{aligned}\ ] ] with @xmath48 , @xmath49 and @xmath50 representing the fields of the light quarks @xmath43 or @xmath44 , @xmath51 is the field for the quark @xmath42 , @xmath52 and @xmath53 are color indices and @xmath54 represents the dirac matrix . considering the isospin relations of eq .
( [ isos ] ) , the currents for the @xmath0 system for the cases of total isospin @xmath2 and @xmath27 are thus , @xmath55,\label{cur}\\ j^{3/2}_{\mu\nu}(x)&=-(\bar{d}_a\gamma_\mu c_a)(\bar{d}_b\gamma_\nu u_b),\nonumber\end{aligned}\ ] ] where we have omitted the @xmath56 dependence of the quark fields for simplicity . using these currents
, we construct the function @xmath57|0\rangle.\label{corr2}\end{aligned}\ ] ] the currents in eq .
( [ corr2 ] ) and , thus , @xmath58 , do not have a well defined spin and need to be projected .
we will elaborate on the spin projection of eq .
( [ corr2 ] ) in the next section , which will lead to a correlation function with a well defined spin , @xmath59 . following the standard procedure of the qcd sum rule method , contracting all the quark anti - quark pairs present in eq .
( [ corr2 ] ) through the currents in eq .
( [ cur ] ) , we can rewrite @xmath58 in terms of the quark propagators , and then we can make the ope expansion of these propagators . in terms of the quark propagators ,
( [ corr2 ] ) adopts the form @xmath60 , \label{corr4}\end{aligned}\ ] ] where tr@xmath61 is the trace operator , @xmath62 represents the propagator of a light quark ( @xmath43 or @xmath44 ) , @xmath63 is the propagator of the quark @xmath42 and @xmath64 is a coefficient whose value is @xmath30 for isospin @xmath30 and @xmath65 for @xmath27 . for convenience , we work in the momentum space for heavy quarks and in the configuration space for light quarks .
the expressions for these propagators , including the terms related to quark and gluon condensates , can be found in ref .
@xcite . making the ope expansion for the propagators involved in eq .
( [ corr4 ] ) , and projecting the result on a particular spin configuration , we can express the correlation function @xmath59 in terms of the following dispersion relation @xmath66 with @xmath67 being the spectral density , which is related to the imaginary part of the correlation function through @xmath68 $ ] . in this study of the @xmath0 system
we have considered condensates up to dimension 7 in the ope description . in this way : @xmath69 the first term in eq .
( [ rho ] ) or perturbative term corresponds to dimension 0 in the expansion . the next term , i.e. , the quark condensate contribution ( @xmath70 ) has dimension 3 .
the two gluon condensate contribution ( @xmath71 ) corresponds to dimension 4 in the expansion .
the mixed condensate or @xmath72 term contributes with dimension 5 .
the two quark and the three gluon condensates ( @xmath73 and @xmath74 , respectively ) have dimension 6 and , finally , the contribution associated to the condensate @xmath75 has dimension 7 .
the result for each of the terms of eq .
( [ rho ] ) for the different isospins and spins can be found in the appendix of this manuscript . in the phenomenological description
, the correlation function for the @xmath0 system can be written in terms of a spectral density as @xmath76 all hadrons with the same quantum numbers as the ones associated to the current @xmath33 of eq .
( [ corr ] ) contribute to the density of eq .
( [ corr5 ] ) . therefore , to extract information about the states we are interested in , the spectral density @xmath77 needs to be parametrized conveniently .
commonly , the density of eq .
( [ corr5 ] ) is expressed as a sum of a narrow , sharp state , which represents the one we are looking for , and a smooth continuum @xcite @xmath78 with @xmath79 the coupling of the current to the state we are interested in and @xmath80 its mass .
the density related to the continuum of states is assumed to vanish below a certain value of @xmath81 , @xmath82 , called continuum threshold . above this threshold
it is common to consider the ansatz @xcite @xmath83 therefore , using this parametrization of the spectral density of eq .
( [ corr5 ] ) , the correlation function in the phenomenological description adopts the form @xmath84 ideally , the result from the evaluation of eqs .
( [ corrope ] ) and ( [ corrphen ] ) should be same at some range of @xmath46 at which we could just directly equate both expressions . however , this is not completely true : the correlation function in eq .
( [ corrope ] ) suffers from divergent contributions originated from long range interactions , while the one in eq .
( [ corrphen ] ) contains contributions arising from the continuum .
these problems can be solved by applying the borel transformation to both correlation functions and then equating them , which gives the relation @xmath85 where @xmath86 is the borel mass parameter . calculating the derivative of eq .
( [ mrule ] ) with respect to @xmath87 and dividing the resulting expression by eq .
( [ mrule ] ) , we can determine the mass as @xmath88 once the mass is obtained , we can calculate the coupling @xmath89 through eq .
( [ mrule ] ) @xmath90
let us discuss now how to make the spin projection of the function @xmath58 of eq .
( [ corr2 ] ) .
as we discussed in section [ eft ] , it was shown in ref .
@xcite that the expressions given in eq .
( [ proj ] ) project the scattering matrix on spin 0 , 1 , and 2 for small momentum of the external hadrons . recalling that in quantum mechanics the wigner - eckart theorem establishes that all vectors ( tensors ) have the same matrix elements up to a reduced matrix coefficient , it is possible to show that in the non relativistic limit , taking zero momentum for the quarks , the @xmath91 matrices present in the currents of eq .
( [ cur ] ) behave as the polarization vectors of eq .
( [ proj ] ) .
therefore , by analogy to the projectors of eq .
( [ proj ] ) , the combinations @xmath92 with @xmath93 being the delta of kronecker , project the currents of eq .
( [ cur ] ) and , thus , the function @xmath58 , on spin @xmath35 , @xmath4 , and @xmath5 in the limit of zero momentum for the quarks .
however , this limit is not applicable to the present study .
it is simple to generalize the projectors of eq .
( [ delproj ] ) for a situation in which the quarks have a nonzero momentum .
to do this , we need to establish the covariant extrapolation of the delta of kronecker , which we call @xmath94 .
since the delta of kronecker is symmetric under the exchange of the two indices , @xmath94 should also be and to construct it we just have the metric tensor , @xmath95 , and the four momentum @xmath38 .
thus , the only possibility is @xmath96 as can be seen , in the limit of @xmath97 , @xmath98 , @xmath99 and @xmath100 . in this form , replacing @xmath101 for @xmath102 in eq .
( [ delproj ] ) we arrive to the following projectors : @xmath103 which project the currents of eq .
( [ cur ] ) and , therefore , the function @xmath58 , on spin @xmath35 , @xmath4 , and @xmath5 for any value of the four momentum @xmath38 .
note that @xmath104 , which is , with the exception of the minus sign , the same normalization as the delta of kronecker . therefore , the normalization of the projectors of eq .
( [ delproj ] ) also holds for the projectors of eq .
( [ deproj ] ) .
the deduction and use of spin projectors in the qcd sum rules method is , up to our knowledge , a novelty .
although , work in this direction has been done in ref .
@xcite , where operators choosing the largest spin component of a particle which couples to the corresponding current were deduced in a study related to the @xmath105 state .
it is easy to show , using the phenomenological description of the correlation function , that the expressions in eq .
( [ deproj ] ) indeed project eq .
( [ corr2 ] ) on spin 0 , 1 , and 2 , respectively .
as mentioned in section [ sumrulesect ] , within the phenomenological description , the calculation of the correlation function is done by inserting intermediate states for the hadron of interest . in the present case , the current @xmath106 of eq .
( [ corr2 ] ) couples to spin 0 , 1 , and 2 states , with the respective couplings defined through : @xmath107 in eq .
( [ states ] ) @xmath108 , @xmath109 , @xmath110 and @xmath111 represent a scalar , a vector , an axial and a tensor state , respectively , @xmath112 , @xmath113 , @xmath114 , @xmath115 are the couplings of the current to the respective states , @xmath116 and @xmath117 are the polarization vectors of the vector , axial and tensor states , respectively , @xmath118 the levi - civita tensor and @xmath52 , @xmath53 are arbitrary constants . saturating the function of eq .
( [ corr2 ] ) with these states and summing over the polarizations , we get the expression @xmath119 with @xmath120 , @xmath121 , @xmath122 , and @xmath123 being the parts of the correlation function associated with the scalar , vector , axial and tensor states .
now , it is easy to see that the action of @xmath124 , @xmath125 and @xmath126 over eq .
( [ expan ] ) selects the scalar , axial and tensor part , respectively , of the correlation function . in summary ,
the correlation function to be used to study the @xmath0 system for a certain isospin @xmath127 and spin @xmath128 is given by @xmath129
the masses and couplings calculated with eqs .
( [ mass ] ) and ( [ lambda ] ) , respectively , are functions of the borel mass @xmath86 .
ideally , the results should not depend on the value of the borel mass used to determine them .
but , in reality , this is not the case and one relies on the existence of a range of borel masses ( or borel window " ) in which the results obtained for the mass and the coupling can be relied upon .
the determination of this borel window is based on the following constraints : ( 1 ) in the phenomenological description , the contribution to the correlation function arising from the pole term , which represents the state in which we are interested in , should dominate over the contribution from the continuum of states with the same quantum numbers .
this condition fixes the maximum value for the borel mass , @xmath130 , at which the results obtained for the mass and coupling are meaningful ( 2 ) .
the other constraint is to guarantee the convergence of the ope expansion .
this fixes the minimum borel mass at which the sum rule is reliable , @xmath131 . here
, in this manuscript , if @xmath132 is the maximum dimension taken into account in the ope expansion , we consider that the right side of eq .
( [ mrule ] ) converges when the relative contribution associated to the condensate of dimension @xmath133 , i.e , the result of dividing the contribution to the integral in eq .
( [ mrule ] ) of all terms of the series up to dimension @xmath133 by the contribution to the integral associated to all the terms , differs , in modulus , by no more than 10 - 25 % ( depending on the case ) from the relative contribution to the condensate of dimension n. using these conditions , a valid borel window is established , and the mass and coupling of the states for the different isospin and spin configurations can be calculated through eqs .
( [ mass ] ) and ( [ lambda ] ) , respectively .
note that these results , as well as the borel mass window , depend on the continuum threshold , @xmath134 .
this continuum threshold is a parameter of the model , but its value is not completely arbitrary . as can be seen in eq .
( [ ansatz ] ) , @xmath135 is related to the onset of the continuum in the current @xmath33 under consideration and a reasonable value is about 450 - 500 mev above the mass of the hadron we are looking for @xcite .
for this case , where we are searching for possible @xmath0 molecular states of masses 2.450 - 2.600 gev , we will use values for @xmath136 gev .
, @xmath137 as function of the squared borel mass .
these results are obtained for a value of the continuum threshold of 3.15 gev and using the central values of table [ table ] for the quark masses and condensates.,scaledwidth=48.0% ] in fig .
[ polecont ] we show as a function of the borel mass and for the phenomenological description of the correlation function the contributions of the pole term and the continuum , weighted by their sum , for the case @xmath2 , @xmath137 , and a value of @xmath138 gev .
as can be seen , the pole term dominates over the continuum of states for a value of the squared borel mass of 1.87 gev@xmath139 , thus , @xmath140 gev@xmath139 .
similarly , for the ope description of the correlation function , we show in fig .
[ ope ] the result of the study of the convergence of the ope series considering the central values of the quark masses and condensates of table [ table ] .
we find a convergence of the ope series for a value of the squared borel mass above 1.07 gev@xmath141 , thus , @xmath142 gev@xmath139 .
once a borel window is found , we calculate the mass resulting from eq .
( [ mass ] ) , and obtain , as can be seen in fig .
[ mass2 ] , that the mass sum rule is stable and gives as a result and @xmath137 as a function of the squared borel mass . these results are obtained for a value of the continuum threshold of 3.15 gev and using the central values of table [ table ] for the quark masses and condensates.,scaledwidth=48.0% ] and @xmath137 as a function of the squared borel mass . the crosses in the figure indicate the borel mass window . this result is found using a value of the continuum threshold of 3.15 gev and considering the central values of table [ table ] for the quark masses and condensates.,scaledwidth=48.0% ] @xmath143 the value shown in eq .
( [ mres ] ) is obtained by averaging the mass over the borel window and by calculating the standard deviation to determine the error .
similarly , using eq .
( [ lambda ] ) , we can get the coupling of the state to the current used and we find @xmath144 to estimate the uncertainty of the results of eqs .
( [ mres ] ) and ( [ lres ] ) related to the @xmath145 value used and to the quark masses and condensates , we change @xmath146 in the range 3.00 - 3.15 gev and consider the range for the quark masses and condensates allowed by the error related to them ( shown in table [ table ] ) . taking into account all these sources of errors , we obtain @xmath147 analogously , we repeat this process for the cases @xmath2 and spin @xmath35 and @xmath4 , respectively . in both cases , we find the valid borel window considering the dominance of the pole term over the continuum in the phenomenological description and the convergence of the series in the ope description and averaging the results over the borel window .
we also change the value of the continuum threshold in the range @xmath148 3.00 - 3.15 gev , and the quark masses and condensates in the range shown in table [ table ] .
the results found for the masses and couplings are : @xmath149 as an example , we show in figs .
[ mass0 ] and [ mass1 ] the result of the mass sum rule calculation , for spin 0 and 1 , respectively , for a value of @xmath150 gev .
the borel mass window for both cases is indicated by crosses in the figures . and @xmath3 as a function of the squared borel mass .
the crosses in the figure indicate the borel mass window .
this result corresponds to a value of the continuum threshold of 3.15 gev and the central values of table [ table ] for the quark masses and condensates.,scaledwidth=48.0% ] but for @xmath2 and @xmath151.,scaledwidth=48.0% ] therefore , we can conclude that three states are found for the case @xmath2 , each with a different spin @xmath3 , @xmath4 , and @xmath5 , with the corresponding masses and couplings given by eqs .
( [ m2 ] ) and ( [ m01 ] ) .
it is interesting to note that the coupling of the spin 2 state to the current is bigger than the respective ones for spin 0 and 1 .
these results are in striking agreement with the ones obtained in ref .
the state found in this manuscript with spin 2 can be associated with the @xmath152 listed by the pdg . for the states with spin 0 and 1 , there is only one candidate listed by the pdg and that is @xmath10 . however , nothing is known about the spin and parity of this state . in ref .
@xcite , the widths related to the resonances found were calculated , obtaining a width of around 40 mev for the state with spin 2 , 60 mev for the state with spin 0 and practically zero width for the state with spin 1 . since the width listed by the pdg for @xmath10 is @xmath153 mev , the authors of ref .
@xcite associated the state with spin 1 to @xmath10 and predicted the existence of a state with spin 0 and a similar mass , but with a much larger width . in this manuscript , we have calculated the masses and couplings for the different states . to make a proper identification of @xmath10 with one of the states with spin 0 and 1 , a qcd sum rule calculation to determine the width of each of the states might be helpful . however
, this is beyond the scope of the present manuscript .
thus , with the information at hand , we can only confirm the existence of two nearly degenerate states with mass around @xmath154 mev and spin 0 and 1 , respectively , one of which can probably be related to @xmath10 .
the study of the @xmath0 system for the case @xmath11 is particularly interesting , since if a state its generated , its quantum numbers can not be obtained in terms of a quark and anti - quark configuration and we can qualify this state as an exotic one .
the study made by the authors of ref .
@xcite disfavor the formation of molecular resonances in the @xmath0 system for total isospin @xmath27 .
this is due to the fact that the kernel which enters in the resolution of the bethe - salpeter equation , eq .
( [ bs ] ) , is repulsive for this particular isospin and for the three possible spins . actually , this result found for the @xmath0 system is similar to the one found in the study of other hadronic systems using an approach with the same spirit of ref .
thus , in general , in such approaches whenever the system can generate a resonance whose quantum numbers can not be explained in terms of the constituent quark model , a repulsive interaction is found , preventing the formation of such a state .
we have studied the configuration @xmath11 of the @xmath0 system for the three possible spins using the qcd sum rule approach explained in sect .
[ sumrulesect ] . in this case
, it is possible to find a valid borel window and calculate the mass and the coupling of the corresponding states .
considering the different sources of error and averaging over the borel window we get @xmath155 the experimental search of these states will be definitely very useful to clarify the existence of possible exotic molecular hadron resonances formed due to the interaction of a @xmath1 and a @xmath36 and , at the same time , a test of the qcd sum rule approach used in this manuscript .
using an approach based on qcd sum rules , we have studied the interaction of a @xmath1 and a @xmath36 mesons and investigated the existence of resonances for the different isospins and spins . for the isospin @xmath30 case , we have found three states with masses @xmath156 gev , @xmath157 gev and @xmath158 gev and spin 0 , 1 and 2 , respectively .
the state with spin 2 can be associated with the meson @xmath152 listed by the pdg .
one of the other two resonances with spin @xmath35 or @xmath4 can be related to the meson @xmath10 of the pdg , whose spin - parity is unknown . for the case of isospin @xmath27
, we have obtained three states with masses @xmath159 gev , @xmath160 gev , @xmath161 gev and spin 0 , 1 , and 2 , respectively .
these states can be considered exotic in the sense that their quantum numbers can not be obtained from a quark plus antiquark configuration . while the results found for the isospin 1/2 case are in good agreement with the ones obtained by a previous study ( see ref .
@xcite ) of the same system with a different approach , the results for the isospin 3/2 sector are different
. the experimental studies of this system will be very relevant to confirm the results found by the model of ref .
@xcite and the one of this manuscript in the isospin @xmath30 sector and the possible existence of @xmath0 molecular resonances with isospin @xmath27 and spin 0 , 1 , and 2 .
we thank professors altu zpineci and juan nieves for very useful discussions and for a careful reading of the manuscript .
the authors would like to thank the brazilian funding agencies fapesp and cnpq for the financial support .
this work is partly supported by the spanish ministerio de economia y competividad and european feder fund under the contract number fis2011 - 28853-c02 - 01 and the generalitat valenciana in the program prometeo , 2009/090 .
e. oset acknowledges the support of the european community - research infrastructure integrating activity study of strongly interacting matter ( acronym hadronphysics3 , grant agreement n. 283286 ) under the seventh framework programme of eu .
in figs . [ diag1 ] and [ diag2 ] we show the different feynman diagrams considered to determine the different contributions to eq .
( [ rho ] ) . the results for the spectral density for a certain isospin @xmath127 , spin @xmath128 and dimension @xmath15 can be written as : @xmath162 where @xmath163 is a number representing the dimension of the terms considered in the ope expansion and @xmath164 is associated with the condensate of dimension @xmath15 . in particular , the dimension @xmath165 corresponds to the perturbative term , @xmath166 . the dimension @xmath167 is related to the quark condensate or @xmath168 contribution .
the two gluon condensate corresponds to @xmath169 , i.e. , @xmath170 .
the mixed condensate contributes to the dimension 5 , @xmath171 . for dimension
@xmath172 we have contributions from the two quark condensate , @xmath173 , and the three gluon condensate , @xmath74 .
the contributions to the dimension @xmath174 arise from a quark and a two gluon condensates , @xmath175 .
the coefficients @xmath176 are : @xmath177 , @xmath178 , @xmath179 , @xmath180 and @xmath181 .
the functions @xmath182 and coefficients @xmath183 can be found in table [ racoef ] , while the @xmath184 functions are given in tables [ tcoef1h ] and [ tcoef3h ] for isospin @xmath2 and @xmath27 , respectively . as can be seen in these tables , we can have two different integral variables in eq .
( [ spec ] ) : @xmath185 , in this case the integral limit @xmath186 , and @xmath187 , for which @xmath188 .
the results for the @xmath182 functions given in table [ racoef ] are expressed in terms of the quantities defined as follows : @xmath189^k,\quad\textrm{if $ x=\alpha$.}\nonumber\\ \mathcal{h}^{i , j , k , l}\equiv ( m_c)^i \frac{\eta^j}{(\eta-1)^l ( m^2)^k } e^{\frac{m^2_c}{(\eta-1)m^2}},\quad\textrm{if $ x=\eta$.}\nonumber\end{array}\right.\end{aligned}\ ] ] where the indices @xmath32 , @xmath33 , @xmath190 , @xmath191 are different for each isospin , spin and dimension . 99 g. goldhaber , f. pierre , g. s. abrams , m. s. alam , a. boyarski , m. breidenbach , w. c. carithers and w. chinowsky _ et al .
_ , phys .
* 37 * , 255 ( 1976 ) .
g. goldhaber , j. wiss , g. s. abrams , m. s. alam , a. boyarski , m. breidenbach , w. chinowsky and j. dorfan _ et al .
_ , phys .
b * 69 * , 503 ( 1977 ) .
p. avery _
[ cleo collaboration ] , phys .
d * 41 * , 774 ( 1990 ) .
k. abe _ et al . _ [ belle collaboration ] , phys .
d * 69 * , 112002 ( 2004 ) .
del amo sanchez _ et al .
_ [ babar collaboration ] , phys .
d * 82 * , 111101 ( 2010 ) .
fayyazuddin and riazuddin , phys .
d * 69 * , 114008 ( 2004 ) .
m. nielsen , f. s. navarra and s. h. lee , phys .
* 497 * , 41 ( 2010 ) . for a review and references to original works ,
see e.g. , s. narison , _ qcd as a theory of hadrons , cambridge monogr .
* 17 * , 1 ( 2002 ) ; _ qcd spectral sum rules , world sci .
notes phys .
_ * 26 * , 1 ( 1989 ) ; acta phys .
b * 26 * , 687 ( 1995 ) ; riv .
* 10n2 * , 1 ( 1987 ) ; phys . rept . * 84 * , 263 ( 1982 ) . | we present a study of the @xmath0 system made by using the method of qcd sum rules to determine the mass of possible resonances generated in the same system . using isospin and spin projectors
, we investigate the different configurations and obtain evidences for three @xmath1 mesons with isospin @xmath2 , spin @xmath3 , @xmath4 , @xmath5 and with masses @xmath6 mev , @xmath7 mev , and @xmath8 mev , respectively .
the last state can be associated with @xmath9 ( spin 2 ) listed by the particle data group , while one of the first two might be related to @xmath10 , with unknown spin - parity . in the case of @xmath11
we also find evidences of three states with spin 0 , 1 and 2 , respectively , with masses @xmath12 mev , @xmath13 mev , and @xmath14 mev .
the results for the sector @xmath2 and @xmath3 , @xmath4 , @xmath5 , are intriguingly similar to a previous study of the @xmath0 system based on effective field theories , supporting in this way a molecular picture for the resonances @xmath10 and @xmath9 , while the results for @xmath11 hint towards the existence of exotic mesons since a multiquark configuration is required to get the quantum numbers of the states found . |
the evolution of galaxies is driven by the intricate power play between interstellar turbulence and gravity controlling the formation of stars in giant molecular clouds @xcite .
the rate at which stars form is hence a pivotal quantity in tracing a galaxy s fundamental properties and distribution of matter and activity .
nonetheless , the functional dependence of the column density of star formation ( @xmath0 ) has been a highly debated topic for more than two decades , with historical parameterizations including the mean column density of available gas ( @xmath1 ) @xcite , as well as the ratio between @xmath1 and the mean freefall time @xmath3 @xcite .
however , significant scatter remains in both these approaches , such that @xmath0 can vary by more than an order of magnitude for any given input @xmath1 or @xmath4 .
follow - up analyses of these theoretical models via computer simulations have determined that the observed scatter may be primarily attributed to the physical variations in the sonic mach number ( @xmath5 ) of the turbulence in the star - forming clouds @xcite .
this results in separate @xmath5-dependent relations for the sfr , which cover the observed range of @xmath0 .
the aim of this work is hence to unify these separate laws and develop a unique , universal relation . in section [
sec : theory ] we introduce and derive our new sf law based on the pdf and mach number of interstellar turbulence .
section [ sec : obs ] describes the observational sample utilized to assess our new sf law . in section [ sec : results ] we present our findings and compare our new model to previous parameterizations . in section [ sec : appl ] we invert our new sf law to make sonic mach number predictions for extragalactic sources .
finally , section [ sec : conclusions ] summarizes our conclusions .
here we derive an improved sf law , which takes into account the distribution of gas densities ( pdf ) present in galactic clouds and the interstellar medium ( ism ) .
this follows from the finding that the observed scatter in the sf law can be principally attributed to the variation in the sonic mach number of the clouds and galaxies observed @xcite . in 2012
krumholz , dekel and mckee initiated the refinement of the traditional kennicutt - schmidt law by dividing the gas surface density by the freefall time of the gas ( @xmath4 ) , indeed achieving a stark improvement in star formation rate correlation @xcite .
however , because this model implements only the division between the _ average _ surface density and the _ average _ freefall time of the gas , important information regarding the wide distribution of densities within the ism and molecular clouds @xcite is not taken into account .
we hence extend upon the important previous work by @xcite to incorporate the density pdf , with the ultimate goal of eliminating the remaining scatter in the previous relation . in our new sf law , we must consider the fact that denser gas forms stars at a higher rate @xcite because gas with higher density @xmath6 has a shorter freefall time than more diffuse gas , @xmath7 where @xmath8 is the gravitational constant . note that the concept of a multi - freefall time was pioneered by @xcite and @xcite .
it suggests that , due to the clumpy nature of molecular clouds , the typical timescale for star formation is not the average timescale of the clouds , but the _ density - dependent _ timescale of each collapsing substructure within the clouds .
this _ multi - freefall concept _ has recently received support from numerical simulations in @xcite and from observations in @xcite .
in contrast , @xcite and @xcite had only correlated @xmath0 with @xmath9 ( hereby denoted as @xmath10 ) ; i.e. only at a _ single _ freefall time evaluated at the mean density @xmath11 .
it is conceivable , however , that the true correlation must be in the form of an integral over the whole density pdf , @xmath12\,p_\rho d\rho$ ] . in the following ,
we compute this integral by making the standard assumption that the pdf follows a log - normal distribution as the initial condition for star formation , @xmath13 where the density is written in logarithmic form , normalized to the mean density @xmath14 this transformation of variables from @xmath6 to @xmath15 has advantages during the integration step below and allows us to write the pdf , equation ( [ eq : pdf ] ) , in its standard log - normal form .
the mean logarithmic density @xmath16 is related to the variance @xmath17 by @xmath18 @xcite . as derived by @xcite and @xcite ,
the logarithmic density variance is given by @xmath19 parameterized by the sonic mach number @xmath5 , the turbulent driving parameter @xmath20 @xcite , and the ratio of thermal to magnetic pressure , plasma @xmath21 .
the log - normal pdf , equation ( [ eq : pdf ] ) , and the associated density variance
mach number relation , equation ( [ eq : sigs ] ) , are observed to provide a good approximation to the real pdf in molecular clouds @xcite , in the galactic centre @xcite , on large galactic scales @xcite , and even in extra - galactic systems @xcite . for a rigorous derivation , we start by defining a cartesian coordinate system in which the line - of - sight toward the cloud or galaxy is in the @xmath22-direction and maps of column density and star formation rate surface density are in the @xmath23-plane .
first , the gas column density is defined as @xmath24 , where @xmath25 is the scale height of the cloud or galaxy for which @xmath1 is to be determined . from observations and simulations discussed in the previous section , we know that @xmath1 and @xmath6 follow closely a log - normal distribution , equation ( [ eq : pdf ] ) .
thus , in order to compute an average of a density - dependent variable , we simply have to integrate that variable over the entire pdf .
for the average gas column density , for example , we would evaluate @xmath26 , which is indeed the average column density @xmath27 .
now we follow exactly the same mathematical procedure , but for the combined density - dependent variable @xmath28 , i.e. , @xmath29 note that in the second step , we use the exact scaling @xmath30 , because @xmath31 and @xmath32 .
the third step transforms variables from @xmath6 to @xmath15 via equation ( [ eq : s ] ) and via the identity @xmath33 .
the last step , equation ( [ eq : sol ] ) , is the analytic solution of the integral over all densities .
we call this new quantity derived above , the _ maximum gas consumption rate _ or _ multi - freefall gas consumption rate _ ( mgcr ) , denoted @xmath34 .
we see that equation ( [ eq : sol ] ) includes the multi - freefall correction factor given by @xmath35 , which corrects the single - freefall average @xmath36 used in @xcite and @xcite for the underlying pdf of densities .
we can now insert the density variance from equation ( [ eq : sigs ] ) into equation ( [ eq : sol ] ) and obtain the multi - freefall correction factor , @xmath37 [ cols="<,<,^,^,^,^,^,^ " , ] * notes .
* column 1 : reference .
column 2 : redshift range .
column 3 : galaxy type .
column 4 : geometric mean @xmath5 prediction from inverting the improved sf law , equation ( [ eq : newsflaw ] ) , as well as the upper and lower limits of this @xmath5 estimate .
[ tab : appl ] the @xcite sf law had utilized extragalactic data of disk and starburst galaxies @xcite in addition to milky way observations .
however , it is difficult to accurately gauge the values of @xmath5 for extragalactic sources as we are not able to well resolve individual molecular clouds , especially at high redshifts .
@xmath5 estimates require measurements of the sound speed and turbulent velocity dispersion of a cloud , which in turn requires high - resolution molecular line data and a temperature measurement .
such data are hard to obtain for molecular clouds in external galaxies , as a complete census of the star - forming molecular cloud population in a given galaxy is needed at sufficiently high resolution to capture the cloud - scale mach number .
there are only very few studies of nearby galaxies ( such as m51 ) which start to resolve molecular clouds in external galaxies @xcite , but even those do not necessarily yield a complete census of clouds and it is not clear that all cloud properties are converged at the telescope resolutions available to date
. we were hence unable to include extragalactic sources that lack @xmath5 estimates in the deduction of our new sf law .
they were similarly omitted in our analysis of previous laws in order to arrive at a fair comparison between the different parameterizations of @xmath0 in figure [ fig : sflaws ] and table [ tab : fits ] .
instead , we invert our new model , equation ( [ eq : newsflaw ] ) , in order to compute @xmath5 estimates for these extragalactic sources ; predictions to be tested by future observations of resolved co maps in extragalactic sources .
our @xmath5 predictions are summarized in table [ tab : appl ] .
since the presence of vigorous star formation in starburst galaxies would result in more violent turbulence of gas compared to disk galaxies , it is expected that the starburst galaxies have greater @xmath5 than their disk galaxy counterparts of the same redshift range , which is indeed the systematic trend we find in table [ tab : appl ] .
this trend is consistent with the analyses and conclusions presented in @xcite .
we derived a new sf law given by equation ( [ eq : newsflaw ] ) , which is based primarily on the sonic mach number of turbulence , as a result of the density pdf of molecular , star - forming gas clouds .
we find that the sfr is equal to @xmath38 of the _ maximum ( multi - freefall ) gas consumption rate _ ( mgcr ) , which we derived in section [ sec : theory ] .
we compared our new model to previous parameterizations of the sfr and determined quantitatively that our new sf law provides a tight linear relation between @xmath0 and mgcr , with a factor of 34 less scatter compared to any previous sf law . by inverting equation ( [ eq : newsflaw ] )
, we predict the mach numbers ( @xmath5 ) of the star - forming molecular clouds in extragalactic sources .
our predictions are summarized in table [ tab : appl ] , which anticipate testing via future submillimetre observations .
we thank christopher mckee and the anonymous referee for useful comments , which improved this work .
c.f . acknowledges funding provided by the australian research council s ( arc ) discovery projects ( grants dp130102078 and dp150104329 ) .
l.j.k . gratefully acknowledges the support of an arc future fellowship and arc discovery project dp130103925 . | whilst the star formation rate ( sfr ) of molecular clouds and galaxies is key in understanding galaxy evolution , the physical processes which determine the sfr remain unclear .
this uncertainty about the underlying physics has resulted in various different star formation laws , all having substantial intrinsic scatter . extending upon previous works that define the column density of star formation ( @xmath0 ) by the gas column density ( @xmath1 )
, we develop a new universal star formation ( sf ) law based on the multi - freefall prescription of gas .
this new sf law relies predominantly on the probability density function ( pdf ) and on the sonic mach number of the turbulence in the star - forming clouds . by doing
so we derive a relation where the star formation rate ( sfr ) correlates with the molecular gas mass per multi - freefall time , whereas previous models had used the average , single - freefall time .
we define a new quantity called _ maximum ( multi - freefall ) gas consumption rate _ ( mgcr ) and show that the actual sfr is only about 0.4% of this maximum possible sfr , confirming the observed low efficiency of star formation .
we show that placing observations in this new framework ( @xmath0 vs. mgcr ) yields a significantly improved correlation with 34 times reduced scatter compared to previous sf laws and a goodness - of - fit parameter @xmath2 . by inverting our new relationship
, we provide sonic mach number predictions for kpc - scale observations of local group galaxies as well as unresolved observations of local and high - redshift disk and starburst galaxies that do not have independent , reliable estimates for the turbulent cloud mach number . |
light - by - light ( lbl ) scattering is a prediction of the quantum theory @xcite which thus - far has not been directly observed , mainly due to smallness of the cross section . on the other hand , the process of @xmath4 fusion ( by quasi - real photons @xmath1 or virtual photons @xmath2 ) into leptons and hadrons
has been observed at nearly all high - energy colliders , see e.g. @xcite for reviews .
the two phenomena lbl scattering and @xmath5 fusion must be related by causality , similar to how the refraction index of light is related to its absorption in the kramers - kronig relation .
the main goal of this work is to establish such relations and use them to investigate the structure of hadrons in the realm of quantum chromo - dynamics ( qcd ) .
the electromagnetic interaction provides a clean probe and the two - photon state allows to produce hadrons with nearly all quantum numbers ( with @xmath6 ) , in contrast to the well studied single - photon scattering or production processes , which only accesses the vector states .
when producing exclusive final states such as in the @xmath7 process , one accesses meson transition form factors ( ffs ) , which are some of the simplest observables where the approach to the asymptotic limit of qcd is studied along with the quark content of mesons described by distribution amplitudes ( das ) .
the non - perturbative dynamics of qcd is also playing a profound role in these ffs at low momentum transfers .
for example , the transition ffs of the @xmath8 and @xmath9 mesons depend on the interplay of various symmetry breaking mechanisms in qcd , i.e. : @xmath10 symmetry breaking @xcite , dynamical and explicit chiral symmetry breaking . in addition , the @xmath7 transition ffs are important for providing and improving constraints on the light - by - light hadronic contribution to the anomalous magnetic moment of the muon , @xmath11 .
the hadronic contributions to @xmath11 are at present the major uncertainty in the search for new , beyond standard model , physics in this high - precision quantity @xcite . in recent years , new experiments at high luminosity @xmath12 colliders such as babar and belle
have vastly expanded the field of @xmath3 physics .
the result of a measurement of the @xmath13 ff at large momentum transfers by the babar collaboration @xcite came as a surprise , as this form factor seems to rise much faster than the perturbative qcd predictions for momentum transfers up to 40 gev@xmath14 .
a @xmath3 physics program is planned now by the bes - iii collaboration @xcite , which will allow to provide high - statistics results at intermediate momentum transfers for a multitude of @xmath15 observables . in this work
we use the dispersion theory to relate the two phenomena of lbl scattering and @xmath0 fusion , and express the low - energy lbl scattering as integrals over the @xmath0-fusion cross sections , where one photon is real while the second may have arbitrary ( space - like ) virtuality .
these integrals , or ` sum rules ' , lead to interesting constraints on @xmath3 decay widths or @xmath0 transition ffs of @xmath16 states , and more general meson states .
the first sum rule of this type involves the helicity - difference cross - section for real photons and reads as : @xmath17 = 0 , \label{eq : sr0}\end{aligned}\ ] ] where @xmath18 is the total energy squared , @xmath19 is the first inelastic threshold for the @xmath3 fusion process , and the subscripts @xmath20 or @xmath21 for the @xmath3 cross sections indicate the total helicity of the state of two circularly polarized photons .
this sum rule was originally production appears on the right - hand side ( _ rhs _ ) of eq .
( [ eq : sr0 ] ) , while integration on the _ lhs _ starts at the 2@xmath22 production threshold . that version would be fully compatible with eq .
( [ eq : sr0 ] ) , if it were not for the sign of the @xmath23 contribution obtained in @xcite . ] inferred @xcite from the the gerasimov drell
hearn ( gdh ) sum rule , using the fact that the photon has no anomalous moments . parameterizing the lowest energy lbl interaction by means of an effective lagrangian ( which contains operators of dimension eight at lowest order ) as @xmath24 with @xmath25 and @xmath26 being the electromagnetic field strength and its dual , one finds sum rules for the lbl low - energy constants ( lecs ) @xcite : @xmath27 where the subscripts @xmath28 or @xmath29 indicate if the colliding photons are polarized parallel or perpendicular to each other .
while the gdh - type sum rule provides a stringent constraint on the polarized @xmath3 fusion , the sum rules for the lecs allow one in principle to fully determine the low - energy lbl interaction through measuring the linearly polarized @xmath3 fusion . in this work
we extend the gdh type sum rule to the case where one of the colliding photons is virtual , with arbitrary ( space - like ) virtuality .
furthermore , we find two additional sum rules , involving the longitudinally polarized @xmath0 cross sections .
all details of sum rule derivation are gathered in sec .
[ sec2 ] . in sec .
[ pert ] , all of the newly derived sum rules are verified at leading order in scalar and spinor quantum electrodynamcis ( qed ) .
next we apply these results to the @xmath4 fusion to mesons . using the available data , we quantitatively study the new sum rules derived in this paper for the case of production of light quark mesons as well as mesons containing charm quarks , both by real photons in sec . [ sec8 ] , and by virtual photons in sec .
we demonstrate the intricate cancellations that must occur among the ( pseudo- ) scalar , tensor , and axial - vector mesons in order to satisfy these sum rules . in the case of production of virtual photons
, we use these relations to provide estimates of hitherto unmeasured @xmath0 transition form factors of tensor mesons , such as @xmath30 and @xmath31 .
the conclusion and outlook is given in sec .
[ sec : concl ] .
the appendices contain ( [ app : gaga ] ) a review of the kinematical notations and @xmath32 cross section conventions ; ( [ app : perturb ] ) expressions for the tree - level @xmath4 cross sections for the case of scalar and spinor qed ( sec .
[ app : perturb ] ) ; ( [ sec7 ] ) general formalism for the @xmath33 transitions with different quantum numbers ( @xmath34 ) , i.e. : pseudo - scalars ( @xmath35 ) , scalars ( @xmath36 ) , axial - vectors ( @xmath37 ) , and tensors ( @xmath38 ) .
in the most general case we consider the forward scattering of virtual photons on virtual photons : @xmath39 where @xmath40 , @xmath41 are photon four - momenta , and @xmath42 ( @xmath43 ) are the helicities of the initial ( final ) virtual photons , which can take on the values @xmath44 ( transverse polarizations ) and zero ( longitudinal ) .
the total helicity in the @xmath4 c.m .
system is given by @xmath45 .
to define the kinematics , we firstly introduce the photon virtualities @xmath46 , @xmath47 , the mandelstam invariants : @xmath48 , @xmath49 , and the following crossing - symmetric variable : @xmath50 such that @xmath51 , @xmath52 .
the @xmath53 forward scattering amplitudes , denoted as @xmath54 , are functions of @xmath55 , @xmath56 , @xmath57 .
parity invariance ( @xmath58 ) and time - reversal invariance ( @xmath59 ) imply the following relations among the matrix elements with different helicities : @xmath60 which leaves out only eight independent amplitudes @xcite : @xmath61 we next look at the constraint imposed by crossing symmetry , which requires that the amplitudes for the process ( [ eq : process1 ] ) equal the amplitudes for the process where the photons with e.g. label 2 are crossed : @xmath62 as under photon crossing @xmath63 , one obtains @xmath64 it becomes convenient to introduce amplitudes which are either even or odd in @xmath55 ( at fixed @xmath56 and @xmath57 ) .
one easily verifies that the following six amplitudes are _ even _ in @xmath55 : @xmath65 whereas the following two amplitudes are _ odd _ in @xmath55 : @xmath66 the optical theorem allows one to relate the absorptive part of the @xmath53 forward scattering amplitudes to cross sections for the process @xmath67 , where x stands for any possible final state . denoting the absorptive part as @xmath68 the optical theorem yields : @xmath69 where @xmath70 denotes the invariant amplitude for the process @xmath71 as a result
, the absorptive parts are expressed in terms of eight independent @xmath67 cross sections ( see ref .
@xcite for details ) : [ eq : vcross ] @xmath72 where the virtual photon flux factor is defined through @xmath73 in eq .
( [ eq : vcross ] ) , @xmath74 are the @xmath67 cross sections for total helicity 0 ( 2 ) respectively , and @xmath75 are the cross sections for linear photon polarizations with both photon polarization directions parallel ( perpendicular ) to each other respectively .
the remaining cross sections ( positive definite quantities @xmath76 ) involve either one transverse ( @xmath59 ) and one longitudinal ( @xmath77 ) photon polarization , or two longitudinal photon polarizations , with @xmath78 and @xmath79 related as : @xmath80 the quantities @xmath81 denote interference cross sections ( which are not sign - definite ) with either both photons transverse ( @xmath82 ) , or for one transverse and one longitudinal photon ( @xmath83 ) , where the superscript @xmath84 indicates the combinations which are odd in @xmath55 .
the principle of ( micro-)causality is known to translate into exact statements about analytic properties of the scattering amplitude in the complex energy plane . in our case this principle translates into the statement of analyticity of the forward @xmath4 scattering amplitude in the entire @xmath55 plane , except for the real axis where the branch cuts associated with particle production are located . assuming that the threshold for particle production is @xmath85
, one can write down the usual dispersion relations , in which the amplitude is given by integrals over the non - analyticities , which in this case are branch cuts extending from @xmath86 to @xmath87 .
finally , for amplitudes that are even or odd in @xmath55 we can write ( for any fixed values of @xmath88 ) : @xmath89 where @xmath90 is an infinitesimal positive number .
these dispersion relations are derived with the provision that the integrals converge .
if they do not , subtractions must be made ; e.g. , the once - subtracted dispersion relation for the even amplitudes reads : @xmath91 we are thus led to examine the high - energy behavior ( @xmath92 at fixed @xmath93 ) of the absorptive parts given by eq .
( [ eq : vcross ] ) . in ref .
@xcite , a regge pole model assumption for the high - energy asymptotics of the light - by - light forward amplitudes yielded : @xmath94 where @xmath95 is the intercept of the pomeron trajectory , and @xmath96 is the intercept of the pion trajectory .
this means that for all the even amplitudes , except @xmath97 , one can only use the subtracted dispersion relation eq .
( [ eq : drsubt ] ) .
we therefore need the information about these amplitudes at zero energy @xmath55 . anticipating the discussion of the low - energy expansion of the lbl scattering
, we can state that at @xmath98 these amplitudes vanish when one of the photons is real [ cf .
( [ eq : lex ] ) ] . using eq .
( [ eq : vcross ] ) then to substitute the cross sections in place of the absorptive parts , we obtain the following sum rules for the case of one real and one virtual photon ( when the virtual photon flux factor becomes @xmath99 ) : @xmath100 we can not write such a subtracted sum rule for @xmath101 , since it trivially vanishes when one of the photons is real . instead ,
considering an unsubtracted dispersion relation , we find the following sum rule : m_00,00 ( ) & = & _ _ 0^ d ^ , with @xmath102 . at least in perturbative qed calculations ( cf .
appendix [ app : perturb ] ) , the above integral converges which seems to validate this sum rule in a renormalizable , perturbative field theory .
we emphasize however that this observation is in contradiction with the expectation of non - convergence from the regge pole model shown above .
a validation of this sum rule in non - perturbative field theory , particularly in qcd , is therefore an open issue .
for all the remaining amplitudes the asymptotic behavior of eq .
( [ eq : highen ] ) justifies the use of unsubtracted dispersion relations which , upon substituting eq .
( [ eq : vcross ] ) , lead to the following sum rules , valid for both photon virtual : @xmath103 } { \nu^{\prime \ , 2 } - \nu^2-i0^+ } , \label{eq : sr1 } \\
m_{++,00}(\nu ) - m_{0+,-0}(\nu ) & = & \frac{8\nu}{\pi } \int_{\nu_0}^\infty \!\!d \nu^\prime \,\frac{\sqrt{x^\prime } \ , \tau^a_{tl } ( \nu^\prime)}{\nu^{\prime \ , 2 } - \nu^2-i0^+ } , \label{eq : sr1b}\\ m_{++,00 } ( \nu)+ m_{0+,-0 } ( \nu ) & = & \frac{8}{\pi } \int_{\nu_0}^\infty \!\!d \nu^\prime \ , \frac{\nu^\prime \sqrt{x^\prime}\ , \tau_{tl } ( \nu^\prime ) } { \nu^{\prime \ , 2 } - \nu^2-i0^+ } , \label{eq : sr2 } \end{aligned}\ ] ] where the dependence on virtualities @xmath56 , @xmath57 is tacitly assumed .
[ eq : sumrules ] the above sum rules , relating all the forward @xmath4 elastic scattering amplitudes to the energy integrals of the @xmath4 fusion cross sections , should hold for any space - like photon virtualities in the unsubtracted cases , and for one of the virtualities equal to zero in the subtracted cases . in the following we examine the low - energy expansion of these sum rules . to obtain more specific relations from the sum rules established in eq .
( [ eq : sumrules ] ) , we parametrize the low - energy ( small @xmath55 ) behavior of the @xmath53 forward scattering amplitudes @xmath104 . at lowest order in the energy , the self - interactions of the electromagnetic field
are described by an effective lagrangian ( of fourth order in the photon energy and/or momentum , and fourth order in the electromagnetic field ) : @xmath105 where @xmath106 , @xmath107 , and where @xmath108 are two low - energy constants ( lecs ) which contain the structure dependent information .
it is often referred to as euler - heisenberg lagrangian due to the seminal work @xcite . at the next order in energy ,
one considers the terms involving two derivatives on the field tensors , corresponding with the sixth order in the photon energy and/or momentum .
writing down all such dimension - ten operators and reducing their number using the antisymmetry of the field tensors , the bianchi identities , as well as adding or removing total derivative terms , we find that there are 6 independent terms at that order , which we choose as : @xmath109 where @xmath110 are the new lecs arising at this order . only @xmath111 and @xmath112 appear in the case of real photons .
we can now specify the low - energy limit of the light - by - light scattering amplitudes in terms of the lecs describing the low - energy self - interactions of the electromagnetic field : [ eq : lex ] @xmath113 \nn\\ & + & 8 \nu^2\left [ 8 ( c_1 + c_2 ) + \left ( q_1 ^ 2 + q_2 ^ 2 \right ) ( -c_3 + 3c_{6}+4c_{7 } ) + { \cal o}(q^4 ) \right ] + { \cal o}(\nu^4 ) , \label{eq : lex4 } \\ m_{++,-- } & = & q_1 ^ 2q_2 ^ 2 \left [ 64c_2 + 4(q_1 ^ 2+q_2 ^ 2)(-c_3 + 2c_6-c_8)+{\cal o}(q^4 ) \right ] \nn\\ & + & 8\nu^2 \left [ 8 ( c_1 - c_2 ) + \left ( q_1 ^ 2 + q_2 ^ 2\right ) ( c_{6}+4c_7 ) + { \cal o}(q^4 ) \right ] + { \cal o}(\nu^4 ) , \label{eq : lex5 } \\ m_{0+,0 + } & = & q_1 ^ 2q_2 ^ 2\left[-32 c_1 + 4q_1 ^ 2 c_8 + 4(q_1 ^ 2+q_2 ^ 2)(c_3 + 4c_4 + 2c_5 - 2c_6 - 4c_7 ) + { \cal
o}(q^4 ) \right ] \nn\\ & + & \nu^2 \left [ -4q_1 ^ 2 c_{8 } + { \cal o}(q^4 ) \right ] + { \cal o}(\nu^4 ) , \\
m_{+0,+0 } & = & q_1 ^ 2q_2 ^ 2\left[-32 c_1 + 4q_2 ^ 2 c_8 + 4(q_1 ^ 2+q_2 ^ 2)(c_3 + 4c_4 + 2c_5 - 2c_6 - 4c_7 ) + { \cal
o}(q^4 ) \right ] \nn\\ & + & \nu^2 \left [ -4q_2 ^ 2 c_{8 } + { \cal o}(q^4 ) \right ] + { \cal o}(\nu^4 ) , \\
m_{00,00 } & = & q_1 ^ 2 q_2 ^ 2 \left [ 96 c_1 + 4(q_1 ^ 2+q_2 ^ 2)(-2c_3 - 4c_4 - 2c_5 + 6c_6 + 12c_7-c_8 ) + { \cal o}(q^4 )
\right ] \nn\\ & + & { \cal o}(\nu^2 ) , \label{eq : lex7 } \\ m_{++,++ } - m_{+-,+- } & = & 8 \nu q_1 ^ 2 q_2 ^ 2 \left [ -c_3 - 4 c_{5 } + c_{8 } + { \cal o}(q^2 ) \right ] + \n^3\left[-64c_4 + { \cal o}(q^2 ) \right]+ { \cal o}(\nu^5 ) , \label{eq : lex1 } \\
m_{++,00 } - m_{0+,-0 } & = & \nu q_1 q_2 \left [ - 64 c_1 + \left(q_1 ^ 2 + q_2 ^ 2 \right ) ( 4c_3 - 16c_{6}-32c_{7}+4c_{8 } ) + { \cal
o}(q^4 ) \right ] \nn\\ & + & { \cal o}(\nu^3 ) , \label{eq : lex2 } \\
m_{++,00 } + m_{0+,-0 } & = & q_1 ^ 3 q_2 ^ 3 \left [ 4c_5 - 12c_{8 } + { \cal o}(q^2 ) \right ] + 4\n^2 q_1q_2 \left [ 2 c_3 + 16c_4 + 4c_5+c_8 + { \cal o}(q^2 ) \right ] \nn\\ & + & { \cal o}(\nu^4 ) .
\label{eq : lex3 } \label{eq : lex6 } \end{aligned}\ ] ] these expressions can be treated as a simultaneous expansion in @xmath55 and the virtualities @xmath114 of the _ lhs _ of the sum rules eq .
( [ eq : sumrules ] ) .
concerning the @xmath115 dependence , it is important that the leading in @xmath55 term , in any of the amplitudes , is proportional to @xmath116 and hence vanishes for at least one real photon .
the latter statement is valid for any values of virtualities , not just when they are small .
for example , let us show for the amplitude @xmath117 its leading term in @xmath55 is proportional to the combination @xmath118 , to all orders in @xmath119 and @xmath120 .
since all photons are transversely polarized the only non - vanishing structures involving polarization vectors of photons @xmath121 are their mutual scalar products @xmath122 . due to gauge invariance
, the electromagnetic fields enter the lagrangian through the field tensor @xmath123 , which contributes to the amplitude as @xmath124 .
thus an arbitrary term in the effective lagrangian contributes to @xmath117 as : @xmath125 where the tensor @xmath126 is constructed from four - vectors @xmath127 and the metric tensor .
since this amplitude is odd with respect to @xmath55 , it is required to be proportional to at least @xmath128 . assuming that one factor @xmath55 comes from contraction of two of the @xmath129 s in eq .
( [ eq : qt ] ) , we are left with @xmath130 .
now , if we suppose that @xmath40 is contracted with @xmath41 we obtain an extra power of @xmath55 , and such an amplitude vanishes when taking the limit @xmath131 . thus , both @xmath40 and @xmath41 must be contracted with another @xmath40 and @xmath41 respectively , giving a global factor @xmath118 .
we are now in position to examine the sum rules in eq .
( [ eq : sumrules ] ) order by order in @xmath55 . for this
we expand the _
rhs _ of eq .
( [ eq : sumrules ] ) using @xmath132 . as the result we obtain from eqs .
( [ eq : sr1],[eq : sr1b],[eq : sr2 ] ) the following set of super - convergence relations , valid for at least one real photon ( e.g. , @xmath133 , @xmath134 ) : @xmath135_{q_2 ^ 2 = 0 } , \label{s0rule4 } \\ 0 & = & \int\limits_{s_0}^\infty d s \ , \left [ \frac{\tau_{tl } ( s , q_1 ^ 2 , q_2 ^ 2 ) } { q_1 q_2 } \right]_{q_2 ^ 2 = 0}. \label{s0rule6}\end{aligned}\ ] ] and the following set of sum rules for the lecs of the dimension-8 ( euler - heisenberg ) lagrangian , valid when both photons are quasi - real : @xmath136_{q_1 ^ 2 = q_2 ^ 2 = 0 } , \label{s0rule3 } \\ & = & \frac{1 } { 8 \pi } \int\limits_{s_0}^\infty d s\ , \left [ \frac{\si_{ll } ( s , q_1 ^ 2 , q_2 ^ 2 ) } { q_1 ^ 2 q_2 ^ 2 } \right]_{q_1 ^ 2 = q_2 ^ 2 = 0 } , \label{s1rule8 } \\
c_2 & = & \frac{1}{8 \pi } \int\limits_{s_0}^{\infty } \frac{d s}{s^2}\ , \sigma_\perp(s,0,0 ) , \label{s0rule2}\end{aligned}\ ] ] where @xmath137 .
we emphasize again that , unlike the other sum rules , the sum rule of eq .
( [ s1rule8 ] ) is only shown to hold in perturbative field theory .
there are as well the sum rules for the lecs of the dimension-10 lagrangian , most notably : @xmath138 but presently they are of far lesser importance and we do not write them out here explicitly . let us remark again that the relation of eq .
( [ s0rule1 ] ) , obtained by combining eqs .
( [ eq : sr1 ] ) and ( [ eq : lex1 ] ) , is essentially a gdh sum rule for the photon target , see @xcite . for large virtuality @xmath56
, it leads to the sum rule for the photon structure function @xmath139 @xcite : @xmath140 . the sum rules in eqs .
( [ s1rule1 ] ) and ( [ s0rule2 ] ) , first established in @xcite , are obtained by combining eqs .
( [ eq : sr4 ] ) with ( [ eq : lex4 ] ) and eqs .
( [ eq : sr5 ] ) with ( [ eq : lex5 ] ) , respectively .
all the other relations presented above are new . in the following section we verify these sum rules in qed at leading order in the fine - structure constant @xmath141 .
we will subsequently discuss a pair production in scalar qed ( e.g. , born approximation to @xmath142 ) and in spinor qed ( @xmath143 where @xmath129 stands for a charged lepton or a quark ) .
the response functions for the case of scalar qed at lowest order in the electromagnetic coupling can be found in appendix [ app : scalar ] .
we firstly study the three sum rules of eqs .
( [ s0rule1 ] , [ s0rule4 ] , [ s0rule6 ] ) for the case of one real or quasi - real photon ( @xmath144 ) and for arbitrary space - like virtuality ( @xmath145 ) of the other photon . to better see the cancellation which must take place in these sum rules between contributions at low and higher energies , we show the integrands of the three sum rules in figs .
[ fig : scalarsr1 ] , [ fig : scalarsr2 ] , [ fig : scalarsr3 ] multiplied by @xmath18 . in this way , when plotted logarithmically , one can clearly see how the low and high energy contributions cancel each other . for the sum rule of eq .
( [ s0rule4 ] ) , we denote the integrand as : @xmath146_{q_2 ^ 2 = 0}. \label{eq : inti}\end{aligned}\ ] ] all three sum rules of eqs .
( [ s0rule1 ] , [ s0rule4 ] , [ s0rule6 ] ) are exactly verified in scalar qed for arbitrary space - like values of @xmath56 .
one notices from figs .
[ fig : scalarsr1 ] , [ fig : scalarsr2 ] , [ fig : scalarsr3 ] that for larger values of @xmath56 the zero crossing of the integrands shifts to larger values of @xmath18 , requiring higher energy contributions for the cancellation to take place . for the helicity difference sum rule of eq .
( [ s0rule1 ] ) , one notices that at low energies @xmath147 dominates while with increasing energies @xmath148 overtakes . besides exactly verifying the sum rules which integrate to zero , we can also use the above derived sum rules to study the low - energy coefficients for light - by - light scattering in scalar qed . using eqs .
( [ s1rule1 ] , [ s0rule2 ] ) , we obtain for the tree - level contributions to the lowest order coefficients @xmath149 and @xmath150 in scalar qed : @xmath151 the response functions for the case of spinor qed at lowest order in the electromagnetic coupling can be found in appendix [ app : spinor ] .
we again study the three sum rules of eqs .
( [ s0rule1 ] , [ s0rule4 ] , [ s0rule6 ] ) for the case of one real or quasi - real photon ( @xmath144 ) for different space - like virtualities of the other photon . as the tree level contribution to @xmath152 in spinor qed vanishes for one quasi - real photon ,
one notices that the sum rule of eq .
( [ s0rule6 ] ) is trivially satisfied . for the sum rules involving the helicity difference of eq .
( [ s0rule1 ] ) , and involving the integrand @xmath153 of eq .
( [ eq : inti ] ) , we show the corresponding integrands multiplied by @xmath18 in figs . [
fig : spinorsr1 ] , [ fig : spinorsr3 ] for the case of one real or quasi - real photon and for different virtualities of the other photon .
we again verify that the sum rules involve an exact cancellation between low and high energy contributions .
using eqs .
( [ s1rule1 ] , [ s0rule2 ] ) , we obtain for the tree - level contributions to the lowest order coefficients @xmath149 and @xmath150 for light - by - light scattering in spinor qed : @xmath154 in these case we also were able to verify the sum rule in eq .
( [ s1rule9 ] ) , yielding c_4 = - , in agreement with the result obtained in ref .
@xcite for the low - energy photon - photon scattering . a more detailed study of the lbl sum rules in field theory , including loop effects , production of vector bosons , etc .
, is the subject of our forthcoming publication @xcite .
in the previous section , the sum rules of eqs .
( [ s0rule1 ] , [ s0rule4 ] , [ s0rule6 ] ) integrating to zero have been shown to hold exactly in perturbative calculations ( e.g. , in qed or qcd in the perturbative regime ) .
however as their derivation is general , their realization in qcd , in its non - perturbative regime , allows to gain insight in the @xmath155 cross - sections .
this was illustrated in ref .
@xcite for the sum rule of eq .
( [ s0rule1 ] ) . in the remainder of this paper , we will elaborate on the discussion of ref .
@xcite and extend it to the other sum rules presented above .
the required non - perturbative input for the absorptive parts of the sum rules are the @xmath155 response functions . in this paper
, we will perform a first analysis by estimating the hadronic contributions to these response functions by the corresponding @xmath156 ( with @xmath104 a meson ) production processes , which are described in terms of the @xmath156 transition form factors . in appendix [ sec7 ]
we detail the formalism and the available data for the @xmath156 transition ffs , and successively discuss the @xmath157-even pseudo - scalar ( @xmath158 ) , scalar ( @xmath159 ) , axial - vector ( @xmath160 ) , and tensor ( @xmath161 ) mesons .
we first consider the helicity sum rule of eq .
( [ s0rule1 ] ) with two real photons producing a meson , as well as the sum rules of eq .
( [ s0rule2 ] ) for the mesonic contributions to the low - energy constants @xmath149 and @xmath150 describing the forward light - by - light scattering amplitude .
when producing mesons , the sum rules will hold separately for states of given intrinsic quantum numbers . therefore , we will separately study the sum rule contributions for light quark isovector mesons ( table [ table_isovector ] ) , for light quark isoscalar mesons ( table [ table_isoscalar ] ) , as well as @xmath162 mesons ( table [ table_ccbar ] ) .
for the isoscalar mesons , one could in principle separate the contributions according to singlet or octet states ( or alternatively according to @xmath163 or @xmath164 states ) .
the corresponding mesons involve mixings however which complicate such separation , as this mixing is not known well enough for some of the states .
we will postpone such a separation for a future work and add all isoscalar meson contributions in the present work .
the pseudo - scalar mesons contribute to the helicity-0 cross section only , given by eq .
( [ eq : pscross ] ) .
the corresponding contributions to the helicity sum rule of eq .
( [ s0rule1 ] ) as well as the @xmath149 and @xmath150 sum rules are shown for the @xmath23 in table [ table_isoscalar ] , for the @xmath165 in table [ table_isoscalar ] , and for the @xmath166 state in table [ table_ccbar ] .
besides the pseudo - scalar mesons , also scalar mesons can only contribute to @xmath147 .
we show the contributions of the @xmath167 in table [ table_isoscalar ] , for the @xmath168 and @xmath169 in table [ table_isoscalar ] , and for the @xmath170 state in table [ table_ccbar ] . for the scalar mesons
, only the @xmath169 state gives a sizable contribution due to its large @xmath171 decay width .
for the helicity sum rule , one notices that in order to compensate the large negative contribution from the pseudo - scalar mesons , and to lesser extent from the scalar meson states , an equal strength is required in the helicity-2 cross section , @xmath148 . for light quark mesons , the dominant feature of the helicity-2
cross section in the resonance region arises from the multiplet of tensor mesons @xmath172 , @xmath31 , and @xmath173 . for @xmath162 tensor mesons ,
the dominant tensor contribution is given by the @xmath174 state .
measurements at various @xmath12 colliders , notably the recent high statistics measurements by the belle collaboration of the @xmath3 cross sections to @xmath175 , @xmath176 , @xmath177 , and @xmath178 channels @xcite have allowed to accurately establish their parameters . for the light quark mesons , the experimental analyses of decay angular distributions have found @xcite that the tensor mesons are produced predominantly ( around 95% or more ) in a state of helicity @xmath179 .
we will therefore assume in all of the following analyses that @xmath180 , and that @xmath181 in tables [ table_isovector ] , [ table_isoscalar ] , [ table_ccbar ] .
we show all tensor meson contributions to the helicity difference sum rule as well as the @xmath108 sum rules for which the @xmath171 decay widths are known . for the isovector meson contributions to the helicity sum rule , shown in table [ table_isovector ] ,
we conclude that the lowest isovector tensor meson composed of light quarks , @xmath31 , compensates to around 70% the contribution of the @xmath23 , which is entirely governed by the chiral anomaly .
for the isoscalar states composed of light quarks , the cancellation is even more remarkable : the sum of @xmath172 and @xmath173 , within the experimental accuracy , entirely compensates the combined contribution of the @xmath8 and @xmath9 mesons .
.@xmath3 sum rule contributions of the light quark isovector mesons based on the present pdg values @xcite of the meson masses ( @xmath182 ) and their @xmath171 decay widths @xmath183 .
fourth column : @xmath184 sum rule of eq .
( [ s0rule1 ] ) .
fifth , sixth columns : @xmath108 sum rules of eqs .
( [ s1rule1 ] , [ s0rule2 ] ) respectively . [ cols="^,^,^,^,^,^ " , ] the process @xmath185 , describing the transition from an initial state of two virtual photons to a tensor meson @xmath186 ( @xmath161 ) with mass @xmath187 and helicity @xmath188 ( defined along the direction of @xmath189 ) , is described by five independent structures @xcite , and can be parameterized as : @xmath190 \ , \frac{\nu}{m_t } \,t^{(2)}(q_1 ^ 2 , q_2 ^ 2 ) \right .
\nonumber \\ & & + \ , r^{\nu \alpha}(q_1 , q_2 ) ( q_1 - q_2)^\beta \left ( q_1^\mu + \frac{q_1 ^ 2}{\nu } q_2^{\mu }
\right ) \ , \frac{1}{m_t } \ , t^{(1)}(q_1 ^ 2 , q_2 ^ 2 ) \nonumber \\ & & + r^{\mu \alpha}(q_1 , q_2 ) ( q_2 - q_1)^\beta \left ( q_2^\nu + \frac{q_2 ^ 2}{\nu } q_1^{\nu }
\right ) \ , \frac{1}{m_t } \ , t^{(1)}(q_2 ^ 2 , q_1 ^ 2 ) \nonumber \\ & & + \ , r^{\mu \nu}(q_1 , q_2 ) ( q_1 - q_2)^\alpha \ , ( q_1 - q_2)^\beta \ , \frac{1}{m_t } \ , t^{(0 , t)}(q_1 ^ 2 , q_2 ^ 2 ) \ , \nonumber \\ & & \left . + \ , \left ( q_1^\mu + \frac{q_1 ^ 2}{\nu } q_2^{\mu } \right ) \left ( q_2^\nu + \frac{q_2 ^ 2}{\nu } q_1^{\nu } \right ) ( q_1 - q_2)^\alpha ( q_1 - q_2)^\beta \ , \frac{1}{m_t^3 } \ , t^{(0 , l)}(q_1 ^ 2 , q_2 ^ 2 ) \right\ } , \label{eq : tensorff}\end{aligned}\ ] ] where @xmath191 is the polarization tensor for the tensor meson with four - momentum @xmath192 and helicity @xmath193 . furthermore in eq . ( [ eq : tensorff ] ) @xmath194 are the @xmath195 transition form factors , for tensor meson helicity @xmath193 . for the case of helicity zero
, there are two form factors depending on whether both photons are transverse ( superscript @xmath59 ) or longitudinal ( superscript @xmath77 ) . from eq .
( [ eq : tensorff ] ) , we can easily calculate the different helicity amplitudes as : @xmath196 the transverse ffs @xmath197 and @xmath198 at @xmath199 describe the two - photon decay widths of the tensor meson with helicities @xmath179 and @xmath200 respectively : in this work , we study the sum rules involving cross sections for one real photon and one virtual photon . for one quasi - real photon ( @xmath144 )
, we can obtain from the above helicity amplitudes and using eq .
( [ eq : abs ] ) the tensor meson contributions to the response functions of eq .
( [ eq : vcross ] ) as : @xmath202_{q_2 ^ 2 = 0 } & = & \delta(s - m_t^2 ) \ , 16 \ , \pi^2 \frac{5 \ , \gamma_{\gamma \gamma}({\cal t}(\lambda = 2))}{m_t }
\ , \left ( 1 + \frac{q_1 ^ 2}{m_t^2 } \right ) \ , \left [ \frac{t^{(2)}(q_1 ^ 2 , 0)}{t^{(2)}(0 , 0 ) } \right]^2 \ , , \nonumber \\ \left [ \sigma_0 \right]_{q_2 ^ 2 = 0 } & = & \delta(s - m_t^2 ) \ , 16 \ ,
\pi^2 \ , \frac{5 \ , \gamma_{\gamma \gamma}({\cal t}(\lambda = 0))}{m_t } \ , \left ( 1 + \frac{q_1 ^ 2}{m_t^2 } \right)^3 \ , \left [ \frac{t^{(0 , t)}(q_1 ^ 2 , 0)}{t^{(0 , t)}(0 , 0 ) } \right]^2 \ , , \nonumber \\ \left [ \sigma_\parallel \right]_{q_2 ^ 2 = 0 } & = & \left [ \frac{1}{2 }
\sigma_2 + \sigma_0 \right]_{q_2 ^ 2 = 0 } \ , , \nonumber \\ \left [ \sigma_\perp
\right]_{q_2 ^ 2 = 0 } & = & \left [ \frac{1}{2 } \sigma_2 \right]_{q_2 ^ 2 = 0 } \ , , \nonumber \\ \left [ \sigma_{lt } \right]_{q_2 ^ 2 = 0 } & = & \delta(s - m_t^2 ) \ , 8 \ , \pi^3 \alpha^2 \ , \frac{q_1 ^ 2}{m_t^2 } \ , \left ( 1 + \frac{q_1 ^ 2}{m_t^2 } \right ) \ , \left [ t^{(1)}(q_1 ^ 2,0 ) \right]^2\ , , \nonumber \\ \left [ \frac{1}{q_1 q_2 } \tau_{tl } \right]_{q_2 ^ 2 = 0 } & = & \delta(s - m_t^2 ) \,8 \ , \pi^3 \alpha^2 \ , \frac{1}{m_t^2 } \ , \left ( 1 + \frac{q_1 ^ 2}{m_t^2 } \right ) \nonumber \\ & \times & \left\ { \frac{2}{3 } \left ( 1 + \frac{q_1 ^ 2}{m_t^2 } \right)^2 t^{(0 , t)}(q_1 ^ 2 , 0 ) \ , t^{(0 , l)}(q_1 ^ 2 , 0 ) - \frac{1}{2 } t^{(1)}(q_1 ^ 2,0 ) \ , t^{(1)}(0 , q_1 ^ 2 ) \right\ } \
, , \nonumber \\ \left [ \frac{1}{q_1 q_2 } \tau^a_{tl } \right]_{q_2 ^ 2 = 0 } & = & \delta(s - m_t^2 ) \,8 \ , \pi^3 \alpha^2 \ , \frac{1}{m_t^2 } \ , \left ( 1 + \frac{q_1 ^ 2}{m_t^2 } \right ) \nonumber \\ & \times & \left\ { \frac{2}{3 } \left ( 1 + \frac{q_1 ^ 2}{m_t^2 } \right)^2 t^{(0 , t)}(q_1 ^ 2 , 0 ) \ , t^{(0 , l)}(q_1 ^ 2 , 0 ) + \frac{1}{2 } t^{(1)}(q_1 ^ 2,0 ) \ , t^{(1)}(0 , q_1 ^ 2 ) \right\ } \ , , \nonumber \\ \left [ \sigma_{ll } \right]_{q_2 ^ 2 = 0 } & = & 0\ , .
\label{eq : tcross}\end{aligned}\ ] ] f. jegerlehner and a. nyffeler , phys .
* 477 * , 1 ( 2009 ) .
b. aubert _ et al . _ [ babar collaboration ] , phys .
d * 80 * , 052002 ( 2009 ) .
d. m. asner , t. barnes , j. m. bian , i. i. bigi , n. brambilla , i. r. boyko , v. bytev and k. t. chao _ et al .
_ , int . j. mod
. phys .
a * 24 * , s1 ( 2009 ) .
s. gerasimov and j. moulin , nucl .
b * 98 * , 349 ( 1975 ) .
s. j. brodsky and i. schmidt , phys .
b * 351 * , 344 ( 1995 ) .
roy , phys .
d * 9 * , 2631 ( 1974 ) .
v. pascalutsa , m. vanderhaeghen , phys . rev .
lett . * 105 * , 201603 ( 2010 ) .
s. d. bass , s. j. brodsky and i. schmidt , phys .
b * 437 * , 417 ( 1998 ) ; a.v .
efremov and o.v .
teryaev , phys .
b * 240 * , 200 ( 1990 ) ; s. d. bass , int .
j. mod .
phys . a * 7 * , 6039 ( 1992 ) ; s. narison , g. m. shore and g. veneziano , nucl .
b * 391 * , 69 ( 1993 ) .
d. a. dicus , c. kao and w. w. repko , phys .
d * 57 * , 2443 ( 1998 ) . v. pauk , v. pascalutsa , m. vanderhaeghen , in preparation .
p. achard _ et al .
_ [ l3 collaboration ] , phys .
b * 526 * , 269 ( 2002 ) .
p. achard _ et al .
_ [ l3 collaboration ] , jhep * 0703 * , 018 ( 2007 ) . r. n. cahn , phys . rev .
d * 35 * , 3342 ( 1987 ) ; phys .
d * 37 * , 833 ( 1988 ) . | relating the forward light - by - light scattering to energy weighted integrals of the @xmath0 fusion cross sections , with one real photon ( @xmath1 ) and one virtual photon ( @xmath2 ) , we find two new exact super - convergence relations .
they complement the known super - convergence relation based on the extension of the gdh sum rule to the light - light system .
we also find a set of sum rules for the low - energy photon - photon interaction .
all of the new relations are verified here exactly at leading order in scalar and spinor qed .
the super - convergence relations , applied to the @xmath0 production of mesons , lead to intricate relations between the @xmath3 decay widths or the @xmath0 transition form factors for ( pseudo- ) scalar , axial - vector and tensor mesons .
we discuss the phenomenological implications of these results for mesons in both the light - quark sector and the charm - quark sector . |
in recent years different methods @xcite have been proposed to map the structure and underlying dynamics of a given time series into an associated graph representation , with the aims of exploiting the modern tools of network science @xcite in the traditional task of time series analysis @xcite , thereby building a bridge between the two fields .
+ in this context , visibility graphs have been proposed @xcite as a tool to extract a graph from the relative positions of an ordered series , from which several graph features can be extracted and used for description and classification problems .
very recently we have advanced the concept of sequential visibility graph motifs @xcite , building on the idea of network motifs @xcite to explore the decomposition of visibility graphs into sequentially restricted subgraphs .
these motifs induce a graph - theoretical symbolization of a given time series into a sequence of subgraphs .
we have shown that the marginal distribution of the motif sequence -the so - called motif profile- is an informative feature to describe different types of dynamics and is useful in the task of classifying empirical time series . for large classes of dynamical systems , we were able to develop a theory to analytically compute the frequency of each motif when these are extracted from a so - called horizontal visibility graph ( hvg ) , this being a modified and simpler version of the original ( natural ) visibility graph ( vg ) which has often shown analytical tractability @xcite .
as a matter of fact , in the case of vgs to obtain analytical insight has shown to be a challenging task , and besides few exception @xcite most of the works that make use of this statistic are computational . in this paper
we bridge this gap and advance a theory to analytically compute the complete motif profile in the natural case ( vg motifs ) .
we focus on motifs of size @xmath0 as this was shown to be the simplest case which gives nontrivial results @xcite .
we validate this theory by deriving explicit motif profiles for several classes of dynamics which we show to be in good agreement with numerical simulations .
we also study the robustness of this feature when the time series is short and polluted with measurement noise , and compare its performance with the case of hvg motifs .
+ the rest of the paper is as follows : after recalling the definitions of natural and horizontal visibility graphs , in section ii we present the concept and main properties of sequential visibility graph motifs , as well as recalling the theoretical framework where the motif profile from the horizontal version was derived . in section iii
we focus on natural visibility and develop the theory to compute analytically the motif profile associated to processes where the dynamics are either bounded or unbounded .
we test this theory by assessing the predictions for different dynamical systems , and we also show that white noise with different marginals can be distinguished using the natural version instead of the horizontal one . in section
iv we show that the visibility graph motif profile is a robust feature in the sense of ( i ) having a fast convergence to asymptotic values for short series size @xmath1 and ( ii ) being robust against contamination with measurement noise ( white and colored ) . in section v we conclude .
let @xmath2 be a real - valued time series of @xmath1 data .
the natural visibility graph ( vg ) @xcite extracted from the series is the graph @xmath3 where each datum @xmath4 in the series is associated to a node @xmath5 ( thus @xmath6 and @xmath7 is a totally ordered set ) and an edge @xmath8 between node @xmath9 and node @xmath10 exists if @xmath11 $ ] for each @xmath12 such that @xmath13 .
this is called the visibility criterion , which for the natural version is indeed a convexity criterion .
analogously , the horizontal visibility graph ( hvg ) @xcite extracted from the series is the graph @xmath14 , with the same vertex set than @xmath15 and a smaller edge set @xmath16 , where an edge @xmath8 exists between nodes @xmath9 and @xmath10 if @xmath17 .
this visibility criterion is in turn an ordering one .
it was indeed shown that the hvg associated to a time series is invariant under monotonic transformations in the series @xcite , thus the hvg is an order statistic of the series .
@xmath18 is indeed a non crossing graph @xcite which by construction is always also a sub - graph of @xmath19 ( although @xmath19 is not in general planar ) .
both vg and hvg are connected graphs with a trivial hamiltonian path given by the sequence of vertices @xmath20 .
an illustration of how to construct a vg and hvg from a given time series is shown in figure [ fig:0 ] .
+ the set of sequential vg motifs of size @xmath21 , @xmath22 $ ] is defined as the set of all the @xmath23 possible sub - graphs with @xmath21 consecutive vertices along the hamiltonian path of a vg ( similarly , the set of hvg motifs of size @xmath21 is the set of all the @xmath24 admissible sub - graphs with @xmath21 consecutive vertices along the hamiltonian path of a hvg ) . accordingly ,
sequential vg motifs are also visibility graphs . for @xmath0
, there are in principle a total of @xmath25 possible motifs ( see table [ tab:1 ] for an enumeration ) , although as we will show below the number of admissible ones is just 6 . given a vg @xmath19 , its sequential motifs can be detected using a sliding window of size @xmath21 which slides along the hamiltonian path of the graph with @xmath26 consecutive overlapping steps . at each step a particular motif
is detected inside the window .
we can accordingly estimate @xmath27 , the frequency of appearance of a certain motif @xmath28 , and define the _
n - motif profile _ @xmath29 .
the process of extracting a vg / hvg and its sequential visibility motif set is illustrated in figure [ fig:1 ] ( the concept is analogous for hvg , although the set of admissible motifs is different in both cases ) . note that since @xmath30 can be understood as a discrete probability distribution and is therefore a vector with unit norm ( we use the @xmath31 norm here ) @xmath32 , the number of degrees of freedom of @xmath30 is @xmath33 ( again , as we will see below , in the case considered here it is even more reduced as the number of admissible motifs will be less than 8) .
+ in a recent work @xcite we introduced the concept of sequential hvg motifs and advanced a theory to compute in an exact way @xmath34 in the case of the hvg .
it was shown that the @xmath0 motif statistic was useful to discriminate across different types of dynamics .
the case of uncorrelated noise was shown to yield a universal motif profile , independent of the marginal distribution of the i.i.d .
process and this enabled the definition of randomness test .
we also found for some deterministic dynamics some _ forbidden motifs _ , which represented a persistent characteristic to test the randomness of a process ( note that if a motif of size @xmath21 does nt occur , then also all the motifs of size @xmath35 which incorporate that motif wo nt occur either ) .
since vg and hvg @xmath21 motif profiles are a temporally constrained feature ( they are evaluated along @xmath21 consecutive nodes on the hamiltonian path ) its extraction can be seen as a process of dynamic symbolization . under this perspective , the relation between hvg motifs of size @xmath0 and the so - called _ ordinal patterns _ ( ops ) @xcite was acknowledged in @xcite .
@xmath21-ops are symbols extracted from a time series representing the possible ranking output of @xmath21 consecutive data and are extracted from a specific time series by comparing the value of all the set of @xmath21 consecutive data along the series @xcite .
it was not unexpected to find a link between hvg @xmath21-motifs and @xmath21-ops as hvg is known to be an order statistic , much as ops .
indeed , in the particular case of a time series for which data do nt repeat @xmath36 there exists a mapping between each appearing hvg motif and a specific set of ordinal patterns @xcite ; in this scenario the forbidden motifs selected by the horizontal visibility are , in general , set of the so called _ forbidden ordinal patterns _ @xcite .
of course both vg and hvg motifs analysis can be applied without requiring any further assumption to time series taking values from finite sets ( namely when @xmath37 ) , while the ordinal patterns approach -based uniquely on the ranking statistics- require further assumptions in that case . here
we focus in the _ natural _ version of the algorithm and explore vg motifs instead .
as vgs are not invariant under monotonic transformations in the series @xcite , in general they depend on the marginal probability distribution of the time series and are not an order statistic . accordingly , there is no obvious correspondence between @xmath21-ops and vg @xmath21-motifs and both approaches in principle represent two independent symbolization methods that encode temporal information in a different way . in what follows we recall the theoretical framework for hvg motifs and in the next section we extend this theory to deal with vg motifs .
.the set of size-4 hvg motifs are defined according to a set of relations between 4 arbitrary consecutive data @xmath38 , @xmath39 $ ] in the time series . [ cols="^,^,^ " , ] we start by considering the fully chaotic logistic map @xmath40 $ ] , with invariant density @xmath41 . as
this process is deterministic , it fulfils a trivial markov property such that @xmath42 .
the hvg motif profile for this process was computed exactly in @xcite , here we compute the vg motif profile . before proceeding to compute each probability contribution , it is important to highlight a subtle point . since for this process @xmath43 $ ] is bounded , in principle one should use the inequality set depicted for bounded variables in table [ tab:2 ] .
however , in this particular case it is actually not necessary to explicitly consider the restriction @xmath43 $ ] .
as we will see in a moment , this is already taken into account implicitly in the computation of each integral and therefore one can use the ( simpler ) inequality set for unbounded variables given in table [ tab:1 ] . + we start by computing @xmath44 : @xmath45 which gives the following conditions : + @xmath46 + @xmath47 + which are satisfied for @xmath48 $ ] .
note at this point that the latter conditions are also satisfied in other ranges , but we only consider those ranges that belong to @xmath49 $ ] , and this is indeed the reason why we do nt need to use in this case the inequality set for bounded variables .
we thus have @xmath50}\left(\frac{1}{2},\frac{1}{2}\right)\simeq0.0591\ ] ] as @xmath51 by construction , we proceed by calculating @xmath52 : @xmath53 which gives the following conditions : + @xmath54 + @xmath55 + which are satisfied for @xmath56\cup[0.75,0.929]$ ] . therefore : @xmath57}\left(\frac{1}{2},\frac{1}{2}\right)+b_{\left[0.75,0.929\right]}\left(\frac{1}{2},\frac{1}{2}\right))\simeq0.289.\ ] ] similarly for @xmath58 we have @xmath59 which gives the following conditions : + @xmath60 + @xmath47 + which are satisfied for @xmath61 $ ] , and therefore @xmath62}\left(\frac{1}{2},\frac{1}{2}\right)\simeq0.3333\ ] ] for @xmath63 we have @xmath64 which gives the following conditions : + @xmath65 + @xmath55 + which are satisfied for @xmath66 $ ] , and thus @xmath67}\left(\frac{1}{2},\frac{1}{2}\right)+b_{\left[0.9239,0.9543\right]}\left(\frac{1}{2},\frac{1}{2}\right)\simeq0.0439\ ] ] for @xmath68 we have @xmath69 which gives the following conditions : + @xmath70 + @xmath47 + which are never satisfied and thus @xmath71 for @xmath72 we have @xmath73 which gives the following conditions : + @xmath74 + @xmath55 + which are satisfied for @xmath75\cup[0.9544,1]$ ] , and thus @xmath76}\left(\frac{1}{2},\frac{1}{2}\right)+b_{\left[0.95,1\right]}\left(\frac{1}{2},\frac{1}{2}\right))\simeq0.2741\ ] ] finally , by construction @xmath77 .
altogether , we find the vg motif profile of a fully chaotic logistic map @xmath78 note that while the result is in this case an approximation , our theory allows for numerical estimates with arbitrary precision ( the result is not exact because the location of fixed points of the map is only approximate , although this approximation is arbitrarily close to the true values ) .
for white uniform noise @xmath79 , @xmath80 $ ] we have a probability density @xmath81 and transition probability @xmath82 given by @xmath83 in this case the computations are more cumbersome since we need to make use of the inequality set for bounded variables described in table [ tab:2 ] .
the @xmath28th component of @xmath84 is given by @xmath85 where the sum runs over all the set @xmath86 of conditions @xmath87 which contribute to evaluate the probability for motif @xmath28 in table [ tab:2 ] .
all the integrals can nonetheless be solved analytically in closed form and give the following motif profile @xmath88 several comments are in order .
first , this profile is different from the one found for the chaotic logistic map .
second , this motif profile turns to be independent from the bounding values @xmath89 and @xmath90 ( where @xmath91 $ ] ) , meaning that white , uniform noise has a vg motif profile which is invariant under various transformations in the original distribution of the time series .
this is not a trivial property and is indeed a peculiarity of the uniform distribution , in other words the vg motif profile of white noise extracted a from bounded distribution _ generally _ depends on the bounds of the distribution .
for standard white gaussian noise @xmath79 , @xmath92 the probability density @xmath81 and transition probability @xmath82 are given by @xmath93 the @xmath28 component of @xmath84 is given by @xmath94 where @xmath95 are now the top and bottom conditions for the variable @xmath96 in motif @xmath28 reported in table for unbounded variables [ tab:1 ] .
the integrals can be evaluated numerically up to arbitrary precision and they give the following results @xmath97 at odds with what happens for hvg motifs @xcite , this result is different from the benchmark result for uniformly distributed white noise , thus there is not a universal vg motif profile for white noise as previously anticipated .
gaussian colored ( red ) noise with exponentially decaying correlations can be simulated using an @xmath98 process : @xmath99 where @xmath92 is gaussian white , and @xmath100 is a parameter that tunes the correlation .
the auto - correlation function @xmath101 decays exponentially @xmath102 , where the characteristic time @xmath103 .
this model is markovian and stationary , with a probability density @xmath81 and transition probability @xmath82 given by @xmath104}{\sqrt{2\pi ( 1-r^2)}}\ ] ] the @xmath28 component of @xmath84 is given by @xmath105 where , again , @xmath95 are the top and bottom conditions for the variable @xmath96 in motif @xmath28 reported in table table [ tab:1 ] .
once set the parameter @xmath106 the profile can be evaluated numerically up to arbitrary precision ; here we give the profile for three possible values @xmath107 and @xmath108 @xmath109 in all these examples , theoretical results are in very good agreement with results obtained with numerical simulations reported in figure [ fig:1_2 ] . differently from the hvg motifs , vg motifs statistics
does not depend uniquely on the ranking statistics of the data and therefore the vg motif profile could be able in principle to discriminate white noises with different marginals . in the latter sections we have been able to distinguish between gaussian and uniform white noise . in figure [ fig:2 ]
we summarize the motif frequencies @xmath110 of vg motifs forming @xmath34 , extracted from i.i.d .
series with different marginals : @xmath111;\quad f(x_i)\sim1\\ \text{gaussian}\rightarrow x_i\in(-\infty,\infty);\quad f(x_i)\sim\frac{\exp(-x_i^2/2)}{\sqrt{2\pi}}\\ \text{power - law}\rightarrow x_i\in[1,\infty);\quad f(x_i)\sim x_i^{-k } , \quad k= 2.5\\ \text{exponential}\rightarrow x_i\in[0,\infty);\quad f(x_i)\sim \exp(-k x_i ) , \quad k= 2.5\\ \end{cases}\ ] ] in every case we extract series of @xmath112 data . the universal profile obtained for hvg is also plotted for comparison . as expected , motif profiles are different for different marginals .
motifs which are symmetric to each other ( 3 and 4 , 5 and 6 ) occur with equal probabilities , something that does nt occur when the series is chaotic ( eq .
[ log ] ) . according to the values obtained for the components of @xmath34
, one can extract some heuristic conclusions : * @xmath113 , @xmath114 encode information on the marginal distribution of the process as well as its autocorrelation structure .
* @xmath115 is null as this motif is not a vg .
this is at odds with the hvg case , where this is an admissible motif provided the probability of finding consecutive equal data in the series is finite ( e.g. for discrete - valued series ) .
* the motifs associated to the pairs ( @xmath116,@xmath117 ) ( @xmath118,@xmath119 ) have chiral symmetry .
in other words , the motifs associated to @xmath116 and @xmath117 are isomorphic , the correct permutation being @xmath120 ( the same holds for @xmath118 and @xmath119 ) .
accordingly , for any process which is statistically time reversible , we expect these probabilities to be equal .
reversible processes include linear stochastic processes ( and both white and red noise belong to this family ) , while non - invertible chaotic processes are usually time irreversible ( the fully chaotic logistic map is an example ) .
time irreversibility of the process is therefore encoded in these terms . *
@xmath121 as this is not a vg and therefore does not appear ( not admissible ) .
when dealing with empirical time series , the practitioner usually faces two different but complementary challenges , namely ( i ) the size of the series and ( ii ) the possible sources of measurement noise .
the first challenge can be a problem when the statistics to be extracted from the series are strongly affected by finite - size effects , whereas for the second one needs to evaluate the robustness of those statistics against noise contamination . for a statistic or feature extracted from a time series to be not just informative but useful one usually requires that statistic or feature to be robust against both problems
: it needs to have fast finite - size convergence speed and to be robust against reasonably large amounts of additive noise . + in @xcite it has been already shown that the hvg motif profile has good convergence properties respect to the series size @xmath1 and it is also robust respect to noise contamination . here
we explore these very same problems for the case of the vg motif profile and we make a detailed comparison of its performance with the hvg motif profile in a range of situations . in general , due to finite size effects , the estimated value of any feature fluctuates and deviates with respect to its asymptotic , expected value . for classical features such as the mean or the variance of a distribution ,
these deviations are bounded and vanish with series size with a speed quantified by the central limit theorem .
the estimation of the motif frequencies can be quantitative effected by finite - size fluctuations and one can even observe missing motifs ( motifs with estimated frequency @xmath122 ) which are not actually forbidden by the process but have not appeared by chance .
this situation can be overemphasized in the presence of certain types of measurement noise . + following an approach analogous to the one followed for the forbidden ordinal patterns in @xcite ,
we first perform a test to study the decay of missing motifs with the series size both in stochastic uncorrelated and correlated processes . in figure [ fig:3 ]
panel a ) we plot @xmath123 , the average number of missing motifs in a series of size @xmath1 in the case of gaussian white noise and colored ( red ) gaussian noise ( for the red noise we consider the ar(1 ) process with correlation length @xmath124 discussed in section iii ) . for both types of noise @xmath123 decays exponentially to zero and already with a series of about 80 - 100 data points we can exclude the possibility of detecting missing motifs ( for both hvg and vg ) due to finite size fluctuations even in the case of correlated noise .
+ as a second analysis , we explore the convergence speed of the estimated motif profile of uncorrelated and correlated stochastic series and of chaotic series ( fully chaotic logistic map ) of size @xmath1 to the asymptotic profile solution given in section iii . to do this we define the distance between the estimated @xmath125-motif probabilities @xmath126 and the asymptotic value @xmath127 .
we use @xmath31 norm and accordingly define @xmath128 in figure [ fig:3 ] panel b ) we show the trend of @xmath129 in log - log scale ( results are averaged over 300 realizations ) .
the average distance decreases like a power - law for all the processes considered , in agreement with a central - limit - theorem - like argument . for a series of @xmath130 points @xmath131 is less than @xmath132 and the average distance @xmath133 for each of the single components is less than @xmath134 ( not shown ) .
these results suggest that vg and hvg motif profiles have very good convergence properties and are thus robust against finite size fluctuations . to test and compare the robustness of vg and hvg motif profiles when the effect of noise contamination combines with the finite size fluctuations we consider the fully chaotic logistic map dynamics @xmath135 polluted with measurement ( additive ) noise @xmath136 @xmath137 in the two cases where @xmath136 is respectively white gaussian noise ( @xmath138 ) or colored gaussian noise ( @xmath124 ) . for both cases
@xmath139 $ ] is the parameter which tunes the noise - to - signal ratio ( nsr ) of the process defined as @xmath140}{\sigma^2[x]}\ ] ] where @xmath141 $ ] and @xmath142 $ ] are respectively the theoretical variance of the white gaussian noise @xmath143 and the theoretical variance of the dynamics ( signal ) @xmath144 ( note that with this definition we are underestimating the nsr in the case of correlated noise where @xmath145=\sigma^2[\sqrt{\alpha}\cdot\xi]/(1-r^2)$ ] ) .
the robustness of the observed motif profile @xmath146 $ ] for a single realization of the process with given @xmath1 and @xmath147 can be defined as the distance between this profile and the theoretical profile @xmath148 $ ] of the noise @xmath149 for the given @xmath147 .
@xmath150-\phi^4_m[\eta(\alpha)]|.\ ] ] with such definition we expect @xmath151 for low values of the nsr ( dominant signal , @xmath152\simeq\phi^4_m[x]$ ] ) and @xmath153 for high values of the nsr ( dominant noise , @xmath152\simeq\phi^4_m[\eta]$ ] ) .
furthermore , @xmath154 is affected by finite size effects : if we assume to have few realizations @xmath155 of the process @xmath156 of small series size @xmath1 , then we expect the variance @xmath157 calculated over the realizations to be high .
in particular we have to consider that a resolution limit @xmath158 exists , such that when @xmath159 we can not say any more if the distance we measured is discriminating the signal @xmath144 from the noise @xmath149 or it is simply due to finite - size effects of the contamination noise @xmath149 .
we define this threshold @xmath158 as the sum of the standard deviations of the estimated profile components @xmath160 $ ] given @xmath155 realizations of the noise process alone @xmath161-\langle\hat{\phi}^4_m[\eta(n,\alpha)]\rangle_{n_r})^2\rangle_{n_r}}\ ] ] it is thus convenient to work with the relative distance @xmath162 ; when @xmath163 we say that the resolution limit for the process @xmath164 -given its @xmath155 realizations- is reached , and @xmath165 is not any more a reliable indicator .
+ in figure [ fig:4 ] panel a ) we show the quantity @xmath166 averaged over 300 realizations of the process @xmath156 for different level of contamination nsr=@xmath167 $ ] at fixed size @xmath1=6400=@xmath168 ( notice that for @xmath169 we missing motifs are not found anymore ) respectively for white gaussian noise and colored gaussian noise and for the hvg and the vg motif profile .
the red solid line represents the resolution limit threshold for the process .
we can see that the hvg and the vg motif profiles are more robust respect to noise contamination when this noise is correlated .
in this situation the hvg motif profile seems to perform better than the vg motif profile , while in the case of uncorrelated gaussian noise the vg profile seems in turn slightly more robust than the hvg profile .
+ the last step of this robustness analysis is to consider the usual situation where only very few realizations ( often a single one ) of the same process are available .
our aim is to define a useful indicator @xmath170 which estimates for any given value of the size @xmath1 the maximum amount of noise contamination level for which a measure @xmath171 computed with only one realization of the process @xmath156 can be considered somewhat reliable .
we define this to be the value of the nsr such that @xmath172 , and thus @xmath173 @xmath174 measures ( in units of noise - to - signal ratio ) the ( statistical ) reliability of the motif profile extracted form a single time series of size @xmath1 of the signal @xmath144 in the presence of measurement noise @xmath149 . + in figure
[ fig:4 ] ( panel a ) we plot @xmath175 for white gaussian noise in the case of vg by considering the @xmath176 curve marked by orange squares and by taking the smallest value of nsr for which an orange error bar intersect the red line ( the blue box highlights the region ) .
wee find @xmath177 , meaning that when working with a single time series of the process @xmath156 with size @xmath178 , the @xmath179 distance measured by using the vg motif profile is reliable up to a level of white gaussian noise contamination @xmath147 such that @xmath180 . in figure [ fig:4 ] ( panel b )
we report the estimated value of @xmath170 for the vg and hvg motif profiles in the case of white gaussian noise and correlated gaussian noise in function of the series size @xmath1=@xmath181 ( maximum noise contamination level considered was nsr(@xmath147)=8 ) .
we can see that the motif profile is in general a robust measure respect to the combined effect of measurement noise and finite size : working with a single time series of only @xmath182 points of the process @xmath156 we can extract both the vg and the hvg motif profiles and expect those features to be informative respect to the underlying chaotic signal @xmath144 up to a level of measurement noise for which nsr=1.5 in the case of uncorrelated gaussian noise and nsr=3 in the case of correlated gaussian noise .
+ also and as observed before ( figure [ fig:4]b ) ) , given the case of white gaussian noise contamination the vg motif profile ( orange squares ) seems to perform slightly better than the hvg motif profile ( green circles ) . for colored gaussian noise
the situation is the opposite and the hvg motif profile ( reversed gray triangles ) performs much better ( almost a gap of one unit of nsr for @xmath183 ) than the vg motif profile ( blue triangles ) . for both type of visibility graphs the motif profile is coherently more robust when polluted with colored noise than with white noise .
this is probably due to the fact that white noise breaks up the correlation structure of the signal faster ( respect to the size @xmath1 ) than correlated noise .
it is also interesting that both types of motif profiles are very sensible to the noise correlations although the different nature of the visibility algorithms .
sequential visibility graph motifs are small subgraphs where nodes are in consecutive order within the hamiltonian path that appear with characteristic frequencies for different types of dynamics . this concept was introduced recently @xcite and a theory was developed to analytically compute the motif profiles in the case of horizontal visibility graphs ( hvgs ) . in this work
we have extended this theory to the realm of natural visibility graphs ( vgs ) , a family of graphs where the previous amount of known exact results was practically null .
we have been able to give a closed form for the 4-node vg motif profile associated to general one dimensional deterministic and stochastic processes with a smooth invariant measure or continuous marginal distribution , for the cases where the variables belong to a bounded or unbounded interval . in the case where the time series is empirical and one
does not have access to the underlying dynamics , the methodology still provides a linear time ( @xmath184 ) algorithm to estimate numerically such profile .
we have shown that the theory is accurate and that vg motifs have similar robustness properties as hvg , yet they depend on the marginal distribution of the process and as such yield different profiles for different marginals .
this is at odds with the results found for hvgs , where the motif profiles did not depend on the marginals as they behave as an order statistic .
+ the detection of such motifs from a visibility graph extracted from a time series can be seen as a process of dynamic symbolization of the series itself , where the alphabet of symbols is composed by different subgraphs ( motifs ) which encode information about both data relations and their temporal ordering in their link structure . the deep similarity between hvg motifs and the so called ordinal pattern analysis -which holds mainly due to the fact that hvg is an order statistic- vanishes for vg motifs , which therefore stand as a complementary tool for time series analysis , specially relevant when the marginals play a role in the analysis . | the concept of sequential visibility graph motifs -subgraphs appearing with characteristic frequencies in the visibility graphs associated to time series- has been advanced recently along with a theoretical framework to compute analytically the motif profiles associated to horizontal visibility graphs ( hvgs ) . here
we develop a theory to compute the profile of sequential visibility graph motifs in the context of natural visibility graphs ( vgs ) .
this theory gives exact results for deterministic aperiodic processes with a smooth invariant density or stochastic processes that fulfil the markov property and have a continuous marginal distribution .
the framework also allows for a linear time numerical estimation in the case of empirical time series . a comparison between the hvg and the vg case ( including evaluation of their robustness for short series polluted with measurement noise ) is also presented . |
in this section the distance between the subsystem of the dephased rectangular decoupled state and the gibbs state @xmath82 is computed .
the discussion is similar to that in ref . @xcite . as all eigenvectors of @xmath23
are given by tensor products of the eigenstates of @xmath21 and @xmath22 , @xmath140 , it is possible to trace the degrees of freedom of the bath of the decoupled rectangular state to get @xmath141 with @xmath142 here , @xmath143 denotes the number of states of the bath with an energy contained in the interval @xmath144}$ ] . for simplicity
we assume that @xmath145 .
notice that the probabilities @xmath146 only depend on the shape of the spectrum of the bath .
we aim at bounding the trace distance @xmath147 where @xmath148 are the eigenvalues of the gibbs state and @xmath149 .
let us define a function @xmath0 as the logarithm of the number of states of the bath @xmath150 the hilbert space of the bath is finite dimensional and thus the spectrum of the bath is discrete , from now on it is however assumed that the bath is large enough such that @xmath151 can be well approximated by a twice differentiable function @xmath152 . for natural systems and reasonable
energy ranges the additional error from this approximation is exponentially small in the size of the bath . for the sake of conciseness we ignore such exponentially small errors .
we discuss this continuous approximation more closely in appendix b. from now on we will hence focus on @xmath153 , where @xmath154 it a twice differentiable approximation of @xmath155 .
taylor s theorem ensures that for every @xmath156 there exists some @xmath157 $ ] , such that @xmath158 where @xmath159 is the inverse temperature and @xmath160 a linear expansion of @xmath161 is equivalent to an exponential fit of the smoothed number of states of the bath @xmath162 , @xmath163 thus , the probabilities @xmath146 can be written as @xmath164 where @xmath165 .
therefore , the distance between @xmath37 and the gibbs state depends on how well the density of states of the bath can be approximated by an exponential curve .
the difference between the probabilities reads @xmath166 where the fraction @xmath167 can be rewritten as @xmath168 introducing the notation @xmath169 } \gamma_k(\xi_k ) , \\
\gamma_{\mathrm{max } } & \coloneqq \max_k \max_{\xi_k\in[e - e_k^s , e ] } \gamma_k(\xi_k ) , \end{aligned}\ ] ]
we can write @xmath170 the trace distance between the reduced dephased decoupled rectangular state @xmath37 and the gibbs state is thus bounded from above by @xmath171 where @xmath88 is exponentially small in the bath size . we explicitly bound @xmath172 for a specific bath of uncoupled spins in appendix b.
in order to have a more explicit expression for the right hand side of eq .
let us consider a particular model for the bath .
we will also use this bath for the algorithm presented in the last part of the article .
we start with the natural choice of @xmath83 non - interacting uncoupled spin-@xmath84 particles with energies @xmath85 and @xmath86 .
the spectrum of this model is discrete and the eigenvalues are integer multiples of @xmath86 between @xmath85 and @xmath173 .
the system is highly degenerate and the number of states with energy @xmath174 follows a binomial distribution @xmath175 .
this degeneracy makes it impossible to find a sufficiently good smoothed approximation @xmath176 for the number of states @xmath177 such that even for intervals whose width @xmath81 scales like @xmath178 with @xmath179 the error in the approximation of the local density @xmath180 goes to zero for large @xmath6 .
that is , the smoothed approximation would cause additional errors that would not go down exponentially fast with the bath size . therefore , we need to assume that the degeneracy of the levels are lifted by a suitable perturbation
. as we are only concerned about the spectrum and not the eigenstates of the bath , essentially any perturbation of adequate strength , such as basically any arbitrary weak interaction , will be sufficient .
a convenient way to perturb the model for a theoretical study is to replace the fixed local field strength by a normal distributed random field strength such that on average the energy of the local excited state is still @xmath86 above the local ground state .
if the width of the normal distribution is chosen in a suitable way , we get , with overwhelmingly high probability , a number of states @xmath177 in the interval @xmath26 $ ] that can be well approximated by @xmath181 in the aforementioned sense , where @xmath182 our aim is to bound @xmath172 in eq . for this model . to do this
, we first consider the case @xmath183 in which we can approximate @xmath176 as follows @xmath184 we can easily compute the second derivative of the logarithm of the above expression and find @xmath185 if @xmath81 is not much smaller than @xmath186 , @xmath187 deviates from the parabolic form .
but , larger @xmath81 only make the curvature of @xmath187 smaller . in the general case we thus have @xmath188 and since @xmath189
we arrive at @xmath190 where @xmath88 is exponentially small in the bath size . in natural situations
the right hand side of the above equation will be very small as @xmath191 .
the rectangular states are a quite artificial class of states .
so what happens if instead of @xmath37 we take states with a different energy distribution ?
will we still get something close to a gibbs state as long as the width of the energy distribution is not too small and the density of states is exponentially increasing ?
the exponential increase in the density of states ensures that most of the contribution to @xmath36 comes from the upper edge of the interval @xmath192 in the case of a rectangular distribution .
thus , it can be expected that one will get a state that is close to a gibbs state for any energy distribution that has a region with a bounded variation followed by a sufficiently sharp cutoff to higher energies . where by sufficiently sharp we mean that it must drop from a value much larger than @xmath193 to a value much smaller than @xmath193 in an energy interval that is small compared to @xmath194 .
the bounded variation is required as for certain systems that violate the so called _ eigenstate thermalization hypothesis _ ( eth )
@xcite details of the energy distribution can have a huge impact on certain properties of the state .
that bounded variation of the energy distribution is required to guarantee thermalization in such systems can be seen for example in the model studied in ref .
in this section we discuss the average number of times the algorithm must be run before the energy window corresponding to the desired temperature is hit .
as we are only interested in the number of repetitions we can assume that the algorithm runs `` perfectly '' in that it generates exactly a rectangular state with fixed width @xmath81 at some position @xmath80 of the spectrum .
furthermore , we are only interested in cases where the reduced state of the rectangular state can be guaranteed to be close to a gibbs state .
thus , we may assume that the spectrum of the bath is dense enough such that the number of states of the bath in the interval @xmath26 $ ] can be well approximated by a continuous and twice differentiable function @xmath176 that increases exponentially to higher energies .
the probability of getting a rectangular state at energy @xmath80 is given by @xmath195 where @xmath196 is the dimension of the hilbert space of the total system .
using that the density of states of the bath can be locally approximated by an exponential , we obtain @xmath197 then , the probability of getting energy @xmath80 can be bounded from above by @xmath198 with @xmath199 the gap of the system .
the number of times that the program must be run on average in order to get the rectangular state at the position @xmath80 is thus lower bounded by @xmath200 this last equation shows that it is exponentially hard to go to low temperatures .
this is to be expected , as no structure of the hamiltonian is being used and generic local hamiltonian systems can presumably not be efficiently cooled arbitrarily close to their ground state , because this would efficiently solve qma - hard problems on a quantum computer @xcite . needless to say , for specific hamiltonians , with some additional structure , an algorithm can well be more efficient .
the coefficient of the exponential is related to the features of the spectrum at low energies . in the case
in which there is a gap , the ground state could encode the solution of a satisfiability problem that is expected to be hard to solve .
in this section we calculate how the inverse temperature @xmath17 and its error @xmath201 depend on the energy value obtained while running the algorithm and the number of qubits @xmath106 of register @xmath95 that are measured . after running the algorithm a value @xmath107
is obtained and a state close to the rectangular state at energy @xmath202 is prepared .
the system is then thermalized at an inverse temperature @xmath203 inserting the formula for @xmath176 for the bath described in appendix b we find @xmath204 the algorithm will be run repeatedly until the value @xmath107 corresponding to the desired @xmath17 is obtained . for the bath considered here
this value is @xmath205 it follows from eq .
that @xmath206 to reach a given precision @xmath201 in the temperature it is thus sufficient to choose @xmath207
in this section the error of the quantum algorithm is derived and a bound on the average number of repetitions that are necessary to reach the desired energy is given .
the total error is bounded by the sum of two errors @xmath208 the deviation of the reduced rectangular state from the gibbs state @xmath209 has been calculated already in appendix a and b @xmath210 here we bound the deviation @xmath211 of the final state of the algorithm from the rectangular state . our final result for
the bath discussed in appendix b will be @xmath212 ( again @xmath88 is exponentially small in the bath size ) .
it is convenient to write @xmath86 in terms of a dimensionless parameter @xmath213 @xmath214 because then the two errors can be written in the form : @xmath215 to reduce the trace distance @xmath82 it is favorable to choose a large @xmath136 .
intuitively this captures the fact that the energy content in the bath must be much larger than that of the system in order to get a gibbs state .
however , at the same time this increases the error in the approximation of the rectangular state , because performing the phase estimation on a system with a more dense spectrum is harder .
this increase of the error can however be compensated by increasing the accuracy of the phase estimation by linearly increasing the number of qubits @xmath96 in the ancilla register .
similarly , lower temperatures can be reached by a quadratic increase of the number of ancilla qubits of order @xmath137 .
as the phase estimation requires a total number of @xmath131 controlled unitary operations , this comes at an exponential cost in terms of runtime .
this is expected to be a problem common to all algorithms using phase estimation on many particle systems .
before we go into the derivation of @xmath216 , let us quickly give the average number of runs that are required to get a certain temperature for the bath discussed in appendix b in terms of @xmath136 . from eqs . and and the explicit formula for the smoothed number of states given in eq
it is easy to see that @xmath217 here we bound the trace distance @xmath216 between the state @xmath218 generated by the circuit with @xmath96 ancilla qubits when , after measuring @xmath106 of them , the outcome @xmath107 is obtained , and the rectangular state @xmath219 in the interval @xmath26 $ ] with @xmath220 and @xmath221 is prepared .
this trace distance can be written in terms of the one norm distance of the distributions of the diagonal elements @xmath222 where the @xmath223 and @xmath146 are the eigenvalues of @xmath218 and @xmath219 respectively , which can be written as @xmath224 where @xmath225 and @xmath226 are the phases corresponding to the eigenvalues of the hamiltonian and the energy of the rectangular state , @xmath227 and @xmath228 are normalization constants , and the functions @xmath229 and @xmath230 are defined as @xmath231 with @xmath232 and @xmath233 the step function ( see main text and fig .
[ fig : qpe - circuit ] ) .
plot of the @xmath230 and @xmath229 functions . as @xmath96 increases
, @xmath229 tends to the rectangular function @xmath230 . ]
is bounded by @xmath234 where we have used that - + f_r , q(x ) -and z_f - z_g_k=1^d | f_r , q(_k- ) - g_q(_k- ) | .
notice that @xmath235 , where @xmath236 is the number of states of the total system in the interval @xmath237 $ ] . using that the spectrum of the whole system is sufficiently dense it is possible to approximate the upper bound in eq . by an integral . to make this rigorous
we decompose the sum into bins of width @xmath238 , @xmath239 where @xmath240 is a function introduced to simplify the notation and @xmath241 is the number of bins in which the spectrum has been divided .
the idea is to take an @xmath241 as large as possible such that the number of energy values in a bin @xmath242 can still be well approximated by its smoothed version @xmath243 ( compare the discussion of the twice differentiable approximation in appendix a and b ) . for baths for which such a smooth approximation is possible
, one can always take an @xmath241 proportional to the dimension of the hilbert space @xmath244 or , at least , to some power @xmath245 with @xmath246 such that @xmath241 scales exponentially with the size of the bath .
as @xmath247 is continuos differentiable except at @xmath80 and @xmath192 , taylor s theorem ensures that for all bins @xmath248 ( except for the two bins that contain @xmath80 and @xmath192 , which we can simply ignore because they only contribute an error of order @xmath249 ) there exists some value @xmath250 for which @xmath251 where @xmath252 .
then , the contribution of the @xmath248-th bin in eq . can be bounded from above by @xmath253 the last term in the parenthesis in eq
. decreases with @xmath241 and is thus exponentially small in the bath size .
this can be verified with a lengthy , but straightforward calculation .
notice that the number of states can be written as @xmath254 where @xmath255 putting eqs . , and together
, the error of the circuit can be bounded from above by @xmath256 notice that the upper bound of the previous equation converges to an integral for @xmath257 with deviations that scale as @xmath258 .
thus , the trace distance between @xmath218 and @xmath29 is bounded from above by @xmath259 where @xmath260 we will bound @xmath261 for the model described in appendix b in the next section . next , the error of the circuit for a bath described in appendix b is bounded .
the one norm @xmath262 from eq . can be bounded by @xmath263 where the definition of @xmath247 given in eq . has been used .
we now find an upper bound on the supremum and a lower bound on the smoothed number of states .
the density of states of the total system can be written in terms of the density of states of the bath as @xmath264 inserting the density of states of our bath in the previous expression and writing it as a function of the inverse temperature given in eq .
we get @xmath265 as we only consider positive temperatures @xmath266 it is easy to see that the density of states attains its maximum value at @xmath267 , thus @xmath268 where @xmath269 is the dimension of the hilbert space of the total system .
the smoothed number of states @xmath270 can be lower bounded as follows @xmath271 where again we have assumed that @xmath80 is in a position of the spectrum with positive temperature .
finally we obtain @xmath272 in order for this error to become small the one norm @xmath273 must be made small enough such that it compensates the exponential prefactors .
as we will see in the next section this can be achieved by using a polynomially large ancilla register @xmath95 . in this section
we bound the one norm between @xmath229 and @xmath230 introduced in eqs . and .
the one norm between @xmath274 and @xmath55 is defined as @xmath275 by a simple change of variables it is easy to show that @xmath276 remember that both @xmath229 and @xmath230 are normalized on the interval @xmath277 $ ] and that @xmath278 is a step function that is non - zero only in the interval @xmath279 , and that in this interval @xmath280 .
using this , the previous integral can be rewritten as @xmath281 due to the symmetry and periodicity of @xmath282 , the contribution to the previous integral of the right tail of @xmath283 is the same as the contribution of the left tail of @xmath284 for @xmath285 , thus @xmath286 this implies that @xmath287 the integral of the previous equation can be bounded by @xmath288 where the bound @xmath289 has been used .
therefore , the distance between the functions @xmath229 and @xmath230 can be bounded by @xmath290 where the sum has been split up in the @xmath291 case , which is exactly @xmath84 , and the rest .
the sum of the previous equation can be bounded in two steps by @xmath292 thus , the one norm between @xmath274 and @xmath55 is bounded by @xmath293 inserting this into eq . , we finally get @xmath294 where @xmath88 is exponentially small in the bath size .
this completes the discussion of an upper bound of the error made in the quantum algorithm .
@xcite presents a novel approach towards thermalizing quantum systems using an iterative approach , in which pre - thermalized parts are put together in a suitable fashion in order to arrive at a gibbs state of a quantum system with a local hamiltonian .
this argument provides a new intuition on how one can think of thermalizing local quantum systems , different from a quantum monte carlo approach ( and the one presented here ) . in this appendix , however , we point out a subtle mistake of the published version of the paper ; however , this challenge can be overcome and a new proof can be formulated that retains key elements of the previous structure of the argument @xcite . the intuition behind the argument and in fact the algorithm itself hence turn out to be correct .
each merging step consists of two steps .
the first is a probabilistic step that updates the probability weights of the gibbs state by means of postselection .
the second one aims at rotating the eigenbasis of the old hamiltonian to the one of the new hamiltonian by means of an instance of dephasing .
each step has as input a chosen @xmath295 .
for the entire algorithm to work , this procedure has to be correct up to errors of @xmath296 .
in what follows we refer to the equation numbering of the preprint v2 . in eq .
( 5 ) , perfect dephasing is being achieved when @xmath297 .
this is approximated by imperfect dephasing with a finite @xmath298 . from the dyson series of second order ( 6)-(8 )
, it follows that ( in the asymptotic notation ) @xmath299 this is not made explicit in the paper , but appears to be crucial . intuitively speaking ,
if @xmath298 becomes larger , the dephasing is more exact , but then ( 6)-(8 ) can no longer be used .
this step is then used in the procedure following eq .
a @xmath300 is introduced and dephasing between eigenstates with relative gap larger than @xmath300 is considered
. then a new hamiltonian @xmath301 is constructed , which has the following feature : it has the same eigenbasis as @xmath302 , but with eigenvalues grouped in bins , such that the smallest gap between bins is @xmath300 . for the following procedure to work ,
one has to take ( again in the asymptotic notation ) @xmath303 so both @xmath304 and @xmath305 ( although only @xmath304 is made explicit , both bounds are actually needed ) . this
, however , seems to already fix @xmath306 .
so there is no longer the freedom to have @xmath307 which would mean that @xmath308 .
this appears to contradict the above statement .
again , intuitively , _ @xmath298 is forced to be small for the dyson approach to work _ , but at the same time this gives a constraint to @xmath300 which _ requires @xmath298 to be large_.
as we have seen in the main text in theorem 1 the distinguishability of the microcanonical states @xmath36 and @xmath29 corresponding to an interval @xmath26 $ ] of the hamiltonians @xmath23 and @xmath7 is bounded by @xmath309 where @xmath41 and @xmath42 are the maximum , and the difference , of the dimensions of the support of @xmath36 and @xmath29 , and @xmath43 is the total number of eigenstates of @xmath20 and @xmath23 in the intervals @xmath44 $ ] and @xmath45 $ ] . in the main text , in order to give a more comprehensible interpretation to eq .
, an approximately constant density of states was assumed and it was shown that @xmath310 however , we have seen that thermal states emerge in situations where the density of states is locally well approximable by an exponential . thus , new conditions that ensure the indistinguishability of the microcanonical states @xmath311 and @xmath37 must be derived for the exponential density of states @xmath312 in order to do this , let us notice that both terms in the upper bound of eq .
are positive and must be simultaneously and independently small . for the first term , this is only possible if @xmath313 . to find the condition for the second term ,
let us assume that the interaction does not shift excessively the energy levels such that @xmath314 can be neglected . then , @xmath315 and the condition @xmath316 is required .
these two necessary conditions can be summarized as @xmath317 it is easy to see that they are also essentially sufficient conditions by using @xmath318 to bound eq .
as follows @xmath319 equation ensures the indistinguishability of the interacting and non - interacting microcanonical states as long as @xmath322 is not too small and the condition v _
t , where @xmath323 is the boltzmann constant and @xmath324 the absolute temperature , is fulfilled .
that is , eq .
gives us a physical intuition about when an interaction is _ weak _ in the sense of theorem 1 : an interaction is weak if it is small compared to the thermal energy @xmath325 , which is a measure of the intensive energy content of the system . in conclusion
we see that how strong an interaction feels for a system depends on how much energy it contains . | in this work , we show how gibbs or thermal states appear dynamically in closed quantum many - body systems , building on the program of dynamical typicality .
we introduce a novel perturbation theorem for physically relevant weak system - bath couplings that is applicable even in the thermodynamic limit . we identify conditions under which thermalization happens and discuss the underlying physics .
based on these results , we also present a fully general quantum algorithm for preparing gibbs states on a quantum computer with a certified runtime and error bound .
this complements quantum metropolis algorithms , which are expected to be efficient but have no known runtime estimates and only work for local hamiltonians .
how do thermal quantum states
cornerstones of a description in canonical ensembles in quantum statistical physics arise from the underlying theory of quantum physics ?
this question , a long tradition as it obviously has , is in many ways still surprisingly wide open .
indeed , much progress was made only recently @xcite ; this is motivated and triggered both by new mathematical @xcite and numerical @xcite techniques becoming available , as well as by new experiments with quantum many - body systems in nonequilibrium @xcite . in this work we present a set of precise sufficient conditions for the emergence of gibbs states from the underlying microscopic theory of quantum mechanics .
our results go beyond previous approaches in that they apply in a physically relevant weak coupling limit and constitute the key insight leading to the invention of a quantum algorithm that prepares gibbs states with certified precision and runtime .
the three ingredients that enter the standard textbook proof of the canonical ensemble in classical statistical physics are : ( i ) the _ equal a priori probability postulate _
( also known as microcanonical ensemble ) and an equilibration postulate ( such as the second law ) , ( ii ) the assumption of _ weak coupling _ , and ( iii ) an assumption about the _ density of states _ of the bath , namely , that it grows faster than exponentially with the energy and that it can be locally well approximated by an exponential @xcite . here
each of these steps is translated to the _ pure state quantum statistical mechanics _
approach @xcite .
in particular ( i ) can be replaced by either a typicality argument , or a statement about dynamical relaxation that follows directly from quantum mechanics and ( ii ) is made precise by proving a novel perturbation theorem that has applications far beyond the scope of the present article .
our new technical results allow us to design a _ quantum algorithm preparing gibbs states _ with explicit error and runtime bounds , invoking a new variant of phase estimation .
our algorithm complements another algorithm with certified runtime that was proposed in ref .
@xcite and recent developments on quantum metropolis algorithms @xcite .
[ [ setting - and - notation . ] ] setting and notation .
+ + + + + + + + + + + + + + + + + + + + + we consider a system @xmath0 weakly coupled to an environment @xmath1 .
the hilbert space reads @xmath2 , where @xmath3 and @xmath4 are the hilbert spaces of the subsystem and the `` bath '' ( with finite dimensions @xmath5 and @xmath6 ) .
the evolution of the total system is governed by the hamiltonian @xmath7 , with eigenvalues and eigenvectors @xmath8 and @xmath9 consisting of an uncoupled hamiltonian @xmath10 , with eigenvalues and eigenvectors @xmath11 and @xmath12 , and a coupling hamiltonian @xmath13 .
we give conditions under which the reduced state @xmath14 , with @xmath15 , of the subsystem @xmath0 relaxes for most times to a gibbs state @xmath16 with inverse temperature @xmath17 under unitary time evolution @xmath18 . by this
we mean that for most times their trace distance @xmath19 , which measures the physical distinguishability @xcite , is small .
note that the decomposition of a given @xmath20 into @xmath21 , @xmath22 , and @xmath13 is not unique .
this freedom can be used to optimize the bounds in our results , and the correct @xmath21 naturally results from this optimization .
we assume that the hamiltonians @xmath20 and @xmath23 are non degenerate such that _ time averaging _ and _ dephasing _ in the eigenbasis give the same result @xmath24 whenever an expectation value equilibrates , it does so to the expectation value in @xmath25 @xcite .
[ [ natural - thermalization - conditions - for - gibbs - states - to - appear . ] ] `` natural thermalization '' : conditions for gibbs states to appear .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in this section we go through points ( i)(iii ) .
the final conclusion is summarized in corollary [ theorem : summaryofnaturalthermalization ] .
the central point of the argument is a novel perturbation theorem that relates spectral projectors of weakly interacting and noninteracting hamiltonians in a physically relevant weak coupling limit .
it allows us to connect results on dynamical equilibration and measure concentration with classical counting arguments and thereby prove a set of natural sufficient conditions for thermalization in quantum mechanics
. a stepping stone in the argument will be states that have an energy distribution that is flat in an interval @xmath26 $ ] and vanishes otherwise .
we indicate such states , and their dephased states , by a subscript @xmath27 like in @xmath28 or @xmath29 and call them _ rectangular states_. this class of states includes both _ mixed states _ ( in particular the microcanonical state @xmath29 ) and _ pure states _ and thus , because of the freedom to choose the phases , usually also initial states that can be locally out of equilibrium .
the _ equal a priori probability postulate _
( i ) can be replaced by a typicality argument using results from refs .
@xcite . in ref .
@xcite powerful concentration of measure techniques are used to show that almost all states from a microcanonical subspace corresponding to a microcanonical energy window @xmath26 $ ] locally look like the reduction of the corresponding microcanonical state ; i.e. , @xmath30 is small for all but exponentially few of the states @xmath31 from the subspace , where @xmath29 is the microcanonical state on the subspace .
alternatively one can use the results concerning the dynamics of states with a high effective dimension of refs . @xcite . under one assumption on the spectrum of the hamiltonian ( nondegenerate energy gaps )
, it is shown that all reduced states on small subsystems of such states tend to the time - averaged equilibrium state @xmath32 and stay close to it for most times .
in many - body systems , natural initial states have a high effective dimension , and this is provably true for all but exponentially few states from a microcanonical subspace @xcite .
the delicate issue , which has up to now not been addressed in the literature in a general and rigorous way , is the _ weak coupling _ approximation ( ii ) @xcite .
the problem is that due to the exponential growth of the hilbert space dimension and the at most polynomial growth of the energy content , the spectrum of the noninteracting hamiltonian @xmath23 becomes exponentially dense with increasing bath size .
therefore , the _ perturbative _ limit , in which the coupling @xmath13 is weak compared to the _ gaps _ of the noninteracting hamiltonian @xmath23 , and in which it can be guaranteed that the energy eigenvectors @xmath33 of the full hamiltonian @xmath7 are close to product states , is arguably _ not the physically relevant weak coupling limit_. even worse , in this limit memory effects provably prevent thermalization @xcite .
as in the classical setting , a coupling should be considered to be _ weak _ as long as it does not change the total energy in a _
noticeable way_. that is to say , the energy stored in the interaction is much less than our ( microcanonical ) uncertainty about the energy of the system , i.e. , @xmath34 , or for thermalizing systems much less than the thermal energy @xmath35 .
this is the _ relevant weak coupling limit _ in which we prove equilibration towards a gibbs state .
we do this by relating the dephased / microcanonical state @xmath29 to the state @xmath36 dephased with respect to the non interacting hamiltonian , for which we can easily perform the partial trace to obtain @xmath37 and thereby an approximation to @xmath38 .
[ theorem : interactingvsnoninteractingcase ] let @xmath36 and @xmath29 be the dephased / microcanonical states belonging to the interval @xmath26 $ ] with respect to @xmath23 and @xmath7 ; then for every @xmath39 @xmath40 where @xmath41 and @xmath42 are the maximum , and the difference , of the dimensions of the supports of @xmath36 and @xmath29 , and @xmath43 is the total number of eigenstates of @xmath20 and @xmath23 in the intervals @xmath44 $ ] and @xmath45 $ ] .
the theorem shows that , for any two initial ( possibly pure ) states that have a flat energy distribution in the interval @xmath26 $ ] with respect to the hamiltonians @xmath23 and @xmath20 with @xmath34 , the distance of their reduced dephased states @xmath37 and @xmath38 is small . in particular , assuming an approximately constant density of states such that @xmath46 and @xmath47 , the best choice for @xmath48 is @xmath49 which gives @xmath50 in cases with an exponential density of states , for which we will get equilibration towards @xmath51 , we can guarantee that @xmath52 is small whenever @xmath53 ( compare appendix h ) . first note that by monotonicity of the trace distance and the triangle inequality @xmath54 where @xmath55 and @xmath56 are the projectors onto the support of @xmath29 and @xmath36 respectively and @xmath57 , and @xmath58 .
it remains to bound @xmath59 .
let @xmath60 and @xmath61 ; then @xmath62 , and thus @xmath63 . to bound @xmath64 we decompose @xmath65 into an interior part @xmath66 which is the projector onto the eigenstates from the interval @xmath67 $ ] and the exterior part @xmath68 and find @xmath69 ( see fig .
[ fig : projectors ] ) . using the inequality @xmath70 , submultiplicativity of the rank , and that @xmath71 this can be recast into @xmath72 .
finally , from theorem v.ii.3.1 in ref .
@xcite it follows that @xmath73 . repeating the argument for @xmath74 , introducing the notation @xmath75 , and putting everything together gives the desired result .
definition of the projectors used in the proof of theorem [ theorem : interactingvsnoninteractingcase ] . ] the level counting argument ( iii ) with is ultimately the reason for the exponential form of @xmath51 carries over to the quantum case in a straightforward way in the _ absence _ of coupling between system and bath @xcite , and with a bit more work one can also obtain a rigorous trace norm error bound .
if the number of states of the bath @xmath76 in the interval @xmath77 $ ] is such that the proportion @xmath78 is close to @xmath79 for the given @xmath80 and @xmath81 and some @xmath17 , then the distance of @xmath82 is small .
this can be guaranteed under a set of natural assumptions that are satisfied by a wide range of natural quantum many - body systems and that resemble the ones commonly used in classical statistical physics , such as an exponential increase of the density of states ( appendix a ) . in particular , for a bath consisting of @xmath83 noninteracting spin-@xmath84 particles with a slightly varying on site field strength and average local energies of @xmath85 and @xmath86 one finds ( appendix b ) @xmath87 with @xmath88 exponentially small in the bath size .
we will later use this bath in our algorithm . in summary , eq . ,
theorem [ theorem : interactingvsnoninteractingcase ] , and the results on dynamical equilibration and random states from the unitary invariant measure derived in refs .
@xcite lead to the following conclusions : [ theorem : summaryofnaturalthermalization ] ( kinematic ) almost all pure states @xmath31 from a microcanonical subspace corresponding to an energy interval @xmath26 $ ] of a weakly interacting , sufficiently large quantum system are locally close to a gibbs state in the sense that for every @xmath89 the probability that @xmath90 drops of exponentially with @xmath91 .
( dynamic ) moreover , if the hamiltonian in addition has nondegenerate energy gaps @xcite , all initial states @xmath92 , even those locally out of equilibrium , with a flat energy distribution in the interval , locally equilibrate towards @xmath93 in the sense that @xmath94 both inequalities are robust against deviations from the rectangular distribution .
if the bath has an exponentially increasing density of states only a region of bounded variation followed by a sharp cutoff towards higher energies should be sufficient ( for details see appendix c ) .
[ [ artificial - thermalization - a - quantum - algorithm - for - gibbs - state - preparation . ] ] `` artificial thermalization '' : a quantum algorithm for gibbs state preparation .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + it follows from eq . and theorem [ theorem : interactingvsnoninteractingcase ] that all one has to do to prepare a gibbs state is to prepare a state close to @xmath29 or @xmath36 on a suitable combination of system plus bath .
this is the central idea behind the quantum circuit shown in fig .
[ fig : qcircuit ] , which prepares thermal states without using any knowledge about the eigenstates of the hamiltonian . quantum algorithms that prepare thermal states have several advantages over classical simulation methods : quantum monte - carlo methods offer a way to , for example , estimate correlation functions of thermal states on a classical computer . however , such methods are restricted to certain types of hamiltonians as they suffer from the sign problem . a procedure that certifiably prepares gibbs states in a quantum computer does not only overcome the sign problem , but moreover makes it possible to use thermal state in experiments addressing questions of nonequilibrium dynamics in quantum simulators , for example to study quenches .
our algorithm requires two registers ( see fig .
[ fig : qcircuit ] ) .
the first register @xmath95 consists of @xmath96 qubits initially in @xmath97 and is used to perform quantum phase estimation .
the second register @xmath98 holds the quantum system plus bath and is put into a rectangular state by performing the following steps : \1 . _ initialization_. the register @xmath98 is initialized into the completely mixed state @xmath99 .
\2 .
_ partial quantum phase estimation_. a new form of quantum phase estimation is performed , which comprises three steps : the application of @xmath96 hadamard gates on the qubits of @xmath95 , the application of @xmath96 controlled-@xmath100 operations ( with @xmath100 raised to successive powers of two ) , and an inverse fourier transform on @xmath95 . after this operation ,
the state of the registers is @xmath101 where @xmath102 and @xmath103 .
note that @xmath104 is a probability distribution that becomes more and more peaked around @xmath105 as @xmath96 increases .
\3 .
_ measurement_. measuring the first @xmath106 qubits of @xmath95 , some binary string @xmath107 of length @xmath106 is obtained and the system is left in the state @xmath108 where @xmath109 is the number of states of the ancilla register @xmath95 compatible with the measurement . by choosing @xmath96 one can determine the width @xmath110 of the rectangular state that is prepared .
the measured value of @xmath107 determines the energy @xmath111 of the rectangular state , and thereby the inverse temperature @xmath17 of the gibbs state . to thermalize the subsystem at some particular temperature
, the previous steps must be repeated until the desired energy is measured .
the number of runs increases exponentially with the inverse temperature @xmath17 .
this prevents us from preparing thermal states at very low temperatures ( see appendix d ) .
this is not a deficit of the algorithm , for otherwise qma - hard problems ( the quantum analog of np @xcite ) could be efficiently solved .
any general algorithm will presumably have this feature @xcite .
the final state of @xmath98 is @xmath112 for large enough @xmath96 , this state is close to the desired state @xmath29 with @xmath111 and @xmath110 . the precise deviation of @xmath113 from @xmath93
, @xmath114 depends on the density of states of system plus bath . a good candidate for the bath
is the system of @xmath83 non interacting spin-@xmath84 particles discussed before ( appendix b ) and we give explicit results for the errors and the complexity for this bath : quantum circuit that generates a dephased rectangular state @xmath115 .
@xmath116 is the initialization gate , @xmath117 are hadamard gates , @xmath118 , @xmath119 with @xmath10 , and @xmath120 is the inverse fourier transform . ] for any chosen @xmath121 , any given inverse temperature @xmath17 and system hamiltonian @xmath21 , the algorithm presented in fig .
[ fig : qcircuit ] , using the bath with @xmath83 spin-@xmath84 particles and energy scale @xmath122 discussed before ( appendix b ) , prepares the system @xmath0 of @xmath123 qubits in a state within trace norm distance bounded by @xmath124 with @xmath88 exponentially small in @xmath83 , to a gibbs state @xmath93 with a temperature in the interval @xmath125 $ ] , where @xmath126 this is achieved using @xmath96 ancilla qubits and running the algorithm an average number of @xmath127 times , where each run requires the application of @xmath128 hadamard gates , @xmath96 controlled single qubit gates , @xmath129 ( with @xmath130 ) single qubit measurements , and @xmath131 controlled unitary time evolutions under @xmath10 for a time @xmath132 .
notice that the time evolution under @xmath22 can be implemented with @xmath83 gates as the bath is a model of uncoupled spins . in practice , in absence of an oracle for the hamiltonian of the system , the error produced to perform the @xmath100 gate carries a second source of error that comes from the trotter - suzuki approximation .
nevertheless , this error can be suppressed at a polynomial cost for local hamiltonians @xcite .
the two contributions to the trace distance error are computed in appendices b and f , the average number of runs is computed in appendix f and the error in the temperature comes from the discrete nature of the energy measurement via quantum phase estimation and is calculated in appendix e. as is clear form fig . [ fig : qcircuit ] , we need @xmath133 of the @xmath100 gates , and the part of the circuit that does not correspond to the controlled time evolution , i.e. , the initialization and the inverse fourier transform , only requires the implementation of @xmath134 hadamard gates , @xmath96 controlled single qubit gates , and @xmath135 single qubit measurements . for a fixed large @xmath136 , the error of the algorithm can be made small by choosing an ancilla register of size @xmath137 .
the exponential scaling of the runtime with @xmath138 caused by this is not a deficit of the algorithm , as any general efficient algorithm would contradict hardness results such as the local hamiltonian problem @xcite . unlike our approach
, metropolis algorithms @xcite are expected to be efficient in some cases , but no rigorous runtime estimates are known and they are only applicable to local hamiltonians . an interesting step towards constructing an efficient ( in system size ) and certified algorithm for local systems
was recently made in ref .
@xcite ( compare appendix g ) .
[ [ conclusions . ] ] conclusions .
+ + + + + + + + + + + + a set of sufficient conditions for thermalization in quantum mechanics has been presented .
the conditions are a natural translation of the standard assumptions from classical statistical physics . along the way ,
a perturbation argument for realistic weak coupling has been proven that we expect to have significant applications beyond the scope of this letter . by using our technical results we are able to design a quantum algorithm preparing thermal states of arbitrary hamiltonians with rigorous runtime and error bounds .
[ [ acknowledgments . ] ] acknowledgments .
+ + + + + + + + + + + + + + + + we would like to thank m. p. mller for discussions , p. reimann for comments , the anonymous referee a for the well justified criticism , and the eu ( qessence , compas , minos ) , and the euryi for support .
34 s. popescu , a. j. short , and a. winter , nature phys .
* 2 * , 754 ( 2006 ) .
s. goldstein , j. l. lebowitz , r. tumulka , and n. zangh , phys .
rev .
lett . * 96 * , 050403 ( 2006 ) .
h. tasaki , phys .
rev .
lett . * 80 * , 7 ( 1998 ) .
h. dong , s. yang , x. f. liu , and c. p. sun , phys .
rev .
a * 76 * , 044104 ( 2007 ) .
p. reimann , phys .
rev .
lett .
* 101 * , 190403 ( 2008 ) ; n. linden , s. popescu , a. j. short , and a. winter , phys .
rev .
e * 79 * , 061103 ( 2009 ) ; a. j. short , arxiv:1012.4622 .
p. reimann , new j. phys .
* 12 * , 055027 ( 2010 ) .
n. linden , s. popescu , a. j. short , and a. winter , new j. phys .
* 12 * , 055021 ( 2010 ) ; c. gogolin , phys . rev .
e * 81 * , 051127 ( 2010 ) . c. gogolin , m. p. mller , and j. eisert , phys .
rev .
lett . * 106 * , 040401 ( 2011 ) ; o. lychkovskiy , phys .
rev .
e * 82 * , 011123 ( 2010 ) .
m. cramer , c. m. dawson , j. eisert , and t. j. osborne , phys .
rev .
lett . * 100 * , 030602 ( 2008 ) ; t. barthel and u. schollwck , ibid .
* 100 * , 100601 ( 2008 ) .
m .- h . yung , d. nagaj , j. d. whitfield , and a. aspuru - guzik , phys .
rev . a * 82 * , 060302(r ) ( 2010 ) .
k. temme _ et al .
_ , arxiv:0911.3635
. m. p. mller , d. gross , and j. eisert , commun .
math .
phys . * 303 * , 785 ( 2011 ) . c. kollath , a. luchli , and e. altman , phys .
rev .
lett . * 98 * , 180601 ( 2007 ) ; s. manmana _ et al .
_ , ibid .
* 98 * , 210405 ( 2007 ) ; a. flesch _ et al . _ ,
phys .
rev .
a * 78 * , 033608 ( 2008 ) ; m. rigol , phys . rev .
lett . * 103 * , 100403 ( 2009 ) ; m. c. banuls , j. i. cirac , and m. b. hastings , ibid . *
106 * , 050405 ( 2011 ) .
m. greiner _ et al .
_ , nature * 415 * , 39 ( 2002 ) ; l. e. sadler _ et al .
_ , ibid .
* 443 * , 312 ( 2006 ) ; s. hofferberth _ et al . _ ,
ibid . * 449 * , 324 ( 2007 ) ; s. trotzky _ et al .
_ , arxiv:1101.2659 .
l. d. landau and e. m. lifschitz , _ course of theoretical physics _ , 3rd edition , pergamon , oxford ( 1980 ) .
d. poulin and p. wocjan , phys .
rev .
lett . * 103 * , 220502 ( 2009 ) .
e. bilgin and s. boixo , phys .
rev .
lett . * 105 * , 170405 ( 2010 ) .
r. bhatia , _ matrix analysis _ ( springer verlag , new york , 1997 ) .
j. kempe , a. kitaev , and o. regev , siam journal of computing , * 35 * , 1070 ( 2006 ) .
d. w. berry _ et al .
_ , comm . math
. phys . *
270 * , 359 ( 2007 ) .
see the introduction of ref .
@xcite .
reference @xcite only deals with the non interacting case . in ref .
@xcite weak means that @xmath139 is much smaller than the level spacing of @xmath21 , which is well defined in the limit of large baths , but still breaks down quickly even for relatively small subsystems and moreover the rigorous results are limited to cases where @xmath13 only couples neighboring eigenstates . ref .
@xcite contains a non rigorous discussion of the general case and numerical results suggesting that couplings much larger than the level spacing can be sufficiently weak for gibbs states to emerge ; a fact made rigorous in this work . private communication with the authors of ref . @xcite . |
the absence of the hydrogen gunn - peterson trough in quasar absorption spectra ( @xcite ) strongly indicates that the smoothly distributed hydrogen in the intergalactic medium ( igm ) is already highly ionized by @xmath10 ( e.g. , @xcite ; @xcite ) .
recently , becker et al . ( 2001 ) reported the first detection of the gunn - peterson test through in a @xmath11 qso spectrum .
however , even a small fraction of neutral hydrogen in the igm would result in an undetectable flux because of the large cross section to ly@xmath12 photons , and thus further careful investigations will be necessary .
these results also imply the presence of very luminous ionizing sources such as qsos and high mass stars at high redshifts to cause photoionizations of hydrogen by @xmath13 . on the other hand
, it is known that earlier reionization would have brought the last scattering surface to lower redshift , smoothing the intrinsic anisotropies in cosmic microwave background ( cmb ) radiation generated at the recombination epoch @xmath14 ( @xcite ; @xcite ) .
the recent measurements of the cmb anisotropies on subdegree angular scales ( @xcite ; @xcite ; @xcite ) can be used to set an upper limit on reionization redshift such as @xmath15 ( @xcite ) .
thus , under the currently favored adiabatic cold dark matter ( cdm ) scenario for the structure formation , reionization of the universe is expected to occur in the range of @xmath16 . revealing how reionization proceeds in the inhomogeneous universe stands out as one of the most important problems in the observational cosmology
( e.g. , see @xcite for a recent review ) .
secondary cmb anisotropies should be induced by scattering of cmb photons off electrons in the ionized medium undergoing bulk motion in the large - scale structure .
it is therefore expected that the secondary anisotropies can provide a laboratory for studying structure formation in the reionized epoch .
kaiser ( 1984 ) investigated the cmb distortion induced by the bulk velocity to pure linear order of @xmath3 and found that the induced anisotropies are small at arcminite scales resulting from cancellations between positive and negative shifts along the line of sight , since the gravitationally induced bulk velocities are irrotational .
ostriker & vishniac ( 1986 ) then pointed out that the coupling of the doppler effect with spatial variations in the free electron density produces @xmath17 anisotropies on arcminite scales at the order of @xmath18 , which is the so - called ostriker - vishniac ( ov ) effect , because the effect is a second - order contribution of perturbations that does not suffer from the cancellation ( see also @xcite ; @xcite ; @xcite ; @xcite ) . for
the same reason , spatial inhomogeneities of ionization fraction associated with a realistic patchy reionization scenario should lead to the secondary cmb anisotropies by the doppler effect ( @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) .
in particular , benson et al . ( 2001 ) investigated the secondary cmb fluctuations under a realistic model of the reionization of igm caused by starts in high - redshift galaxies , based on a semi - analytic model of galaxy formation and a high - resolution n - body simulation of the dark matter clustering .
they showed that measurements of cmb anisotropies could actually put stringent constraints on the reionization processes . on the other hand , from theoretical aspects ,
it is not so evident to answer the problem which second - order effect of perturbations produces dominant contribution to the secondary cmb fluctuations .
hu , scott , & silk ( 1994 hereafter hss ; see also dodelson & jubas 1995 ) derived a complete boltzmann equation including the quadratic doppler effects @xmath19 of bulk velocity that governs redshift evolution of the cmb temperature fluctuation field . consequently , they found that , although the quadratic doppler effect should induce spectral distortion of cmb like the thermal sunyaev - zeldovich ( sz ) effect ( @xcite ) as a new qualitative feature , the magnitude is always smaller than that of the ov effect for the cdm models . a great uncertainty in models for structure formation in the reionized epoch is the extent and nature of reionization , because this requires the understanding of various complicated astrophysical processes such as the nature of first ionizing sources , the fraction of ionizing photons that escape the sources to become available for photoionization of the hydrogen in igm , and the radiative balance between the escaped emissivity and the recombination rate in the inhomogeneous igm .
for example , even a naive question which high - density or low - density region in the large - scale structure becomes first photoionized bas not still been answered definitely .
assuming _ ab initio _ that the distribution of the ionization fraction is given as a function of gas density , miralde - escud , hanehnelt , & rees ( 2000 ; hereafter mhr ) suggested that the uv radiation may escape into and ionize the low - density region , where the recombination rates are the lowest , even though it is expected that the ionizing sources could be associated with the peaks in the density field .
this conclusion has also been confirmed by the recent numerical simulations taking into account the radiative transfer approximately ( @xcite ) .
while the relation between the dark matter and ionized medium distributions is thus a crucial clue to understanding the reionization history connected with the gravitational instability scenario of structure formation , the relation seems difficult to be directly observed by methods so far proposed .
therefore , the purpose of this paper is to investigate the cross - correlation between the dark matter distribution and the secondary cmb temperature fluctuation field induced by the ionized medium in the mildly nonlinear structure .
it is now recognized that the weak gravitational lensing effect due to the large - scale structure can be a unique tool to observationally reconstruct the dark matter distribution ( e.g. , see @xcite for a review ) .
there are two feasible methods often considered in the literature ; one is using the distortion effects on distant galaxy images that have been actually detected ( e.g. , @xcite ) , and the other is using the weak lensing effects on the cmb map ( @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite ; @xcite , b ) .
in particular , it is expected that the method of using the distortion effects on the cmb map developed by seljak & zaldarriaga ( 1999 ; also @xcite ) allows us to reconstruct the projected dark matter distribution with high accuracy .
based on these considerations , in this paper we assume that the weak lensing field is _ a priori _ reconstructed by those methods , for example , using the future cmb data from the satellite mission _
planck surveyor_. then , taking a cross - correlation between the weak lensing field and the cmb map including the secondary fluctuations would be a challenging test .
furthermore , the cross - correlation method could improve the signal - to - noise ( s / n ) ratio of the doppler effect compared with the original s / n in the cmb map itself if the secondary cmb field correlates with the weak lensing field , which is analogous to , for example , the cross - correlation between the cmb temperature fluctuation and polarization fields ( @xcite ; see also @xcite and @xcite for other examples ) . in this paper , based on the small angle approximation
, we first develop a formalism for calculating the cross - correlation function and its angular power spectrum between the weak lensing field and the secondary cmb fluctuation field including the doppler effects up to the second - order of the bulk velocity field .
we then estimate the magnitude of the cross - correlation under the currently favored cdm models and the simplest assumption of homogeneous reionization history after some epoch @xmath20 .
the layout of this paper is as follows . in [ prel ] we review the relevant properties and parameters of the adiabatic cdm scenario for structure formation . in
[ boltz ] we briefly review the boltzmann equation including the doppler effect up to the second - order of bulk velocity and discuss the physical implications of the effect following hss . in
[ form ] we develop a formalism for calculating the cross - correlation function and its angular power spectrum between the weak lensing field and the secondary cmb temperature fluctuation field induced by the bulk motion of the ionized medium in the large - scale structure .
results for the currently favored cdm models are presented in [ result ] . in [ conc ]
we present the discussions and conclusions . throughout this paper , we employ unit of @xmath21 for light speed @xmath22 .
the expansion rate of the universe can be described by the friedmann equation : @xmath23 where @xmath24 denotes the scale factor , @xmath25 is the hubble constant , @xmath26 and @xmath27 denote the density parameters of the present - day non - relativistic matter and the contribution of the cosmological constant relative to the critical density @xmath28 , respectively .
here we have set @xmath29 at present for convenience . throughout this paper , we consider only a flat universe model as supported by the recent high precision cmb experiments ( e.g. , @xcite ; @xcite ) and then we have @xmath30 . the conformal time which we will often use in the following discussions
is defined by @xmath31 .
it is convenient to introduce the comoving angular diameter distance @xmath32 with respect to @xmath33 or equivalently redshift @xmath34 , and for a flat universe model it is given by @xmath35 where @xmath36 is the present conformal time and @xmath34 is given by @xmath37 .
it is currently believed that the reionization of the universe occurred at @xmath16 such that the reionized medium is optically thin to the thomson scattering of cmb photons , which is characterized as @xmath38 in terms of the optical depth , @xmath39 .
assuming homogeneous ionization after some reionization epoch @xmath40 , @xmath39 can be analytically expressed as a function of @xmath34 for a flat universe model : @xmath41 , \label{eqn : opt}\ ] ] with @xmath42 here @xmath43 is the thomson cross section , @xmath44 is the mean number density of free electron , @xmath45 is the mean ionization fraction , which does not depend on time here , @xmath46 is the present - day density parameter of baryon in units of the critical density , @xmath47 is the optical depth to thomson scattering to the hubble distance today under the assumption of full hydrogen ionization and homogeneous distribution of baryon , and @xmath48 is the primordial helium fraction .
we fix @xmath49 throughout this paper . again in the case of homogeneous and full hydrogen ionization , the visibility function that gives the probability of last scattering within @xmath50 of @xmath33 is expressed as @xmath51 where overdot denotes derivative with respect to the conformal time @xmath33 .
the evolution of the large - scale structure after the decoupling epoch @xmath52 is characterized by the time evolution and scale dependence of the density contrast field @xmath53 mainly caused by the dark matter distribution .
the gravitational instability then leads to bulk motions of structure with the velocity field @xmath54 , and we expect that the ionized baryonic medium also follows the bulk motion on large scales , leading to the doppler effects on cmb photons through the compton scattering by electrons .
since we focus our investigations on the secondary cmb fluctuations up to the second - order of @xmath2 , we need to consider the fields of @xmath55 and @xmath56 up to the second - order that are valid in the mildly nonlinear regime such as @xmath57 : @xmath58 the quantities with superscripts ` ( 1 ) ' and ` ( 2 ) ' denote the first- and second - order perturbation quantities , respectively .
the time dependences of these fields can be expressed in terms of cosmological parameters based on the ( eulerian ) perturbation theory ( e.g. , @xcite ) . in the linear regime ,
the time dependence of @xmath59 is given by the growing factor as @xmath60 with @xmath61 where we have neglected the radiation energy contribution , and will normalize @xmath62 at the present time @xmath63 for convenience in the followings .
the fourier transformation of @xmath64 is expressed as @xmath65 if the linear density fluctuations are gaussian , the statistical properties can be completely specified by the power spectrum @xmath66 defined by @xmath67 where @xmath68 denotes an ensemble average over all realizations .
note that time evolution of the power spectrum is given by @xmath69 .
in this paper , assuming the adiabatic cdm scenario , we employ the harrison - zeldovich spectrum and the bbks transfer function ( @xcite ) with the shape parameter given by sugiyama ( 1995 ) as for the shape of @xmath66 .
the free parameter is then the normalization of @xmath66 only , which is conventionally expressed in terms of the rms mass fluctuations of a sphere of @xmath70mpc radius , i.e. , @xmath71 . from the continuity equation
, we can obtain the fourier components of the bulk velocity filed @xmath72 at the first order as @xmath73 likewise , based on the perturbation theory , we can obtain the second - order fourier components of density fluctuation and bulk velocity fields in terms of @xmath74 and @xmath75 ( e.g. , see @xcite ) : @xmath76 with @xmath77 where we have neglected extremely weak dependences of the functions @xmath78 and @xmath79 on cosmological parameters @xmath26 and @xmath27 for simplicity , and considered the scalar type perturbations only .
it is worth noting that , even if perturbations start from gaussian in the linear regime , the mildly nonlinear evolution leads to non - vanishing third - order moments such as @xmath80 and @xmath81 , because those quantities are essentially the fourth - order moments of linear perturbation quantities like @xmath82 .
here , following hss , we briefly review the derivation of the secondary cmb temperature fluctuation field induced by the doppler effect of ionized medium up to the second order @xmath19 of the bulk velocity .
let us first write down the boltzmann equation in terms of the cmb photon energy @xmath83 and direction @xmath84 in the presence of compton scattering between the cmb photons and free electrons that could lead to energy changes ; @xmath85 where @xmath86 is the photon occupation number and @xmath87 is the collision term by the compton scattering .
hss derived the collision term that leads to the doppler effect up to the second order @xmath19 as @xmath88p\frac{\partial f}{\partial p } + \left(\frac{3}{20}v^2+\frac{11}{20}({\mbox{\boldmath$\gamma$}}\cdot{\mbox{\boldmath$v$}})^2\right ) p^2\frac{\partial^2f}{\partial p^2}\right\ } \right ] , \label{eqn : collision}\ ] ] where we have ignored the term by thermal electron scattering , leading to the thermal sz effect . the first term in the large brackets on the r.h.s represents the doppler effect of @xmath3 that produces the ov effect from the coupling with the electron density fluctuation field @xmath9 at the order @xmath89 .
the other terms give the quadratic doppler effect of @xmath19 .
it is interesting that the effect leads to energy exchanges between cmb photons and electrons like the thermal sz effect , and therefore causes the compton @xmath90-distortion . as shown by hss
, we can get more physical insight into this effect by considering the limit of many scattering blobs or equivalently the multiple scattering limit .
if we can average equation ( [ eqn : collision ] ) over the direction of bulk velocity field @xmath56 , the @xmath3 doppler effect primarily cancels in this case and then we can obtain @xmath91 .
\label{eqn : manyblob}\ ] ] if introducing the effective temperature @xmath92 defined by @xmath93 ( @xmath94 is the electron mass ) , this collision term reduces to the same form of that in the kompaneets equation with the thermal temperature @xmath95 of hot electron , which gives the thermal sz effect .
this implies that many ionized blobs distributed in the large - scale structure can be considered like hot electrons inside one cluster , and thus those blobs induce the energy exchange to the cmb photons characterized by the effective temperature @xmath92 . under the cdm scenario for the structure formation
, the amplitude of compton @xmath90-distortion induced by the quadratic doppler effect is below that due to the thermal sz effect mainly caused by intracluster hot plasma .
this is because we can typically estimate @xmath96 for the gravitationally induced bulk flow and @xmath97 for intracluster hot plasma , and the fact of @xmath98 leads to the conclusion , even though clusters are rare objects in the universe . in this paper
, we do not employ the many scattering blobs limit for the purpose of generality as done by hss .
namely , we shall no longer be concerned with frequency dependence of the temperature fluctuation field .
integration of equation ( [ eqn : boltz1 ] ) over photon energy @xmath83 leads to the boltzmann equation up to the order @xmath19 to govern the time evolution of temperature fluctuation field , @xmath99 , along the line of sight with direction of @xmath100 : @xmath101,\ ] ] where we have neglected the gravitational effect and the correction due to the quadrupole temperature fluctuations . by solving the equation above formally ,
we can obtain the secondary fluctuation field induced by the doppler effect of ionized medium during propagations of the cmb photons from the decoupling @xmath52 to present : @xmath102 \left[{\mbox{\boldmath$\gamma$}}\cdot{\mbox{\boldmath$v$}}-v^2 + 7({\mbox{\boldmath$\gamma$}}\cdot{\mbox{\boldmath$v$}})^2 \right ] , \label{eqn : fullcmb2nd}\ ] ] where @xmath103 is the conformal time at the decoupling epoch @xmath104 , @xmath105 is the visibility function given by equation ( [ eqn : visib ] ) , and we have used the fact that the _ free _ electron density field can be expressed as @xmath106\nonumber\\ & = & \bar{n}_{\rm h0}a^{-3}(\eta)x_e(\eta)(1+\delta_e ) .
\label{eqn : elenum}\end{aligned}\ ] ] the @xmath107 is the ionization fraction as a function of @xmath33 , @xmath9 is the density fluctuation field of electron along the line of sight , and @xmath108 is the mean hydrogen number density at present given by @xmath109 . note that , as explicitly introduced by gnedin & jaffe ( 2001 ) , @xmath110 can also be expressed as @xmath111 [ 1+\delta_b({\mbox{\boldmath$x$}},\eta)]$ ] , where @xmath112 and @xmath113 are the fluctuation fields of the ionization fraction and the baryon density , respectively , as a function of the spatial position and time .
it is thus clear that @xmath9 is different from @xmath113 in a general reionization history .
note that we will deal with @xmath9 throughout this paper .
in this section , we develop a formalism for calculating the cross - correlation function and its angular power spectrum between the weak lensing field and the secondary cmb anisotropies field induced by the doppler effect of the ionized medium .
the weak gravitational lensing due to the large - scale structure is now recognized as a unique tool for directly measuring the dark matter distribution .
seljak & zaldarriaga ( 1999 ) ( also see @xcite ) developed a useful method for reconstructing the projected field of dark matter fluctuations between the last scattering surface and present from the lensing distortions to the product of gradients of the temperature filed .
this method could be successfully performed by the satellite mission planck .
it has been also shown that a new power of temperature fluctuations generated by the weak lensing effect on arcminute scales below the silk damping scale can be directly used to reconstruct the projected dark matter field , provided that we have data with arcminute - scale spatial resolution and sensitivity of @xmath114 ( @xcite ; @xcite ; @xcite ) and we can remove other secondary signals such as the sz effect by taking advantage of the specific spectral properties or by combining the cmb measurement with the @xmath115-ray or the sz effect itself . furthermore ,
if measurements of the cmb polarization become possible , combinations of the lensing effects on the temperature fluctuation and the @xmath116- and @xmath117-type polarization fields can improve the reconstruction ( @xcite ; @xcite ) .
this is because the lensing effect induces the @xmath117-type polarization on the observed sky from the non - linear coupling to the primary @xmath116-mode , evin if the primary cmb contains the @xmath116-mode only as suggested by starndard inflationary scenarios , and there is negligible cosmological contamination of the polarization field in the arcminute regime unlike the temperature fluctuations ( @xcite ) . the projected dark matter field
is called the convergence field , @xmath0 , and can be expressed in the form of a weighted projection of the three - dimensional density fluctuation field along the line of sight ; @xmath118 where @xmath119 is the lensing weight function defined by @xmath120 on the other hand , the lensing distortion effect on distant galaxy images can be similarly used to measure the relatively low redshift dark matter distribution such as @xmath121 , depending on typical redshift of source galaxies ( @xcite ) . in this case , the lower limit in the integration ( [ eqn : conv ] ) becomes the conformal time corresponding to the redshift of source galaxies @xmath122 .
however , we should bear in mind that the lensing effect on galaxies is relatively irrelevant for the purpose of this paper , because this method usually restricts the convergence field to low redshifts ( e.g. see @xcite ) , while the reionization signal in the cmb map is imprinted at moderate and high redshifts .
the weak lensing effects on cmb thus has a great advantage of probing the dark matter distribution up to high redshift , which can not be attained by any other means ( see also @xcite ; @xcite ; @xcite ) . in this paper
, we assume that the convergence field @xmath123 with respect to direction @xmath100 is _ a prior _ reconstructed from measurements of the lensing effects on cmb . from equation ( [ eqn : fullcmb2nd ] )
, the secondary temperature fluctuation field induced by the @xmath3 doppler effect can be written as @xmath124 the pure linear order doppler effect , @xmath125 , is suppressed on small scales because the gravitational instability generates irrotational flow , leading to cancellations between positive and negative doppler shifts along the line of sight ( @xcite ) .
for this reason , leading contribution to the secondary temperature fluctuations due to the doppler effect comes from second order perturbations that do not suffer from the cancellation like the ov effect of @xmath89 ( @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) and the patchy ionization effect of @xmath126 ( @xcite ; @xcite ; @xcite ; @xcite ) , where @xmath112 is the spatial inhomogeneities of ionization fraction @xmath127 as discussed below equation ( [ eqn : elenum ] ) . from equations
( [ eqn : conv ] ) and ( [ eqn:1stdp ] ) , a cross - correlation between the convergence field and the secondary temperature fluctuation field due to the doppler effects of @xmath3 and @xmath89 depend on the moments @xmath128 and @xmath129 of perturbations , respectively , where the latter moment seems to produce some contributions in the quasi nonlinear regime as discussed in [ lss ] .
however , it is clear that , since those moments has linear dependence on the bulk velocity field @xmath56 , the cross - correlation is suppressed on relevant small angular scales because of the cancellations between positive and negative contributions from the bulk velocity field for the ensemble average .
note that the two - point correlation function of the ov effect or equivalently its angular power spectrum arises from the quadratic contribution of the velocity field as @xmath130 , whereas it does not suffer from the cancellation ( @xcite ; @xcite ; @xcite ) .
we next consider secondary temperature fluctuations induced by the quadratic doppler effect of the bulk velocity . in the case of assuming the homogeneous ionization history ,
the secondary fluctuations are divided into two contributions ; the doppler effect of @xmath19 and the density - modulated effect of @xmath131 . as will be shown later
, the effect of @xmath131 generally produces dominant contribution on small angular scales @xmath6 , where the nonlinear structures with @xmath132 are important . from equation ( [ eqn : fullcmb2nd ] )
, the secondary fluctuation field of @xmath19 is given by @xmath133 .
\label{eqn:2dpaniso}\ ] ] the bulk flow in adiabatic cdm models arises mainly from the linear regime . in the @xmath134cdm model ,
half the contribution to @xmath135 comes from scales @xmath136 , and the density fluctuations go nonlinear at @xmath137 .
this is the reason that the secondary cmb anisotropies by the doppler effects of @xmath19 and @xmath131 are the order of magnitude smaller than the anisotropies by the ov effect of @xmath89 on relevant angular scales .
however , from the fact that there is no cross - correlation between the convergence field and the ov effect , the quadratic doppler effects can provide primary contribution to the cross - correlation between the convergence field and the secondary cmb temperature fluctuations induced by the ionized medium . since we are interested in the cross - correlation on small angular scales such as @xmath138 , it is useful to employ the small angle approximation originally developed by bond & efstathiou ( 1987 ) . in this case ,
the unit vector @xmath100 in equation ( [ eqn:2dpaniso ] ) , which denotes the direction of line of sight , can be considered to have component as @xmath139 for @xmath140 in the small region around the north pole on the celestial sphere .
it is then convenient to introduce the two - dimensional vector @xmath141 , which has @xmath142 . within the context of the small angle approximation ,
the cross - correlation function between the convergence field @xmath0 and the temperature fluctuation field @xmath143 separated by the angle of @xmath144 is defined by @xmath145 where @xmath146 and @xmath147 are the two - dimensional vectors corresponding to unit vectors @xmath148 and @xmath149 for @xmath0 and @xmath143 , respectively .
furthermore , if using the two - dimensional fourier transformation as , for example , @xmath150 , the cross - correlation function can be expressed in terms of its angular power spectrum by the hankel transform of zeroth order as @xmath151 where @xmath152 is the zeroth order bessel function .
the angular power spectrum @xmath153 can be also expressed in terms of the two - dimensional fourier components of @xmath0 and @xmath143 as @xmath154 from equations ( [ eqn : conv ] ) and ( [ eqn:2dpaniso ] ) , the cross - correlation function ( [ eqn : twogl2dp ] ) can be further calculated as @xmath155 where @xmath156 with @xmath157 , and the functions @xmath158 and @xmath159 are given by equations ( [ eqn : visib ] ) and ( [ eqn : lensweight ] ) , respectively . in the second line on the r.h.s
we have replaced the ionization fraction of @xmath127 , which generally depends on conformal time @xmath33 , with the average fraction , @xmath160 , between @xmath161 assuming that @xmath162 is a slowly varying function with respect to conformal time after a given reionized epoch @xmath163 . from equation ( [ eqn : pert ] ) , the ensemble average @xmath164 can be rewritten in terms of third order moments of perturbations as @xmath165({\mbox{\boldmath$\theta$}}_b)}\rangle}\nonumber\\ & & + { \langle{\delta^{(2)}({\mbox{\boldmath$\theta$}}_a ) [ -v^{(1)}_iv^{(1)}_i+7v_z^{(1)}v^{(1)}_z]({\mbox{\boldmath$\theta$}}_b)}\rangle}+o((\delta^{(1)})^6 ) .
\end{aligned}\ ] ] here we have used @xmath166 and , from the fact that the propagation direction of cmb photons is almost parallel to @xmath167-direction in our setting , we have used @xmath168 , where @xmath169 denotes the @xmath34 component of bulk velocity field . therefore , using equations ( [ eqn : pecvel ] ) and ( [ eqn:2pert ] ) , we can obtain @xmath170(\eta ' ) \int\!\!\frac{d^3{\mbox{\boldmath$k$}}}{(2\pi)^3}\int\!\!\frac{d^3{\mbox{\boldmath$k$}}'}{(2\pi)^3 } \frac{7k'_z(k_z+k'_z)-k^{\prime 2}-{\mbox{\boldmath$k$}}\cdot{\mbox{\boldmath$k$ } } ' } { k^{\prime 2}|{\mbox{\boldmath$k$}}+{\mbox{\boldmath$k$}}'|^2}\nonumber\\ & & \times \left[4g(-{\mbox{\boldmath$k$}},{\mbox{\boldmath$k$}}+{\mbox{\boldmath$k$}}')+f(-{\mbox{\boldmath$k$}}',{\mbox{\boldmath$k$}}+{\mbox{\boldmath$k$ } } ' ) \right]p(k)p(|{\mbox{\boldmath$k$}}+{\mbox{\boldmath$k$}}'|)e^{i{\mbox{\boldmath$k$}}_\perp\cdot ( r{\mbox{\boldmath$\theta$}}_a - r'{\mbox{\boldmath$\theta$}}_b)-ik_z(\eta-\eta ' ) } , \label{eqn : glv2}\end{aligned}\ ] ] where @xmath171 are @xmath172-components of @xmath173 that are perpendicular to the line of sight , and @xmath174 and @xmath175 denote the comoving angular distances from the present time to @xmath33 and @xmath176 , respectively .
one can readily see that the integration of the term @xmath177 over @xmath178 on the r.h.s of equation ( [ eqn : glv2 ] ) leads to the dirac delta function @xmath179 as long as we are concerned with wavelength modes much smaller than the comoving hubble distance characterized by @xmath33 .
this leads to the consequence that dominant contribution to the cross - correlation function arises from structures of dark matter and ionized medium that are distributed at the same redshift and are separated by @xmath144 on the observed sky .
this is the so - called the fourier - space analogue of limber s equation ( @xcite ; @xcite ) .
it is known that this flat sky approximation causes errors no greater than @xmath180 at @xmath6 ( @xcite ; @xcite ) . under this approximation
, the cross - correlation function @xmath181 can be calculated as @xmath182 with @xmath183p(k ' ) p(\sqrt{k^2+k^{\prime 2}+2kk'\cos\phi\sqrt{1-\mu^2}}),\end{aligned}\ ] ] where we have assumed that @xmath184 is along the @xmath185-coordinate from the statistical isotropy and then @xmath186 .
hence , from equations ( [ eqn : defangcl ] ) and ( [ eqn : twobisp ] ) we can finally derive the angular power spectrum of the cross - correlation between the convergence field and the secondary temperature fluctuation field by the quadratic doppler effect of @xmath19 : @xmath187 this is the first main result of this paper .
it should be noted that the integrand quantity in @xmath188 depends on the cosmological parameters and on the dark matter power spectrum .
namely , it is not affected by the unknown biasing relation between the dark matter and ionized medium distributions . in this sense
, we can accurately predict the magnitude of the integrand quantity once we fix the dark matter power spectrum such as the cdm model and the cosmological parameters , which will be well constrained in the near future by other observations such as measurements of the cmb anisotropies , the weak lensing survey and so on . we therefore expect that measurements of @xmath153 can set precise constraints on the average ionization fraction @xmath160 between the reionization epoch and present .
next , we consider the secondary temperature fluctuation field induced by the density - modulated quadratic doppler effect of @xmath189 . from equation ( [ eqn : fullcmb2nd
] ) we have @xmath190.\ ] ] in the following , to relate @xmath9 to the dark matter fluctuation field @xmath55 , we introduce the linear biasing parameter @xmath191 expressed as @xmath192 for simplicity .
note that @xmath191 generally depends on time and spatial position .
it is relatively straightforward to calculate the cross - correlation function between the convergence field and the temperature fluctuations field @xmath193 : @xmath194(\eta',{\mbox{\boldmath$\theta$}}_b)}\right\rangle}\nonumber\\ & \approx & { \langle{x_e b}\rangle}\int\!\!d\eta \int\!\!d\eta ' w(\eta,\eta_\ast)g(\eta ' ) { \left\langle{\delta(\eta,{\mbox{\boldmath$\theta$}}_a)[\delta\left\{-v^2 + 7({\mbox{\boldmath$\gamma$}}_b\cdot { \mbox{\boldmath$v$}})^2\right\}](\eta',{\mbox{\boldmath$\theta$}}_b)}\right\rangle } \nonumber\\ & = & { \langle{x_eb}\rangle}\int\!\!d\eta \int\!\!d\eta ' w(\eta,\eta_\ast)g(\eta ' ) { \langle{\delta\delta}\rangle}{\langle{-v^2 + 7({\mbox{\boldmath$\gamma$}}_b\cdot{\mbox{\boldmath$v$}})^2}\rangle}\nonumber\\ & = & { \langle{x_eb}\rangle}\int\!\!d\eta \int\!\!d\eta ' w(\eta,\eta_\ast)g(\eta ' ) { \langle{\delta(\eta,{\mbox{\boldmath$\theta$}}_a)\delta(\eta',{\mbox{\boldmath$\theta$}}_b)}\rangle } \frac{4}{3}{\langle{v^2}\rangle},\end{aligned}\ ] ] where @xmath195 is the average quantity of @xmath196 between the reionization epoch and present . after the similar procedure as used in the derivation of equation ( [ eqn : clbis ] ) , we can obtain the angular power spectrum @xmath197 where @xmath198 is the rms of the present - day bulk velocity field and
can be expressed in terms of the linear density power spectrum as @xmath199 here the factor @xmath200 is due to the three - dimensional freedom of bulk velocity field such that @xmath198 represents the rms of one - dimensional motion .
the most important result of equation ( [ eqn : cldens ] ) is that the angular power spectrum @xmath201 is proportional to not quadratic but linear term of @xmath195 for all @xmath202 and therefore could be both positive and negative , depending on the sign of @xmath195 or more specifically on the sing of @xmath203 , since the integrand quantity in equation ( [ eqn : cldens ] ) is always positive .
the quantity @xmath195 should be a crucial clue to understanding of how the ionized medium distribution is related to the dark matter distribution in the large - scale structure , which still remains large uncertainties both from theoretical and observational aspects . the measurements of @xmath201 can thus be a new direct probe of revealing the projected biasing relation .
it is worth noting that the angular power spectrum of the ov effect depends on the quadratic term @xmath204 and is always positive irrespective of the sing of @xmath205 .
the simple assumption that the ionized medium traces the dark matter distribution allows us to place the upper limit on @xmath206 using the nonlinear dark matter power spectrum , which has been well studied with n - body simulations for the full range of adiabatic cdm models .
equation ( [ eqn : cldens ] ) can be then rewritten as @xmath207 where @xmath208 is the power spectrum of nonlinear density fluctuations for which we will employ the fitting formula developed by peacock & dodds ( 1996 ) . note that we here consider a possible case that electrons in reionized medium with non - linear density fluctuations are distributed along the line of sight moving according to a large - scale coherent bulk velocity field , as suggested by the cdm power spectrum .
this is in fact an essential mechanism that the ostriker - vishniac effect can avoid cancellations between blue- and red shifts along the line of sight from the coupling of the density and velocity fields .
based on this consideration , in figure [ fig : cl_nl ] we use the non - linear power spectrum only for the calculation of @xmath201 and use the second - order perturbation theory for that of @xmath153 , because the latter arises from the ensemble average of the density fluctuation field and the bulk velocity filed on the mildly non - linear regime .
we present numerical results for the angular power spectra given by equations ( [ eqn : clbis ] ) and ( [ eqn : cldens ] ) for the @xmath134cdm model .
we adopt a set of cosmological parameters of @xmath209 , @xmath210 , @xmath211 , @xmath212 , and @xmath213 , which is consistent with the recent cmb experiments ( @xcite ) and the cluster abundance ( @xcite ; @xcite ) .
we simply consider the homogeneous hydrogen ionization model after a given reionization epoch in the range of @xmath214 . in this case
we will consider @xmath215 with respect to time during @xmath216 and equations ( [ eqn : opt ] ) and ( [ eqn : visib ] ) can be used for calculations of the optical depth and visibility function , respectively . although more realistic model is a patchy or inhomogeneous reionization ( e.g. , @xcite ) , our model allows us to compute the magnitude of effects in the simplest way .
we first estimate the order of magnitude of secondary temperature fluctuations by the quadratic doppler effect . using the @xmath134cdm model above
, the present - day rms of bulk velocity field can be estimated from equation ( [ eqn : rmsvel ] ) as @xmath217 in the unit of @xmath21 .
equation ( [ eqn : fullcmb2nd ] ) tells us that the magnitude of temperature fluctuations induced by the quadratic doppler effect can be roughly estimated as @xmath218 for @xmath219 , since we have @xmath220 for the model considered in this paper .
similarly , the magnitude of temperature fluctuations induced by the ov effect can be estimated as @xmath221 , which is consistent with the numerical results done by hu & white ( 1996 ) ( see also @xcite and @xcite ) .
we thus conclude @xmath222 in the observed cmb map itself , and the magnitude of @xmath143 is much below sensitivities of the satellite missions _
map _ and _ planck surveyor_. figure [ fig : cl ] shows the angular power spectra of the cross - correlations between the weak lensing field and the secondary temperature fluctuations induced by the quadratic doppler effect of @xmath19 ( dashed lines ) and the density - modulated effect of @xmath223 ( solid ) , which are computed using equations ( [ eqn : clbis ] ) and ( [ eqn : cldens ] ) , respectively .
we here assumed @xmath224 for the unknown biasing parameter between the dark matter and electron distributions in the large - scale structure .
both the magnitudes of the doppler effects of @xmath19 and @xmath8 are comparable on @xmath225 in the case of no biasing @xmath224 .
this is because leading contribution to the cross - correlation between the weak lensing field of @xmath226 and the doppler effect of @xmath19 comes from the fourth - order moments of linear perturbations like @xmath227 resulting from the vanishing of the third - order moments @xmath228 , while the cross - correlation from the density - modulated doppler effect of @xmath8 also arises from the moments of @xmath229 .
it is clear that the contribution from the density - modulated effect of @xmath131 dominates the effect of @xmath19 on angular scales of @xmath138 .
the total magnitudes of the angular power spectra peak @xmath230 around @xmath231 for the @xmath134cdm model and for the five cases of reionization epoch @xmath232 .
the peak location in @xmath202-space results from the shape of cdm power spectrum .
on the other hand , cooray ( 2000 ) investigated the angular power spectrum of the cross - correlation between the weak lensing field and the thermal sz temperature fluctuation field based on the halo approach of dark matter clustering ( @xcite ) and the simple model of intracluster gas distribution .
the magnitude of the angular power spectrum from the sz effect amounts to @xmath233 at @xmath234 for the similar models as considered in this paper .
the temperature fluctuation induced by the quadratic doppler effect is over an order of magnitude smaller than that by the sz effect because of @xmath235 and @xmath236 .
from this fact , it seems unlikely that the magnitude of the cross - correlation between @xmath0 and the quadratic doppler effect is smaller only by @xmath237 than that from the sz effect .
this is because the cross - correlation with the doppler effect comes from wide - range redshift structures such as @xmath238 , while the sz effect has dominant contributions from low redshift massive dark halos such as clusters of galaxies with @xmath239 at @xmath240 ( @xcite ) .
these results therefore imply that , if the weak lensing field is _ a priori _ reconstructed by measurements of lensing effects on the cmb , the cross - correlation could improve the s / n of the quadratic doppler effect , even though the angular power spectrum itself of the doppler effect has a small power of @xmath241 only .
this is an analogous statement to the cross - correlation between the primary cmb temperature fluctuation and polarization fields ( e.g. , @xcite ) . to explicitly illustrate how the cross - correlation considered could be useful to reveal the biasing relation between the dark matter and electron distributions , we here employ a toy model of the antibiasing relation , motivated by the reionization scenario of low - density region prior to the high - density region as suggested by mhr .
the homogeneous full hydrogen ionization with @xmath242 seems to lead to @xmath243 from @xmath244 , which seems inconsistent with the antibiasing relation , and thus we here assume the partly homogeneous ionization fraction of @xmath245 after @xmath163 .
although the antibiasing process depends on how reionization proceeds in the large - scale structure in detail , we consider the simplest model of linear and constant antibiasing relation given by @xmath246 with @xmath247 .
it should be then noted that the condition of @xmath248 leads to @xmath249 .
accordingly , we have taken into account the range of @xmath55 for the calculation of the ensemble average of @xmath250 used in equation ( [ eqn : cldens ] ) .
figure [ fig : antibias ] shows the result of this toy model , and one can readily see that we should observe the negative total angular power spectra ( solid lines ) of the cross - correlation between the weak lensing field and the secondary temperature fluctuation field induced by the doppler effect on @xmath251 .
note that this critical scale of @xmath202 depends on the value of the antibiasing parameter @xmath191 , because the cancellation between the spectra from the quadratic doppler effect of @xmath19 and the density - modulated effect of @xmath131 also depends on @xmath191 .
as expected , the cross - correlation function could thus be a direct probe of the antibiasing relation , which can not be attained by measurements of the angular power spectrum of the @xmath252 doppler effect such as the ov effect because of the quadratic dependence on the biasing relation .
figure [ fig : cl_nl ] shows the maximal estimate of the angular power spectrum of the cross - correlation using the nonlinear dark matter power spectrum under the simplest assumption that the gas traces dark matter distribution in the large - scale structure .
we here again considered the simplest model of @xmath242 and @xmath224 as in figure [ fig : cl ] .
the nonlinear enhancement could be important on small angular scales of @xmath253 .
however , it is not still clear if the electron distribution traces dark matter distribution in the nonlinear regime , because physical processes of the gas pressure and radiative dynamics should also play an important role in such nonlinear regions ( e.g. , @xcite ) . for this reason , the curves in figure [ fig : cl_nl ] provide us with upper limits on a realistic angular power spectrum of the cross - correlation
in this paper , we have investigated the cross - correlation between the weak lensing field and the secondary cmb temperature fluctuations induced by the doppler effect of the ionized medium in the reionized epoch of the universe .
the main result of this paper is to develop a formalism for calculating the cross - correlation function and its angular power spectrum based on the small angle approximation .
the leading order to the cross - correlation comes from the secondary cmb anisotropies due to the quadratic doppler effects of the bulk motions , since the cross - correlation with the @xmath3 doppler effect such as the ov effect depends linearly on the bulk velocity field and thus suffers from cancellations between positive and negative contributions from the bulk velocity field for the ensemble average .
there are two main contributions to the cross - correlation function : one is contribution from the pure quadratic doppler effect of @xmath19 and the other is that from the density - modulated effect of @xmath131 , where @xmath9 is the electron density fluctuation field . for the cdm models , the latter contribution dominates the former on angular scales of @xmath138 if the ionized medium traces the dark matter distribution .
we found that the cross - correlation of @xmath254 from the doppler effect of @xmath19 is sensitive to the average ionization fraction @xmath160 between the reionization epoch and present in the sense that the bulk velocity and the weak lensing field depends only on the dark matter fluctuations , through the gravitational instability scenario , and on the cosmological parameters , which will be well constrained by the future precise data such as the primary cmb anisotropies and the weak lensing survey . on the other hand
, we showed that the cross - correlation of @xmath255 between the weak lensing field and the density - modulated quadratic doppler effect of @xmath131 depends linearly on the biasing relation between the dark matter and free electron distributions , and therefore its angular power spectrum could be both positive and negative , depending on the biasing and antibiasing , respectively .
if the low density region of igm is ionized prior to the high density region in the reionization history as suggested by mhr ( see also @xcite ) , we would measure the negative angular spectrum of the cross - correlation .
our results thus offer a new opportunity to understand the unknown problem how reionization proceeds in the inhomogeneous intergalactic medium connected to the dark matter clustering in principle , although the magnitude of the signal is very small .
let us briefly comment on other sources of cross - correlation between the weak lensing field and the observed cmb map .
one significant source is the thermal sz effect , and it has been indeed shown that the cross - correlation could have the magnitude of @xmath256 for the similar cdm model ( @xcite ) as investigated in this paper .
the sz effect thus produces much larger contribution than the cross - correlation with the doppler effect estimated as @xmath257 .
however , since the thermal sz signal arises mainly from hot plasma in clusters of galaxies at low redshift @xmath258 , it is expected that we can remove this signal from the cross - correlation maps , combined with the x - ray data or the sz maps .
the other possible sources are the cmb foregrounds such as synchrotron radiation from extragalactic dust that may correlate with the dark matter distribution in the large - scale structure .
isolating the signal shown in this paper observationally from the foregrounds will be a great challenge by using the different frequency - dependence ( @xcite ) , though this problem is beyond the scope of this paper .
m. t. acknowledges financial support from the japan society for promotion of science ( jsps ) research fellowships .
this research was supported by japanese grant - in - aid for science research fund of the ministry of education , science , sports and culture grant nos . 06107 and 11640235 , and sumitomo fundation .
aghanim , n. , dsert , f. x. , puget , j. l. , & gispert , r. 1996 , , 311 , 1 becker , r. h. , et al .
2001 , astro - ph/0108097 bardeen , j. m. , bond . , j. r. , kaiser , n. , & szalay , a. s. 1986 , , 304 , 15 ( bbks ) bartelmann , m. & schneider , p. 2001
, phys . rep .
2001 , 340 , 291 barkana , r. , & loeb , a. 2001 , phys . rep .
, in press , astro - ph/0010468 benson , a. j. , nusser , a. , sugiyama , n. , & lacey , c. g. 2001 , , 320 , 1 de bernardis , et al . , p. , et al .
2000 , , 404 , 955 bond , j. r. , & efstathiou , g. p. 1987
, , 226 , 655 bond , j. r. , efstathiou , g. p. , lubin , p. m. , & meinhold , p. r. 1991 , , 66 , 2179 cooray , a. 2000 , , 62 , 103506 dodelson , s. , & jubas , j. m. 1995 , , 439 , 503 efstathiou , g. p. 1988
, in large - scale motions , edited by v. rubin and s. j. coyne ( princeton univ .
press ) eke , v. , cole , s. , & frenk , c. s. 1996 , , 282 , 263 fan , x. , et al . 2000 , , 120 , 1167 gnedin , n. y. 2000 , , 535 , 530 gnedin , n. y. , & jaffe , a. h. 2001 , , 551 , 3 gruzinov , a. , & hu , w. 1998 , , 508 , 435 , gunn , j. e. , & peterson , b. a. 1965 , , 142 , 1633 jaffe , a. , & kamionkowski , m. 1998 , , 58 , 043001 hanany , s. , et al .
2000 , , 545 , l5 hu , w. 2000 , , 529 , 12 hu , w. 2000 , , 62 , 043007 hu , w. 2001a , , 557 , l79 hu , w. 2001b , astro - ph/0111606 hu , w. , & white , m. 1996 , , 315 , 33 hu , w. , & white , m. 1997 , , 56 , 596 hu , w. , scott , d. , & silk , j. 1994 , , 49 , 648 ( hss ) kaiser , n. 1984 , , 282 , 374 kaiser , n. 1992 , , 388 , 272 kitayama , t. , & suto , y. 1997 , , 490 , 557 knox , l. , scoccimarro , r. , & dodelson , s. 1998 , , 81 , 2004 komatsu , e. , & kitayama , t. 1999 , , 526 , l1 lanzetta , k. m. , wolfe , a. m. , & tunshek d. a. 1996 , , 440 , 435 miralde - escude , j. , haehnelt , m. , & rees , m. 2000 , , 530 , 1 ( mhr ) netterfield , c. b. , et al .
2001 , astro - ph/0104460 ostriker , j. p. , & vishniac , e. t. 1986 , , 306 , l51 peacock , j. a. , & dodds , s. j. 1996 , , 280 , l19 peebles , p. j. e. 1980 , the large scale structure of the universe ( princeton univ . press ) peiris , h. v. , & spergel , d. n. 2000 , , 540 , 605 schmalzing , j. , takada , m. , & futamase , t. 2000 , , 544 , l83 seljak , u. , & zaldarriaga , m. , 1999 , , 82 , 2636 seljak , u. , & zaldarriaga , m. 2000 , , 538 , 57 sugiyama , n. 1995 , 100 , 281 sugiyama , n. , silk , j. , & vittorio , n. 1993 , , 419 , l1 takada , m. 2001 , , 558 , 29 takada , m. , & futamase , t. 2001 , , 546 , 620 takada , m. , komtasu , e. , & futamase , t. 1999 , , 533 , l83 tegmark , m. , & zaldarriaga , m. 2000 , , 85 , 2240 tegmark , m. , eisenstein , d. j. , hu , w. , & de oliveira - costa , a. o. 2000 , , 530 , 133 vishniac , t. 1987 , , 322 , 597 van waerbek , l. , bernardeau , f. , & benabed , k. 2000 , , 540 , 14 van waerbeke , l. et al .
2000 , , 358 , 30 zaldarriaga , m. 2000 , , 62 , 063510 zaldarriaga , m. , & seljak , u. 1998 , , 58 , 023003 zaldarriaga , m. , & seljak , u. 1999 , , 59 , 123507 zeldovich , ya .
b. , & sunyaev , r. a. 1969 , astrophys . | we investigate a cross - correlation between the weak gravitational lensing field of the large - scale structure , @xmath0 , and the secondary temperature fluctuation field , @xmath1 , of cosmic microwave background ( cmb ) induced by thomson scattering of cmb photons off the ionized medium in the mildly nonlinear structure .
the cross - correlation is expected to observationally unveil the biasing relation between the dark matter and ionized medium distributions in the large - scale structure .
we develop a formalism for calculating the cross - correlation function and its angular power spectrum based on the small angle approximation . as a result
, we find that leading contribution to the cross - correlation comes from the secondary cmb fluctuation field induced by the quadratic doppler effects of the bulk velocity field @xmath2 of ionized medium , since the cross - correlations with the @xmath3 doppler effect and the ostriker - vishniac effect of @xmath4 are suppressed on relevant angular scales because the linear dependence on the bulk velocity field leads to cancellations between positive and negative contributions among the scatterers for the ensemble average .
the magnitude of the cross - correlation can be estimated as @xmath5 on small angular scales ( @xmath6 ) under the currently favored cold dark matter model of the structure formation and the simplest scenario of homogeneous reionization after a given redshift of @xmath7 .
although the magnitude turns out to be small , we find several interesting aspects of the effect .
one of them is that the density - modulated quadratic doppler effect of @xmath8 produces the cross - correlation with the weak lensing field that has _ linear _ dependence on the electron density fluctuation field @xmath9 or equivalently on the biasing relation between the dark matter and electron distributions . in other words ,
the angular power spectrum could be both positive or negative , depending on the positive biasing or antibiasing , respectively .
detection of the cross - correlation thus offers a new opportunity to observationally understand the reionization history of intergalactic medium connected to the dark matter clustering . |
this paper discusses simple graphs .
let @xmath2 be a graph with vertex - set @xmath3 and edge - set @xmath4 .
we say a vertex @xmath5 is a _ neighbor _ of another vertex @xmath6 if @xmath7 .
if @xmath8 , then @xmath9 denotes the degree of @xmath6 and furthermore , @xmath6 is called a _ @xmath1-vertex _ ( or _ @xmath10-vertex _ or _
@xmath11-vertex _ ) if @xmath12 ( or @xmath13 or @xmath14 ) .
similarly , a _ @xmath1-circuit _ ( or _ @xmath10-circuit _ or _ @xmath11-circuit _ ) is a circuit of length @xmath1 ( or at least @xmath1 or at most @xmath1 ) , and if @xmath2 is planar then a _ @xmath1-face _ ( or _ @xmath10-face _ or _ @xmath11-face _ ) is a face of size @xmath1 ( or at least @xmath1 or at most @xmath1 ) .
let @xmath15 $ ] denote a @xmath1-circuit with vertices @xmath16 in cyclic order .
if @xmath17 , then @xmath18 $ ] denotes the subgraph of @xmath2 induced by @xmath19 , and @xmath20 denotes the set of edges between @xmath19 and @xmath21 .
let @xmath2 be a graph and @xmath22 be a mapping .
the pair @xmath23 is called a _ signed graph _ , and @xmath24 is called a _ signature _ of @xmath2 .
an edge @xmath25 is _ positive _ ( or _ negative _ ) if @xmath26 ( or @xmath27 ) .
denote by @xmath28 the signed graph @xmath23 with @xmath26 for each @xmath29 .
a graph with no signature is usually called an _
unsigned graph_. a circuit of a signed graph is balanced ( unbalanced ) if it contains an even ( odd ) number of negative edges .
zaslavsky @xcite defines a ( signed ) coloring of a signed graph @xmath23 with @xmath1 colors or with @xmath30 signed colors to be a mapping @xmath31 such that for every edge @xmath32 of @xmath2 , @xmath33 if @xmath34 , and @xmath35 if @xmath36 .
recently , majov , raspaud and koviera @xcite introduced a _ @xmath1-coloring _ of @xmath23 as a proper coloring of @xmath23 using colors from @xmath37 if @xmath38 , and ones from @xmath39 if @xmath40 . a signed graph
@xmath23 is _
@xmath1-colorable _ if it admits a @xmath1-coloring .
the _ chromatic number _ of @xmath23 is the minimum number @xmath1 such that @xmath23 is @xmath1-colorable .
we follow the approach of @xcite to define list colorings of signed graphs .
given a signed graph @xmath23 , a _ list - assignment _ of @xmath23 is a function @xmath41 defined on @xmath3 such that @xmath42 for each @xmath8 .
an _ @xmath41-coloring _ of @xmath23 is a proper coloring @xmath43 of @xmath23 such that @xmath44 for each @xmath8 .
a list - assignment @xmath41 is called a _
@xmath1-list - assignment _ if @xmath45 for each @xmath8 .
we say @xmath23 is _
@xmath1-choosable _ if it admits an @xmath41-coloring for every @xmath1-list - assignment @xmath41 .
the _ choice number _ of @xmath23 is the minimum number @xmath1 such that @xmath23 is @xmath1-choosable . clearly , if a signed graph is @xmath1-choosable , then it is also @xmath1-colorable .
let @xmath23 be a signed graph , @xmath41 be a list assignment of @xmath23 , and @xmath43 be an @xmath41-coloring of @xmath23 .
let @xmath46 .
we say @xmath47 and @xmath48 are obtained from @xmath49 and @xmath43 by a _ switch _ at @xmath19
if @xmath50 @xmath51 two signed graphs @xmath23 and @xmath52 are _ equivalent _ if they can be obtained from each other by a switch at some subset of @xmath3 .
let @xmath53 is equivalent to @xmath54 .
[ switch ] let @xmath23 be a signed graph , @xmath41 be a list - assignment of @xmath2 and @xmath43 be an @xmath41-coloring of @xmath23 . if @xmath47 and @xmath48 are obtained from @xmath49 and @xmath43 by a switch at a subset of @xmath3 , then @xmath48 is an @xmath55-coloring of @xmath56 . furthermore , two equivalent signed graphs have the same chromatic number and the same choice number .
let @xmath2 be a graph . by definition , @xmath2 and
@xmath28 have the same chromatic number and the same choice number .
hence , the following statement holds .
[ swith_corollary ] if @xmath57 , then @xmath2 and @xmath23 have the same chromatic number and the same choice number .
this paper focusses on the choosability of signed planar graphs and generalizes the results of @xcite to signed graphs .
section [ 5c ] proves that every signed planar graph is 5-choosable .
furthermore , there is a signed planar graph @xmath23 which is not 4-choosable , but @xmath28 is 4-choosable .
section [ 4c ] proves for every @xmath0 that every signed planar graph without @xmath1-circuits is 4-choosable .
section [ 3c ] proves that every signed planar graph with neither 3-circuits nor 4-circuits is 3-choosable .
furthermore , there exists a signed planar graph @xmath23 such that @xmath2 has girth 4 and @xmath23 is not 3-choosable but @xmath28 is 3-choosable .
[ 5_choosable ] every signed planar graph is 5-choosable .
we use the method described in @xcite to prove following theorem which implies theorem [ 5_choosable ] . a plane graph
@xmath2 is a _ near triangulation _ if the boundary of each bounded face of @xmath2 is a triangle .
[ 5_choosable_extend ] let @xmath23 be a signed graph , where @xmath2 is a near - triangulation .
let @xmath58 be the boundary of the unbounded face of @xmath2 and @xmath59 $ ] . if @xmath41 is a list - assignment of @xmath23 such that @xmath60 , @xmath61 and @xmath62 , and that @xmath63 for @xmath64 and @xmath65 for @xmath66 , then @xmath23 has an @xmath41-coloring .
let us prove theorem [ 5_choosable_extend ] by induction on @xmath67 . if @xmath68 , then @xmath69 and @xmath70 .
choose a color from @xmath71 for @xmath72 .
so we proceed to the induction step .
if @xmath58 has a chord which divides @xmath2 into two graphs @xmath73 and @xmath74 , then we choose the notation such that @xmath73 contains @xmath75 , and we apply the induction hypothesis first to @xmath73 and then to @xmath74 .
hence , we can assume that @xmath58 has no chord .
let @xmath76 be the neighbors of @xmath77 in cyclic order around @xmath77 .
since the boundary of each bounded face of @xmath2 is a triangle , @xmath2 contains the path @xmath78 . since @xmath58 has no chord , @xmath79 is a circuit @xmath80 .
let @xmath81 and @xmath82 be two distinct colors of @xmath83 .
define @xmath84 for @xmath85 , and @xmath86 for @xmath87 .
let @xmath88 be the restriction of @xmath24 to @xmath89
. by the induction hypothesis , signed graph @xmath90 has an @xmath55-coloring .
let @xmath43 be the color vertex @xmath91 receives .
we choose a color from @xmath92 for @xmath77 , giving an @xmath41-coloring of @xmath23 .
voigt @xcite constructed two planar graphs which are not 4-choosable . by corollary [ swith_corollary ]
these two examples generate two group of signed planar graphs which are not 4-choosable .
we extend this result to signed graphs .
there exists a signed planar graph @xmath23 such that @xmath23 is not 4-choosable but @xmath2 is 4-choosable .
we construct @xmath23 as follows .
take a copy @xmath73 of complete graph @xmath93 and embed it into euclidean plane .
insert a claw into each 3-face of @xmath73 and denote the resulting graph by @xmath74 . once again , insert a claw into each 3-face of @xmath74 and denote by @xmath94 the resulting graph .
a vertex @xmath6 of @xmath94 is called an initial - vertex if @xmath95 , a solid - vertex if @xmath96 and a hollow - vertex if @xmath97 ( figure [ fig1 ] illustrates graph @xmath94 ) .
a 3-face of @xmath94 is called a special 3-face if it contains an initial - vertex , a solid - vertex and a hollow - vertex .
clearly , @xmath94 has 24 special 3-faces , say @xmath98 .
let @xmath99 be the plane graph as shown in figure [ fig2 ] , which consists of a circuit @xmath100 $ ] and its interior . for @xmath101
, replace @xmath102 by a copy @xmath103 of @xmath99 such that @xmath104 and @xmath105 are identified with the solid - vertex , hollow - vertex and initial - vertex of @xmath102 , respectively .
let @xmath2 be the resulting graph .
clearly , @xmath2 is planar . ,
title="fig:",width=188 ] + , title="fig:",width=226 ] + define a signature @xmath24 of @xmath2 as follows : @xmath106 for @xmath107 and @xmath26 for @xmath108 .
let @xmath41 be a 4-list - assignment of signed graph @xmath23 defined as follows : @xmath109 for @xmath110 , and @xmath111 , @xmath112 , @xmath113 , @xmath114 , @xmath115 , @xmath116 , @xmath117 and @xmath118 for @xmath119 .
we claim that signed graph @xmath23 has no @xmath41-coloring .
suppose to the contrary that @xmath120 is an @xmath41-coloring of @xmath23 . by the construction of @xmath94 , precisely one of the special 3-faces of @xmath94
is assigned in @xmath120 color 1 to its solid - vertex , color 2 to its hollow - vertex and color 3 to its initial - vertex . without loss of generality ,
let @xmath121 be such a special 3-face .
let us consider @xmath120 in @xmath122 . clearly , @xmath123 and @xmath124 .
it follows that @xmath125 .
notice that the odd circuit @xmath126 $ ] is balanced and the even circuit @xmath127 $ ] is unbalanced , and thus both of them are not 2-choosable .
it follows that if @xmath128 , then @xmath120 is not proper in @xmath126 $ ] , and that if @xmath129 , then @xmath120 is not proper in @xmath127 $ ] .
therefore , @xmath23 has no @xmath41-coloring and thus is not 4-choosable .
let @xmath55 be any 4-list - assignment of @xmath2 . by the construction , it is not hard to see that @xmath94 is 4-choosable .
let @xmath43 be an @xmath55-coloring of @xmath94 . clearly , for @xmath119 , each of vertices @xmath104 and @xmath105 receives a color in @xmath43 .
let @xmath130 and @xmath131 be two distinct colors from @xmath132 .
choose a color from @xmath133 for @xmath134 , and then vertices @xmath135 and @xmath136 can be list - colored by @xmath55 in turn . since circuit @xmath137 $ ]
is 2-choosable , it follows that vertices @xmath138 and @xmath139 can also be list - colored by @xmath55 .
therefore , @xmath43 can be extended to an @xmath55-coloring of @xmath2 .
this completes the proof that @xmath2 is 4-choosable .
a graph @xmath2 is _ @xmath140-degenerate _ if every subgraph @xmath99 of @xmath2 has a vertex of degree at most @xmath140 in @xmath99 .
it is known that every @xmath141-degenerate graph is @xmath140-choosable .
this proposition can be extended for signed graphs .
[ degenerate_choosability ] let @xmath23 be a signed graph . if @xmath2 is @xmath141-degenerate , then @xmath23 is @xmath140-choosable .
( induction on @xmath67 ) let @xmath41 be any @xmath140-list - assignment of @xmath2 .
the proof is trivial if @xmath142 . for @xmath143 , since @xmath2 is @xmath141-degenerate , @xmath2 has a vertex @xmath6 of degree at most @xmath144 and moreover , graph @xmath145 is @xmath141-degenerate .
let @xmath88 and @xmath55 be the restriction of @xmath24 and @xmath41 to @xmath145 , respectively . by applying the induction hypothesis to @xmath146 , we conclude that @xmath146 is @xmath140-choosable and thus has an @xmath55-coloring @xmath120 . since @xmath6 has degree at most @xmath144 , we can choose a color @xmath130 for @xmath6 such that @xmath147 .
we complete an @xmath41-coloring of @xmath23 with @xmath120 and @xmath130 .
it is an easy consequence of euler s formula that every triangle - free planar graph contains a vertex of degree at most 3 . therefore , the following statement is true : [ 3-degenerate3 ] planar graphs without 3-circuits are 3-degenerate .
moreover , we will use two more lemmas .
[ @xcite ] [ 3-degenerate4 ] planar graphs without 5-circuits are 3-degenerate .
[ @xcite ] [ 3-degenerate5 ] planar graph without 6-circuits are 3-degenerate .
[ 4_choosable ] let @xmath23 be a signed planar graph .
for all @xmath148 , if @xmath2 has no @xmath1-circuit , then @xmath23 is 4-choosable . for @xmath149
we deduce the statement from theorem [ degenerate_choosability ] , together with lemmas [ 3-degenerate3 ] , [ 3-degenerate4 ] and [ 3-degenerate5 ] , respecvitely .
it remains to prove theorem [ 4_choosable ] for the case @xmath150 .
suppose to the contrary that the statement is not true .
let @xmath23 be a counterexample of smallest order , and @xmath41 be a 4-list - assignment of @xmath23 such that @xmath23 has no @xmath41-coloring .
clearly , @xmath2 is connected by the minimality of @xmath23 .
[ minimal degree ] @xmath151 .
let @xmath5 be a vertex of @xmath2 of minimal degree .
suppose to the contrary that @xmath152 .
let @xmath88 and @xmath55 be the restriction of @xmath24 and @xmath41 to @xmath153 , respectively . by the minimality of @xmath23 , the signed graph @xmath154 has an @xmath55-coloring @xmath43 . since every neighbor of @xmath5 forbids one color for @xmath5 no matter what the signature of the edge between them is , @xmath155 still has a color left for coloring @xmath5 .
therefore , @xmath43 can be extended to an @xmath41-coloring of @xmath23 , a contradiction .
[ 5face ] @xmath2 has no 6-circuit @xmath58 such that @xmath156 $ ] and @xmath157 , and @xmath158 and all other vertices of @xmath58 are of degree 4 .
suppose to the contrary that @xmath2 has such 6-circuit @xmath58 .
since @xmath2 has no 4-circuit , @xmath159 is the only chord of @xmath58 .
there always exists a subset @xmath19 of @xmath160 such that all of the edges @xmath161 and @xmath162 are positive after a switch at @xmath19 .
let @xmath88 and @xmath55 be obtained from @xmath24 and @xmath41 by a switch at @xmath19 , respectively .
proposition [ switch ] implies that signed graph @xmath56 has no @xmath55-coloring .
hence , @xmath56 is also a minimal counterexample .
let @xmath163 and @xmath164 be the restriction of @xmath88 and @xmath55 to @xmath165 , respectively .
it follows that @xmath166 has an @xmath164-coloring @xmath120 .
we obtain a contradiction by further extending @xmath120 to an @xmath55-coloring of @xmath56 as follows . by the condition on the vertex degrees of @xmath58 ,
there exists a list - assignment @xmath167 of @xmath168 $ ] such that @xmath169 for @xmath170 , and @xmath171 and @xmath172 for @xmath173
. let @xmath174 .
suppose that @xmath175 has a color , say @xmath130 , not appear in at least two of lists @xmath176 and @xmath177 .
we color @xmath178 with @xmath130 , and then all other vertices of @xmath58 can be list - colored by @xmath167 in some order .
for example , if @xmath130 does not appear in @xmath179 and @xmath180 , then we color @xmath160 in the order @xmath181 .
hence , we may assume that @xmath182 , @xmath183 and @xmath184 .
if @xmath185 , then color @xmath186 with @xmath187 , @xmath188 with @xmath131 , and @xmath178 with @xmath130 , and the remaining vertices of @xmath58 can be list - colored by @xmath167 in the order @xmath189 .
hence , we may assume @xmath190 .
it follows that @xmath191 and @xmath192 . if @xmath193 , then color both @xmath186 and @xmath188 with @xmath130 , and the remaining vertices of @xmath58 can be list - colored by @xmath167 in the order @xmath194 .
hence , we may assume @xmath195 .
now color 0 is included in list @xmath179 but no in list @xmath177 .
thus there exists an integer @xmath196 in set @xmath197 such that @xmath198 and @xmath199 ( index is added modular 6 ) .
we color @xmath200 with color 0 , and then the remaining vertices of @xmath58 can be list - colored by @xmath167 in cyclic order on @xmath58 ending at @xmath201 .
[ 6-vertex ] @xmath2 has no 10-circuit @xmath58 such that @xmath202 $ ] and @xmath203 , and vertex @xmath178 has degree 6 and all other vertices of @xmath58 have degree 4 .
suppose to the contrary that @xmath2 has such a 10-circuit @xmath58 .
let @xmath88 and @xmath55 be the restriction of @xmath24 and @xmath41 to graph @xmath165 , respectively . by the minimality of @xmath23 , signed graph @xmath204 has an @xmath55-coloring @xmath120 .
a contradiction is obtained by further extending @xmath120 to an @xmath41-coloring of @xmath23 as follows .
we shall list - color the vertices of @xmath58 by @xmath41 in the cyclic order @xmath205 . for @xmath206 ,
let @xmath207 . clearly , @xmath208 is the set of forbidden colors by the neighbors of @xmath201 not on @xmath58 to be assigned to vertex @xmath201 . since @xmath209 and moreover ,
if there is any other chord of @xmath58 then the list @xmath208 will not become longer , it follows that @xmath210 and @xmath211 .
hence , we can let @xmath130 and @xmath131 be two distinct colors from @xmath212 , and let @xmath213 .
color vertex @xmath186 with @xmath187 . for @xmath214 ,
vertex @xmath201 has at most 3 neighbors colored before @xmath201 in this color - assigning process and thus , @xmath215 still has a color available for @xmath201 .
denote by @xmath216 the color vertex @xmath217 receives .
we complete the extending of @xmath120 by assigning a color from @xmath218 to @xmath219 .
consider an embedding of @xmath2 into the euclidean plane .
let @xmath2 denote the resulting plane graph .
we say two faces are _ adjacent _ if they share an edge .
two adjacent faces are _ normally adjacent _ if they share an edge @xmath220 and no vertex other than @xmath221 and @xmath222 . since @xmath2 is a simple graph , the boundary of every 3-face or 5-face is a circuit .
since @xmath2 has no 4-circuits , we can deduce that if a 3-face and a 5-face are adjacent , then they are normally adjacent .
a vertex is _ bad _ if it is of degree 4 and incident with two nonadjacent 3-faces .
bad 3-face _ is a 3-face containing three bad vertices .
a 5-face @xmath223 is _ magic _ if it is adjacent to five 3-faces , and if all the vertices of these six faces have degree 4 except one vertex of @xmath223 .
we shall obtain a contradiction by applying discharging method .
let @xmath224 , @xmath225 , and @xmath226 be the set of faces of @xmath2 .
denote by @xmath227 the size of a face @xmath223 of @xmath2 . give initial charge @xmath228 to each element @xmath221 of @xmath229 , where @xmath230 for @xmath231 , and @xmath232 for @xmath233 .
discharge the elements of @xmath229 according to the following rules : 1 .
every vertex @xmath5 sends each incident 3-face charge 1 if @xmath5 is a bad vertex , and charge 2 otherwise .
2 . every 5-vertex sends @xmath234 to each incident 5-face .
3 . every 6-vertex sends each incident 5-face @xmath223 charge 1 if @xmath223 is magic , charge @xmath235 if @xmath223 is not magic but contains four 4-vertices , charge @xmath234 if @xmath223 contains at most three 4-vertices .
every @xmath236-vertex sends 1 to each incident 5-face .
every 3-face sends @xmath234 to each adjacent 5-face if this 3-face contains at most one bad vertex .
every @xmath237-face sends @xmath238 to each adjacent bad 3-face , where @xmath1 is the number of common edges between them .
let @xmath239 denote the final charge of each element @xmath221 of @xmath229 when the discharging process is over .
on one hand , by euler s formula we deduce @xmath240 since the sum of charge over all elements of @xmath229 is unchanged , we have @xmath241 on the other hand , we show that @xmath242 for @xmath243 .
hence , this obvious contradiction completes the proof of theorem [ 4_choosable ] .
it remains to show that @xmath242 for @xmath243 .
[ claim_vertex ] if @xmath231 , then @xmath244 .
let @xmath245 be the number of 3-faces that contains @xmath6 . since @xmath2 has no 4-circuit , @xmath246 .
moreover , @xmath247 by claim [ minimal degree ] .
suppose @xmath248 .
we have @xmath249 .
if @xmath250 , then @xmath6 is a bad vertex and thus we have @xmath251 by r1 ; otherwise , we have @xmath252 by r1 again . if @xmath253 , then @xmath249 and thus by r1 and r2 , we have @xmath254 .
suppose that @xmath255 .
thus @xmath256 . by r1 and r3 ,
if @xmath249 then we have @xmath257 , and if @xmath6 is incident with no magic 5-face then we have @xmath258 .
hence , we may assume that @xmath69 and that @xmath6 is incident with a magic 5-face @xmath223 .
for any other @xmath237-face @xmath259 containing @xmath6 than @xmath223 , claim [ 6-vertex ] implies that if @xmath259 has size 5 then it contains at most three 4-vertices , and thus @xmath6 sends at most @xmath234 to @xmath259 by r3 .
hence , we have @xmath260 .
it remains to suppose @xmath261 . by r1 and r4
, we have @xmath262 . if @xmath233 , then @xmath263 .
suppose @xmath264 recall that in this case the boundary of @xmath223 is a circuit .
we have @xmath265 by r1 and r5 when @xmath223 has at most one bad vertex , and @xmath266 by r1 when @xmath223 has precisely two bad vertices .
it remains to assume that @xmath223 has precisely three bad vertices , that is , @xmath223 is a bad 3-face . in this case
, @xmath223 receives charge 1 in total from adjacent faces by r6 , and charge 3 in total from incident vertices by r1 .
hence , we have @xmath267 .
suppose @xmath268 .
recall in this case that the boundary of @xmath223 is a circuit and that if @xmath223 is adjacent to a 3-face then they are normally adjacent .
let @xmath269 be the number of bad 3-faces adjacent to @xmath223 .
clearly , @xmath223 sends charge only to adjacent bad 3-faces by r6 , and possibly receives charge from incident @xmath237-vertices and adjacent 3-faces by rules from r2 to r5 .
hence , we have @xmath270 when @xmath271 . claim [
5face ] implies that @xmath272 and that @xmath223 contains a @xmath237-vertex @xmath5 , which sends at least @xmath234 to @xmath223 .
hence , we have @xmath273 when @xmath274 .
first suppose @xmath275 .
if @xmath223 has a @xmath237-vertices different from @xmath5 , then we are done by @xmath276 .
hence , we may assume that @xmath223 contains four 4-vertices .
it follows that if @xmath277 , then @xmath223 receives at least @xmath235 from @xmath6 by r3 or r4 and thus we are done .
hence , we may assume that @xmath278 .
through the drawing of 3-faces adjacent to @xmath223 , we can assume @xmath5 is incident with a 3-face @xmath279 $ ] that is adjacent to @xmath223 on edge @xmath32 .
5face ] implies that @xmath280 .
hence , @xmath223 receives @xmath234 from face @xmath279 $ ] by r5 , and thus we are done .
let us next suppose @xmath281 .
we may assume @xmath282 $ ] such that @xmath283 , @xmath284 and @xmath285 are the three common edges between @xmath223 and bad 3-faces .
since both vertices @xmath286 and @xmath287 are bad , edges @xmath288 and @xmath289 are contained in 3-faces @xmath290 $ ] and @xmath291 $ ] , respectively .
if @xmath292 then claim [ 5face ] implies that @xmath293 , and thus @xmath223 receives @xmath234 from each of faces @xmath290 $ ] and @xmath291 $ ] by r5 , we are done .
if @xmath294 , then @xmath223 receives 1 from @xmath5 and thus we are done .
hence , we may assume that @xmath295 .
if both @xmath296 and @xmath297 has degree 4 , that is , @xmath223 is a magic 5-face , then @xmath223 receives 1 from @xmath5 by r3 ; otherwise , @xmath223 receives @xmath235 from @xmath5 and @xmath234 from at least one of faces @xmath290 $ ] and @xmath291 $ ] by r3 again .
we are done in both cases .
it remains to suppose @xmath298 .
remind that @xmath223 has no charge moving in or out except that it sends @xmath299 in total to adjacent bad 3-faces by r6 .
hence , we have @xmath300 .
the proof of theorem [ 4_choosable ] is completed .
in 1995 , thomassen @xcite proved that every planar graph of girth at least 5 is 3-choosable .
and then in 2003 , he @xcite gave a shorter proof of this result .
we find out that the argument used in @xcite also works for signed graphs .
hence the following statement is true .
[ 3_choosable ] every signed planar graph with neither 3-circuit nor 4-circuit is 3-choosable . for the sake of completeness , the proof is given in the appendix .
there exists a signed planar graph @xmath23 such that @xmath2 has girth 4 and @xmath23 is not 3-choosable but @xmath2 is 3-choosable .
let @xmath301 be a plane graph consisting of two circuits @xmath302 $ ] and @xmath303 $ ] of length 4 and four other edges @xmath304 and @xmath305 , as shown in figure [ fig3 ] .
take nine copies @xmath306 of @xmath301 , and identify @xmath307 into a vertex @xmath308 and @xmath309 into a vertex @xmath80 .
let @xmath2 be the resulting graph .
clearly , @xmath2 is planar and has girth 4 . ,
title="fig:",width=151 ] + define a signature @xmath24 of @xmath2 as : @xmath27 for @xmath310 , and @xmath26 for @xmath311 . for @xmath312 ,
let @xmath313 and @xmath314 define a 3-list - assignment @xmath41 of @xmath2 as follows : @xmath315 ; for @xmath316 , let @xmath317 and @xmath318 .
we claim that signed graph @xmath23 has no @xmath41-coloring .
suppose to the contrary that @xmath43 is an @xmath41-coloring of @xmath23 .
let @xmath319 and @xmath320 .
consider subgraph @xmath321 .
it follows that @xmath322 .
furthermore , the circuit @xmath323 $ ] is unbalanced and thus not 2-choosable .
hence , @xmath321 is not properly colored in @xmath43 , a contradiction .
this proves that @xmath23 has no @xmath41-coloring and therefore , @xmath23 is not 3-choosable .
we claim that graph @xmath2 is 3-choosable . for any 3-list - assignment of @xmath2 , choose any color for vertices @xmath308 and @xmath80 from their color lists , respectively .
consider each subgraph @xmath324 .
both vertices @xmath325 and @xmath136 can be list colored .
the 2-choosability of circuit @xmath326 $ ] yields a list coloring of @xmath102 and hence a list coloring of @xmath2 .
this proves that @xmath2 is 3-choosable .
99 , choosability in graphs , congr . numer .
* 26 * ( 1980 ) 125 - 157 , planar graphs without cycles of specific lengths , european j. combin .
* 23 * ( 2002 ) 377 - 388 , the 4-choosability of plane graphs without 4-cycles , j. combin .
theory , ser .
b * 76 * ( 1999 ) 117 - 126 , the chromatic number of a signed graph ( 2014 ) arxiv : 1412.6349 , every planar graph is 5-choosable , j. combin . theory , ser .
b * 62 * ( 1994 ) 180 - 181 , 3-list - coloring planar graphs of girth 5 , j. combin . theory , ser .
b * 64 * ( 1995 ) 101 - 107 , a short list color proof of grtzsch s theorem , j. combin .
theory , ser .
b * 88 * ( 2003 ) 189 - 192 , coloring the vertices of a graph in prescribed colors , diskret .
analiz * 29 * ( 1976 ) 3 - 10 ( in russian ) , list colourings of planar graphs , discrete math .
* 120 * ( 1993 ) 215 - 219 , on 3-colorable non-4-choosable planar graphs , j. graph theory * 24 * ( 1997 ) 233 - 235 , choosability and edge choosability of planar graphs without five cycles , appl .
lett . * 15 * ( 2002 ) 561 - 565 , signed graph coloring , discrete math .
* 39 * ( 1982 ) 215 - 228 , how colorful the signed graph ? , discrete math .
* 52 * ( 1984 ) 279 - 284 , signed graphs and geometry ( 2013 ) arxiv : 1303.2770
[ thomassen ] let @xmath23 be a signed plane graph of girth at least 5 , and @xmath327 be the outer face boundary of @xmath2 .
let @xmath328 be a path or circuit of @xmath2 such that @xmath329 and @xmath330 , and @xmath331 be the restriction of @xmath24 to @xmath328 .
assume that @xmath332 has a 3-coloring @xmath43 .
let @xmath41 be a list - assignment of @xmath2 such that @xmath333 if @xmath334 , @xmath335 if @xmath336 , and @xmath63 if @xmath337 .
assume furthermore that there is no edge joining vertices whose lists have at most two colors except for the edges in @xmath328 .
then @xmath43 can be extended to an @xmath41-coloring of @xmath23 .
we may assume that @xmath2 is connected , since otherwise we apply the induction hypothesis to every connected component of @xmath2 .
similarly , @xmath2 has no cutvertex in @xmath328 .
moreover , @xmath2 has no cutvertex at all .
suppose to the contrary that @xmath5 is a cutvertex contained in an endblock @xmath338 disjoint from @xmath328 .
we first apply the induction hypothesis to @xmath339 .
if @xmath338 has vertices with only two available colors joined to @xmath5 , then we color each such vertex .
these colored vertices of @xmath338 together with the edges joining them to @xmath5 divide @xmath338 into parts each of which has at most three colored vertices inducing a path .
now we apply the induction hypothesis to each of those parts .
this contradiction proves claim [ 2-connected ] .
if some edge @xmath25 of @xmath328 is a chord of @xmath2 , then @xmath25 divides @xmath2 into two parts , and we apply the induction hypothesis to each of those two parts .
this contradiction proves claim [ p_nonchord ] .
let @xmath19 be a set of colored vertices of @xmath2 . to save writing we just say `` delete the product colors of @xmath19 from @xmath2 ''
instead of `` for @xmath343 , delete all of the colors in @xmath344 from the list of @xmath6 '' . if @xmath346 , then we delete any vertex from @xmath327 , and delete the product color of that vertex from @xmath2 . if @xmath347 and @xmath348 , then we color the vertices of @xmath327 not in @xmath328 , we delete them together with their product colors from @xmath2 .
now we apply the induction hypothesis to the resulting graph @xmath349 , if possible . as @xmath2 has grith at least 5 , the vertices with precisely two available colors are independent . for the same reason
, such a vertex can not be joined to two vertices of @xmath328 .
such a vertex may be joined to precisely one vertex of @xmath328 .
we then color it . now
the colored vertices of @xmath349 divide @xmath349 into parts each of which has at most 6 precolored vertices inducing a path .
we then apply induction hypothesis to each of those parts .
this contradiction proves claim [ path ] .
suppose to the contrary that @xmath220 is a chord of @xmath327 .
then @xmath220 divides @xmath2 into two graphs @xmath350 , say .
we may choose the notation such that @xmath74 has no more vertices of @xmath328 than @xmath73 has , and subject to that condition , @xmath351 is minimum .
we apply the induction hypothesis first to @xmath73 .
in particular , @xmath221 and @xmath222 receive a color .
the minimality of @xmath74 implies that the outer cycle of @xmath74 is chordless .
so @xmath74 has at most two vertices which have only two available colors and which are joined to one of @xmath221 and @xmath222 .
we color any such vertex , and then we apply the induction hypothesis to @xmath74 .
this contradiction proves claim [ chord ] .
[ 2-path ] @xmath2 has no path of the form @xmath352 where @xmath5 lies inside @xmath327 , except possibly when @xmath353 and the path is of the form @xmath354 or @xmath355 .
in particular , @xmath5 has only two neighbors on @xmath327 .
we define @xmath73 and @xmath74 as in the proof of claim [ chord ] .
we apply the induction hypothesis first to @xmath73 .
although @xmath5 may be joined to several vertices with only two available colors , the minimality of @xmath74 implies that no such vertex is in @xmath356 .
there may be one or two vertices in @xmath356 that have only two available colors and which are joined to one of @xmath357 and @xmath358 .
we color any such vertex , and then at most six vertices of @xmath74 are colored .
if possible , we apply the induction hypothesis to @xmath74 . this is possible unless the coloring of @xmath73 is not valid in @xmath74 .
this happens only if @xmath328 has a vertex in @xmath74 joined to one of @xmath357 and @xmath358 .
this happens only if we have one of the two exceptional cases described in claim [ 2-path ] .
[ 3-path ] @xmath2 has no path of the form @xmath359 such that @xmath5 and @xmath360 lie inside @xmath327 , and @xmath361 .
also , @xmath2 has no path @xmath359 such that @xmath5 and @xmath360 lie inside @xmath327 , @xmath362 , and @xmath363 .
if @xmath364 , then we complete the proof by deleting @xmath365 and its product color from @xmath2 , and apply the induction hypothesis to @xmath366 and obtain thereby a contradiction .
so we assume @xmath367 . by claim [ path ] , @xmath368 and thus @xmath369 .
if @xmath370 , then we first color @xmath371 and @xmath372 , then we delete them and their product colors from @xmath2 .
we obtain a contradiction by applying the induction hypothesis to the resulting graph . by claims
[ chord ] and [ 2-path ] this is possible unless @xmath353 and @xmath2 has a vertex @xmath5 inside @xmath327 joined to both @xmath373 and @xmath374 .
in this case we color @xmath5 and delete both @xmath375 and @xmath376 before we apply the induction hypothesis .
hence , we may assume that @xmath377 .
we give @xmath378 a color not in @xmath379 and then color @xmath371 and @xmath372 , and finally we delete @xmath357 and the product color of @xmath357 from @xmath2 for @xmath380 .
we obtain a contradiction by applying the induction hypothesis to the resulting graph . if @xmath353 and @xmath2 has a vertex @xmath5 inside @xmath327 joined to @xmath373 and @xmath374 , then , as above , we color @xmath5 and delete @xmath375 and @xmath376 before we use induction .
if @xmath381 , and @xmath2 has a vertex @xmath382 inside @xmath327 joined to @xmath72 and @xmath383 , then we also color @xmath382 and delete @xmath384 and @xmath385 before we use induction .
finally , there may be a path @xmath386 where @xmath360 and @xmath387 lies inside @xmath327 . by claim
[ empty cycle ] , this path is unique .
we color @xmath360 and @xmath387 and delete them together with their product colors from @xmath2 before we use induction .
note that @xmath5 and @xmath382 may also exist in this case . if there are vertices joined to two colored vertices , then we also color these vertices before we use induction .
the colored vertices divide @xmath2 into parts , and we shall show that each part satisfies the induction hypothesis . by second statement of claim [ 3-path ]
, there are at most six precolored vertices in each part , and they induce a path . claim [ 2-path ] and the first statement of claim [ 3-path ] imply that there is no vertex with precisely two available colors on @xmath327 which is joined to a vertex inside @xmath327 whose list has only two available colors after the additional coloring . since @xmath2 has girth at least 5 and by claim [ empty cycle ] , there is no other possibility for two adjacent vertices @xmath387 and @xmath297 to have only two available colors in their lists , as both @xmath387 and @xmath297 must be adjacent to a vertex that has been colored and deleted . | this paper studies the choosability of signed planar graphs .
we prove that every signed planar graph is 5-choosable and that there is a signed planar graph which is not 4-choosable while the unsigned graph is 4-choosable . for each @xmath0 , every signed planar graph without circuits of length @xmath1 is 4-choosable .
furthermore , every signed planar graph without circuits of length 3 and of length 4 is 3-choosable .
we construct a signed planar graph with girth 4 which is not 3-choosable but the unsigned graph is 3-choosable . |
the optical theorem should be applied for the unitary @xmath1-matrix since it follows from unitarity condition .
it is frequently used for the non - hermitian hamiltonians as well . in this case
the @xmath1-matrix should be unitarized . however , this requirement breaks down for a number of well - known models since the unitarization is the non - trivial aspect of the problem . in this paper
the possible consequences are studied by the example of @xmath0 transitions [ 1 - 3 ] in a medium followed by annihilation @xmath2 where @xmath3 are the annihilation mesons .
the matter is that in the standard calculations of this process ( see [ 4 - 9 ] , for example ) the optical theorem is applied for the essentially non - unitary @xmath1-matrix . as an alternative to the standard calculation mentioned above we consider the simple model ( see fig .
1 ) with unitary @xmath1-matrix ( later on reffered to as model with the hermitian hamiltonian ) . for both unitary and non - unitary models
the solution in the analytical form is available which permits the generalization of the results .
it is shown that an application of optical theorem for the non - unitary @xmath1-matrix leads to the qualitative error in the result .
the alternative approach gives the result which depends critically on the details of the model . due to
this the lower limit on the free - space @xmath0 oscillation time @xmath4 lies in the broad range @xmath5 .
the explanation of the huge distinction between the values of @xmath4 ( @xmath6 and @xmath7 ) is the second purpose of this paper .
we recall that unitarity condition @xmath8 @xmath9 , gives @xmath10 from this equation the optical theorem and expression for the decay width @xmath11 are obtained . here
@xmath12 is the normalization time , @xmath13 .
the non - unitarity of @xmath1-matrix implies that @xmath14 or , what is the same @xmath15 @xmath16 , resulting in @xmath17 since the value @xmath18 can be very small .
instead of ( 2 ) we have ( 5 ) and eq .
( 3 ) and optical theorem are inapplicable .
( 5 ) means the probability non - conservation : @xmath19 .
let us consider the process ( 1 ) .
the background potential of neutron - medium interaction @xmath20 is included in the neutron wave function : @xmath21 , @xmath22 . in the standard calculation
[ 4 - 9 ] the @xmath23-medium interaction is described by optical potential ( potential model ) .
the interaction hamiltonian is @xmath24 @xmath25 @xmath26 here @xmath27 and @xmath28 are the hamiltonians of @xmath0 conversion [ 4,5 ] and @xmath23-medium interaction , respectively ; @xmath29 is a small parameter with @xmath30 , @xmath31 is the antineutron optical potential , @xmath32 is the annihilation width of @xmath23 . in eq .
( 9 ) we have put @xmath33 , @xmath34
. the model can be realized by means of equations of motion [ 4 - 9 ] , or diagram technique .
the full in - medium antineutron propagator @xmath35 is @xmath36 @xmath37 , @xmath38 .
the on - diagonal matrix element @xmath39 is shown in fig .
1c . for the total decay width @xmath40 eq . ( 4 ) gives the well - known result [ 4 - 9 ] : @xmath41 the lower limit on the free - space @xmath0 oscillation time @xmath42 derived by means of eq .
( 11 ) is @xmath43 [ 7 ] .
however , the basic eq .
( 4 ) is inapplicable since @xmath1-matrix is non - unitary ( see ( 9 ) ) .
the model should be revised .
in the one - particle model described above the total decay width can be obtained by means eq .
( 4 ) only .
we calculate directly the off - diagonal matrix element in the framework of field - theoretical approach .
the process model is shown in fig .
the interaction hamiltonian is @xmath44 @xmath45 here @xmath46 is the hamiltonian of @xmath23-medium interaction , @xmath47 is the effective annihilation hamiltonian in the second quantization representation , @xmath48 is the residual scalar field , @xmath27 is given by ( 8) .
we consider the model shown in fig .
1a ( model * a * ) .
the amplitude of antineutron annihilation in the medium @xmath49 is given by @xmath50 here @xmath51 is the state of the medium containing the @xmath23 with the 4-momentum @xmath52 ; @xmath53 denotes the annihilation mesons , @xmath54 includes the normalization factors of the wave functions .
@xmath49 includes the all orders in @xmath47 .
transition in the medium followed by annihilation .
* b * same as * a * but the annihilation amplitude is given by ( 21 ) . the blocks @xmath49 and @xmath55 involve the all orders in @xmath47 .
* c * the on - diagonal matrix element @xmath39 ( see text),title="fig : " ] in the lowest order in @xmath27 the process amplitude is _ uniquely _ determined by the hamiltonian @xmath56 : @xmath57 where the antineutron green function @xmath58 is @xmath59 @xmath60 since @xmath37 , @xmath38 .
the hamiltonian @xmath47 acts in the block @xmath49 only and so @xmath58 is completely determined by @xmath48 . for the total process width @xmath61 one obtains @xmath62 @xmath63
the normalization multiplier @xmath64 is the same for @xmath61 and @xmath32 .
the time - dependence is determined by the exponential decay law : @xmath65 the realistic set of parameters is @xmath66 mev , and @xmath67 mev . for the lower limit on the free - space @xmath0 oscillation time @xmath68 eq .
( 20 ) gives @xmath69 .
we consider the model shown in fig .
1b ( model * b * ) .
if @xmath49 is determined by ( 14 ) , the process amplitude ( 15 ) follows uniquely from ( 12 ) , ( 13 ) . on the other hand , for the one - step process of the antineutron annihilation in the medium @xmath70
, the annihilation amplitude @xmath55 can be defined through the hamiltonian @xmath46 and not @xmath47 : @xmath71 @xmath55 contains the all @xmath23-medium interactions including antineutron rescattering in the initial state . in this case
the amplitude of process ( 1 ) is ( see fig .
1b ) @xmath72 the definition of annihilation amplitude through eq .
( 21 ) is natural since it corresponds to the observable values .
there are many physical arguments in support of the model ( 22 ) .
however , this model contains infrared singularity @xmath73 since @xmath74 .
this problem has been considered in [ 8,9 ] .
( in [ 8,9 ] only model * b * has been studied . ) for the purposes of this paper it is essential that model ( 22 ) gives linier @xmath32-dependence @xmath75 , as well as model ( 15 ) .
consequently , the model with hermitian hamiltonian gives linear @xmath32-dependence at any definition of annihilation amplitude .
the lower limit obtained in the framework of the model * b * is @xmath76 [ 8,9 ] .
if @xmath77 , model * a * converts to model * b*. the huge distinction between the values of @xmath68 and @xmath78 is due to the @xmath79-dependence of ( 18 ) .
the @xmath32-dependence of the models with hermitian and non - hermitian hamiltonians differs fundamentally : @xmath80 , whereas @xmath81 . at the same time the annihilation is the main effect which defines the process speed .
one of two models is wrong .
we assert that model with non - hermitian hamiltonian @xmath28 is wrong since ( 11 ) follows from ( 4 ) which is inapplicable for non - unitary @xmath1-matrix .
besides , @xmath82 is unknown ( see below ) .
notice that the result ( 11 ) takes place in the all standard calculations [ 4 - 9 ] because they are based on the optical potential .
we compare ( 18 ) and ( 11 ) : @xmath83 for the parameters used in ( 20 ) ( @xmath66 mev and @xmath67 mev ) we have @xmath84 .
when @xmath85 as well as in the case of the model * b * , eqs .
( 18 ) and ( 23 ) are invalid . however , in that event @xmath86 as well [ 9 ] . on the other hand , for small @xmath32 eq .
( 11 ) _ coincides _ with ( 18 ) : @xmath87 this is because the hamiltonian @xmath28 is _ practically hermitian _ in this case .
if @xmath88 , the results of models with hermitian and non - hermitian hamiltonians coincide . we would like to stress this fact .
it can be considered as a test for the model given in sect .
also we believe that the hamiltonian @xmath28 describes correctly the @xmath0 transition with @xmath23 in the final state @xmath89 since eq .
( 4 ) is not used in this case [ 10 ] .
consequently , for the non - unitary models the optical theorem can be used for the estimations if the absorption is small : @xmath90 this is not the case for the @xmath0 transition in the nuclear matter .
because of this we performed the calculations in the framework of unitary models
. if the optical potential is used for the problems described by schrodinger - type equation ( optical model ) , the unitarization takes place [ 10 ] : the matrix elements and optical potential are fitted to @xmath91-atom ( @xmath92-atom , @xmath93-atom ) and low energy scattering data .
however , the optical potential is the effective one .
the @xmath0 transition is described by the system of coupled equations [ 5,8 - 10 ] .
the corresponding @xmath1-matrix differs principally [ 10 ] .
even the physical meaning of @xmath94 is uncertain : one can not get continuity equation from the system of coupled equations .
_ there are no experimental data and unitarization _ in this case .
equations ( 3 ) and ( 4 ) are inapplicable . besides , since @xmath82 is unknown , eq .
( 3 ) is inoperative in principle .
( we also note that it is meaningless to impose the condition of probability conservation @xmath95 since @xmath96 is unknown . ) at the same time , in the model described in sect . 3 eq .
( 4 ) is used , where @xmath39 is unknown .
the consequences are illustrated by eqs .
( 11 ) , ( 18 ) and ( 23 ) .
if the optical theorem is not applied , the range of applicability of optical potential is considerably wider . as a first approximation
, it can be used in the calculation of the diagrams with @xmath23 in the intermediate or final states [ 10 ] .
for example , the channel @xmath89 . in these cases
the off - diagonal matrix elements are calculated directly without use of optical theorem . in view of the uncertainty in the annihilation amplitude we can not decide between models
* a * and * b*. the same is true for the value of the antineutron self - energy @xmath97 in ( 16 ) .
these problems are general in the theory of reactions . however , in the problem under study the values of @xmath61 and @xmath68 are extremely sensitive to @xmath48 ( see ( 18 ) ) .
this is because the amplitude ( 22 ) is in the peculiar point @xmath73 .
the small change of @xmath48 affects the result vastly : @xmath98 .
owing to this @xmath68 and @xmath78 differ greatly .
the main results are as follows .
\(a ) for the non - unitary models the optical theorem and condition of probability conservation can be used for the estimations if the absorption is small . for the models with the essentially non - hermitian hamiltonians
they can be applied only if @xmath1-matrix is unitarized .
this is also true for the @xmath99 oscillations in particular .
\(b ) the huge distinction between the values of @xmath68 and @xmath78 stems from the fact that @xmath98 and @xmath77 . at present
it is impossible to decide between models * a * and * b * as well as to determine the value of @xmath48 exactly .
so the values @xmath100 and @xmath101 yr are interpreted as the estimations from below ( conservative limit ) and from above , respectively .
the realistic limit @xmath4 can be in the range @xmath5 .
the estimation from below @xmath100 exceeds the restriction given by the grenoble reactor experiment [ 11 ] by a factor of 14 and the lower limit given by potential model by a factor of 5 .
further investigations are desirable . 99 v.a .
kuzmin , jetf lett .
* 12 * , 228 ( 1970 ) s.l .
glashow , preprint hutp-79/a059 ( harvard , 1979 ) , unpublished r.n .
mohapatra and r.e .
marshak , phys .
. lett . * 44 * , 1316 ( 1980 ) k.g .
chetyrkin , m.v .
kazarnovsky , v.a .
kuzmin and m.e .
shaposhnikov , phys .
b * 99 * , 358 ( 1981 ) p.g.h .
sandars , j. phys . *
g6 * , l161 ( 1980 ) j. arafune , o. miyamura , prog .
. phys . * 66 * , 661 ( 1981 ) k. s. ganezer , presented at the international workshop on search for baryon and lepton number violations , berkley , september 2007 ( http://inpa.lbl.gov/blnv/blnv.htp ) v.i .
nazaruk , phys .
c * 58 * , r1884 ( 1998 ) v.i .
nazaruk , eur . phys . j. c * 53 * , 573 ( 2008 ) v.i .
nazaruk , eur .
j. a * 31 * , 177 ( 2007 ) m. baldo - ceolin _ et al .
_ , z. phys .
c * 63 * , 409 ( 1994 ) | the applicability of the optical theorem in the models with the non - hermitian hamiltonian is studied . by way of example
we consider the @xmath0 transition in a medium followed by annihilation .
it is shown that an application of optical theorem for the non - unitary @xmath1-matrix leads to the qualitative error in the result . the alternative model which is free from drawback given above
is studied as well .
-0.6 in 11.30.fs ; 13.75.cs keywords : unitarity , optical theorem , infrared divergences * e - mail : [email protected] |
it is by now a well - appreciated fact that non - abelian gauge theories display many interesting and beautiful properties which are , no doubt , indicative of their deeply geometrical nature and symmetry .
already at the level of perturbation theory , many of the scattering amplitudes , in quantum chromodynamics ( qcd ) for example , have a simple and elegant form , although large numbers of feynman diagrams have to be summed up to arrive at these @xcite .
this feature was originally understood in terms of calculational techniques based on string theory as well as recursion rules @xcite-@xcite .
( a field theoretic understanding of the string - based techniques has also emerged @xcite . )
there are also indications that there are integrable field theories hidden in non - abelian gauge theories which describe some aspects of these theories , such as scattering for certain specific choices of helicities as well as scattering in the regge regime @xcite-@xcite . given these features ,
it is clear that the theoretical exploration of the structure of even the perturbative scattering matrix can be quite useful . within a field - theoretic approach ,
recursion rules for scattering amplitudes have been very useful in understanding color factorization and many other properties .
these rules were originally derived for tree - level amplitudes using diagrammatic analyses by berends and giele and have later been extended upto the one - loop level @xcite-@xcite . elaboration and extension of this technique , we feel , can be very fruitful . in this letter
, we present a derivation of the recursion rules within a functional formalism without reliance on diagrammatic analysis and in a way valid for any field theory . from the basic equations ,
recursion rules valid upto arbitrary loop order can be obtained , although at the expense of increasing algebraic complexity .
we explicitly display the tree - level and one - loop recursion rules .
renormalization constants , which one must consider beyond the tree - level are also easily incorporated in a functional derivation .
we start by considering a scalar field theory with action of the form @xmath0=s_0[\phi]+ s_{int}[\phi]\,,\ ] ] where @xmath1 $ ] is the free action of the form @xmath2=\int\frac{1}{2}\phi k\phi$ ] and @xmath3 is the interaction part of the action .
specifically , @xmath4 can be taken to be of the form @xmath5 .
the functional for the s - matrix elements can be written as @xcite @xmath6=e^f~\int [ d \phi ] \
e^{is_0[\phi]+is_{int}(\phi+\varphi)}\,,\ ] ] where @xmath7 can be expanded as @xmath8 @xmath9 are the one - particle wave functions and are solutions of the free field equation @xmath10 .
@xmath11 is given by @xmath12 and is introduced in ( [ 2 ] ) to give the matrix elements for subscattering processes where some of the particles fly by unscattered .
the matrix element for a process where particles of momenta @xmath13 scatter to particles of momenta @xmath14 is given by @xmath15\right]_{\varphi=0}\,.\ ] ] since @xmath11 is not particularly relevant for our calculations below we can drop it in what follows
. further we consider @xmath7 to be an arbitrary function rather than a solution to @xmath10 .
eventually , in obtaining the s - matrix elements we can choose it to be a solution to @xmath10 .
we thus define @xmath16 & = & \int[d\phi]\ e^{is_0[\phi]+is_{int}(\phi+\varphi ) -i\frac12\int\varphi k\varphi}\nonumber\\ & = & \int [ d\phi]\ e^{is[\phi]-i\int\varphi k\phi}. \label{5}\end{aligned}\ ] ] where we added a term @xmath17 for simplifications in what follows ; this of course does not contribute when @xmath18 . in terms of this @xmath19 ,
the s - matrix elements are given by @xmath20\right]_{\varphi=0}\ , .
\label{6}\ ] ] where @xmath21 from the second of equations ( [ 5 ] ) , we see that @xmath19 is the generating functional of the connected green functions with source @xmath22 and ( [ 6 ] ) represents the well - known lsz reduction formula .
consider now @xmath23 .
differentiating with respect to @xmath24 , we find @xmath25~ \phi e^{is[\phi]-i\int\varphi k\phi}.\ ] ] effectively , @xmath26 s inside the integral behave as @xmath27 . from ( [ 5 ] ) we also have the equation of motion @xmath28\ \left[\frac{\delta s}{\delta \phi } -k\varphi\right]e^{is[\phi]-i\int\varphi k\phi}=0,\ ] ] which can be written as @xmath29 where @xmath30 and @xmath31 .
@xmath19 itself generates connected as well as disconnected scattering processes .
if we write @xmath32=e^{ic[\varphi]}$ ] , where @xmath33 $ ] describes connected processes only , ( [ df ] ) reduces to @xmath34 this can be regarded as a nonlinear functional differential equation for the s - matrix generating functional and can be used for deriving systematic recursion rules for scattering amplitudes .
we shall do this for yang - mills theory in the next section .
although it is not crucial to the discussion of recursion rules , one may also work with the quantum effective action or the generating functional for one - particle irreducible vertices @xmath35 $ ] defined by @xmath36=c[\varphi]+\int(k\varphi)\phi\,,\ ] ] with the connecting relations @xmath37 then , ( [ dc ] ) becomes @xmath38 or , in terms of @xmath35 $ ] , @xmath39 the right hand side involves the derivative of @xmath40 with respect to @xmath24 which is the two - point correlator @xmath41 given by @xmath42 it satisfies the basic schwinger - dyson equation for the theory , viz .
, @xmath43\widetilde{g}(y , z)=i\delta^{(4)}(x - z)\,.\ ] ] equation ( [ dg ] ) supplemented by ( [ sd ] ) can thus be regarded as a nonlinear equation for @xmath35 $ ] . from the connecting relations ( [ connecting ] )
we find that @xmath44 for @xmath10 , as is appropriate for s - matrix elements . in this case , from the definition of @xmath45 , we have @xmath46 = \gamma[\phi]\vert_{\frac{\delta\gamma}{\delta\phi}=0 } \,.\ ] ] the s - matrix is then given by @xmath47}\right]_{\frac{\delta\gamma}{\delta\phi}=0}\,.\ ] ] this relation gives a nonperturbative definition of the s - matrix .
the free data in the solutions to @xmath44 are the quantities on which @xmath19 depends .
( perturbatively , the free data are the amplitudes @xmath48 in the solution for @xmath24 . )
the fact that the s - matrix can be obtained as the exponential of ( @xmath49 times ) the action evaluated on solutions of the equations of motion is rather well known @xcite , @xcite .
we start with the gauge - fixed lagrangian @xmath50 of an @xmath51-yang - mills theory given by @xmath52 where @xmath53\,,\nonumber\\ d_\mu c&=&\partial_\mu c+g[a_\mu , c]\,,\end{aligned}\ ] ] with @xmath54=f^{abc}t^c$ ] and @xmath55 .
@xmath56 are matrices in the fundamental representation of @xmath51 .
the free part and the interaction parts of the action are respectively identified as @xmath57 + \frac{g^2}2{\rm tr\;}[a_\mu , a_\nu]^2 \\ { \cal l}^{(2)}_{int}&= & 2g{\rm tr\;}\partial_\mu\bar{c}[a^\mu , c]\ , .
\label{lag}\end{aligned}\ ] ] following the discussion of sec .
2 , the functional @xmath19 which gives the s - matrix can be constructed as @xmath58 = \int e^{is - i\int a_\mu^a(-\partial^2)a^{\mu a}}[da][dc][d\bar{c}]\,,\ ] ] the transition amplitude for @xmath59 gluons of momenta @xmath60 , polarizations @xmath61 and colors labelled @xmath62 to go to @xmath63 gluons of momenta @xmath64 , polarizations @xmath65 and color labels @xmath66 is @xmath67 = \left.\int\prod_i e^{-ik_ix_i}\epsilon_{\mu_i}^{(i ) } \frac{\delta}{\delta a_{\mu_i}^{a_i } } \prod_je^{ip_jy_j}\epsilon_{\nu_j}^{(j ) } \frac{\delta}{\delta a_{\nu_j}^{a_j}}{\cal f}[a]\right|_{a=0}\,.\ ] ] from ( [ df ] ) @xmath19 satisfies the equation @xmath68e^{is}\,,\end{aligned}\ ] ] where @xmath69+g[f_{\mu\nu},a^\nu]\ , .
\end{aligned}\ ] ] and @xmath70 .
as we have written it , the term in ( [ df2 ] ) involving ghosts can not immediately be replaced by derivatives on @xmath19 . for this
, we proceed as follows . integrating out the ghost fields in @xmath19 we get faddeev - popov determinant @xmath71 , which is equivalent to a term in the action @xmath72 .
this leads to a ghost - current of the form @xmath73_{adj } \partial^{\nu_1}g(x , y_1)\cdots \partial^{\nu_n}g(y_{n-1},y_n)\partial_\mu g(y_n , x)\,,\nonumber\\ & & \end{aligned}\ ] ] where we have introduced the adjoint representation of @xmath56 s by @xmath74 .
( @xmath75 in this notation reads @xmath76 . )
the trace in the above equation may be written in terms of the generators in fundamental representation using @xmath77 $ ] , @xmath78\ldots]\,.\ ] ] therefore we can finally write the ghost part of the current ( as a matrix in the fundamental representation ) as @xmath79\ldots ] , \nonumber\end{aligned}\ ] ] where @xmath80 then ( [ df2 ] ) finally becomes @xmath81_{a = ig\frac{\delta}{\delta a}}{\cal f}\,.\ ] ] for connected part @xmath82=-i\ln{\cal f}[a]$ ] , @xmath83_{a=-\hbar g\frac{\delta c}{\delta a } + i\hbar g\frac{\delta}{\delta a}}1\,.\ ] ] this is the basic equation for the scattering amplitudes .
the @xmath84-expansion of this equation leads to recursion rules upto arbitrary order .
we have explicitly displayed the factors of @xmath84 in ( [ dc2 ] ) .
recall that the interaction part of the action carries @xmath85 and that each propagator @xmath86 carries @xmath84 .
the ghost terms , arising from the determinant , have @xmath87 .
@xmath82 $ ] in ( [ dc2 ] ) has an expansion of the form @xmath88\equiv - i\ln{\cal f}[a]=\frac1\hbar c^{(0)}+c^{(1)}+\hbar c^{(2)}+\cdots\,.\ ] ] starting from ( [ dc2 ] ) and using the above expansion for @xmath82 $ ] , one can , in principle , systematically derive recursion relations to any desired order in @xmath84 .
let us begin with a tree - level recursion formula for @xmath89 .
this is simply given by @xmath90 where @xmath91 i.e. , @xmath92 this is just the equation of motion as it should be .
@xmath93 is essentially the same object as the current that berends and giele @xcite used to derive tree - level recursion relation for gluon scattering processes ; if one expands @xmath93 in powers of @xmath94 , the coefficient function of each term gives one - gluon off - shell current of ref .
@xcite when multiplied by polarization vectors of on - shell external gluons .
more generally , if one considers @xmath94 s off shell , it gives the generalized current with off - shell gluons which has been used for @xmath95 process and some one - loop calculations @xcite .
also , notice that from the form of current @xmath96 in ( [ current ] ) , it is obvious that color factors are factorized from coefficient functions and we can write @xmath97 the coefficient functions @xmath89 s do not carry color indices .
we also have @xmath98 . using ( [ a ] ) in ( [ tree ] ) , @xmath99a(x)a(y)\nonumber\\ & & -\sum\int [ 2c_\nu^{(0)}(x , x)c_\mu^{(0)}(x , y)c^{\nu ( 0)}(x , z ) -c_\nu^{(0)}(x , x)c^{\nu ( 0)}(x , y)c_\mu^{(0)}(x , z)\nonumber\\ & & \qquad\quad -c_\mu^{(0)}(x , x)c_\nu^{(0)}(x , y ) c^{\nu ( 0)}(x , z)]a(x)a(y)a(z)\ , , \label{xrecursion}\end{aligned}\ ] ] where @xmath100 and @xmath101 are collective indices and @xmath102 stands for @xmath103 etc .
if @xmath94 is restricted to be on - shell , the terms containing @xmath104 in the third line vanish because of current conservation
. we can transform this equation to momentum space by writing @xmath105 where momentum conservation has been taken into account .
then ( [ xrecursion ] ) becomes @xmath106 where @xmath107 and @xmath108 , @xmath109 are color - ordered vertices , @xmath110 this equation is the recursion relation for currents with off - shell gluons derived in @xcite,@xcite . in deriving higher order recursion relations for s - matrix elements , we must correct ( [ dc2 ] ) by including the appropriate renormalization constants . as usual
, we interpret fields and coupling constants as renormalized ones and assume that renormalization counterterms are included in @xmath111 of ( [ lag ] ) .
explicitly , @xmath112
-\frac{g^2}2z_1 ^ 2z_3^{-1}{\rm tr\;}[a_\mu , a_\nu]^2\nonumber\\ & & -2\delta { \tilde z_3}{\rm tr\;}\bar{c}(-\partial^2)c -2gz_1{\tilde z_3}/{z_3}{\rm tr\;}\partial_\mu\bar{c}[a_\mu , c]\,,\end{aligned}\ ] ] where @xmath113 , @xmath114 and @xmath115 are bare quantities with @xmath116 .
correspondingly , @xmath96 becomes @xmath117+g[f_{\mu\nu},a_\nu]\nonumber\\ & & -\frac12\delta z_3\partial_\nu(\partial_\mu a_\nu-\partial_\nu a_\mu ) -g(z_1 - 1)(\partial_\nu[a_\mu , a_\nu]-[f_{\mu\nu},a_\nu])\nonumber\\ & & + g^2z_1(z_1/z_3 -1)[[a_\mu , a_\nu],a_\nu]\,.\end{aligned}\ ] ] the extra counterterms contribute to one - loop or higher orders and will cancel ultraviolet divergences from loop integrals .
in addition , to account for the freedom of arbitrary finite renormalization we include a finite wavefunction renormalization constant @xmath118 in the external field , i.e. , @xmath119 .
first we study the action of @xmath129 on @xmath128 . from ( [ a ] ) , @xmath130 & = & \sum_{n=1}^\infty\sum_{m=1}^n\int c_{\nu\nu_1\ldots\nu_{m-1}\mu\nu_{m+1}\ldots\nu_n}^{(0 ) } ( x , y_1,\ldots , y_{m-1},y , y_{m+1},\ldots , y_n)g(x , y)\nonumber\\ & & \hspace{15mm}\times [ t^a , a_{\nu_1}(y_1)\cdots a_{\nu_{m-1}}(y_{m-1})t^a a_{\nu_{m+1}}(y_{m+1})\cdots a_{\nu_n}(y_n)]\,.\end{aligned}\ ] ] with the help of fierz identity for @xmath51 @xmath131 it becomes @xmath132 = \sum_{n=1}^\infty\sum_{m=0}^n\int \widetilde{c}_{\nu\mu}(x,1,\ldots , m;m+1\ldots , n ) a(1)\cdots a(m){\rm tr\;}[a(m+1)\cdots a(n)]\,,\ ] ] where @xmath133\,.\end{aligned}\ ] ] ( here the @xmath134-term in ( [ fiertz ] ) does not contribute because of the commutator ; notice also that the summation in ( [ gda ] ) starts with @xmath135 . )
thus differentiation of @xmath136 produces terms with trace over substrings of @xmath94 s . also , from ( [ j_ghost ] )
, we see that the ghost current has the same structure , @xmath137 where @xmath138 is the sum over all permutations of @xmath139 such that @xmath140 and @xmath141 . using ( [ fiertz ] ) , we get @xmath142,\end{aligned}\ ] ] where @xmath143 is the set of all permutations of @xmath144 that preserves the ordering of @xmath145 and the cyclic ordering of @xmath146 , while allowing for all possible relative orderings between elements in the two sets ( for example , @xmath147 is in @xmath143 but @xmath148 is not . ) ; @xmath149 is given by @xmath150 since ( [ oneloop ] ) is at most linear in @xmath126 , it is clear that @xmath126 can be written as @xmath151\,.\ ] ] then in ( [ oneloop ] ) we can separately equate terms with the same trace structure and @xmath152 s with different @xmath153 s do not mix with each other .
the functions we need are then @xmath154 corresponding to the term in ( [ a1 ] ) with no trace .
these obey the recursion rule obtained by substituting @xmath155 in ( [ oneloop ] ) .
the other amplitudes can be obtained from @xmath156 .
the coefficient functions corresponding to different trace structures have simple relations among them ( which allow us to construct the amplitudes with subtraces from @xmath156 ) .
we have @xcite,@xcite @xmath157 for this , we show that the right hand side of ( [ cnck ] ) satisfies the same equation as that for the left hand side .
this is most easily seen for the ghost current which we shall consider first .
obviously , an equation like ( [ cnck ] ) holds for @xmath149 s ( with the identification of the summation indices @xmath158 in ( [ j_ghost2 ] ) and @xmath153 in ( [ cnck ] ) ) , if @xmath159 in ( [ j_ghost2 ] ) is replaced by its lowest order term @xmath94 . for the terms with more than one @xmath94 s from one @xmath160
, there are potential discrepancies between both sides .
this is because in ( [ cnck ] ) the sumation is over all @xmath161 while , in ( [ j_ghost2 ] ) , we sum only over @xmath162 ( @xmath163 ) which does not include such permutations that mix indices from both @xmath164 set and @xmath165 set within one tree structure . however , the extra terms in ( [ cnck ] ) cancel out thanks to the symmetry properties of color - ordered vertices , @xmath166 which , when applied to ( [ precursion ] ) together with induction , leads to identities for @xmath167 , @xmath168 where @xmath169 is the same as @xmath170 defined above except that it does not include the cyclic permutations of @xmath165 set .
[ this equation can be considered as a generalization of the so - called `` dual ward identity '' for tree amplitudes which corresponds to @xmath171 case @xcite,@xcite . ] thus ( [ cnck ] ) holds for ghost current .
we shall now show that the relation ( [ cnck ] ) connecting amplitudes with different trace structures holds for the nonghost terms in ( [ a1 ] ) as well . towards this , consider the terms with differentiation @xmath172 in ( [ oneloop ] ) .
diagramatically , those terms correspond to one - loop gluon diagrams with the external leg @xmath173 directly connected to the loop . in our setting
, we can proceed as follows . applying ( [ precursion ] ) repeatedly to @xmath174
we can identify the internal line for each term from ( [ precursion ] ) .
it connects @xmath173 and @xmath175 and the other external legs are ordered in clockwise direction in such a way that @xmath176-th legs are below the line while @xmath177-th legs are above the line ( fig . 1 ) .
then we draw a line connecting @xmath173 and @xmath175 which enclose @xmath177-th legs so that it represents that the legs inside the loop are traced . then , it is easy to see which terms generate which color structures
. obviously , given such a diagram , we get the same subtrace structure from diagrams made by altering the relative order of trees belonging to different sets while keeping the order of @xmath178 and the cyclic order of @xmath179 ( fig.2 ) .
moreover , they are trivially related to terms which contribute to @xmath156 , i.e. , we can simply pull out the trees inside the loop to outside with minus sign , using symmetry property of vertices ( [ symmetry ] ) .
notice that during this procedure the order of @xmath179 is also reversed .
thus , essentially we have the same situation as in the case of ghost current and the relation of the type ( [ cnck ] ) holds in this case as well .
now it remains to consider the term from @xmath180 which contains @xmath181 s with less number of legs .
it corresponds to the gluon - loop diagram with the external leg @xmath173 attached to a tree .
this case , however , is not much different from the previous one and we can argue in the same way with the help of ( [ identity ] ) if needed . finally , there are counterterm contributions for @xmath182 case in contrast to @xmath183 @xmath184 .
but again the identity ( [ identity ] ) guarantees that those terms cancel out when the summation in ( [ cnck ] ) is done .
this completes the proof that both the left and the right hand sides of ( [ cnck ] ) satisfy the same recursion relation .
since ( [ cnck ] ) trivially holds for @xmath185 , they are indeed equal to each other . beyond one - loop ,
it is clear from ( [ dc2 ] ) that @xmath186 will in general have terms with @xmath158 subtraces , @xmath187\cdots{\rm tr\;}[a(m_k+1)\cdots a(n)]\,,\nonumber\\ & &
\end{aligned}\ ] ] and each @xmath188 with different @xmath158 will satisfy its own equation . it might be possible to find simple relations between them as in one - loop case . s. j. parke and t. r. taylor , phys .
* 56 * ( 1986 ) 2459 ; nucl .
* b269 * ( 1986 ) 410 .
f. berends and w. giele , nucl .
* b306 * ( 1988 ) 759 .
d. a. kosower , b .- h .
lee and v. p. nair , phys .
* 201b * ( 1988 ) 85 .
see , for example , z. bern , l. dixon and d. a. kosower , slac - pub-7111 , hep - ph/9602280 ( to appear in ann .
z. bern and d. c. dubna nucl .
* b379 * ( 1992 ) 562 ; m. j. strassler , nucl .
* b385 * ( 1992 ) 145 .
l. n. lipatov , phys .
lett . * 309b * ( 1993 ) 394 ; l. d. faddeev and g. p. korchemsky , phys .
* 342b * ( 1995 ) 311 .
v. p. nair , phys .
* 214b * ( 1988 ) 215 .
w. a. bardeen , fermilab - conf-95 - 379-t , presented at yukawa international seminar 95 : from the standard model to grand unified theories , kyoto , japan , 21 - 25 aug .
1995 ; k.g .
selivanov , itep-21 - 96 , hep - ph/9604206 .
d. cangemi , ucla-96-tep-16 , hep - th/960508 ; g. chalmers and w. siegel , itp - sb-96 - 29 , hep - th/9606061 ; v. e. korepin and t. oota , yitp-96 - 33 , hep - th/9608064 . l. d. faddeev and a. a. slavnov , _ gauge fields , introduction to the quantum theory _
, benjamin - cummings , ma ( 1980 ) .
g. mahlon , t .-
yan and c. dunn , phys .
* d48 * ( 1993 ) 1337 .
g. mahlon phys .
* d49 * ( 1994 ) 2197 ; 4438 .
a. jevicki and c. lee , phys .
* d37 * ( 1988 ) 1485 ; r. fukuda , m , komachiya , and m. ukita , phys .
* d38 * ( 1988 ) 3747 .
z. bern and d. a. kosower , nucl .
b362 * ( 1991 ) 389 . z. bern , l dixon , d. c. dubna , and d. a. kosower , nucl .
* b425 * ( 1994 ) 217 .
z. bern , l. dixon , and d. a. kosower , nucl .
* b437 * ( 1995 ) 259 . c. dunn and t .- m .
yan , nucl .
* b352 * ( 1991 ) 402 . | we present a functional derivation of recursion rules for scattering amplitudes in a non - abelian gauge theory in a form valid to arbitrary loop order .
the tree - level and one - loop recursion rules are explicitly displayed . |
magnetars are highly magnetized neutron stars with magnetic fields @xmath24@xmath1610@xmath25 g , at least two orders of magnitude higher than the crab - like pulsars ( @xmath24@xmath1610@xmath26 g ) .
two types of objects are thought to be magnetars the anomalous x - ray pulsars ( axps ) and the soft gamma repeaters ( sgrs ) , having spin periods @xmath27@xmath162@xmath112 s and x - ray luminosities , @xmath28 , much larger than their spin - down luminosities , @xmath2 ( woods & thompson 2006 ) . there is now growing evidence of a new and small class of pulsars , the high magnetic field pulsars ( hbps ) , with spin and magnetic properties intermediate between the rotation - powered pulsars and the magnetars , and with magnetic fields @xmath29 ; where @xmath30 g is the so - called quantum critical field .
it is still under debate whether these hbps form transient objects between the classical pulsars and the magnetars or whether they stand as a separate class of population .
so far , there are six hbps discovered with @xmath29 , 5 of which are discovered as radio pulsars ( see table 3 of safi - harb & kumar 2007 and references therein for a summary of their properties ) .
psr j1846@xmath10258 is an energetic hbp discovered in x - rays , powering a bright pulsar wind nebula ( pwn ) , at the center of the composite type snr kes 75 ( snr [email protected] ) of @xmath16 3.5 in diameter ( gotthelf et al . 2000 ) .
it has a rotation period @xmath27=324 ms , a period derivative @xmath31=7.1@xmath310@xmath12 s s@xmath5 , a magnetic field @xmath24=5@xmath310@xmath6 g , and a spin - down luminosity @xmath2=8.3@xmath310@xmath4 erg s@xmath5 . with a characteristic age of 723 yrs ,
it is likely the youngest of all known rotation - powered pulsars ( gotthelf et al .
no radio pulsations have been detected so far from this pulsar .
it has been classified as a crab - like pulsar because of its non - thermal hard x - ray spectrum and for powering a bright pwn ( gotthelf et al .
2000 , helfand et al . 2003 hereafter h03 ) .
recent timing measurements using the _ rossi x - ray timing explorer ( rxte ) _ placed an upper limit on the spin - down age of 884 yrs for the pulsar , the smallest estimated age for any rotation - powered pulsar based on a braking index @[email protected] ( livingstone et al .
2006 ) . new _
chandra _ and _ spitzer space telescope _
observations of the snr kes 75 suggests a wolf - rayet progenitor for the supernova explosion ( morton et al . 2007 ) . until now ,
all calculations for the psr j1846@xmath10258 were based on a distance of 19 kpc obtained from neutral hydrogen absorption measurements ( becker & helfand 1984 ) .
also , its x - ray efficiency ( @xmath33=@xmath17/@xmath2=5@xmath22 ; calculated using luminosities quoted by h03 at a distance of 19 kpc ) is very high compared to rotation - powered pulsars .
recently , leahy & tian ( 2008 ) obtained a more accurate distance of 6 kpc for the snr kes 75 using hi+@xmath6co observations of the vla galactic plane survey .
revision of the distance from 19 kpc to 6 kpc lowers the luminosity and the x - ray efficiency by a factor of 10 .
@xmath34-ray observations using @xmath35 detected soft @xmath34-rays from the pulsar and its pwn with a combined efficiency of @xmath162.7@xmath22 ( recalculated using a distance of 6 kpc ) in the @xmath0 ( 210 kev ) and @xmath35 ( 20100 kev ) bands ( mcbride et al .
2008 ) . psr
j1846@xmath10258 s spin properties are remarkably similar to those of the radio pulsar psr j1119@xmath16127 , however their x - ray properties are very different ( see [ 4.2 ] ) .
although psr j1846@xmath10258 has been classified as a crab - like pulsar , its spin properties and unusually high x - ray efficiency have long been suspected to be linked to its magnetar - strength field ( e.g. gotthelf et al . 2000 ) . in order to compare the properties of these two hbps and shed light on the intriguing nature of psr j1846@xmath10258
, we analyzed four recent _ chandra _ archived observations . in this letter , we report the results from this analysis and show the first evidence of variability from the pulsar , revealing an activity most likely linked to its magnetar - strength field .
psr j1846@xmath10258 was observed with the _ advanced ccd imaging spectrometer ( acis ) _ onboard the _ chandra x - ray observatory ( cxo ) _ in 2000 oct 15 ( _ obsid _ : 748 ) , and again in 2006 june 7 - 12 ( _ obsid _ : 6686 , 7337 , 7338 , 7339 ) .
the source was positioned at the aimpoint of the back - illuminated s3 chip of the acis .
the ccd temperature was @xmath1120 c , with a ccd frame readout time of 3.2 s. the data were reduced using the standard ciao 3.4 routines .
the resulting effective exposure time was 192.2 ks for all the five observations .
figure 1 shows the combined acis - s3 image of the thermally emitting plasma from the snr kes 75 ( studied in detail by morton et al .
2007 ) and the psr j1846@xmath10258 ( @xmath36=18@xmath3746@[email protected]@[email protected] , @xmath41=@[email protected]@xmath180.2 ; h03 ) surrounded by a bright and hard pwn . the background - subtracted images were divided in the soft ( 0.5 - 1.15 kev ; red ) , medium ( 1.15 - 2.3 kev ; green ) and hard ( 2.3 - 10.0 kev ; blue ) energy bands , adaptively smoothed using a gaussian with @xmath43=1@xmath44 - 3@xmath44 for a significance of detection 2 to 5 , and finally combined to produce the image shown in fig .
a detailed spatially resolved spectroscopic study of the pwn features is beyond the scope of this paper . here
, we will focus on the overall spectrum of the pwn to study any changes in its properties from 2000 to 2006 in the context of the variability detected from the pulsar ( see [ 4.1 ] ) .
the spectroscopic study of psr j1846@xmath10258 and its pwn was carried out in the 0.510 kev energy range using xspec version 12.4.1 .
the spectrum of the pulsar was extracted by selecting a circular region of 2-radius centered on the pulsar ( fig .
1 ) , encompassing 90@xmath22 of the encircled energy .
the background region was selected from an annular ring from 3 to 4 centered on the pulsar to subtract the contamination by the surrounding bright pwn .
the spectrum of the pwn was extracted from a circular region of 18-radius ( fig .
the background was extracted from a source - free region to the north - west of the pulsar .
the spectra extracted were grouped by a minimum of 20 and 50 counts per bin for the 2000 and 2006 data , respectively , and the errors were calculated at the 90@xmath22 confidence level . a simple absorbed power - law ( pl ) model to the pulsar gave a poor fit due to ccd pileup7edavis / papers / pileup2001.ps ] leading to the artificial flattening of the observed spectrum .
psr j1846@xmath10258 , being a bright source , is affected by pileup with an estimated pileup fraction of @xmath1610@xmath22 for the 2000 data ( h03 ) and 18@xmath22 for the 2006 data . to account for pileup in our analysis
, we included a convolution model ( `` pileup '' ) in xspec .
the resulting model ( fig .
2(a ) , table 1 ) produced an excellent fit .
the spectra were further examined by adding a blackbody ( bb ) component to the model ( after freezing the pl parameters ) to account for any surface thermal emission from the pulsar .
we obtained a bb temperature @xmath45=0.74@xmath46 kev ( @xmath47=8.6@xmath48@xmath310@xmath49 k ) and an unabsorbed thermal flux of ( 4.7@xmath50 erg @xmath15 s@xmath5 ( 0.510 kev ) , which is only @xmath162@xmath22 of the total non - thermal flux .
however , the current statistics did not allow us to confirm or constrain the additional bb component . using the bb component and assuming isotropic emission
, we infer a bb radius @xmath51=0.23@xmath52 km , which is in agreement with the conventional polar cap radius r@[email protected]@xmath54=0.27 km .
we also independently carried out the spectral analysis of the 2000 data by freezing @xmath55=3.96@xmath310@xmath56 @xmath15 to compare with the values obtained by h03 .
our observed photon index @xmath7=1.32@xmath8 is in good agreement with that obtained by h03 ( @[email protected] ) . due to the limited number of counts
, we could not add a bb component to the 2000 data .
the variability of the pulsar s spectrum from 2000 to 2006 is discussed in [ 4.1 ] .
the pwn is well fitted with an absorbed pl as summarized in table 1 . by comparing the 2000 and the 2006 spectral parameters ,
we note that the pwn s luminosity and photon index have increased by @xmath1611@xmath57@xmath22 and @xmath163@xmath58@xmath22 , respectively ; however within error these changes are not significant .
the pl fit to the pulsar clearly shows a flux enhancement and a spectral softening from 2000 to 2006 ( fig .
2(a ) , table 1 ) . the corresponding increase observed in the luminosity by more than a factor of 6 provides evidence for brightening of the pulsar , suggesting for the first time an activity most likely associated with its ultra - high @xmath24-field .
we now discuss the implications of the observed variations in the spectrum of psr j1846@xmath10258 considered as a crab - like pulsar so far .
firstly , rotation - powered pulsars are known to be steady while psr j1846@xmath10258 has brightened by a factor of 6 , a behavior seen in magnetars .
for example , the transient magnetar xte j1810 - 197 brightened by 2 orders of magnitude in 2003 ( ibrahim et al .
2004 ) ; and the most recently discovered 2-s magnetar , 1e 1547.0@xmath15408 ( camilo et al .
2007 ) , had its x - ray flux vary by a factor of 7 between 1980 and 2006 ( gelfand & gaensler 2007 , halpern et al . 2007 ) .
secondly , the pulsar s spectrum softened as it brightened ( @xmath7 increased from 1.32 to 1.97 ) , indicating a spectral change that could be tied to a change in its @xmath24-field configuration or a magnetar - like burst .
thirdly , there is a hint that the pwn s spectrum has also changed as seen from the increase in its @xmath28 by @xmath1611@xmath22 and a slight softening of its spectrum . while these changes are not significant within error , they support the emergence of a magnetar - like burst from psr j1846@xmath10258 which would have injected relativistic particles in the pwn , thus enhancing its brightness observed in 2006 .
furthermore , as pointed out by kargaltsev & pavlov ( 2008 ) , it is possible that the unusually high efficiency of the pwn could be explained by a series of magnetar - like bursts .
therefore , based on the observed changes , we conclude that psr j1846@xmath10258 revealed its identity as a magnetar , likely after a magnetar - like burst .
like the young rotation - powered pulsars , psr j1846@xmath10258 s x - ray spectrum is described by a pl spectrum with a hard photon index , even at its steeper phase ( @xmath7=1.97 ) .
the only 2 magnetars having hard pl x - ray spectra with @xmath7@xmath162 are sgr 1806@xmath120 and sgr 1900 + 14 ( woods & thompson , 2006 ) .
as well , its x - ray efficiency , despite being high in comparison to rotation - powered pulsars , is still smaller than 1 a characteristic that differentiates rotation - powered pulsars from magnetars .
although these properties pose some difficulty in interpreting the pulsar as a newly emerging magnetar , they can be attributed to the pulsar being young and highly energetic . in particular ,
magnetars x - ray luminosities can easily exceed their spin - down luminosity because they have a much lower @xmath2 ( @xmath1610@xmath59 erg s@xmath5 ) , ranging from @xmath166@xmath310@xmath60 erg s@xmath5 for 1e 2259 + 586 to 5@xmath310@xmath19 erg s@xmath5 for sgr 1806@xmath120 .
the highest known @xmath2 of a magnetar is that of the 2-s magnetar , 1e 1547.0@xmath15408 , with @xmath2=1.0@xmath310@xmath21 erg s@xmath5 ( camilo et al .
2007 ) , still @xmath162 orders of magnitude smaller than that of psr j1846@xmath10258 .
the x - ray luminosity of 1e 1547.0@xmath15408 only just exceeded its @xmath2 during outburst .
furthermore , magnetars are known to spin down fast to a period of a few seconds in @xmath6110,000 yrs , but psr j1846@xmath10258 is only @xmath61900 years and hence , its @xmath2 ( @xmath62@xmath63 ) is still high .
therefore , based on pure energetic grounds , psr j1846@xmath10258 s rotational energy loss can still power its @xmath28 and the surrounding pwn , like the rotation - powered pulsars . on the other hand , the unusually high x - ray efficiency ( of the pulsar and pwn ) and the variability observed between 2000 and 2006 suggests that at least a fraction of the x - ray luminosity is powered by magnetic energy that would be supplied e.g. during a magnetar - like burst .
finally we note that we can not rule out the possibility that the pulsar s @xmath28 exceeded its @xmath2 prior to the 2006 @xmath0 observations during a magnetar - like burst that was missed by the @xmath0 observations .
of all the hbps , psr j1846@xmath10258 has the highest @xmath2=8.3@xmath310@xmath4 erg s@xmath5 ( gotthelf et al .
2000 ) and @xmath64=1.4@xmath22 ( using the 2006 flux from this work ) .
it has spin parameters similar to the hbp psr j1119@xmath16127 ( @xmath27=407 ms , @xmath31=4@xmath310@xmath12 s s@xmath5 , b=5@xmath310@xmath6 g ; camilo et al . 2000 ) and both pulsars are now believed to be at about the same distance , @xmath65@xmath166 kpc ( gonzalez & safi - harb 2003 , leahy & tian 2008 ) . despite the remarkable similarities between their spin properties
, they have very different x - ray properties .
the resolved spectrum of psr j1119@xmath16127 is best fitted by a two - component bb+pl model , with the bb component dominating the x - ray emission ( safi - harb & kumar 2007 ) , and characterized by a high pulsed fraction of 74@xmath1814@xmath22 in the 0.52.0 kev ( gonzalez et al . 2005 ) ; however , the emission from psr j1846@xmath10258 is dominated by a pl model with a hard photon index .
furthermore , the x - ray efficiency of psr j1119@xmath16127 is only @xmath160.05@xmath22 ( safi - harb & kumar 2007 ) , much lower than that of psr j1846@xmath10258 .
moreover , psr j1119@xmath16127 powers a weak pwn ( @xmath660.02@xmath22 , safi - harb & kumar 2007 ) while psr j1846@xmath10258 powers a much brighter pwn ( @xmath661.85@xmath22 ) . while the data presented here show the first evidence of spectral changes from psr j1846@xmath10258 between 2000 and 2006 , no changes were observed in the @xmath0 data of psr j1119@xmath16127 taken in 2002 & 2004 ( safi - harb & kumar 2007 ) .
these differences indicate that the b - field strength is not the sole parameter that determines their characteristics .
we attribute these differences mainly to their evolutionary state ( psr j1119@xmath16127 being twice as old and less energetic ) , their different @xmath55 ( psr j846 - 0258 s @xmath55 is at least twice as high as psr j1119@xmath16127 s thus burying any soft bb component ) , and likely different progenitor stars .
morton et al .
( 2007 ) suggests that kes 75 could have resulted from the rare explosion of a wolf - rayet star during a type ib or ic event .
furthermore , psr j1846@xmath10258 has revealed its magnetar - like nature only now . since
two transient magnetars have now been observed in the radio ( camilo et al .
2006 , 2007 ) , monitoring of psr j1846@xmath10258 in the radio and other wavelengths will shed more light on its magnetar nature . as well , monitoring of psr j1119@xmath16127 in the x - ray may reveal in the future a magnetar - like activity as observed for psr j1846@xmath10258 , and as recently inferred for 1e 1547.0 - 5408 ( gelfand & gaensler 2007 ) .
we have used the archival @xmath0 data to carry out a spectroscopic study of the hbp psr j1846@xmath10258 , classified as a crab - like pulsar , and its pwn in the 0.510 kev energy band .
our analysis of the pulsar s spectrum revealed flux enhancement by a factor of more than 6 and x - ray softening from 2000 to 2006 , accompanied by a hint of flux enhancement of the pwn .
this provides the first observational evidence of a magnetar - like emission associated with this hbp .
it is still intriguing that the x - ray luminosity , even at its observed peak , did not exceed the pulsar s rotational energy power , suggesting that spin - down can energetically power the x - ray emission from the pulsar and its pwn .
further observations of this source at all wavelengths and monitoring of other hbps will answer the question whether hbps are indeed transient magnetars disguised in other forms , and will shed more light on the nature of these sources and their link to magnetars and the rotation - powered pulsars .
we acknowledge a recent private communication with m. gonzalez who , prior to the submission of this paper , made us aware of the gavriil et al .
( 2008 ) finding ( paper posted on the arxiv after submission of our manuscript ) , which confirm the results and conclusions we had reached independently prior to the conversation , and which further establish the magnetar nature of the pulsar . it was also brought to our attention , after submission of our paper , that the variability detected from the pulsar with @xmath0 was also independently found by ng et al .
2008 ( p. slane , private communication ) .
we thank f. camilo for commenting on our manuscript . s.
safi - harb acknowledges support by the natural sciences and engineering research council of canada and the canada research chairs program .
this research made use of nasa s astrophysics data system and of nasa s heasarc maintained at the goddard space flight center .
becker , r .h . , & helfand , d. j. 1984 , , 283 , 154 camilo , f. , kaspi , v. m. , lyne , a. g. , manchester , r. n. , bell , j. f. , damico , n. , mckay , n. p. f. , crawford , f. 2000 , , 541 , 367 camilo , f. , ransom , s. m. , halpern , j. p. , reynolds , j. , helfand , d. j. , zimmerman , n. , & sarkissian , j. 2006 , nature , 442 , 892 camilo , f. , ransom , s. m. , halpern , j. p. , reynolds , j. 2007 , , 666 , l93 gavriil , f. p. , gonzalez , m. e. , gotthelf , e. v. , kaspi , v. m. , livingstone , m. a. , & woods , p. m. 2008 , science , in press gaensler , b. m. , & slane , p. o. 2006 , , 44 , 17 gelfand , j. d. , & gaensler , b. m. 2007 , , 667 , 1111 gonzalez , m. e. , & safi - harb , s. 2003 , , 591 , l143 gonzalez , m. e. , kaspi , v. m. , camilo , f. , gaensler , b. m. , & pivovaroff , m. j. 2005 , , 630 , 489 gotthelf , e. v. , vashisht , g. , boylan - kolchin , m. , & torri , k. 2000 , , 542l , 37 halpern , j. p. , & gotthelf , e. v. 2005 , , 618 , 874 halpern , j. p. , gotthelf , e. v. , reynolds , j. , ransom , s. m. , & camilo , f. 2007 , , in press helfand , d. j. , collins , b. f. , & gotthelf , e. v. 2003 , , 582 , 783 ibrahim , a. i. , et al .
2004 , , 609 , l21 kargaltsev , o. , & pavlov , g. g. 2008 , _ pre - print _ arxiv:0801.2602 ( astro - ph ) leahy , d. a. , & tian , w. w. 2008 , a&a ( letters ) , in press livingstone , m. a. , kaspi , v. m. , gotthelf , e. v. , & kuiper , l. 2006 , , 647 , 1286 mcbride , v. a. , et al .
2008 , a&a , 477 , 249 morrison , r. , mccammon , d. 1983 , , 270 , 119 morton , t. d. , slane , p. o. , borkowski , k. j. , reynolds , s. p. , helfand , d. j. , gaensler , b. m. , hughes , j. p. 2007
, , 667 , 219 ng , c .- y . ,
slane , p. o. , gaensler , b. m. , hughes , j. p. 2008
, abstract submitted to the head 2008 meeting safi - harb , s. , & kumar , h. s. 2007 , , submitted woods , p. m. , & thompson , c. 2006 , compact stellar x - ray sources . eds .
walter lewin & michiel van der klis .
cambridge university press , p.547586 | we present results from the archival @xmath0 observations of the 0.3 s x - ray pulsar psr j1846@xmath10258 associated with the supernova remnant ( snr ) kes 75 .
the pulsar has the highest spin - down luminosity ( @xmath2=8.3@xmath310@xmath4 erg s@xmath5 ) among all the high magnetic field pulsars ( hbps ) and has been classified as a crab - like pulsar despite its magnetic field ( 5@xmath310@xmath6 g ) being above the quantum critical field .
it is the only hbp described by a non - thermal crab - like spectrum , powering a bright pulsar wind nebula ( pwn ) .
our spectroscopic study shows evidence of spectral softening ( photon index @xmath7=1.32@xmath8 to 1.97@xmath9 ) and temporal brightening ( unabsorbed flux @xmath10=(4.3@xmath1110@xmath12 to 2.7@xmath1310@xmath14 erg @xmath15 s@xmath5 ) of the pulsar by @xmath166 times from 2000 to 2006 .
the 0.510 kev luminosity of the pulsar at the revised distance of 6 kpc has also increased from @xmath17=([email protected])@xmath310@xmath19 to ( 1.16@xmath20)@xmath310@xmath21 erg s@xmath5 , and the x - ray efficiency increased from [email protected]@xmath22 to 1.4@xmath23@xmath22 . the observed x - ray brightening and softening of the pulsar suggests for the first time that this hbp is revealing itself as a magnetar . |
the galactic x - ray binary gx 3394 was discovered @xmath330 years ago ( markert et al .
1973 ) , and it has been extensively studied since then . its optical companion is undetectable and most of the observed optical emission originates in an accretion disc . the system is classified as a low - mass x - ray binary from upper limits on the luminosity of the companion star ( e.g. , shahbaz , fender & charles 2001 ) .
gx 3394 is a transient source exhibiting state transitions between soft , hard and off states .
based on its temporal and spectral characteristics , it is considered to be a black hole binary ( sunyaev & revnivtsev 2000 ; zdziarski et al .
1998 , hereafter z98 ) , which is in agreement with an estimate of its mass function @xmath4 at the 95 per cent confidence ( hynes et al . 2003 ) .
most of its x@xmath5-ray spectra in luminous states are remarkably similar to those of cyg x-1 , and well fitted by comptonization ( z98 ; wardziski et al .
2002 , hereafter w02 ) . however ,
while the two sources share common properties on time scales @xmath6 a few days , their evolution patterns are rather different on longer time scales ( e.g. , smith , heindl & swank 2002b ) . in particular , the spectral evolution of cyg x-1 is directly correlated with the luminosity changes , whereas the spectral changes in gx 3394 have been observed to lag the variations of the luminosity .
these lags constitute a hysteresis in the lightcurve of the source ; the hard - to - soft state transitions during the rise phase occur at higher luminosities than the soft - to - hard ones during the decline phase .
a major difference between gx 3394 and cyg x-1 is that the latter is a persistent source in a high - mass x - ray binary .
thus , accretion in cyg x-1 is via a focused wind from the high - mass companion , while in gx 3394 it is due to roche - lobe overflow from the low - mass companion .
the different behaviour of low and high - mass black - hole binaries is likely to be due to different disc sizes ( e.g. , smith et al .
2002b ) . in a low - mass x - ray binary
, the disc is expected to be large .
this then leads to delays in the propagation of the accretion rate from the disc outer regions to the hot inner flow .
another consequence of the different types of accretion flows in the two systems is that whereas cyg x-1 is a persistent source , in which the bolometric luminosity ( and presumably the accretion rate ) varies only by @xmath15 or so ( zdziarski et al . 2002 , hereafter z02 ) ,
gx 3394 is a transient , in which accretion is unstable and the accretion rate in the vicinity of the black hole varies by at least several orders of magnitude .
consequently , gx 3394 goes through a wider range of spectral states than cyg x-1 . in this paper
, we first estimate the distance to the source based on recent optical / ir data .
then we study the long - term variability of gx 3394 during 19872004 and its theoretical interpretation .
we use data from the all - sky monitors ( asm ) on board _ ginga _ and _ rossi x - ray timing explorer ( rxte ) _ satellites , the burst and transient source experiment ( batse ) on board _ compton gamma ray observatory ( cgro ) _ , and the currrently available data from the proportional counter array ( pca ) and high - energy transient x - ray experiment ( hexte ) on board _
the distance , @xmath7 , to a source is , obviously , of major importance for interpretation of its observations .
therefore , we update here the constraints obtained by z98 .
they obtained @xmath8 kpc based on the extinction , which , however , does not impose an upper limit . in particular , there is a star at @xmath18 kpc within @xmath9 of gx 3394 with less extinction than it ( fig . 1 of z98 ) .
on the other hand , z98 found a kinematic distance of @xmath10 kpc , but they noted that this result was less secure than the extinction limit due to a possible peculiar velocity of the system ( see below ) . based on vlt observations of gx 3394 , shahbaz et al .
( 2001 ) estimated the relative contribution of the secondary to the gunn @xmath11-band magnitude as @xmath12 .
thus , the secondary must have an @xmath11-band magnitude @xmath321.4 ( note an error in shahbaz et al .
, who quoted a value of 20.4 on p. l19 ) , which corresponds , after correcting for @xmath13 ( z98 ) , to a late k - type star with an apparent diameter of @xmath14 , where we assumed the effective temperature of @xmath15 .
this estimate is also consistent with the lowest recorded near - ir magnitudes ( work in preparation ) . in particular , @xmath16 observed in 2003 jan.feb .
( buxton & bailyn 2004 ) , combined with the upper limit on the @xmath17 magnitude of shahbaz et al .
( 2001 ) is consistent with an early m - type star with @xmath18 .
since the system is a lmxb and the secondary is filling its roche lobe , the mass ratio , @xmath19 , is given by ( eggleton 1983 ) , @xmath20 where @xmath21 is the secondary s radius and @xmath22 is the separation . adopting then the orbital parameters of hynes et al .
( 2003 ) , namely the secondary semi - amplitude of @xmath23 , the period @xmath24 d , and @xmath25 , we obtain @xmath26 , @xmath27 , and @xmath28 , where @xmath29 is the inclination .
the distance is then @xmath30 , which is @xmath31 kpc for @xmath14 .
the system is not eclipsing , so @xmath32 .
on the other hand , the relatively large secondary mass function and the presence of moderate orbital modulation of the @xmath33 light during its optically - high state ( work in preparation ) , is inconsistent with a very low @xmath29 .
assuming then a plausible lower limit of @xmath34 , we obtain then an inclination - dependent lower limit on @xmath7 of @xmath35 this result is consistent with the recent lower limit of @xmath36 kpc for the kinematic distance of hynes et al .
( 2004 ) based on analysis of high - resolution optical spectra of the na d lines .
specifically , they find at least 9 line components arising from different clouds located along the line of sight .
they derive a range of mean cloud velocity of @xmath37 to @xmath38 km s@xmath39 , which they attribute to the galactic rotation , and argue that gx 3394 lies at or beyond the tangent point , i.e. , @xmath40 kpc .
they also suggest that the slightly positive systemic velocity of gx 3394 and the presence of the absorption component in na d lines with positive velocity , 2535 km s@xmath39 , indicates a location close to the solar circle on the far side of the galaxy , i.e. , at @xmath41 kpc , which distance they favour .
although they note that for such a large @xmath7 , gx 3394 would be at a @xmath42 kpc below the galactic plane , they still associate gx 3394 and the redshifted absorption with the warped material of the outer galactic disc . however , we prefer the location in the galactic bulge for the following reasons .
( i ) the analysis of high resolution spectra na d and ca k lines towards ob stars ( sembach et al . 1993 ) revealed similar redshifted components in most of their sample stars located in the galactic bulge , at @xmath43 kpc , @xmath44 .
( ii ) the maximum galactic disc warping on the southern hemisphere corresponds to the azimuthal angle of @xmath45 , where the warping starts at the radial distance from the galactic center of @xmath46 kpc and the hi disc is bending down to @xmath47 kpc at @xmath48 kpc ( nakanishi & sofue 2003 ) .
since both @xmath49 and @xmath17 are the galactocentric coordinates , gx 3394 can not be associated with this warped material because at a distance of @xmath50 kpc , its @xmath51 and @xmath52 kpc .
moreover , @xmath49 increases with the increasing @xmath7 , and a larger @xmath7 would locate gx 3394 even farther away from the galactic plane .
( iii ) the rather high peculiar velocity ( @xmath1140 km s@xmath39 if gx 3394 lies at the tangent point , hynes et al .
2004 ) is not unusual if gx 3394 is a member of the spheroidal galactic bulge population .
thus , we hereafter use @xmath53 kpc , compatible with the location in the galactic bulge . using equation ( [ d_min ] ) ,
this is also compatible with @xmath54 and @xmath55 , given the mass function of @xmath56 of hynes et al .
we note that the relatively strong compton reflection measured in x - ray spectra of gx 3394 ( e.g. , z98 ; w02 ; corongiu et al . 2003 ) appears incompatible with @xmath57 ; however , an investigation of this issue is beyond the scope of this paper .
the _ ginga_/asm ( tsunemi et al . 1989 ) monitored gx 3394 in the 120 kev band from 1987 march 6 to 1991 october 2 , and the countrate lightcurve was presented in kong et al .
( 2002 ) . here
we use instead energy fluxes in the bands of 11.6 , 1.63.3 , 3.35.0 , 5.012.5 , and 12.519.8 kev converted from the instrumental counts using the response matrix of tsunemi et al .
( 1989 ) . using eq .
( 1 ) of z02 , we measure the hardness across two adjacent energy bands in those and subsequent data by a photon spectral index , @xmath58 , corresponding to a power law that would yield the same hardness ratio as observed .
even if the actual spectrum is different from a power law , this @xmath58 still gives a measure of the overall hardness across the two energy bands that avoids the dependence on their relative sizes .
we do not correct for the interstellar absorption ( @xmath59 @xmath60 , z98 ) .
this results in the values of @xmath58 at @xmath63 kev being somewhat lower than the intrinsic ones . the used _ rxte_/asm ( levine et al .
1996 ) data cover the interval of 1996 january 62004 january 29 . during that time ,
gx 3394 was observed by the asm @xmath110 times a day on average with each ( dwell ) exposure of 90 s. here , we use the data binned over one day ( from xte.mit.edu/asm ) .
the daily averages are based on dwell data selected using a number of criteria ( see that web page ) , in particular the @xmath61 of the solution . here
, we have imposed an additional criterion of the minimum of 4 dwells per day ( derived using the asm crab data ) , to avoid possible spurious fluctuations .
the asm data have a positive bias of 1 mcrab ( heasarc.gsfc.nasa.gov ) , which we have removed based on the average asm crab counts .
we then convert the instrumental counts into energy fluxes in the ranges of 1.53 , 35 , and 512 kev using the conversion matrix given by eq .
( a1 ) of z02 , derived based on cyg x-1 data , but applicable to gx 3394 as the conversion is independent of the direction .
for this work , we have additionally confirmed its correctness using crab data .
we have compared the asm crab data with the _
bepposax_/mecs data ( from www.asdc.asi.it ) acquired on 1997 october 8 .
we have found that the asm conversion gives the energy fluxes in the above bands @xmath62 per cent higher than the mecs data .
this is a discrepancy smaller than typical ones between various x - ray instruments ( e.g. 30 per cent for hexte vs. pca ) .
then , the obtained 312 kev effective spectral index for the asm data is in the 2.042.10 range for 200-day averages , in good agreement with @xmath63 for the mecs data . on the other hand ,
the mecs data give the effective ( absorbed ) @xmath645 kev@xmath65 , somewhat softer than the range obtained for asm of 1.791.97 .
the low relative level of this discrepancy can be illustrated by the fact that it can be completely removed if we modify the lower energy bound from 1.5 kev to 1.55 kev .
in addition , the asm indices above 3 kev of gx 3394 are in excellent agreement with those from the pca , see below . the batse instrument aboard _ cgro _ monitored the x@xmath5-ray sky 1991 may 312000 may 25 .
it used earth occultation to identify point sources ( e.g. , harmon et al .
2002 ) . on average , 1215 occultations of gx 3394 were observed daily .
we use the data binned on a one - day timescale , and reject days with @xmath66 occultations . we use four energy bins , 2040 , 4070 , 70160 and 160430 kev .
the counts have been converted to energy fluxes based on the crab observations of jung ( 1989 ) , with the resulting conversion coefficients from countrate to energy flux for the four bands of 7.567 , 5.830 , 8.079 , @xmath67 erg , respectively . given that gx 3394 is a transient source , the data cover wide ranges of fluxes , and often are of limited statistics .
we have then rebinned adjacent ( i.e. , avoiding averaging over large gaps ) time bins until achieving a specified accuracy in the 1030 per cent ( 10@xmath68 ) range ( depending on the considered case ) , weighing the fluxes with the inverse square error .
however , during the off states , achieving those accuracies would require averaging over a very large number of days and often extending the average to include a part of a luminous state .
thus , we also limited the maximum number of averaged days to 100200 .
we sum the fluxes and their square errors when adding adjacent energy channels .
we also analyze all currently available pca and hexte data , following the method of done & gierliski ( 2003 , hereafter dg03 ) .
we extract one pca ( detectors 0 , 2 and 3 , top layer only , except for observations since mjd 52763 , when only detectors 0 and 2 were available ) and hexte ( cluster 0 ) spectrum per pointed observation , using ftools 5.3 .
we then fit the resulting 236 spectra with positive detections by a model consisting of a blackbody disc and thermal comptonization , thcomp ( zdziarski , magdziarz & johnson 1996 ) .
the effects of compton reflection are approximated by adding smeared edge and gaussian line .
the model is absorbed by @xmath69 @xmath60 .
we use the resulting unabsorbed ( unlike the case of the asm data ) energy fluxes in the 34 , 46.4 , 6.49.7 and 9.716 kev energy bands ( as in dg03 ) , normalized to the pca . as above
, we convert the hardness ratios to the average spectral indices .
this method proved to give robust and fairly model - independent fluxes as long as the model gives an adequate spectral fit , with @xmath70 ( dg03 ) .
this is the case for most of the data , except for a few soft - state spectra ( with @xmath713 ) , where the high - energy tail can not be properly reproduced by thermal comptonization , an indication of the non - thermal emission present in soft states ( see e.g. gierliski et al .
1999 ; zdziarski et al .
this , however , very weakly affects neither the 316 kev flux nor the bolometric one ( see below ) .
we also use the model bolometric flux , @xmath72 , for these data .
the systematic error related to the uncertainty in the method of calculating @xmath72 is not included in its quoted error ( based only on the countrate error ) .
[ pca_flux ] shows the resulting relation between the 316 kev flux ( with low systematic errors ) to the obtained model bolometric flux .
we see two different relationships in the hard and soft states , related to the distinctly different intrinsic spectra of the two states , peaking at @xmath1100 kev and @xmath11 kev , respectively ( see , e.g. , zdziarski & gierliski 2004 for a recent review ) . finally , we note that we do not study differences in absolute flux calibration of different instruments .
this adds a systematic error to the relative level of different lightcurves .
fig . [ lc_all](a ) shows the lightcurves based on our data , rebinned to the @xmath73 significance . fig .
[ lc_all](b ) shows the corresponding evolution of selected average spectral indices ( based on data rebinned to @xmath74 for clarity of display ) .
the indices of @xmath75 ( at energies @xmath33 kev ) correspond to the hard state , and those with @xmath76 , to soft states .
ginga_/asm first saw some weak hard states , and then a strong soft outburst around mjd 47400 , classified as including the very high state ( vhs ) based on _ ginga_/lac observations ( miyamoto et al .
( see zdziarski & gierliski 2004 for spectra and discussion of that state . ) after the flux declined by a factor of at least several , another soft peak ( possibly a continuation of the first outburst ) occured at mjd @xmath147600 .
the next outburst occured at mjd @xmath148100 , with the hard state followed by the soft one .
another outburst occured at mjd @xmath148500 .
the initial hard state seen by the batse was followed by the soft one , only beginning of which was recorded by _
ginga_. the next two outbursts took place at mjd @xmath149000 and @xmath149400 .
the source became very weak between the outbursts , with @xmath7710 kev@xmath78 kev @xmath60 s@xmath39 measured by _
asca _ on mjd 49246 ( asai et al .
1998 ) . until mjd 49400
, our data had shown 5 major outbursts occuring every @xmath1500 d. however , the character of the variability changed thereafter .
the outbursts became progressively more frequent , weaker , and with the increasing interim minimum fluxes ( rubin et al .
the intervals between the peaks became @xmath79 d. also , the _ rxte_/asm and pca / hexte monitoring showed only the hard state until mjd @xmath150800 , see fig .
[ lc_all ] . then , the variability pattern changed again with gx 3394 entering an extended soft state lasting for @xmath80 d. after its decline , a hard state took place , and then the source gradually faded .
this phase was well covered by the pca / hexte , which showed very low fluxes during the resulting off state . also ,
the _ bepposax _ observation on mjd 51403 showed @xmath8110 kev@xmath82 kev @xmath60 s@xmath39 ( kong et al .
we note there were pca observations on mjd 51396 and 51418 , which both show the 316 kev flux @xmath14 times higher
. this may be due to variability , or may indicate some problems with pca observations of gx 3394 at such low fluxes . on the other hand , rebinning of the _
rxte_/asm data gave the flux level of @xmath83 kev @xmath60 s@xmath39 , which may be an overestimate due to , e.g. , a residual contribution of the galactic ridge emission ( see w02 ) .
an indication of such a problem is also a discrepancy between the derived asm hard indices and those from the pca , much softer , see fig .
[ lc_all ] .
thus we hereafter not take into account the fluxes obtained by rebinning the _ rxte_/asm data during off states .
this off state lasted @xmath11000 d , until mjd @xmath152360 .
then another outburst took place ( smith et al .
2002a ) , with the hard - soft - hard spectral variability pattern .
the outburst ended at mjd @xmath152800 .
its flux level also has exceeded ( zdziarski & gierliski 2004 ) for the first time the highest previously recorded level of mjd @xmath147400 .
the gx 3394 went then to a quiescent state , with an observation by _
chandra _ showing the unabsorbed @xmath8411 kev@xmath85 kev @xmath60 s@xmath39 and @xmath86 on mjd 52911 ( gallo et al .
this is identical to the lowest previous upper limit from _ asca _ ( see above ) .
the newest outburst was detected on mjd 53036 ( smith et al .
2004 ; belloni et al .
2004 ; kuulkers et al .
2004 ) , as shown in fig .
[ lc_all ] till mjd 53068 .
the initial pca data show the source in its standard hard state , with @xmath87 . of particular interest
are the two recent complete outbursts covered by the _ rxte_/asm and pca / hexte , showing the hard - soft - hard spectral evolution .
[ outbursts ] shows the asm and pca / hexte fluxes and spectral indices in detail .
the two soft outbursts ( red and black symbols ) are similar in shape , but their 1.512-kev flux and duration are higher and lower by factors of @xmath12 and @xmath11.5 , respectively , in the second outburst .
a likely cause of this difference is the behaviour before the soft state . before the first outburst ,
gx 3394 underwent an extended period of a variable hard state , during which its broad - band flux was persistently relatively high for @xmath88 d , whereas the second outburst followed a @xmath88 d period of the off state , see fig .
[ lc_all ] .
the vertical lines marking the state transitions in fig . [ outbursts ] are at mjd 50812 , 51211 , 52401 and 52725 .
the last number is based on the asm data , which show a strong hardening on that day , see fig .
[ outbursts](b ) . on the other hand , the pca / hexte data favour a slightly later soft - to - hard transition day , mjd 52731 , consistent with a softening of the 312 kev asm spectrum after the hardening on mjd 52725 .
apart from the different maximum fluxes in the two outbursts , another remarkable property is the varying flux level corresponding to transitions between the states .
the hard - to - soft transition of the second outburst occured at the level higher by a factor of @xmath13 ( similar to the ratio between the maximum 1.512 kev fluxes of the two outbursts ) than that of the first one . in both cases , this level was higher than that of the soft - to - hard transition , which happened at the very similar flux levels for both outbutsts . the lagging of the effect in a physical system when the acting force is changed is called hysteresis . in the present case , the acting force is apparently the accretion rate and the effect is the spectral state .
the presence of hysteresis in gx 3394 has been found before by , e.g. , miyamoto et al .
( 1995 ) , smith et al .
( 2002b ) , nowak , wilms & dove ( 2002 ) , and by , e.g. , maccarone & coppi ( 2003 ) and rossi et al .
( 2004 ) in some other black - hole and neutron - star binaries .
we see in fig .
[ outbursts ] that the 312 and 316 kev fluxes change only slightly but the 1.55 kev one changes by a large factor during each state transition .
this implies a transition pivot at @xmath110 kev , similarly to cyg x-1 ( zhang et al .
1997 ) .
another interesting effect is seen during the soft - to - hard transitions .
namely , the hardening of the 312 kev spectrum is accompanied by an initial softening of the 1.55 kev one during the first outburst , see figs .
[ outbursts](a ) .
such softening is also seen in the 36.4 kev spectrum near the end of the second soft state , see fig .
[ outbursts](d ) .
this effect likely corresponds to disappearance of inner parts of the optically - thick disc when its luminosity declines at the end of a soft state .
then , at the beginning of a hard state , a strong soft excess is still present in the spectrum .
fig . [ asm_index ] presents correlations in the asm data , including those shown in fig .
[ outbursts](a b ) .
we plot the indices decreasing along the axes to keep a correspondence to hardness ratio diagrams . in the 1.55 kev range ,
the @xmath72-@xmath58 region occupied by the data forms roughly a circle , see fig .
[ asm_index](a ) .
evolution is clockwise , see fig .
[ asm_index](b ) .
the second outburst is extended to fluxes higher than the first .
interestingly , both the hard and soft state ( defined based on the 312 kev spectrum ) occupy similar ranges of the 1.55 kev spectral indices and differ mostly in their fluxes .
this situation reverses in the 312 kev range , see figs .
[ asm_index](c ) , ( d ) .
the hard state occupies an almost horizontal strip for @xmath892 clearly separated from the soft state but the range of @xmath72 is similar in both states , only slightly shifted to higher values for the soft state .
in fact , the first soft state ( red symbols ) shows fluxes typically lower than the hard state preceding the second soft state .
there is at most slight @xmath90 correlation in the hard state , with the best fit slope of @xmath91 .
we note , however , that our @xmath92 does not separate the compton reflection component , which strength positively correlates with the intrinsic slope ( ueda , ebisawa & done 1994 ; revnivtsev , gilfanov & churazov 2001 ) .
thus , the range of variability of the intrinsic @xmath58 will be slightly larger than that shown in fig .
[ asm_index](c ) .
[ asm_index](d ) shows clearly the presence of hysteresis , with the transitions to the soft state occuring at much larger fluxes than those of out that state .
a part of the horizontal extent covered in each state is due to this effect , in which the spectrum remains relatively stable within a large range of the flux .
we also see a separation between the regions corresponding to the two outbursts , consistent with the difference between their flux scale , fig .
[ outbursts ] . in both 1.55 and 312 kev bands ,
the spectral index is anticorrelated with the flux in the soft states , and the anticorrelation is much stronger during the second outburst ( black symbols ) .
this effect is probably caused by the spectral shape varying due to the variable amplitude of the high energy tail beyond the blackbody component ( as it is the case in cyg x-1 , z02 ) .
for many points , especially during the first outburst , the 312 kev power law is very soft , up to @xmath93 , i.e. , the emission in the tail becomes very weak .
[ asm_index](e ) , ( f ) show the corresponding colour - colour diagram .
the two indices are positively correlated in the soft state , which simply means that softening at lower energies is accompanied by softening at higher energies . the dotted line in fig .
[ asm_index](e ) corresponds to the two indices being equal .
thus , we see that in the soft state the 312 kev spectrum is in most cases softer than the 1.55 kev one .
this softening at high energies probably corresponds to both the cutoff in the disc blackbody entering the range of energies @xmath33 kev and absorption .
our strongest hard - state ( anti)correlation involves the 1.55 and 312 kev spectral indices , see fig .
[ asm_index](e ) .
we see that the harder the 312 kev spectrum , the softer the 1.55 kev one , i.e. , the stronger the soft x - ray excess .
the slope of this correlation is @xmath94 ( obtained by fitting including errors in both variables , e.g. , press et al .
1992 ) , and the spearman correlation coefficient is @xmath95 , corresponding to the probability of occuring by chance of @xmath96 . the part of the hard - state data to the right of the dotted line appears to correspond to the hardening of the lower - energy spectrum due to absorption . on the other hand
, there is a soft x - ray excess to the left of the dotted line , strong enough to be seen in spite of the absorption . as seen in fig .
[ asm_index](f ) , the left - most part of the hard state corresponds to the return from the soft state , and the soft excess here is probably a residual from that state .
still , similar soft excesses are also clearly present in the hard - state data ( blue symbols ) before the first soft state .
we discuss a theoretical explanation of this correlation in section [ hard ] .
overall , this diagram strongly resembles those for atoll neutron - star binaries , e.g. , aql x-1 ( fig . 4 in dg03 ) , but
_ not _ the colour - colour diagrams for black - hole binaries , which are usually quite diagonal ( fig . 2 in dg03 ) , which is also the case for the corresponding asm plot for cyg x-1 ( fig . 5 in z02 ) . however , an important difference between colour - colour diagrams based on the asm and pca ( used by dg03 ) data is the energy range .
the asm covers energy @xmath971.5 kev , whereas the coverage of the pca is for @xmath973 kev .
in particular , the colours used by dg03 are for 36.4 and 6.416 kev , i.e. , at energies above the lowest asm channel , 1.53 kev .
the hard - state soft excess in gx 3394 is clearly constrained to energies @xmath63 kev . fig .
[ pca_index ] shows diagrams corresponding to those in fig .
[ asm_index](c ) , ( e ) but based on the pca / hexte , see fig .
[ outbursts](c d ) .
we see in fig .
[ pca_index](c ) that the hard state colour - colour diagram for gx 3394 is approximately diagonal , like for other black - hole binaries ( dg03 ) .
given that gx 3394 is definitely a black - hole system ( hynes et al .
2003 ) , the difference between the hard state of atoll and black - hole binaries in their colour - colour diagrams at energies @xmath973 kev ( with the former and latter forming horizontal and diagonal regions ) can not be attributed solely to the cooling action of the neutron star surface , which was proposed by dg03 . instead
, the difference is , at least partly , related to the different blackbody temperature , @xmath98 , in the two classes of sources .
the upper horizontal branch appears due to the disc blackbody emission affecting the colours , which obviously depends on the maximum @xmath98 .
this temperature is significantly higher in neutron - star systems due to their lower masses , and thus the effect of blackbody emission is seen above 3 kev , whereas it is not seen in black - hole systems with a lower temperature .
however , the upper horizontal branch , related to the hard - state blackbody , still appears in our black - hole system when low enough energies are studied .
thus , black - hole and neutron - star binaries are more similar than the results dg03 would suggest .
another interesting property is that whereas fig .
[ asm_index](c ) shows a softening from the lower to the higher energies for virtually all soft - state data , the soft state in the corresponding fig . [ pca_index](c ) shows either hardening or softening from energies below 6.4 kev to those above it .
this difference is also related to the different energy ranges in the two diagrams . in the asm range , we see mostly the effect of the high - energy cutoff of the disc blackbody , whereas in the pca range , we see the variable high - energy tail ( which can be quite hard ) beyond this cutoff . figs .
[ pca_index](a ) , ( b ) show the two pca indices as a function of the model bolometric flux .
in the hard state , both indices are almost constant at high fluxes , and appear to soften at low fluxes ( @xmath60.1 kev @xmath60 s@xmath39 ) .
although the soft state extends to the highest fluxes , we see many soft - state points corresponding to the bolometric fluxes _ below _ the maximum fluxe of the hard state .
this clearly shows the presence of the hysteresis , now based on the broad - band pca / hexte data .
[ pca_index ] includes the data for the beginning of the 2004 outburst , showing it to be in the standard hard state .
[ ginga_index ] shows the colour - flux diagrams based on the _
ginga_/asm analogous for those for the _ rxte_/asm , figs .
[ asm_index](a ) , ( c ) .
we see virtually identical dependences , showing that no significant change has occured in gx 3394 in about a decade separating those observations .
[ batse_index ] shows the relationships between the effective spectral indices in three bands vs. fluxes for the batse data in the hard state covered by the _ rxte_/asm ( blue ) and for earlier epochs ( red ) .
most of the symbols also correspond to the hard state , as the soft state is hardly detected by the batse .
those diagrams represent continuation of those for the _ rxte_/asm , figs .
[ asm_index](a ) , ( c ) , towards higher energies . in the 312 kev band , @xmath58 was almost independent of @xmath99 , with the best fit slope of @xmath91 .
now , we see clear @xmath72-@xmath58 correlations from 20 to 160 kev , with its strength increasing with energy .
linear @xmath100 fits to all the hard - state points give the slopes of @xmath101 and @xmath102 for the 2070 , 40160 kev bands , respectively , see figs .
[ batse_index](a ) , ( b ) .
the corresponding spearman correlation coefficients are 0.39 and 0.62 , with the chance probabilities of the correlation of @xmath103 and @xmath104 , respectively .
thus , it is very unlikely that those correlations are due to a chance occurence , and the strength of the correlation given by the slope increases with the photon energy .
( note that the strength of the correlations is due to the batse data preceding the asm monitoring , when the strongest fluxes were recorded . )
this result means that the softening of the spectrum with increasing flux becomes progressively stronger with increasing energy .
it indicates the high energy cutoff in the spectrum decreasing with the increasing flux , as also found by w02 based on broad - band pointed observations .
we discuss a theoretical explanation of this result in section [ hard ] .
the dependence of @xmath105 on the flux is more complex .
there is now a change of the sign of the correlation in the middle of the 70430 kev energy band , which , as we have checked , does not appear in the previous two cases .
namely , there is still very strong positive correlation of this index on the 70160-kev flux , with the slope of @xmath106 , and the spearman coefficient of 0.54 corresponding to the chance probability of @xmath107 , see fig .
[ batse_index](c ) .
the sign of the correlation , however , reverses for the dependence on the 160430 kev flux , as seen in fig .
[ batse_index](d ) , with the best - fit slope of @xmath108 .
these effects are seen for both the batse data simultaneous with the _ rxte_/asm and the earlier ones .
this reversal of the correlation sign shows the hard - state variability can be described by a variable slope at high energies with the pivot point at @xmath1200 kev .
note that the pivoting variability also contributes to a part of the softening of the 20160 kev slopes with increasing flux .
however , this softening can not be solely due to pivoting as its degree increases with the photon energy , opposite to the pivoting , where the @xmath72-@xmath58 correlation becomes stronger in the opposite direction , i.e. , away from the pivot energy .
we then considered the joint asm / batse simultaneous data , dominated by the hard state .
we have found that any of the three asm fluxes is well linearly correlated with any of the batse fluxes in the bands of 2040 , 4070 and 70160 kev energy bands .
however , there is no clear correlation of the asm fluxes with the 160430 kev band .
this is consistent with fig .
[ batse_index](d ) , showing a reversal of the positive flux - index correlation above 160 kev .
additional information about the physical states of the source is provided by the fractional rms variability as a function of energy .
we calculate it separately for the hard state covered jointly by the _ rxte_/asm and batse , i.e. , until the first soft state , and for each of the two soft states using only the asm data . in general
, there are several ways to compute the rms , e.g. , using data rebinned up to a given significance .
this , however , depends on the chosen significance level .
thus , we have decided to show results obtained using the raw 1-day average data .
then , we have calculated it for both weighted ( see appendix ) and unweighted ( edelson et al .
2002 ) averages . in fig .
[ rms ] , we show results for the unweighted averages . however , this method yields results similar to those using rebinned data and/or weighted averages , even if the values of the rms are somewhat different .
for example , the hard - state data yield the weighted unrebinned fractional rms values of 0.68 , 060 , 0.53 , 0.67 , 0.52 and 0.42 in the 6 consecutive channels of fig .
[ rms ] . during the hard state ,
the fractional rms of the asm data decreases with energy .
then , there is a slight increase of the rms from the last asm channel to the first batse one .
this may be due to the presence of variable compton reflection ( e.g. , revnivtsev et al .
2001 ) , which spectral component peaks at @xmath130 kev ( e.g. , magdziarz & zdziarski 1995 ) .
still , the trend of the rms decreasing with energy continues within the batse data , consistent with the pivot at @xmath109 kev ( see section [ highe ] ) .
we do not show the point for the 160430 kev channel because of its large error .
on the other hand , the rms increases with @xmath110 for either of the soft states .
interestingly , it is much higher for the second outburst , correlated with its substantially higher flux .
the increase of the rms with energy in both cases can be interpreted as the high - energy tail being more variable than the blackbody component , as also found for cyg x-1 ( z02 ) .
we have studied periodograms ( lomb 1976 ; scargle 1982 ; as implemented in press et al .
1992 ) of the _ rxte_/asm ( using the dwell data ) , batse and _ ginga_/asm data .
we find the most significant peaks in the periodograms at 215 and 227 d in the first two data sets limited to the hard state ( mjd @xmath111 ) , and at 230 d for the _ ginga _ data , see fig .
[ period ] , confirming the results of kong et al .
( we note that their _
rxte_/asm peak power was much lower due to the use of the daily - averaged data ; also , we refer the reader to that paper a thorough discussion of the sigificance levels . )
we see that within their widths , all the peaks are compatible with being identical , @xmath112 d. kong et al .
( 2002 ) have attributed this period to the repetition time scale of the outbursts . however , visual inspection of the lightcurve , fig .
[ lc_all ] , does not seem to confirm it .
we see , e.g. , that the first three batse outbursts were separated by @xmath1450 d , which apparently corresponds to the broad peak at 433 d in the batse periodogram . to test it further ,
we have made the periodogram for the batse data constrained to those three outbursts , mjd @xmath113 , and obtained results virtually identical to those in fig .
[ period ] .
thus , we consider it likely that the 220 d period corresponds to the precession of the accretion disc in the system .
this issue certainly deserves further study , which , however , is beyond the scope of this paper . on the other hand
, no @xmath1220 d peak appears in either the entire _ rxte_/asm data set or the data constrained to any of the two recent soft outbursts .
thus , it appears that the first soft altered the accretion flow so significantly that the precession pattern changed .
we have also searched for modulation corresponding to the orbital period , @xmath114 d. we have not found any in the asm data for either the hard ( fig . [ period ] ) or the soft states . on the other hand , there is a clear peak at @xmath115 d , which is also present in the asm data for other sources , e.g. , cyg x-1 ( lachowicz et al .
, in preparation ) .
it is an artefact related to the scheduling cycle of the _ rxte _ observations , in particular many pointings with a 1-day periodicity , e.g. , during numerous agn monitoring campaigns .
coupling of this periodicity with long - term secular changes introduces a powerful signal at @xmath11 d.
as our results above show , there are many important similaraties and differences between gx 3394 and cyg x-1 .
\(i ) cyg x-1 shows no hysteresis .
state transitions in that source follow the same trajectory in each direction on index - flux and index - index diagrams ( z02 ) . the presence of hysteresis in gx 3394 is certainly related to its overall range of variability ( and presumably the accretion rate ) much wider than in cyg x-1 , both in terms of the fluxes and spectral indices .
\(ii ) the 312 kev spectral index in the hard state correlates with the corresponding flux much less in gx 3394 ( fig .
[ asm_index]c ) , than in cyg x-1 ( fig .
4b in z02 ) .
however , the range of this @xmath58 in the hard state is almost the same in both source , @xmath11.22.0 , whereas the range of the flux is @xmath9710 times higher in gx 3394 .
this then explains the weaker dependence .
the correlation in cyg x-1 was interpreted by z02 as due to variable source geometry in the hard state , with a variable relative rate of cooling of the comptonizing plasma .
such variable geometry is also likely to be present in gx 3394 , governing its spectral hardness , but with an additional hysteretic dispersion of the hard - state flux .
\(iii ) the 312 kev and 1.55 kev indices in the hard state are anticorrelated in gx 3394 ( resembing some atoll sources ) , whereas they are strongly positively correlated in cyg x-1 ( fig . 5 in z02 ) .
this differerence is likely to be related to the @xmath98 in the hard state being much higher in gx 3394 than in cyg x-1 . indeed , corongiu et al .
( 2003 ) obtain @xmath116 kev in the hard state of gx 3394 , compared to @xmath117 kev in cyg x-1 ( di salvo et al .
2001 ; frontera et al .
2001 ) kev in gx 3394 based on _ asca _ data . however , those data cover a much narrower energy band than that of _ bepposax _ , as well as they used used a model with a broken power law , whereas corongiu et al .
( 2003 ) found only a single power law necessary . ] .
then , the variability pattern of the blackbody emission ( see section [ hard ] ) affects the asm band much more in gx 3394 than in cyg x-1 .
\(iv ) the hard - state fluxes in bands from 20 to 160 kev are positively correlated with the local spectral indices in gx 3394 .
the sign of the correlation changes above 160 kev , indicating a pivot around this energy . on the other hand ,
the hard - state pivot in cyg x-1 is at a lower energy , @xmath150 kev ( z02 ) . as discussed by zdziarski et al .
( 2003 ) , a pivot at energies close to the high - energy cutoff ( or higher ) indicates the behaviour in which an increase of the power in soft photons providing seeds for thermal comptonization is accompanied by an increase of the power supplied to the hot comptonizing plasma . in the case of cyg x-1 , that power is close to constant in the hard state on long time scales ( unlike gx 3394 ) .
we also note that cyg x-1 shows a pivot at @xmath1120 kev similar to that of gx 3394 when studied on hour time scales ( bazzano et al .
2003 ) .
\(v ) consistent with the pivot at high energies , the hard - state batse data show the rms decreasing with energy , unlike cyg x-1 , where it has a minimum and then increases again above 100 kev ( z02 ) .
\(vi ) in the soft state , the flux - index anticorrelation observed in cyg x-1 was explained by z02 as a result of constant luminosity in the soft component and varying luminosity of the hard , power - law spectrum .
the stronger the power - law , the less visible the cutoff in the soft component and thus the harder spectrum . in gx
3394 , this behaviour is also seen in the two soft outbursts ( fig [ asm_index ] ) . however , the blackbody flux is also variable in gx 3394 ( fig . [ outbursts ] ) .
\(vii ) the dependence of the rms on energy in the first soft outburst is similar in gx 3394 and in cyg x-1 during its 1996 soft state ( z02 ) . on the other hand ,
the second soft state shows the daily time scale rms higher than that of cyg x-1 in any of its soft states .
\(viii ) no modulation of x - rays with the orbital period is present in gx 3394 in any state , compared to the strong modulation of soft x - rays in the hard state of cyg x-1 ( e.g. , wen et al . 1999 ) .
this is again likely to be related to the different nature of the accretion flow in the two objects .
no x - ray eclipses or dips in gx 3394 is due to inflow of matter being contained in the roche - lobe overflow , unlike the case of the wind accretion in cyg x-1 .
our results show that luminosity changes in gx 3394 are not simultaneous with changes of the spectral shape .
obviously , a physical model to describe this behaviour is desired .
the data with best time coverage we have are from the _ rxte_/asm instrument , which , however , provides us only with information in three channels in the limited energy band of 1.512 kev . given that , our theoretical model should be rather simple .
thus , we have developed a model of an accretion flow composed of an inner , hot , optically thin disc and an outer , cold , optically thick one with only two free parameters .
the first of them is the accretion rate ( assumed constant throughout the flow ) , which also represents the normalization of the spectrum , and the second , the transition radius between the hot and cold disc .
the dimensionless accretion rate is defined in the standard way ( e.g. , shakura & sunyaev 1973 ) as @xmath118 where @xmath119 is the accretion rate and @xmath120 is the eddington luminosity , @xmath121 then @xmath122 , where @xmath123 is an accretion efficiency .
we assume that an optically thick disc of shakura & sunyaev ( 1973 ) extends down to a radius @xmath124 , expressed in the units of gravitational radius , @xmath125 .
the luminosity of the disc ( in soft blackbody radiation ) is @xmath126 below @xmath124 , we assume the flow becomes optically thin .
this is intended to approximate optically thin accretion , e.g. , cooling - dominated hot flows ( shapiro , lightman & eardley 1976 ) , adaf ( e.g. , narayan & yi 1996 ) or cdaf ( e.g. , narayan , igumenshchev & abramowicz 2000 ) .
we assume the luminosity of the optically thin flow is @xmath127 i.e. , take the radiative efficiency of the hot disc equal to that of the cold one assuming no dissipation below @xmath128 .
this yields the accretion efficiency of @xmath129 . the spectrum of the optically thick , outer
, disc is approximated as a blackbody with the total luminosity of @xmath130 .
the colour temperature , @xmath98 , is calculated from the temperature of the shakura - sunyaev disc at @xmath124 , with the inner zero - torque boundary condition neglected for simplicity , @xmath131 where the colour correction of @xmath132 is assumed ( shimura & takahara 1995 ) .
since the inclination of gx 3394 is unknown , we consider here the angle - averaged emission . for the hot disc spectrum we assume an exponentially cut off power law , which approximates the spectrum produced by unsaturated comptonization of photons emitted by the disc .
the power - law photon index is , @xmath133 ( beloborodov 1999 ) where @xmath134 depends on the temperature of seed photons , with @xmath135 for stellar - mass black holes , and where we have assumed a half of the blackbody photons is comptonized in the hot flow ( and the other half escapes unscattered ) .
the normalization of the power - law spectrum at the energy of the peak of the blackbody spectrum is calculated using eq .
( 7 ) in zdziarski ( 1986 ) .
the energy of the cutoff in the power - law spectrum ( depending on the electron temperature ) is calculated by the normalization of that spectrum given by @xmath136 .
we note that our model allows for any @xmath124 ( @xmath137 ) for a given @xmath138 .
thus , it allows for either soft or hard state to take place at a given accretion rate , consistent with our observational result of hysteresis in gx 3394 .
we have implemented the above model into the xspec package ( arnaud 1996 ) .
the _ rxte_/asm fluxes ( in physical units and rebinned to @xmath73 significance , see section [ data ] ) were converted to fits - format files , assuming a diagonal response matrix .
the above model was then fitted to the data assuming @xmath53 kpc , @xmath139 , and @xmath140 @xmath60 .
the fit quality corresponded to the typical @xmath141 ( although a higher @xmath142 was found for a number of days ) .
the results the fits with the above model are shown in fig .
[ asm_model ] , in which we plot @xmath143 ( @xmath144 ) instead of @xmath138 . the maximum model @xmath143 ( reached during the soft state ) are then 0.10 and 0.30 erg @xmath60 s@xmath39 , which is 2.4 times our asm estimate .
that overestimate is fully explained by the incorrect application of the asm bolometric correction obtained for the hard state to the soft state ( maccarone , private communication ) . ] ( for the assumed @xmath7 and @xmath145 ) for the first and second outbursts , respectively .
the corresponding global maximum of the daily asm countrate was @xmath146 s@xmath39 on mjd 52472 , and the maximum asm dwell countrate was @xmath147 s@xmath39 on mjd 52472.9 , i.e. , only marginally higher than the daily one .
the maximum @xmath143 in the hard state of 0.26 was achieved in the second outburst just before the state transition .
the transition @xmath143 corresponding to the transition to and out of the soft state were @xmath10.07 , @xmath10.015 , and @xmath10.25 , @xmath10.010.025 , for the first and second outburst , respectively .
these asm results can be compared to our estimated bolometric luminosities from the pca / hexte .
the highest ( model ) @xmath143 achieved during the first outburst ( at mjd 5086550874 , see figs .
[ lc_all ] , [ outbursts ] ) is 0.12 ( @xmath148 erg @xmath60 s@xmath39 ) , in good agreement with our model result ( above ) of 0.10 .
then , the @xmath143 some time before ( mjd 51201 ) and shortly after the soft - to - hard state transition ( mjd 51222 ) is 0.034 and 0.017 ( @xmath149 , @xmath150 erg @xmath60 s@xmath39 ) , respectively , in agreement with our asm model results . during the second outburst ,
the pca / hexte data show the maximum in the hard state before the transition on mjd 5238790 of @xmath151 erg @xmath60 s@xmath39 , @xmath152 , compared to the corresponding asm model value of @xmath153 .
then , the @xmath143 during observations before [ mjd 52400 , @xmath154@xmath155 and after the hard - to - soft transition [ mjd 52402 , @xmath154@xmath156 ) are 0.14 and 0.21 ( @xmath157 , @xmath158 erg @xmath60 ) . on the other hand , the same values for observations just before [ mjd 52727 , @xmath154@xmath159 and after the soft - to - hard state transition [ mjd 52734 , @xmath154@xmath160 ) are 0.031 and 0.020 ( @xmath161 , @xmath162 erg @xmath60 ) , with the transition itself occuring on mjd 52731 [ @xmath154@xmath156 ] with @xmath164 erg @xmath60 , @xmath165 . the pca / hexte global maximum was on mjd 52483 with @xmath166 erg @xmath60 s@xmath39 , @xmath167 , in excellent agreement with our asm model maximum of 0.30 on mjd 52472 .
thus , the estimates of our asm model , given all its simplifications , are in good agreement with the available broad - band data .
the good agreement in the hard state also indicates the proper estimate of the high - energy cutoff energy in the asm model .
thus , the soft - to - hard state transitions in gx 3394 occur just around the average value of @xmath168 found for black - hole binaries by maccarone ( 2003 ) provided @xmath169 kpc .
overall , @xmath170 follows changes in the observed 1.512 kev flux , but different values of @xmath124 are possible for the same @xmath170 .
the variations of @xmath124 follow those of @xmath154@xmath171 , which due to the ratio @xmath172 being determined by @xmath173 in our model .
the inner disc radius assumes values of @xmath110100 and @xmath1746 in the hard and soft state , respectively .
the average accretion rate in gx 3394 is low enough for the accretion to be unstable , similarly to standard x - ray novae ( e.g. , mineshige 1996 ) , but with a difference of much shorter recurrence time scales .
however , the long - term behaviour , see fig . [ lc_all ] , does not consist of regularly repeating outbursts , which suggests variability of the long - term average @xmath119 .
the batse lightcurve appears consistent with an increase of that @xmath119 from an epoch of strong outbursts every @xmath1500 d separated by periods of a very low flux ( asai et al .
1998 ) to the source becoming more persistent , with weaker peaks and brighter minima . to check it ,
we have calculated the average batse flux in the first of the two epochs , during mjd 4840049500 , which is @xmath175 erg @xmath60 s@xmath39 . based on the pca data
, we can estimate the bolometric correction as a factor @xmath1761.5 .
a further correction is due to soft states appearing after the hard outbursts .
based on the _
ginga_/asm data , the duration of soft states at that epoch was relatively short , and we estimate it as @xmath177 .
thus , our best estimate of the average flux is @xmath178 erg @xmath60 s@xmath39 , which corresponds to @xmath179 . due to systematic errors ( e.g. , in the absolute calibration of the batse )
, the numerical coefficient above may be somewhat lower , @xmath10.01 .
this @xmath180 clearly corresponds to accretion in this source being strongly unstable .
the average batse flux during the next epoch , mjd 4950050800 , characterized by weaker and more often hard otbursts with higher minimum fluxes , had then the average flux of about 1.6 times the preceding one , corresponding to @xmath181 . a further increase lead to the first of the two recent soft states .
we estimated the bolometric correction based on the pca data , which yielded @xmath182 erg @xmath60 s@xmath39 during its duration .
this soft outburst apparently emptied the inner region of matter , and a @xmath11000 d off state followed .
the accumulated matter powered the last strong outburst , with @xmath183 of its soft state of about twice the preceding one , and a @xmath1300 d duration .
averaging from the beginning of the first outburst ( mjd 50800 ) until now mjd 53000 yields @xmath184 erg @xmath60 s@xmath39 , corresponding to @xmath185 , i.e. , about three and two times that during the preceding epochs , mjd 4840049500 and 4950050800 , respectively .
this indicates variability of the accretion rate on the time scale of @xmath88 d. interestingly , the average luminosity , @xmath183 , during the initial outbursts observed by batse , @xmath1(0.010.015)@xmath186 , is very similar to that of the hard state in a number of persistent sources , e.g. , cyg x-1 , where @xmath187 during 199194 ( z02 ) , but also in 1e 1740.72942 and grs 1758258 ( heindl & smith 1998 ) . in the case of cyg x-1 , its stability as opposed to the instability of gx 3394 may be related to the different sizes of accretion discs in those systems .
the instabilities in gx 3394 certainly develop in the outer disc , whereas the accretion disc in cyg x-1 may form out of the quasispherical wind relatively close to the black hole ( beloborodov & illarionov 2001 ; illarionov & beloborodov 2001 ) . the hysteresis in gx 3394 is obviously related to the existence of a range of @xmath170 at which two stable accretion solutions , optically thick and optically thin , exist .
the state of the source in a given moment is then determined not solely by @xmath170 but also by the past history .
the maximum possible luminosity of the hard state is larger than the minimum possible luminosity of the soft state .
we could then expect the transitions to / out of the soft state to appear simply at the limiting luminosities .
the transition out of the soft state did take place in a narrow range of @xmath170 during the last two outbursts .
such a uniformity is also characteristic to other sources ( maccarone 2003 ) .
this luminosity is likely to represent the minimum possible soft state @xmath170 , limited probably by evaporation ( e.g. , raska & czerny 2000 ; meyer , liu & meyer - hofmeister 2000 ) . in this case , the past history determining the source behaviour is simply the presence of an optically - thick disc .
however , it is less clear what determines the moment of collapse of the hard - state hot flow and its transformation into an optically - thick disc .
the luminosity at which this happens varied by a factor of @xmath13 between the two last outbursts of gx 3394 .
the cause of this difference appears to be the previous history of the source .
the transition to the first of the two soft states happened after several years of sustained hard - state activity . on the other hand ,
the second outburst followed a @xmath11000 d of quiescence .
one possibility is that the inner radius of the outer cold disc was much closer to the black hole when the @xmath119 started to increase leading to the first soft outburst .
before the second one , the cold disc was initially much further out , and its buildup towards lower radii did not keep pace with the rapidly increasing local accretion rate .
this , in turn , allowed a much higher luminosity in the hard state .
thus , the past history in the case of the hard - to - soft transitions appears to be not just existence a hot flow but also the state of the surrounding cold disc .
a way of testing the above hypothesis would be studying the evolution of the solid angle covered by the disc by measuring the strength of compton reflection in the source .
indeed , an increasing strength of reflection with decreasing @xmath124 is expected from consideration of energy balance ( zdziarski , lubiski & smith 1999 ) , and consistent with the observed associated increase of the degree of relativistic broading of fe k spectral features and the qpo frequency ( revnivstev et al .
nowak et al . ( 2002 ) noted also a correlation of the reflection strength with the x - ray flux before the first soft outburst .
we would then expect different time histories of compton reflection and the flux in the case of the hard state preceding the first and the second of the soft outbursts .
we have found several interesting aspects of the variability of gx 3394 during the hard state .
first of all , the strongest correlation in the _
rxte_/asm data is a negative one between the 1.55 kev and 312 kev average slopes , forming an upper horizontal branch in the colour - colour diagram , fig .
[ asm_index](e ) .
we find it can be very well explained in the framework of thermal comptonization .
as discussed , e.g. , by z02 , the main factor determining the 312 kev slope is the ratio between the power in soft photons incident on the comptonizing plasma to the power supplied to that plasma , likely determined by a variable source geometry .
a hardening of that slope corresponds to a reduction of that ratio .
however , only a geometry - dependent fraction of the emitted soft flux is incident on the hot flow .
the main variability pattern may consist of changing this fraction rather than the total soft flux .
then , a reduction of the incident fraction would both harden the comptonized spectrum and decrease the point at which that spectrum intersects with the spectrum of the soft photons , as illustrated in fig .
[ patterns](a ) .
the change may be caused by a variable overlap of the cold disc by the hot flow ( see fig . 15 in z02 ) . in order to model this effect quantitatively ,
we have used a comptonization model of coppi ( 1999 ) ( described in detail in gierliski et al .
1999 ) , in which the plasma electron temperature is determined self - consistently from energy balance between the powers in the hot plasma and the incident seed photons
. the x - ray slope is determined by @xmath188 and the thomson optical depth , @xmath189 .
we have assumed that there is a constant total power in the soft blackbody spectrum , but its fraction irradiating the hot plasma is variable , as discussed above .
we assumed the soft photon spectrum is a disc blackbody with the maximum temperature of @xmath190 kev ( corongiu et al .
2003 ) , and @xmath191 ( w02 ) . at one extremum , all the blackbody photons irradiate the plasma ( i.e.
, the cold disc is embedded in the hot flow ) , which power is assumed to be 5 times that in the blackbody photons .
this spectrum , absorbed by @xmath140 @xmath60 , is shown in fig .
[ patterns ] by the upper solid curve . based on it
, we calculate the fluxes in the asm energy channels and the effective spectral indices , obtaining @xmath64@xmath192 and @xmath154@xmath193 .
we now reduce the fraction of the blackbody photons incident on the hot plasma by a factor of 10 ( e.g. , due to a radial separation between the cold disc and hot flow ) .
this strongly reduces the cooling rate , leading to both a spectral hardening of the power - law emission ( dotted curve ) and the appearance of the unscattered part of the blackbody ( dashed curve ) .
the resulting total spectrum ( the lower solid curve in fig . [ patterns ] ) has @xmath64@xmath194 and @xmath154@xmath195 .
the two points corresponding to those spectral solutions are joint by a heavy black line in fig .
[ asm_index](e ) .
we see this effect leads to a correlation with a slope identical to the best fit to the data .
second , we have found the high - energy spectrum softens with the increasing luminosity . in the thermal comptonization model , which fits broad - band data for gx
3394 well ( z98 , w02 ) , the energy of the cutoff is directly related to @xmath188 . as in w02 , we consider a hot accretion model of zdziarski ( 1998 ) , parametrized by the compton parameter , @xmath196 where @xmath197 is the electron rest energy .
this parameter determines the slope , @xmath58 , which is weakly dependent on the flux in the data , see fig .
[ asm_index](c ) , indicating a constant @xmath198 .
then , @xmath199 and @xmath200 ( where @xmath170 is the luminosity ) in the advection and cooling dominated cases , respectively ( zdziarski 1998 ) .
this behaviour reflects more efficient cooling at higher accretion rates . in order to test those predictions ,
we first fit a spectrum from a pointed observation ( the _ ginga_/osse spectrum no . 1 of z98 ) by a model including the intrinsic continuum due to thermal comptonization .
we now use a comptonization model of poutanen & svensson ( 1996 ) , also used by w02 , in which , in contrast to the model of coppi ( 1999 ) , @xmath188 is a free parameter .
the resulting points are plotted by the rightmost asterisk in each of figs .
[ asm_index](c ) and [ batse_index](a d ) .
we then reduce the luminosity by a factor of 5 and increase @xmath188 according to the above two dependences .
we also consider the case with @xmath170 recuced by a factor of 10 , and @xmath199 .
the resulting three model points are shown by the three left - hand - side asterisks in each of figs .
[ asm_index](c ) , [ batse_index](a d ) .
we see in fig .
[ asm_index](c ) that the three points correspond indeed to almost the same x - ray slope as the original data .
we also see that this model roughly reproduces the index increasing with the flux above 40 kev , figs .
[ batse_index](b c ) .
it predicts an almost constant @xmath58 in the 2070 kev range , not reproducing a slight increasing trend in the data , fig .
[ batse_index](a ) .
however , this increase of @xmath58 is solely due to the batse data preceding the _ rxte_/asm monitoring , for which we do not know the behaviour in the 312 kev range .
thus , this model provides a likely explanation of the behaviour seen in the hard - state batse data .
third , we have found a pivot point around @xmath1200 kev , figs .
[ batse_index](c d ) , which is similar to the variability pattern found in a number of seyfert galaxies ( zdziarski et al .
this implies that though a major driver of the variability is at low energies due to the variable flux of seed photons for thermal comtonization , an increase / decrease of that flux is accompanied by a similar increase / decrease of the power in the hot comptonizing plasma .
[ patterns](b ) illustrates such a pattern , resulting in a pivot at 200 kev .
we used the same models as those of fig .
14a of z02 , but adjusted their normalization as above .
in that model , @xmath188 is self - consistently adjusted to varying irradiation by soft photons . both above patterns ,
the decreasing @xmath201 , modelled for a constant x - ray slope , and pivoting , which changes the slope , act together .
the pivoting may explain the softening of the 2070 kev slope with the increasing flux , not explained by the @xmath201 dependence .
it also explains the disagreement of the trend predicted by the @xmath201 model at energies above 160 kev , see the asterisks in fig .
[ batse_index](d ) , where pivoting dominates .
in fact , the variability pattern shown in fig . [ patterns](b ) includes softening of the spectrum with increasing luminosity , as the increase is accompanied by a decreasing @xmath188 ( see above and z02 ) .
altogether , thermal componization explains well both the long - term spectral variability of gx 3394 in its hard state and its broad - band x@xmath5-ray spectra ( z98 , w02 ) . on the other hand , synchrotron emission by nonthermal power law electrons in a jet
was proposed to be responsible for its hard - state x@xmath5-ray spectra ( markoff et al .
this model is motivated by the presence of very good correlations between the radio and x - ray fluxes in a few bands above 3 kev found in gx 3394 ( corbel et al .
however , the nonthermal synchrotron model for the x@xmath5-ray spectra suffers from many problems ( e.g. , zdziarski et al .
2003 ) , in particular it requires the high - energy cutoff in the electron distribution to be both very sharp and highly fine - tuned .
our present results on spectral correlation add to the problems of this model .
especially , we have shown that the hard x - ray power law responds to the variability in the soft x - ray excess , with an increase of the relative strength of the excess associated with a hardening of the power law . in the nonthermal model , the power law extends down to the turnover frequency in the infrared , and it is not clear which mechanism would cause such behaviour ( explained by thermal comptonization cooling in our model ) .
it is also unclear why the nonthermal power law should pivot around 200 kev . as discussed in zdziarski et al .
( 2003 ) , the likely origin of the radio - x - ray flux correlation is the formation of the radio jet out from energetic particles in the hot inner flow ( i.e. , the base of the jet being that flow ) , which also radiates x - rays .
in particular , heinz & sunyaev ( 2003 ) derived power law dependences between the jet radio flux and @xmath119 . on the other hand
, the accretion luminosity also follows a power law dependence on @xmath119 , with the exponent @xmath2021 ( e.g. , shapiro et al .
1976 ; narayan & yi 1996 ; zdziarski 1998 ) .
consequently , there will be a power law dependence between the jet radio flux and the bolometric accretion luminosity .
in order to test this theoretical prediction , we plot in fig .
[ xte_radio ] the dependence between the 8.6 ghz radio flux of corbel et al .
( 2003 ) and the bolometric flux estimated by us from the pca / hexte data .
we have used only the radio measurements for which same - day x - ray measurements exist . with the current calibration of the pca ( see section [ data ] ) , we also obtain an x - ray detection on mjd 51388 ( with the 316 kev flux of @xmath203 erg @xmath60 s@xmath39 ) , for which corbel et al .
( 2003 ) reported an upper limit .
we obtain indeed an excellent correlation for all the points with the radio measurements . a least - square fit symmetric in both fluxes yields @xmath204 where we have included systematic errors with a fractional value ( of 0.25 ) as implied by the condition of @xmath205 , and the units are as in fig .
[ xte_radio ] .
observationally , this systematic error is mostly due to systematic errors of calculating the bolometric @xmath170 ( see section [ data ] ) , and the observations being not fully simultaneous .
interestingly , we see that the correlation breaks down at lowest fluxes as implied by the radio upper limit ( which was not included in the fit ) .
we note that our explanation of the correlation of corbel et al .
( 2003 ) is different from that of markoff et al .
( 2003 ) , who attributed it to the x - ray flux and the optically - thin jet emission being the same power law of the nonthermal synchrotron . in our explanation
, we do not require the jet optically - thin emission and the x - rays to lie on the same power law .
we have found that the current optical / ir data imply that the distance to gx 3394 is @xmath0 kpc .
its most likely location is in the galactic bulge .
we also argue against the recently proposed distance of @xmath115 kpc , in which case gx 3394 would be a halo object .
then , we have performed a comprehensive analysis of 16 years of the long - term variability of gx 3394 .
we found the long - term @xmath183 from this transient source has increased about threefold since 1991 , corresponding to marked changes in the character of the outbursts .
the epoch mjd 4840049500 was characterized by a @xmath1500 d cycle of strong hard outbursts separated by deep quiescence and @xmath206 .
the epoch mjd 4950050800 was characterized by a persistent hard state and @xmath183 about 1.5 times the above value .
this increase lead then to a long soft state , a @xmath11000 d queiescence , and the most recent @xmath1400 d very strong outburst . during this epoch ,
mjd 5080053000 , @xmath207 .
we have studied in detail hysteresis in this source .
especially interesting is the strong dependence of the flux of the hard - to - soft transition on the preceding behaviour of the hard / off state .
the 1998 transition , which followed the long persistent hard - state period , took place at the flux about three times lower than the 2002 one , which followed the long queiescent epoch .
we model the hysteretic variations by independent variability of the accretion rate and the inner radius of the cold disc .
the maximum possible @xmath119 in the hard state is higher than the minimum possible one in the soft state , but transitions from the hard to soft state can occur below that maximum .
we have found a new x - ray spectral correlation in gx 3394 , between the relative strength of the soft excess ( due to blackbody disc emision ) in the hard state and the hardness of the main power law component .
the effect is explained by cooling of the hot comptonizing plasma by a variable fraction of the disc emission , probably related to a variable overlap of the disc by the hot flow .
interestingly , the correlation forms an upper horizontal branch in the colour - colour diagram , previously thought to be specific to neutron - star binaries .
in contrast to cyg x-1 , we do not find any orbital modulation of the x - ray flux .
we have also found a pivoting pattern of hard - state x - ray variability , with the pivot energy at @xmath1200 kev , and confirmed the existence of an anticorrelation between the bolometric flux and the high - energy cutoff , presumably related to the temperature of the comptonizing electrons .
the weak correlation of the x - ray slope and the flux ( in contrast to the case in cyg x-1 ) is explained by the wide range of the fluxes at which the the hard state appears , related to the transient nature of gx 3394 .
our results provide strong support to the thermal - comptonization nature of x - rays in the hard state .
we explain that the correlation between the x - ray and radio fluxes by the total power of the accretion emission being correlated with the power in outflowing particles forming the radio jet , which is , in turn , correlated with the radio flux .
this research has been supported by the kbn grants 5p03d00821 , 2p03c00619p1,2 , pbz - kbn-054/p03/2001 .
we thank t. belloni , c. done , p. grandi , p. lachowicz , p. lubiski , t. maccarone and p. ycki for valuable and inspiring discussions , and the referee for valuable comments .
we also thank s. corbel for the radio data and acknowledge the use of data obtained through the heasarc online service provided by nasa / gsfc , and , in particular , thank the _
rxte_/asm team for their data .
arnaud k. a. , 1996 , in jacoby g. h. , barnes j. , eds . , astronomical data analysis software and systems v , asp conf .
series vol . 101 , san francisco , p. 17
asai k. , dotani t. , hoshi r. , tanaka y. , robinson c. r. , terada k. , 1998 , pasj , 50 , 611 bazzano a. , et al . , 2003 , a&a , 411 , l389 belloni t. , homan j. , cui w. , swank j. , 2004 , atel , 236 beloborodov a. m. , 1999 , in poutanen j. & svensson r. , eds . , high energy processes in accreting black holes .
asp conf .
vol . 161 , san francisco , asp , p. 295
beloborodov a. m. , illarionov a. f. , 2001 , mnras , 323 , 167 bevington p. r. , robinson k. d. , 1992 .
data reduction and error analysis for the physical sciences , 2nd ed .
, new york , mcgraw - hill brandt s. , 1997 .
statistical and computational methods in data analysis , 3rd ed . , springer , new york buxton m. , bailyn c. , 2004 , in kaaret p .. lamb f. k. & swank j. h. , x - ray timing 2003 : rossi and beyond , melville , ny , aip , in press ( astro - ph/0312392 ) coppi p. s. , 1999 , in poutanen j. & svensson r. , eds . , high energy processes in accreting black holes , asp conf .
161 , san francisco , asp , p. 375
corbel s. , nowak m. a. , fender r. p. , tzioumis a. k. , markoff s. , 2003 , a&a , 400 , 1007 corongiu a. , chiapetti l. , haardt f. , treves a. , colpi m. , belloni t. , 2003 , a&a , 408 , 347 di salvo t. , done c. , ycki p. t. , burderi l. , robba n. r. , 2001 , apj , 547 , 1024 done c. , gierliski m. , 2003 , mnras , 342 , 1041 ( dg03 ) edelson r. , turner t. j. , pounds k. , vaughan s. , markowitz a. , marshall h. , dobbie p. , warwick r. , 2002 , apj , 568 , 610 eggleton p. p. , 1983 , apj , 268 , 368 frontera f. et al . , 2001 , apj , 546 , 1027 gallo e. , fender r. p. , corbel s. , 2003 , atel , 196 gierliski m. , zdziarski a. a. , poutanen j. , coppi p. s. , ebisawa k. , johnson n. w. , 1999 , mnras , 309 , 496 heinz s. , sunyaev r. a. , 2003 , mnras , 343 , l59 harmon b. a. et al . , 2002 , apjs , 138 , 149 heindl w. a. , smith d. m. , 1998 , apj , 506 , l35 hynes r. i. , steeghs d. , casares j. , charles p. a. , obrien k. , 2003 , apj , 583 , l95 hynes r. i. , steeghs , d . , casares j. , charles p. a. , obrien k. , 2004 , apj , submitted ( astro - ph/0402408 ) illarionov a. f. , beloborodov a. m. , 2001 , mnras , 323 , 159 jung g. v. , 1989 , apj , 338 , 972 kong a. k. h. , kuulkers e. , charles p. a. , homer l. , 2000 , mnras , 312 , l49 kong a. k. h. , charles p. a. , kuulkers e. , kitamoto s. , 2002 , mnras , 329 , 588 kuulkers e. , et al . , 2004 , atel , 240 levine a. m. , bradt h. v. , cui w. , jernigan j. g. , morgan e. h. , remillard r. , shirley r. e. , smith d. a. , 1996 , apj , 469 , l33 lomb n. r. , 1976 , ap&ss , 39 , 447 maccarone t. j. , 2003 , a&a , 409 , 697 maccarone t. j. , coppi p. s. , 2003 , mnras , 338 , 189 magdziarz p. , zdziarski a. a. , 1995 , mnras , 273 , 837 markert t. h. , canizares c. r. , clark g. w. , lewin w. h. g. , schnopper h. w. , sprott g. f. , 1973 , apj , 184 , l67 markoff s. , nowak m. a. , corbel s. , fender r. p. , falcke h. , 2003 , a&a , 397 , 645 meyer f. , liu b. f. , meyer - hofmeister e. , 2000 , a&a , 354 , l67 mineshige s. , 1996 , pasj , 48 , 93 miyamoto s. , kimura k. , kitamoto s. , dotani t. , ebisawa k. , 1991 , apj , 383 , 784 miyamoto s. , kitamoto s. , hayashida k. , egoshi w. , 1995 , apj , 442 , l13 nakanishi h. , sofue y. , 2003 , pasj , 55 , 191 narayan r. , yi i. , 1995 , apj , 452 , 710 narayan r. , igumenshchev i. v. , abramowicz m. a. , 2000 , apj , 539 , 798 nowak m. a. , wilms j. , dove j. b. , 2002 , mnras , 332 , 856 poutanen j. , svensson r. , 1996 , apj , 470 , 249 press w. h. , teukolsky s. a. , vetterling w. t. , flannery b. p. , 1992
, numerical recipes .
cambridge univ . press , cambridge revnivtsev m. , gilfanov m. , churazov e. , 2001 , a&a , 380 , 520 rossi s. , homan j. , miller j. m. , belloni t. , 2004 , in van den heuvel e. p. j. , in t zand j. j. m. , wijers r. a. m. j. , eds .
, the restless high - energy universe , in press ( astro - ph/0309129 ) raska a. , czerny b. , 2000 , a&a , 360 , 1170 rubin b. c. , harmon b. a. , paciesas w. s. , robinson c. r. , zhang s. n. , fishman g. j. , 1998 , apj , 492 , l67 scargle d. , 1982 , apj , 263 , 835 sembach k.r . , danks a.c . ,
savage b.d . , 1993 ,
a&as , 100 , 107 shahbaz t. , fender r. , charles p. a. , 2001 , a&a , 376 , l17 shakura n. i. , sunyaev r. a. , 1973 , a&a , 24 , 337 shapiro s. l. , lightman a. p. , eardley d. m. , 1976 , apj , 204 , 187 shimura t. , takahara f. , 1995 , apj , 445 , 780 smith d. m. , belloni t. , heindl w. a. , kalemci e. , remillard r. , nowak m. , swank j. h. , corbel s. , 2002a , iauc , 7912 , 2 smith d. m. , heindl w. a. , swank j. h. , 2002b , apj , 569 , 362 smith d. m. , heindl w. a. , swank j. h. , wilms j. , pottschmidt k. , 2004 , atel , 231 sunyaev r. a. , revnivtsev m. , 2000 , a&a , 358 , 617 tsunemi h. , kitamoto s. , makoto m. , miyamoto s. , yamashita k. , 1989 , pasj , 41 , 391 ueda y. , ebisawa k. done c. , 1994 , pasj , 46 , 107 wardziski g. , zdziarski a. a. , gierliski m. , grove j. e. , jahoda k. , johnson w. n. , 2002 , mnras , 337 , 829 ( w02 ) wen l. , cui w. , levine a. m. , bradt h. v. , 1999 , apj , 525 , 968 wilms j. , nowak m. a. , dove j. b. , fender r. p. , di matteo t. , 1999 , apj , 522 , 460 zdziarski a. a. , 1986 , apj , 303 , 94 zdziarski a. a. , 1998 , mnras , 296 , l51 zdziarski a. a. , gierliski m. , 2004 , progr . theor .
, in press zdziarski a. a. , johnson w. n. , magdziarz p. , 1996
, mnras , 283 , 193 zdziarski a. a. , poutanen j. , mikoajewska j. , gierliski m. , ebisawa k. , johnson w. n. , 1998 , mnras , 301 , 435 ( z98 ) zdziarski a. a. , lubiski p. , smith d. a. , 1999 , mnras , 303 , l11 zdziarski a. a. , grove j. e. , poutanen j. , rao a. r. , vadawale s. v. , 2001 , apj , 554 , l45 zdziarski a. a. , poutanen j. , paciesas w. s. , wen l. , 2002 , apj , 578 , 357 ( z02 ) zdziarski a. a. , lubiski p. , gilfanov m. , revnivtsev m. , 2003 , mnras , 342 , 355 zhang s. n. , cui w. , harmon b. a. , paciesas w. s. , remillard r. e. , van paradijs j. , 1997 , apj , 477 , l95
we consider a lightcurve of flux measurements , @xmath208 , @xmath209 , each with a corresponding measurement standard error of @xmath210 .
its weighted average is , @xmath211 which unbiased variance estimate is @xmath212 ( e.g. , bevington & robinson 1992 ) .
the weighted average measurement - error variance is @xmath213 the variance of this variance is ( e.g. , brandt 1997 ) , @xmath214 the unbiased estimate of the total flux variance ( including both the intrinsic variability and measurement errors ) is , @xmath215 with its variance of @xmath216 the intrinsic source variance is , @xmath217 where its variance is the sum of the variances of the two factors , @xmath218 the fractional variability is then , @xmath219 considering contributions to its standard deviation from the three variances of its components , the relative error on @xmath17 is @xmath220^{1/2 } } \left [ { \langle \sigma^2_{\rm t}\rangle^2 + \langle \sigma_i^2\rangle^2 \over ( \langle \sigma^2_{\rm t}\rangle- \langle \sigma_i^2\rangle)^2 } + { \langle \sigma_i^2\rangle^2 \over \langle f\rangle^2 } \right ] ^{1/2}.\ ] ] note that when the average measurement error is small , the relative error is simply @xmath221 .
see edelson et al .
( 2002 ) for similar calculations for unweighted averages , where , however , they neglected contributions to the error on @xmath17 from the variance of the average flux and average measurement error . | we study x - ray and variability and distance of gx 3394 .
we derive the distance of @xmath0 kpc , based on recent determination of the binary parameters .
we study data from the asm aboard _ ginga _ , the batse aboard _ cgro _ , and the asm , pca and hexte aboard _ rxte_. from 1987 to 2004 , gx 3394 underwent @xmath115 outbursts and went through all known states of black - hole binaries .
for the first time , we present the pca data from the initial hard state of the outburst of 2004 .
we then study colour - colour and colour - flux correlations . in the hard state
, there is a strong anticorrelation between the 1.55 and 312 kev spectral slopes , which we explain by thermal comptonization of disc photons .
there is also a softening of the spectrum above 3 kev with the increasing flux that becomes stronger with increasing energy up to @xmath1200 kev .
this indicates an anticorrelation between the electron temperature and luminosity , explained by hot accretion models .
in addition , we see a variable broad - band slope with a pivot at @xmath1200 kev . in the soft state , there is a high energy tail with varying amplitude beyond a strong and variable blackbody component .
we confirm the presence of pronounced hysteresis , with the hard - to - soft state transitions occurring at much higher ( and variable ) luminosities than the soft - to - hard transitions .
we fit the _
rxte_/asm data with a model consisting of an outer accretion disc and a hot inner flow .
state transitions are associated then with variations in the disc truncation radius , which we fit as @xmath2 in the soft state and several times that in the hard state .
the disappearence of the inner disc takes place at a lower accretion rate than its initial appearance due to the dependence of the transitions on the source history .
we provide further evidence against the x - ray emission in the hard state being nonthermal synchrotron , and explain the observed radio - x - ray correlation by the jet power being correlated with the accretion power .
[ firstpage ] accretion , accretion discs binaries : general stars : individual : gx 3394 x - rays : binaries x - rays : observations x - rays : stars . |
as an offspring of the wide interest in frame representations and sparsity promoting techniques for data recovery , proximal methods have become popular for solving large - size non - smooth convex optimization problems @xcite .
the efficiency of these methods in the solution of inverse problems has been widely studied in the recent signal and image processing literature ( see for instance @xcite and references therein ) . even if proximal algorithms and the associated convergence properties have been deeply investigated @xcite , some questions persist in their use for solving inverse problems .
a first question is : how can we set the parameters serving to enforce the regularity of the solution in an automatic way ?
various strategies were proposed in order to address this question @xcite , but the computational cost of these methods is often high , especially when several regularization parameters have to be set .
alternatively , it has been recognized for a long time that incorporating constraints directly on the solutions @xcite , instead of considering regularized functions , may often facilitate the choice of the involved parameters .
indeed , in a constrained formulation , the constraint bounds are usually related to some physical properties of the target solution or some knowledge of the degradation process , e.g. the noise statistical properties .
note also that there exist some conceptual lagrangian equivalences between regularized solutions to inverse problems and constrained ones , although some caution should be taken when the regularization functions are nonsmooth ( see @xcite where the case of a single regularization parameter is investigated ) .
another question is related to the selection of the most appropriate algorithm within the class of proximal methods according to a given application .
this also raises the question of the computation of the proximity operators associated with the different functions involved in the criterion . in this context , the objective of this paper is to propose an efficient splitting technique for solving some constrained convex optimization problems of the form : [ p : gen ] @xmath2 where @xmath3 is a real hilbert space , and 1 . for every @xmath4
, @xmath5 is a bounded linear operator from @xmath3 to @xmath6 , 2 . for every @xmath4 , @xmath7-\infty,+\infty\right]}}$ ]
is a proper lower - semicontinuous convex function , 3 . for every @xmath8
, @xmath9 is a bounded linear operator from @xmath3 to @xmath10 , 4 . for every @xmath11
, @xmath12 is a nonempty closed convex subset of @xmath10 .
more precisely , the present work aims at designing a method to address problem [ p : gen ] when some convex constraints are expressed as follows : for some @xmath8 , @xmath13 where @xmath14 and @xmath15 is a proper lower - semicontinuous convex function from @xmath10 to @xmath16-\infty,+\infty\right]}}$ ] .
indeed , the projection onto the convex set @xmath12 defined in often does not have a closed form expression . in the present work
, we will show that : 1 . when the function @xmath15 in corresponds to a _
decomposable loss _ , i.e. it can be expressed as the sum of functions evaluated over different blocks of the vector @xmath17 , the problem of computing the projection onto the associated convex set @xmath12 can be addressed by resorting to a splitting approach that decomposes the set @xmath12 into a collection of epigraphs and a half - space ; 2 . the projection operator associated with an epigraph ( namely the _ epigraphical projection _ ) has a closed form for some functions of practical interest , such as the absolute value raised to a power @xmath18 , the distance to a convex set and the @xmath19-norm with @xmath20 ; 3 . in the context of image restoration ,
regularity constraints based on total variation @xcite and non - local total variation @xcite can be efficiently handled by the proposed epigraphical splitting , which significantly speeds up the convergence ( in terms of execution time ) with respect to standard iterative solutions @xcite .
the paper is organized as follows . in section
[ sec : rec ] , we review the algorithms which are applicable for solving large - size convex optimization problems , so motivating the choice of proximal methods , and we review the variable - splitting techniques commonly used with these methods . in order to deal with a constraint expressed under the form , we propose in section [ sec : proposed ] a novel splitting approach involving an epigraphical projection .
in addition , closed form expressions for specific epigraphical projections are given .
experiments in two different contexts are presented in section [ sec : exp ] .
the first ones concern an image reconstruction problem , while the second ones are related to pulse shape design for digital communications .
finally , some conclusions are drawn in section [ sec : con ] . + * notation * : let @xmath3 be a real hilbert space endowed with the norm @xmath21 and the scalar product @xmath22 . @xmath23 denotes the set of proper lower - semicontinuous convex functions from @xmath3 to @xmath16-\infty,+\infty\right]}}$ ] .
the epigraph of @xmath24 is the nonempty closed convex subset of @xmath25 defined as @xmath26 and the lower level set of @xmath27 at height @xmath28 is the nonempty closed convex subset of @xmath3 defined as @xmath29 .
a subgradient of @xmath27 at @xmath30 is an element of its subdifferential defined as @xmath31 .
when @xmath27 is gteaux - differentiable at @xmath32 , @xmath33 where @xmath34 is the gradient of @xmath27 at @xmath32 .
let @xmath35 be a nonempty closed convex subset @xmath35 of @xmath3 .
the relative interior of @xmath35 is denoted by @xmath36 . for every @xmath37 ,
the indicator function @xmath38 of @xmath35 is given by @xmath39 the projection onto @xmath35 reads @xmath40 , and the distance to @xmath35 is given by @xmath41 .
the first methods for finding a solution to an inverse problem were restricted to the use of a differentiable cost function @xcite , i.e. problem [ p : gen ] where @xmath42 and , for every @xmath4 , @xmath43 denotes a differentiable function . in this context ,
gradient - based algorithms appear to be the most efficients solutions when an iterative procedure is required ( see @xcite and references therein ) .
however , in order to model additional properties , sparsity promoting penalizations ( @xmath44 ) or hard constraints ( @xmath45 ) may be introduced and the diffentiability property is not satisfied anymore .
one way to circumvent this difficulty is to resort to smart approximations in order to smooth the involved non - differentiable functions @xcite .
if one wants to address the original nonsmooth problem without approximation errors , one may apply some specific algorithms @xcite , the convergence of which is guaranteed under restrictive assumptions .
interior point methods @xcite can also be employed for small to medium size optimization problems . on the other hand , in order to solve convex feasibility problems , i.e. to find a vector belonging to the intersection of convex sets ( problem [ p : gen ] with @xmath46 ) , iterative projection methods were developed .
the projection onto convex sets algorithm ( pocs ) is one of the most popular approach to solve data recovery problems @xcite .
a drawback of pocs is that it is not well - suited for parallel implementations .
the parallel projection method ( ppm ) and method of parallel projections ( mopp ) are variants of pocs making use of parallel projections .
moreover , these algorithms were designed to efficiently solve inconsistent feasibility problems ( when the intersection of the convex set is empty ) .
thorough comparisons between projection methods have been performed in @xcite .
computing the projection @xmath47 onto a nonempty closed convex subset @xmath35 of a real hilbert space @xmath3 requires to solve a constrained quadratic minimization problem .
however , it turns out that a closed form expression of the solution to this problem is available in a limited number of instances .
some well - known examples are the projections onto hyperplanes , closed half - spaces and @xmath48-norm balls @xcite .
when an expression of the direct projection is not available , the convex set @xmath35 can be approximated by a half - space , which leads to the concept of subgradient projection .
an efficient block iterative surrogate splitting method was proposed in @xcite in order to solve problem [ p : gen ] when @xmath45 and , for every @xmath4 , @xmath49 where @xmath50 . a main limitation of this method is that the global objective function must be strictly convex . for recent works about subgradient projection methods , the readers may refer to @xcite . a way to overcome this difficulty consists of considering proximal approaches .
the key tool in these methods is the proximity operator @xcite of a function @xmath51 , defined as @xmath52 the proximity operator can be interpreted as a sort of subgradient step for the function @xmath27 , as @xmath53 is uniquely defined through the inclusion @xmath54 proximity operators enjoy many interesting properties @xcite . in particular , they generalize the notion of projection onto a closed convex set @xmath35 , in the sense that @xmath55 .
hence , proximal methods provide a unifying framework that allows one to address non - smooth penalizations ( @xmath56 ) and hard constraints ( @xmath45 ) .
the class of proximal methods includes primal algorithms @xcite and primal - dual algorithms @xcite .
primal algorithms generally require to inverse some linear operators ( e.g. @xmath57 ) , while primal - dual ones only require to compute @xmath58 , @xmath59 and their adjoints . consequently , primal - dual methods are often easier to implement than primal ones , but their convergence may be slower @xcite .
note also that some of these methods are closely related to augmented lagrangian approaches @xcite . with the development of proximal methods for solving convex optimization problems
, many techniques have been developed to cope with the intrinsic limitations of these methods .
in particular , the computation of the proximity operator becomes intractable in general when considering the sum of several functions or a function composed with a linear operator . in this context ,
variable - splitting constitutes a very effective way to design easily implementable algorithms .
one of the most popular examples is given by the approaches inspired by the alternating direction method of multipliers ( admm ) , which deal with optimization problems of the form @xmath60 by introducing an auxiliary variable @xmath61 , the problem is reformulated as @xmath62 this kind of splitting technique has been often used in image restoration @xcite and more recently for distributed optimization problems @xcite .
a similar form of splitting has been considered in @xcite , where the constraint @xmath63 is handled by computing the projection onto the nullspace of the linear operator @xmath64 $ ] , which has a closed - form expression for some specific choices of @xmath65 , such as circulant matrices involved in image restoration .
the solution that we will propose in this work also introduces auxiliary variables . however , our objective is not to deal with linear transformations of the data but with a projection which does not have a closed - form expression .
consequently , the proposed solution departs from the usual splitting methods , in the sense that our approach leads to a collection of epigraphs and a half - space constraint sets , while the usual splitting techniques yield linear constraints .
[ sec : spl ] we now turn our attention to convex sets for which the associated projection does not have a closed form and we show that , under some appropriate assumptions , it is possible to circumvent this difficulty . in problem
[ p : gen ] , assume that @xmath66 denotes such a constraint and that it can be modelled as : for every @xmath67 , @xmath68 where @xmath69 .
hereabove , the generic vector @xmath32 has been decomposed into blocks of coordinates as follows @xmath70^\top\ ] ] and , for every @xmath71 , @xmath72 and @xmath73 is a function in @xmath74 such that @xmath75 .
the idea underlying our approach consists of introducing an auxiliary vector @xmath76 , so that constraint can be equivalently rewritten as@xmath77 align & _
= 1^l_1 _ 1^ ( ) _ 1,[e : const1 ] + & ( \{1, ,l_1})h_1^()(y^ ( ) ) _ 1^()[e : const2 ] .
let us now introduce the closed half - space of @xmath78 defined as @xmath79 with @xmath80 , and the closed convex set @xmath81 then , constraint means that @xmath82 , while constraint is equivalent to @xmath83 . in other words ,
the constraint @xmath66 can be split into the two constraints @xmath84 and @xmath85 , provided that an additional vector @xmath86 is introduced in problem [ p : gen ] .
the resulting criterion takes the form : [ p : epi ] @xmath87 note that the additional constraints can be easily handled by proximal algorithms as far as the projections onto the associated constraint sets can be computed . in the present case , the projection onto @xmath84 is well - known @xcite , whereas the projection onto @xmath85 is given by @xmath88 where @xmath89 , vector @xmath90 is blockwise decomposed as @xmath91^\top$ ] like in , and @xmath92 hence , the problem reduces to the lower - dimensional problem of the determination of the projection onto the convex subset @xmath93 of @xmath94 for each @xmath71 .
an example of an algorithm that converges to a solution to problem [ p : epi ] will be provided in section [ s : algo ] .
the key point in the proposed splitting is the introduction of some epigraphs in the minimization process . in the following ,
we provide some results concerning the projection onto the epigraph of a convex function .
[ p : newprox ] let @xmath3 be a real hilbert space and let @xmath25 be equipped with the standard product space norm .
let @xmath27 be a function in @xmath23 such that @xmath95 is open .
the projector @xmath96 onto the epigraph of @xmath27 is given by : @xmath97 where @xmath98 for every @xmath99 , let @xmath100 .
if @xmath101 , then @xmath102 and @xmath103 .
in addition , @xmath104 which shows that holds .
let us now consider the case when @xmath105 . from the definition of the projection , we get @xmath106 from the karush - kuhn - tucker theorem ( * ? ? ?
* theorem 5.2),@xmath77 there exists @xmath107 such that @xmath108 where the lagrange multiplier @xmath109 is such that @xmath110 . since the value @xmath111 is not allowable ( since it would lead to @xmath112 and @xmath113 )
, it can be deduced from the above equality that @xmath114 .
in addition , differentiating the lagrange functional in w.r.t .
@xmath115 yields @xmath116 hence , @xmath117 given by is such that @xmath118 furthermore , as @xmath105 , we have @xmath119 where we have used the fact that @xmath120 belongs to the boundary of @xmath121 which is equal to @xmath122 since @xmath27 is lower - semicontinuous and @xmath95 is open ( * ? ? ?
* corollary 8.38 ) .
we have then @xmath123 altogether , and lead to @xmath124 which is equivalent to since @xmath125 .
note that alternative characterizations of the epigraphical projection can be found in ( * ? ? ?
* propositions 9.17 , 28.28 ) . from the previous proposition
, we see that the proximity operator in plays a prominent role in the calculation of the projection onto @xmath126 .
we now provide an example of function @xmath27 for which this proximity operator admits a simple form . [ p : valbeta ] let @xmath127 , @xmath1280,+\infty\right[}}$ ] .
assume that @xmath129 if @xmath130-\infty,0]$ ] , then for every @xmath131 @xmath132 where @xmath133 is the unique solution on @xmath134 of the equation @xmath135 if @xmath1360,+\infty\right[}}$ ] , then , for every @xmath131 , @xmath137 where @xmath138 is the unique solution on @xmath139 of .
since @xmath140 is an even function , @xmath141 is an odd function ( * ? ? ?
* remark 4.1(ii ) ) . in the following , we thus focus on the case when @xmath1420,+\infty\right[}}$ ] .
if @xmath130-\infty,0]$ ] , then @xmath143 .
when @xmath144 , @xmath145 and , from ( * ? ? ?
* example 4.6 ) , it can be deduced that @xmath146 when @xmath147 , @xmath148 is differentiable and , according to , @xmath149 is uniquely defined as @xmath150 where , according to ( * ? ? ?
* corollary 2.5 ) , @xmath151 .
this allows us to deduce that @xmath152 .
let us now focus on the case when @xmath1300,{\ensuremath{{+\infty}}}[$ ] .
if @xmath1530,(\zeta/\tau)^{1/\beta}[$ ] , it can be deduced from ( * ? ? ?
* corollary 2.5 ) , that @xmath154 . since @xmath155 @xmath156 , yields @xmath112 .
on the other hand if @xmath157 , as the proximity operator of a function from @xmath158 to @xmath158 is continuous and increasing ( * ? ? ?
* proposition 2.4 ) , @xmath159 .
since @xmath140 is differentiable in this case , and @xmath160 @xmath161 , allows us to deduce that @xmath162 is the unique value in @xmath163 satisfying .
it can be concluded that , when @xmath1640,{\ensuremath{{+\infty}}}[$ ] , holds .
note that , when @xmath165 is a rational number , is equivalent to a polynomial equation for which either closed form solutions are known or standard numerical solutions exist .
the previous propositions allow us to establish the following results concerning the epigraphical projection in : @xmath77 * distance functions of the form @xmath166 where @xmath167 , @xmath1680,+\infty\right[}}$ ] , @xmath169 $ ] , and @xmath170 is a nonempty closed convex subset of @xmath171 .
+ [ ex : epidistl ] assume that @xmath172 is given by .
then , for every @xmath173 , @xmath174 where @xmath175 and @xmath176 , where @xmath177 and the expression of @xmath178 is provided by proposition [ p : valbeta ] .
+ let us notice that @xmath179 where @xmath180 . according to ( * ? ? ?
* proposition 2.7 ) , for every @xmath181 , @xmath182 where @xmath183 .
in addition , we have @xmath184 , & \mbox{if $ \zeta^{(\ell ) } < 0 $ and $ \beta^{(\ell ) } = 1$},\\ \{0\ } , & \mbox{otherwise } , \end{cases}\ ] ] and , according to proposition [ p : valbeta ] , when @xmath185 , @xmath186 and @xmath187 , @xmath188 .
these show that reduces to . * * euclidean norms , as particular case of distance functions in when @xmath189 and @xmath190 : @xmath191 where @xmath167 and @xmath1680,+\infty\right[}}$ ] .
the resulting epigraphical projection is given below .
+ [ ex : norm_l2 ] assume that @xmath172 is given by .
+ then , for every @xmath173 , @xmath192 where @xmath193 .
+ the epigraph of the euclidean norm is the so - called lorentz convex symmetric cone @xcite and the above result is actually known in the literature @xcite . as it will be shown in section [ sec : exp ] , this expression of the epigraphical projection is useful to deal with multivariate sparsity constraints @xcite or total variation bounds @xcite , since such constraints typically involve a sum of functions like composed with linear operators corresponding to analysis transforms or gradient operators . * * infinity norms defined as : for every @xmath167 and @xmath194 , @xmath195 where @xmath1960,+\infty\right[}}^{m^{(\ell)}}$ ] . + [ ex : norm_linf ] assume that @xmath172 is given by where the values @xmath197 are in ascending order , and set @xmath198 and @xmath199 .
then , for every @xmath200 , @xmath201 is such that @xmath202 , with @xmath203 @xmath204 and @xmath205 is the unique integer in @xmath206 such that + @xmath207 + for every @xmath208 , in order to determine @xmath209 we have to find @xmath210 for every @xmath211 , the inner minimization is achieved when , for every @xmath212 , @xmath213 is the projection of @xmath214 onto @xmath215 $ ] , which is given by . then , the problem reduces to @xmath216 which is also equivalent to calculate @xmath217 , where @xmath218 is such that , for every @xmath219 , @xmath220 by using ( * ? ? ?
* proposition 12 ) , we have @xmath221 . the function @xmath218 belongs to @xmath222 since for every @xmath223 , @xmath224 is finite convex and @xmath225 is finite convex and increasing on @xmath134 .
in addition , @xmath218 is differentiable and it is such that , for every @xmath226 and every @xmath227 , @xmath228 for every @xmath229 , as @xmath230 is characterized by , there exists @xmath231 such that @xmath232 and @xmath233 this yields @xmath234 , hence , and we have : @xmath235 @xmath236 . the uniqueness of @xmath231 satisfying this inequality follows from the uniqueness of @xmath237 . + when @xmath238 , the function @xmath172 in reduces to the standard infinity norm @xmath239 for which the expression of the epigraphical projection has been recently given in @xcite .
note that this proposition can be employed to efficient deal with @xmath240 regularization which has attracted much interest recently @xcite .
[ sec : exp ] in this section , we provide numerical examples to illustrate the usefulness of the proposed epigraphical projection method .
the first presented experiment focuses on applications in image restoration involving projections onto @xmath241-balls where @xmath20 . the second experiment deals with a pulse shape design problem based on proposition [ ex : epidistl ] .
set @xmath242 .
denote by @xmath243 the signal of interest , and by @xmath244 an observation vector such that @xmath245 .
it is assumed that @xmath246 is a linear operator , @xmath247 is a decimation operator,@xmath77 and @xmath248 is a realization of a zero - mean white gaussian noise vector .
the recovery of @xmath249 from the degraded observations is performed by following a variational approach which aims at solving the following problem @xmath250^{\overline{n}}}}}{\operatorname{minimize}}\;\;\norm}}{dax - z}^2 \quad\operatorname{s.t.}\quad \sum_{\ell = 1}^l \norm{\omega_\ell \
, b_\ell \ , f \ , x}_p \le \eta,\ ] ] where @xmath251 with @xmath252 , @xmath253 is a real positive constant , and @xmath254 is the linear operator associated with an analysis transform .
furthermore , for every @xmath167 , @xmath255 is a _ block - selection linear operator _ which selects a block of @xmath256 data from its input vector.@xmath77 for every @xmath167 , @xmath257 denotes an @xmath258 diagonal matrix of real positive weights .
the term @xmath259 is the _ data fidelity _ corresponding to the negative log - likelihood of @xmath260 .
the bounds @xmath261 and @xmath262 allow us to take into account the _ value range _ of each component of @xmath249 .
the second constraint involved in problem promotes solutions having a sparse analysis representation .
indeed , it reduces to the weighted @xmath263-norm criterion found in @xcite when each block reduces to a singleton ( i.e. @xmath264 , and , for every @xmath167 , @xmath265 and @xmath266 ) .
it captures the @xmath267 criteria present in @xcite when @xmath268 .
it matches the @xmath269 criterion proposed in @xcite when @xmath270 .
note that overlapping blocks in constraint are dealt with by increasing the dimensionality of the problem ( through the linear transform @xmath271 ) and then using an usual non - overlapping block selection operator ( denoted @xmath272 ) .
let us define @xmath273 and @xmath274 where @xmath275 and the same decomposition as in is performed .
then , it can be observed that problem is a particular case of problem [ p : gen ] where @xmath276 , @xmath277 , @xmath278 , @xmath279 , @xmath280 , @xmath66 is the above @xmath0-ball , @xmath281 and @xmath282^{\overline{n}}$ ] . as already mentioned in section [ ssec : algprox ]
, various proximal algorithms can be used to solve non - smooth convex optimization problems and would potentially benefit from the proposed epigraphical splitting technique . in this work , we will consider the monotone+lipschitz forward backward forward ( m+lfbf ) algorithm , which is a primal - dual method recently proposed in @xcite , and the simultaneous - direction method of multipliers algorithm ( sdmm ) @xcite , which is a parallelized version of admm .
their convergence is guaranteed ( under weak conditions ) and their structure makes them suitable for implementation on parallel architectures .
although both algorithms address a wide class of convex optimization problems , sdmm requires to invert the matrix @xmath283 .
this is a well - known limitation of admm - like algorithms , which can be circumvented in the case when the matrix @xmath284 is diagonalizable in the dft domain ( e.g. see @xcite ) . in the present case ,
the matrix @xmath284 is not diagonalizable since @xmath285 .
consequently , sdmm requires to introduce some auxiliary variables into the minimization problem ( see @xcite ) in order to solve the original problem .
as we will see in tables [ tab : nltv3_2]-[tab : nltv3_inf ] , this makes admm - like approaches much slower than primal - dual approaches .
the main difficulty in solving problem stems from the constraint in .
the point is that proximal algorithms ( as well as most of the applicable algorithms ) require to compute the projection onto @xmath66 .
specific numerical methods @xcite have been developed for this purpose .
the aim of this section is to propose an alternative method based on the splitting principle presented in section [ sec : spl ] .
so doing , the resulting problem can be efficiently addressed by proximal algorithms .
the two possible approaches are now detailed . * _ epigraphical method _ the principle of this method is to decompose @xmath66 into the closed half - space defined by and the closed convex set defined by with @xmath286 .
we have thus to @xmath287 the advantage of this decomposition is that the projections onto @xmath84 and @xmath85 have closed forms .
indeed , @xmath288 is the projection onto a half - space , while @xmath289 is given by proposition [ ex : norm_l2 ] for @xmath268 and proposition [ ex : norm_linf ] for @xmath270 .
we are then able to solve problem by means of algorithms such as m+lfbf or sdmm .
the associated iterations are given in algorithm [ algo : epi ] .
note that replacing the inequality in the definition of @xmath84 by an equality constraint @xmath290 was observed to make little difference in the numerical behaviour of sdmm .
+ @xmath291 } , \nu^{[0 ] } } \in { \ensuremath{\mathbb r}}^{m } \times { \ensuremath{\mathbb r}}^l\\ \parens{x^{[0 ] } , \zeta^{[0 ] } } \in { \ensuremath{\mathbb r}}^{\overline{n } } \times { \ensuremath{\mathbb r}}^l\\ \theta = 2\vert a \vert^2 + \max\{\norm{{\lambda } f},1\}\\ \epsilon \in ] 0 , \frac{1}{\theta+1}[\\ \end{array } \right.\\[5 mm ] \mathrm{for}\ ; i = 0 , 1 , \dots\\ \left\lfloor \begin{array}{l } \displaystyle\gamma_i \in \bracks*{\epsilon , \frac{1-\epsilon}{\theta}}\\\\ \parens[\big]{\widehat{x}^{[i]},\widehat{\zeta}^{[i ] } } = \parens[\big]{x^{[i]},\zeta^{[i ] } } - \gamma_i \parens[\big]{a^\top \nabla g_1(a x^{[i ] } ) + f^\top\lambda^\top v^{[i ] } , \nu^{[i]}}\\ \parens[\big]{p^{[i]},\rho^{[i ] } } = p_{c_2 \times v_1}\parens[\big]{\widehat{x}^{[i]},\widehat{\zeta}^{[i]}}\\ \parens[\big]{\widehat{v}^{[i]},\widehat{\nu}^{[i ] } } = \parens[\big]{v^{[i]},\nu^{[i ] } } + \gamma_i \parens[\big]{{\lambda } f \ , x^{[i]},\zeta^{[i]}}\\ \parens[\big]{a^{[i]},\alpha^{[i ] } } = \parens[\big]{\widehat{v}^{[i]},\widehat{\nu}^{[i ] } } - \gamma_i p_{e_1}\parens[\big]{\widehat{v}^{[i]}/\gamma_i,\widehat{\nu}^{[i]}/\gamma_i}\\ \parens[\big]{v^{[i+1]},\nu^{[i+1 ] } } = \parens[\big]{a^{[i]},\alpha^{[i ] } } + \gamma_i \parens[\big]{\lambda f ( p^{[i]}-x^{[i ] } ) , \rho^{[i]}-\zeta^{[i]}}\\ \parens[\big]{\widetilde{x}^{[i]},\widetilde{\zeta}^{[i ] } } = \parens[\big]{p^{[i]},\rho^{[i ] } } - \gamma_i \parens[\big]{a^\top \nabla g_1(a p^{[i ] } ) + f^\top\lambda^\top a^{[i]},\alpha^{[i]}}\\ \parens[\big]{x^{[i+1]},\zeta^{[i+1 ] } } = \parens[\big]{x^{[i]}- \widehat{x}^{[i]}+ \widetilde{x}^{[i]},\zeta^{[i ] } - \widehat{\zeta}^{[i ] } + \widetilde{\zeta}^{[i]}}\\ \end{array } \right .
\end{array}\ ] ] * _ direct method _ for completeness , we also consider the projection onto @xmath66 with the algorithm in @xcite when , @xmath77 or the iterative algorithm in @xcite when @xmath270 . in this case ,
proximal methods ( such as m+lfbf and sdmm ) can be used to solve directly problem . according to the general results in (
* theorem 4.2 ) and @xcite , the sequence @xmath292}\big)_{i\in { \ensuremath{\mathbb n}}}$ ] generated by m+lfbf or sdmm is guaranteed to converge to a ( global ) minimizer of problem . in the context of image restoration ,
the quality of the results obtained through a variational approach strongly depends on the ability to model the regularity present in images . since natural images are often piecewise smooth , popular regularization models tend to penalize the image gradient . in this regard , _ total variation _ ( tv )
@xcite has emerged as a simple , yet successful , convex optimization tool .
however , tv fails to preserve textures , details and fine structures , because they are hardly distinguishable from noise . to improve this behaviour ,
the tv model has been extended by using a non - locality principle @xcite .
another approach to overcome these limitations is to replace the gradient operator with a frame representation which yields a more suitable sparse representation of the image @xcite .
the connections between these two different approaches have been studied in @xcite .
it is still unclear which approach leads to the best results .
however , there are some evidences that _ non - local _ ( nl ) tv may perform better in some image restoration tasks @xcite .
we thus focus our attention on nltv - based constraints , although our proposed algorithm is quite general and it can also be adapted to frame - based approaches . by appropriately selecting the operators @xmath271 , @xmath272 and @xmath257 in problem , we can integrate the nltv measures in a constrained convex optimization approach . in our experiments , we propose to evaluate the performances of two nltv constraints that constitute particular cases of the one considered in when @xmath293 .
they are described for 2d data in the following . *
_ @xmath48-nltv _ this constraint has the form @xmath294 where @xmath295 is the _ neighbourhood support _ at position @xmath296 and @xmath297 is the set of positions @xmath298 located into a @xmath299 window centered at @xmath296 , where @xmath300 is odd .
this constraint is a particular case of the one considered in where @xmath301 and @xmath271 is a concatenation of discrete difference operators @xmath302 with @xmath303 .
more precisely , for every @xmath304 , @xmath302 is a 2d filter with impulse response : for every @xmath305 , @xmath306 in addition , for every @xmath307 , , @xmath272 selects the components of @xmath308 corresponding to differences @xmath309 , and the positive weights @xmath310 are gathered in the diagonal matrix @xmath257 . * _ @xmath311-nltv_ we consider the following constraint @xmath312 we proceed similarly to the previous constraint , except that the @xmath311-norm is now substituted for the @xmath48-norm .
note that the classical isotropic tv constraint ( designated by @xmath48-tv in the following ) constitutes a particular case of the @xmath48-nltv one , where each neighbourhood @xmath295 only contains the horizontal / vertical neighbouring pixels ( @xmath313 ) and the weights are @xmath314 .
similarly , the @xmath311-tv constraint is a special case of the @xmath311-nltv one . to set the weights , we got inspired from the non - local means approach originally described in @xcite . here , for every @xmath307 and @xmath315
, the weight @xmath316 depends on the similarity between patches built around the pixels @xmath296 and @xmath317 of the image . since our degradation process involves some missing data , a two - step approach has been adopted . in the first step ,
the @xmath48-tv approach is used in order to obtain an estimate @xmath318 of the target image .
this estimate is subsequently used in the second step to compute the weights through a _ self - similarity _ measure , yielding @xmath319 where @xmath320 , @xmath3210,+\infty\right[}}$ ] , @xmath322 ( resp .
@xmath323 ) selects a @xmath324 patch centered at position @xmath296 ( resp .
@xmath317 ) and @xmath325 ( resp .
@xmath326 ) is a linear processing of the image depending on the position @xmath296 ( resp .
@xmath317 ) . the constant @xmath327
is set so as to normalize the weights ( i.e. @xmath328 ) .
the measure in generalizes the one proposed in @xcite , which corresponds to the case when @xmath325 ( resp .
@xmath326 ) reduces to a gaussian function with mean @xmath296 ( resp .
@xmath317 ) . in the present work ,
we consider the _ foveated self - similarity _ measure recently introduced in @xcite , due to its better performance in denoising .
this approach can be derived from by setting @xmath325 ( resp .
@xmath326 ) to a set of low - pass gaussian filters whose variances increase as the spatial distance from the patch center @xmath296 ( resp .
@xmath317 ) grows . for every @xmath307 , the neighbourhood @xmath295 is built according to the procedure described in @xcite . in practice , we limit the size of the neighbourhood , so that @xmath329 ( a possible choice of @xmath330 is given in the next paragraph ) . in this section ,
the execution time of the proposed epigraphical technique is evaluated w.r.t . the direct method involving standard numerical solutions . in the following experiments , if not specified otherwise , the degradation matrix @xmath331 is a convolution which consists of a @xmath332 uniform blur and the decimation matrix @xmath333 randomly removes @xmath334 of the pixels ( @xmath335 ) .
the standard deviation of the additive white gaussian noise is equal to @xmath336 .
since we deal with natural images , the data range bounds are @xmath337 and @xmath338 . for the smoothness constraint , we set @xmath339 , @xmath340 , @xmath341 and @xmath342 .
we present the results obtained with the image _ boat _ cropped at @xmath343 ( @xmath344 ) , since a similar behaviour was observed for other images .
the stopping criterion is set to @xmath345}-x^{[i ] } } \le 10^{-4 } \norm{x^{[i]}}$ ] . for the @xmath0-ball projectors needed by the direct method
, we used the software publicly available on - line @xcite .
note that sdmm requires to invert the matrix @xmath283 , which we address by resorting to the solution proposed in @xcite . in order to make the operators @xmath331 and @xmath271 diagonalizable in the dft domain , a periodic extension of the image is performed . in practice
, the constraint bound @xmath253 may not be known precisely . although it is out of the scope of this paper to devise an optimal strategy to set this bound , it is important to evaluate the impact of its choice on our method performance . in the following ,
we compare the epigraphical approach with the direct computation of the projections ( via standard iterative solutions ) for different choices of regularization constraints and values of @xmath253 . * _ total variation _
tables [ tab : tv_eta ] and [ tab : tv_eta_inf ] report a comparison between the direct and epigraphical methods for @xmath346-tv and @xmath347-tv , respectively . for more readability ,
the values of @xmath253 are expressed as a multiplicative factor of the @xmath348-tv - semi - norm of the original image .
the convergence times indicate that the epigraphical approach yields a faster convergence than the direct approach for sdmm and m+lfbf .
moreover , the numerical results show that errors within @xmath349 from the optimal value for @xmath253 lead to snr variations within @xmath350 .
[ fig : tv_prof : time ] and [ fig : tv_prof_inf : time ] show the relative error @xmath351}-x^{[\infty]}}/\norm{x^{[\infty]}}$ ] as a function of the computational time , where @xmath352}$ ] denotes the solution computed after a large number of iterations ( typically , 5000 iterations ) .
the dashed line presents the results for the direct method while the solid line refers to the epigraphical one .
these plots show that the epigraphical approach is faster despite it requires more iterations in order to converge .
this can be explained by the computational cost of the subiterations required by the direct projections onto the @xmath0-ball .
+ + c@^c @ ^c@^c ^c@^c @^c @^c@^c @ ^c@^c ^c@^c @^c @ ^c & & & & + ( r)10 - 15 & & & & & & & & & & + ( r)3 - 4(lr)5 - 6(r)10 - 11(lr)12 - 13 & & # iter . & sec .
& sec . & & & & # iter . & sec . &
# iter . & sec . & & + 0.45 & 19.90 0.733 & 107 & 6.07 & 174 & 2.03 & & 2.99 & & 113 & 6.15 & 182 & 3.49 & & 1.76 + 0.50 & 20.18 0.745 & 117 & 6.95 & 159 & 1.95 & & 3.57 & & 116 & 6.97 & 168 & 3.44 & & 2.03 + * * 0.56 & 20.23 0.745 & 129 & 8.36 & 153 & 1.90 & & 4.41 & & 124 & 8.17 & 159 & 3.01 & & 2.72 + 0.62 & 20.16 0.737 & 141 & 9.44 & 155 & 1.83 & & 5.16 & & 131 & 8.62 & 159 & 3.26 & & 2.65 + 0.67 & 20.00 0.724 & 154 & 10.20 & 162 & 2.17 & & 4.71 & & 140 & 10.00 & 164 & 2.84 & & 3.52 + + [ tab : tv_eta ] + + c@^c @ ^c@^c ^c@^c @^c @^c@^c @ ^c@^c ^c@^c @^c @ ^c & & & & + ( r)10 - 15 & & & & & & & & & & + ( r)3 - 4(lr)5 - 6(r)10 - 11(lr)12 - 13 & & # iter . & sec . &
# iter . & sec . & & & & # iter . & sec
& # iter . & sec . & & + 0.45 & 19.52 0.726 & 160 & 312.55 & 231 & 3.89 & & 80.43 & & 183 & 347.10 & 252 & 6.43 & & 53.96 + 0.50 & 19.71 0.734 & 168 & 342.01 & 215 & 3.75 & & 91.31 & & 185 & 368.24 & 236 & 5.83 & & 63.17 + * * 0.56 & 19.71 0.728 & 180 & 373.60 & 211 & 3.49 & & 106.93 & & 189 & 386.29 & 229 & 5.53 & & 69.91 + 0.62 & 19.59 0.715 & 196 & 412.68 & 216 & 3.67 & & 112.50 & & 198 & 411.04 & 229 & 5.86 & & 70.15 + 0.67 & 19.39 0.698 & 211 & 448.77 & 223 & 3.76 & & 119.27 & & 207 & 437.66 & 234 & 5.76 & & 75.96 + + [ tab : tv_eta_inf ] + + c@^c @ ^c@^c ^c@^c @^c @^c@^c @ ^c@^c ^c@^c @^c @ ^c & & & & + ( r)10 - 15 & & & & & & & & & & + ( r)3 - 4(lr)5 - 6(r)10 - 11(lr)12 - 13 & & # iter . & sec . &
# iter . & sec . & & & & # iter . & sec . & # iter . & sec . & & + + 0.43 & 20.82 0.757 & 208 & 20.67 & 211 & 10.93 & & 1.89 & & 82 & 6.95 & 93 & 3.76 & & 1.85 + 0.49 & 20.97 0.765 & 167 & 16.84 & 177 & 9.01 & & 1.87 & & 75 & 6.61 & 83 & 3.47 & & 1.91 + 0.54 & 21.02 0.767 & 147 & 15.31 & 157 & 7.93 & & 1.93 & & 71 & 6.45 & 77 & 3.15 & & 2.04 + 0.59 & 20.98 0.764 & 134 & 14.44 & 148 & 7.67 & & 1.88 & & 72 & 6.58 & 77 & 3.24 & & 2.03 + 0.65 & 20.88 0.757 & 133 & 14.82 & 136 & 7.11 & & 2.08 & & 76 & 7.53 & 80 & 3.27 & & 2.30 + + 0.43 & 21.00 0.766 & 301 & 56.03 & 343 & 45.18 & & 1.24 & & 82 & 8.51 & 90 & 5.43 & & 1.57 + 0.49 & 21.15 0.773 & 260 & 49.03 & 302 & 39.64 & & 1.24 & & 75 & 7.90 & 81 & 4.90 & & 1.61 + * * 0.54 & 21.20 0.775 & 242 & 46.31 & 283 & 37.72 & & 1.23 & & 71 & 8.26 & 75 & 4.47 & & 1.85 + 0.59 & 21.17 0.773 & 231 & 46.20 & 268 & 36.56 & & 1.26 & & 70 & 7.94 & 74 & 4.49 & & 1.77 + 0.65 & 21.08 0.767 & 220 & 44.64 & 252 & 34.46 & & 1.30 & & 73 & 8.40 & 76 & 4.59 & & 1.83 + + [ tab : nltv3_2 ] + + c@^c @ ^c@^c ^c@^c @^c @^c@^c @ ^c@^c ^c@^c @^c @ ^c & & & & + ( r)10 - 15 & & & & & & & & & & + ( r)3 - 4(lr)5 - 6(r)10 - 11(lr)12 - 13 & & # iter . & sec . & # iter . & sec . & & & & # iter
# iter . & sec . & & + + 0.43 & 20.78 0.762 & 434 & 1470.46 & 449 & 25.03 & & 58.76 & & 225 & 730.26 & 244 & 12.35 & & 59.15 + 0.49 & 20.86 0.764 & 395 & 1319.64 & 413 & 22.86 & & 57.72 & & 221 & 692.25 & 237 & 11.92 & & 58.08 + 0.54 & 20.83 0.760 & 363 & 1193.61 & 382 & 21.46 & & 55.62 & & 217 & 667.50 & 233 & 11.46 & & 58.22 + 0.59 & 20.73 0.752 & 340 & 1093.26 & 354 & 19.77 & & 55.30 & & 216 & 653.79 & 230 & 11.67 & & 56.01 + 0.65 & 20.58 0.740 & 322 & 1007.55 & 336 & 18.64 & & 54.06 & & 216 & 643.00 & 229 & 11.45 & & 56.18 + + 0.43 & 20.91 0.769 & 384 & 2069.62 & 452 & 64.42 & & 32.13 & & 233 & 863.01 & 252 & 18.47 & & 46.73 + 0.49 & 20.98 0.771 & 326 & 1700.34 & 412 & 58.66 & & 28.99 & & 231 & 822.06 & 247 & 18.36 & & 44.77 + * * 0.54 & 20.97 0.767 & 290 & 1476.98 & 389 & 55.35 & & 26.69 & & 229 & 787.61 & 245 & 17.90 & & 43.99 + 0.59 & 20.88 0.759 & 276 & 1336.16 & 374 & 52.64 & & 25.38 & & 230 & 772.42 & 245 & 17.57 & & 43.96 + 0.65 & 20.75 0.749 & 268 & 1220.14 & 362 & 51.45 & & 23.72 & & 231 & 760.86 & 245 & 17.81 & & 42.72 + + [ tab : nltv3_inf ] * _ @xmath48-nltv _ table [ tab : nltv3_2 ] collects the results of @xmath48-nltv for different values of neighbourhood size @xmath284 . to set the weights ,
the first tv estimate is computed with @xmath353 .
the convergence times show that the epigraphical approach is faster than the direct one for both considered algorithms .
moreover , it can be noticed that errors within @xmath349 from the optimal bound value lead to snr variations within @xmath354 . in [ fig : nltv2_prof : time ] , a plot similar to those in figs .
[ fig : tv_prof : time ] and [ fig : tv_prof_inf : time ] show the convergence profile .
the epigraphical method requires about the same number of iterations as the direct one in order to converge .
this results in a time reduction , as a single iteration of the epigraphical method is faster than one iteration of the direct method . * _ @xmath311-nltv _ table [ tab : nltv3_inf ] and [ fig : nltv_inf_prof : time ] show the results obtained with the @xmath311-nltv constraint . similarly to @xmath311-tv , the epigraphical approach greatly speeds up the convergence times . in this section ,
the quality of images reconstructed with our variational approach is evaluated for different choices of regularization constraints and comparisons are made with a state - of - the - art method .
extensive tests have been carried out on several standard images of different sizes .
the snr and ssim @xcite results obtained by using the various previously introduced tv - like constraints are collected in table [ tab : all ] .
in addition , a comparison is performed between our method using an m+lfbf implementation and the gradient projection for sparse reconstruction ( gpsr ) method @xcite , which also relies on a variational approach .
the constraint bound for both methods was hand - tuned in order to achieve the best snr values .
the best results are highlighted in bold .
a visual comparison is made in [ fig : tv_images ] , where two representative images are displayed .
these results demonstrate the interest of considering non - local smoothness measures .
indeed , nltv with @xmath267-norm proves to be the most effective constraint with gains in snr and ssim ( up to 1.82 db and 0.042 ) with respect to @xmath48-tv , which in turn outperforms gpsr
. the better performance of nltv seems to be related to its ability to better preserve edges and thin structures present in images . in terms of computational time , gpsr is about twice faster than @xmath48-nltv .
our codes were developed in matlab@xmath77 , the operators @xmath271 and @xmath355 being implemented in c using mex files . in order to complete the analysis ,
we report in [ fig : noise ] snr / ssim comparisons between @xmath48-nltv and @xmath48-tv for different blur and noise configurations .
these plots show that @xmath48-nltv provides better results regardless of the degradation conditions . [
cols="<,^,>,>,>,>,^,>,>",options="header " , ]
we have proposed a new epigraphical technique to deal with constrained convex variational formulations of inverse problems with the help of proximal algorithms . in this paper , our attention has been turned to constraints based on distance functions and weighted @xmath0-norms with @xmath20 . in the context of 1d signals , we have shown that constraints based on distance functions are useful for pulse shape design . in the context of images ,
we have used @xmath0-norm constraints to promote block - sparsity of analysis representations .
the obtained results demonstrate the better performance of non - local measures in terms of image quality .
our results also show that the @xmath267-norm has to be preferred over the @xmath269-norm for image recovery problems . however , it would be interesting to consider alternative applications of @xmath269-norms such as regression problems @xcite .
furthermore , the experimental part indicates that the epigraphical method converges faster than the approach based on the direct computation of the projections via standard iterative solutions .
parallelization of our codes should even allow us to accelerate them @xcite .
note that , although the considered application involves two constraint sets , the proposed approach can handle an arbitrary number of convex constraints .
the epigraphical approach could also be used to develop approximation methods for addressing more general convex constraints .
10 p. l. combettes and j .- c .
proximal splitting methods in signal processing . in h.
h. bauschke , r. s. burachik , p. l. combettes , v. elser , d. r. luke , and h. wolkowicz , editors , _ fixed - point algorithms for inverse problems in science and engineering _ , pages 185212 .
springer - verlag , new york , 2011 . c. chaux , m. el gheche , j. farah , j .- c .
pesquet , and b. pesquet - popescu . a parallel proximal splitting method for disparity estimation from multicomponent images under illumination variation .
, 47(3):167178 , november 2013 .
s. ono and i. yamada .
poisson image restoration with likelihood constraint via hybrid steepest descent method . in _ proc .
acoust . , speech signal process .
_ , pages x+5 , vancouver , canada , may 26 - 31 2013 . y. censor , w. chen , p. l. combettes , r. davidi , and g. t. herman . on the effectiveness of projection methods for convex feasibility problems with linear inequality constraints . , 51(3):10651088 , 2012 .
i. yamada .
the hybrid steepest descent method for the variational inequality problem over the intersection of fixed point sets of nonexpansive mappings . in
_ inherently parallel algorithms for feasibility and optimization and their applications _ , pages 473504
elsevier , 2001 .
a. foi and g. boracchi .
foveated self - similarity in nonlocal image filtering . in _ proc .
spie electronic imaging 2012 , human vision and electronic imaging xvii _ ,
volume 8291 , burlingame ( ca ) , usa , jan . 2012 .
r. gaetano , g. chierchia , and b. pesquet - popescu .
parallel implementations of a disparity estimation algorithm based on a proximal splitting method . in _ visual communication and image processing _ , san diego , usa , 2012 . | we propose a proximal approach to deal with a class of convex variational problems involving nonlinear constraints . a large family of constraints , proven to be effective in the solution of inverse problems ,
can be expressed as the lower level set of a sum of convex functions evaluated over different , but possibly overlapping , blocks of the signal .
for such constraints , the associated projection operator generally does not have a simple form .
we circumvent this difficulty by splitting the lower level set into as many epigraphs as functions involved in the sum .
a closed half - space constraint is also enforced , in order to limit the sum of the introduced epigraphical variables to the upper bound of the original lower level set . in this paper , we focus on a family of constraints involving linear transforms of distance functions to a convex set or @xmath0 norms with @xmath1 . in these cases , the projection onto the epigraph of the involved function has a closed form expression .
the proposed approach is validated in the context of image restoration with missing samples , by making use of constraints based on non - local total variation .
experiments show that our method leads to significant improvements in term of convergence speed over existing algorithms for solving similar constrained problems . a second application to a pulse shape design problem
is provided in order to illustrate the flexibility of the proposed approach . |
detailed measurements of fluctuations in the cosmic microwave background ( cmb ) have established cosmology as a high - precision science .
one striking illustration of this is the fact that it is today possible to predict a vast number of observables based on six numbers only , with only a few ( but nevertheless intriguing ) `` glitches '' overall .
the key to this success has been making accurate measurements of the cmb power spectrum , perhaps most prominently exemplified by wilkinson microwave anisotropy probe ( wmap ; bennett et al .
2003 ; hinshaw et al . 2007 ,
hinshaw et al . 2008 ) .
the primary connection between theoretical models and cmb observations is made through the cmb likelihood , @xmath18 .
this is a multivariate , non - gaussian function that quantifies the match between the data and a given power spectrum , @xmath6 .
unfortunately , it is impossible to evaluate this function explicitly for modern high - resolution data sets , due to the sheer size of the problem , and one therefore instead typically resolves to various approximations .
however , given the importance of the cmb in modern cosmology , it is of critical importance to characterize this likelihood accurately , and all approximations must be thoroughly verified .
one example is the approximation of the large angular scale likelihood , where @xmath7 is strongly non - gaussian .
this turned out to be a non - trivial issue after the original analysis of the 3-year wmap temperature data by @xcite , in which a master - based @xcite approximation was used at @xmath19 .
an exact likelihood analysis @xcite later demonstrated that this sub - optimal approximation , when applied to harmonic modes between @xmath20 and 30 , biased the spectral index of scalar perturbations , @xmath15 , low by @xmath16 .
a second example is that of non - cosmological foregrounds . unless properly accounted for , such foregrounds bias the observed power spectrum to high values , and can seriously compromise any cosmological conclusions . while important for temperature observations , this is an absolutely crucial issue for polarization observations , as the desired cmb in amplitude is comparable to or weaker than the interfering foregrounds over most of the sky . in recent years , a new set of statistical methods
have been developed that allows the user to address these issues within a single well - defined framework @xcite .
the heart of this method is the gibbs sampling algorithm ( see , e.g , gelfand and smith 1990 ) , in which samples from a ( typically complicated ) joint distribution are drawn by alternately sampling from ( simpler ) conditional distributions . in the cmb setting , this is realized by drawing joint samples from @xmath21 , by alternately sampling from @xmath22 , where @xmath6 is the cmb power spectrum , @xmath23 is the cmb sky signal , and @xmath24 are the observed data .
in addition to allow for exact likelihood analysis at reasonable computational cost , an equally important feature of this framework is its unique capability of including additional degrees of freedom , such as non - cosmological foregrounds , into the analysis @xcite .
further , very recently an additional metropolis - hastings mcmc sampling step was introduced by @xcite , that effectively resolves the previously described inefficiency of the gibbs sampler at low signal - to - noise @xcite .
the framework has also been extended to handle polarization @xcite and anisotropic universe models @xcite . by now , the cmb gibbs sampler is well established and demonstrated to sample efficiently from the exact cmb posterior . however , a long - standing issue has been the characterization of the joint likelihood , given a set of such samples .
originally , @xcite proposed to use the so - called blackwell - rao ( br ) estimator for this purpose , and this approach was later implemented and studied in detail by @xcite . while highly accurate for the large angular scale and high signal - to - noise temperature likelihood , it suffers from one major drawback : because it attempts to describe the full @xmath25-dimensional likelihood without any constraints on allowed correlations , the number of samples required for convergence scales exponentially with @xmath25 . in practice , this limits the br estimator to @xmath26 for temperature data , and just @xmath27 for low signal - to - noise polarization data . in this paper
, we introduce a new temperature likelihood approximation based on samples drawn from the cmb posterior , by modifying the original br estimator in a way that restricts the allowed @xmath28-point functions of @xmath7 , but still captures most of the relevant information .
explicitly , this is done through a specific change of variables , such that the observed marginal posterior for each multipole , @xmath29 , is transformed into a gaussian .
then , in these new variables the joint distribution is approximated by a multivariate gaussian .
as long as the correlation between any two multipoles is reasonably small , as is the case for nearly full - sky experiments such as wmap and planck , we shall see that this provides an excellent approximation to the exact joint likelihood . as a result
, the new approach greatly reduces the overall number of samples required for convergence , and allows us to obtain a highly accurate likelihood approximation to arbitrary @xmath25 .
generalization to a full polarized likelihood will be discussed in a future paper ( eriksen et al .
, in preparation ) .
this paper is organized as follows : in [ sec : review ] , we first briefly review the gibbs sampling algorithm together with the original blackwell - rao estimator , and in [ sec : method ] we introduce the new gaussianized blackwell - rao estimator .
next , in
[ sec : application ] , we apply the new estimator to simulated data , and compare results with brute - force likelihood evaluations in pixel space . in [ sec : analysis ] , we analyze the 5-year wmap temperature data , and provide an updated power spectrum and set of cosmological parameters .
we summarize and conclude in [ sec : conclusion ] .
we start by reviewing the current state of the cmb gibbs sampling framework , as previously developed through a series of papers @xcite , and highlight the problem of likelihood modelling as currently presented in the literature .
first , we assume that our observations , @xmath24 , in direction @xmath30 may be modelled in terms of a signal , @xmath23 and a noise , @xmath31 , component , @xmath32 further , we assume that both @xmath23 and @xmath31 are gaussian distributed with vanishing mean and covariances @xmath33 and @xmath34 , respectively .
the cmb is in this paper additionally assumed to be isotropic , such that in spherical harmonic space ( @xmath35 ) the cmb covariance matrix may be written as @xmath36 , where @xmath37 is the angular power spectrum .
our goal is now to map out the cmb posterior distribution @xmath38 and the cmb likelihood @xmath39 .
note that we in this paper are concerned with the problem of likelihood characterization only , which is a post - processing step relative to the gibbs sampler . for notational transparency , we therefore neglect issues such as foreground marginalization , instrumental beams , multi - frequency observations etc . for full details on these issues ,
see , e.g. , eriksen et al . 2008a .
when working with real - world cmb data , there are a number of issues that complicate the analysis .
two important examples are anisotropic noise and galactic foregrounds .
first , because of the scanning motion of a cmb satellite , the pixels in a given data set are observed over unequal amounts of time .
this implies that the effective noise is a function of pixel location on the sky .
second , large regions of the sky are obscured by galactic foregrounds ( e.g. , synchrotron , free - free and dust emission ) , and these regions must be rejected from the analysis by masking . because of such issues , the total data covariance matrix @xmath40 is dense in both pixel and harmonic space . as a result ,
it is computationally unfeasible to evaluate and sample directly from @xmath38 .
fortunately , this problem was originally solved by @xcite , @xcite and @xcite , who developed a particular cmb gibbs sampler for precisely this purpose . for full details on this method
, we refer the interested reader to the original papers , and in the following we only describe the main ideas .
the practical implementation of the algorithm used in this paper is called `` commander '' , and has been described in detail by @xcite . the idea behind the cmb gibbs sampler is to draw samples from the joint posterior by alternately sampling from the two corresponding conditionals . the sampling scheme may thus be written on the symbolic form @xmath41 where the left arrow implies sampling from the distribution on the right - hand side .
then , after some burn - in period , @xmath42 will be drawn from the desired distribution .
the only remaining step is to write down sampling algorithms for each of the two above conditional distributions , both of which are readily available for our problem , since the former is simply a multivariate gaussian , and the second is a product of independent inverse gamma distributions .
for one possible general sampling algorithm for @xmath43 , see , e.g. , @xcite .
the gibbs sampler produces a set of samples drawn from the joint cmb posterior , @xmath38 .
however , for these samples to be useful for estimation of cosmological parameters , we have to transform the information contained in this sample set into a smooth approximation to the likelihood @xmath18 . in principle
, we could simply generate a multi - variate histogram and read off corresponding values , but this does not work in practice because of the large dimensionality of the parameter space . in the current literature ,
the best approach for handling this problem is the blackwell - rao ( br ) estimator @xcite , which attempts to smooth the sampled histogram by taking advantage of the known analytic distribution , @xmath43 : first , we define the observed power spectrum , @xmath44 , of the current cmb sky gibbs sample , @xmath45 , @xmath46 then the br estimator is derived as follows , @xmath47 in other words , the br estimator is nothing but the average of @xmath48 over the sample set , where @xmath44 refers to the power spectrum of a full - sky noiseless cmb signal gibbs sample .
this distribution has a simple analytic expression ( e.g. , * ? ? ?
* ) , @xmath49 while equation [ eq : br ] does constitute a computationally convenient and accurate approximation to the full likelihood for some special applications , it suffers badly from poor convergence properties with increasing dimensionality of the sampled space .
this behaviour may be understood in terms of relative distribution widths : suppose we want to map out an @xmath25-dimensional distribution , and each of the univariate blackwell - rao functions [ i.e. , @xmath48 ] have a standard deviation of , say , 90% of the corresponding marginal distributions .
the total volume fraction spanned by a single sample in @xmath25 dimensions is then @xmath51 , an exponentially decreasing function with @xmath25 .
therefore it also takes an exponential number of samples in order to build up the full histogram , and this becomes computationally unfeasible for realistic data sets already at @xmath52 @xcite .
the main problem with this approach is that one attempts to map out all possible @xmath28-point correlation functions between all multipoles .
the number of such @xmath28-point functions is obviously overwhelming with increasing dimensionality . but
this also hints at a possible resolution of the problem : we know by experience that the cmb likelihood is a reasonably well behaved function , in that 1 ) there are only weak correlations between multipoles for data sets with nearly full - sky coverage , and 2 ) that even including just two - point correlations ( in transformed variables ) produces very reasonable results ( e.g. , bond , jaffe & knox 2000 ; verde et al . 2003 ) .
this intuition will be used in the next section to define a stable likelihood estimator .
we now introduce a new gibbs - based likelihood estimator we call the `` gaussianized blackwell - rao estimator '' ( gbr ) . the basic idea behind this approach is similar to that employed by , e.g. , @xcite and @xcite , namely approximation by a multivariate gaussian in a transformed set of variables . explicitly , our approximation is defined by transforming the univariate marginal distributions @xmath29 into gaussianized variables , @xmath53 , and then assuming a multivariate gaussian distribution in these transformed variables , @xmath54 here @xmath55 is the jacobian of the transformation , and @xmath56 is a gaussian random vector with mean @xmath57 and covariance matrix @xmath58 .
thus , the approximation of our likelihood estimator relies on the assumption that @xmath59 note that this is by construction exact for the full - sky uniform noise case , because the covariance matrix in this case is diagonal , and the full expression factorizes in @xmath0 ; in that case we are only performing an identity operation .
the first step in our algorithm is to compute the change - of - variables rule from @xmath6 to @xmath53 that transforms the marginal distribution , @xmath29 , for each @xmath0 into a gaussian distribution , @xmath60 .
the data used for this process are the @xmath44 samples drawn from the joint posterior @xmath61 by the cmb gibbs sampler .
we use two different methods of estimating the marginal distributions from these samples .
the first approach is to estimate @xmath29 with the blackwell - rao estimator as defined by equation [ eq : br ] , over a grid in @xmath6 for each @xmath0 .
then , a cubic spline is fitted to the resulting distribution .
this is the preferred approach for high signal - to - noise or low-@xmath0 modes .
however , for low signal - to - noise and high-@xmath0 modes one observes similarly poor convergence properties of this marginal estimator as for the full joint estimator . in these cases
we therefore instead compute a simple histogram directly from the @xmath6 samples , and fit a smooth spline @xcite through this histogram .
for further stability , we also produce @xmath62 @xmath6 samples from @xmath48 based on the same @xmath44 set as used for the br estimator .
this essentially corresponds to computing the blackwell - rao estimator by monte carlo , and the computational cost of producing these extra samples is small .
( the computational expense of the gibbs sampler is driven by sampling from @xmath63 , not by @xmath43 . )
note that this approach naturally supports arbitrary @xmath6 binning schemes @xcite , and also interfaces naturally with the hybrid mcmc scheme described by @xcite . given these spline approximations
to @xmath29 for each @xmath0 , we compute the corresponding cumulative distributions by numerical integration , @xmath64 this is subsequently identified with a standard gaussian distribution with zero mean and unity variance .
explicitly , we find @xmath65 over a grid in @xmath66 such that @xmath67 where @xmath68 is the error function .
this equation is straightforward to solve using standard numerical root - finding routines .
the result is a convenient set of look - up tables @xmath69 , again stored in the form of cubic splines , that allows for very efficient transformation from standard to gaussian variables for arbitrary values of @xmath6 . from these splines
, it is also easy to compute the derivatives required for the jacobian in equation [ eq : transformation ] .
having defined a change - of - variables for each @xmath0 , the remaining task is to estimate the joint distribution , @xmath70 , in the new variables . in this paper , we approximate this distribution by a joint gaussian , but any parametric function could of course serve this purpose . for example
, we implemented support for the skew - gaussian distribution ( e.g. , * ? ? ?
* ) in our codes , but found that the improvement over a simple gaussian was very small .
the only free parameters in this multivariate gaussian distribution are the mean , @xmath71 , and the covariance , @xmath72 .
these are again estimated from the samples produced by the gibbs sampler .
first , we draw @xmath73 @xmath6 samples from @xmath48 , as described above , but this time including all @xmath0 s for each sample
. then we gaussianize these @xmath0-by-@xmath0 , by evaluating @xmath69 for each sample and multipole moment .
finally , we compute the corresponding means and standard deviations , @xmath74 where the sums run over sample index .
before applying the machinery described in the previous section to the 5-year wmap data , we verify the method by a analyzing a simulated low - resolution data set .
the reason for considering a low - resolution simulation is that only in this case is it possible to evaluate the exact likelihood by brute force in pixel space , without making any approximations .
the simulation is made by drawing a gaussian realization from the best - fit 5-year wmap @xmath9cdm power spectrum @xcite , smoothing this with a @xmath75 fwhm gaussian beam , and projecting it on an @xmath76 healpix grid .
finally , @xmath77 rms white noise is added to each pixel , and the ( degraded ) wmap kq85 sky cut @xcite is applied to the data .
the maximum multipole considered in this analysis was @xmath78 , and the spectrum was binned with a bin size of @xmath79 from @xmath80 .
the signal - to - noise is unity at @xmath81 , and negligible beyond @xmath82 .
we now compute slices for each @xmath0 through the full multivariate likelihood , both with the method described in [ sec : method ] and by brute - force pixel space evaluation ( e.g. , * ? ? ?
* ) , fixing all other @xmath0 s at the input @xmath9cdm spectrum .
for comparison , we also compute the the marginal distributions for each @xmath0 .
the results from this exercise are shown in figure [ fig : verification ] .
black lines indicate the brute - force likelihoods , and red lines show the gaussianized blackwell - rao likelihoods .
the green lines show the marginal distributions , visualizing the effect of mode coupling due to the sky cut .
first , we see that all distributions agree very closely at @xmath83 . in this very large - scale regime ,
all harmonic modes are sufficiently well sampled with the kq85 sky cut that mode coupling is negligible . however , from @xmath84 the marginal distributions are noticeably different from the likelihood slices , with a typical shift in peak position of @xmath85 s .
we also see that these correlations are accurately captured by the gaussian approximation implemented in the gbr estimator , as the gbr likelihoods are essentially identical to the brute - force slices up to @xmath86 . at the very high @xmath0 and low signal - to - noise end , we see slight differences between the gbr and the pixel space slices , and in fact , the agreement is better with the marginal distributions .
this is caused by poor convergence of the covariance matrix in this particular run , and is included here for pedagogical purposes only : in a real analysis , one must always make sure that all distributions have converged well , typically by analyzing different chain sets separately .
note also that with sufficiently wide bins , the correlations to neighboring bins eventually vanish , and in this case it may be better to remove these correlations by hand from the covariance matrix , rather than trying to estimate them by sampling . whether this is the case or not for a given set can again be estimated by jack - knife tests . finally , for the 5-year wmap analysis presented in this paper
, we will only use the gbr estimator in the high signal - to - noise regime , and in that case the distributions converge very quickly .
we now apply the tools described in [ sec : method ] to the 5-year wmap temperature data .
we only consider @xmath8 in this paper , to avoid issues with error propagation for unresolved point sources and beam estimation .
however , we do correct for the mean spectrum of unresolved point sources , as described below .
we analyze the foreground reduced 5-year wmap v - band temperature sky maps , which are available from lambda .
the v - band data was chosen because these are considered to be the cleanest in terms of foregrounds out of the five wmap bands @xcite .
further , at @xmath8 the v - band alone is strongly cosmic variance dominated , and one does therefore not gain any significant statistical power by co - adding with other bands .
instead , one only increases the chance of introducing foreground biases by adding more frequencies .
we work with the individual differencing assembly ( da ) maps @xcite , and take into account the beam and noise pattern for each map separately .
the wmap sky maps are pixelized at a healpix resolution of @xmath87 , corresponding to a pixel size of @xmath88 , and the instrumental beam of the two v - band channels has a fwhm of 21. we therefore impose an upper harmonic mode limit of @xmath89 in the gibbs sampling ( commander ) step , probing deeply into the noise dominated regime .
note , however , that we only use @xmath8 in the gbr estimator , to avoid high-@xmath0 complications , such as beam and point source error propagation , in the cosmological parameter stage .
we correct the spectrum for unresolved point sources using the wmap model .
explicitly , the mean spectrum due to unresolved point sources in a single frequency , @xmath90 , for the 5-year wmap data is modelled as @xcite @xmath91 where @xmath92 is the point source amplitude relative to the q - band channel ( @xmath93 ) , @xmath94 is the best - fit spectral index of the point sources , and @xmath95 is the conversion factor between antenna and thermodynamic temperature units . to correct for this in our analysis ,
we subtract @xmath96 , evaluated at @xmath97 , from each @xmath44 sample before computing the gbr estimator .
finally , we impose the wmap kq85 sky cut @xcite on the data that masks point sources , removing 18% of the sky .
note that we adopt the template corrected maps provided by the wmap team in this analysis , and postpone an internal gibbs sampling based foreground analysis to a future paper ; for now our main focus is the new likelihood approximation , not the impact of foregrounds .
the analysis consists of the following steps : 1 .
generate 4000 @xmath44 samples with commander from the 5-year v1 and v2 differencing assemblies , including @xmath0 s up to @xmath89 , divided over 8 chains .
generate 500000 @xmath6 samples from these @xmath44 s , including @xmath0 s between @xmath98 and 250 .
compute the corresponding gbr parameters , i.e. , transformation tables , means @xmath71 and covariance matrix @xmath72 .
4 . modify the 5-year wmap temperature likelihood by replacing the existing low-@xmath0 part with equation [ eq : transformation ] , with the parameters given in ( 3 ) .
the transition multipole between the low-@xmath0 and high-@xmath0 is increased from @xmath99 to 200 .
multipoles between @xmath100 and 250 are included in the gbr estimator to avoid truncation effects , but the spectrum in this range is kept fixed at a fiducial spectrum , in order not to count these multipoles twice .
cosmological parameters are estimated using cosmomc @xcite . before presenting the results from the wmap analysis
, we consider the question of convergence .
first , we compute the gelman - rubin statistic @xcite for each @xmath44 using the eight chains computed with commander and removing the first 20 samples for burn - in .
we find that @xmath101 is less than 0.01 for @xmath102 and less than 1.1 for @xmath103 , indicating very good convergence in terms of power spectra .
however , the fact that each @xmath44 individually is well converged does not automatically imply that the full likelihood is well converged , since the latter depends crucially on the correlations between @xmath44 s . to assess the convergence in terms of cosmological parameters , we therefore analyse a toy model , by fitting a simple two - parameter amplitude and tilt , @xmath104 and @xmath105 , model , @xmath106 to the wmap data between @xmath98 and 250 with the gbr likelihood . here
@xmath107 is a fiducial power spectrum , which is chosen to be the best - fit 5-year wmap @xmath9cdm power spectrum @xcite , and @xmath108 .
we then map out the likelihood in a grid over @xmath104 and @xmath105 .
this is repeated twice , first including samples from chains number 1 to 4 and then from chains number 5 to 8 .
the results from this exercise are shown in figure [ fig : qn_model ] in terms of two sets of likelihood contours , corresponding to each of the two chain sets , respectively .
the agreement between the two is excellent , indicating that we also have good convergence in terms of cosmological parameters with the existing sample set .
note also that the point @xmath109 lies well inside the @xmath110 confidence region , indicating that the best - fit wmap model , which is obtained including @xmath0 s between @xmath98 and 1024 , also is a good fit to @xmath98 to 250 separately .
third , as described in
[ sec : method ] , we construct the gbr covariance matrix from @xmath111 @xmath6 samples drawn from the ( smaller ) set of @xmath44 samples . an outstanding question is how large @xmath28 should be in order for this covariance matrix to reach convergence , as a function of @xmath25 . to settle this question ,
we carry out the following simple exercise : first we produce two @xmath6 sample sets , each containing @xmath28 samples , and all drawn from a single @xmath44 sample .
second , we compute the two corresponding covariance matrices , invert these , then subtract them from each other , and finally compute the standard deviation of all elements .
third , we define the inverse covariance matrix to be converged if the rms mcmc noise is less than 0.005 , corresponding to 0.5% of the diagonal elements .
( we have checked that this produces robust parameter estimates . )
we then find the smallest @xmath28 such that this is satisfied , as a function of @xmath25 .
the results from this exercise are shown in figure [ fig : covar_convergence ] . here
we see that the number of samples required for convergence rises rapidly up to @xmath112 , reaching a maximum of @xmath113 samples , and then flattens to a plateau .
to be on the safe side , we therefore always use either @xmath114 or @xmath115 samples in the wmap analysis .
the reason for this behaviour becomes intuitive when considering the structure of the actual matrix .
this is shown in figure [ fig : covar ] , on the form of a correlation matrix @xmath116 the main features of this matrix are negative correlations around the diagonal , with the largest amplitudes observed between @xmath0 and @xmath117 .
this is expected : first , two modes separated by @xmath118 have different parity , and can therefore not easily mimic each other . on the other hand , modes separated by @xmath119 have both identical parity and similar angular scale , and it is therefore possible to add power to one mode and subtract it from the other , and still maintain an essentially unchanged image .
the result is a noticeable anti - correlation between @xmath0 and @xmath117 . at larger separations in @xmath0 ,
the correlations die off rapidly , since it is difficult for a large - scale mode to mimic a small - scale mode with a reasonably small sky cut . and
this explains the convergence behaviour seen in figure [ fig : covar_convergence ] : the covariance matrix is strongly band - limited .
therefore , once one has a sufficiently large number of @xmath6 samples for a sub - block to converge , there is also enough samples for a sub - block further away to converge .
these are essentially uncorrelated .
we now present the main results derived from the 5-year wmap temperature data with the gbr estimator between @xmath98 and 200 .
first , in the top panel of figure [ fig : powspec ] we plot the power spectrum obtained by maximizing the gbr likelihood together with the official 5-year wmap power spectrum .
the bottom panel shows the difference between these two , and the gray band indicates the standard deviation of @xmath44 , i.e. , the uncertainty due to noise and sky cut , but not to cosmic variance . clearly , the agreement between the two power spectra is very good .
next , in figure [ fig : likelihood_comparison ] we compare a few selected slices through the gbr likelihood with slices through the wmap likelihood . all other @xmath0 s than the one currently considered are kept fixed at the best - fit @xmath9cdm spectrum . here
we see that there are small shifts in peak positions , corresponding to the small differences seen in the power spectra in figure [ fig : powspec ] .
however , a main point in this plot is that the gbr likelihood slices are well behaved even at the highest @xmath0 s , and this is not the case for the standard br estimator @xcite .
finally , in table [ tab : parameters ] and figure [ fig : parameters ] we summarize the marginal cosmological parameter posteriors obtained with the two likelihood codes from cosmomc .
interestingly , there are some notable differences at the 0.3@xmath12 level , with the most striking example being the spectral index of scalar perturbations , @xmath120 .
this is only @xmath11 away from unity , and @xmath12 higher than the official wmap values .
we have presented a new likelihood approximation to be used within the cmb gibbs sampling framework .
this approximation is defined by gaussianizing the observed marginal power spectrum posteriors , @xmath29 , through a specific change - of - variables , and then coupling these univariate posteriors into a joint distribution through a multivariate gaussian in the new variables .
this process is exact , i.e. , an identity operation , in the uniform and full - sky coverage case , and it is also an excellent approximation in for the moderate sky cuts relevant to satellite missions such as wmap and planck .
our new approach relies on the previously described cmb gibbs sampling framework @xcite , and thereby inherits many important advantages from that .
first and foremost , this framework allows for seamless propagation of uncertainties from various systematic effects ( e.g. , foregrounds , beam uncertainties , calibration or noise estimation errors ) to the final cosmological parameters .
this is not straightforward in the hybrid scheme used by the wmap code .
second , this new approach corresponds to the exact low-@xmath0 pixel space likelihood part of the wmap code , not the approximate high-@xmath0 master part .
still , our method can handle arbitrary high @xmath0 s .
third , once the one - time pre - processing step has been completed , the computational expense of our estimator is determined by the cost of @xmath25 spline evaluations , while a pixel space approach requires a matrix inversion , and therefore scales as @xmath121 . for the cases considered in this paper , the cpu time required for the gbr wmap estimator up to @xmath122 was @xmath123 milliseconds , while it was @xmath5 seconds for the pixel space approach up to @xmath99 , for a map with @xmath124 pixels .
in order to validate our estimator , we applied it to a low - resolution simulated data set , and compared it to slices through the exact joint likelihood as computed by brute - force evaluation in pixel space .
the agreement between the two approaches was excellent .
we then applied the same estimator to the 5-year wmap temperature data , and estimated both a new power spectrum and new cosmological parameters within a standard six - parameter @xmath9cdm model .
the results from these calculations are interesting .
first , our power spectrum is statistically very similar to the official wmap spectrum , with no visible biases seen and relative fluctuations within the level predicted by noise and sky cut .
nevertheless , we do find significant differences in terms of cosmological parameters , and most notably in the spectral index of scalar perturbations , @xmath15 .
specifically , we find @xmath125 , which is only @xmath11 away from unity and @xmath12 higher than the official wmap result , @xmath126 .
this result resembles very much the outcome of a re - analysis we did with the 3-year wmap temperature data @xcite , for which we found a bias of @xmath16 in @xmath15 compared to the official wmap results .
this bias was due to the sub - optimal master - based likelihood approximation @xcite used by the wmap team between @xmath14 and 30 , whereas we used an exact estimator in the same range .
this study later prompted the wmap to change their codes to use an exact likelihood evaluator up to @xmath17 .
in the same study , we also tried to increase the @xmath0-range for our exact estimator to @xmath127 , but found small differences .
we therefore concluded , perhaps somewhat prematurely , that an exact estimator up to @xmath17 was sufficient for obtaining accurate results . on the contrary , in this paper we find
still find significant changes when increasing the exact estimator up to @xmath122 .
in retrospect , this should perhaps not come as a complete surprise , when realizing that the impact on a particular cosmological parameter typically depends logarithmically on @xmath0 .
for instance , @xcite considered a simple power spectrum model with a single free amplitude , @xmath128 , and found that , for a given likelihood estimator to be `` statistically unbiased '' , the systematic errors in that same estimator must fall off faster than @xmath129 .
cccc @xmath130 & @xmath131 & @xmath132 & 0.4 + @xmath133 & @xmath134 & @xmath135 & -0.3 + @xmath136 & @xmath137 & @xmath137 & 0.0 + @xmath138 & @xmath139 & @xmath140 & 0.3 + @xmath15 & @xmath141 & @xmath142 & 0.6 + @xmath143 & @xmath144 & @xmath144 & 0.0 a similar consideration holds for @xmath15 .
intuitively , @xmath15 is as much affected by @xmath98 to 10 as it is between @xmath80 and 100 . in the previous 3-year wmap re - analysis paper , we increased the range of the accurate likelihood estimator from @xmath14 to 30 , corresponding to a factor of 2.5 in @xmath0 , and removed a bias of @xmath145 in @xmath15 . in this paper
, we increase the range from @xmath17 to 200 , corresponding to a factor of 6.7 in @xmath0 , and find an additional bias of @xmath12 .
however , increasing @xmath0 from 30 to 50 corresponds only to a factor of 1.7 in @xmath0 , and this appears to be too small to produce a statistically significant result .
the main conclusions from this work are two - fold .
first , it seems that an accurate likelihood description is required to higher @xmath0 s than previously believed , and at least up to @xmath122 , in order to obtain unbiased results . by extrapolation
, it also does not seem unlikely that even higher multipoles should be included .
this issue will be revisited in a future publication .
our second main conclusion is that we find a spectral index only @xmath11 away from unity , namely @xmath146 . to us
, it therefore seem premature to make strong claims concerning @xmath147 ; the statistical significance of this is rather low , and there are likely still unknown systematic errors in this number . in a future publication
we will generalize the gbr estimator to polarization .
once completed , this will enable a fully gibbs - based cmb likelihood analysis at low @xmath0 s , and remove the need for likelihood techniques based on matrix operations , i.e. , inversion and determinant evaluation .
the computational cost of a standard cosmological parameter mcmc analysis ( e.g. , cosmomc ) will then once again be driven by the required boltzmann codes ( e.g. , camb or cmbfast ) and not by the likelihood evaluation . in turn , this will increase the importance of fast interpolation codes such as pico @xcite or cosmonet @xcite .
with such fast algorithms for both spectrum and likelihood evaluations ready at hand , the cpu requirements for cosmological parameter estimation may possibly be reduced by orders of magnitude .
we thank tony banday , ben wandelt and graca rocha for useful and interesting discussions .
we acknowledge use of the healpix software @xcite and analysis package for deriving the results in this paper .
we acknowledge use of the legacy archive for microwave background data analysis ( lambda ) .
this work was partially performed at the jet propulsion laboratory , california institute of technology , under a contract with the national aeronautics and space administration .
r , neg and hke acknowledge financial support from the research council of norway . | we introduce a new cmb temperature likelihood approximation called the gaussianized blackwell - rao ( gbr ) estimator .
this estimator is derived by transforming the observed marginal power spectrum distributions obtained by the cmb gibbs sampler into standard univariate gaussians , and then approximate their joint transformed distribution by a multivariate gaussian .
the method is exact for full - sky coverage and uniform noise , and an excellent approximation for sky cuts and scanning patterns relevant for modern satellite experiments such as wmap and planck .
the result is a stable , accurate and computationally very efficient cmb temperature likelihood representation that allows the user to exploit the unique error propagation capabilities of the gibbs sampler to high @xmath0 s .
a single evaluation of this estimator between @xmath1 and 200 takes @xmath2 cpu milliseconds , while for comparison , a singe pixel space likelihood evaluation between @xmath3 and 30 for a map with @xmath4 pixels requires @xmath5 seconds .
we apply this tool to the 5-year wmap temperature data , and re - estimate the angular temperature power spectrum , @xmath6 , and likelihood , @xmath7 , for @xmath8 , and derive new cosmological parameters for the standard six - parameter @xmath9cdm model .
our spectrum is in excellent agreement with the official wmap spectrum , but we find slight differences in the derived cosmological parameters .
most importantly , the spectral index of scalar perturbations is @xmath10 , @xmath11 away from unity and @xmath12 higher than the official wmap result , @xmath13 .
this suggests that an exact likelihood treatment is required to higher @xmath0 s than previously believed , reinforcing and extending our conclusions from the 3-year wmap analysis . in that case
, we found that the sub - optimal likelihood approximation adopted between @xmath14 and 30 by the wmap team biased @xmath15 low by @xmath16 , while here we find that the same approximation between @xmath17 and 200 introduces a bias of @xmath12 in @xmath15 . |
interaction of a two - level system with a plane traveling wave leads to one - photon transitions between energy levels and consequent changes of total atomic momentum in limits of one photon momentum @xmath1 .
what can be said about distributions and mean values of momentum for each individual atomic level ?
the answer is well - known and is trivial , if the atom before the interaction is on one of the energy levels : distribution on the other level sets shift @xmath1 and mean values of momentum distinguished also by @xmath0 ; @xmath2 , where @xmath3 and @xmath4 are mean values of momentum on ground and excited levels consequently ( @xmath5 case).what would we have in general case , that is , when the atom before the interaction with the travelling wave is in the superposition state of ground and excited levels with mutually different momentum distributions there ? in the following just this question will be elucidated in details : in general form in sec . 2 , and the special case of preparing standing wave , latter .
2 it will be shown that in one - photon absorption / emission process , in general , large scale redistributions of the energy level s momenta takes place .
this entails the time - changes of the @xmath6 @xmath7 and these changes can greatly exceed the value of one photon momentum . in general , the relations @xmath2 are not true , too . in sec .
3 we will discuss from the practical point of view a very important case , when the preliminary superposition state of the atom is realized by the coherent diffraction of the atom in the field of resonant standing wave , which is being often referred to us the resonant kapitza - dirac effect .
will be pointed out , that the redistribution of momenta in the travelling can be received as a transition from the resonant kapitza - dirac splitting to the stern - gerlach type splitting . in sec .
4 will be discussed the temporal behavior of the energy level s mean momenta .
the results are summarized in sec . 5 , where the possibility of experimental observation of this phenomenon will be sketched too .
let s discuss the resonant interaction of a two - level atom with the radiation field @xcite .
for the sake of simplicity , suppose the field has plane wavefront , linear polarization ( these assumptions will be conserved for the standing wave , discussed in sec.3 ) , turns on instantly and after on it s amplitude remains constant .
let the wave functions of free two - level atom s ground ( @xmath8 ) and excited ( @xmath9 ) levels be @xmath10 and @xmath11 respectively , where@xmath12 is atomic internal coordinate ( radius - vector of the optical electron relative to the atomic center - of - mass )
. the wave function of interacting atom will be @xcite @xmath13 where @xmath14 and @xmath15 are the probability amplitudes of the atom to be on the ground and excited levels correspondingly .
when taking into account translational motion of the atomic center of mass , it is necessary to separate respective parts ( wave functions ) in @xmath14 and @xmath15 coefficients .
if , for example , the atom on thw energy level has the well - defined value of momentum @xmath16 , the respective wave function is given by function @xmath17 that is , by the exponential function with imaginarily degree .
in general , when on energy levels an atom hasnt definite values of momentum , @xmath14 and @xmath15 coefficients will be expressed by the series of @xmath18-states : @xmath19 with probability amplitudes @xmath20 and @xmath21 of atom to have momentum @xmath22 ( at the @xmath23 moment of time ) , simultaneously being on the ground or excited energy levels , correspondingly . inserting the expressions ( [ 1])-([3 ] ) into the quantum mechanical determination of atom momentum @xmath24 and doing the standard transformations we arrive to @xmath25 the first member specifies the contribution of the ground energy level in total momentum , @xmath26 accordingly
, the second member specifies the excited level s contribution , @xmath27 both these momenta are time - dependent and their change for @xmath23 time of interaction will be @xmath28 when the atom interacts with the travelling wave , the coefficient @xmath20 of ground level relates with the coefficient @xmath29 of excited level ( spontaneous transitions are not taken into account ) ( fig .
[ fig11 ] ) . as a result
we get @xmath30 by this relation we can connect @xmath31 with @xmath32 : @xmath33 where @xmath34 @xmath35 is the change of ground level s population , or which is the same , the population change @xmath36 of excited level with the opposite sign ( see [ 16 ] ) . from the equality of the first and last parts of ( [ 8 ] )
directly follows the well known inequality between the momentum of photon and total atom : @xmath37 let , nevertheless , note that this one photon demarcafationpertains to the total momentum of the atom , but not to the momentum of ground and excited levels , separately .
their changes , in accordance with ( [ 6a ] ) and ( [ 6b ] ) , in principle , may be arbitrary , depending on distributions of @xmath38 and @xmath39 in momentum space . from the expressions ( [ 6a ] ) and ( [ 6b ] ) also is obvious that for @xmath40 @xmath41 to get great values , is necessary the distribution of @xmath38 ( of @xmath39 ) to be strictly non - symmetric , relative to the replacement @xmath42 and to have a gathering in the range of great values of @xmath43 . and now let s show that one photon absorption/ emission process in the field of travelling wave really allows a behavior , mentioned above .
the hamiltonian of the system , in dipole approximation , may be written as @xmath44 where @xmath45 is the free atom hamiltonian , and @xmath46 is the dipole moment operator and @xmath47 is the electric field , whose @xmath48 frequency is equal to the @xmath49 frequency of bohr transition . from schrodinger equation for @xmath50 and
@xmath51 amplitudes we arrive to @xmath52 the rabi - solutions of which are @xcite @xmath53 where @xmath54 represents the rabi frequency , @xmath55 .
performing @xmath56-expansion ( see ( 3 ) ) in ( [ 13a ] ) and ( [ 13b ] ) , we obtain @xmath57 at first , it is readily verified that if the atom is on one of energy levels before the interaction , the extraordinary things does nt take place . really , if for example @xmath58 , then @xmath59 , \label{15b}\end{aligned}\ ] ] that is the contribution of momentum per energy level evolves periodically and this is the evolution merely caused by periodic exchange of population between the energy levels ( posed by the term @xmath60 .
note also that in conditions under consideration the momentum distributions coincide with each other with a shift @xmath1 : @xmath61 @xmath20 tg@xmath62 , as was mentioned in introduction .
the situation is totally diverse , if the atom is initially in superposition state of ground and excited levels , because now the initial momentum distributions on the ground and excited levels are not under necessity to be identical with shift @xmath1 : @xmath63 @xmath64 in general ( @xmath65 is some constant , independent of @xmath16 ) .
then , it follows unavoidably from ( [ 14a ] ) and ( [ 14b ] ) that the optical transition , besides the changes on the energy levelspopulations , leads also to periodic evolutions in the form of momentum distributions there . herewith the atomic amplitudes @xmath20 and @xmath21 are nt mutually proportional ( with any constant shift ) . to wash out the contributions , appropriate to evolution of the energy level populations , let introduce a pair of new quantities , @xmath3 and @xmath66 , which would be scaled in units of level - populations @xmath67 and @xmath68 respectively : @xmath69 @xmath70 since these new quantities are already independent on level populations , their possible evolutions would be stipulated by form - deformations in the energy levels momentum distributions .
afterwards we will call them mean momenta of ground energy level ( @xmath3 ) and of excited energy level ( @xmath4 ) correspondingly .. thereby the total momentum of atom , in addition to ( [ 5 ] ) , can be presented in the following more convenient form : @xmath71 these mean momenta , @xmath3 and @xmath4 , remain constant , of course , if the atom is initially on one of the energy levels .
they remain constant also when the initial distributions are mutually proportional with constant shift @xmath1 : @xmath72 really , putting ( [ 18 ] ) in relations ( [ 14a ] ) and ( [ 14b ] ) making the obvious substitutions , we arrive to @xmath73 for ground energy level , and @xmath74 for excited energy level . in these circumstances in ( [
17 ] ) the time evolution exhibit only energy level populations @xmath68 and @xmath67 . in the general case , nevertheless , the evolution of state is due to interference of non - similarly distributed amplitudes , the atomic amplitudes distributions on energy levels are nt proportional with each other yet and , as a consequence , mean momenta @xmath4 and @xmath3 get temporal evolution , too . for acquisition of more concrete and quantitative results , let s note that for the intentions of atom optics and interferometry@xcite the coherent scattering of atoms in the resonant field of standing wave is the routine for preparation of large spreaded momentum distribution .
the probability amplitudes , prepared in a such way , ca nt satisfy the undesirablecondition ( [ 18 ] ) in principle , since in the field of standing wave .
as is well known , any state with momentum @xmath16 on one energy level is connected with the two - states on other energy level with momenta @xmath75 and @xmath76 simultaneously .
therefore , any atom prepared by means of resonant kapitza - dirac effect , during its later interaction with the travelling wave , ought to implicitly change the momentum distributions on energy levels with following to it above mentioned consequences .
let before the interaction with the travelling wave , during time @xmath77 , the atom had coherent interaction with the resonant ( @xmath78 ) standing wave @xcite .
we restrict ourselves to the relatively simple case , when the interaction proceeds by the well known scheme of mutually orthogonal atom - standing wave beams .
moreover , the raman - nath approximation will be employed , which permit to put out from the problem at hand the kinetic energy term in the hamiltonian ( note that the kinetic energy term has not been included into ( [ 10 ] ) too ) .
although the scheme of calculation is well known and presented in details ( see , for example , in @xcite , @xcite , we find it convenient to give an account of main intermediate formulas , too . to describe the interaction in the preparing standing wave , the electric field ( [ 11 ] ) in the hamiltonian ( [ 10 ] )
must be exchanged by @xmath79 in consequence , the atomic amplitudes @xmath80 and @xmath81 have to fulfill ( [ 12a],[12b])-type equations where the following replacements must be performed : @xmath82 ( which is mean rabi frequency in the standing wave ) , @xmath83 . allowing that the atom has been on the ground level before the interaction @xmath84 , we arrive to @xmath85 these amplitudes at the moment @xmath86 , when the standing wave is turned off , just present the initial amplitudes @xmath87 and @xmath88 for interaction with the travelling wave ( see the formulas ( [ 13a],[13b ] ) . using their @xmath18-expantions
@xcite . @xmath89 where @xmath90 is bessel function , for atomic center - of - mass motion probability amplitudes @xmath20 and @xmath21 we get the following expressions : @xmath91 , \label{22a } \\
b((2m+1)\hbar k , t ) & = & i^{2m+1}\left [ \cos \nu
t\text { } j_{2m+1}(2\nu _ s\tau _ s)+\sin
\nu t\text { } j_{2m1}(2\nu _ s\tau _ s)\right ] , \label{22b } \\
a((2m+1)\hbar k , t ) & = & b(2m\hbar k , t)=0 \label{22c}\end{aligned}\ ] ] the superposition state , created as a result of interaction with the standing wave , present the discrete mainfields of states , where the space between the adjacent values of momentum is @xmath92 , herewith the mainfields for ground and excited levels are totally shifted with respect to each other by @xmath1 ( the half of @xmath92 ) @xcite .
the formulas ( [ 22a])-([22c ] ) contain explicitly the seeking result about the evolution of momentum distributions . to exhibit this evolution ,
let first note that the initial momentum distribution for both energy levels are symmetric relative to value @xmath93 . really , they are specified by @xmath94 and @xmath95 functions for ground and excited energy levels respectively and are symmetric , relative to @xmath96 , @xmath97 transformations , that is just relative to value @xmath98 @xmath99 .
this symmetry signifies that the momentum of each energy level ( as the incremental ( [ 5 ] ) , as the mean ( [ 17 ] ) values ) is zero @xcite before the interaction with the travelling wave .
nevertheless the symmetry breaks under the influenceof travelling wave : one photon absorption / emission process , in accordance with ( [ 22a ] ) and ( [ 22b ] ) , gives the beginning of asymmetric transformations in the form of momentum distributions , periodically running in opposite directions for ground and excited energy levels . a typical form of initial distributions and the following redistributions ( due to single - photon process )
are depicted on fig .
[ fig22 ] and fig .
[ fig33 ] for ground and excited energy levels consequently .
single - photon large - scale changes are apparent .
now let notice that in conditions of fig .
[ fig22 ] and fig .
[ fig33 ] we get almost one - side distributions : the translational states with @xmath100 for ground energy level only , and the translational states with @xmath101 for excited level only .
so , the state of total atom has been explited into two sub - groups , where one sub - group presents ground - level atoms with negative values of momentum , and the second sub - group presents vice - versa , the excited - level atoms with negative values of momentum .
of course , this is a stern - gerlach type splitting .
that is , one - photon optical transition implements the resonant kapitza - dirac splitting into the stern - gerlach type splitting .
the phenomenon of one - photon coherent accumulation of momentum on energy levels ( op - camel ) , may get some expansion in exhibition , if the initial momentum distributions would be taken in asymmetric form .
this kind of distributions also can be obtained by the standing wave , but if some travelling wave is previous to it @xcite . such sequence of pulses is got , if the standing wave is formed by means of reflection of a laser pulse from the distanted ( from atomic beam ) mirror ( see , for example @xcite , ) . to avoid the overloading of the text we are nt going to bring formulas and the behavior of the op - camel in these conditions
will be given only by some graphs . in fig .
[ fig44 ] and fig .
[ fig55 ] is presented the case when the momentum distribution in resonant kapitza - dirac splitting is maximum asymmetric .
as is seen from the graphs , in this special case op - camel appears already as a accumulation of asymmetry on one ( ground ) energy level for account of its suppressing on the other ( excited ) energy level .
let us now return to the preparation only by the standing wave and discuss the evolution of momenta @xmath3 and @xmath4 . by means of expressions for the quantities defining @xmath3 and @xmath4 ( see ( [ 16 ] ) ,
( [ 16 ] ) and ( [ 5 ] ) , ( [ 5 " ] ) ) we will have @xmath102 ^2= \label{23 } \\
& = & -\hbar k\left [ \frac{1-j_0(2u)}2\sin ^2\nu t+\frac{u - j_1(2u)}4\sin 2\nu t\right ] \nonumber\end{aligned}\ ] ] @xmath103 ^2= \label{23 ' } \\ & = & \frac 12+\frac{j_0(2u)}2\cos 2\nu t-\frac{j_1(2u)}2\sin 2\nu t\text { } \nonumber\end{aligned}\ ] ] on the ground energy level , and @xmath104 ^2= \label{24 } \\ & = & \hbar k\left [ \frac{1+j_0(2u)}2\sin ^2\nu t+\frac{u+j_1(2u)}4\sin 2\nu t\right ] , \nonumber\end{aligned}\ ] ] @xmath105 ^2= \label{24 ' } \\ & = & \frac 12-\frac{j_0(2u)}2\cos 2\nu t+\frac{j_1(2u)}2\sin 2\nu t=1-n_g\text { } \nonumber\end{aligned}\ ] ] on the excited energy level . here
@xmath106 the last forms of ( [ 23])-([24 ] ) are got by using for formulas of summations of bessel functions @xcite ) . as @xmath107 , @xmath108 , ( the same for @xmath109 and @xmath110 ) ,
then their values at any next moment t present simultaneously their changes for the time @xmath23 : @xmath111 , @xmath112 . in fig .
[ fig66 ] and [ fig77 ] are given the temporal evolutions of momenta , corresponding to each energy level , while they are interacting with travelling ( accumulating ) wave .
population changes , which also are responsible for the evolutions in general , are depicted in figures by dashed lines . in presented case
the populations are constant in practice , which follows from ( [ 23 ] ) and ( [ 24 ] ) too , if we take into account @xmath113 for @xmath114 . and ,
finally , the temporal evolution of mean momenta @xmath109 and @xmath115 conditioned solaly by redistributions of momentumon the energy levels , is shown already in fig .
[ fig88 ] and [ fig99 ] .
the parameters of the preparing standing wave are the same as in fig .
[ fig22 ] , where the distance between left - hand and right - hand maximums ( the width of momentum distribution ) is about @xmath116 @xmath1 .
such magnitudes for the resonant kapitza - dirac splitting are totally in limits of experimental realizations @xcite .
note also , that the comparison of the deviation of @xmath109 or @xmath115 ( from the fig . [ fig88 ] and [ fig99 ] ) and the width of momentum distribution ( from the [ fig22 ] ) shows the same order of magnitude for them .
since the width of momentum distribution has multiphoton nature ( created by means of multiphoton process of reemission of photons from one wave into the counterpropagating one ) , the large -scale variations in op - camel may be called as multiphoton .
multiphoton op - camel manifests itself in ( [ 23 ] ) , ( [ 24 ] ) formulas in following way .
when the initial momentum distribution is sufficiently widespread , that is @xmath117 then @xmath118 ( from the theory of resonant kapitza - dirac effect the connection between the momentum width @xmath119 and the number of rabi - flops @xmath120 is @xmath121 ) ) . under this condition the members @xmath122 @xmath1 @xmath123 in ( [ 23 ] ) and @xmath124 @xmath1 @xmath125 in ( [ 24 ] ) , being proportional to @xmath126 , stand out as the previal terms and just present the multiphoton op - camel .
a simple theoretical consideration of optical transition in general conditions , when the atom initially has in superposition state of ground and excited energy levels different momentum distributions , shows that one one - photon optical transition leads to radical asymmetric changes in momentum distributions on each energy level . saying in images , in general one photon changes the mean momentum of energy level more than photon s own momentum .
in important case , when the preliminary superposition state of the atom is being prepared by coherent scattering at the resonant standing wave , the phenomenon can be presented as a transition from resonant kapitza - dirac splitting of atomic translational states to stern - gerlach type splitting .
this is schematically depicted in fig .
[ fig1010 ] . finally , let s give an account of some remarks about the possibility of experimental observation of phenomenon .
firs let s notice that the non - opticalmethods , which registrate the total atom ( for example hot - wiremethod ) , ca nt be used for this purpose , because the phenomenon deals with each individual energy level ; the momentum distribution of total atom does nt change or which will be more right , it changes only in one - photon momentum limits .
it is preferable to use registration methods , that will deal only with one of resonantly connected energy levels , such as the adjacent optical transitions .
then the phenomenon will appear itself as a pronounced asymmetry in profile of doppler broadening , relative to bohr frequency .
another possibility we see in using of long - living energy levels , so long that the atomic translational states can be distinguished in space till the spontaneous emission ( c zone in fig .
[ fig1010 ] ) . in this case
the space - sensitive schemes of spontaneous emission collection or probe pulse absorption will bring to desirable result .
l. d. landau and e. m. lifshitz quantum mechanics : nonrelativistic theory , nauka , moscow , 1963 , p.173 ( in russian ) ; l. allen and j. h. eberly optical resonance and two - level atoms , chap .
2 , wiley , new york , 1978 .
r. j. cook and bernhardt phys .
a * 18 * , 2533 ( 1978 ) ; a. p. kazantsev , g. i. surdutovich and v. p. yakovlev , pisma zh . eks . eks .
. fiz . * 31 * , 542 ( 1980 ) ; a. f. bernhardt and b. w. shore , phys .
rev . a * 23*,1290 ( 1981 ) ; j. dalibard and c. cohen - tannoudji , j. opt .
b * 2 * , 1707 ( 1985 ) ; a. zh .
muradyan , izvestia an arm ssr , fizika * 10 * , 361 ( 1975 ) ; p. j. martin , p. l. gould , b. g. oldaker , a. h. miklich and d. e. pritchard , phys .
a * 36 * , 2495 ( 1987 ) . | it is shown that a two - level atom , being initially in general superposition state of ground and excited energy levels with mutually different momentum distributions there , gets a large scale evolution in the energy levels momentum distribution . as a consequence
the mean momentum of each individual energy level also gets large scale changes , more than the own momentum @xmath0 of the photon .
thoroughly is discussed the special case , when the atom s preliminary superposition state is created as a result of interaction of the atom with the resonant standing wave .
also it is pointed that in such conditions the mentioned phenomenon can be presented as a transformation of the resonant kapitza - dirac splitting of atomic states into the stern - gerlach type splitting . |
this paper studies a particular example of the interaction between topology and number theory in the context of finite volume hyperbolic @xmath2-manifolds . if a _ noncompact _
hyperbolic @xmath2-manifold @xmath0 is irreducible and atoroidal , and homeomorphic to the interior of a compact @xmath2-manifold with torus boundary components , then @xmath0 admits a unique complete hyperbolic structure .
this structure determines a faithful representation from @xmath3 into @xmath4 which can be taken to have image in @xmath5 for some smallest number field @xmath6 , called the _ trace field _ of @xmath0 .
see @xcite , theorem 4.2.3 .
( note that the noncompactness of @xmath0 is important for the identification of @xmath6 with the trace field of @xmath3 . )
it is an important question to study which number fields @xmath6 can arise from this construction , and conversely , to understand the relationship between the topology of @xmath0 and the algebra of @xmath6 .
this question seems to be wide open , and very few nontrivial relationships are known . in this paper
we show by an elementary geometric argument that when @xmath0 is a hyperbolic once - punctured torus bundle over @xmath1 , the trace field @xmath6 has no real places ( i.e. it is not galois conjugate to a subfield of @xmath7 ) . in
2 we give an exposition of the relation between euler classes , stiefel - whitney classes , orientations , and boundary traces for hyperbolic surfaces and @xmath2-manifolds .
all this material is classical , but it seems that there is no explicit and thorough account of it in the literature which takes our particular viewpoint .
this discussion takes up more than half the length of the paper we feel that its inclusion is justified by its potential interest as a reference for the @xmath2-manifold community . note that some facts which might be otherwise somewhat obscure , are clarified by a thorough discussion of this foundational material .
in particular , it is immediate from our viewpoint that for a hyperbolic knot complement @xmath0 , the trace of a longitude is @xmath8 , after lifting a geometric representation @xmath9 to @xmath10 .
in 3 we specialize to once - punctured torus bundles . after a preliminary discussion of trace fields and invariant trace fields , we prove our main theorem : let @xmath11 be a hyperbolic oriented once - punctured torus bundle over @xmath1 with monodromy @xmath12 .
let @xmath6 denote the trace field of @xmath11 , and @xmath13 the invariant trace field .
then : 1 . @xmath6 admits no real places 2 .
if @xmath13 admits a real place , then @xmath14 contains 2-torsion in 4 we make some related observations .
most interesting is the observation that if @xmath0 contains a _ pseudo - rational surface _
i.e. a surface @xmath15 with all traces rational then such a surface has maximal euler class with respect to any @xmath16 representation .
in particular , a @xmath2-manifold containing a pseudo - rational surface which is not thurston norm minimizing in its homology class ( e.g. it might be separating ) has a trace field with no real place .
a nice corollary , pointed out by a. reid , is that this equality lets us construct examples of knot complements which contain no totally geodesic immersed surfaces , for example , the knot @xmath17 in rolfsen s tables .
we also give an example showing that for every @xmath18 and every @xmath19 with @xmath20 and @xmath21 there are incompressible surfaces @xmath22 with euler characteristic @xmath23 in some hyperbolic @xmath2-manifold @xmath0 with a real place @xmath24 for which @xmath25 .
the milnor - wood inequality implies that @xmath20 is sharp , and the mod 2 condition is implied by the fact that the representations are galois conjugate into a geometric representation , so these are all the possibilities which can arise .
somewhat amazingly , all these examples can be found in a single manifold @xmath0 , the complement of the link @xmath26 in rolfsen s tables .
thanks to j. button , n. dunfield , o. goodman , a. reid and the anonymous referee for comments , computations and corrections .
extra special thanks to a. reid for nursing this paper through an problematic gestation , and for the idea of corollary [ no_totally_geodesic ] .
in what follows , we frequently deal with complete noncompact hyperbolic surfaces @xmath15 with finite area . such surfaces are homeomorphic to the interior of a compact surface @xmath27 with boundary . by abuse of notation
, we will refer to the boundary components of the ( compactifying ) surface @xmath27 as the boundary components of @xmath15 . for the convenience of the reader
, we give a thorough exposition of the relationship between @xmath28 , stiefel - whitney , euler , and hyperbolic geometry .
related references are @xcite , @xcite .
let @xmath15 be a noncompact orientable surface of finite type , and let @xmath29 be a pseudo - anosov homeomorphism .
then the mapping torus @xmath30 admits a complete hyperbolic structure . corresponding to this hyperbolic structure there is a discrete faithful representation @xmath31 which is unique up to conjugacy and orientation .
we fix one such representation , and call it the _ geometric representation_. an orientable surface @xmath15 with negative euler characteristic itself admits a hyperbolic structure , and any such hyperbolic structure determines some discrete faithful representation @xmath32 for which the image of each boundary curve is parabolic . the space of discrete faithful representations from @xmath33 into @xmath4 up to conjugacy and orientation , and for which the image of each boundary curve maps to a parabolic element , is connected , and contains an open dense subset homeomorphic to a ball consisting of _ quasifuchsian _ representations . for each quasifuchsian representation
@xmath34 the ideal circle of the universal cover @xmath35 maps to a quasicircle in the sphere @xmath36 .
see @xcite and @xcite for background and details on the theory of kleinian groups . a surface @xmath15 might be orientable and yet admit representations into @xmath37 with nonorientable holonomy .
the group of all isometries of @xmath38 is @xmath39 which embeds into @xmath40 .
we may embed both @xmath16 and @xmath39 into the group @xmath41 by considering their action on the ideal circle @xmath42 of @xmath38 .
the image of @xmath16 is contained in @xmath43 ( the superscript @xmath44 denotes orientation preserving homeomorphisms ) .
these inclusion maps are homotopy equivalences .
the circle @xmath1 embeds in @xmath45 as the equator , and the group @xmath41 embeds in @xmath46 .
we have two commutative diagrams @xmath47 where the left diagram includes into the right diagram by a homotopy equivalence .
it is further true that the inclusions @xmath48 and @xmath49 where @xmath50 are open balls , obtained by coning to the center and throwing away the boundary , are homotopy equivalences .
the first case is straightforward ; the second follows from hatcher s proof of the smale conjecture @xcite and the ( homotopy ) equivalence of the categories diff and top in dimension @xmath2 . for any group @xmath51 , and any space @xmath52 ,
a representation @xmath53 induces a homotopy class of maps to the classifying space @xmath54 .
there is a tautological @xmath51 bundle over @xmath54 called @xmath55 which pulls back to a @xmath51-bundle over @xmath52 called @xmath56 .
topologically , we form @xmath56 as the quotient bundle @xmath57 where @xmath58 ranges over elements of @xmath59 .
the group @xmath43 is homotopy equivalent to the subgroup @xmath61 consisting of rotations .
@xmath41 is homotopy equivalent to the subgroup @xmath62 which is homotopic to two disjoint circles .
@xmath46 is homotopy equivalent to @xmath63 .
it follows that we can compute the homotopy groups of @xmath64 : @xmath65 @xmath66 and is torsion for @xmath67 .
cohomology on @xmath54 pulls back to cohomology classes on @xmath52 which represent the first obstruction to trivializing the bundle @xmath56 .
since @xmath68 is a @xmath69 and therefore has the homotopy type of @xmath70 , the cohomology ring is generated by a single element @xmath71 . identifying @xmath43 with @xmath16 up to homotopy
, this element represents the obstruction to lifting a representation from @xmath16 to @xmath72 , its universal covering group .
the mod 2 reduction of @xmath73 is the obstruction to lifting to @xmath74 .
since @xmath75 is a @xmath76 below dimension @xmath77 , only the class @xmath78 is relevant to @xmath2-manifolds . identifying @xmath46 with @xmath4 up to homotopy
, this represents the obstruction to lifting a representation from @xmath4 into @xmath10 .
if we include @xmath16 into @xmath4 then we see that @xmath79 is the image of @xmath73 under @xmath80 .
the short exact sequence @xmath82 gives rise to a fibration of spaces @xmath83 which exhibits @xmath81 up to homotopy as a twisted @xmath70 bundle over @xmath84 .
the generator of @xmath85 acts on @xmath86 by multiplication by @xmath87 .
since this multiplication is trivial with @xmath88 coefficients , the @xmath88 cohomology can be computed from the knneth formula . in low dimensions we get @xmath89 the generator in dimension
@xmath90 is the orientation class @xmath91 , and the generators in dimension @xmath92 are @xmath93 and @xmath94 which is the mod 2 reduction of the euler class , and is obtained by pulling back @xmath79 from @xmath95 .
suppose we have a pair of spaces @xmath96 and a representation @xmath53 such that @xmath97 is trivial .
then the resulting map to the classifying space maps @xmath98 to the basepoint of @xmath54 , and there is a well - defined relative homotopy class of pairs .
this defines a canonical trivialization of the restricted bundle @xmath99 .
it follows that we can pull back reduced cohomology of @xmath54 to a relative class in @xmath100 which represents the obstruction to extending this trivialization of @xmath56 over @xmath98 to all of @xmath52 .
more generally , suppose @xmath51 contains a closed contractible subgroup @xmath101
. then the coset space @xmath102 is homotopy equivalent to @xmath51 , and the bundle @xmath56 can be replaced by a homotopy equivalent bundle @xmath103 .
a representation @xmath104 for which @xmath105 is contained in @xmath101 defines a canonical trivialization of @xmath103 , and therefore a relative cohomology class in @xmath100 . in the groups @xmath16 and @xmath4 the maximal parabolic subgroups
are contractible ; these are the @xmath106 subgroups with respect to the @xmath107 ( or iwasawa ) decomposition of these groups . in @xmath43
the stabilizer of a point is contractible . in @xmath46 , the stabilizer of a point
is not contractible , being homotopic to @xmath1 . for @xmath15 a closed orientable hyperbolic surface , the euler class @xmath108 of the representation @xmath109 associated to any hyperbolic structure on @xmath15 satisfies @xmath110 ) = \pm \chi(s)\ ] ] where @xmath111 $ ] represents the fundamental class in @xmath112 , and the sign depends on the choice of orientation .
note in this case that @xmath113 is isomorphic to the unit tangent bundle of @xmath15 .
if @xmath15 is orientable and complete with finite area , but possibly with punctures , then there is a relative fundamental class in @xmath114 .
if @xmath113 represents the associated circle bundle , then @xmath113 restricts to a finite union of tori fibering over the boundary components of @xmath15 . by our assumption ,
the holonomy around a boundary component is parabolic , and has a ( unique ) fixed point in @xmath1 .
this fixed point suspends to a canonical section of @xmath113 over @xmath115 , and defines a trivialization of this restricted bundle .
the _ relative euler class _ of @xmath116 is the element of @xmath117 which represents the obstruction to extending this trivialization of @xmath118 over all of @xmath113 . in @xcite ,
thurston defined a @xmath92-cocycle for a group @xmath51 acting in an orientation - preserving way on @xmath1 .
if @xmath119 is a representation , choose a point @xmath120 and for every triple @xmath121 define @xmath122 equivalently , think of @xmath1 as the ideal boundary of @xmath38 .
then for @xmath123 a triple of elements in @xmath51 , define @xmath124 to be the ( signed ) area of the ideal triangle with vertices at the @xmath125 , divided by @xmath126 . from this geometric definition ,
it is easy to see that @xmath127 is coclosed , and that the cohomology class it defines is independent of @xmath128 .
suppose that @xmath129 and @xmath130 is @xmath131 for each component @xmath132 of @xmath133 . if each @xmath134 has a fixed point in @xmath1 , then for some choice of basepoint , @xmath127 is identically zero on @xmath134 .
it follows that @xmath127 defines a relative class in @xmath135 .
[ thurston_and_euler ] the thurston cocycle @xmath127 is related to the euler class @xmath73 by @xmath136 = 2[e ] \text { in } h^2(m,\partial m;\z)\ ] ] for a proof , see e.g. @xcite . for more details , and related constructions ,
see @xcite or @xcite .
for a closed hyperbolic @xmath2-manifold , the @xmath45 bundle coming from the geometric representation @xmath137 is isomorphic to the unit tangent bundle ( to see this , use the exponential map ) .
the pullback of the generator of @xmath138 can therefore be identified with the second stiefel - whitney class of @xmath0 .
if @xmath0 is orientable , @xmath139 is parallelizable , so this class must vanish .
if @xmath0 is noncompact with finite volume , the geometric representation on each boundary torus group is parabolic , with a fixed point @xmath140 .
the subgroup @xmath141 of parabolic elements fixing @xmath128 is contractible , so there is a trivialization of the associated bundle over @xmath133 and we get a relative second stiefel - whitney class @xmath142 which might not be trivial .
note that the image of @xmath143 in @xmath144 is the ordinary second stiefel - whitney class which is always trivial for orientable @xmath0 , as above .
it follows from the long exact sequence in cohomology that @xmath143 is the image of a distinguished class in @xmath145 , where @xmath146 denotes the homomorphism induced by the inclusion map @xmath147 .
since the ordinary second stiefel - whitney class vanishes , the geometric representation lifts to the double cover @xmath148 since every element of @xmath149 is parabolic , the preimages either have trace @xmath92 or @xmath8 .
different choices of lift to @xmath10 might change the value on elements of @xmath150 which are in the image of @xmath151 but do not change the values on elements which are homologically nontrivial in @xmath133 but bound in @xmath0 ( mod 2 ) . in particular , if @xmath152 represents zero in @xmath151 , then the trace is _ independent _ of the choice of lift .
we deduce that if @xmath153 where @xmath154 for some properly embedded @xmath155 then @xmath143 is nontrivial .
in general , for any representation @xmath156 which sends @xmath149 to parabolic elements , we get a relative class @xmath157 if @xmath158 lifts to @xmath159 then the image of @xmath160 in @xmath144 is zero . if @xmath161 for some @xmath152 which represents a trivial class in @xmath151 then @xmath160 is nontrivial .
we summarize this as a lemma : [ stiefel_nontrivial ] suppose that @xmath0 is a compact orientable @xmath2-manifold with torus boundary components , and suppose @xmath158 is some representation @xmath156 for which elements of @xmath149 map to parabolic transformations .
suppose further that @xmath158 lifts to @xmath162 let @xmath163 denote the relative stiefel - whitney class of @xmath158 , where the trivialization on @xmath133 comes from the fixed point of the corresponding parabolic subgroup .
further , suppose there is some @xmath152 representing zero in @xmath151 for which @xmath164 then @xmath160 is nontrivial .
let @xmath0 be a manifold containing an incompressible surface @xmath15 with a single boundary component .
it turns out that there is a very simple relationship between @xmath166 and @xmath167)$ ] ; in fact , @xmath167)$ ] is just the mod 2 reduction of @xmath166 .
the simplest way to see this is topological .
the value of @xmath167)$ ] depends only on the topology of the @xmath45 bundle over @xmath15 obtained from the representation @xmath168 by using the exponential map , we may identify this bundle with the restriction of the unit tangent bundle @xmath169 .
in particular , the value of @xmath167)$ ] only depends on the topology of a tubular neighborhood of @xmath15 in @xmath0 .
if @xmath15 is two - sided and embedded , then this tubular neighborhood is just a product @xmath170 and therefore the value of @xmath167)$ ] depends only on @xmath15 .
now , for a geometric representation @xmath171 coming from a hyperbolic structure on @xmath15 , we have already seen that @xmath172 ) = \chi(s ) \text { mod } 2 $ ] .
in particular , we have the following lemma : [ stiefel_class ] let @xmath0 be a complete finite - volume orientable hyperbolic @xmath2-manifold .
if @xmath155 is incompressible and orientable , then @xmath173 ) = \chi(s ) \mod 2\ ] ] where @xmath111 $ ] represents the fundamental class in @xmath174 .
the following corollary is folklore , and seems to have been observed first by w. thurston , at least for seifert surfaces of knots in @xmath175 .
there appears to be some confusion in the literature about whether it is well - known , and therefore we state it for completeness : let @xmath0 be a complete noncompact orientable hyperbolic @xmath2-manifold , and let @xmath176 be the boundary of a @xmath92-sided incompressible surface @xmath155 .
if @xmath177 is any lift of the geometric representation , then @xmath178 it follows that the longitude of any knot has trace @xmath8 .
this answers question 6.2 in @xcite .
we remark that by the proof of the ending lamination conjecture @xcite , for @xmath15 incompressible , the geometric representation @xmath179 coming from the hyperbolic structure on @xmath0 and a geometric representation @xmath171 coming from a hyperbolic structure on @xmath15 are homotopic through representations of @xmath33 into @xmath4 which send boundary curves to parabolic elements , since @xmath179 is always in the closure of the space of quasifuchsian representations .
this gives another , more high - powered proof of lemma [ stiefel_class ] .
if @xmath0 is a complete hyperbolic @xmath2-manifold of finite volume , we have the corresponding geometric representation @xmath180 the _ trace field _ @xmath6 of @xmath0 is the field generated by the traces of @xmath181 , as @xmath58 varies over the elements of @xmath3 .
of course , the trace of an element in @xmath4 is only determined up to sign , but the field generated by these elements is independent of sign .
it turns out that the trace field @xmath6 is always a _ number field _
( i.e. some finite algebraic extension of @xmath182 ) and as remarked in the introduction , _ for @xmath0 noncompact _
, @xmath137 can be conjugated into @xmath5 .
the _ invariant trace field _ is the subfield @xmath13 of @xmath6 generated by the squares of the traces of @xmath181 as above . in general ,
@xmath13 and @xmath6 are not equal , and the degree satisfies @xmath183 = 2^n$ ] for some @xmath23 .
the field @xmath13 is an invariant of the commensurability class of @xmath0 , where two hyperbolic manifolds @xmath184 are said to be commensurable if they have a common finite cover .
see @xcite for details and proofs .
let @xmath186 denote the galois closure of @xmath6 in @xmath187 .
if @xmath188 is the minimal polynomial of a generating element of @xmath6 , then @xmath186 is obtained from @xmath182 by adjoining all roots of @xmath188 . the galois group @xmath185 of @xmath186 over @xmath182 acts on @xmath186 by field automorphisms , conjugating @xmath6 into different subfields of @xmath187 . the various ( galois conjugate )
embeddings of @xmath6 into @xmath187 are called _ places _ , and can be real or complex .
complex places come in pairs , interchanged by complex conjugation . if the degree @xmath189=d$ ] , then @xmath190 where @xmath191 is the number of real places , and @xmath192 is the number of conjugate pairs of complex places . in general ,
if @xmath193 is some representation which is parabolic on @xmath133 , and if the relative stiefel - whitney class @xmath160 has zero image in @xmath144 , then @xmath158 lifts to @xmath194 , and parabolic elements lift to elements of @xmath195 with trace equal to @xmath196 .
if @xmath197 is obtained from @xmath158 by galois conjugating @xmath6 to @xmath198 , then @xmath197 lifts to @xmath199 , so the relative stiefel - whitney class @xmath200 has zero image in @xmath144 .
moreover , since @xmath196 are in the fixed field of @xmath24 , we have equality @xmath201 since the relative class is determined by these traces , we have equality @xmath202 the invariance of @xmath160 under the galois group implies that @xmath160 can be pulled back from the cohomology of @xmath203 where @xmath204 has the discrete topology .
a real place for @xmath6 determines an embedding of @xmath3 in @xmath16 , which lifts to @xmath74 , by the vanishing of the second stiefel - whitney class for a @xmath2-manifold , and the invariance of @xmath79 under the action of the group @xmath185 .
let @xmath205 be the group generated by squares of elements of @xmath3 where @xmath0 is noncompact as before . a real place for @xmath13 determines an embedding of @xmath205 into @xmath16 which extends to an embedding of @xmath3 into @xmath4 for which every element of @xmath3 has a trace which is real or pure imaginary .
since the square of every element @xmath58 of @xmath3 stabilizes @xmath38 in @xmath206 , it follows that @xmath58 either fixes @xmath38 ( possibly reversing orientation ) or takes it to an orthogonal copy of @xmath38 . in the second case
, the square @xmath207 takes @xmath38 to itself by an orientation - reversing isometry , contrary to the fact that @xmath207 is in @xmath16 by hypothesis .
it follows that @xmath3 preserves @xmath38 , and therefore has image in @xmath39 .
again , this discussion depends on the noncompactness of @xmath0 . in general , for noncompact @xmath0 , a similar argument shows that the representation of @xmath3 can always be conjugated into @xmath208 where @xmath13 is the invariant trace field .
this fact is implicit e.g. in @xcite , page 278 .
let @xmath11 be a hyperbolic once - punctured torus bundle with monodromy @xmath12 .
then @xmath12 induces an automorphism on @xmath209 , represented by some matrix @xmath210 since @xmath12 is pseudo - anosov , the trace @xmath211 satisfies @xmath212 .
then @xmath213 is isomorphic to the kernel of the map @xmath214 which has rank @xmath90 , and therefore @xmath215 is isomorphic to @xmath216 , generated by the relative class of the fiber .
let @xmath11 be a hyperbolic surface bundle over @xmath1 with fiber a once - punctured torus @xmath217 .
the following theorem relates topology , homological algebra , and number theory : let @xmath11 be a hyperbolic oriented once - punctured torus bundle over @xmath1 with monodromy @xmath12 .
let @xmath6 denote the trace field of @xmath11 , and @xmath13 the invariant trace field .
then : 1 . @xmath6 admits no real places 2 .
if @xmath13 admits a real place , then @xmath14 contains 2-torsion we suppose after conjugating that the image of @xmath137 lies in @xmath5 . we know
that @xmath137 lifts to @xmath218 moreover , if @xmath217 denotes the fiber of the fibration , and @xmath219 are standard ( free ) generators for @xmath220 , then @xmath221 ) = -2\ ] ] as in lemma [ stiefel_class ] .
suppose @xmath222 is a real place , and let @xmath223 be obtained by galois conjugating @xmath6 into @xmath7 .
let @xmath224 be obtained by composing @xmath225 with the covering map @xmath226 .
we denote the relative euler class of @xmath197 by @xmath73 : @xmath227 since @xmath228 we must have that @xmath229)$ ] is odd .
triangulate @xmath217 by two ideal triangles @xmath230 .
the representation @xmath197 determines a developing map from the universal cover @xmath231 to @xmath38 @xmath232 if the image of both triangles has the same orientation , then the developing map is a homeomorphism , and we obtain a complete hyperbolic structure on @xmath217 which is invariant under @xmath12 . but
this implies that @xmath12 has finite order , which is incompatible with the existence of a complete hyperbolic structure on @xmath11 .
it follows that the orientations on the images of the @xmath233 disagree . by thurston s formula for @xmath234 ( lemma [ thurston_and_euler ] )
, we have @xmath235 ) = \sum_i \text{sign of orientation on } d(\delta_i ) = 0\ ] ] this gives a contradiction , and shows that @xmath6 has no real place . now
if @xmath13 admits a real place , then there is some @xmath236 such that @xmath237 as above , we get a developing map from @xmath231 to @xmath38 for which the orientations on the ideal triangles must disagree , and the rational relative euler class of the action must vanish . since @xmath197 is conjugate into @xmath39 but not @xmath16 the orientation class @xmath238 must be nontrivial .
in fact , since the traces of elements of @xmath239 are @xmath196 , boundary elements map to the subgroup @xmath16 , and therefore the orientation class @xmath240 is a nontrivial class in @xmath241 . since @xmath242 is torsion for a punctured torus bundle , we are done .
note that any field of odd degree admits a real place ; in particular , the degree of @xmath6 is always even .
if @xmath0 is a ( compact or noncompact ) hyperbolic surface bundle , and @xmath15 is any fiber , then the same argument shows that if @xmath6 has a real place @xmath222 , then @xmath243 j. button has studied trace fields of punctured torus bundles with monodromy of the form @xmath244 in @xcite .
he showed for positive @xmath245 and for all odd @xmath23 that the invariant trace field of @xmath246 has no real places .
for @xmath23 odd , the homology of @xmath247 has no @xmath92-torsion , but for @xmath23 even and not divisible by @xmath77 , this does not follow from theorem a , but rather from an explicit computation .
one might further ask whether every punctured torus bundle whose invariant trace field has a real place has @xmath77-torsion in @xmath248 .
in fact , we posed exactly this question in an earlier version of this paper .
j. button has found a counterexample to this question : the census manifold s299 is a once - punctured torus bundle with monodromy @xmath249 whose invariant trace field @xmath13 has degree 3 , and whose trace field @xmath6 has degree 12 , and which has first homology @xmath250 . in @xcite ,
w. goldman characterizes geometric representations of once punctured torus groups amongst all @xmath16 representations in terms of trace data .
this gives an alternate proof of the first part of theorem a , without using the geometric formula for the euler class .
part ( 2 ) of theorem a also follows from corollary 2.3 of @xcite . amongst the cusped manifolds in the hodgson weeks census ( see @xcite ) , m039 is a torus bundle with monodromy @xmath251 , @xmath252 and invariant trace field with minimal polynomial @xmath253 .
it has a degree @xmath92 cover v3225 for which this is the trace field ; this cover is fibered with fiber a twice - punctured torus , so necessarily the euler class of the representation associated to the real place must vanish on this fiber .
some other examples : m040 is a torus bundle with monodromy @xmath254 , @xmath255 and invariant trace field with minimal polynomial @xmath253 , and v2231 is a torus bundle with monodromy @xmath256 , @xmath257 and invariant trace field with minimal polynomial @xmath258 .
the invariant trace fields were found with the help of the program snap ( @xcite ) .
we have seen from [ characteristic_class_section ] and [ torus_section ] that @xmath259 for @xmath15 a fiber of @xmath0 , and @xmath260 for any incompressible surface @xmath15 , whenever @xmath222 is a real place . in @xcite
, thurston introduced a norm on @xmath261 for @xmath0 irreducible and atoroidal . for a homology class
@xmath111 $ ] , the norm satisfies @xmath262\| = \inf_s -\chi(s)\ ] ] where the infimum is taken over all ( possibly disconnected ) representatives @xmath15 of @xmath111 $ ] with no spherical components . a generalization of this norm , due to gromov , measures a similar complexity amongst all _ immersed _ surfaces with no spherical components representing a given homology class .
a theorem of gabai ( @xcite ) shows that these two norms are equal ( after a suitable normalization ) ; i.e. any immersed surface may be replaced by an embedded surface of no larger norm .
the key properties of the norm @xmath263 are that the unit ball @xmath264 is a finite sided polyhedron , whose vertices are _ rational _ , and that there are a finite ( possibly empty ) collection of top dimensional faces @xmath265 with the property that the integral homology classes @xmath111 $ ] representing fibrations of @xmath0 over @xmath1 are exactly those whose ( positive ) projective rays intersect the interiors of the @xmath265 . such @xmath265 are called _ fibered faces _ of @xmath264 . [ norm_ball ]
let @xmath0 be a cusped hyperbolic @xmath2-manifold , and suppose @xmath222 is a real place with associated relative euler class @xmath266 .
then for every fibered face of @xmath264 there is a vertex @xmath267 such that @xmath268 .
similarly , there is a vertex @xmath269 such that @xmath270 .
a discrete finite covolume pseudo - rational subgroup acts on @xmath38 with quotient a pseudo - rational surface @xmath15 .
such a surface in a hyperbolic @xmath2-manifold is necessarily totally geodesic .
note that for us , pseudo - rational surfaces are always orientable .
if @xmath6 is the trace field of @xmath0 and @xmath222 is a real place , the traces of a pseudo - rational subsurface do not change .
it follows that @xmath272 ) = \pm \chi(s)\ ] ] for any pseudo - rational surface , and any real place @xmath24 .
consequently , we have the following corollary : [ pseudo_theorem ] let @xmath0 be a cusped hyperbolic @xmath2-manifold , and suppose @xmath155 is a pseudo - rational surface ( possibly immersed ) .
if @xmath15 is not ( gromov or thurston ) norm minimizing in its homology class , @xmath6 has no real places
. one method of constructing ( pseudo)-rational surfaces is by covering thrice punctured spheres .
a thrice - punctured sphere is always homologically essential , and therefore so is its preimage in a finite cover .
but if such a finite cover has suitable symmetries , one might be able to find a ( low genus ) surface with a @xmath92-fold orientation - reversing fixed - point free symmetry , in the same homology class as the pseudo - rational surface .
one can then cut along such a surface , and reglue the resulting boundary components to themselves to get a new manifold with the same trace field as the old , in which the pseudo - rational surface is homologically trivial .
[ no_totally_geodesic ] let @xmath0 be a fibered knot complement in a rational homology sphere whose trace field @xmath6 has odd prime degree .
then @xmath0 does not contain an immersed totally geodesic surface .
since @xmath6 has prime degree , it has no proper subfields other than @xmath182 ; in particular , any immersed totally geodesic surface @xmath15 has rational traces . since @xmath0 is a knot complement in a rational homology sphere , its rational second homology is @xmath90 dimensional .
since it is fibered , the rational homology is generated by the fiber @xmath273 .
it follows that @xmath111 = n[f]$ ] in homology for some nonzero integer @xmath23 . since @xmath273 is a fiber of a fibration ,
it is thurston ( and gromov ) norm - minimizing , and therefore @xmath274 let @xmath222 be a real place , and let @xmath266 be the associated relative euler class .
then we have @xmath275 contrary to theorem [ pseudo_theorem ] .
if @xmath278 is a closed , orientable surface of genus @xmath279 , goldman @xcite showed that the @xmath16 representation variety of @xmath280 has @xmath281 components , indexed by values of the euler class on @xmath278 satisfying @xmath282)| \le -\chi(\sigma)\ ] ] for representations @xmath197 coming from real places @xmath24 of trace fields @xmath6 , we have the additional constraint that @xmath286 ) = \chi(\sigma ) \mod 2 $ ] .
modulo this constraint , we will see how to construct simple examples which realize every possible compatible combination of euler characteristic and euler class .
let @xmath0 be a manifold , and @xmath264 the unit ball of the thurston norm .
a _ big diamond _ is a symmetrical @xmath77-gon @xmath287 which is the intersection of @xmath264 with a two - dimensional plane @xmath126 , and whose vertices are integer lattice points which generate the lattice of integral points in @xmath126 .
let @xmath0 be a cusped hyperbolic @xmath2-manifold .
suppose the trace field @xmath6 has a real place @xmath24 with associated relative euler class @xmath266 , and suppose further that the unit ball in the thurston norm @xmath264 contains a big diamond @xmath288 .
then for every integer @xmath18 and every integer @xmath19 with @xmath289 and @xmath290 there is an immersed incompressible connected surface @xmath22 in @xmath0 satisfying @xmath291 ) = m\ ] ] let @xmath292 be surfaces representing the vertices of the big diamond @xmath288
. then @xmath293 and therefore @xmath294 .
after replacing @xmath295 and/or @xmath296 with their negatives if necessary , we can assume @xmath297 for @xmath298 let @xmath299 denote the thurston norm - minimizing surface representing the homology class @xmath300 + q[v_2]$ ] .
since @xmath288 is a diamond , we have @xmath301 since @xmath266 is linear , we have @xmath302 if @xmath303
are coprime , then @xmath299 is represented by a connected surface . otherwise , we have @xmath304 for some @xmath305 where @xmath306 are coprime
. then @xmath307 has a subgroup of index @xmath308 which gives a connected incompressible immersed surface in @xmath0 with euler characteristic @xmath309 and euler class @xmath310 .
together with finite index subgroups of @xmath311 , this shows that every possibility is realized .
the link @xmath26 in rolfsen s tables @xcite has a complement whose unit ball in the thurston norm is a big diamond , and has trace field generated by a root of @xmath312 , which has a real place because the degree is odd ( thanks to n. dunfield for finding this example . ) 99 j. brock , r. canary and y. minsky , _ the classification of kleinian surface groups , ii : the ending lamination conjecture _ , eprint , math.gt/0412006 j. button , _ invariant trace fields of once - punctured torus bundles _ , kodai math . j.
* 28 * ( 2005 ) , no .
1 , 181190 d. calegari , _ circular groups , planar groups and the euler class _ , geom . topol .
* 7 * ( proceedings of the casson fest ) ( 2004 ) , 431491 d. cooper and d. long , _
remarks on the @xmath313-polynomial of a knot _ , j. knot theory ramifications * 5 * ( 1996 ) , no . 5 , 609628 m. culler , _ lifting representations to covering groups _ , adv . in math .
* 59 * ( 1986 ) , no .
1 , 6470 d. gabai , _ foliations and the topology of @xmath2-manifolds _ , j. diff . geom . * 18 * ( 1983 ) , no .
3 , 445503 e. ghys , _
groupes dhomomorphismes du cercle et cohomologie borne _ , _ the lefschetz centennial conference , part iii ( mexico city , 1984 ) _ , 81106 contemp . math . , 58 , iii amer .
providence , ri , 1987 w. goldman , _ topological components of spaces of representations _ , invent . math .
* 93 * ( 1988 ) , no .
3 , 557607 o. goodman , snap , available at http://www.ms.unimelb.edu.au/`~snap`/ a. hatcher , _ a proof of a smale conjecture , @xmath314 _ , ann . of math . ( 2 )
* 117 * ( 1983 ) , no .
3 , 553607 s. jekel , _ a simplicial formula and bound for the euler class _ , israel j. math .
* 66 * ( 1989 ) , no . 1 - 3 , 247259 d. long and a. reid , _ pseudomodular surfaces _ , j. reine angew
* 552 * ( 2002 ) , 77100 c. maclachlan and a. reid , _ the arithmetic of hyperbolic 3-manifolds _ , graduate texts in mathematics 219 springer - verlag , new york , 2003 k. matsuzaki and m. taniguchi , _ hyperbolic manifolds and kleinian groups _ , oxford mathematical monographs , oxford science publications the clarendon press , oxford university press , new york , 1998 j. milnor , _ on the existence of a connection with curvature zero _ , comment . math .
* 32 * ( 1958 ) , 215223 w. neumann and a. reid , _ arithmetic of hyperbolic manifolds _ , topology 90 ( columbus , oh , 1990 ) , 273310 , ohio state univ .
, 1 , de gruyter , berlin , 1992 .
d. rolfsen , _ knots and links _ , corrected reprint of the 1976 original mathematics lecture series , 7 . publish or perish , inc .
, houston , tx , 1990 w. thurston , _ a norm for the homology of @xmath2-manifolds _ , mem .
* 59 * ( 1986 ) , no .
339 , i vi and 99130 w. thurston , _ three - manifolds , foliations and circles ii _ , privately circulated preprint , ( 1997 ) j. weeks , snappea , available at http://www.geometrygames.org/snappea/ j. wood , _ bundles with totally disconnected structure group _
, comment .
* 46 * ( 1971 ) , 257273 | if @xmath0 is a hyperbolic once - punctured torus bundle over @xmath1 , then the trace field of @xmath0 has no real places .
[ section ] [ theorem]lemma [ theorem]corollary [ theorem]conjecture [ theorem]question [ theorem]problem [ theorem]definition [ theorem]construction [ theorem]notation [ theorem]convention [ theorem]remark [ theorem]example h p u |
the cp or t violation is a fundamental and important problem of the particle physics and cosmology .
the cp study of the lepton sector , though it has been less examined than that of quark sector , is indispensable , since the neutrinos are allowed to have masses and complex mixing angles in the electroweak theory . the neutrino oscillation search is a powerful experiment which can examine masses and mixing angles of the neutrinos .
in fact the several underground experiments have shown lack of the solar neutrinos@xcite and anomaly in the atmospheric neutrinos@xcite , implying that there may occur the neutrino oscillation .
the atmospheric neutrino anomaly suggests mass difference around @xmath0 ev@xmath1@xcite , which encourages us to make long base line neutrino experiments .
recently such experiments are planned and will be operated in the near future@xcite .
it seems necessary for us to examine whether there is a chance to observe not only the neutrino oscillation but also the cp or t violation by long base line experiments . in this short paper
we study such possibilities taking account of the atmospheric neutrino experiments and also considering the solar neutrino experiments and others .
we briefly review cp and t violation in vacuum oscillation@xcite to clarify our notation .
let s denote the mass eigenstates of 3 generations of neutrinos by @xmath2 with mass eigenvalues in vacuum . ] @xmath3 and the weak eigenstates by @xmath4 corresponding to electron , @xmath5 and @xmath6 , respectively .
they are connected by a unitary transformation : @xmath7 where @xmath8 is a unitary ( @xmath9 ) matrix similar to the ckm matrix for quarks .
we will use the parametrisation for @xmath8 by chau and keung@xcite , @xmath10 where the @xmath11 s are the gell - mann matrices .
the evolution equation for the weak eigenstate is given by @xmath12 where @xmath13 s are the momenta , @xmath14 is the energy and @xmath15 .
a term proportional to a unit matrix like @xmath16 in eq.[eq : evoleq ] is dropped because it is irrelevant to the transition probability .
the solution for the equation is @xmath17 the transition probability of @xmath18 at distance @xmath19 is given by @xmath20 the t violation gives the difference between the transition probability of @xmath21 and that of @xmath22@xcite : @xmath23 where @xmath24 the unitarity of @xmath8 gives @xmath25 with the sign @xmath26 ( @xmath27 ) for @xmath28 in cyclic ( anti - cyclic ) order(@xmath26 for @xmath29 or @xmath30 ) . in the following
we assume the cyclic order for ( @xmath31 , @xmath32 ) for simplicity .
there are bounds for @xmath33 and @xmath34 , @xmath35 where the equality holds for @xmath36 where the equality holds for @xmath37 ( mod @xmath38 ) . in the vacuum
the cpt theorem gives the relation between the transition probability of anti - neutrino and that of neutrino , @xmath39 which relates cp violation to t violation : @xmath40 let s consider how large the t(cp ) violation can be in the `` disparate '' mass difference case and the smaller one by @xmath41 in the case that the mass differences have a large ratio .
] , say @xmath42 . in this case
the following two situations are interesting@xcite , since in the case @xmath43 we have too small @xmath44 due to eq.[eq : fv ] ) to observe the t(cp ) violation effect : situation 1 .
@xmath45 .
because @xmath46 in this case , the oscillatory part @xmath34 becomes @xmath47 : @xmath48 fig.[fig1 ] shows the graph of @xmath49 .
the approximation eq.[eq : fapp ] works very well up to @xmath50 . in the following
we will use eq.[eq : fapp ] instead of eq.[eq : f ] .
we see many peaks of @xmath51 in fig.[fig1 ] . in practice , however , we do not see such sharp peaks but observe the value averaged around there , for @xmath52 has a spread due to the energy spread of neutrino beam ( @xmath53 ) . in the following
we will assume @xmath54@xcite as a typical value .
table [ table1 ] gives values of @xmath55 at the first several peaks and the averaged values around there .
we see the t violation effect , @xmath56 at peaks for neutrino beams with 20 % of energy spread .
note that the averaged peak values decrease with the spread of neutrino energy .
which peak we can reach depends on @xmath57 and @xmath14 .
the first peak @xmath58 is reached , for example by @xmath59 ev@xmath1 , @xmath60 km ( for kek - kamiokande long base line experiment ) and neutrino energy @xmath61 gev . in this case
we see the t(cp ) violation effect at best of @xmath62 since we have a bound on @xmath33 as eq.[eq : jmax ] .
( 16,6 ) ( 25,80)@xmath63 ( 150,5)@xmath52 .the peak values of @xmath55 and the corresponding averaged values .
here @xmath64 is a value of @xmath65(see eq.[eq : fapp ] ) averaged over the range @xmath66 ( @xmath67 ) . [ cols="^,^,^,^",options="header " , ] situation 2 .
@xmath68 .
because @xmath69 and @xmath70 oscillate rapidly and vanish after being averaged over the energy spread in this case , the oscillatory part @xmath34 is dominated by @xmath71 .
since @xmath34 has now a bound @xmath72 instead of eq.[eq : fmax ] , the t violation effect @xmath73 is bounded as @xmath74 .
( for energy spread of @xmath75 of neutrino beam@xcite , @xmath76 is enough for @xmath69 and @xmath70 to oscillate rapidly and vanish after being averaged . )
there are a variety of possible combinations of the parameters , three mixing angles , two mass differences and a cp violating phase .
when we consider only the atmospheric neutrino anomaly to be attributed to the neutrino oscillation , we can take the mass differences , @xmath41 and @xmath77(and hence @xmath78 ) , to be comparable , while when we consider both the solar and the atmospheric neutrino anomalies to be attributed to the neutrino oscillation , we expect @xmath41 and @xmath77 to be `` disparate '' , @xmath79 .
we investigate how large the cp violation effect can be in neutrino oscillation for the above two cases .
let s examine the case of mass differences to be of the same order of magnitude .
we use a parameter set that @xmath80ev@xmath1 , @xmath81 and @xmath82 is arbitrary , derived by yasuda@xcite through the analysis of the atmospheric neutrino anomaly . here
the matter effect@xcite is negligibly small and eq.[eq : cpt ] is available . with use of eq.[eq : jf ] , eq.[eq : j ] , and eq.[eq : cpt ] this parameter set gives the cp violation effect @xmath83 where @xmath84 and @xmath85 ( 16,6 ) ( 135,38)@xmath86 ( 0,75)@xmath87 fig.[fig2 ] shows the oscillatory part @xmath88 .
there are many peaks @xmath88 showing the possibility to observe the large cp violation effect .
for example , we may see very large difference between the transition probabilities , @xmath89 @xmath90 for @xmath60 km(for kek - kamiokande experiment ) and @xmath914.5 gev corresponding to @xmath92 , if we have large @xmath93 .
incidentally we may remark that the survival probability of solar neutrino is calculated to be 0.45 for those mixing angles .
this value is consistent with both two gallium experiments@xcite and kamiokande experiment@xcite , but it is inconsistent with homestake result@xcite , if all the solar neutrino anomaly should be attributed to the neutrino oscillation@xcite .
in conclusion we may see a large cp violation effect when we have comparable mass differences . in this respect
we note that the long base line experiments are urgently desirable .
next we consider the `` disparate '' mass difference case @xmath94 . the case @xmath95 ev@xmath1 and @xmath96 ev@xmath1 is favoured by the hot dark matter scenario@xcite and the atmospheric neutrino anomaly .
this case is already analysed by tanimoto@xcite and we will not discuss it here .
the case @xmath97 ev@xmath1 and @xmath98 ev@xmath1 could typically explain the anomalies of the atmospheric and the solar neutrinos@xcite . in this case we can not neglect the matter effect@xcite @xmath99 where @xmath100 is the electron number density of the earth and @xmath101 is the matter density of the surface of the earth , since it is greater than @xmath41 .
it requires to subtract such effect in order to deduce the pure cp violation effect@xcite . in principle
it is possible , because the matter effect is proportional to @xmath14 while @xmath41 is constant .
in the matter with constant density requires the exchange of the production point and the detection point and the time reversal of @xmath102 in matter is in general different from @xmath103@xcite . ]
, we have a pure t violation effect @xmath104 , though we do not observe a pure cp violation effect because of an apparent cp violation due to matter .
when a neutrino is in matter , its matrix of effective mass squared @xmath105 of weak eigenstates is@xcite @xmath106 where @xmath107 and @xmath8 is given by eq.[eq : mixingmatrix ] .
this is diagonalized by a mixing matrix @xmath108 as @xmath109 .
it is written with a real unitary ( orthogonal ) matrix @xmath110 as @xmath111 with arguments analogous to 2.1 we have the t violation effect , @xmath112 where @xmath113 we get @xmath114 where @xmath115 and @xmath116 . an exact result for @xmath108 and @xmath117
is given in @xcite , though their result is rather complicated .
here we show a simple expression for @xmath108 and @xmath117 in the case @xmath118 .
we derive @xmath108 and @xmath117 in this case using perturbation with respect to small @xmath119 .
first we decompose @xmath120 and diagonalize by @xmath121 the first term of the parenthesis \ { } of eq.[eq : matter ] , the eigenvalues of which we denote by @xmath122 s .
we find @xmath123 and @xmath124 we have @xmath125 next we diagonalize the whole @xmath105 by @xmath126 with perturbation with respect to small @xmath41 . at the zeroth order of @xmath119
, we have @xmath127 @xmath128 , and @xmath129 which gives @xmath130 and hence @xmath131(see eq.[eq : jm ] ) . at the first order of perturbation
, we have @xmath132 and with eq.[eq : jm ] @xmath133 it seems most likely to be realized that @xmath134 as is discussed in 3.2 . here
we study this case in detail . since @xmath135 is @xmath136 we neglect @xmath136 in estimating @xmath137 .
we also neglect @xmath138 since @xmath139
. then we have the effective masses @xmath140 and `` mass difference ratio '' @xmath141 note that @xmath142 .
we find @xmath143 and @xmath144 using the argument similar to that used to derive eq.[eq : atbest2 ] , we obtain the t violation effect @xmath145 at peaks , where we choose the mean neutrino energy @xmath14 to satisfy ( see table [ table1 ] ) @xmath146 according to the analysis by fogli _
et al_@xcite , @xmath147 @xmath148 and @xmath149 are allowed and @xmath150 .
] for example
. then @xmath151
we have examined the cp and t violation in the neutrino oscillation , and analysed how large the violation can be by taking account of the constraints of the neutrino experiments . in case of the comparable mass differences of @xmath152 and @xmath78 in the range @xmath153 to @xmath154 ev@xmath1 ,
which is consistent with the analysis of the atmospheric neutrino anomalies , it is found that there is a possibility that the cp violation effect is large enough to be observed by 100 @xmath155 1000 km base line experiments if the cp violating parameter sin@xmath82 is sufficiently large . in case that @xmath41 is much smaller than the matter parameter `` @xmath156 '' and @xmath77 , which is favoured both by the solar and atmospheric neutrino anomalies , we have derived a simple formula for the t violation effect .
we note that the probability of cp or t violation effect should vanish for @xmath157 0 , and therefore be proportional to @xmath158 , @xmath159 or @xmath160 by the dimensional analysis .
our calculation confirms this expectation .
if the solar and atmospheric neutrino anomalies are both attributed to the neutrino oscillation , the cp violation test is found difficult since matter effect is larger than the pure cp violation effect .
how to extract the matter effect in such a case will be discussed in a separate paper@xcite . in conclusion the long base line neutrino oscillation experiments are very important and desirable to study not only neutrino masses and mixings but the cp or t violation in the lepton sector and
there is some possibility to find such effect explicitly . 99 gallex collaboration , p. anselmann _ et al _ , phys
. lett . * b357*,(1995 ) 237 .
sage collaboration , j. n. abdurashitov _
et al _ , phys
b328 * , ( 1994 ) 234 .
kamiokande collaboration , nucl , phys .
proc . suppl . *
b38 * , ( 1995 ) 55 .
homestake collaboration , nucl , phys .
. suppl . *
b38 * , ( 1995 ) 47 .
kamiokande collaboration , k.s .
hirata et al . , phys . lett .
* b205 * ( 1988 ) 416 ; _ ibid . _ * b280 * ( 1992 ) 146 ; y. fukuda et al .
, phys . lett . *
imb collaboration , d. casper et al .
* 66 * ( 1991 ) 2561 ; + r.
becker - szendy et al . , phys . rev .
* d46 * ( 1992 ) 3720 .
nusex collaboration , europhys .
lett . * 8*(1989)611 ; _ ibid _ * 15*(1991)559 .
soudan2 collaboration , nucl .
* b35 * ( 1994 ) 427 ; _ ibid _ * 38 * ( 1995 ) 337 .
frjus collaboration , z. phys . *
g. l. fogli , e. lisi , d. montanino and g.scioscia , preprint iassns - ast 96/41 ( hep - ph/9607251 ) . g. l. fogli , e. lisi and d. montanino , phys .
rev * d49 * , 3626 ( 1994 ) .
o. yasuda , preprint tmup - hel-9603 ( hep - ph/9602342 ) .
k. nishikawa , ins - rep-924(1992 ) .
s. parke , fermilab - conf-93/056-t(hep - ph/9304271)(1993 ) . for a review , m. fukugita and t. yanagida , in _ physics and astrophysics of neutrinos _ , edited by m. fukugita and a. suzuki , ( springer - verlag , tokyo , 1994 ) .
s. m. bilenky and s. t. petcov , rev . mod
59 * , 671 ( 1987 ) .
s. pakvasa , in high energy physics-1980 ( 20th int .
conf . on high energy physics ) , ed.l.durand and l.pondrom , vol.2,p.1164 .
chau and w .- y .
keung , phys .
* 53 * , 1802 ( 1984 ) .
t.k kuo and j. pantaleone , phys . lett . * b198 * , 406 ( 1987 ) .
s. toshev , phys . lett . *
b226 * , 335 ( 1989 ) .
v. barger , k. whisnant and r. j. n. phillips , phys . rev
. lett . * 45 * , 2084 ( 1980 ) .
n. cabibbo , phys . lett . *
b72 * , 333 ( 1978 ) .
nishikawa , private communication .
l. wolfenstein , phys . rev . *
d17 * , 2369 ( 1978 ) .
s. p. mikheev and a. yu .
smirnov , sov .
. phys . * 42 * , 913 ( 1985 ) h. minakata , preprint tmup - hel-9602 .
primack , j. holtzman , a. klypin and d. o. caldwell , phys .
. lett . * 74 * 2160 ( 1995 ) and references therein .
m. tanimoto , preprint ehu-96 - 6 ( hep - ph/9605413 ) .
j. arafune and j. sato , in preparation .
h. w. zaglauer and k. h. schwarzer , z. phys * c 40 * , 273 ( 1988 ) . | we examine how large violation of cp and t is allowed in long base line neutrino experiments .
when we attribute only the atmospheric neutrino anomaly to neutrino oscillation we may have large cp violation effect .
when we attribute both the atmospheric neutrino anomaly and the solar neutrino deficit to neutrino oscillation we may have a sizable t violation effect proportional to the ratio of two mass differences ; it is difficult to see cp violation since we ca nt ignore the matter effect .
we give a simple expression for t violation in the presence of matter .
-0.4 cm 0.3 cm = 20.5pt plus 0.2pt minus 0.1pt |
the departure of the equation of state ( eos ) of neutron - rich matter ( @xmath0 ) from the symmetric limit ( @xmath1 ) is governed by the nuclear symmetry energy .
this makes the knowledge of the symmetry energy important for a variety of areas of nuclear physics ( masses , neutron densities , heavy ion collisions , etc . ) and of astrophysics ( supernovae , neutron stars , nucleosynthesis , etc . ) @xcite .
however , the available experimental data on nuclei , which correspond mostly to isotopes with small and moderate neutron - proton asymmetries , do not constrain very precisely the value of the symmetry energy at the saturation density .
rather , it seems to be better constrained at a lower density around 0.10 @xmath2 @xcite , corresponding to an average between the bulk and surface symmetry contributions in the semi - empirical mass formula @xcite .
all in all , the density dependence of the symmetry energy is a basic nuclear property that still is not sufficiently well known @xcite .
this situation has prompted a recent ongoing effort to characterize more closely the symmetry energy in the regime of subsaturation densities , relevant for finite nuclei , from the phenomenology of different nuclear observables @xcite .
a proper understanding of the symmetry energy is particularly desirable in order to predict ground - state and excitation properties of atomic nuclei far from the naturally - occurring region of stability @xcite .
the study of collective excitations of nuclei near and away from the stability line is a salient source of information for nuclear structure @xcite .
for instance , the isoscalar giant monopole resonance ( gmr ) constrains the compression modulus @xmath3 of the eos of symmetric nuclear matter .
this resonance has been accurately measured in heavy and medium mass nuclei through recently improved @xmath4-particle scattering experiments @xcite . by comparing the 1p-1h rpa results predicted by gogny @xcite , skyrme @xcite and relativistic mean field ( rmf ) interactions @xcite with the excitation energy measured in @xmath5pb ( @xmath6 mev @xcite )
, it is concluded that the value of the nuclear incompressibility @xmath3 must lie in the range of 200280 mev .
performing measurements of the gmr in exotic nuclei may deliver other valuable informations and enhance our knowledge of the physics of isospin - rich and loosely - bound nuclear systems @xcite .
however , it represents an important experimental challenge due to e.g. the low intensities in the radioactive ion beams . with the advent of new technologies
it is hoped that measurements of the breathing mode in short - lived unstable nuclei will become feasible in the future @xcite .
note that promising first successes have already been demonstrated @xcite .
although the value of the excitation energy of the gmr is driven to a large extent by the compression modulus of symmetric matter , it is known to be influenced also by the density dependence of the symmetry energy in a neutron - rich nucleus such as @xmath5pb @xcite .
this influence is expected to be amplified in isotopes with large neutron numbers , but how and to which extent has not actually been assessed .
our purpose in this paper is to investigate the effects of the density dependence of the symmetry energy on the average excitation energy of the gmr in isospin - rich nuclei , of heavy and medium mass , when moving away from the stability valley .
we ought to mention , however , that our aim here is not pursuing to constrain the nuclear eos from the present calculations , but rather to analyze the effects that arise from the existing uncertainty in the knowledge of the density dependence of the symmetry energy .
we advance that the calculations will show that the impact on the excitation energy of the gmr is somehow moderate for isotopes not far from stability , while the more notable effects are found for nuclei with large neutron numbers , thereby ( unfortunately ) rendering them more difficult to see in experiment .
to describe the effective nuclear interaction , we found convenient to use the model of @xcite that consists of a standard rmf lagrangian plus an isoscalar - isovector nonlinear interaction between the @xmath7 and @xmath8 meson fields .
this coupling provides an efficient way to change in a controlled manner the density dependence of the symmetry energy without modifying the eos of symmetric matter .
in particular , the compression modulus @xmath3 is not modified . when the model is applied to finite nuclei , the variation of the strength of the isoscalar - isovector coupling changes the properties of the neutron density distribution ( for example , the neutron rms radius ) , which have large experimental uncertainties , but the predicted ground - state binding energies and charge rms radii , which are accurately known in experiment , remain basically unaltered .
thus , this rmf model allows one to modify the density content of the symmetry energy while preserving the agreement with the accurate ground - state information on which the model has been calibrated previously ( see @xcite ) .
the consistent microscopic framework for the calculation of the properties of the gmr is furnished by the hartree - fock plus rpa theory , which corresponds to the small - amplitude limit of the time - dependent hartree - fock method .
the generalization of this theory to deal with pairing in open - shell nuclei is known as quasiparticle rpa ( qrpa ) .
the qrpa provides the full information on the response function and centroid energies of the gmr .
these microscopic calculations are manageable but rather demanding if they are fully self - consistent , which is accentuated when they are performed using the canonical basis or in isotopes with large neutron numbers as it has been recently discussed in @xcite
. we will carry out our analysis by studying the estimates of the average excitation energy of the gmr that can be obtained without performing explicit rpa calculations but which carry meaningful physical information on the rpa strength .
the typical example of this procedure is the sum - rule approach to the rpa developed in the nonrelativistic framework using skyrme forces @xcite .
the sum - rule techniques provide average properties of the gmr such as the centroid energy and the resonance width and have been successfully applied to study the gmr in different nuclei and isotopic chains @xcite . in the sum - rule approach one obtains the cubic and inverse energy - weighted moments of the rpa strength function by means of , respectively , scaling and constrained calculations of the uncorrelated hartree - fock nuclear ground state .
note that these ground - state scaling and constrained values are the _ exact _ rpa moments @xcite .
because highly - accurate ground - state calculations are simpler , the sum - rule techniques also have been used in the literature as a means to check the self - consistency of explicit rpa calculations of the strength function @xcite .
the validity of the sum - rule theorems when pairing is active ( qrpa ) has been proven in recent publications @xcite .
capelli et al .
@xcite have made a careful study of constrained hartree - fock ( chf ) , constrained hartree - fock - bogoliubov ( chfb ) and qrpa calculations of the excitation energy of the gmr in the light isotopic chains of ca and ni .
the calculations have reached up to the theoretical neutron drip line of the skyrme skm * force , i.e. , up to the isotopes @xmath9ca and @xmath10ni .
admitting that these isotopic chains are known to terminate earlier in experiment than in the predictions of skm * and of most of the nuclear effective forces , and knowing also that it will be difficult to measure the gmr in even less exotic isotopes , the calculations spanning the whole theoretical isotopic chains are very useful to test the predictions of nuclear forces under extreme isospin conditions ( which can take place in astrophysical scenarios @xcite ) , as well as to test the different many - body frameworks employed to deal with the problem .
it may be noted also that khan @xcite has recently concluded that knowing the gmr in a single isotope such as @xmath5pb may not suffice to extract @xmath3 of nuclear matter and that the whole isotopic chain should be measured in order to have a grasp on the various effects acting on the gmr @xcite . from the results of @xcite for ca and ni
, it is seen that the inclusion of pairing ( chfb and qrpa calculations ) does not modify significantly the behaviour shown by the gmr excitation energies obtained in the chf calculations along the isotopic chains . in the case of the ni chain , the atomic number not being as small as in ca , the chfb and chf values agree fairly well for most isotopes from @xmath11ni up to the neutron drip line included , as noticed in figure 5 of @xcite .
of course , superfluidity is to be taken into account if one wants to describe in detail the excitation spectra or if precise values of the excitation energy within a few hundred kev in specific nuclei are needed for comparison with accurate measurements @xcite . it has been pointed out in @xcite that the qrpa calculations for isotopes having large neutron numbers become much time consuming and that ensuring proper numerics in such isospin - rich nuclei becomes somewhat difficult due to the weakly bound neutrons .
indeed , the authors of @xcite state that in the case of large neutron numbers very accurate qrpa calculations may become almost prohibitive in keeping with the size of the basis .
the gmr is a collective excitation whose average properties in general vary smoothly with the mass number @xmath12 and are rather insensitive to shell effects , at least for not too light isotopes .
thus , a description of the average properties of this resonance through semiclassical approaches like the extended thomas - fermi ( etf ) method can be , for some purposes , a practical and interesting alternative .
it has been applied in different studies in the nonrelativistic skyrme - hartree - fock framework @xcite .
the results have shown that the average excitation energies of the gmr obtained from scaling and constrained quantal hf calculations are nicely described by the corresponding semiclassical etf counterparts .
the etf predictions of gmr excitation energies become increasingly closer to the quantal predictions with increasing atomic number , as it can be expected in semiclassical methods . indeed , in the calculations along the pb isotopic chain with the skyrme force shown in figure 12 of @xcite , the etf scaling and constrained excitation energies differ from the corresponding quantal values by less than 0.4 mev and they reproduce quite accurately the change of the quantal results in going from the proton to the neutron drip line .
thus , from the experience in nonrelativistic calculations , we believe that the semiclassical scaling and constrained estimates of the average excitation energy will be a useful and realistic procedure to have a quick representation of the general trends of the gmr along isotopic chains also in the relativistic mean field framework . and in heavy systems such as the pb isotopes
, we expect that the semiclassical calculations of the gmr average excitation energy will yield reliable predictions practically on the quantitative level .
certainly , for a precise microscopic description , the more demanding qrpa calculations would have to be performed .
the semiclassical relativistic extended thomas - fermi ( retf ) approach @xcite has recently been applied in scaling and constrained calculations of the gmr energy in stable isotopes in @xcite .
the results turn out to be in good agreement with the values extracted from full relativistic rpa calculations @xcite .
let us mention as a side comment that the retf scaling procedure has been helpful in obtaining the virial theorem for the relativistic mean field model @xcite and that it has allowed self - consistent calculations of the surface contribution @xmath13 to the nuclear incompressibility in the relativistic theory @xcite . in this paper we will use the relativistic nuclear mean field model with an isoscalar - isovector coupling mentioned previously together with the retf approach for describing the average excitation energy of the gmr in the pb and zr isotopic chains , through scaling and constrained calculations .
the present results are expected to provide a realistic indication on how the density dependence of the symmetry energy impacts on the behaviour of the excitation energies and widths of the gmr along isotopic chains of heavy and medium mass . in the next section the basic theory
is presented , in the third section the results are discussed and the summary and conclusions are laid in the fourth section .
some details about the semiclassical retf calculations are given in the appendices .
a detailed description of the scaling and constrained calculations of the excitation energy of the gmr in the relativistic framework using the retf approach has been presented elsewhere @xcite . here
, we shall restrict ourselves to outline the more relevant expressions , incorporating the new contributions that arise from the mixed @xmath7-@xmath8 meson interaction .
the basic model for the present study is the relativistic mean field theory of nucleons and mesons . as in the nonrelativistic problem ,
the theory can be formulated in the thomas - fermi approach .
when corrections of order @xmath14 ( in gradients of the nucleon densities and of the dirac effective mass ) to the pure thomas - fermi contribution are included in the energy density , it is known as the relativistic extended thomas - fermi approach @xcite .
the local energy density that we employ in the present work can be written as follows : @xmath15 + \frac{\kappa}{3 ! }
\phi^3 + \frac{\lambda}{4 ! } \phi^4
\nonumber \\[3 mm ] & & \mbox { } - \frac{1}{2 g_v^2 } \left [ ( { \mbox{\boldmath$\nabla$ } } w)^2 + m_{v}^2 w^2 \right ] - \frac{1}{2 g_{\rho}^2 } \left [ ( { \mbox{\boldmath$\nabla$ } } b)^2 + m_\rho^2 b^2 \right ] - \lambda_v b^2 w^2 - \frac{1}{2 } \left({\mbox{\boldmath$\nabla$ } } { \cal a } \right)^2 , \label{eq0a}\end{aligned}\ ] ] where @xmath16 is the baryon density , @xmath17 is the isovector density , and the term @xmath18 will be described immediately below .
the energy density ( [ eq0a ] ) contains scalar @xmath19 , vector @xmath20 and vector - isovector @xmath21 meson fields , as well as the coulomb field @xmath22 .
these fields represent , respectively , @xmath23-meson , @xmath7-meson , @xmath8-meson and photon exchange .
the cubic @xmath24 and quartic @xmath25 self - interactions of the scalar field soften the eos of symmetric nuclear matter around the saturation point .
in addition , the model is supplemented by a nonlinear mixed coupling between the @xmath7-meson and the @xmath8-meson fields .
this @xmath26 interaction was introduced in @xcite to soften the density dependence of the symmetry energy and of the eos of neutron matter , which are two quantities that are predicted to be stiff by the majority of relativistic mean field models . in the quantum approach ,
the contribution @xmath18 to the rmf energy density in equation ( [ eq0a ] ) is given by @xmath27 \varphi_\nu$ ] . here , the @xmath28 are single - particle dirac wave functions , @xmath29 and @xmath30 are the usual dirac matrices and @xmath31 denotes the rest mass of the nucleons .
the retf representation of @xmath18 consists of a pure thomas - fermi part @xmath32 plus a part @xmath33 , which is of order @xmath14 compared to @xmath32 .
the nucleon variables of the retf energy density are the neutron and proton densities ( @xmath34 and @xmath35 ) instead of the single - particle wave functions @xmath28 .
thereby , @xmath18 is a functional of the nucleon densities and of the dirac effective mass @xmath36 in the retf approach .
the gradient corrections contained in the @xmath33 term provide an improved description of the nuclear surface with respect to the thomas - fermi treatment with only the @xmath32 contribution .
the reader can find the full retf expression of the functional @xmath37 in appendix a. we also provide there the variational euler - lagrange equations derived from equation ( [ eq0a ] ) in the retf approach for the nucleon densities and for the meson and photon fields .
the energy density @xmath38 is to be integrated over space to compute the total energy of the nucleus . by means of partial integrations , and on account of the variational meson field equations given in appendix a , it is possible to write the relativistic energy density of a finite nucleus in the following form : @xmath39 here , we have introduced the effective scalar density @xmath40 defined as @xmath41 the form given in equation ( [ eq1 ] ) for @xmath42 turns out to be more practical to perform the scaling of the equations in order to compute the excitation energy of the gmr in the scaling formalism .
the excitation energy of the isoscalar giant monopole resonance of a finite nucleus of mass number @xmath12 is customarily written as @xmath43 where @xmath44 is called the compression modulus or incompressibility of the finite nucleus , and @xmath45 is called the mass or inertia parameter of the monopole oscillation . in the scaling approach , the incompressibility @xmath44 is obtained from the restoring force of the monopole vibration modeled through a scaling transformation of the nucleon densities : @xmath46 here , @xmath4 denotes the scaling parameter and @xmath47 refers to neutrons or protons .
the meson fields do not scale as simple powers of @xmath4 because of the finite - range character of the meson interactions ; we collect in appendix b the variational field equations for the scaled meson fields of the rmf model with the mixed @xmath7-@xmath8 interaction .
following @xcite , it is helpful to write the scaled dirac effective mass @xmath48 in the form @xmath49 because the energy density @xmath50 and the scalar density @xmath51 of the retf model scale , respectively , as @xmath52 \equiv \alpha^4 \ , { \tilde{\cal e } } ( \alpha { \mbox{\boldmath$r$}})$ ] and as @xmath53 \equiv \alpha^3\ , { \tilde\rho}_{\rm s } ( \alpha{\mbox{\boldmath$r$}})$ ] . the tilded quantities @xmath54 and @xmath55 have the same expressions as @xmath18 and @xmath56 given in appendix a if one replaces @xmath57 by @xmath58 in them .
taking into account the above , the scaled form of the energy density of the present model becomes @xmath59 , \label{eq7}\ ] ] with the definition @xmath60 .
all densities and fields on the r.h.s . of ( [ eq7 ] ) depend on the variable @xmath61 .
the calculation of the first and second derivatives of the scaled energy @xmath62 with respect to @xmath4 proceeds along the same lines described in @xcite .
the final result for the scaling incompressibility is @xmath63_{\alpha=1 } \nonumber \\[3 mm ] & = & \frac{1}{a } \int d { \mbox{\boldmath$r$ } } \left [ - m \frac{\partial { \tilde\rho}_{s}}{\partial \alpha } + 3 \left ( \frac{m_s^2}{g_s^2 } \phi^2 - \frac{m_v^2}{g_v^2 } w^2 - \frac{m_\rho^2}{g_\rho } b^2 + \frac{\kappa}{3 ! }
\phi^3 \right ) \right .
\nonumber \\[1.5 mm ] & & \left . \mbox { } - \left ( 2 \frac{m_s^2}{g_s^2 } \phi + \frac{\kappa}{2 }
\phi^2 \right ) \frac{\partial \phi_\alpha}{\partial\alpha } + 2 \frac{m_v^2}{g_v^2 } w \frac{\partial w_\alpha}{\partial\alpha } + 2 \frac{m_\rho^2}{g_\rho^2 } b \frac{\partial b_\alpha}{\partial\alpha } \right]_{\alpha=1 } , \label{eqfn30b}\end{aligned}\ ] ] where @xmath64 $ ] .
it is interesting to note that all of the nucleonic contributions to @xmath65 are summarized in the term with the scalar density and that the massless photon field and the quartic meson couplings such as @xmath66 and @xmath26 do not provide explicit contributions . the expression ( [ eqfn30b ] ) requires the calculation of the derivatives of the meson fields with respect to the scaling parameter @xmath4 at equilibrium ( @xmath67 .
these derivatives can be computed by differentiation of the scaled meson field equations with respect to @xmath4 and by solving them at @xmath68 , as we describe in appendix b. we will use the notation @xmath69 to refer to the excitation energy of the gmr calculated in the scaling approach , namely , @xmath70 the mass parameter of the monopole resonance is given by the expression @xcite @xmath71 we note that in the nonrelativistic limit it reads @xmath72 where @xmath31 is the nucleon rest mass . because of the large contribution of the nucleon rest mass term @xmath73 to the energy density @xmath42 in the integrand of equation ( [ eq13 ] ) , the value of the excitation energy of the resonance is little modified by using either @xmath45 or @xmath74 in the calculations .
one also can study the average excitation energy of the giant monopole resonance through constrained calculations @xcite .
for that purpose , one solves the problem associated with the constrained functional @xmath75 = e(\eta ) - \eta \int d { \mbox{\boldmath$r$ } } r^2 \rho \ , .
\label{eqfn56}\ ] ] the densities , fields and energy obtained from the solution of the variational equations deduced from equation ( [ eqfn56 ] ) depend on the value of the parameter @xmath76 at which one performs the calculation . by expanding @xmath77 in a harmonic approximation about its minimum , located at the ground - state rms radius @xmath78 ( corresponding to @xmath79 ) ,
one obtains the compression modulus of the finite nucleus in the constrained approach : @xmath80 here , @xmath81 is the square mass radius of the nucleus computed with the density @xmath8 of equation ( [ eqfn56 ] ) .
one can show that equation ( [ eqfn60 ] ) can be rewritten in the following equivalent forms : @xmath82 we have used ( [ eqfn60x ] ) to check the numerical accuracy of our calculations of @xmath83 .
finally , the excitation energy of the constrained isoscalar monopole vibration is computed as @xmath84 we would like to point out that there is a parallelism between the scaling and constrained expressions for the gmr energy derived above for the relativistic model and the known results in the nonrelativistic formalism of sum rules .
that is , one can define suitable average excitation energies of the resonance such as @xmath85 in terms of ratios of the integral moments @xmath86 of the rpa strength function @xmath87 . in the nonrelativistic rpa theory of the gmr @xcite , it has been demonstrated that the cubic energy - weighted moment @xmath88 of the rpa strength function is exactly equivalent to the second derivative of the energy obtained from a monopole scaling transformation .
it is also known that the rpa energy - weighted moment @xmath89 is equivalent to @xmath90 ( `` thouless theorem '' ) , and that the exact rpa inverse energy - weighted moment @xmath91 can be obtained from a constrained calculation of the hartree - fock ground state : @xmath92 ( `` dielectric theorem '' ) .
taking into account these results and the fact that @xmath93 , one sees an interesting analogy between @xmath94 and @xmath95 of the nonrelativistic rpa formalism of sum rules and @xmath69 of equations ( [ eqescal ] ) and ( [ eqfn30b ] ) and @xmath96 of equations ( [ eqecons ] ) and ( [ eqfn60x ] ) given previously for the gmr average excitation energy in the relativistic model , though to our knowledge the sum - rule approach has not been derived in the relativistic rpa theory .
it is worth mentioning that the integral moments of the nonrelativistic rpa strength function obey certain inequalities @xcite . in particular
, one has @xmath97 where @xmath98 is the centroid energy of the gmr evaluated from the ratio of the @xmath99 moment to the @xmath100 moment .
that is , @xmath95 is a lower limit for the mean energy ( centroid ) of the resonance , whereas @xmath94 is an upper limit .
it is clear from the definition of the @xmath101 and @xmath102 moments that @xmath94 and @xmath95 explore different energy domains of the collective excitation , i.e. , @xmath94 is more influenced by the high - energy region of the strength function @xmath87 , whereas @xmath95 is more sensitive to the low - energy components of @xmath87 .
it can be shown @xcite that the quantity @xmath103 provides an upper bound for the width of the giant monopole resonance in the rpa approximation .
the calculations carried out for stable nuclei show that @xmath94 and @xmath95 become increasingly close with increasing nuclear mass .
therefore , in a stable heavy nucleus the value of the width @xmath23 is small and the monopole strength becomes concentrated around a single peak @xcite , which is the situation found in experiment .
following equation ( [ width ] ) and the above comments , in the discussion of our results the spread between the calculated @xmath69 and @xmath96 values will provide an indication about the width of the gmr .
first , in this section we test with some examples the results of the relativistic extended thomas - fermi ( retf ) approach vis - - vis relativistic hartree and rpa calculations . to this end
, we report in table 1 semiclassical retf results for the binding energy and the excitation energy of the gmr in the nuclei @xmath5pb and @xmath104zr .
we also include in the comparison the value of the neutron skin thickness @xmath105 ( the difference between the neutron and proton rms radii of the nucleus ) because this quantity is very sensitive to the density dependence of the symmetry energy @xcite .
the results have been computed with the nl3 mean field interaction @xcite supplemented by the nonlinear @xmath106 coupling between the @xmath7-meson and @xmath8-meson fields @xcite . as pointed out in the introduction , this model , through the isoscalar - isovector meson coupling ,
allows one to change the symmetry energy leaving unchanged the properties of the eos of symmetric nuclear matter . by modifying the value of @xmath106
one can investigate the changes in finite nuclei due to the variation of the density dependence of the symmetry energy almost without affecting the previous quality of the model for binding energies and charge radii @xcite .
.relativistic hartree - rpa and relativistic extended thomas - fermi values obtained with the nl3 model plus the isoscalar - isovector coupling @xmath106 for the binding energy per particle ( mev ) , neutron skin thickness @xmath105 ( fm ) , and excitation energy of the gmr ( mev ) in @xmath5pb and @xmath104zr . the gmr centroid energies @xmath107 are taken from @xcite . [
cols="<,^,^,^,^,^,^,^ " , ] for the coupling constant @xmath106 we use the values 0.00 , 0.01 , 0.02 , and 0.03 ( models denoted as nl3.00 , nl3.01 , nl3.02 and nl3.03 in table 1 ) .
it is known that the binding energy of nuclei constrains an average between the symmetry energy at saturation and the surface symmetry energy , rather than the individual value of the symmetry energy at saturation @xcite .
thus , as done in earlier literature @xcite , we constrain the symmetry energy @xmath108 of the models to take a value of 25.67 mev at a fermi momentum @xmath109 @xmath110 ( i.e. , at a subsaturation density @xmath111 @xmath2 ) .
this implies that the coupling constant @xmath112 of the nl3 interaction has to be suitably adjusted for each value of the isoscalar - isovector @xmath106 coupling . as a consequence , the value @xmath113 of the symmetry energy at the saturation density @xmath114 changes with the @xmath106 parameter .
we find the results @xmath115 , 35.0 , 33.2 and 31.7 mev when we vary @xmath106 from 0.00 to 0.03 .
that is , the symmetry energy @xmath108 of the model around saturation becomes softer ( it increases more slowly with the nuclear density ) when the value of the @xmath106 parameter is larger .
further details on the behaviour of @xmath108 in the nl3 model supplemented by the @xmath106 coupling can be found in @xcite .
table i shows that by increasing the @xmath106 coupling the neutron skin thickness of a neutron - rich nucleus is significantly reduced , i.e. , the models with a softer symmetry energy give rise to smaller neutron skins . here , this is achieved while maintaining the proton rms radius of the nucleus essentially intact .
for example , in the quantal hartree calculation with the nl3.00 model we find @xmath116 fm for @xmath5pb and 4.199 fm for @xmath104zr , while the respective values using nl3.03 are 5.478 fm and 4.210 fm .
the changes in @xmath117 are thus less than 0.5% .
moreover , one sees in table i that also the binding energies are little altered by @xmath106 .
both @xmath5pb and @xmath104zr are slightly more bound at large @xmath106 but the changes in @xmath118 are again no larger than 0.5% .
thus , the predictions for charge radii and binding energies of the original nl3 model are not seriously compromised by the additional @xmath106 coupling , which otherwise largely softens the symmetry energy . from table 1
we observe that in the semiclassical retf approach the binding energies of @xmath5pb and @xmath104zr are larger and the neutron skins smaller , than in the corresponding quantal hartree calculations . in principle
, this fact is not to be considered a lack of accuracy of the thomas - fermi - like methods because semiclassical and quantal calculations must differ by shell effects .
it is well known that the semiclassical ground - state calculations by construction smooth out the quantal shell effects and provide the average part of the quantal binding energy @xcite .
actually , in the semiclassical calculations one works with the particle densities and completely avoids to calculate single - particle wave functions . by this same reason
, the semiclassical approach yields smooth average densities and energies that do not show the quantum shell oscillations . the nuclear etf functionals with gradient corrections up to order @xmath14 always produce some overbinding ( between a @xmath1192% in heavy nuclei and a @xmath11910% in light nuclei ) and yield smaller rms radii ( between @xmath1191% and @xmath1194% ) than the quantal mean field calculations @xcite .
the quantal corrections , which are more prominent in magic nuclei ( our situation in table 1 ) , can eventually be added perturbatively to the semiclassical binding energies using techniques such as the strutinsky shell - correction method @xcite or the expectation value method @xcite .
our focus here is somehow more about the change of the results with the softness of the symmetry energy than on the absolute values themselves . in this respect
, one has to note that the quantal and semiclassical results reported in table 1 show both practically the same variation as a function of the @xmath106 coupling .
the binding energies remain similarly constant and the neutron skins decrease by almost the same amount of 0.08 fm in @xmath5pb and of 0.03 fm in @xmath104zr in both the hartree and retf calculations when @xmath106 is changed from 0.00 to 0.03 . therefore , the retf approach predicts practically the same variation with @xmath106 as the quantal hartree calculations .
this implies that the rate of change of the results with the softness of the symmetry energy in the present model is well described by the retf method . as a further test , which is motivated also because later we will study retf results along isotopic chains , in figure [ f_skindif ] we plot for pb isotopes the difference between the value of the neutron skin thickness of the nucleus calculated with @xmath120 and with @xmath121 , as a function of the parameter @xmath122
we observe by comparison with the corresponding quantal hartree results , that the retf approach performs quite satisfactorily along the pb isotopic chain in predicting the right size of the modification induced by the change of the symmetry energy of the model . obtained with @xmath123 and with @xmath121 for lead isotopes .
the results of the retf method are compared with the results of quantal hartree calculations . ] with regard to the average excitation energies of the gmr , table 1 shows the scaled and constrained values ( i.e. , @xmath69 of section [ escal ] and @xmath96 of section [ econs ] ) for @xmath5pb and @xmath104zr calculated in the retf approach .
we compare our retf results with the values of the centroid energy @xmath107 of the gmr extracted from the relativistic rpa ( rrpa ) strength , which have been reported in @xcite .
we see that for the @xmath5pb nucleus , the rrpa centroid energies very nicely lie in between the retf constrained and scaled estimates , as it is required in the nonrelativistic sum - rule approach to the rpa @xcite ( recall equation ( [ centroid ] ) of section [ resemblance ] ) . in the case of a lower - mass nucleus such as @xmath104zr ,
the semiclassical excitation energies are not expected to be as successful as for the heavier @xmath5pb .
indeed , we see in table 1 that the retf constrained and scaled estimates in @xmath104zr overestimate a little the rrpa centroid energies
. the discrepancies are nevertheless moderate , with differences no larger than 2% for @xmath96 and 5% for @xmath69 with respect to the rrpa @xmath107 value of @xmath104zr . and , as it happens in the case of @xmath5pb , the semiclassical average excitation energies of @xmath104zr as a function of the @xmath106 coupling display the tendency of the rrpa excitation energies .
in summary , we can conclude from the results discussed in the present section that the retf approach is a reasonable starting point to study the influence of the density dependence of the symmetry energy on the gmr of neutron - rich nuclei .
as mentioned in the introduction , we found in @xcite within the skyrme - hartree - fock framework that the semiclassical calculations are able to describe realistically the tendencies of the excitation energy of the gmr in regions away from the stability valley , especially for heavy nuclei such as the lead isotopes where a semiclassical approach performs better .
in the present section we analyze with the retf approach to the relativistic mean field model the effects of the density dependence of the symmetry energy on the excitation energy of the gmr in heavy neutron - rich nuclei by contrasting the results for the nucleus @xmath124pb , which we take as an example of a heavy system with a large neutron excess away from stability , with the results for the stable isotope @xmath5pb .
we display in figure [ f_pb208 ] for @xmath5pb and in figure [ f_pb266 ] for @xmath124pb the calculated values of the energy per particle @xmath125 , the neutron skin thickness @xmath126 , the compression modulus of the finite nucleus @xmath44 and the average excitation energy @xmath98 of the gmr as a function of the coupling constant @xmath106 in the relativistic model .
we recall that the coupling @xmath106 induces a change of the density dependence of the symmetry energy .
the model has a stiff symmetry energy at @xmath120 , while it has a soft symmetry energy at @xmath121 .
the rmf lagrangian supplemented by the nonlinear interaction between the @xmath7 and @xmath8 meson fields leaves the binding energy of the nucleus almost independent of @xmath106 , at least for moderate values of this coupling .
we note from the results for @xmath124pb in figure [ f_pb266 ] that this happens not only in the case of stable nuclei , which we have seen previously for @xmath5pb and @xmath104zr in table 1 , but also in the case of much more neutron - rich nuclei .
the neutron thickness @xmath127 decreases almost linearly with an increase of @xmath106 in both @xmath5pb and @xmath124pb .
the reduction of the @xmath127 value at @xmath121 amounts to about a @xmath128 of the original neutron skin thickness at @xmath120 .
the proton rms radius @xmath117 of @xmath5pb and @xmath124pb has been found to remain nearly constant with @xmath106 , and therefore the decrease of @xmath127 is due almost exclusively to the decrease of @xmath129 .
it is worth mentioning that the alluded variation of the neutron skin thickness of @xmath5pb in the present calculations is close to roughly a 1% change in the neutron rms radius @xmath129 of this nucleus .
that is , the range of variation we have allowed in the density dependence of the symmetry energy by changing @xmath106 between 0.00 and 0.03 induces a change in @xmath129 of @xmath5pb around 1% . in this sense , it is consistent with the uncertainty that will remain in this observable after the prex experiment at the jefferson laboratory @xcite , which aims to measure @xmath129 of @xmath5pb to 1% accuracy via parity - violating electron scattering . of the energy per particle ( a ) , neutron skin thickness ( b ) , finite nucleus compression modulus ( c ) and average excitation energy of the gmr ( d ) for the @xmath5pb nucleus .
] of the energy per particle ( a ) , neutron skin thickness ( b ) , finite nucleus compression modulus ( c ) and average excitation energy of the gmr ( d ) for the @xmath124pb nucleus . ]
the average excitation energy of the gmr calculated in the retf approach involves the knowledge of the finite nucleus compression modulus @xmath44 , or restoring force of the resonance .
we display @xmath44 in figures [ f_pb208](c ) and [ f_pb266](c ) for @xmath5pb and @xmath124pb , respectively .
let us discuss first the change of @xmath44 with the mass number @xmath12 in isotopes .
we see from the results for @xmath5pb and @xmath124pb that the finite nucleus compression modulus decreases when @xmath12 increases .
this behaviour may be easily understood from the so - called leptodermous expansion of @xmath44 @xcite : @xmath130 where @xmath3 is the compression modulus of symmetric infinite nuclear matter and @xmath13 , @xmath131 and @xmath132 are the leading surface , volume - symmetry and coulomb corrections .
the coefficients @xmath13 , @xmath131 , and @xmath132 in the scaling model are negative , and in particular @xmath131 may be large @xcite . within an isotopic chain , the nuclear charge @xmath133
is fixed and therefore the neutron - proton asymmetry @xmath122 increases when @xmath12 increases . then , although the surface and coulomb contributions in ( [ lepto ] ) become a little less negative with increasing @xmath12 , the increase of the negative volume - symmetry contribution @xmath134 dominates .
therefore , @xmath44 in isotopes decreases with larger @xmath12 .
we stress that the expansion ( [ lepto ] ) is provided here for indicative purposes only and that in all results we have computed @xmath44 self - consistently .
the scaling value @xmath65 of the finite nucleus incompressibility , given by equation ( [ eqfn30b ] ) of section [ escal ] , is displayed in figures [ f_pb208](c ) and [ f_pb266](c ) by the upper curves with solid symbols .
for both isotopes @xmath5pb and @xmath124pb , @xmath65 shows an upward trend with larger @xmath106 .
this variation of @xmath65 with @xmath106 is due to the softening of the density dependence of the symmetry energy .
it modifies the value of @xmath131 and consequently @xmath44 ( of course , @xmath131 does not describe alone the whole effect as there are surface - symmetry and other corrections neglected in ( [ lepto ] ) @xcite ) .
the present models have all @xmath135 mev , while the @xmath131 coefficient changes from @xmath136 mev in nl3.00 to @xmath137 mev in nl3.03 .
the recent empirical extractions of @xmath131 suggest @xmath131 values in this range .
for example , isospin diffusion in nuclear reactions favours the constraint @xmath138 mev @xcite , the breathing mode of sn isotopes suggests @xmath139 mev @xcite ( or @xmath140 mev @xcite ) and neutron skins of nuclei point to @xmath141 mev @xcite .
the excitation energy @xmath69 of the monopole oscillation evaluated in the scaling approach is determined by the ratio of the restoring force @xmath65 to the mass parameter @xmath45 ( section [ formal ] ) .
as indicated , to replace @xmath45 by its nonrelativistic limit @xmath74 is a very good approximation . in practice
, this implies that @xmath45 varies with the parameters of the nuclear interaction essentially in the same way as @xmath142 of the nucleus does , see equation ( [ eq13a ] ) .
when @xmath106 is increased , @xmath143 of the matter distribution of the nucleus decreases because of the reduction of the neutron rms radius originated by the softer symmetry energy . as a consequence , the mass parameter of the gmr ( analogously , the @xmath99 sum rule ( [ eq13b ] ) in the nonrelativistic language ) also decreases .
this effect combined with the growing tendency of @xmath65 with @xmath106 implies that the scaling excitation energy of the gmr will be enhanced with a softer symmetry energy , which is confirmed in figures [ f_pb208](d ) and [ f_pb266](d ) where @xmath69 is plotted .
we recall that this enhancement of @xmath69 is due to the different density dependence of the symmetry energy alone because the incompressibility @xmath3 of symmetric matter is forced to be the same in all the interactions considered here .
let us now analyze the results obtained through the constrained calculations described in section [ econs ] .
for the stable nucleus @xmath5pb , we see in figure [ f_pb208 ] that @xmath144 and @xmath96 have a similar behaviour to the discussed scaling results as a function of the density dependence of the symmetry energy .
the separation between @xmath69 and @xmath145 is indicative of the width of the gmr resonance ( see equation ( [ width ] ) ) , and therefore this quantity is basically constant with the symmetry energy in @xmath5pb .
in contrast to this situation in the stable @xmath5pb nucleus , in the more neutron - rich isotope @xmath124pb we see in figure [ f_pb266 ] that the change of @xmath144 and @xmath96 with the symmetry energy exhibits a different pattern from the corresponding scaling results . for @xmath124pb , both quantities @xmath144 and @xmath145 not only do not increase with @xmath106 , but they decrease significantly . to interpret the reduction of @xmath96 and @xmath144 with @xmath106 in @xmath124pb , it will be helpful to rewrite the expression @xmath146 of equation ( [ eqecons ] ) by approximating @xmath45 with its nonrelativistic form @xmath147 , and to write @xmath148 using the first expression given in equation ( [ eqfn60x ] ) .
this results in @xmath149 which in terms of the defined quantities @xmath150 and @xmath151 has the same form that @xmath96 takes in the nonrelativistic rpa formalism of sum rules ( i.e. , @xmath95 of equation ( [ eqfn60y ] ) ) @xcite .
it is to be noted that the denominator of ( [ emcons2 ] ) is very sensitive to the presence of loosely bound nucleons in the nucleus @xcite , which is precisely the situation that occurs in @xmath124pb .
when one is approaching the neutron drip line , the negative chemical potential @xmath152 of neutrons becomes close to vanishing , implying that the system contains weakly bound neutrons .
these neutrons are very soft against compression and in the constrained calculations they give rise to a large increase of the absolute value of @xmath153 , i.e. , of @xmath151 in ( [ emcons2 ] ) .
this effect has been found in @xcite in constrained hf ( and etf ) calculations for nonrelativistic skyrme forces , where it can be related to the enhancement of the rpa strength in the low - energy region produced by the increase of the transitions to the continuum at large neutron numbers .
the same effect is found in the recently published qrpa and chfb skyrme calculations , which take account of the pairing correlations , performed in the light isotopic chains of ca and ni @xcite . in our present calculations , the increase of @xmath151 at large neutron numbers
is accentuated the softer is the symmetry energy ( note that parameter sets having a softer symmetry energy have also weaker binding energy of the last bound neutrons @xcite and therefore reinforce the effect ) .
this explains the decreasing trend of @xmath96 with @xmath106 in @xmath124pb noticed in figure [ f_pb266 ] .
the decrease of @xmath144 with @xmath106 seen in the same figure is similarly understood by recalling equation ( [ eqfn60x ] ) that relates @xmath154 with @xmath155 . in this section
we would like to investigate at which values of the neutron excess of the nucleus the effects coming from the symmetry energy become more relevant in the excitation energies of the gmr in isotopic chains of heavy ( pb ) and medium ( zr ) mass . within the skyrme - hartree - fock framework , in @xcite we analyzed the excitation energy and the resonance width of the gmr in some isotopic chains with the sum - rule approach ( i.e. , through scaling and constrained calculations ) . in the case of a small atomic number such as the ca isotopes ,
the semiclassical etf results were able to describe the global variation of the quantal hf excitation energies but could not reproduce them quantitatively .
however , the etf results for the gmr excitation energies become progressively more accurate with larger atomic number . in the case of the pb chain ,
the etf scaling and constrained excitation energies were found to be in very good agreement with the quantal hf values on the quantitative level from the proton to the neutron drip line ( the differences between etf and quantal values not exceeding 0.4 mev ) @xcite .
thus , from the experience in the nonrelativistic calculations , as well as suggested by the comparison of retf and rrpa results made in table 1 , the present semiclassical predictions for the gmr are expected to be sufficiently realistic , and rather accurate quantitatively ( compared to quantal calculations ) in the case of the pb isotopes .
we depict in figure [ f_pb_ex ] the scaling and constrained average excitation energies of the gmr for the pb isotopic chain . here
we only use the values 0.00 and 0.03 for the nonlinear @xmath7-@xmath8 interaction @xmath106 , representing stiff and soft density dependences of the symmetry energy . in the semiclassical calculations ,
the vanishing of the neutron chemical potential @xmath152 defines the location of the neutron drip line .
we see in figure [ f_pb_ex ] that the latter is reached at a smaller mass number when the model has a soft symmetry energy , in agreement with the earlier observations of @xcite .
note that shell effects are absent in these semiclassical predictions of the drip line ( a quantal hartree calculation with nl3 predicts the pb neutron drip line at @xmath156 ) .
the proton drip line occurs at neutron number @xmath157 not much larger than @xmath133 .
therefore , because one deals with almost symmetric matter , the position of the proton drip line is unaffected by the density dependence of the symmetry energy .
( stiff symmetry energy ) and @xmath121 ( soft symmetry energy ) , in the scaling and constrained approaches . ]
the calculated gmr excitation energies show in figure [ f_pb_ex ] a downward tendency with increasing mass number @xmath12 when one proceeds along the isotopic chain , by similar reasons to those pointed out in our previous discussion about the @xmath5pb and @xmath124pb isotopes .
the downward trend is more noticeable for @xmath96 than for @xmath69 in the isotopes with larger neutron numbers .
the scaling excitation energies @xmath69 obtained using a parameter set with a soft symmetry energy ( @xmath121 ) are systematically higher in the whole isotopic chain than if the symmetry energy is stiff ( @xmath120 ) .
similarly to the scaling results discussed for @xmath5pb and @xmath124pb , this behaviour of @xmath158 is a consequence of the fact that the matter radius of the nucleus is smaller and the scaling incompressibility @xmath65 larger for a soft symmetry energy . at variance with the uniformity displayed by the @xmath69 results ,
the constrained estimate @xmath96 of the gmr excitation energy has a distinct feature with the symmetry energy below and above @xmath159 .
if the mass number is below @xmath160 , @xmath96 is larger when the parameter set has a soft symmetry energy .
however , this tendency is reversed from @xmath159 to the neutron drip line when the isotopes are more neutron rich .
actually , in a given nucleus the @xmath106 parameter reduces its neutron rms radius , which means that the system becomes on average more compact .
as long as the nucleus does not be very neutron rich , if it is more compact , the effect of the constraining parameter @xmath76 on the nuclear rms radius is less , and therefore the absolute value of @xmath153 in the denominator of equation ( [ emcons2 ] ) for @xmath96 is smaller .
this explains the larger magnitude of @xmath96 at @xmath121 , compared to @xmath120 , in the pb isotopes lighter than @xmath159 .
but a point is reached in mass number ( @xmath159 here ) where the isotopes are so neutron rich that the last neutrons become very weakly bound .
these outer neutrons are very soft against compression . as a result of their specific contribution to @xmath161
when one applies an external constraint , the absolute value of @xmath153 increases sharply instead of decreasing .
this originates the large reduction of @xmath96 seen in figure [ f_pb_ex ] for the more neutron - rich isotopes .
the reduction of @xmath96 at large neutron numbers is enhanced when the symmetry energy is soft compared to the case of a stiff symmetry energy because the outer neutrons are more weakly bound in the former case .
a comparable decrease of the excitation energy of the gmr , however , is not present in the scaling calculations because the discussed effect does not arise in a self - similar scaling of the density of the nucleus . as we mentioned in section [ resemblance ] , the scaling and constrained average energies pertain to different energy regimes of the gmr . while @xmath69 informs more about the relatively high - energy contributions to the rpa strength , @xmath96 carries more information about the low - energy sector which is the region that shows the more noticeable changes in largely neutron - rich nuclei because of the presence of weakly bound neutrons
. we may take @xmath162^{1/2}$ ] as an estimate of the width of the gmr , in analogy with the nonrelativistic framework @xcite ( see section [ resemblance ] ) . from the separation observed in figure [ f_pb_ex ] between @xmath69 and @xmath96 , at either @xmath120 or @xmath121 , the width of the resonance
is expected to be almost independent of the mass number and of the density dependence of the symmetry energy for all of the pb nuclei located in the region between the proton - rich side of the isotopic chain till @xmath163240 .
however , @xmath96 starts to show a pronounced departure from @xmath69 after @xmath164 and , as a consequence , the resonance width should display a sharp increase for the pb isotopes with large neutron numbers .
we see from figure [ f_pb_ex ] that this effect is predicted to be stronger when the nuclear interaction has a soft symmetry energy .
as discussed in @xcite in nonrelativistic calculations with skyrme forces , the value of the estimate @xmath23 of the gmr width provides a qualitative idea about the distribution of the rpa strength .
a small width @xmath23 indicates that the rpa strength of the gmr is basically concentrated in a narrow single peak , while a large width @xmath23 suggests that the peak is broad , or even that the rpa strength may be fragmented into several peaks .
the large value of @xmath165 ( see equation ( [ emcons2 ] ) ) that one finds in the constrained calculations for the more neutron - rich pb isotopes , points towards an enhancement of the rpa strength in the low - energy region in these nuclei , due to transitions from the last weakly bound single - particle levels to the continuum @xcite .
( this effect , which has a quantal origin , is accounted at least on average by the semiclassical calculations @xcite . )
the discussed scenario is actually confirmed by the qrpa calculations of @xcite where the fragmentation as well as the enhancement of the low - energy part of the rpa strength in ca and ni isotopes approaching the neutron drip line can be seen in figures 3 and 6 .
( stiff symmetry energy ) and @xmath121 ( soft symmetry energy ) , in the scaling and constrained approaches . ]
lastly , we present in figure [ f_zr_ex ] the calculated results for the excitation energies of the gmr in the zr isotopic chain .
the variation of the excitation energies obtained in the scaling scheme as a function of the mass number @xmath12 and of the @xmath106 coupling is qualitatively similar to the pattern exhibited by the pb isotopes .
that is , the excitation energies @xmath69 show a monotonic decrease with increasing mass number , and they are larger when calculated with a soft symmetry energy .
the relative differences between the values of @xmath69 at @xmath121 and at @xmath120 are smaller compared to the chain of the pb isotopes because of the smaller isospin content of the zr isotopes .
concerning the results for the constrained excitation energies in zr , the value of @xmath96 shows a stronger dependence on the mass number @xmath12 at large neutron numbers than the scaling value @xmath158 .
as long as @xmath166 , the excitation energy @xmath145 for zirconium is to a large extent independent of the symmetry energy .
but afterwards ( @xmath167 ) , @xmath96 becomes visibly lower for the soft symmetry energy . in consonance with the behaviour of @xmath69 and @xmath145 found in figure [ f_zr_ex ]
, it is predicted that the width @xmath23 of the gmr for the zr nuclei should remain practically constant till @xmath168 .
the width @xmath23 would start increasing moderately between @xmath168 and @xmath169 , because of the slight gradual separation between @xmath69 and @xmath145 in these isotopes .
finally , the resonance width would be much enhanced in the region of the most neutron - rich zr isotopes , where the effect is largely reinforced by having a soft symmetry energy in the nuclear interaction .
though the excitation energy of the isoscalar giant monopole resonance is mainly ruled by the compression modulus of symmetric nuclear matter , it is also affected by the symmetry energy in neutron - rich nuclei . in this paper
we have analyzed the influence of the density dependence of the symmetry energy on the average excitation energy of the gmr . in order to do this study ,
we have used a relativistic mean field model supplemented by an additional @xmath7-@xmath8 nonlinear interaction driven by a coupling constant @xmath106 .
this additional coupling allows one to generate different parameter sets that preserve the properties of the symmetric matter and differ among them , only , in the density dependence of the symmetry energy , which is softened by increasing @xmath106 . as a function of the coupling constant @xmath106
, the model predicts in finite nuclei almost the same binding energy and proton rms radius as in the case where @xmath123 , while the neutron rms radius gets significantly reduced .
the excitation energy of the isoscalar giant monopole resonance has been computed using the semiclassical relativistic extended thomas - fermi theory .
we obtain the energy of the breathing mode in two different ways : through the scaling model and performing constrained calculations .
it is known that these calculations with nonrelativistic skyrme forces nicely reproduce the average energies @xmath94 and @xmath95 obtained from the more microscopic rpa calculations , not only for stable nuclei but also for largely neutron - rich nuclei @xcite . in our study , we have calculated the scaling and constrained semiclassical average energies with the relativistic model along the isotopic chains of pb and zr .
we can also obtain some qualitative information about the distribution of the rpa strength from the resonance width computed using these two semiclassical estimates of the excitation energy .
as it is known , the excitation energy of the breathing mode decreases , at least for stable nuclei , following roughly an @xmath170 law @xcite due to the geometrical enhancement of the mass denominator
. however , when the neutron - proton asymmetry increases , departures from this behaviour appear .
the reason is that the isospin - dependent part of the finite nucleus incompressibility @xmath171 plays a non - negligible role by reducing the restoring force and making nuclei with large neutron numbers softer than the stable isotopes .
this effect in the average excitation energy is present in the scaling estimate but it is particularly strong in the constrained calculation . when the neutron - proton asymmetry increases , the outer neutrons become more and more weakly bound starting to behave almost as extremely asymmetric nuclear matter , which is very soft compared to symmetric matter @xcite .
this effect makes the compression modulus @xmath172 smaller and , therefore , the average excitation energy of the gmr decreases . as discussed in the case of
constrained hf calculations with skyrme forces @xcite , this scenario reflects the important contribution to the low - energy part of the rpa strength distribution coming from the particle - hole transitions to the continuum from the weakly bound neutron energy levels in nuclei having large neutron numbers .
a similar situation has been found recently by capelli , col and li @xcite in qrpa and constrained hfb skyrme calculations performed along the light isotopic chains of ca and ni . for a given nucleus
the excitation energy of the gmr also depends on the density content of the symmetry energy of the model used to compute it . in stable nuclei
the scaled and constrained estimates of the excitation energy are larger for parameter sets that have a symmetry energy with a soft density dependence . in the case of the constrained estimate ,
this tendency is reversed in the nuclei with large neutron - proton asymmetry .
the reason lies in the fact that in these nuclei the outermost and weakly bound neutrons are very compressible and consequently the constrained compression modulus @xmath173 becomes lower when the density dependence of the symmetry energy is soft ( the outer neutrons are then less bound ) than when the symmetry energy is stiff .
thus , the average excitation energies based on the inverse energy - weighted sum rule are predicted to be significantly reduced for large neutron numbers when computed with a parameter set that has a softer density dependence of the symmetry energy . in conclusion , the density dependence of the symmetry energy has a moderate influence on the value of the excitation energy of the isoscalar giant monopole resonance in stable nuclei and in nuclei that are not far from stability .
it has a larger impact on the gmr of isotopes with large isospins , where the calculations reveal that the constrained energies become much reduced in models having a soft symmetry energy .
the results point towards a considerable enhancement of the low - energy part of the rpa strength and to likely different structures in the excitation spectra of the gmr in unstable neutron - rich nuclei if the nuclear symmetry energy is soft .
it may be worth pursuing a detailed characterization of these predictions related with the softness of the symmetry energy by computing the strength distribution in a few selected isotopes at large neutron numbers with the more demanding but more microscopic qrpa theory of the gmr response .
unfortunately , it is difficult that the indicated effects be tested in the laboratory as they become prominent at neutron numbers beyond present reach .
such nuclear systems are otherwise expected to exist in astrophysical conditions , such as those prevailing in the inner crust of neutron stars , where exotic nuclear clusters can be stabilized by an enveloping gas of neutrons and electrons @xcite .
it is thought that low - energy collective excitations play an important role in astrophysical processes that involve nuclei far from stability @xcite , and in that context the modification of the low - energy region of the strength distribution of the monopole resonance by the softness of the symmetry energy could have implications . m.c . , x.r . and x.v .
would like to thank j. piekarewicz for useful conversations .
they acknowledge support from the consolider ingenio 2010 programme cpan csd2007 - 00042 and grants fis2008 - 01661 from mec and feder , and 2009sgr-1289 from generalitat de catalunya . x.r . also acknowledges grant ap2005 - 4751 from mec .
s.k.p . and
acknowledge the support of csir ( no . 03(1060 ) 06/emr - ii ) , government of india .
p.d.s . acknowledges support from the uk science and technology facilities council under grants pp / f000596/1 and st / f012012/1 .
we have introduced the energy density of the relativistic nuclear mean field model supplemented by an isoscalar - isovector mixed @xmath7-@xmath8 interaction in equation ( [ eq0a ] ) of the text . following @xcite , in the relativistic extended thomas - fermi approach the expression of the nucleonic part @xmath18 is written as @xmath174 , where @xmath175 \label{eq0c}\ ] ] and @xmath176 .
\label{eq0d}\ ] ] the leading term @xmath32 is the contribution of the pure thomas - fermi part , while the term @xmath33 contains the gradient corrections of order @xmath14 . the variables @xmath177 , @xmath57 , and @xmath178 are , respectively , the local fermi momentum @xmath179 , the dirac effective mass @xmath180 , and the dispersion relation @xmath181 of the effective particle , with @xmath182 for neutrons and @xmath183 for protons . the coefficients
@xmath184 stand for the following analytic functions of @xmath177 and @xmath180 : @xmath185 c_{2q } & = & \frac{m^*}{6 k_{{\rm f}q } \epsilon_{{\rm f}q}^2 } \ln \frac{k_{{\rm f}q } + \epsilon_{{\rm f}q}}{m^ * } , \nonumber\\[3 mm ] c_{3q } & = & \frac{k_{{\rm f}q}^2}{24 \pi^2 \epsilon_{{\rm f}q}^2 } \left[\frac { \epsilon_{{\rm f}q}}{k_{{\rm f}q } } - \left ( 2 + \frac{\epsilon_{{\rm f}q}^2}{k_{{\rm f}q}^2 } \right ) \ln \frac{k_{{\rm f}q } + \epsilon_{{\rm f}q}}{m^ * } \right ] .
\label{eq0f}\end{aligned}\ ] ] the retf ground - state densities and meson fields are calculated by solving by numerical iteration the euler
lagrange equations associated to the energy density ( [ eq0a ] ) , with the constraint of baryon number conservation ( which is imposed through lagrange multipliers @xmath186 , the chemical potentials of neutrons and protons ) : @xmath187 \nonumber \\[3 mm ] & & - 2 c_{1q}\delta\rho_{q } - c_{2q}\delta m^ * - \frac{\partial c_{1q}}{\partial \rho_{q}}({\mbox{\boldmath$\nabla$ } } \rho_q)^2 - 2 \frac{\partial c_{1q}}{\partial m^ * } ( { \mbox{\boldmath$\nabla$ } } \rho_{q } \cdot { \mbox{\boldmath$\nabla$ } } m^ * ) \nonumber \\[3 mm ] & & - \left ( \frac{\partial c_{2q}}{\partial m^ * } - \frac{\partial c_{3q}}{\partial \rho_{q } } \right ) ( { \mbox{\boldmath$\nabla$ } } m^ * ) ^2 = m + \mu_q \ , , \label{eq0h}\end{aligned}\ ] ] @xmath188 ( \delta - m_{v}^2 ) w & = & -g_{v}^2 \left ( \rho - 2 \lambda_v b^2 w \right ) , \label{eqfn5 } \\[3 mm ] ( \delta - m_\rho^2 ) b & = & - g_\rho^2 \left ( \rho_3 - 2 \lambda_v b w^2 \right ) , \label{eqfn6 } \\[3 mm ] \delta { \cal a } & = & - \rho_{p } \ , . \label{eqfn7}\end{aligned}\ ] ] we have @xmath189 for protons and @xmath190 for neutrons . the baryon density is @xmath16 , while @xmath17 is the isovector density of the nucleus . finally , the retf expression of the scalar density @xmath191 entering equation ( [ eqfn4 ] ) is given by @xmath192 & = & \sum_{q } \frac{m^*}{2\pi^2}\left[k_{{\rm f}q}\epsilon_{{\rm f}q}-{m^*}^2 \ln\frac{k_{{\rm f}q } + \epsilon_{{\rm f}q } } { m^*}\right ] \nonumber \\[3 mm ] & & - \sum_q \left [ c_{2q}\delta\rho_{q}+2c_{3q}\delta m^ * + \left(\frac{\partial c_{2q } } { \partial \rho_{q}}-\frac{\partial c_{1q}}{\partial m^*}\right ) ( { \mbox{\boldmath$\nabla$ } } \rho_q)^2 \right .
\nonumber \\[3 mm ] & & \left . \mbox { } + 2 \frac{\partial c_{3q}}{\partial \rho_{q } } ( { \mbox{\boldmath$\nabla$ } } \rho_{q } \cdot { \mbox{\boldmath$\nabla$ } } m^ * ) + \frac{\partial c_{3q}}{\partial m^ * } ( { \mbox{\boldmath$\nabla$ } } m^ * ) ^2 \right ]
. \label{eq0g}\end{aligned}\ ] ]
in this appendix we illustrate how to calculate the derivatives of the scaled meson fields upon the scaling parameter @xmath4 that we have used to represent the collective coordinate of the monopole oscillation of the nucleus in the scaling approach . in view of equation ( [ eq4 ] ) for the scaled baryon density @xmath193 ,
i.e. , @xmath194 , and of the field equation ( [ eqfn5 ] ) obeyed by the omega - meson field @xmath195 , the scaled field @xmath196 fulfills the klein
gordon equation @xmath197 where we have introduced the coordinate @xmath198 . on differentiating equation ( [ eqap1 ] ) with respect to the scaling parameter @xmath4 we have @xmath199 & & -g_{v}^2 \left ( \rho + \frac{2 m_{v}^2}{g_v^2}\frac{w_\alpha } { \alpha^3 } + \frac{4 \lambda_v}{\alpha^3}b_\alpha^2 w_\alpha - \frac{4 \lambda_v}{\alpha^2}b_\alpha w_\alpha \frac{\partial b_\alpha}{\partial\alpha } - \frac{2 \lambda_v}{\alpha^2}b_\alpha^2 \frac{\partial w_\alpha}{\partial\alpha } \right ) .
\label{eqap2}\end{aligned}\ ] ] if we now set @xmath200 , the solution of this equation allows us to obtain the value of the derivative latexmath:[$\partial w_\alpha /\partial\alpha in the case of the scalar field one has to take account of the terms that arise due to the fact that the scalar density @xmath191 is itself a function of the scalar field .
following the same steps as above , from the klein - gordon equation @xmath204 for the scaled scalar field @xmath205 , one readily arrives at the equation @xmath206 whose solution at @xmath68 provides the value of @xmath207 .
99 steiner a w , prakash m , lattimer j m and ellis p j 2005 _ phys .
rep . _ * 411 * 325 + ( arxiv : nucl - th/0410066 ) klhn t
_ et al _ 2006 _ phys .
_ c * 74 * 035802 ( arxiv : nucl - th/0602038 ) lattimer j m and prakash m 2007 _ phys
. rep .
_ * 442 * 109 ( arxiv : astro - ph/0612440 ) li b a , chen l w and ko c m 2008 _ phys . rep . _ * 464 * 113 ( arxiv:0804.3580 ) xu j , chen l w , li b a and ma h r 2009 _ ap . j. _ * 697 * 1549 ( arxiv:0901.2309 ) bednarek i and manka r 2009 _ j. phys .
g : nucl .
phys . _ * 36 * 095201 ( arxiv:0905.0131 ) brown b a 2000 _ phys . rev .
lett . _ * 85 * 5296 horowitz c j and piekarewicz j 2001 _ phys .
lett . _ * 86 * 5647 horowitz c j and piekarewicz j 2001 _ phys .
_ c * 64 * 062802(r ) todd b g and piekarewicz j 2003 _ phys .
_ c * 67 * 044317 danielewicz p 2003 _ nucl .
* 233 centelles m , roca - maza x , vias x and warda m 2009 _ phys .
lett . _ * 102 * 122502 + ( arxiv:0806.2886 ) warda m , vias x , roca - maza x and centelles m 2009 _ phys .
_ c * 80 * 024316 + ( arxiv:0906.0932 ) baran v , colonna m , greco v and di toro m 2005 _ phys
. rep . _ * 410 * 335 + ( arxiv : nucl - th/0412060 ) shetty d v , yennello s j and souliotis g a 2007 _ phys .
_ c * 76 * 024606 ( arxiv:0704.0471 ) tsang m b , zhang y , danielewicz p , famiano m , li z , lynch w g and steiner a w 2009 _ phys . rev .
* 102 * 122701 ( arxiv:0811.3107 ) klimkiewicz a _ et al _ 2007 _ phys .
_ c * 76 * 051603(r ) trippa l , col g and vigezzi e 2008 _ phys .
_ c * 77 * 061304(r ) ( arxiv:0802.3658 ) chowdhury p roy , basu d n and samanta c 2009 _ phys . rev
. _ c * 80 * 011305(r ) ( arxiv:0905.1599 ) khoa dao t , von oertzen w and ogloblin a a 1996 _ nucl . phys . _
a*602 * 98 ozawa a , suzuki t and tanihata i 2001 _ nucl .
a*693 * 32 khoa dao t , than h s , nam t h , grasso m and van giai n 2004 _ phys .
_ c * 69 * 044605 ed seweryniak d and khoo
l 2005 _ aip conf .
_ vol 764 ed gross c j , nazarewicz w and rykaczewski k p 2005 _ eur .
j. _ a * 25 * s01 harakeh m n and van der woude a 2001 _ giant resonances _ ( new york : oxford university press ) harakeh m n 2002 _ eur
. phys .
j. _ a * 13 * 169 youngblood d h , clark h l and lui y - w 1999 _ phys . rev . lett . _ * 82 * 691 li t
_ et al _ 2007 _ phys .
lett . _ * 99 * 162503 ( arxiv:0709.0567 ) monrozeau c _ et al _ 2008 _ phys . rev .
lett . _ * 100 * 042501 blaizot j p , berger j f , decharg j and girod m 1995 _ nucl .
phys . _ a * 591 * 435 farine m , pearson j m and tondeur f 1997 _ nucl .
a * 615 * 135 hamamoto i , sagawa h and zhang x z 1997 _ phys .
_ c * 56 * 3121 col g , van giai n , meyer j , bennaceur k and bonche p 2004 _ phys .
_ c * 70 * 024307 + ( arxiv : nucl - th/0403086 ) ma zhong - yu , van giai n , wandelt a , vretenar d and ring p 2001 _ nucl .
phys . _ a * 686 * 173 todd - rutel b g and piekarewicz j 2005 _ phys . rev . lett .
_ * 95 * 122501 ( arxiv : nucl - th/0504034 ) piekarewicz j 2002 _ phys .
_ c * 66 * 034305 vretenar d , niki t and ring p 2003 _ phys .
_ c * 68 * 024310 agrawal b k , shlomo s and au v k 2003 _ phys .
_ c * 68 * 031304(r ) piekarewicz j 2004 _ phys .
_ c * 69 * 041301(r ) ( arxiv : nucl - th/0312020 ) piekarewicz j 2007 _ phys .
_ c * 76 * 031301(r ) ( arxiv:0705.1491 ) sil tapas , centelles m , vias m and piekarewicz j 2005 _ phys .
_ c * 71 * 045502 + ( nucl - th/0501014 ) bohigas o , lane a and martorell j 1979 _ phys .
_ * 51 * 267 lipparini e and stringari s 1989 phys .
* 175 * 103 gleissl p , brack m , meyer j and quentin p 1990 _ ann .
( ny ) _ * 197 * 205 centelles m , vias x , patra s k , de j n and sil tapas 2005 _ phys . rev
. _ c * 72 * 014304 ( arxiv : nucl - th/0502055 ) sil tapas , shlomo s , agrawal b k and reinhard p - g 2006 _ phys .
_ c * 73 * 034316 + ( arxiv : nucl - th/0601091 ) khan e , sandulescu n , grasso m and giai n v 2002 _ phys . rev .
_ c * 66 * 024309 sil tapas _ et al _ 2002 _ phys . rev .
_ c * 66 * 045803 khan e 2009 _ phys .
_ c * 80 * 057302 ( arxiv:0907.3423 ) khan e 2009 _ phys .
_ c * 80 * 011307(r ) ( arxiv:0905.3335 ) centelles m , pi m , vias x , garcias f and barranco m 1990 _ nucl .
a*510 * 397 guo - quiang li and gong - ou xu 1990 _ phys .
_ c * 42 * 290 guo - quiang li 1991 _ j phys .
g : nucl . part .
* 17 * 1 centelles m , vias x , barranco m and schuck p 1993 _ ann .
( ny ) _ * 221 * 165 centelles m , vias x , barranco m , marcos s and lombard r j 1992 _ nucl . phys . _
a*537 * 486 speicher c , engel e and dreizler r m 1993 _ nucl .
a*562 * 569 centelles m , del estal m and vias x 1998 _ nucl . phys . _
a*635 * 193 centelles m _ et al _ 1993 _ phys .
_ c * 47 * 1091 patra s k , vias x , centelles m and del estal m 2001 _ phys .
_ b * 523 * 67 patra s k , vias x , centelles m and del estal m 2002 _ nucl .
a*703 * 240 patra s k , centelles m , vias x and del estal m 2002 _ phys . rev .
_ c * 65 * 044304 nishizaki s , kurasawa h and suzuki t 1987 _ nucl .
a*462 * 689 zhu chaoyuan and qiu xi - jun 1991 _ j. phys .
g : nucl .
* 17 * l11 maruyama t and suzuki t 1989 _ phys . lett . _
b*219 * 43 boersma h f , malfliet r and scholten o 1991 _ phys .
b*269 * 1 stoitsov m v , ring p and sharma m m 1994 _ phys .
_ c * 50 * 1445 stoitsov m v , cescato m l , ring p and sharma m m 1994 _ j phys .
g : nucl . part .
phys . _ * 20 * l149 lalazissis g a , knig j and ring p 1997 _ phys .
rev . _ c * 55 * 540 ring p and schuck p 1980 _ the nuclear many body problem _ ( berlin : springer - verlag ) brack m , guet c and hkansson h - b 1985 _ phys . rep . _ * 123 * 275 brack m and bhaduri r k 1997 _ semiclassical physics _
( reading : addison - wesley ) blaizot 1980 j p _ phys . rep .
_ * 64 * 171 chossy t and stocker w 1997 _ phys .
_ c * 56 * 2518 sagawa h , yoshida s , zeng g m , gu j z and zhang x z 2007 _ phys .
_ c * 76 * 034327 ( arxiv:0706.0966 ) piekarewicz j and centelles m 2009 _ phys .
_ c * 79 * 054311 ( arxiv:0812.4499 ) bombaci i and lombardo u 1991 _ phys .
_ c * 44 * 1892 paar n , vretenar d , khan e and col g 2007 _ rep . prog . phys .
_ * 70 * 691 + ( arxiv : nucl - th/0701081 ) | we analyze the influence of the density dependence of the symmetry energy on the average excitation energy of the isoscalar giant monopole resonance ( gmr ) in stable and exotic neutron - rich nuclei by applying the relativistic extended thomas - fermi method in scaling and constrained calculations . for the effective nuclear interaction ,
we employ the relativistic mean field model supplemented by an isoscalar - isovector meson coupling that allows one to modify the density dependence of the symmetry energy without compromising the success of the model for binding energies and charge radii .
the semiclassical estimates of the average energy of the gmr are known to be in good agreement with the results obtained in full rpa calculations .
the present analysis is performed along the pb and zr isotopic chains . in the scaling calculations ,
the excitation energy is larger when the symmetry energy is softer .
the same happens in the constrained calculations for nuclei with small and moderate neutron excess .
however , for nuclei of large isospin the constrained excitation energy becomes smaller in models having a soft symmetry energy .
this effect is mainly due to the presence of loosely - bound outer neutrons in these isotopes .
a sharp increase of the estimated width of the resonance is found in largely neutron - rich isotopes , even for heavy nuclei , which is enhanced when the symmetry energy of the model is soft .
the results indicate that at large neutron numbers the structure of the low - energy region of the gmr strength distribution changes considerably with the density dependence of the nuclear symmetry energy , which may be worthy of further characterization in rpa calculations of the response function . |
quantum key distribution can enable two authentic parties , the sender ( alice ) and the receiver ( bob ) , to obtain unconditional secret keys without restricting the power of the eavesdropper ( eve ) @xcite .
on the premise of unconditional security , the higher key rate and the longer distance are constantly pursued @xcite . to enhance the tolerable excess noise of the continuous - variable quantum key distribution ( cv qkd ) @xcite ,
the two - way cv qkd protocols are proposed @xcite , where bob initially sends a mode to alice , and alice encodes her information by applying a random displacement operator to the received mode and then sends it back to bob .
bob detects both his original mode and received mode to decode alice s modulations .
although the two - way cv qkd protocols can remarkably enhance the tolerable excess noise @xcite , it needs to implement the tomography of the quantum channels to analyze the security under general collective attack @xcite , which is complicated in practice .
therefore , we proposed a feasible modified two - way protocol by replacing the displacement operation of the original two - way protocol with a passive operation on alice s side @xcite .
however , the source noise @xcite and both detection efficiency and detection noise @xcite on bob s side are not considered in the modified protocol .
it has been proved that adding a proper noise on bob s detection side in one - way cv qkd can enhance the tolerable excess noise and the secret key rate in reverse reconciliation @xcite .
this idea has been applied to the original two - way protocol in @xcite , while the scheme did not consider the correlation between the two channels .
the correlated noise affects the secret key rate @xcite . in this paper ,
we apply the idea of adding noise to our modified two - way protocol @xmath1 @xcite to enhance the tolerable excess noise and the secret key rate . considering the correlation between the channels ,
the security of the two - way cv qkd with added noise against entangling cloner collective attacks @xcite is analysed and numerically simulated .
the entanglement - based ( eb ) scheme of the two - way cv - qkd protocol @xmath2 with added detection noise is shown in ( a ) , where the dashed box at @xmath3 is the added noise and the other part is our original two - way protocol @xmath1 @xcite . the added noise is equivalent to an einstein - podolsky - rosen ( epr ) pair with the variance of @xmath4 coupled into the channel by a beam splitter with the transmittance of @xmath5 . the protocol @xmath2 is described as follows .
_ step one_. bob initially keeps one mode @xmath6 of an epr pair with the variance of @xmath7 while sending the other mode @xmath8 to alice through the forward channel . _
step two_. alice measures one mode @xmath9 of her epr pair ( variance : @xmath10 ) to get the variables \{@xmath11 , @xmath12 } with a heterodyne detection , and couples the other mode of her epr pair with the received mode @xmath13 from bob by a beam splitter ( transmittance : @xmath14 ) .
one output mode @xmath15 of the beam splitter is measured with homodyne detection and the other output mode @xmath16 is sent back to bob through the backward channel . _
step three_. with a beam splitter ( transmittance : @xmath5 ) , bob couples another epr pair ( variance : @xmath4 ) which is equivalent to the added noise with his received mode .
the two modes @xmath17 and @xmath18 of this epr pair are measured .
bob performs homodyne detections on both modes @xmath6 and @xmath3 to get the variables @xmath19 ( or @xmath20 ) and @xmath21 ( or @xmath22 ) , respectively .
_ step four_. alice and bob implement the reconciliation and privacy amplification @xcite . in this step ,
the measurement values of the modes @xmath3 , @xmath6 , @xmath15 , @xmath9 , @xmath18 and @xmath17 are used to estimate the channel s parameters and bob uses @xmath23 ( @xmath24 ) to construct the optimal estimation to alice s corresponding variables @xmath11 ( @xmath12 ) , where @xmath25 is the channel s total transmittance .
the prepare - and - measure ( pm ) scheme of the two - way protocol can be equivalent to the eb scheme .
in fact , alice heterodyning one half of the epr pair at @xmath9 is equivalent to remotely preparing a coherent state , and bob performing homodyne detection on one half of the epr pair at @xmath6 is equivalent to remotely preparing a squeezed state @xcite .
the homodyne detection preceded by an epr pair coupled by a beam splitter at @xmath3 is equivalent to bob s real homodyne detection with efficiency @xmath5 and electronic noise @xcite .
note that @xmath26 and @xmath27 quadratures are randomly measured in homodyne detection and only @xmath26 quadrature is analysed in the following .
( a ) the eb scheme of @xmath28 protocol .
bob keeps one half of the epr pair ( epr ) and sends the other half to alice .
alice measures one mode of her epr pair ( @xmath29 ) and one mode @xmath15 from a beam splitter @xmath14 .
the other mode from this beam splitter is returned back to bob .
the letters ( e.g. @xmath30 ) beside arrows : the mode at the arrow ; e : eve s whole mode ; the dashed box at @xmath9 : the heterodyne detection ; the dashed box at @xmath3 : the added noise . ( b ) the equivalent scheme to ( a ) with postprocessing . bob uses a symplectic transformation @xmath31 to change the modes @xmath3 and @xmath6 into @xmath32 and @xmath33.,width=396 ]
first , we show that the gaussian attack is optimal to the two - way protocol @xmath28 in general collective attack . in ( a ) ,
since all modes of alice and bob are measured , eve can get the purification of the state of alice and bob .
in addition , the @xmath26 and @xmath27 quadratures of alice and bob s modes are not mixed via heterodyne or homodyne detection and alice and bob use the second - order moments of the quadratures to bound eve s information .
therefore , the two - way protocol @xmath28 can satisfy the requirement of optimality of gaussian collective attack ( i.e. , continuity , invariance under local gaussian unitary and strong subadditivity ) @xcite . when the corresponding covariance matrix of the state @xmath34 is known for alice and bob
, the gaussian attack is optimal @xcite .
therefore , only eve s gaussian collective attack is needed to be considered in the following security analysis . in ( a ) , the secret key rate of the two - way protocol @xmath28 in reverse reconciliation is @xcite @xmath35 where @xmath36 is the reconciliation efficiency , @xmath37 [ @xmath38 is the mutual information between bob and alice ( eve ) , @xmath39 and @xmath40 are alice s variance and conditional variance , @xmath41 and @xmath42 are eve s von neumann entropy and conditional von neumann entropy on bob s data , respectively . in the following , @xmath41 and @xmath42
are calculated by the methods in @xcite . for gaussian state , the entropy can be calculated from its corresponding covariance matrix @xcite . since the state @xmath43 is a pure state , then @xmath44 .
the corresponding covariance matrix of the state @xmath34 is @xmath45 where @xmath46 is a @xmath47 identity matrix , the diagonal elements correspond to the variances of @xmath26 and @xmath27 quadratures of the modes @xmath3 , @xmath6 , @xmath18 , @xmath17 , @xmath15 and @xmath9 in turn , e.g. @xmath48 , and the nondiagonal elements correspond to the covariances between modes , e.g. @xmath49 .
therefore , eve s entropy @xcite @xmath50 where @xmath51 and @xmath52 is the symplectic eigenvalue of @xmath53 which is the function of the element @xmath54 of @xmath53 @xcite , seen in [ appendixa ] .
bob uses @xmath55 to estimate alice s variable , which is equivalent to that bob uses a symplectic transformation @xmath31 to change the modes @xmath3 and @xmath6 into the modes @xmath33 and @xmath32 where the @xmath26 quadrature of the mode @xmath33 is @xmath56 @xcite , as shown in ( b ) . since ( b ) is equivalent to ( a ) with postprocessing , we use ( b ) to calculate @xmath42 in the following .
after the symplectic transformation @xmath31 , the corresponding covariance matrix of the mode @xmath57 is @xmath58\gamma_{b_2b_1n_2n_1a_2a_1}[\gamma_k\oplus\mathbb{i}_4]^t,\ ] ] where @xmath59 , @xmath60 is a continuous - variable c - not gate @xcite @xmath61 since the state @xmath62 is a pure state when bob gets @xmath63 by measuring the modes @xmath33 , then @xmath64 .
the corresponding covariance matrix of the state @xmath65 conditioned on @xmath63 is @xcite @xmath66^{mp}c_{b_4},\ ] ] where @xmath67 and @xmath68 are the corresponding reduced matrixes of the states @xmath65 and @xmath33 in @xmath69 , respectively , @xmath70 is their correlation matrix , @xmath71 and @xmath72 denotes the inverse on the range .
therefore , we have @xmath73 where @xmath74 is the symplectic eigenvalue of @xmath75 which is the function of the element @xmath76 of @xmath75 @xcite , seen in [ appendixa ] .
by substituting equations ( [ se ] ) and ( [ s(e|xbpb ) ] ) into equation ( [ kr ] ) , the secret key rate is obtained @xmath77 in experiment , alice and bob can calculate the element @xmath54 and @xmath76 of equations ( [ rhoab ] ) and ( [ gammconditionxb ] ) by the measurement values of the modes @xmath3 , @xmath6 , @xmath18 , @xmath17 , @xmath15 and @xmath9 .
therefore , according to equation ( [ kr(amn ) ] ) , the secret key rate in general collective attack is obtained without the assumption that the two channels are uncorrelated .
the analytic representations of equation ( [ kr(amn ) ] ) is too complex to give here .
we give a numerical simulation in the following .
the eb scheme of @xmath28 protocol against entangling cloner attacks on correlated channels .
@xmath78 , @xmath79 : the modes introduced into the channels ; @xmath80 : half beam splitter ; @xmath81 : beam splitter .
alice and bob are the same as ( a).,width=396 ] ( a ) tolerable excess noise as a function of the transmission distance for @xmath2 , @xmath1 , @xmath82 and @xmath83 protocols , where @xmath84 .
( b ) tolerable excess noise as a function of the transmission distance for @xmath2 protocol , where @xmath85 .
the curves of ( a ) and ( b ) are plotted for @xmath86 , @xmath87 and @xmath88.,width=396 ] ( a ) secret key rate as a function of the transmission distance for @xmath2 , @xmath1 , @xmath89 , @xmath82 and @xmath83 protocols , where @xmath86 , @xmath90 , @xmath84 , @xmath87 and @xmath88 .
( b ) optimal choice of the added noise @xmath91.,width=396 ] ( a ) secret key rate as a function of the transmission distance for @xmath2 protocol , where @xmath90 , @xmath84 , @xmath87 , @xmath88 and @xmath92 .
( b ) optimal choice of the added noise @xmath91.,width=396 ] ( a ) tolerable excess noise as a function of the transmission distance for high modulation for @xmath2 protocol .
( b ) secret key rate as a function of the transmission distance for high modulation for @xmath2 protocol , where @xmath93 .
the curves of ( a ) and ( b ) are plotted for @xmath94 , @xmath86 , @xmath87 and @xmath95.,width=396 ]
for simplicity in numerical simulation , when there is no eve , the forward and the backward channels are assumed to be independent with the identical transmittances @xmath96 and noises referred to the input @xmath97 , where @xmath98 is the channel excess noises referred to the input . it is equivalent to eve implementing two independent collective entangling cloner attacks which are a gaussian collective attack investigated in detail in @xcite . when eve implements more complicated two - mode attack @xcite , the correlation between the two channels is induced .
shows that eve implements two correlated entangling cloner attacks . on condition
that eve introduces the equivalent variances of the modes @xmath78 and @xmath79 into the two channels , the noise referred to the input of the backward channel is @xmath99 , where the second item on the right - hand side is induced extra by the correlation between the two channels , i.e. , the part of the mode introduced into the backward channel correlating with the forward channel interferes with the mode from alice , @xmath100 is the coefficient representing the degree of the correlation , e.g. , @xmath86 represents that the two channels are uncorrelated .
the added noise is @xmath101 .
we can calculate the elements of equation ( [ rhoab ] ) @xmath102 \ } \mathbb{i},\nonumber \\ & \gamma_{n_2}=\{t\!_n v\!_n + t ( 1 - t\!_n ) [ v\!_a - t\!_a v\!_a +
t t\!_a ( v + \chi ) + \chi_2 ] \ } \mathbb{i},\nonumber \\ & \gamma_{a_2}= [ t\!_a v\!_a + t ( 1- t\!_a ) ( v + \chi ) ] \mathbb{i},\nonumber \\ & c_1\!= -\eta c_6=t\sqrt{t\!_a t\!_n ( v^2 - 1 ) } \sigma_{\!z } , \nonumber \\ & c_2\!=\!\sqrt{(1\!-\ ! t\!_n ) t\!_n } \{v\!_n - t [ v\!_a - t\!_a v\!_a + t t\!_a ( v + \chi ) + \chi_2]\ } \mathbb{i } , \nonumber \\ & c_3\!=\frac{1}{\eta}c_8=\sqrt{(1- t\!_n ) ( v_n^2 - 1 ) } \sigma_{\!z } , \nonumber \\ & c_4\!=-\eta c_9=\!\!\!\sqrt{t ( \!1\!-\ ! t\!_a\ ! ) t\!_n } \left[\!\!\sqrt{t\!_a}v\!_a\ ! -\
! t\ ! \sqrt{t\!_a}(\!v \!+\ !
\chi ) \!-\!n_c\!\sqrt{t}\varepsilon\right ] \mathbb{i } , \nonumber \\ & c_5\!=-\eta c_{10}=\sqrt{t ( 1- t\!_a ) t\!_n ( v^2\!_a-1 ) } \sigma_{\!z } , \nonumber \\ & c_7\!=-\sqrt{t ( 1 - t\!_a ) ( v^2 - 1 ) } \sigma_{\!z } , \nonumber \\ & c_{11}\!=\sqrt{t\!_a ( v^2\!_a-1 ) } \sigma_{\!z},\end{aligned}\ ] ] and @xmath103 where @xmath104 the typical fiber channel loss is assumed to be 0.2 db / km . @xmath7 and @xmath98 are in shot - noise units . substituting equations ( [ elementgmmaab ] ) and ( [ iba ] ) into equation ( [ kr(amn ) ] ) , the optimal secret key rate @xmath105 and the optimal tolerable excess noise @xmath98 of the two - way protocol @xmath106 can be obtained by adjusting the added noise @xmath107 .
when @xmath86 , the two channels are uncorrelated , which is equivalent to eve implementing two independent gaussian cloner attacks . for comparison , the heterodyne protocol ( @xmath83 ) @xcite and the homodyne protocol ( @xmath82 ) @xcite of one - way cv - qkd protocol with coherent state and the original modified two - way protocols @xmath108 and @xmath109 @xcite are also given in figures [ fig4tolerablenoiseuncorrelated](a ) and [ fig5kuncorrelated](a ) .
( a ) shows the tolerable excess noise as a function of the transmission distance , where @xmath88 , @xmath87 , @xmath84 and @xmath86 .
the proper added noise @xmath107 is chosen to make @xmath98 of @xmath106 protocol optimal .
the numerical simulation result indicates that the tolerable excess noise of the two - way protocol with added noise is more than that without added noise and surpasses that of the one - way cv - qkd protocol .
therefore , it indicates that properly added noise is useful to enhance @xmath98 in the two - way protocol .
( b ) shows the tolerable excess noise @xmath98 of @xmath106 protocol with different @xmath36 , which indicates that the tolerable excess noise @xmath98 increases with the increase of @xmath36 .
\(a ) shows the secret key rate as a function of the transmission distance , where @xmath110 , @xmath87 , @xmath90 , @xmath84 and @xmath86 . to make the secret key rate of @xmath2 optimal , the proper added noise @xmath107 is chosen , as shown in ( b ) . in ( a )
, the simulation result indicates that the two - way protocol with added noise has higher secret key rate than that without added noise .
especially , the achievable transmission distance of the two - way protocol @xmath106 is over 60 km when @xmath36 is @xmath0 , which is much longer than that of the one - way protocol .
the reason is that the added noise not only lowers the mutual information between alice and bob , but also lowers that between bob and eve .
when the effect on eve is more than that on alice and bob , the secret key rate is enhanced . when @xmath111 , the two channels are correlated .
( a ) shows the secret key rate as a function of the transmission distance for @xmath2 protocol with different @xmath112 . considering the practical experiment @xcite , we choose @xmath90 , @xmath84 , @xmath87 , @xmath88 and @xmath92 .
to make @xmath105 optimal , the proper added noise @xmath107 is chosen , as shown in ( b ) . in ( a ) , the simulation result indicates that the distance of the secret key distribution decreases with the increase of @xmath112 .
the reason is that the correlation between the two channels induces the change of the excess noise in the backward channel , which affects the secret key rate .
( a ) shows that the decrease of the secret key rate induced by this effect is small .
in addition , comparing with the one - way protocol in ( a ) , despite the transmission distance of the two - way protocol decreases slightly due to the correlation , the performance of the two - way protocol is still far beyond that of the one - way protocols .
( b ) shows that the optimal added noise decreases with the decrease of @xmath112 . in the following ,
we compare the two - way protocol with the one - way protocols in high modulation .
figures [ fig6ek](a ) and ( b ) show the tolerable excess noise and the secret key rate as a function of the transmission distance for high modulation , where @xmath94 , @xmath87 , @xmath95 and @xmath86 .
the proper added noise @xmath107 is chosen to make the tolerable excess noise and the secret key rate of @xmath2 protocol optimal .
the numerical simulation result indicates that both the tolerable excess noise and the secret key rate of the two - way protocol with added noise are much more than that of the one - way cv - qkd protocols for high modulation .
in conclusion , we improve the two - way cv - qkd protocol by adding a proper noise on bob s detection side .
the security of the two - way cv - qkd protocol with added noise in homodyne detection against general collective attack is analysed .
the numerical simulation under the collective entangling cloner attack is given for the correlated and the uncorrelated channels .
the simulation result indicates that despite the secret key rate for the correlated channels is slightly lower than that for the uncorrelated channels when eve inputs equivalent variance of the modes into the two channels , the performance of the two - way protocol is still far beyond that of the one - way protocols .
in addition , the properly added noise is beneficial for enhancing the secret key rate and the tolerable excess noise of the two - way cv qkd .
the optimal tolerable excess noise of the two - way cv qkd with added noise is much more than that of the one - way cv qkd . with the reasonable reconciliation efficiency of @xmath0 , the two - way cv qkd with added noise allows the distribution of secret keys over 60 km fibre distance , which is difficult to reach for the one - way cv - qkd protocols with gaussian modulation in experiment .
this work is supported by the national science fund for distinguished young scholars of china ( grant no .
61225003 ) , national natural science foundation of china ( grant no .
61101081 ) , and the national hi - tech research and development ( 863 ) program .
we use @xmath113 to denote the elements of the corresponding covariance matrix @xmath114 of a @xmath115-mode state .
the symplectic invariants @xmath116 of @xmath114 for @xmath117 are defined as @xcite @xmath118 where @xmath119 ( @xmath120 standing for the @xmath121 pauli matrix ) and @xmath122 is the principal minor of order @xmath123 of the @xmath124 matrix @xmath125 which is the sum of the determinants of all the @xmath126 submatrices of @xmath127 @xcite .
the symplectic eigenvalues of the matrix corresponding to a four - mode state are the solution of the four - order equation on the symplectic invariants @xcite @xmath128 where @xmath129 from equation ( [ rhoab ] ) , the covariance matrix @xmath130 of the modes @xmath131 can be obtained by permuting the corresponding elements of @xmath132 . applying a unitary transformation @xmath133 to equation ( [ rhoab ] )
, we can obtain @xmath134 where @xmath135 and @xmath136/t_n$ ] , @xmath137 for @xmath138 .
therefore , the eigenvalues of @xmath132 are @xmath139 , where @xmath140 are the eigenvalues of @xmath141 given by equation ( [ lamda1234 ] ) . the symplectic invariants of @xmath142 are denoted as @xmath143 for j=1 ... 5 .
it can be proved that @xmath144 .
therefore , one of the eigenvalues of @xmath142 is 1 and the others have the same forms of equation ( [ lamda1234 ] ) , which needs the replacement @xmath145
99 scarani v , bechmann - pasquinucci h , cerf n j , duek m , ltkenhaus n and peev m 2009 _ rev .
phys . _ * 81 * 1301 weedbrook c , pirandola s , lloyd s and ralph t c 2010 _ phys . rev . lett . _
* 105 * 110501 shen y , yang j and guo h 2009 _ j. phys .
b : at . mol .
phys . _ * 42 * 235506 shen y , peng x , yang j and guo h 2011 _ phys . rev . a _ * 83 * 052304 | we propose an improved two - way continuous - variable quantum key distribution ( cv qkd ) protocol by adding proper random noise on the receiver s homodyne detection , the security of which is analysed against general collective attacks .
the simulation result under the collective entangling cloner attack indicates that despite the correlation between two - way channels decreases the secret key rate relative to the uncorrelated channels slightly , the performance of the two - way protocol is still far beyond that of the one - way protocols .
importantly , the added noise in detection is beneficial for the secret key rate and the tolerable excess noise of this two - way protocol . with the reasonable reconciliation efficiency of @xmath0 , the two - way cv qkd with added noise
allows the distribution of secret keys over 60 km fibre distance . |
ultra - compact dwarfs are low mass stellar systems with properties in between those of bright globular clusters and the most compact dwarf elliptical galaxies . since their discovery ucds have been linked to galaxy clusters . initially found in the fornax cluster ( hilker et al . 1999 ; drinkwater et al .
2000 ) ucds have been reported in virgo , centaurus , hydra , coma and the more distant clusters abell s040 and abell 1689 ( hasegan et al .
2005 ; evstigneeva et al . 2008 ;
chilingarian & mamon 2008 ; mieske et al .
2004 , 2007 ; wehner & harris 2007 ; misgeld et al . 2008 ; madrid et al . 2010 ; blakeslee & barber degraaff 2008 ) . the search for ucds has been focused in galaxy clusters and observations seeking to discover ucds in other environments are critically needed to understand their formation process and to determine how common these low mass stellar systems truly are
. only recent accounts of ucds outside galaxy clusters with spectroscopic confirmation exist in the literature .
evstigneeva et al . (
2007 ) search for ucds in six galaxy groups using ground based data and found only one definite candidate in the dorado group .
romanowsky et al .
( 2009 ) describe ucds in the galaxy group of ngc 1407 while hau et al .
( 2009 ) find a ucd located in the vicinity of the sombrero galaxy , a low density environment . da rocha et al .
( 2011 ) give a detailed account of the ucds present in two hickson compact groups .
recently , and after discovering ucds in field ( ngc 4546 ) and group ( ngc 3923 ) environments norris & kannappan ( 2011 ) claim the ubiquity of ucds in galaxies with populous globular cluster systems ( gcs ) .
a search for ucds is undertaken in the isolated giant elliptical ngc 1132 .
this galaxy is likely the only remnant of a merged galaxy group , or fossil group .
fossil groups have a defining characteristic gap of two magnitudes between the brightest galaxy and the second brightest galaxy .
the temperature , metallicity , and luminosity of the x - ray halo surrounding ngc 1132 are similar to those of galaxy groups ( mulchaey & zabludoff 1999 ) .
analysis of the millenium simulation ( springel et al .
2005 ) show that fossil groups assembled their dark matter halos early , accumulating 80% of the present day mass 4 gyr ago ( dariush et al . 2007 ) .
the absence of parent galaxies and thus lack of substructure in fossil groups has been debated as a possible challenge to the @xmath5cdm paradigm , similar to the missing galaxy problem of the local group ( sales et al .
2007 and references therein ) .
fossil groups are dynamically evolved environments and due to their early assembly and their fast evolution the dominant galaxy has had time to accrete the most massive surrounding galaxies through dynamical friction ( mulchaey & zabludoff 1999 ) .
dwarf galaxies are exempt from the consequences of dynamical friction given that this effect is proportional to the mass of the satellite galaxy . also , given that ram pressure stripping is milder in a fossil group than in the core of a galaxy cluster , ucds formed by downsizing dwarf galaxies ( bekki et al .
2003 ) are expected to survive up to the present day in fossil groups .
this work uses the same method employed by madrid et al .
2010 ( hereafter paper i ) to determine the presence of ucds in the coma cluster .
the brightest ucd candidates reported in paper i have been spectroscopically confirmed by chiboucas et al .
( 2010 ) using the keck telescope .
the fainter end of the luminosity distribution of ucd candidates presented in paper i lacks spectroscopy confirmation simply due to the onerous challenge of obtaining their spectra at the coma distance .
collobert et al .
( 2006 ) give a redshift of @xmath6 for ngc 1132 , and distance modulus of ( m - m)=34.86 mag equivalent to a luminosity distance of 99.5 mpc using h@xmath7=71 km.s@xmath8mpc@xmath8 , @xmath9 , and @xmath10 ( wright 2006 ) .
the observations of ngc 1132 are carried out with the advanced camera for surveys wide field channel ( acs / wfc ) on 2006 august 22 .
the data is acquired in two bands : @xmath11 , similar to sloan @xmath0 , and @xmath12 , sloan @xmath1 ( mack et al .
four exposures are obtained in each band for a total exposure time of 4446 @xmath13 in @xmath11 and 6885 @xmath13 in @xmath12 .
the pixel scale of the acs / wfc is @xmath14/pixel ( 23pc / pixel ) and the physical scale of the field of view is @xmath15 kpc at the distance of ngc 1132 . the same physical scale is covered in the study of paper i. eight flatfielded science images ( flt.fits ) are retrieved from the public archive hosted at the space telescope science institute .
these files are pre - processed through the standard pipeline that corrects bias , dark current , and flatfield ( sirianni et al .
the pyraf task multidrizzle is used to combine the different exposures within the same filter , remove cosmic rays , and perform distortion correction .
detection of sources is carried out with sextractor ( bertin & arnouts 1996 ) using a detection threshold of 3@xmath16 above the background .
a careful round of visual inspection allows for the discarding of contaminants such as background galaxies .
as recently shown by the galaxy zoo project the human eye remains a powerful tool to discriminate between different types of galaxy morphology ( lintott et al .
2011 ) . as in paper
i the empirical point spread function ( psf ) is created by running the pyraf tasks pstselect , psf , and seepsf .
these tasks respectively select bright unsaturated stars from the image , build the psf by fitting an analytical model ( moffat function ) , and subsample the psf as required by the software used to determine the structural parameters .
structural parameters are determined using ishape ( larsen 1999 ) .
this specialized software convolves the psf with an analytical model of the surface brightness profile of slightly resolved stellar structures such as globular clusters and yields their effective radius , ellipticity , position angle , signal - to - noise ratio , and an estimate of the goodness of fit .
an important parameter to consider when studying low mass stellar systems in different environments is the effective radius or half - light radius ( @xmath17 ) . a king profile ( king 1962 , 1966 ) with a concentration parameter , or tidal to core radius , of c=30 is the most suitable analytical model for fitting the surface brightness profile of globular clusters and ucds ( see paper i ) .
the structural parameters of 1649 point - like sources are derived in both bands .
the values of structural parameters quoted hereafter are those corresponding to the @xmath12 band given its longer exposure time .
this filter is also a better tracer of the mass of globular clusters and ucds .
photometry is performed using an aperture of 4 pixel in radius with the task phot within the daophot package of pyraf , aperture correction is applied using the formulas of sirianni et al .
photometric measurements are obtained for all 1649 stellar objects identified as globular clusters and ucd candidates in the acs images of ngc 1132 .
foreground extinction for ngc 1132 is @xmath18 mag ( from ned ) , the specific extinction corrections applied to each hst filter are obtained following the prescriptions of sirianni et al .
( 2005 ) which give @xmath19 mag and @xmath20 mag .
up - to - date photometric zeropoints for acs / wfc are obtained from the stsci website : @xmath21 , @xmath22 .
these zeropoints yield magnitudes in the vega magnitude system .
ucds are selected based on their color , magnitude and structural parameters .
a metallicity break found by mieske et al .
( 2006 ) at the onset of the luminosity - size relation sets at @xmath23 the luminosity cutoff to distinguish between simply bright globular clusters and ucds . at the distance of ngc 1132 this luminosity requirement
translates to @xmath24 22.3 mag , this value is obtained using calcphot within the synphot package of stsdas ( laidler et al .
2005 ) . in color
, ucd candidates must be in the same range defined by the globular cluster system , this is shown in the next section to be @xmath25 in size , ucds are conventionally , but arbitrarily , defined as stellar systems having an effective radius between 10 and 100 parsecs ( see paper i and references therein ) .
ucds must be consistently resolved in both bands , at the distance of 100 mpc this condition is satisfied for sources with @xmath26pc with the deep images analyzed in this work . in photometric studies , and as a safeguard against contamination by background galaxies ,
the ellipticity of ucds is expected to be between @xmath27 and @xmath28 ( blakeslee & barber de graff 2008 , paper i ) .
all ucd candidates fall within that range of ellipticity .
additionally , a signal - to - noise ratio of @xmath29 is a requirement for obtaining a reliable estimate of structural parameters with ishape ( harris 2009 ) .
the color - magnitude diagram ( cmd ) of the globular cluster system associated with ngc 1132 is used to define the parameter space where ucds are expected to be present , a detailed analysis of this cmd is not the aim of this paper . for most globular clusters ( @xmath30 of them ) their colors are in the range @xmath31 and their magnitudes between @xmath32 as shown in figure 1 .
this broad color range translates into a metallicity range of @xmath33<-0.10 $ ] given by the color - metallicity transformation of peng et al .
the deduced metallicity range is in agreement with previous metallicity estimates of ucds ( mieske et al .
2004 , evstigneeva et al .
2007 ) .
the vast majority of 1649 globular clusters and ucd candidates identified in the acs images remain unresolved or their @xmath34 is too low to obtain a truly reliable estimate of their structural parameters .
however , 11 sources satisfy the color , magnitude , size , and @xmath34 criteria outlined above to be classified as ucd candidates .
the effective radii of these 11 ucd candidates range from 77.1 pc , for the largest and brightest candidate , to 8.5 pc .
the median effective radius for ucds is @xmath35 pc with a standard deviation of @xmath36 pc . excluding the brightest candidate brings these numbers to @xmath37 pc and @xmath38 pc .
all 11 ucd candidates have a @xmath39 and are plotted as red dots in the color - magnitude diagram of figure 1 . in color and magnitude most ucds
overlap with the brightest globular clusters , the very same is the case for ucds in the core of the coma cluster , i. e. they share the same parameter space of the brightest globular clusters associated with ngc 4874 ( paper i ) . the brightest ucd candidate is a m32-like object located , in projection , @xmath406.6 kpc away from the center of the galaxy .
its effective radius is @xmath3 pc , its magnitude @xmath41 , or two magnitudes brighter than the second brightest ucd candidate , and its is color particularly red @xmath42 . in the framework of a dual formation mechanism for ucds proposed by da rocha et al .
( 2011 ) and norris & kannappan ( 2011 ) this candidate is the remaining nucleus of a stripped companion of ngc 1132 due to the clear gap of two magnitudes between the brightest globular clusters and this ucd .
this object is similar to sdss j124155.3 + 114003.7 , the ucd reported by chilingarian & mamon ( 2008 ) at a distance of 9 kpc from m59 .
as stated by norris & kannappan ( 2011 ) no globular cluster system has a luminosity function continuously extending up to such high luminosity .
as shown in paper i , not even the extremely rich globular cluster system of ngc 4874 in the core of the coma cluster has members with magnitudes similar to the brightest ucd candidate .
this object was catalogued by the two micron all sky survey as 2mass-02525121 - 0116193 , its k band magnitude is @xmath43 mag ( skrutskie et al .
2006 ) . a size estimate for the m32 counterpart was also carried out with galfit ( peng et al .
2002 ) using a sersic model with an initial sersic index of @xmath44 ( sersic 1968 ) .
using galfit the best fit for this object has a reduced @xmath45=1.008 and yields @xmath46 , and @xmath47 pc .
the original image of the brightest ucd and the residual after model subtraction with both galfit and ishape is given in figure 2 .
the best fit of two different analytical models and different software leaves a small residual in the core .
similar residuals are found by price et al .
( 2009 ) studying several compact elliptical galaxies in the core of the coma cluster .
the brightest ucd candidate is comparable , but slightly smaller than the compact elliptical galaxies studied by price et al .
( 2009 ) which have effective radii of @xmath48 pc .
this object is also fainter than the price compact ellipticals which have magnitudes ranging from @xmath49 to @xmath50 mag .
photometric and structural parameters of the eleven ucd candidates are presented in table 1 .
an additional 39 sources are positively resolved in both bands and have effective radii between 8.2 pc and 59.7 pc , with a median @xmath51 pc .
while these sources have the size characteristic of ucds their magnitudes fall short of the minimum threshold for the selection criteria .
these extended stellar systems have received several denominations , here we refer to them as extended star clusters .
extended star clusters are plotted as blue squares on the cmd of figure 1 . even if these extended star clusters are not as bright as ucds , in luminosity some of these objects are comparable to @xmath52 centauri .
the position of globular clusters , extended star clusters , and ucds on the acs frame are plotted in the top panel of figure 3 .
all these low - mass stellar systems congregate towards the central elliptical arguing against background contamination , this is also shown in the bottom panel .
five ucds and seven extended star clusters are found within the inner 12 kpc ( in projection ) to the center of the galaxy . in paper
i no ucds were found within the inner 15 kpc of the center of ngc 4874 .
as noted in the introduction a milder tidal field and ram pressure stripping within a fossil group can allow ucds and extended star clusters to survive up to the present day at small galactocentric distances .
ccccc 77.1 & 0.09 & 1168 & 18.49 & 2.28 + 23.2 & 0.11 & 323 & 20.72 & 1.61 + 13.3 & 0.22 & 287 & 20.98 & 2.12 + 9.6 & 0.16 & 164 & 21.49 & 2.23 + 19.8 & 0.37 & 213 & 21.50 & 1.89 + 16.7 & 0.05 & 190 & 21.53 & 1.92 + 8.5 & 0.25 & 166 & 21.80 & 2.22 + 13.0 & 0.10 & 153 & 21.88 & 1.92 + 9.2 & 0.08 & 133 & 22.01 & 1.69 + 9.6 & 0.12 & 89 & 22.16 & 1.97 + 9.9 & 0.33 & 95 & 22.21 & 1.80 +
one of the defining characteristics of ucds is their magnitude - size relation .
ucds are the smallest stellar systems that show a correlation between luminosity ( or mass ) and size ( hasegan et al .
2005 ) , with masses of @xmath53 these objects are indeed at the onset of this relation . in figure 4
the magnitude - size relation for the 11 ucd candidates plotted as red cirlces is evident .
magnitude and @xmath17 for the 11 ucds in figure 4 have a spearman rank correlation coefficient of @xmath54 .
the correlation is not perfect ( i. e. @xmath55 ) due to the scatter characteristic of such plots .
no magnitude - size correlation is found for extended star clusters , with @xmath56 , these objects are plotted as blue squares in figure 4 .
the effective radius of globular clusters with masses below @xmath53 do not show any correlation with luminosity either , with a coefficient of @xmath57 derived by mclaughlin ( 2000 ) , that is , almost completely uncorrelated ( i. e. @xmath58 ) .
a second correlation for stellar systems with masses of @xmath59 and above is the mass - metallicity relation ( mmr , harris et al .
2006 , strader et al.2006 ) .
ucds also obey this relation .
the mmr is evident in two simple ways in the cmd of figure 1 . at the distance of ngc 1132 ( @xmath60 mpc )
the paradigmatic bimodality of gcs is blurred , however if we consider the measurements presented by da rocha et al .
( 2011 ) for gcs of compact groups at lower redshift the color of the metal - poor subpopulation of globular clusters is @xmath61 . in the cmd of figure 1
blue globular clusters do not extend to luminosities higher than @xmath62 . with only one exception , no clusters or extended objects
are found to have the colors of blue globular clusters and luminosities higher than @xmath62 .
the second brightest ucd , located 20.6 kpc away from the galaxy center , is exceptionally blue ( f475w - f850lp)=1.61 and stands out of the mmr .
a second manifestation of the mmr quantified in a simple way is the median color of ucds and extended star clusters .
self - enrichment models ( bailin & harris 2009 ) predict that more massive ( and bigger ) objects migrate in color towards the red .
ucds that are more massive than extended globular clusters have indeed redder colors on average .
the median color for ucds is @xmath63 with @xmath64 while the median color for extended globular clusters is @xmath65 .
the brightest ucd candidate is also the reddest with a color of @xmath42 , this is in agreement with the findings of norris & kannappan ( 2011 ) .
in contrast , the largest extended globular cluster with @xmath66 has a blue color with @xmath67 .
ucds extend the mmr to brighter magnitudes than the end of the globular cluster luminosity function .
in paper i and in this study two highly evolved environments with fundamentally different densities are probed : the core of the coma cluster , the richest galaxy cluster of the nearby universe and an isolated elliptical galaxy in a fossil group . both systems exhibit
similar characteristics of their ucd population with the exception that ucds are present at small galactocentric distance ( projected ) in the fossil group and not in the galaxy cluster .
the presence of ucds at the bright tip of the globular cluster systems of elliptical galaxies appears independent of environment .
given also the evidence given by the works cited in section 1 on the presence of ucds outside galaxy clusters it is natural to believe that ucds are commonly present in all environments .
the results of this work support the thesis put forward by norris & kannappan ( 2011 ) on the ubiquity of ucds in galaxies with populous globular cluster systems irrespective of environment .
only the superior resolution of the hubble space telescope allows for the determination of the structural parameters of ucds in the local universe .
evstigneeva et al . (
2007 ) searched for ucds in five galaxy groups to no avail using ground based data .
the fact that ucds escaped detection for almost a decade while hst was operational is not a coincidence .
hst can resolve ucds only at relatively low redshift ( d@xmath40 tens to hundreds of mpc ) however at these distances first and second generation instruments onboard hst ( e.g. wfpc2 ) have small field of views that only cover physical scales of @xmath68 kpc .
ucds are difficult to find within the innermost 10 kpc of the host galaxy ( paper i , bekki et al .
2003 , da rocha et al .
2011 ) not only due to the crowding produced by starlight but also due to the destructive effects of the tidal field of the host galaxy ( bekki et al .
an artificial maximum luminosity cutoff usually used in globular cluster studies has also certainly contributed to under - reporting of ucds in the past ( norris & kannappan 2011 ) .
many thanks to l. spitler , j. hurley , w. harris , l. schwartz , and d. croton for their input .
the referee gave a detailed and constructive report that helped to improve the manuscript .
this research has made use of the nasa / ads , the nasa / ipac extragalactic database , simbad , and google .
stsdas and pyraf are products of the stsci , which is operated by aura for nasa . | eleven ultra - compact dwarf ( ucd ) and 39 extended star cluster candidates are found to be associated with the galaxy ngc 1132 .
this giant elliptical galaxy is the remnant of a fossil group .
ucd and extended star cluster candidates are identified through the analysis of their structural parameters , colors , spatial distribution , and luminosity using deep hubble space telescope observations in two filters : the f475w ( sloan @xmath0 ) and f850lp ( sloan @xmath1 ) .
the median effective radius of these ucds is @xmath2 pc .
two types of ucds are identified in the vicinity of ngc 1132 , one type shares the same color and luminosity of the brightest globular clusters and traces the onset of the mass - size relation .
the second kind of ucd is represented by the brightest ucd candidate , a m32-type object , with an effective radius of @xmath3 pc , located at @xmath4 kpc from the nucleus of ngc 1132 .
this ucd candidate is likely the remaining nucleus of a minor merger with the host galaxy . with the exception of a particularly blue ucd candidate , ucds
are found to extend the mass - metallicity relation found in globular clusters to higher luminosities .
the results of this work support the growing body of evidence showing that ucds are not circumscribed to galaxy clusters as previously thought .
ucds are likely to be a common occurrence in all environments
. the milder tidal field of a fossil group , when compared to a galaxy cluster , allows ucds and extended star clusters to survive up to present time at small galactocentric distances . |
experimental studies of interacting quantum systems are increasingly focusing on non - equilibrium setups , e.g. , driving cold atom systems or electronic materials dynamically through various finite temperature and quantum phase transitions @xcite .
theoretical modeling of systems under these conditions is even more challenging than the already difficult problem of computing equilibrium properties of quantum systems away from perturbative regimes .
exact numerical calculations are possible for small systems , and there has been some success in studying issues such as thermalization and many - body localization @xcite .
reaching system sizes sufficiently large enough for modeling experiments is still difficult in most cases , with the exception of some one - dimensional ( 1d ) systems where the density - matrix renormalization group ( dmrg ) method ( or the closely related matrix - product states ) now allows for time - evolution studies on relatively large system sizes and long times @xcite .
for higher - dimensional systems the challenges in real - time calculations are formidable . given the difficulties with real - time evolution ,
alternative ways to extract non - equilibrium dynamical properties of quantum systems have been explored in the imaginary - time domain , where quantum monte carlo ( qmc ) methods can be applied . in equilibrium , there are solid relationships between real- and imaginary - time correlation functions which can be exploited in numerical analytic continuation of qmc data @xcite .
much less is known about practical ways to infer real - time properties from imaginary - time calculations out of equilibrium , though some progress has been made on this front recently .
examples include studies of systems driven through a quantum critical point @xcite at different velocities or according to non - linear protocols , where kibble - zurek scaling @xcite can be used to extract the dynamic exponent and other important quantities such as the quantum geometric tensor @xcite .
an important observation here is that real- and imaginary - time evolution are identical not only in the adiabatic limit , where both dynamics keep systems in their instantaneous ground states , but also including the leading non - adiabatic effects .
another example is the phenomenon of `` initial slip '' @xcite , where a random product state is evolved by a hamiltonian tuned to a quantum - critical point and the state initially becomes increasingly ordered , before developing critical fluctuations and vanishing long - range order .
the transient states produced before one reaches the critical equilibrium state have interesting properties that can be probed in imaginary - time @xcite .
imaginary - time evolution has also been used to investigate the emergence of topological conservation laws @xcite . in this paper
we use the imaginary - time approach to study the relaxation mechanism of the order parameter in quantum antiferromagnets with o(@xmath17 ) rotationally - invariant order parameters .
the system , described by a hamiltonian @xmath1 , is initially prepared in a fully saturated antiferromagnetic state with the order parameter along the @xmath3 spin axis . evolving this state in imaginary - time @xmath5 with the operator @xmath0 , the rotational symmetry
will eventually be restored marked by the expectation value of the @xmath3 component of the order parameter decaying to zero .
we identify short- and long - time behavior of the dynamics and develop theoretical and practical tools for analyzing emergent scaling behaviors by defining an effective dynamic exponent for a given threshold value of the @xmath3 component of the order parameter .
as we shall show in this paper , this effective dynamic exponent converges to the true dynamic exponent as the threshold of the order parameter is lowered to @xmath18 .
in addition to delivering the dynamic exponent , which is not very surprising ( though useful in its utility as a tool to extract the exponent ) , we show that the asymptotic long - time , large - system scaling behavior contains valuable information on the nature of the high - energy states . in a clean 2d heisenberg antiferromagnet
we observe fast convergence of the effective dynamic as the threshold order parameter is decreased , which we argue is indicative of the expected `` anderson tower '' of quantum rotor states @xcite . in other words ,
the scaling is characterized by an constant effective dynamic exponent @xmath19 ( @xmath20 being the dimensionality of the system ) for a large range of values of the order parameter .
in contrast , in a 2d system randomly diluted at its percolation point , we find that the effective dynamic exponent increases as the threshold order parameter goes to 0 , converging toward a fixed value only when the order parameter is small .
this demonstrates a hierarchy of excitations which forms tower governed by a common dynamic ( size - scaling ) exponent only at low energies .
the ultimate low - energy value of the dynamic exponent is @xmath21 , which improves in previous estimates @xmath22 obtained using different methods @xcite .
these results reinforce the notion that lowest excitations of the system at the percolation point are not conventional anderson quantum rotor states ( goldstone modes @xcite ) although the system breaks the @xmath23 spin symmetry spontaneously in the thermodynamic limit @xcite .
the outline of the rest of the paper is as follows : in sec .
[ sec : theory ] we will describe the theoretical underpinnings of our approach . in secs .
[ sec : cleanheisenberg ] and [ sec : percolating ] we discuss results for the pure 2d heisenberg model and the diluted system , respectively .
we summarize our study and provide some further remarks in sec .
[ summary ] .
as mentioned in the introduction our setup will be the following : we prepare our system initially in a fully saturated antiferromagnetic state denoted by @xmath24 , and evolve in imaginary time with @xmath25 where @xmath1 is a heisenberg hamiltonian describing a gapless quantum antiferromagnet . because the ground state obeys the symmetry of @xmath1 we know that this initial state will not be an eigenstate of @xmath1 but it will have some overlap with the ground state , as they share the same ordering .
as the state evolves in imaginary - time it will eventually decay to the ground state , restoring the rotational symmetry of our system .
from the theory of spontaneous symmetry breaking we know that in the limit of system size tending to infinity , a set of excited states just above the rotationally symmetric ground state become degenerate allowing the system to spontaneously align along a particular axis when subjected to an infinitesimal perturbation @xcite . in imaginary - time this phenomena will manifest itself as a divergence in the relaxation time to reach the ground state as the system size is increased .
we will use the staggered magnetization @xmath26 along the @xmath3 spin axis @xmath27 as a measure of the restoration of rotational symmetry as our state is evolving in imaginary - time .
the expectation value of this operator as a function of imaginary - time is given by : @xmath28 where @xmath29 is the imaginary - time evolved state : @xmath30 expanding in eigenstates of @xmath1 , denoting the eignestates and eigenenergies by @xmath31 and @xmath32 , respectively , and defining the gap , @xmath33 ; in the limit @xmath34 , the expectation value in eq .
( [ eq : exp ] ) will vanish as @xmath35 where @xmath36 .
since the ground state is symmetric under rotations we have that @xmath37 .
let us define the relaxation time @xmath38 as the time where @xmath39 drops below some threshold @xmath40 .
equation suggests that as this threshold goes to 0 , @xmath41 .
therefore in this limit , by calculating the scaling @xmath38 with system size , we can infer the scaling of the low energy gap of @xmath1 .
one can characterize this scaling of the low energy gap by the dynamic exponent @xmath42 which has different interpretations depending on its value .
@xmath43 implies that the system has a finite gap in the thermodynamic limit , while finite @xmath3 means the system has gapless excitations , and finally @xmath44 denotes exponential scaling of the gap .
if we consider systems which have gapless excitations , a finite dynamic exponent implies the relaxation time should scale as a power law : @xmath45 . in other words ,
if one rescales the time axis by @xmath9 , @xmath46 should show scaling collapse at small @xmath40 .
for some hamiltonians there exists a large , but sub - extensive , number of low energy states which have energies ( relative to the ground state ) that have the same scaling as the low energy gap @xcite .
the existence of these states will have the effect that there will be a larger window of @xmath40 for which this scaling collapse holds .
this is because higher order terms in eq .
( ) will have energy exponentials which all scale in a similar manner with system size .
this argument will be important later when we discuss the differences between the clean and diluted 2d heisenberg models in later sections .
the preceding arguments , however , become valid asymptotically in the limit @xmath47 and so it is necessary to take into account finite size corrections if one would like quantitative estimates for the dynamic exponent . as we will explain in the rest of this section , it is possible to control for the effects of finite size deviations by calculating the `` flow '' of the dynamic exponent from finite - size systems in a similar manner as the techniques used in finite - size scaling near critical points in equilibrium @xcite . using these methods one can extrapolating the finite - size results to the thermodynamic limit . to estimate the dynamic exponent from finite size systems we start by calculating the relaxation times @xmath38 and @xmath48 ( at some finite value of @xmath40 ) for two different system sizes @xmath6 and @xmath49 respectively . from this
, the dynamic exponent can be estimated by rescaling the two times by their respective system sizes such that the rescaled results are equal : @xmath50 from this expression we can define a finite size exponent for the pair of system sizes @xmath51 : @xmath52 as @xmath53 , this finite size exponent will converge to what we shall call the effective dynamic exponent ( recall that @xmath40 is finite ) which we will denote as @xmath54 .
the manner in which @xmath51 converges to infinite size is determined by corrections to the @xmath55 scaling of @xmath56 . to see this , one can parametrize the corrections to the low energy gap with correction exponents @xmath57 : @xmath58 now modifying eq .
( [ eq : dyn_z ] ) to include corrections we get : @xmath59 using this equation and eq .
( ) we find that @xmath51 converges to @xmath54 as : @xmath60 the equation above allows one to extract both @xmath57 s and @xmath61 s needed to obtain the scaling corrections in @xmath56 .
note that we have explicitly suppressed the fact that the effective dynamic exponent as well as the finite size corrections are functions of @xmath40 .
we will revisit this functional dependence in later sections , but it is important to recall our previous argument that in the limit @xmath62 , @xmath54 converges to the true dynamic exponent of the model . in the following we will test this scaling hypotheses on two examples .
first we study the heisenberg antiferromagnetic on the simple square lattice .
the low - energy physics of this model is very well understood and provides a good benchmark for our scaling approach @xcite .
we then go on to apply this method on the site - diluted heisenberg antiferromagnet on a square lattice at the percolation point , where there have been previous studies but not as clear of a consensus as to the low energy physics of the model @xcite . to compute the imaginary - time evolution
, we use a projector qmc method that in practice shares many similarities with the common stochastic series expansion ( sse ) method .
we sample contributions to @xmath63 and define @xmath64 where @xmath65 is the total number of spins on the lattice .
this time definition is equivalent , up to a factor , to the conventional imaginary - time appearing in the schrdinger evolution operator @xcite .
( a ) evolution of @xmath66 as a function of imaginary - time in the clean 2d heisenberg model for different system sizes .
( b ) the same data as in ( a ) but plotted vs @xmath67 ( @xmath68 ) , showing how the asymptotic relaxation is governed by the anderson tower of rotor states .
in both ( a ) and ( b ) the dotted line is the magnitude of the order parameter of the heisenberg model in the thermodynamic limit at @xmath69 .
this is the value the order parameter would relax to if the system size was infinite and would remained in the symmetry broken ground state.,width=283 ] ) space .
the panels ( i)-(iii ) correspond to the similarly marked scaling regimes in fig .
[ fig : rotor](a ) .
the radius of the white circle corresponds to the magnitude of the order parameter in the ground state , and the arrow marks the average order parameter @xmath70 . in ( i )
the system is close to the initial state , where the magnitude of the order parameter is larger than in the ground state . in ( ii )
the magnitude has decayed to its asymptotic value but the direction of the order parameter is still close to the initial one . in ( iii )
the system is close to the asymptotic finite - size state with uniformly fluctuating ( not symmetry - broken ) order parameter.,scaledwidth=48.0% ]
for the first example we consider the standard two - dimensional ( 2d ) antiferromagnetic heisenberg model , defined by the hamiltonian @xmath71 where @xmath72 is a spin-@xmath10 operator at lattice site @xmath73 and the sum is over nearest neighbors on an @xmath6 by @xmath6 square lattice in @xmath74 with periodic boundary conditions .
we set @xmath75 for the rest of the paper .
this model is a part of a broad class of antiferromagnet models on translationally - invariant bipartite lattices which are known to have quantum - rotor like low - energy excitations @xcite .
these excited states have energy levels which become degenerate with the ground state in the limit @xmath47 as a power law : @xmath76 .
here @xmath19 is the dimension of the lattice and @xmath6 is the linear dimension of the system .
one can think of these quantum rotor states as the quantization of the global angular fluctuations of the order parameter which in the thermodynamic limit form a basis in the ground state manifold used to create the symmetry broken ground state which is the vacuum of the gapless spinwave excitations @xcite .
[ fig : rotor](a ) shows @xmath77 versus @xmath5 for various system sizes . in this figure
one clearly sees that initially all the system sizes relax at the same rate but then eventually break off from one another when the sublattice magnetization reaches @xmath78 , with the larger system sizes taking longer to relax to @xmath79 .
one may recall that in the ground state of this model the thermodynamic value of @xmath80 @xcite which roughly corresponds to the value where the different system sizes begin to relax at different rates ( see the dotted line in fig [ fig : rotor ] ) .
one can make use of the continuum field theory description of the long wavelength behavior of this model to understand the evolution of the order parameter in imaginary - time .
the lowest energy states of the continuum field theory are uniform in space .
since the initial state is also uniform , and the system is translationally invariant , the evolution will occur within the subspace of uniform configurations .
the dynamics in this subspace simplify to that of a quantum particle relaxing in the `` mexican hat '' potential . here
the probability distribution of the particle in space represents the probability distribution of order parameter of the system .
since the initial order parameter value is @xmath81 , the initial probability distribution of the order parameter is localized away from the minimum of the potential which for the heisenberg model on a square lattice is close to @xmath82 , as illustrated schematically in fig .
[ fig : density](@xmath73 ) .
as the system evolves in imaginary - time the energy of the system decreases and order parameter decays until it reaches the bottom of the potential which in the thermodynamic limit corresponds to the the symmetry broken `` ground state '' , as in fig .
[ fig : density](@xmath83 ) .
however , because the system is finite , this is not a true ground state so the energy of the system continues to decay and the mean magnetization along the z - axis relaxes to @xmath18 as the probability distribution spreads out over all possible solid angles ; fig .
[ fig : density](@xmath84 ) .
this second part of the relaxation is governed by the rotor state as they are the states which make up the quantization of the angular part of the order parameter ( and sets the effective moment of inertia which scales as @xmath85 , @xcite ) .
we can confirm this intuition for the lattice model by observing that in fig .
[ fig : rotor](b ) , by rescaling the @xmath5 axis with @xmath86 one finds that the second section of the relaxation shows scaling collapse .
a consequence of this is that in the limit @xmath47 the order parameter never relaxes to @xmath18 implying that the system remains in the symmetry broken ground state .
next let us discuss how to numerically extract the dynamic exponent from finite size data .
first one must numerically determine the intersection points of @xmath46 with the threshold value @xmath87 required to calculate the finite size exponent @xmath51 between two system sizes .
here we choose @xmath6 and @xmath88 , defining @xmath89 . to extract @xmath38 we fill a window around the threshold with qmc data and then use a polynomial ( or some other appropriate function ) to interpolate the data and
numerically find the crossing point of the interpolation and @xmath40 .
next we must derive the finite size corrections to @xmath90 which is dependent on the parameterization the finite size corrections in @xmath56 .
for this model it is natural to assume that @xmath56 should be an analytic function in @xmath91 because the model has no critical fluctuations . keeping terms up to order @xmath92 in @xmath56 we find the finite size corrections to @xmath90 are @xmath93 the extrapolated dynamic exponents are shown as a function @xmath87 in fig .
[ fig : clean_exp ] ( note that the values in fig .
[ fig : clean_exp ] are correlated because @xmath40 values can be arbitrarily close to one another ) .
the extrapolated values of @xmath54 for low @xmath40 are in excellent agreement with the analytic result of @xmath94 . for higher values for @xmath40
the disagreement is natural because at short times the many - body wavefunction still has an overlap with high - energy states .
as the system evolves in imaginary - time ( as @xmath40 decreases ) , the overlap with these higher energy states decay and so their effects on the effective dynamic exponent vanish .
the parameters @xmath61 s coming from the exponent flow give us the same coefficients which parameterize the finite size corrections of @xmath56 .
figure [ fig : clean_col](a ) focuses on the final relaxation time scale and more clearly shows the finite size corrections to @xmath95 . by including the finite size corrections calculated from the extrapolation , we find much better scaling collapse ; see fig .
[ fig : clean_col](b )
. extrapolated values of the asymptotic dynamic exponent @xmath54 of the clean 2d heisenberg antiferromagnet as a function of the threshold value @xmath40 used to perform the @xmath5-axis rescaling.,scaledwidth=45.0% ] scaling collapse of @xmath46 in the heisenberg model on a square lattice without ( a ) and with ( b ) finite - size corrections to the leading scaling form @xmath55 .
the dotted black lines denote the threshold value along which the curves are collapsed ; @xmath96 . in ( a ) , in the leading power - law @xmath55
the exponent is @xmath97 . in ( b ) the finite - size correction to @xmath56
are calculated from the fit to the size dependence of the exponent shown in the inset.,scaledwidth=45.0% ]
the second example has the same heisenberg interactions with nearest neighbors as the previous section , but the boundaries are no longer periodic and spins on the sites are randomly removed from the lattice with probability @xmath98 .
this dilution is particularly interesting when it is set to the percolation point .
it has been shown that at this dilution , the characteristics of the low energy states are different from that of the standard quantum rotor picture @xcite .
the volume of these clusters scale as @xmath99 where @xmath15 is the fractal dimension and @xmath6 is a linear size of the cluster @xcite .
it has been conjectured that the low energy states of this model are dominated by a generalization of the `` dangling spin '' concept local sub - lattice imbalance in a region of a given cluster @xcite where en effective moment forms due to the inability of spins to pair up in a bipartite manner . on a bethe lattice geometry the same phenomena
was studied using the dmrg method @xcite . in this study
it was shown explicitly that there exists a set of low lying quasi - degenerate ( qd ) eigenstates which remain separated from the higher energy eigenstates by a finite size gap @xmath100 , which goes to 0 slower then the spacing between the qd states .
it was conjectured that these qd states decouple from the bulk and because they are made up of power - law localized magnetic moments which interact with each other across the cluster , as had been previously deduced based on scaling behaviors of quantities probing the low - energy excitations indirectly @xcite .
for this reason , diluted heisenberg models have a larger dynamic exponent than predicted by the quantum rotor picture @xcite . beyond the interesting physics of this model ,
the disorder should prove a more robust test of our scaling hypothesis and method for extracting the dynamic exponent .
the major difference between this model and the last is the type of finite size corrections we see . in the unadulterated 2d case ,
the model is very far away from any sort of critical behavior and so the corrections are analytic in @xmath101 but here we can not assume this as there are fluctuations driven by the classical percolation threshold @xcite .
however , since we are only interested in extracting the dynamic exponent its perfectly reasonable to parameterize the finite size corrections to be analytic in @xmath102 , where @xmath65 is the number of sites making up a cluster .
finite - size scaling analysis of @xmath46 in the heisenberg model on a percolating cluster ( averaged over dilution realizations ) .
the black dotted line denotes the threshold value along which the curves are collapsed ; @xmath103 .
( a ) shows the data collapse with the leading behavior @xmath104 used for rescaling the @xmath105-axis , with @xmath106 .
( b ) shows the scaling collapse including finite - size correction to @xmath107 corresponding to the fit of the finite - size flow shown in the inset.,scaledwidth=45.0% ] to perform the dilution averaging we employ a procedure similar to what is outlined in ref .
each cluster is constructed with a fixed number of sites @xmath108 where we round up to an even integer to insure that the ground state can have total @xmath109 .
the finite - size scaling of the low energy gap is implemented in the following form : @xmath110 the finite - size flow of the effective dynamic exponents are calculated between system sizes of length @xmath6 and @xmath111 , which implies @xmath112 , for the cluster sizes .
keeping terms up to order @xmath113 in finite size corrections to @xmath107 we obtain the following expansion for the finite - size effective dynamic exponent @xmath114 : @xmath115 results of a data - collapse analysis both with and without the scaling corrections are shown in fig .
[ fig : cluster_col ] . for our lowest value of @xmath103 , @xmath116 or @xmath117 , slightly larger than what was found previous studies@xcite .
although we may have introduced some systematic error by assuming the finite size corrections decay as @xmath118 , the inset in fig .
[ fig : cluster_col](b ) shows that for the largest system size ( @xmath119 ) , @xmath120 . because our system sizes are comparable to previous studies we know that this is not an issue of finite size effects .
we also know that in imaginary - time evolution , the weights coming from an eigenstate @xmath31 relaxes on a timescale of @xmath121 , meaning that higher energy states always decay faster than low - energy states and therefore the effective dynamic exponents must be monotonically increasing as @xmath62 .
this is consistent with results in the previous section and with the threshold dependence of the extrapolated dynamic exponent shown in fig .
[ fig : cluster_exp ] . extrapolated values of @xmath54 averaged over dilution realization for the site - diluted heisenberg antiferromagnet as a function of the threshold value @xmath40 used to perform the @xmath5-axis rescaling.,scaledwidth=45.0% ]
in summary we have shown that the relaxation of the order parameter in imaginary - time can be used to quantitatively extract low - energy properties of a given model .
in particular , we have developed a scaling method which allows one to extract the dynamic exponent by performing scaling collapse of the order parameter along the imaginary - time axis .
the imaginary - time evolution of the order parameter in a 2d heisenberg antiferromagnets was used as a test - case to show that one can extract the correct dynamic exponent , which in this case is related to the anderson tower of quantum rotor states .
we also studied the relaxation of the order parameter of the corresponding site dilute heisenberg antiferromagnet at its percolation threshold and found that the scaling theory gives a dynamic exponent even larger than seen in previous studies ; @xmath122 versus @xmath16 from refs .
these results conform with the notion that there is a set of low - energy states due to quasi - localized moments``dangling spins '' and their generalizations to larger regions of local sublattice imbalance @xcite .
as there are very clear numerical results that the ground state of the heisenberg model on the percolating cluster does indeed have long range order @xcite , there is still an open question if there is a quantum - rotor tower of states in these clusters , which presumably should be required for spontaneous breaking of the spin - rotation symmetry in the thermodynamic limit @xcite .
if they do exist they would have relaxation times that would scale as @xmath123 according to the quantum rotor picture , however our method would not be able to isolate these states directly as it can only extract an effective dynamic exponent originating from the mixing of low and high energy states at short times .
the fact that we see a dynamic exponent with significant flow with the threshold value of the mean order parameter in the system much larger than in the unadulterated system may in principle be an indication of the rotor states .
however , the density of the rotor states should be much lower than the low - energy states arising from the quasi - localized moments .
the drift in the exponent may therefore instead be related to a slowly changing dynamic exponent of these quasi - localized states as one goes to higher energies in their tower .
we believe that this is the more likely scenario , with rotor states existing in the bulk of the cluster but not contributing significantly to the low - energy dynamics because of their much higher dynamic exponent and much lower density of states . in this work
we have only studied the spatially averaged order parameter but in the case of the percolating clusters it could be useful to look at the local order as a function of imaginary - time in a similar spirit as refs .
. one would expect that if the localized spins are indeed decoupled from the bulk , that this would manifest itself as very different relaxation times between the localized moments and the bulk .
one would even hope to see that the bulk relaxes to @xmath18 on time scales @xmath124 while the decoupled moments would relax on time scales of @xmath125 .
the qmc methods used here to simulate imaginary - time can be generalized to higher spin representations which , along with other methods , would give a more complete picture as to the low energy excitations of the higher spin versions of the heisenberg antiferromagnets on percolating clusters .
we would like to thank h. changlani , a. iaizzi , p. patil , and a. polkovnikov for useful discussions .
the computational was performed on the shared computing cluster administered by boston university s research computing services .
this research was supported by the nsf under grant no . | we apply imaginary - time evolution with the operator @xmath0 to study relaxation dynamics of gapless quantum antiferromagnets described by the spin - rotation invariant heisenberg hamiltonian ( @xmath1 ) . using quantum monte carlo simulations to obtain unbiased results , we propagate an initial state with maximal order parameter @xmath2 ( the staggered magnetization ) in the @xmath3 spin direction and monitor the expectation value @xmath4 as a function of imaginary - time @xmath5 . results for different system sizes ( lengths ) @xmath6 exhibit an initial essentially size - independent relaxation of @xmath7 toward its value in the infinite - size spontaneously symmetry - broken state , followed by a strongly size - dependent final decay to zero when the o@xmath8 rotational symmetry of the order paraneter is restored .
we develop a generic finite - size scaling theory that shows the relaxation time diverges asymptotically as @xmath9 where @xmath3 is the dynamic exponent of the low energy excitations .
we use the scaling theory to develop a practical way of extracting the dynamic exponent from the numerical finite - size data , systematcally eliminating scaling corrections .
we apply the method to spin-@xmath10 heisenberg antiferromagnets on two different lattice geometries : the standard two - dimensional ( 2d ) square lattice as well as a site - diluted 2d square lattice at the percolation threshold . in the 2d case
we obtain @xmath11 , which is consistent with the known value @xmath12 , while for the site - dilutes lattice we find @xmath13 or @xmath14 , where @xmath15 is the fractal dimensionality of the percolating system .
this is an improvement on previous estimates of @xmath16 .
the scaling results also show a fundamental difference between the two cases ; for the 2d square lattice , the data can be collapsed onto a common scaling function even when @xmath7 is relatively large , reflecting the anderson tower of quantum rotor states with a common dynamic exponent @xmath12 . for the diluted 2d square lattice ,
the scaling works well only for small @xmath7 , indicating a mixture of different relaxation time scaling between the low energy states .
nevertheless , the low - energy dynamic here also corresponds to a tower of excitations . |
t tauri was initially identified as a single optical star , with unusual variability and peculiar emission lines ( barnard 1895 , joy 1945 and references therein ) .
early infrared observations then revealed the existence of a heavily obscured companion ( hereafter t tau s ) located about @xmath4 to the south of the visible star ( dyck , simon & zuckerman 1982 ) , and most likely gravitationally bound to it ( ghez et al .
recently , this infrared companion was itself found to contain two sources ( t tau sa and t tau sb koresko 2000 ; khler , kasper & herbst 2000 ) in rapid relative motion ( duchne , ghez & mccabe 2002 ; furlan et al .
thus , t tau is now acknowledged to be at least a triple stellar system . at radio wavelengths ,
t tau has long been known to be a double source ( schwartz , simon & campbell 1986 ) .
the northern radio component is associated with the optical star , and mostly traces the base of its thermal jet ( e.g. johnston et al .
2003 ) , whereas the southern radio source is related to the infrared companion , and is thought to be the superposition of a compact component of magnetic origin and an extended halo , presumably related to stellar winds ( johnston et al .
2003 , loinard et al .
2003 ) .
the relative motions between the various components of the t tau system have recently been under enhanced scrutiny , following the suggestion , based on multi - epoch vla observations , that one of the components might have seen its path dramatically altered by a recent chaotic encounter ( loinard et al .
this interpretation was disputed by johnston et al .
( 2003 , 2004b ) and tamazian ( 2004 ) , who fitted the same vla data with stable orbits .
it is noteworthy , however , that the residuals between their best fits and the actual radio positions ( @xmath5 @xmath6 ) are often significantly larger than the nominal observational errors ( @xmath7 @xmath8 ) , especially at recent epochs ( see tab .
6 in johnston et al .
2004b ) . to resolve this discrepancy , johnston et al .
( 2004a , 2004b ) proposed that the structure of t tau s at radio wavelengths was affected by erratic internal variations , which made the centroid of the vla source dither around the position of the underlying pms star . to reconcile the radio observations obtained in the last few years with the orbital fits , the vla source centroid needs to have moved about 25 mas yr@xmath0 faster than the associated pms star .
clearly , this would render the existing 20 years of vla observations useless as tracers of the stellar trajectories , in spite of the high quality astrometry naturally provided by radio interferometry .
cccccc + ut date & epoch & synthesized beam & r.m.s . & flux density & peak flux + & ( yr ) & @xmath9 @xmath10 @xmath11 ; p.a . &
( @xmath12jy beam@xmath0 ) & ( mjy ) & ( mjy beam@xmath0 ) + & & ( mas @xmath10 mas ; degrees ) + + + 24.09.2003 & 2003.7300 & 1.88 @xmath10 0.78 ; @xmath138.99 & 70.5 & 1.69 & 1.43 + 18.11.2003 & 2003.8804 & 1.87 @xmath10 0.79 ; @xmath135.19 & 64.3 & 1.70 & 1.17 + 15.01.2004 & 2004.0387 & 2.17 @xmath10 1.00 ; @xmath1411.94 & 68.0 & 1.03 & 0.84 + 26.03.2004 & 2004.2349 & 1.90 @xmath10 0.80 ; @xmath132.24 & 65.5 & 1.51 & 0.76 + 13.05.2004 & 2004.3657 & 1.92 @xmath10 0.76 ; @xmath1310.66 & 76.3 & 1.79 & 1.18 + 08.07.2004 & 2004.5183 & 1.90 @xmath10 0.78 ; @xmath136.00 & 62.9 & 1.13 & 0.54 + 16.09.2004 & 2004.7090 & 2.06 @xmath10 0.80 ; @xmath138.94 & 68.7 & 1.70 & 0.98 + + the non - thermal mechanisms at the origin of the compact radio emission in t tau s require the presence of an underlying , magnetically active star ( skinner 1993 ) .
specifically , the emission is expected to be either gyrosynchrotron radiation associated with reconnection flares in the stellar magnetosphere and at the star - disk interface ; or coherently amplified cyclotron emission from magnetized accretion funnels connecting the disk to the star ( dulk 1985 , feigelson & montmerle 1999 , smith et al .
2003 ) . in all cases ,
the emission is produced within less than about 10 stellar radii ( roughly 30 r@xmath15 ) of the star itself .
indeed , 3.6 cm vlbi observations recently revealed the existence , near the expected position of t tau sb , of a source with an angular size less than about 15 r@xmath15 ( smith et al .
because it is so small , any structural changes in this compact radio component would occur on such small scales that the effects on the astrometry would be very limited .
thus , observations focusing on it should accurately trace the path of the underlying pms star .
here , we present the results of a series of seven continuum 3.6 cm ( 8.42 ghz ) observations of t tau s obtained every two months between september 2003 and september 2004 with the 10-element very long baseline array ( vlba ) of the national radio astronomy observatory ( nrao ; tab . 1 ) .
since the vlba is only sensitive to compact emission structures with high surface brightness , it will effectively filter out the extended radio halo of t tau s , and provide images of the non - thermal source alone .
our phase center was at @xmath16 = @xmath17 , @xmath18 = + @xmath19 , the position of the compact source detected by smith et al .
each observation consisted of a series of cycles with two minutes spent on source , and one minute spent on the phase - referencing quasar j0428 + 1732 ( @xmath16 = @xmath20 , @xmath18 = + @xmath21 ) , located @xmath22 away .
the secondary quasar j0431 + 1731 ( @xmath16 = @xmath23 , @xmath18 = + @xmath24 ) was also observed periodically ( about every 30 minutes ) ; it was not used in the phase calibration process , but served as a check on the final astrometry . based on the dispersion on the measured position of that quasar
, we estimate our astrometric uncertainties to be about 0.25 mas in both @xmath25 and @xmath26 .
data editing , amplitude calibration , and fringe fitting ( carried out only on the calibrators , since the target is much too weak ) were made in a standard way using nrao s aips software .
once calibrated , the visibilities were imaged with a pixel size of 0.25 mas after weights intermediate between natural and uniform were applied .
the compact radio source associated with t tau s was detected at all seven epochs with high signal - to - noise ratio ( tab . 1 ) .
its flux density varies significantly , from a maximum of nearly 1.8 mjy , down to a minimum of just over 1.0 mjy .
moreover , its peak brightness in mjy beam@xmath0 is found to be always significantly smaller than its flux density in mjy ( on average by 35 % tab . 1 ) .
this suggests that the emission is somewhat extended since equal numerical values are expected for a point source .
this possibility gains further credibility when one considers the restored images ( fig .
1a - c ) , where the emission is consistently found to be composed of a compact source ( most certainly associated with a stellar magnetosphere ) and a low brightness `` spur '' extending eastwards .
although this latter component was not reported by smith et al .
( 2003 ) , inspection of their published image ( their fig .
1 ) reveals a source structure quite similar to that found here , with a weak spur extending towards the north - east of a compact component .
when our seven observations are combined , the low brightness spur becomes even more evident ( fig .
1d ) ; its position angle is between + 70@xmath27 and + 80@xmath27 , and its deconvolved extent in that direction is about 1 mas . since the measured parallactic and proper motions ( sect .
3.2 ) imply a maximum displacement of about 0.02 mas during any one of our 6-hour observing runs , the measured size is not affected by smearing effects .
the observed spur is also very unlikely to be a consequence of random errors in the phase calibration , since its characteristics repeat themselves from one epoch to the next .
t tau s contains two infrared sources , so one might be tempted to associate the eastern spur to this preferred direction of the system .
however , extrapolating from the latest published infrared observations ( furlan et al . 2003 ; duchne et al . , in prep . )
, we estimate that the position angle between t tau sb and t tau sa at the median epoch of our vlba observations must have been about 110@xmath27 , significantly larger than the position angle of the spur .
another unlikely possibility is that it would trace a weak secondary component . given the short distance which would then separate the two vlba sources , orbital motions should have been easily detected over the course of our observations .
interestingly , the spur is nearly exactly perpendicular to the jet acknowledged to be powered by t tau s ( solf & bhm 1999 ) .
this might suggest that the low brightness extension is associated with an accretion disk , and results from reconnection flares at the star - disk interface . indeed , duchne et al .
( 2002 ) reported the detection of a wavelength - dependent infrared excess around t tau sb , which strongly suggests the existence of an accretion disk there .
typically , the inner edge of that disk is expected to be at a radius of about @xmath28 , where @xmath29 is the stellar radius ( ostriker & shu 1995 ) . a low - mass pms star such as t tau sb is expected to have a radius of about 3 r@xmath15 ( siess , dufour & forestini 2000 ) .
hence , the inner disk edge is expected to be at 15 r@xmath15 ( 0.07 au ) . at a distance of 141.5 pc ( see sect .
3.2 ) , this corresponds to an angular scale of about 0.5 mas .
after convolution with our beam size in @xmath25 ( on average 0.82 mas ) , we obtain an expected observed extent of 0.9 mas for the spur .
this is in very good agreement with the measured extent . in this scheme ,
it is , however , somewhat puzzling that we only see structure on one side of the star .
as figs .
1a to 1c readily show , the absolute position of the source changes significantly from one epoch to the next .
these displacements most certainly result from parallactic and proper motions , and can be modeled in terms of the source position at epoch j2000.0 ( @xmath16 and @xmath18 ) , its proper motion in right ascension and declination ( @xmath30 and @xmath31 ) , and its parallax @xmath1 .
these five astrometric elements were deduced from the measured positions of the source by minimizing the @xmath32 associated with this description using an iterative scheme ( fig . 2 ) .
the results for the five astrometric parameters are : [ cols= " > , < " , ] the excellent agreement between the infrared and the radio proper motions confirms that the magnetically active pms star generating the compact radio emission is t tau sb , and that the radio data ( from both vla and vlba ) can be used to trace its trajectory .
the vlba data presented here have allowed a detailed study of the structure and astrometry of the compact radio source associated with t tau s. the emission is found to be composed of a compact core most certainly originating from a stellar magnetosphere , and a low - brightness extension which may result from reconnection flares at the star - disk interface .
the accuracy of the absolute astrometry offered by the vlba has made it possible for us to demonstrate that the infrared , the vla , and the vlba observations all trace the same underlying star ( t tau sb ) .
this implies , in particular , that any orbital fit to the t tau system must be able to reproduce the positions measured by the vla in the last 20 years .
finally , the vlba data have allowed us to measure the distance to t tau with unprecedented accuracy : @xmath33 = 141.5@xmath3 pc .
this is in good agreement with the traditionally accepted value for taurus ( @xmath33 = 140 @xmath2 10 pc kenyon , dobrzycka & hartmann 1994 ) , and seemingly rules out the possibility of large distance gradients across taurus .
we are indebted to walter brisken and gaspard duchne who provided vlba pulsar data with which to checked our astrometry results , and infrared positions ahead of publication , respectively .
, l.f.r . , and
r.a.g acknowledge the financial support of dgapa , unam and conacyt , mxico ; a.j.m .
acknowledges support from the nrao .
the national radio astronomy observatory is a facility of the national science foundation operated under cooperative agreement by associated universities , inc . | in this _ letter _ , we present a series of seven observations of the compact , non - thermal radio source associated with t tauri south made with the very long baseline array over the course of one year . the emission is found to be composed of a compact structure most certainly originating from the magnetosphere of an underlying pre - main sequence star , and a low - brightness extension which may result from reconnection flares at the star - disk interface .
the accuracy of the absolute astrometry offered by the vlba allows very precise determinations of the trigonometric parallax and proper motion of t tau south .
the proper motion derived from our vlba observations agrees with that measured with the vla over a similar period to better than 2 mas yr@xmath0 , and is fully consistent with the infrared proper motion of t tau sb , the pre - main sequence m star with which the radio source has traditionally been associated .
the parallax , @xmath1 = 7.07 @xmath2 0.14 mas , corresponds to a distance of 141.5@xmath3 pc . |
the wealth of data from solar @xcite and atmospheric neutrinos @xcite has put neutrino physics in the spotlight and physicists are asking what is it that makes solar and atmospheric neutrinos convert .
the most popular possibility is that of neutrino oscillations , expected in most theories of neutrino mass @xcite .
indeed an excellent joint description of solar and atmospheric data is obtained in this case @xcite . however , more often than not , models of neutrino masses are accompanied by non - standard interactions ( nsi ) of neutrinos @xcite .
these can have non - universal ( nu ) or flavour - changing ( fc ) components , which typically co - exist .
the simplest nsi does not involve additional interactions beyond those mediated by the standard model electroweak gauge bosons : it is simply the nature of the leptonic charged and neutral current interactions which is non - standard because of the complexity of neutrino mixing @xcite . on the other hand nsi
can also be mediated by the exchange of new particles with mass at the weak scale such as in some supersymmetric models with r - parity - violating @xcite interactions .
such non - standard flavour - violating physics can arise even in the absence of neutrino mass @xcite .
such varieties of non - standard interactions of neutrinos affect their propagation in matter @xcite .
the magnitude of the effect depends on the interplay between conventional mass - induced neutrino oscillation features in matter and those genuinely engendered by the nsi , which do not require neutrino mass @xcite .
these may have a variety of phenomenological implications @xcite and have been considered in the context of atmospheric @xcite , as well as astrophysical neutrino sources @xcite .
the impact of non - standard interactions of neutrinos has also been considered from the point of view of future experiments involving solar neutrinos @xcite as well as the upcoming neutrino factories @xcite . in this paper
we re - consider the case of nsi solutions to the solar neutrino problem @xcite taking into account the recent charged current measurement at the sudbury neutrino observatory ( sno ) @xcite , the 1258day super - kamiokande ( sk ) data and the previous solar neutrino data @xcite . in contrast to previous work we consider a three - neutrino nsi analysis and study also the dependence of the solar neutrino nsi solution on the neutrino mixing angle @xmath0 whose value is presently constrained mainly by the reactor neutrino data @xcite .
due to its energy - independent nature , the nsi solution is in excellent agreement with the flat spectral energy distribution observed in the super - kamiokande experiment .
thus we focus first on the determination of the allowed solutions by considering only the total rates of the solar neutrino experiments .
we find that these solutions provide excellent descriptions of the solar rates , including the recent sno charged current ( cc ) result .
then we study the impact of the day - night data from super - kamiokande measurements and show that the nsi solutions are consistent with the non - observation of day - night variations . finally , and more important , using simplified _
anstze _ for the nsi interactions we analyse the impact of the nsi description of solar data on the atmospheric data , showing how they are fully acceptable also for the latter .
this paper is organized as follows . in section [ sec : neutr - evol - pres ] we discuss the neutrino evolution and conversion probabilities in the presence of nsi , in section [ sec : analysis - method ] we summarize the calculational and fit procedures we adopt . in section [ sec : how - about - atmosph ]
we discuss the impact of the nsi solar solution on the atmospheric data analysis , and in section [ sec : conclusions ] we summarize our results .
the most general form of three - neutrino evolution hamiltonian in the flavor base @xmath2 , in the presence of nsi can be given as @xmath3 where e is the energy , @xmath4 fermi s constant , @xmath5 and @xmath6 is the number density of electrons along the neutrino trajectory . in our calculations of the @xmath7 survival probability we use the electron and neutron number densities in the sun from the ssm
@xcite , while for the earth matter effects we use the density profile given in the preliminary reference earth model ( prem ) @xcite . in eq .
@xmath8 parametrize the deviation from standard neutrino interactions .
for example @xmath9 is the forward scattering amplitude of the fc process @xmath10 , while @xmath11 represents the difference between the non - standard component of the @xmath12 and the @xmath13 elastic forward scattering amplitudes .
the quantity @xmath14 is the number density of the fermion @xmath15 along the path @xmath16 of the neutrinos propagating in the sun or the earth .
we consider two cases , depending on the nsi model , in which the neutrino interaction occurs with down type or up type quarks . for the case of non - standard interactions on electrons @xcite one would have also to take into account the effect of the nsi also in the neutrino detection cross section .
note that for simplicity we have set the mass splitting @xmath17 of the first two neutrinos to zero , so that the corresponding mixing angle @xmath18 can be eliminated .
this is justified as we are mainly interested in isolating the effect of the nsi in the description of solar neutrino data .
in contrast we assume the most general nsi flavour structure , consistent with cp conservation . under these assumptions
, we have @xcite : @xmath19 where we used the notation @xmath20 and @xmath21 .
similarly to the usual oscillation case @xcite the @xmath7 survival probability @xmath22 can be written as @xmath23 where @xmath24 is the electron survival probability in an effective @xmath25 model described by the hamiltonian @xmath26_{2\times 2 } = \sqrt{2 } g_f n_e(\vec{r } ) \begin{pmatrix } c_{13}^2 & 0 \\ 0 & 0 \end{pmatrix } + \sqrt{2 } g_f n_f(\vec{r } ) \begin{pmatrix } 0 & { \varepsilon}_{\mathrm{eff}}\\ { \varepsilon}_{\mathrm{eff } } & { \varepsilon}'_{\mathrm{eff}}\end{pmatrix}\ ] ] with the effective @xmath27 and @xmath28 given in terms of the original @xmath8 parameters as @xmath29 \ , , \\ \begin{split } \label{eq:5 } { \varepsilon}'_{\mathrm{eff}}&= { \varepsilon}_{22 } c_{23}^2 - 2 { \varepsilon}_{23 } c_{23 } s_{23 } + { \varepsilon}_{33 } s_{23}^2 + 2 s_{13 } c_{13 } ( { \varepsilon}_{13 } c_{23 } + { \varepsilon}_{12 } s_{23 } ) \\ & \quad - s_{13}^2 ( { \varepsilon}_{33 } c_{23}^2 + { \varepsilon}_{22 } s_{23}^2 + 2 { \varepsilon}_{23 } s_{23 } c_{23 } ) \ , .
\end{split}\end{aligned}\ ] ] note that even though we have assumed the most general nsi flavour structure , the final propagation of solar neutrinos can be described effectively as a two - dimensional evolution hamiltonian depending only on the mixing angle @xmath0 and on two effective nsi parameters @xmath30 .
in our description of the solar neutrino data @xcite we adopt the same analysis techniques already described in refs .
@xcite using the latest theoretical standard solar model ( ssm ) best fit fluxes and estimated uncertainties @xcite . for the neutrino conversion probabilities we use the results calculated numerically as indicated above .
since the parameter space is three - dimensional , the allowed regions for a given c.l . are defined as the set of points satisfying the condition @xmath31 in our numerical calculations we use the survival / conversion probabilities of solar electron neutrinos valid over a generous range of nsi parameters @xmath30 . on the other hand
, we use the relevant reaction cross sections and efficiencies for the all experiments employed in ref .
@xcite . for the sno case
the cc cross section for deuterium was taken from @xcite . for what concerns
the fit of the total rates , we take into account the results of the homestake , gallex / gno , sage and super - kamiokande experiments , together with the latest sno cc result .
thus we have @xmath32 experimental data , which we fit in terms of the @xmath33 parameters @xmath27 , @xmath28 and @xmath0 ; therefore , the total number of degrees of freedom is @xmath34 .
in addition to total rates , we also take into account the super - kamiokande `` zenith angle spectrum '' @xcite , which includes both the spectral information on the final lepton energy ( 8 bins ) and the information on the zenith angle distribution which results from neutrino interactions inside the earth ( 1-day and 6-night bins ) .
this data sample contains a total of @xmath35 bins .
thus the number of degrees of freedom is 5 ( rates ) + 44 ( zenith - spectrum ) - 1 ( free normalization ) - 3 ( fit parameters ) = 45
. allowed nsi regions indicated by the fit of the solar rates ( upper panels ) and global solar data ( lower panels ) for different values of @xmath0 , assuming non - standard interactions of neutrinos with @xmath36-type quarks .
the shaded areas refer to the 90% , 95% , 99% and 99.7% c.l . with 3 degrees of freedom .
the best fit point is indicated by a star.,scaledwidth=85.0% ] same as fig .
[ fig : quark - d ] but for the case of non - standard interactions of neutrinos with @xmath37-type quarks.,scaledwidth=85.0% ] in figs .
[ fig : quark - d ] and [ fig : quark - u ] we present the allowed regions in the @xmath30 parameter space for different values of @xmath38 , assuming non - standard interactions of neutrinos with @xmath36-type ( fig . [ fig : quark - d ] ) and @xmath37-type ( fig .
[ fig : quark - u ] ) quarks .
the shaded areas refer to the 90% , 95% , 99% and 99.7% c.l . with 3 degrees of freedom , and the best fit point
is indicated by a star .
let us now comment some of their features .
the first thing to notice from figs .
[ fig : quark - d ] and [ fig : quark - u ] is that the amount of nu is large .
nevertheless this is allowed by experiment and therefore consistent .
we note also that the nsi strength required in the case of interaction with u - type quarks is smaller than for the case of d - type quarks .
notice also that @xmath27 can only be large over a very narrow @xmath28 region close to 0.57 , a fact which will be used shortly .
more importantly , we note that the quality of the solar fit becomes progressively worse as @xmath0 increases , in a way similar to what happens in a three - neutrino osc scenario @xcite .
this indicates that , although with less sensistivity , @xmath0 can be constrained solely by solar data , irrespective of reactor data , also in the context of the nsi mechanism . in fig .
[ fig : chisq ] we present the dependence of @xmath39 as a function of @xmath40 .
dependence of @xmath39 as a function of @xmath41 , for different nsi solutions.,scaledwidth=70.0% ] the _ solid _ line refers to non - standard interactions of neutrinos with @xmath36-type quarks , and only the information from total rates is used , while for the _ dashed _ line also the spectrum - zenith distribution of super - kamiokande is included in the fit .
the two _ dotted - dashed _ lines refer to non - standard interactions of neutrinos with @xmath37-type quarks , both with and without the spectrum - zenith information .
finally in table [ tab : rates ] we present the best - fit points and goodness - of - fit of oscillation and nsi solutions to the solar neutrino problem . since the neutrino mixing angle @xmath0 is strongly constrained by reactor neutrino data @xcite we set , for definiteness , @xmath42 in what follows . this way we have only two degrees of freedom . ? = ?
lcccccc solution & @xmath43 & @xmath44 & @xmath27 & @xmath28 & @xmath45 & g.o.f .
+ lma & @xmath46 & @xmath47 & @xmath48 & @xmath48 & @xmath49 & @xmath50 + sma & @xmath51 & @xmath52 & @xmath48 & @xmath48 & @xmath53 & @xmath54 + nsi ( d ) & @xmath48 & @xmath48 & @xmath55 & @xmath56 & @xmath57 & @xmath58 + nsi ( u ) & @xmath48 & @xmath48 & @xmath59 & @xmath60 & @xmath61 & @xmath62 + the numbers in the table refer to a restricted analysis using only @xmath43 and the neutrino mixing angle @xmath18 for the pure osc case and only @xmath27 and @xmath28 for the pure nsi case .
we see that the pure two - parameter nsi solution is somewhat better than the corresponding pure osc solution .
so far we have given a description of solar neutrino data in terms of nsi interactions .
we now turn to the issue of atmospheric neutrinos . in principle one
could imagine a pure nsi description of the atmospheric data itself @xcite .
however , it has recently been shown that , while it may fit the contained atmospheric data , such a description can not reconcile them with the up - going muon data .
the sensitivity of the atmospheric data to the nsi follows mainly from the fact that the wide energy range of the atmospheric neutrino data sample , from the sub - gev events up to the up - going samples of macro and super - kamiokande @xcite is enough to manifest the energy - dependence characteristic of the atmospheric neutrino conversion mechanism . on this basis
it has been shown how , taken altogether , the atmospheric data leave very little room for nsi and can tolerate the existence of nsi only at a sub - leading level @xcite . the main question for us then is to evaluate whether the nsi description of solar neutrino data is in conflict with the atmospheric data sample .
the danger lies in the fact that , in contrast to a pure oscillation description of the neutrino data @xcite , even when the value of the third neutrino mixing angle @xmath0 ( presently constrained mainly by the reactor neutrino data ) is taken to zero , the solar and atmospheric sectors are still coupled with each other through the postulated nsi interactions whose strength is fixed , as described above , in order to account for the solar data .
a full description of the atmospheric and solar neutrino data in terms of a hybrid osc+nsi description is , at the moment , prohibitive in view of the complexity of the problem .
it suffices to remind the reader that we expect ten relevant independent parameters in this case : in addition to the five relevant cp conserving oscillation parameters of ref .
@xcite ( two mass splittings and three angles ) there are five new nsi parameters characterizing the magnitude of fc and nu cp conserving non - standard interactions .
thus we are forced to make a simplifying _
ansatz_. for definiteness we consider the two following choices of the nsi matrix : @xmath63 and @xmath64 with each of these choices the five new nsi parameters are reduced to the two parameters involved in the description of the solar data , as explained previously in sections [ sec : neutr - evol - pres ] and [ sec : analysis - method ] .
in fact for both of these choices , once we set @xmath42 and @xmath65 ( see below ) , we have simply @xmath66 and @xmath67 .
for definiteness we focus here on the case of neutrino non - standard interactions with down - type quarks in eq . .
as for the data here we use the totality of sub- and multi - gev ( @xmath68 , @xmath69 ) atmospheric neutrino data @xcite as well as super - kamiokande stop and through - going muon data , and macro through - going muons @xcite . allowed region ( 90% , 95% , 99% , 99.73% c.l . with 2 d.o.f .
) in the plane @xmath70 from the analysis of atmospheric neutrino data , for each of the two _ anstze _ given in eqs .
( left panels ) and ( right panels ) .
the best fit point is denoted as a star .
we fix the undisplayed parameters to @xmath42 , @xmath71 and @xmath72 .
the top panels show the behavior of @xmath73 as a function of @xmath74 when @xmath75 is integrated out.,scaledwidth=99.0% ] in the lower panels of fig .
[ fig : atm - epsi ] we show the allowed region in the plane @xmath70 from the analysis of atmospheric neutrino data .
the remaining undisplayed parameters are set to @xmath42 ( indicated by a combination of reactor and atmospheric data @xcite ) , @xmath71 ( indicated by the standard atmospheric data analysis ) and @xmath76 ( indicated by the solar data analysis presented in sections [ sec : neutr - evol - pres ] and [ sec : analysis - method ] ) . in the upper panels we show the dependence of @xmath77 as a function of @xmath74 when @xmath75 is `` integrated out '' . from fig . [
fig : atm - epsi ] we see that in both cases the value of @xmath75 is essentially unaffected by the inclusion of non - standard interaction in the @xmath78 and @xmath79 channels , up to values of 10 % or so .
this is analogous to what was already found in ref .
@xcite for the @xmath80 channel .
we also note that in model @xmath81 values of @xmath82 are preferred , since in this case a small contribution of fc actually helps in improving the sk contained @xmath7 data .
this phenomenon is analogous to having a small but finite solar mass splitting @xmath17 , see ref .
@xcite . however , a value of @xmath83 is clearly in agreement with the data , as expected by the excellent quality of the oscillation fit for the pure @xmath84 vacuum oscillations .
we see that in both schemes the values of @xmath74 allowed by the solar data analysis are fully acceptable also for the atmospheric data .
however in model @xmath81 there is a small tension between the solar best fit point ( @xmath85 ) and the atmospheric data ( which prefer , with @xmath86 , a value @xmath87 ) , so that some slight change in the shape of the solar allowed region might be expected when atmospheric data are also included .
this effect does not appear in ansatz @xmath88 , for which any value of @xmath74 below @xmath89 is practically equivalent .
in this paper we have re - analysed the non - standard interaction solutions to the solar neutrino problem in the light of the recent sno measurement as well as the 1258day super - kamiokande data .
in contrast to previous work we have used a three - neutrino interpretation of the solar data analysis , displaying how results depend on the neutrino mixing angle @xmath0 which is constrained mainly by the reactor neutrino data .
more importantly , we have checked consistency of the nsi interpretation of solar data with the good agreement that the atmospheric data sample shows with the osc interpretation .
we have found that the status of such energy - independent solution to the solar neutrino problem is slightly better than that of the osc solution . while these nsi solutions exist for reasonable values of the flavour - changing interaction strength @xmath90 they require a somewhat large value of the non - universal parameters , suggesting that the solar conversion channel must involve @xmath91 .
we have also analysed the implications of the solar nsi solution for atmospheric neutrinos , generalizing the study in ref .
@xcite so as to analyse the sensitivity of atmospheric data also to the nsi parameters involved in the solar neutrino channel . by using two
simplified _ anstze _ for the nsi interactions we have explicitly demonstrated that the values of @xmath74 allowed by the solar data analysis are fully acceptable also for the atmospheric data .
this establishes that the nsi description of the solar neutrino data is not in conflict with the atmospheric data sample . for one of these models we noted the existence of a slight conflict between the solar best fit point ( @xmath85 ) and the atmospheric data ( which prefer , with @xmath86 , a value @xmath87 ) , which suggests that some slight change in the shape of the solar allowed region might be expected within a fully global description of both solar the atmospheric data . our simplified _
anstze _ are justified to the extent that the complexity of the analysis makes a full description close to impossible , and possibly un - illuminating .
we find it significant , however , that the present understanding of solar and atmospheric neutrino data does not , as yet , require solar neutrino oscillations or solar neutrino mixing .
this may have profound implications for model - building , especially in view of the difficulties in accomodating bi - large mixing - type solutions within the framework of unified theories . from an experimental point of view
we find it worth pointing that the energy dependence characteristic of the oscillation mechanism has only been demonstrated for the case atmospheric , not solar neutrino conversions , as yet .
one may argue that our nsi hypothesis is somewhat artificial .
however one should bear in mind the fact that most models neutrino masses are accompanied by non - standard interactions of neutrinos and there are , in fact , some in which nsi are unacompanied by neutrino mass effects .
the soundness of hybrid schemes such as the ones indicated here may indicate more subtle new ways for accounting for the presently observed neutrino anomalies from first principles .
we look forward to new solar neutrino experiments to probe neutrino nsi with improved sensitivities , such as suggested in ref .
@xcite , as well as to the prospects of future neutrino factories performing correlated osc - nsi studies , as suggested in @xcite .
this work was supported by spanish dgicyt under grant pb98 - 0693 ( spain ) , fapesp and cnpq ( brazil ) , by the european commission rtn network hprn - ct-2000 - 00148 and by the european science foundation network grant n. 86 .
m.m . was supported by the marie curie contract hpmf - ct-2000 - 01008 .
m.a.t . was supported by the m.e.c.d .
fellowship ap2000 - 1953 .
we thank o. peres , for useful discussions .
the original version of the present paper showed that a pure nsi description of solar data with massless and unmixed neutrinos is slightly better than the favoured lma - msw solution .
moreover , the nsi values indicated by the solar data analysis do not spoil the succesful oscillation description of the atmospheric data , establishing the overall consistency of a hybrid scheme in which only atmospheric data are explained in terms of neutrino oscillations .
however , the recent results of the kamland collaboration @xcite reject non - oscillation solutions in the solar sector under the assumption of cpt invariance , in such a way that solutions based on nsi should be strongly disfavoured by kamland results . in this appendix , we re - evaluate the status of nsi solutions of the solar neutrino problem , in the light of the first kamland data .
the kamland experiment is a reactor neutrino experiment whose detector is located at the kamiokande site .
most of the @xmath92 flux incident at kamland comes from nuclear plants at distances 80 - 350 km from the detector , making the average baseline of about 180 km , long enough to provide a sensitive probe of the lma - msw region .
the target for the @xmath92 flux consists of a spherical transparent balloon filled with 1000 tons of non - doped liquid scintillator , and the antineutrinos are detected via the inverse neutron @xmath93-decay process @xmath94 .
the kamland collaboration has for the first time measured the disappearance of neutrinos produced in a power reactor .
they observe a strong evidence for the disappearance of neutrinos during their flight over such distances , giving the first terrestrial confirmation of the solar neutrino anomaly and also establishing the oscillation hypothesis with man - produced neutrinos . in this appendix
we perform a combined analysis of the solar neutrino data with the first kamland results in terms of neutrino non - standard interactions with matter , setting for definiteness @xmath95 .
the analysis of solar neutrino data includes the most recent results from all experiments @xcite , as well as the latest measurements from sno @xcite presented in the form of 34 data bins ( see ref .
@xcite for a detailed explanation of the new solar analysis ) .
the details of the theoretical monte - carlo and statistical analysis of the kamland results are given in ref .
@xcite ; in particular , the kamland @xmath96-function is calculated assuming a poisson distribution of the experimental data , as described in sec . iv of that paper . in table
[ tab : chi2-kl ] we present the best - fit points for the oscillation and nsi solutions to the solar neutrino problem obtained in our analysis , together with its corresponding value of @xmath96 .
the value of @xmath97 confirms our previous result : nsi picture gives a good description of the solar neutrino data .
in fact , for the case of neutrino nsi with _ up _ quarks we obtain a slightly better fit than for the lma - msw solution .
the inclusion of the latest solar data sample @xcite does not change the status nor the allowed regions of the nsi solution in any significant way .
in contrast , the inclusion of kamland data in the analysis changes dramatically the situation . note that for the kamland experiment matter effects are very small , and can therefore safely be neglected . as a result ,
the nsi solutions predict no reduction in the @xmath92 flux , in conflict with what is observed at kamland .
this worsens the status of the nsi hypothesis , ( reflected in the corresponding @xmath98 values , in table [ tab : chi2-kl ] ) with respect to that of the lma - msw solution , which predicts the correct suppression factor observed at kamland . from our analysis
we obtain that the description in terms of neutrino nsi with _ down _
quarks is rejected with a @xmath99 = 14.2 , corresponding to 3.8@xmath1 , with respect to lma - msw , while nsi of neutrinos with quarks of type _ up _ are rejected at 3.6@xmath1 level ( @xmath99 = 12.8 ) as the explanation for the solar neutrino anomaly plus the kamland disappearance of neutrinos . in summary
, we have shown that the inclusion of kamland data in the analysis of the solar neutrino anomaly excludes the interpretation based on neutrino non - standard interactions with matter at more than 3@xmath1 .
therefore , non - standard interactions may at best play a sub - leading role in solar neutrino propagation .
in fact , one may use the confirmation of the lma - msw oscillation solution together with the experimental data from kamland and solar neutrino experiments in order to determine improved restrictions on nsi parameters .
lcccccc solution & @xmath43 & @xmath44 & @xmath27 & @xmath28 & @xmath97 & @xmath100 + lma - msw & @xmath101 & @xmath102 & @xmath48 & @xmath48 & @xmath103 & @xmath104 + nsi ( d ) & @xmath48 & @xmath48 & @xmath55 & @xmath61 & @xmath105 & @xmath106 + nsi ( u ) & @xmath48 & @xmath48 & @xmath59 & @xmath107 & @xmath108 & @xmath109 + b.t .
cleveland _ et al .
_ , astrophys .
j. * 496 * ( 1998 ) 505 ; k.s .
hirata _ et al .
_ , kamiokande coll .
* 77 * ( 1996 ) 1683 ; w. hampel _ et al .
_ , gallex coll .
b * 447 * ( 1999 ) 127 ; d.n .
abdurashitov _ et al .
_ , sage coll .
* 83 * ( 1999 ) 4686 ; phys .
rev . * c 60 * ( 1999 ) 055801 [ astro - ph/9907113 ] ; v. gavrin ( sage collaboration ) , nucl .
* b 91 * ( 2001 ) 36 ; m. altmann _ et al . _ ( gno collaboration ) , phys . lett .
( 2000 ) 16 ; e. bellotti _ et al . _ ( gno collaboration ) , nucl .
* b 91 * ( 2001 ) 44 ; y. fukuda _ et al . _
( super - kamiokande collaboration ) , phys .
* 81 * ( 1998 ) 1158 ; erratum * 81 * ( 1998 ) 4279 ; phys .
* 82 * ( 1999 ) 1810 ; y. suzuki ( super - kamiokande collaboration ) , nucl .
proc . suppl .
* b 91 * ( 2001 ) 29 ; s. fukuda _ et al .
_ ( super - kamiokande collaboration ) , phys .
* 86 * ( 2001 ) 5651 and * 86 * ( 2001 ) 5656 .
q. r. ahmad _ et al .
_ [ sno collaboration ] , phys . rev . lett .
* 87 * ( 2001 ) 071301 [ nucl - ex/0106015 ] .
super - kamiokande coll .
, y. fukuda _ et al .
_ , phys .
* 81 * ( 1998 ) 1562 ; c. mcgrew in _ neutrino telescopes 2001 _ , venice , italy , march 2001 , to appear ; t. toshito in _ moriond 2001 _ , les arcs , france , march 2001 , to appear .
y. fukuda _ et al .
_ , kamiokande coll . , phys .
b * 335 * ( 1994 ) 237 ; r. becker - szendy _ et al .
_ , imb coll .
. phys .
. suppl . ) * 38 * ( 1995 ) 331 ; w.w.m .
allison _ et al .
_ , soudan coll .
b * 449 * ( 1999 ) 137 .
macro coll .
, m. ambrosio _ et al .
_ , phys .
b * 434 * ( 1998 ) 451 ; m. spurio _ et al .
_ , hep - ex/0101019 ; b. barish , talk given at _
neutrino 2000 _ , 1521 june 2000 , sudbury , canada [ ` http://nu2000.sno.laurentian.ca ` ] .
j. schechter and j. w. valle , phys .
d * 22 * ( 1980 ) 2227 .
d * 23 * ( 1981 ) 1666 .
see , for example , m. c. gonzalez - garcia , m. maltoni , c. pena - garay and j. w. f. valle , phys .
d * 63 * ( 2001 ) 033005 [ hep - ph/0009350 ] , for reviews see s. m. bilenkii , c. giunti and w. grimus , prog .
* 43 * ( 1999 ) 1 [ arxiv : hep - ph/9812360 ]
. g. l. fogli in _ neutrino telescopes 2001 _ , venice , italy , march 2001 ; v. d. barger , arxiv : hep - ph/0005011 . for
a review see j. w. f. valle , prog .
* 26 * ( 1991 ) 91 .
g. g. ross and j. w. f. valle , phys .
b * 151 * , 375 ( 1985 ) .
j. r. ellis _ et al .
b * 150 * , 142 ( 1985 ) .
l. j. hall and m. suzuki , nucl .
b * 231 * , 419 ( 1984 ) .
m. a. diaz , j. c. romao and j. w. f. valle , nucl .
b * 524 * , 23 ( 1998 ) [ hep - ph/9706315 ] .
r. hempfling , nucl .
b * 478 * ( 1996 ) 3 [ arxiv : hep - ph/9511288 ] . m. hirsch _ et al .
d * 62 * ( 2000 ) 113008 [ arxiv : hep - ph/0004115 ] .
d * 61 * ( 2000 ) 071703 [ arxiv : hep - ph/9907499 ] .
o. haug _ et al .
b * 565 * ( 2000 ) 38 [ arxiv : hep - ph/9909318 ] .
r. n. mohapatra and j. w. f. valle , phys .
d * 34 * ( 1986 ) 1642
. l. j. hall , v. a. kostelecky and s. raby , nucl .
b * 267 * ( 1986 ) 415 ; l. wolfenstein , phys .
d * 17 * , 2369 ( 1978 ) ; s.p .
mikheev and a.yu .
smirnov , sov .
j. nucl .
* 42 * , 913 ( 1985 ) .
e. roulet , phys .
* d44 * , 935 ( 1991 ) ; v. barger , r. j. n. phillips and k. whisnant , phys .
rev . * d44 * , 1629 ( 1991 ) ; s. bergmann , nucl . phys .
* b515 * , 363 ( 1998 ) ; e. ma and p. roy , phys .
* 80 * , 4637 ( 1998 ) .
j. bahcall and p. krastev in hep - ph/9703267 .
gonzalez - garcia _ et al .
_ , phys . rev .
* 82 * ( 1999 ) 3202 ; n. fornengo , m.c . gonzalez - garcia and j.w.f .
valle , jhep * 0007 * ( 2000 ) 006 ; p.
lipari and m. lusignoli , phys .
d * 60 * ( 1999 ) 013003 ; g.l .
fogli , e. lisi , a. marrone and g. scioscia , phys .
d * 60 * ( 1999 ) 053006 m.m .
guzzo , h. nunokawa , o.l.g .
peres and r. zukanovich funchal , nucl .
* 87 * ( 2000 ) 201 .
h. nunokawa _ et al .
_ , phys .
d * 54 * ( 1996 ) 4356 [ hep - ph/9605301 ] ; d. grasso , h. nunokawa and j.w.f .
valle , phys .
* 81 * ( 1998 ) 2412 z. berezhiani , r. s. raghavan and a. rossi , nucl .
b * 638 * ( 2002 ) 62 [ arxiv : hep - ph/0111138 ] .
z. berezhiani and a. rossi , phys .
b * 535 * ( 2002 ) 207 [ arxiv : hep - ph/0111137 ] .
guzzo , h. nunokawa , p.c .
de holanda and o.l.g .
peres , phys .
* d * 64 ( 2001 ) 097301 .
p. huber , t. schwetz and j. w. f. valle , phys .
* 88 * ( 2002 ) 101804 [ arxiv : hep - ph/0111224 ] .
a. m. gago _ et al .
_ , phys .
d * 64 * ( 2001 ) 07003 ; p. huber and j. w. f. valle , phys . lett .
b * 523 * ( 2001 ) 151 [ arxiv : hep - ph/0108193 ] .
a. m. gago _ et al .
_ , phys .
d * 65 * ( 2002 ) 073012 [ arxiv : hep - ph/0112060 ] ; s. bergmann _ et al .
_ , phys .
d * 62 * ( 2000 ) 073001 [ hep - ph/0004049 ] .
m. apollonio _ et al . _ [ chooz collaboration ] , phys .
b * 466 * , 415 ( 1999 ) [ hep - ex/9907037 ] .
f. boehm _ et al .
_ , palo verde coll .
d * 64 * ( 2001 ) 112001 [ hep - ex/0107009 ] .
we adopt ssm data from ` http://www.sns.ias.edu/~jnb/sndata/sndata.html ` . j. schechter and j. w. f. valle , phys .
d * 21 * , 309 ( 1980 ) .
lim , in _ proceedings of bnl neutrino workshop _ , p. 249
, february 1987 , new york , usa , edited by m. j. murtagh , bnl-52079 , c87/02/05 . g. l. fogli , e. lisi , and d. montanino , phys .
d * 49 * , 3626 ( 1994 ) ; astropart .
phys . * 4 * , 177 ( 1995 ) .
t. k. kuo and j. pantaleone , phys .
lett . * 57 * , 1805 ( 1986 ) ; phys .
d * 35 * , 3432 ( 1987 ) ; rev .
phys . * 61 * , 937 ( 1989 ) .
m. c. gonzalez - garcia , p. c. de holanda , c. pena - garay and j. w. f. valle , nucl .
phys . * b573 * ( 2000 ) 3 [ hep - ph/9906469 ] and references therein . j. n. bahcall , m. c. gonzalez - garcia and c. pena - garay , jhep * 0108 * ( 2001 ) 014 [ hep - ph/0106258 ] . for other recent papers on the oscillation interpretation of the solar neutrino anomaly after the sno result
see v. barger , d. marfatia and k. whisnant , phys .
* 88 * ( 2002 ) 011302 [ hep - ph/0106207 ] .
a. bandyopadhyay , s. choubey , s. goswami and k. kar , phys .
b * 519 * ( 2001 ) 83 [ hep - ph/0106264 ] .
g. l. fogli , e. lisi , d. montanino and a. palazzo , phys .
d * 64 * ( 2001 ) 093007 [ hep - ph/0106247 ] ; p. creminelli , g. signorelli and a. strumia , hep - ph/0102234v2 ; for a discussion on the status of oscillation solutions including also sn1987a data see m. kachelriess , a. strumia , r. tomas and j. w. f. valle , phys . rev .
d * 65 * ( 2002 ) 073016 [ arxiv : hep - ph/0108100 ] .
beacom and s.j .
parke , phys .
d * 64 * ( 2001 ) 091302 [ hep - ph/0106128 ] ; s. nakamura , t. sato , v. gudkov and k. kubodera , phys .
* c 63 * ( 2001 ) 034617 ; m. butler , j .- w .
chen and x. kong , phys .
* c 63 * ( 2001 ) 035501 ; i.s .
towner , phys .
* c 58 * ( 1998 ) 1288 .
fogli , e. lisi and d. montanino and a. palazzo , phys .
* d * 62 ( 2000 ) 013002 ; a.m. gago , h. nunokawa and r. zukanovich funchal , phys . rev .
63 ( 2001 ) 013005 [ erratum ibid . * d * 64 ( 2001 ) 119902 ] ; a. bandyopadhyay , s. choubey , s. goswami and k. kar , hep - ph/0110307 .
s. fukuda _ et al .
_ [ superkamiokande collaboration ] , phys .
b * 539 * , 179 ( 2002 ) [ arxiv : hep - ex/0205075 ] .
j. n. abdurashitov
_ et al . _ [ sage collaboration ] , j. exp .
phys . * 95 * , 181 ( 2002 ) [ zh .
fiz . * 122 * , 211 ( 2002 ) ] [ arxiv : astro - ph/0204245 ] .
t. kirsten , for the gno collaboration , talk at the neutrino 2002 conference , muenchen ( germany ) may 2002 , to be published on nucl .
b ( proc . suppl . )
q. r. ahmad _ et al . _ [ sno collaboration ] , phys .
lett . * 89 * , 011301 ( 2002 ) ; phys .
lett . * 89 * , 011302 ( 2002 ) .
m. maltoni , t. schwetz , m. a. tortola and j. w. valle , phys .
d * 67 * , 013011 ( 2003 ) [ arxiv : hep - ph/0207227 ] .
m. maltoni , t. schwetz and j. w. valle , arxiv : hep - ph/0212129 .
d , in press . | we re - analyse the non - standard interaction ( nsi ) solutions to the solar neutrino problem in the light of the latest solar as well as atmospheric neutrino data .
the latter require oscillations ( osc ) , while the former do not .
within such a three - neutrino framework the solar and atmospheric neutrino sectors are connected not only by the neutrino mixing angle @xmath0 constrained by reactor and atmospheric data , but also by the flavour - changing ( fc ) and non - universal ( nu ) parameters accounting for the solar data . since the nsi solution is energy - independent the spectrum is undistorted , so that the global analysis observables are the solar neutrino rates in all experiments as well as the super - kamiokande day - night measurements .
we find that the nsi description of solar data is slightly better than that of the osc solution and that the allowed nsi regions are determined mainly by the rate analysis . by using a few simplified _ anstze _ for the nsi interactions we explicitly demonstrate that the nsi values indicated by the solar data analysis are fully acceptable also for the atmospheric data . in the appendix we present an updated analysis combining the latest data from all solar neutrino experiments with the first results from kamland .
we show that , although nsi still gives an excellent description of the solar data , the inclusion of kamland excludes at more than 3@xmath1 the nsi hypothesis as a solution to the solar neutrino problem .
neutrino oscillations , solar and atmospheric neutrinos , neutrino mass and mixing , non - standard neutrino interactions 14.60.pq , 14.60.lm , 26.65.+t , 13.15.+g [ 0pt][0pt ] , , , , and |
it is very well known that poisson s equation is among the most important equations in electromagnetic theory .
this equation fundamentally links the source of the field , i.e. charge distribution @xmath0 , and the electric potential @xmath1 . in theory , this equation allow us to find all we need to know about an electrostatic field given the charge distribution .
the general solution to poisson s equation has the following integral form @xcite @xmath2 once we know @xmath1 the electric field is calculated from the gradient of @xmath1 , i.e. @xmath3 . nevertheless , the integral given in eq .
( [ eqi1 ] ) can only be evaluated analytically for the simplest charge configurations.this is why it is very useful to recast the problem using poisson s equation with appropiate boundary conditions .
for the restricted case of static charges and fields in vacuum , the appropiate boundary conditions for the electric potential are @xcite @xmath4 on the other hand , the standard boundary conditions for the wave function @xmath5 in one dimensional quantum mechanics are : @xcite @xmath6 in particular , if there is a delta function potential at @xmath7 , then the change of slope of the wave function , @xmath8 , at the point @xmath7 is proportional to the amplitude of the delta function as well as the wave function @xmath9 at this point.@xcite this is the type of boundary conditions we expect for the electrostatic potential . as a consequence , the electrostatic potential and the one dimensional wave function share the same boundary conditions for a delta function potential
+ the one dimensional delta - function potential has proved to be a useful model to represent diatomic ion and molecular orbitals,@xcite two electron systems,@xcite electron - dipole interaction,@xcite solids,@xcite the many body problem,@xcite and scattering.@xcite it is well known that the single attractive delta function potential , i.e. @xmath10 , possesses one exponentially localized bound state for all values of the strength of the well @xmath11 .
its existence and stability has been tested for the effects of different boundary conditions@xcite and symmetry - breaking perturbations.@xcite + in this article , a mapping between a three dimensional electrostatic problem described by poisson s equation and the one dimensional schrndinger equation is given through a simple transformation .
this relation will allow us to find the wave function and energy of the ground state of the quantum system by using electrostatic results .
+ the article is organized as follows .
first we will start by transforming the poisson equation into a schrdinger - like equation , and discuss when both equations are equivalent
. then we will apply electrostatic results to solve the schrdinger equation for the case of a potential function given by a combination of attractive and repulsive delta functions .
the conclusions are summarized in the last section .
the poisson equation in three dimensions is given by @xmath12 where @xmath1 is the electric potential , @xmath0 is the volume charge density and @xmath13 is the electric permitivity of free space.@xcite + making the following transformation @xmath14 , where @xmath15 and @xmath16 are constants to ensure dimensional consistency , we obtain @xmath17 where we have used the fact that @xmath18 . multiplying eq .
( [ eq02 ] ) by @xmath19 we end up with @xmath20 equation ( [ eq02a ] ) is a schrdinger - type equation when the magnitude @xmath21 is constant .
we know from basic electrodynamic courses that the only charge distribution to produce uniform electrostatic fields are infinite charge sheets .
since the infinite charge sheet has plane symmetry , the electrostatic potential depends on one variable only , i.e. @xmath22 . for this case
( [ eq02a ] ) is given by @xmath23 making the following substitutions : @xmath24 where @xmath25 is an arbitrary length and @xmath26 is the asymptotic magnitude of the electric field , i.e. @xmath27 , we obtain the time independent schrdinger equation @xmath28 for the case of bound states , i.e. @xmath29.@xcite this relationship means that the wave function and the bound state energy will be given in terms of the electric potential and the energy density by @xmath30 where @xmath16 is the normalization constant and @xmath31 is the energy density , respectively .
note that the wave function has no nodes , therefore only the ground state can be found with this method .
this is consistent with the electrostatic uniqueness theorem which guarantees that the electrostatic potential is uniquely determined for a given charge density distribution and the electrostatic potential at all boundaries are specified.@xcite + in order to have physically realizable quantum states the wave function must be normalized.@xcite + in the following section we solve the schrdinger equation using electrostatic solutions .
we know from basic electrodynamic courses that the only charge distribution to produce a uniform electrostatic field are an odd number of parallel infinite sheets with opposite electric charge densities @xmath32 @xcite , hence we will consider a volume charge density given by @xmath33 where @xmath34 is the dirac delta function at the position @xmath35.@xcite let us consider the case when @xmath36 , for this case the volume charge distribution is of the form @xmath37 where we have taken @xmath38 .
the electric field caused by an infinite sheet of charge density @xmath39 at any space point is @xcite @xmath40 therefore , the electric potential and the energy density are given by @xcite @xmath41 using equation ( [ eq04a ] ) we obtain the wave function and the energy of the bound state @xmath42 the wave function given by equation ( [ eq09 ] ) can be normalized only for @xmath43 , otherwise it blows up when @xmath44 . using equation ( [ eq03 ] ) we get the potential energy function @xmath10 for the quantum mechanical case , where @xmath45 is the strength of the potential . making the appropiate substitutions into equation ( [ eq09 ] ) we get the corresponding bound state solution in the presence of an atractive delta function potential @xmath46 where the normalization constant is @xmath47.@xcite using the wave function given in equation ( [ eq10 ] ) it is easy to obtain the expectation values for the potential energy @xmath48 and kinetic energy @xmath49 , respectively .
+ if we consider the general case for @xmath50 , then the electrostatic potential and energy density are given by @xmath51 interestingly , there is one quantum bound state with exactly the same energy for the charge distributions given by equations ( [ eq05 ] ) and ( [ eq06 ] ) , where the wave function is given by @xmath52 remarkably , the solution given in equation ( [ eq12 ] ) is the product of growing and decaying exponential functions . in order to have a physically realizable quantum bound states we need to normalize the wave function given in equation ( [ eq12 ] ) . for simplicity
, i am going to consider the case consisting of evenly spaced infinite charge sheets , i.e. @xmath53 , where @xmath54 is the distance between adjacent sheets .
for this particular case the wave function is an even function and in order to normalize @xmath5 we need to solve the following integral @xmath55 using the fact that @xmath56 and by changing @xmath57 by @xmath58 in the first sum we have @xmath59 then the integral given in equation ( [ eq13 ] ) reduces to @xmath60dz+\\ \nonumber \int_{na}^{\infty}\exp\left[-\frac{\sigma}{\epsilon_0v_0}\left(\sum_{n=0}^{n}(-1)^n|z+na|+\sum_{n=1}^{n}(-1)^n|z - na|\right)\right]dz . \label{eq15}\end{aligned}\ ] ] the result of the integral given above is ( the evaluation of this integral is given in appendix [ appendixa ] ) @xmath61\sum_{k=0}^{n-1}\sinh\left[(-1)^k\frac{\sigma a}{2\epsilon_0v_0}\right]-\frac{\epsilon_0v_0}{\sigma(-1)^n}\exp\left[-\frac{\sigma(-1)^n}{\epsilon_0v_0}z\right]\bigg|_{na}^{\infty } \label{eq16}\ ] ] the result given in equation ( [ eq16 ] ) will diverge unless @xmath62 , for @xmath43 .
this restriction means that we need to have positive infinite charge sheets on both ends of the configuration in order for the wave function to go to zero when @xmath44 .
therefore , the normalization constant is given by @xmath63+\frac{4\epsilon_0v_0}{\sigma}\exp\left[-\frac{(1 + 2n)\sigma a}{2\epsilon_0v_0}\right]\left[\frac{1-(-1)^n}{2}\right]\sinh\left[\frac{\sigma a}{2\epsilon_0v_0}\right]}\,\,\mbox{for $ n\geq1$. } \label{eq17}\ ] ] where we have used the fact that @xmath64=\left[\frac{1-(-1)^n}{2}\right]\sinh\left[\frac{\sigma a}{2\epsilon_0v_0}\right ] . \label{eq18}\ ] ] therefore , if we make the appropiate substitutions we have the bound state energy and wave function for a quantum particle of mass @xmath65 moving through the potential energy function of a one - dimensional ionic crystal @xcite , i.e. @xmath66 which is given by @xmath67}{\sqrt{\frac{\hbar^2}{m\alpha}\exp\left[-\frac{2m\alpha na}{\hbar^2 } \right]+\frac{2\hbar^2 } { m\alpha}\exp\left[-\frac{(1 + 2n)am\alpha } { \hbar^2}\right]\left[\frac{1-(-1)^n}{2}\right]\sinh\left[\frac{m\alpha a}{\hbar^2}\right ] } } , \label{eq19}\ ] ] for @xmath50 and with energy @xmath68 .
this result shows that it is possible to trap a single electron with a one dimensional ionic crystal modeled by repulsive and attractive delta functions for the negative and positive ions , respectively
. the wave function @xmath69 is plotted in fig .
( [ psi ] ) for different values of @xmath70 . [ cols="^,^ " , ] with the wave function given by equation ( [ eq19 ] ) we can calculate the expectation value of the potential energy , which is given by @xmath71\delta(z - na)dz . \label{eq20}\ ] ] solving the integral given above ( the evaluation of this integral is given in appendix [ appendixb ] ) , we find the expectation value of the potential energy function @xmath72}{1 + 2e^{-m\alpha a/\hbar^2}\left[\frac{1-(-1)^n}{2}\right]\sinh[m\alpha a/\hbar^2]}\right ) \label{eq20}\ ] ] and the average kinetic energy which is given by @xmath73}{1 + 2e^{-m\alpha a/\hbar^2}\left[\frac{1-(-1)^n}{2}\right]\sinh[m\alpha a/\hbar^2]}\right ) . \label{eq21}\ ] ] now we will consider the case when the volume charge density is given by @xmath74 the electric field for this charge configuration is given by @xmath75 and the electrostatic potential will be @xmath76 therefore , the dual quantum problem will have a ground state energy given by @xmath77 with the following wave function @xmath78 for the following quantum potential @xmath79 where @xmath80 is one for @xmath81 and zero otherwise .
notice that the origin of the last term in eq .
( [ eq25 ] ) is due to the difference in electromagnetic energy between the asymptotic region and all space points .
we have shown a relation between poisson and schdinger equation which enable us to find the ground state energy and eigenfunction of a dual quantum system using electrostatic solutions .
we have obtained the general solution of the wave function for a quantum particle moving through a potential energy function that describes a one - dimensional ionic crystal of finite length using electrostatic solutions .
the result shows that forces exerted by regularly spaced positively and negatively ions described by attractive and repulsive dirac delta functions admit only one bound state , i.e. it is possible to trap a single electron in a ionic crystal .
the method can be used for non periodic potentials which can describe one dimensional ionic quasicrystals .
i would like to thank an anonymous referee for useful comments and feedback to improve this article .
we need to evaluate the integral @xmath82 where @xmath83 using the fact that @xmath84 , \quad\mbox{for}\quad ka < z<(k+1)a , \label{eqa3}\ ] ] then we can rewrite the first integral in equation ( [ eqa0 ] ) in the form @xmath85\right]dz . \label{eqa4}\ ] ] integrating equation ( [ eqa4 ] ) we get @xmath86\exp\left[-\frac{\sigma a ( -1)^kk}{\epsilon_0v_0}\right]\left[1-\exp\left[-\frac{\sigma a(-1)^k}{\epsilon_0v_0}\right]\right ] .
\label{eqa5}\ ] ] evaluating the sum over @xmath57 and regrouping terms in @xmath87 we end up with @xmath88\sum_{k=0}^{n-1}\sinh\left[\frac{\sigma a ( -1)^k}{2\epsilon_0v_0}\right ] .
\label{eqa6}\ ] ] in orde to evaluate the second integral in equation ( [ eqa0 ] ) , we use the fact that @xmath89 , \quad\mbox{for}\quad na < z<\infty , \label{eqa7}\ ] ] therefore , evaluating the sum over @xmath57 we end up with @xmath90dz , \label{eqa8}\ ] ] integrating equation ( [ eqa8 ] ) we have @xmath91\bigg|_{na}^{\infty}. \label{eqa9}\ ] ] this integral given in equation ( [ eqa9 ] ) clearly diverges unless we choose @xmath92 , with @xmath43 .
we want to calculate the expectation value of the potential energy , which is given by @xmath71\delta(z - na)dz , \label{eqb1}\ ] ] where @xmath93 is the normalization constant . integrating equation ( [ eqb1 ] ) we have @xmath94 , \label{eqb2}\ ] ] by using the following identity @xmath95 and if we let @xmath96 and @xmath97 in the first and second sum , respectively .
these substitutions change the limits of the sums , and we get @xmath98 evaluating the sums over @xmath87 and @xmath99 in equation ( [ eqb4 ] ) we get @xmath100
. \label{eqb5}\ ] ] substituting equation ( [ eqb5 ] ) into equation ( [ eqb2 ] ) we have @xmath101\sum_{n =- n}^{n}(-1)^{n+n}\exp\left[\frac{ma\alpha}{\hbar^2}(-1)^{n+n}\right ] .
\label{eqb6}\ ] ] using the fact that @xmath102=\sum_{n=0}^{2n}(-1)^{n}\exp\left[\frac{ma\alpha}{\hbar^2}(-1)^{n}\right]=e^{ma\alpha/\hbar^2}+2n\sinh[ma\alpha/\hbar^2 ] , \label{eqb7}\ ] ] and substituting the normalization constant , we end up with @xmath103\left(e^{ma\alpha/\hbar^2}+2n\sinh[ma\alpha/\hbar^2]\right)}{\hbar^2\exp\left[-\frac{2m\alpha na}{\hbar^2 } \right]+2\hbar^2 \exp\left[-\frac{(1 + 2n)am\alpha } { \hbar^2}\right]\left[\frac{1-(-1)^n}{2}\right]\sinh\left[\frac{m\alpha a}{\hbar^2}\right ] } , \label{eqb8}\ ] ] which can be simplified into @xmath72}{1 + 2e^{-m\alpha a/\hbar^2}\left[\frac{1-(-1)^n}{2}\right]\sinh[m\alpha a/\hbar^2]}\right ) .
\label{eqb9}\ ] ] by using conservation of energy we can obtain the expectation value for the kinetic energy of the system . | the relation between the poisson and schrdinger equation is obtained through a simple transformation .
it is pointed out that this analogy between both equations can be only applied for potentials that involve a combination of attractive and repulsive delta function potentials .
this relationship enables us to use elementary electrostatic results to find the ground state energy and wave function of the associated quantum state problem in one dimension .
particularly , the result shows that it is possible to trap a single electron in a one dimensional ionic crystal modeled by repulsive and attractive delta functions . |
let @xmath4 denote the closed disc in @xmath5 centred at @xmath6 with radius @xmath7 .
a finite borel measure @xmath0 is said to be @xmath8-dimensional if @xmath9 , and there exists a constant @xmath10 such that @xmath11 for any @xmath12 and @xmath7 . for a kernel function @xmath13 , and a finite measure @xmath0
, we define the singular integral operator associated to @xmath14 by @xmath15 a well - known problem in harmonic analysis is to determine geometric properties of @xmath0 from regularity properties of the operator @xmath16 , see for instance the monograph of david and semmes @xcite .
this paper concerns the question of characterizing those functions @xmath14 with the following property : @xmath17 the property that @xmath18 is equivalent to the boundedness of @xmath16 as an operator in @xmath2 , see for instance @xcite .
a measure @xmath0 is rectifiable if @xmath19 can be covered ( up to an exceptional set of @xmath20 measure zero ) by a countable union of rectifiable curves .
a measure @xmath0 is purely unrectifiable if its support is purely unrectifiable , that is , @xmath21 for any rectifiable curve @xmath22 .
david and lger @xcite proved that the cauchy kernel @xmath3 has property ( @xmath23 ) .
as is remarked in @xcite , the proof in @xcite extends to the case when the cauchy kernel is replaced by either its real or imaginary part , i.e. @xmath24 or @xmath25 . recently in @xcite , chousionis , mateu , prat , and tolsa extended the result of @xcite and showed that kernels of the form @xmath26 have property ( @xmath23 ) for any odd positive integer @xmath27 .
both of these results use the melnikov - menger curvature method . on the other hand ,
huovinen @xcite has shown that there is a purely unrectifiable ahlfors - david ( ad)-regular set @xmath28 for which the singular integral associcated to the kernel @xmath29 is bounded in @xmath30 .
in fact , an essentially stronger conclusion is proved that the principal values of the associated singular integral operator exist @xmath20-a.e .
on @xmath28 .
huovinen takes advantage of several non - standard symmetries and cancellation properties in this kernel to construct his very nice example .
the result of this paper is that a weakened version of huovinen s theorem holds for a very simple kernel function .
indeed , it is perhaps the simplest example of a kernel for which the menger curvature method fails to be directly applicable . from now on , we shall fix @xmath31 we prove the following result . [ thm1 ] there exists a @xmath8-dimensional purely unrectifiable probability measure @xmath0 with the property that @xmath32 in other words , the kernel @xmath14 in ( [ kernel ] ) fails to satisfy property ( @xmath23 ) . at this point
, we would also like to mention huovinen s thesis work @xcite , regarding the kernel function @xmath33 from ( [ kernel ] ) .
it is proved that if @xmath34 @xmath0-a.e .
( essentially the ad - regularity of @xmath0 ) , then the @xmath0-almost everywhere existence of @xmath35 in the sense of principal value implies that @xmath0 is rectifiable .
this result was proved by building upon the theory of symmetric measures , developed by mattila @xcite , and mattila and preiss @xcite .
unfortunately the measure in theorem [ thm1 ] does not satisfy the ad - regularity condition . in view of huovinen s work it would be of interest to construct an ad - regular measure supported on an unrectifiable set for which the conclusion of theorem [ thm1 ] holds .
we have not been able to construct such a measure ( yet ) . for the measure @xmath0 constructed in theorem [ thm1 ] ,
we show that @xmath35 fails to exist in the sense of principal value @xmath0-almost everywhere .
thus the two properties of @xmath2 boundedness of the operator @xmath16 , and the existence of @xmath35 in the sense of principal value , are quite distinct for this singular integral operator .
* let @xmath36 denote the @xmath37-dimensional lebesgue measure normalized so that @xmath38 .
we let @xmath39 denote the @xmath8-dimensional lebesgue measure . *
a collection of squares are essentially pairwise disjoint if the interiors of any two squares in the collection do not intersect . throughout the paper
, all squares are closed .
* we shall denote by @xmath40 and @xmath41 large and small absolute positive constants .
the constant @xmath40 should be thought of as large ( at least @xmath8 ) , while @xmath41 is to be thought of as small ( smaller than @xmath8 ) . * for @xmath42
, the disc @xmath43 denotes the concentric enlargement of a disc @xmath44 by a factor of @xmath45 .
* we define the @xmath20-measure of a set @xmath28 by + @xmath46 * for @xmath12 and @xmath7 , we define the annulus @xmath47 .
* the set @xmath19 denotes the closed support of @xmath0 .
let us make the key observation that allows us to prove theorem [ thm1 ] .
[ refl ] let @xmath12 , @xmath7 .
for any @xmath48 , @xmath49 without loss of generality , we may set @xmath50 and @xmath51 . if @xmath52 , then @xmath53 so whenever @xmath54 , we have @xmath55 .
( this follows merely from the fact that @xmath56 whenever @xmath57 satisfy @xmath58 . ) on the other hand , if @xmath59 , then @xmath60 therefore , if @xmath61 , then @xmath62= \frac{2\pi}{\omega^3}(t|\omega|^2 - 2t^3).\ ] ] since @xmath63 , the desired conclusion follows .
the next lemma will form the basis of the proof of the non - existence of @xmath35 in the sense of principal value .
[ lebmisbal ] there exists a constant @xmath64 such that for any disc @xmath4 , and @xmath65 , @xmath66 [ balance ] by an appropriate translation and rescaling , we may assume that @xmath67 , and @xmath68 . making reference to figure 1 above ,
we split the domain of integration into three regions , @xmath69\}$ ] , @xmath70\}$ ] and @xmath71\}$ ] .
the regions @xmath72 and @xmath73 are respectively the right and left grey shaded regions in figure 1 .
note that @xmath74 if @xmath75 $ ] , and @xmath76 if @xmath77\cup \bigl[\frac{2\pi}{3 } , \pi \bigl].$ ] furthermore , note that @xmath78 but @xmath79 .
therefore , by setting @xmath80 , the lemma follows .
fix @xmath81 such that @xmath82 and @xmath83 .
[ squarepac ] one can pack @xmath84 pairwise essentially disjoint squares of side length @xmath85 into a disc of radius @xmath86 . we may assume that the disc is centred at the origin .
consider the square lattice with mesh size @xmath85 .
label those squares that intersect @xmath87 as @xmath88 .
these squares are contained in @xmath89 . since @xmath90
, we have that @xmath91 . by throwing away @xmath92 of the least desirable squares ,
we arrive at the desired collection .
[ smalldiff ] consider a disc @xmath93 .
let @xmath94 be the collection of squares contained in @xmath95 found in lemma [ squarepac ] .
then @xmath96 since @xmath97 , the property follows from the fact that both sets are contained in @xmath95 .
let @xmath98 , and choose @xmath99 , @xmath100 , to be a sequence which tends to zero quickly .
assume that @xmath101 , @xmath102 , and @xmath103 for all @xmath104 .
several additional requirements will be imposed on the decay of @xmath99 over the course of the following analysis , and we make no attempt to optimize the conditions
. it will be convenient to let @xmath105 for @xmath106 .
first define @xmath107 .
given the @xmath108-th level collection of @xmath109 discs @xmath110 of radius @xmath111 , we construct the @xmath112-st generation according to the following procedure : fix a disc @xmath110 .
apply lemma [ squarepac ] with @xmath113 and @xmath114 to find @xmath115 squares @xmath116 of side length @xmath117 that are pairwise essentially disjoint , and contained in @xmath118 .
let @xmath119 be the centre of @xmath116 , and set @xmath120 .
this procedure is carried out for each disc @xmath110 from the @xmath108-th level collection .
there are a total of @xmath121 discs @xmath122 in the @xmath112-st level .
the above construction is executed for each @xmath123 .
now , set @xmath124 .
define @xmath125 .
we shall repeatedly use the following properties of the construction : \(a ) @xmath126 , for all @xmath127 .
\(b ) @xmath128 for each @xmath129 .
moreover , @xmath130 .
\(c ) @xmath131 whenever @xmath132 , @xmath127 .
[ discpic ] property ( a ) is immediate . to see property ( b ) , merely note that @xmath133 for property ( c ) , we shall use induction . if @xmath134 , then the claim is trivial . using ( b ) , the claimed estimate is clear if @xmath135 and @xmath136 have been created by an application of lemma [ squarepac ] in a common disc @xmath137 .
otherwise , the squares are born out of applying lemma [ squarepac ] to different discs at the @xmath138-st level , and those parent discs are already separated by @xmath139 .
courtesy of properties ( a ) and ( b ) , we see that @xmath140 for each @xmath127 .
set @xmath141 .
each @xmath142 is contained in a unique disc @xmath143 ( or square @xmath135 ) which we shall denote by @xmath144 ( respectively @xmath145 ) .
if @xmath146 , then @xmath147 is covered by the @xmath148 discs @xmath149 that are contained in @xmath143 , each of which has radius @xmath150 .
therefore @xmath151 .
taking @xmath134 yields @xmath152 .
define @xmath153 . set @xmath154
. then @xmath155 , and @xmath156 for all @xmath108 .
therefore , there exists a subsequence of the sequence of measures @xmath157 that converges weakly to a measure @xmath0 , with @xmath158 and @xmath159 .
the following three properties hold : \(i ) @xmath160 whenever @xmath161 , \(ii ) @xmath162 for @xmath161 , and \(iii ) there exists @xmath163 such that @xmath164 for any @xmath12 , @xmath7 and @xmath127 . properties ( i ) and ( ii ) follow immediately from the construction of @xmath165 . to see the third property , note that since @xmath157 is a probability measure , the property is clear if @xmath166 . if @xmath167 , then @xmath168 for some @xmath169 . if @xmath161 , then @xmath4 intersects at most one disc @xmath143 .
. otherwise @xmath171 . in this case , note that since the discs @xmath172 are @xmath173 separated , @xmath4 intersects at most @xmath174 discs @xmath172 .
hence , by property ( ii ) , we see that @xmath175 r_{m+1},\ ] ] which is at most @xmath176 . the weak convergence of a subsequence of @xmath157 to the measure @xmath0 , along with property ( iii ) , yields that @xmath177 for any disc @xmath4 .
we shall henceforth refer to this property by saying that @xmath0 _ is @xmath178-nice_. we have now shown that @xmath0 is @xmath8-dimensional .
notice that we also have @xmath179
as a simple consequence of the weak convergence of @xmath157 to @xmath0 , the property that @xmath18 will follow from the following proposition . [ maxest ] provided that @xmath180 , there exists a constant @xmath10 so that the following holds : suppose that @xmath181 .
then for any @xmath169 with @xmath182 , @xmath183 to begin the proof , fix @xmath184 with @xmath182 .
let @xmath185 with @xmath186 .
for any @xmath187 , @xmath188 , so @xmath189
. now , let @xmath190 be the least integer with @xmath191 ( so @xmath192 ) . then by property ( ii ) of the previous section , @xmath193 the crux of the matter is the following lemma . [ scale]there exists @xmath10 such that for any @xmath106 with @xmath194 , @xmath195 for the proof of lemma [ scale ] , we shall require the following simple comparison estimate
. [ comparison ] let @xmath196 , and @xmath197 .
fix @xmath198 $ ] with @xmath199 .
suppose that @xmath200 and @xmath201 are borel measures , such that @xmath202 , @xmath203 , and @xmath204 .
then , for any @xmath12 with @xmath205 , we have @xmath206 note that the left hand side of the inequality can be written as @xmath207 d(\nu_1-\nu_2)(\xi)\bigl|.\ ] ] but , under the hypothesis on @xmath6 , we have that @xmath208 for any @xmath209 . plugging this estimate into the integral and taking into account the supports of @xmath200 and @xmath201 , the inequality follows .
write @xmath210 first suppose that @xmath211 for @xmath212
. then the hypothesis of lemma [ comparison ] are satisfied with @xmath213 , @xmath214 , @xmath215 , @xmath216 , and @xmath217 . thus @xmath218 now suppose that @xmath212 and @xmath219 .
since @xmath220 , we must have that @xmath221 .
but as @xmath222 , and @xmath162 , we have the following crude bound @xmath223 ( here it is used that @xmath224 for any borel measurable set @xmath225 of finite @xmath36-measure . ) at most @xmath226 of the essentially pairwise disjoint squares @xmath135 , @xmath212 , can satisfy @xmath227 ( and it can only happen at all if @xmath228 ) .
therefore by summing ( [ smalleps ] ) and ( [ largeps ] ) over @xmath212 in the cases when @xmath229 and @xmath230 respectively , we see that the quantity @xmath231 is no greater than a constant multiple of @xmath232 the first term here is bounded by @xmath233 .
since @xmath234 is @xmath178-nice , we bound the second term by @xmath235 we now wish to estimate @xmath236 . with a slight abuse of notation ,
write @xmath237 if @xmath238
. then @xmath239 is bounded by @xmath240 by lemma [ smalldiff ] , this quantity is no greater than @xmath241 it remains to employ the reflectionless property ( lemma [ refl ] ) . since @xmath242 , we use lemma [ refl ] to infer that @xmath243 this quantity is bounded by @xmath244 .
the lemma follows . with lemma [ scale ] in hand
, we may complete the proof of proposition [ maxest ] .
first write @xmath245 next note that that @xmath246 if @xmath247 , and @xmath248 for @xmath249 .
as @xmath250 it follows from lemma [ scale ] that the sum appearing in the right hand side of ( [ splitup ] ) is bounded in absolute value independently of @xmath190 , @xmath251 and @xmath252 .
the remaining term on the right hand side of ( [ splitup ] ) has already been shown to be bounded in absolute value , see ( [ qest ] ) .
we now turn to consider the operator in the sense of principal value .
the primary part of the argument will be the following lemma .
[ misbalance ] provided that @xmath108 is sufficiently large , there exists a constant @xmath253 such that for any disc @xmath143 , and @xmath12 satisfying @xmath254 , @xmath255 before proving the lemma , we deduce from it that @xmath35 fails to exist in the sense of principal value for @xmath0-almost every @xmath12 . to this end , we set @xmath256 .
it suffices to show that @xmath257 . first note that , with @xmath258 , we have @xmath259 , so it suffices to show that @xmath260 for all @xmath108 . to do this , note that for each @xmath261 , at most @xmath262 squares @xmath263 can intersect @xmath264 .
thus @xmath265 where the last inequality holds provided that @xmath251 is sufficiently large . but
then , as long as @xmath108 is large enough , this inequality may be iterated to yield @xmath266 hence @xmath260 . in preparation for proving lemma [ misbalance ]
, we make the following claim .
[ anncomp ] let @xmath106 . for any disc @xmath143 , and @xmath12 , we have @xmath267 to derive this claim , first suppose that a square @xmath268 .
then from a crude application of lemma [ comparison ] ( see ( [ smalleps ] ) ) , we infer that @xmath269 if it instead holds that @xmath270 , then we have the blunt estimate @xmath271,\ ] ] which is bounded by @xmath272
. there are most @xmath273 squares @xmath274 contained in @xmath275 , and no more than @xmath276 squares @xmath274 can intersect the boundary of @xmath277 .
on the other hand , the set @xmath278 consisting of the points in @xmath279 not covered by any square @xmath274 has @xmath36-measure no greater than @xmath280 ( see lemma [ smalldiff ] ) .
thus @xmath281 .
bringing these estimates together establishes claim [ anncomp ] .
let us now complete the proof of lemma [ misbalance ] note that @xmath282 is a lipschitz continuous function in @xmath5 , with lipschitz norm at most @xmath283 .
thus , we infer from lemma [ lebmisbal ] that there is a constant @xmath253 such that @xmath284 whenever @xmath285 .
but now we apply claim [ anncomp ] to deduce that for all such @xmath6 , @xmath286 ( the only part of the support of @xmath0 that @xmath275 intersects is contained in @xmath143 ) .
the right hand side here is at least @xmath287 for all sufficiently large @xmath108 .
we now show that @xmath28 is purely unrectifiable , that is , @xmath288 for any rectifiable curve @xmath22 .
the proof that follows is a simple special case of the well known fact that any set with zero lower @xmath20-density is unrectifiable ( one can in fact say much more , see for instance @xcite ) .
p. huovinen , _ singular integrals and rectifiability of measures in the plane .
* 109 * ( 1997 ) .
p. huovinen , _ a nicely behaved singular integral on a purely unrectifiable set .
soc . * 129 * ( 11 ) ( 2001 ) 33453351 . | an example is constructed of a purely unrectifiable measure @xmath0 for which the singular integral associated to the kernel @xmath1 is bounded in @xmath2 .
the singular integral fails to exist in the sense of principal value @xmath0-almost everywhere .
this is in sharp contrast with the results known for the kernel @xmath3 ( the cauchy transform ) . |
in classification problems , the goal is to predict output labels for given input vectors . for this purpose ,
a decision function defined on the input space is estimated from training samples .
the output value of the decision function is used for the label prediction . in binary classification problems ,
the label is predicted by the sign of the decision function .
many learning algorithms use loss functions to measure the penalty of misclassifications .
the decision function minimizing the empirical mean of the loss function over training samples is employed as the estimator @xcite .
for example , hinge loss , exponential loss and logistic loss are used for support vector machine ( svm ) , adaboost and logistic regression , respectively . especially in the binary classification tasks , statistical properties of learning algorithms based on loss functions are well - understood due to intensive recent works .
see @xcite for details . as another approach
, the maximum - margin criterion is also applied for the statistical learning . under the maximum - margin criterion , the best separating hyperplane between the two output labels
is employed as the decision function . in hard - margin svm @xcite , a convex - hull of input vectors for each binary label
is defined , and the maximum - margin between the two convex - hulls is considered . for the non - separable case , @xmath0-svm provides a similar picture @xcite . in @xmath0-svm , the so - called reduced convex - hull which is a subset of the original convex - hull
is used for the learning .
a reduced convex - hull is defined for each label , and the best separating hyperplane between the two reduced convex - hulls is employed as the decision function .
not only polyhedral sets such as the convex - hull of finite input points but also ellipsoidal sets are applied for classification problems @xcite . in this paper ,
the set used in the maximum - margin criterion is referred to as _ uncertainty set_. this term is borrowed from robust optimization in mathematical programming @xcite .
there are some works in which the statistical properties of the learning based on the uncertainty set are studied .
for example , @xcite proposed minimax probability machine ( mpm ) using the ellipsoidal uncertainty sets , and studied statistical properties under the worst - case setting . in the statistical learning using uncertainty set ,
the main concern is to develop optimization algorithms under the maximum margin criterion @xcite .
so far , statistical properties of the learning algorithm using uncertainty sets have not been intensively studied compared to the learning using loss functions .
the main purpose of this paper is to study the learning algorithm using the uncertainty set .
we focus on the relation between the loss function and the uncertainty set .
we show that the uncertainty set is described by using the conjugate function of the loss function . for given uncertainty set ,
we construct the corresponding loss function .
we study the statistical properties of the learning algorithm using the uncertainty set by applying theoretical results on the loss function approach .
then , we establish the statistical consistency of learning algorithms using the uncertainty set .
we point out that in general the maximum margin criterion for a fixed uncertainty set does not provide accurate decision functions .
we need to introduce a parametrized uncertainty set by the one - dimensional parameter which specifies the size of the uncertainty set .
we show that a modified maximum margin criterion with the parametrized uncertainty set recovers the statistical consistency .
the paper is organized as follows . in section [ sec : preliminaries ] , we introduce the existing method based on the uncertainty set . in section [ sec : loss_and_uncertainty ] , we investigate the relation between loss functions and uncertainty sets .
section [ sec : revision_uncertainty_set ] is devoted to illustrate a way of revising the uncertainty set to recover nice statistical properties . in section
[ sec : learning_algorithm ] , we present a kernel - based learning algorithm with uncertainty sets . in section [ sec : statistical_properties ] , we prove that the proposed algorithm has the statistical consistency .
numerical experiments are shown in section [ sec : numerical_studies ] .
we conclude in section [ sec : conclusion ] .
some proofs are shown in appendix .
we summarize some notations to be used throughout the paper .
the indicator function is denoted as @xmath1 , i.e. , @xmath1 equals @xmath2 if @xmath3 is true , and @xmath4 otherwise .
the column vector @xmath5 in the euclidean space is described in bold face .
the transposition of @xmath5 is denoted as @xmath6 .
the euclidean norm of the vector @xmath5 is expressed as @xmath7 .
for a set @xmath8 in a linear space , the convex - hull of @xmath8 is denoted as @xmath9 or @xmath10 .
the number of elements in the set @xmath8 is denoted as @xmath11 .
the expectation of the random variable @xmath12 w.r.t .
the probability distribution @xmath13 is described as @xmath14 $ ] .
we will drop the subscript @xmath13 as @xmath15 $ ] , when it is clear from the context .
the set of all measurable functions on the set @xmath16 is denoted by @xmath17 or @xmath18 for short .
the supremum norm of @xmath19 is denoted as @xmath20 .
for the reproducing kernel hilbert space @xmath21 , @xmath22 is the norm of @xmath23 defined from the inner product @xmath24 on @xmath21 .
we define @xmath16 as the input space and @xmath25 as the set of binary labels .
suppose that the training samples @xmath26 are drawn i.i.d . according to a probability distribution @xmath13 on @xmath27 .
the goal is to estimate a decision function @xmath28 from a set of functions @xmath29 , such that the sign of @xmath30 provides an accurate prediction of the unknown binary label associated with the input @xmath31 under the probability distribution @xmath13 . in other word , for the estimated decision function @xmath32
, the probability of @xmath33 is expected to be as small as possible . in this article , the composite function of the sign function and the decision function , @xmath34 , is referred to as classifier . in binary classification problems ,
the prediction accuracy of the decision function @xmath32 is measured by the 0 - 1 loss @xmath35 which equals @xmath2 when the sign of @xmath30 is different from @xmath36 and @xmath4 otherwise .
the average prediction performance of the decision function @xmath32 is evaluated by the expected 0 - 1 loss , i.e. , @xmath37.\end{aligned}\ ] ] the bayes risk @xmath38 is defined as the minimum value of the expected 0 - 1 loss over all the measurable functions on @xmath16 , @xmath39 bayes risk is the lowest achievable error rate under the probability @xmath13 .
given the set of training samples , @xmath40 , the empirical 0 - 1 loss is denoted by @xmath41 the subscript @xmath42 in @xmath43 is dropped if it is clear from the context . in general , minimization of @xmath43 is considered as a hard problem @xcite .
the main difficulty is considered to come from non - convexity of the 0 - 1 loss @xmath35 as the function of @xmath32 .
hence , many learning algorithms use a surrogate loss of the 0 - 1 loss in order to make the computation tractable .
for example , svm uses the hinge loss , @xmath44 , and adaboost uses the exponential loss , @xmath45 .
both the hinge loss and the exponential loss are convex in @xmath32 , and they provide an upper bound of the 0 - 1 loss .
thus , the minimizer under the surrogate loss is also expected to minimize the 0 - 1 loss .
the quantitative relation between the 0 - 1 loss and the surrogate loss was studied by @xcite .
to avoid overfitting of the estimated decision function to training samples , the regularization is considered . by adding the regularization term such as the squared norm of the decision function to the empirical surrogate loss
, the complexity of the estimated classifier is restricted .
the balance between the regularization term and the surrogate loss is adjusted by the regularization parameter @xcite .
then , the deviation of the empirical 0 - 1 loss and the expected 0 - 1 loss is controlled by the regularization .
when both the regularization term and the surrogate loss are convex , the computational tractability of the statistical learning is retained . besides statistical learning using loss functions , there is another approach to the classification problems , i.e. , statistical learning based on the so - called _
uncertainty set_. we briefly introduce the basic idea of the uncertainty set .
we assume that @xmath16 is a subset of euclidean space . in robust optimization problems
@xcite , the uncertainty set describes uncertainties or ambiguities included in optimization problems .
the parameter in the optimization problem may not be precisely determined . instead of the precise information
, we have an uncertainty set which probably includes the parameter in the optimization problem .
the worst - case setting is employed to solve the robust optimization problem with the uncertainty set .
the statistical learning with uncertainty set is considered as an application of the robust optimization to classification problems . in classification problems ,
the uncertainty set is designed such that most training samples are included in the uncertainty set with high probability .
we prepare an uncertainty set for each binary label . for example , @xmath46 and @xmath47 are the confidence regions such that the conditional probabilities , @xmath48 and @xmath49 , are equal to @xmath50 . as the other example , the uncertainty set @xmath46 ( resp .
@xmath47 ) consists of the convex - hull of input vectors in training samples having the positive ( resp .
negative ) label .
the convex - hull of data points is used in hard margin svm @xcite . the ellipsoidal uncertainty set is also used for the robust classification under the worst - case setting @xcite . based on the uncertainty set , we estimate the linear decision function @xmath51 . here , we consider the _ minimum distance problem _ @xmath52 let @xmath53 and @xmath54 be optimal solutions of .
then , the normal vector of the decision function , @xmath55 , is estimated by @xmath56 , where @xmath57 is a positive real number .
figure [ fig : uncertaintyset_approach ] illustrates the estimated decision boundary . when both @xmath46 and @xmath47 are compact subsets satisfying @xmath58 , the estimated normal vector can not be the null vector .
the minimum distance problem appears in the hard margin svm @xcite , @xmath0-svm @xcite and the learning algorithms proposed by @xcite . in section [ subsec : uncertainty_set_nusvm ] , we briefly introduce the relation between @xmath0-svm and the minimum distance problem . in minimax probability machine ( mpm )
proposed by @xcite , the other criterion is applied to estimate the linear decision function , though the ellipsoidal uncertainty set plays an important role also in their algorithm . and @xmath47 . ]
the minimum distance problem is equivalent with the maximum margin principle @xcite .
when the bias term @xmath59 in the linear decision function is estimated such that the decision boundary bisects the line segment connecting @xmath53 and @xmath54 , the estimated decision boundary achieves the maximum margin between the uncertainty sets , @xmath60 . according to @xcite ,
we explain how the maximum margin is connected with the minimum distance .
suppose that @xmath46 and @xmath47 are convex subsets and that @xmath58 holds .
then , the margin of two uncertainty sets along the direction of @xmath55 is given as @xmath61 the maximum margin criterion is described as @xmath62 the equality above follows from the minimum norm duality @xcite .
we study the relation between loss functions and uncertainty sets .
first , we introduce the relation in @xmath0-svm according to @xcite and @xcite .
then , we present an extension of @xmath0-svm to investigate a generalized relation between loss functions and uncertainty sets .
suppose that the input space @xmath16 is a subset of euclidean space @xmath63 .
we consider the linear decision function , @xmath51 , where the normal vector @xmath64 and the bias term @xmath65 are to be estimated based on observed training samples . by applying the kernel trick @xcite
, we obtain rich statistical models for the decision function , while keeping the computational tractability . in @xmath0-svm ,
the classifier is estimated as the optimal solution of @xmath66 where @xmath67 is a prespecified constant which has the role of the regularization parameter . as @xcite pointed out
, the parameter @xmath0 controls the margin errors and number of support vectors . in @xmath0-svm ,
a variant of the hinge loss , @xmath68 , is used as the surrogate loss . in the original formulation of @xmath0-svm , the non - negativity constraint , @xmath69 ,
is introduced . as shown by @xcite
, we can confirm that the non - negativity constraint is redundant .
indeed , for an optimal solution @xmath70 , we have @xmath71 where the last inequality comes from the fact that the parameter , @xmath72 , is a feasible solution of . as a result
, we have @xmath73 for @xmath74 .
we briefly show that the dual problem of yields the minimum distance problem in which the reduced convex - hulls of training samples are used as uncertainty sets .
see @xcite for details .
the problem is equivalent with @xmath75 then , the lagrangian function is defined as @xmath76 where @xmath77 are non - negative lagrange multipliers . for the observed training samples ,
we define @xmath78 and @xmath79 as the set of sample indices for each label , i.e. , @xmath80 by applying min - max theorem , we have @xmath81 where the last equality is obtained by changing the variable from @xmath82 to @xmath83 . for the positive ( resp .
negative ) label , we introduce the uncertainty set @xmath46 ( reps .
@xmath47 ) defined by the reduced convex - hull , i.e. , @xmath84 when the upper limit of @xmath85 is less than one , the reduced convex - hull is a subset of the convex - hull of training samples .
we find that solving the problem is identical to solving the minimum distance problem under the uncertainty set of the reduced convex - hulls , @xmath86 the representation based on the minimum distance problem provides an intuitive understanding of the learning algorithm .
we consider general loss functions , and study the relation between the loss function and the corresponding uncertainty set .
again , the decision function is defined as @xmath51 on @xmath63 .
let @xmath87 be a convex and non - decreasing function .
for the training samples , @xmath88 , we propose a learning method in which the decision function is estimated by solving @xmath89 the regularization effect is introduced by the constraint @xmath90 , where @xmath91 is the regularization parameter which may depend on the sample size .
the statistical learning using is regarded as an extension of @xmath0-svm . to see this , we define @xmath92 .
let @xmath70 be an optimal solution of for a fixed @xmath67 . by comparing the optimality conditions of and
, we can confirm that the problem with @xmath93 has the same optimal solution as @xmath0-svm . in the similar way as @xmath0-svm
, we derive the uncertainty set associated with the loss function @xmath94 in .
we introduce the slack variables @xmath95 satisfying the inequalities @xmath96 .
then , the lagrangian function of is given as @xmath97 where @xmath98 and @xmath99 are the non - negative lagrange multipliers .
the optimality conditions , @xmath100 and the non - negativity of @xmath82 lead to the constraint on lagrange multipliers , @xmath101 we define the conjugate function of @xmath102 as @xmath103 then , by applying min - max theorem , we have @xmath104 in section [ sec : statistical_properties ] , we present a rigorous proof that under some assumptions on @xmath105 , the min - max theorem works in the above lagrangian function , i.e. , there is no duality gap . for each binary label , we define the parametrized uncertainty sets , @xmath106 $ ] and @xmath107 $ ] , by @xmath108=\bigg\{\sum_{i\in{m_o}}\alpha_i\x_i \,:\ , \alpha_i\geq0 , \,\sum_{i\in{m_o}}\alpha_i=1,\ \frac{1}{m}\sum_{i\in{m_o}}\ell^*(m\alpha_i)\leq{c } \bigg\}. \label{eqn : uncertainty - set}\end{aligned}\ ] ] then , the optimization problem in is represented by @xmath109,\,\z_n\in\ucal_n[c_n],\ c_p,\,c_n\in\rbb .
\label{eqn : rcm - representation}\end{aligned}\ ] ] let @xmath110 and @xmath111 be the optimal solution of @xmath112 and @xmath113 in .
let @xmath114 be an optimal solution of @xmath55 in .
the saddle point of the above min - max problem provides the relation between the @xmath110 , @xmath111 and @xmath114 . some calculation yields that , when @xmath115 holds , any vector such that @xmath116 satisfies the kkt condition of . on the other hand ,
when @xmath117 holds , @xmath114 is given by @xmath118 .
hence , an optimal solution of the normal vector in the linear decision function is given as @xmath119 we show a sufficient condition that the equality @xmath120 holds .
suppose that @xmath121\cap\ucal_n[c_n]$ ] is nonempty for all @xmath122 and @xmath123 , whenever @xmath121 $ ] and @xmath124 $ ] are both nonempty .
then , clearly @xmath125\cap\ucal_n[c_n]$ ] is the optimal choice of the objective function in . in @xmath0-svm with a small @xmath74 , the reduced convex - hulls satisfy @xmath58 , and hence , @xmath120 and @xmath126 hold .
the bias term @xmath59 in the linear decision function is not directly obtained from the optimal solution of without knowing the explicit form of the loss function @xmath94 . a simple way of estimating the bias term is to choose @xmath127 , which provides the decision boundary bisecting the line segment connecting @xmath110 and @xmath111 . in the learning algorithm proposed in section [ sec : learning_algorithm ]
, the bias term is estimated by minimizing the error rate @xmath128 since the estimated normal vector @xmath114 is substituted in the above objective function , the optimization is tractable .
based on the argument above , we propose the learning algorithm using uncertainty sets in figure [ fig : simple_learning_algorithm ] .
it is straightforward to apply the kernel method to the algorithm . in order to study statistical properties of the learning algorithm based on uncertainty sets , we need more elaborate description on the algorithm .
details are presented in section [ sec : learning_algorithm ] .
we show some examples of uncertainty sets associated with popular loss functions . in the following examples , the index sets , @xmath78 and @xmath79 , are defined by for the training samples @xmath88 , and let @xmath129 and @xmath130 be @xmath131 and @xmath132 , respectively .
as explained above , the problem is reduced to @xmath0-svm by defining @xmath92 .
the conjugate function of @xmath94 is given as @xmath133,\\ \infty , & \alpha\not\in[0,2/\nu],\\ \end{cases } \end{aligned}\ ] ] and the associated uncertainty set is defined by @xmath134&= \begin{cases } \displaystyle \bigg\ { \sum_{i\in{m_o}}\alpha_i\x_i\,:\,\sum_{i\in{m_o}}\alpha_i=1,\ , 0\leq\alpha_i\leq\frac{2}{m\nu},\,i\in{m_o } \bigg\ } , & c\geq0,\\ \displaystyle \emptyset , & c<0 . \end{cases}\quad
\end{aligned}\ ] ] for @xmath135 , the uncertainty set consists of the reduced convex - hull of training samples , and it does not depend on the parameter @xmath57 .
in addition , the negative @xmath57 is infeasible .
hence , in the problem , optimal solutions of @xmath122 and @xmath123 are given as @xmath136 , and the problem is reduced to the simple minimum distance problem .
[ example : uncertainty_truncated_quadratic ] now consider @xmath137 .
the conjugate function is @xmath138 for @xmath139 , we define @xmath140 and @xmath141 as the empirical mean and the empirical covariance matrix of the samples @xmath142 , i.e. , @xmath143 suppose that @xmath141 is invertible .
then , the uncertainty set corresponding to the truncated quadratic loss is given as @xmath134 & = \bigg\ { \sum_{i\in{m_o}}\alpha_i\x_i\,:\,\sum_{i\in{m_o}}\alpha_i=1,\ , \alpha_i\geq{0},\,i\in{m_o},\,\sum_{i\in{m_o}}\alpha_i^2\leq \frac{4(c+1)}{m } \bigg\}\\ & = \bigg\ { \z\in\conv\{\x_i : i\in{m_o}\}\,:\ , ( \z-\bar{\x}_o)^t\widehat{\sigma}_o^{-1}(\z-\bar{\x}_o)\leq\frac{4(c+1)m_o}{m } \bigg\}. \end{aligned}\ ] ] to prove the second equality , let us define the matrix @xmath144 . for @xmath145 satisfying the constraints , the equality @xmath146 holds , where @xmath147 .
then , the singular value decomposition of the matrix @xmath148 and the constraint @xmath149 yield the second equality . a similar uncertainty set
is used in minimax probability machine ( mpm ) @xcite and maximum margin mpm @xcite , though the constraint , @xmath150 , is not imposed in these learning methods .
the loss function @xmath151 is used in adaboost @xcite .
the conjugate function is equal to @xmath152 hence , the corresponding uncertainty set is defined as @xmath153= \bigg\ { \sum_{i\in{m_o}}\alpha_i\x_i\,:\,\sum_{i\in{m_o}}\alpha_i=1,\ , \alpha_i\geq{0},\,i\in{m_o},\ , \sum_{i\in{m_o}}\alpha_i\log\frac{\alpha_i}{1/m_o}\leq{}c+1+\log\frac{m_o}{m}\bigg\ } \end{aligned}\ ] ] for @xmath139 . in the uncertainty set , the kullback - leibler divergence from the weight @xmath154 to the uniform weight is bounded above . in this section
, we derived parametrized uncertainty sets associated with convex loss functions .
inversely , if the uncertainty set is represented as the form of , there exists the corresponding loss function .
when we consider statistical properties of the classifier estimated based on the uncertainty set , we can study the equivalent estimator derived from the corresponding loss function .
we have many theoretical tools to analyze such estimators .
however , if the uncertainty set does not have the expression of , the corresponding loss function would not exist . in this case , we can not apply the standard theoretical tools to understand statistical properties of learning algorithms based on such uncertainty sets .
one way to remedy the drawback is to revise the uncertainty set so as to possess the corresponding loss function .
the next section is devoted to study a way of revising the uncertainty set .
given a parametrized uncertainty set , generally there does not exist the loss function which corresponds to the uncertainty set . in this section , we present a way of revising the uncertainty set such that there exists a corresponding loss function .
we consider two kinds of representations for parametrized uncertainty sets : one is vertex representation , and the other is level - set representation .
let @xmath155 and @xmath156 be index sets defined in , and we define @xmath131 and @xmath132 .
for @xmath139 , let @xmath157 be a closed , convex , proper function on @xmath158 , and @xmath159 be the conjugate function of @xmath157 .
the argument of @xmath159 is represented by @xmath145 .
vertex representation _ of the uncertainty set is defined as @xmath160= \bigg\ { \sum_{i\in{m_o}}\alpha_i\x_i\,:\ , l_o^*(\alphabold_o)\leq{c } \bigg\},\quad{}o\in\{p , n\}. \label{eqn : uncertainty - vertex - rep}\end{aligned}\ ] ] in example [ example : uncertainty_truncated_quadratic ] , the function @xmath161 is employed . on the other hand ,
let us define @xmath162 as a closed , convex , proper function , and @xmath163 be the conjugate of @xmath164 .
the _ level - set representation _ of the uncertainty set is defined by @xmath160= \bigg\ { \sum_{i\in{m_o}}\alpha_i\x_i\,:\ , h_o^*\big(\sum_{i\in{m_o}}\alpha_i\x_i\big)\leq{c } \bigg\},\quad{}o\in\{p , n\}. \label{eqn : uncertainty - levelset - rep } \ ] ] the function @xmath163 may depend on the population distribution .
we suppose that @xmath163 does not depend on the sample points , @xmath165 . in example
[ example : uncertainty_truncated_quadratic ] , the second expression of the uncertainty set involves the convex function @xmath166 .
this function does not satisfy the assumption , since @xmath163 depends on training samples via @xmath140 and @xmath141 .
instead , the function @xmath167 with the population mean @xmath168 and the population covariance matrix @xmath169 meets the condition .
when @xmath168 and @xmath169 are replaced with the estimated parameters based on a prior knowledge or a set of samples independent of the training samples , @xmath170 , the function @xmath163 with the estimated parameters still satisfies the condition we imposed above . in popular learning algorithms using uncertainty sets such as hard - margin svm , @xmath0-svm and maximum margin mpm ,
the decision function is estimated by solving the minimum distance problem with @xmath171 $ ] and @xmath172 $ ] , where @xmath173 and @xmath174 are prespecified constants . in order to investigate the statistical properties of the learning algorithm using uncertainty sets , we consider the primal expression of a variant of the minimum distance problem . in section [ sec : loss_and_uncertainty ] , we derived the problem as the dual form of . here
, we consider the following optimization problem to obtain the loss function corresponding to given uncertainty sets having the vertex representation , @xmath175\cap\conv\{\x_i : i\in{m_p}\},\\ \displaystyle \phantom{\st}\ \z_n\in\ucal_n[c_n]\cap\conv\{\x_i : i\in{m_n}\}. \end{array } \label{eqn : opt - based - on - uncertainty_set}\end{aligned}\ ] ] in the above problem the constraints , @xmath176 , are added , since the corresponding uncertainty set has the same constraint .
we derive the primal problem corresponding to via the min - max theorem .
a brief calculation yields that is equivalent to @xmath177 if there is no duality gap , the corresponding primal formulation of is given as @xmath178 where @xmath179 is defined as @xmath180 for @xmath139 . in the primal expression , @xmath181 and @xmath182
are regarded as the loss function for the decision function @xmath183 on training samples . in general , however , the loss function is not represented as the empirical mean over training samples .
thus , we can not apply the standard theoretical tools to investigate statistical properties such as bayes risk consistency for the learning algorithm based on or . on the other hand ,
if the problem is described as the empirical loss minimization , we can study statistical properties of the algorithm by applying the statistical theory developed by @xcite . to link
the uncertainty set approach with the empirical loss minimization , we consider a revision of the uncertainty set .
we propose a way of revising uncertainty sets such that the primal form is represented as minimization of the empirical mean of a loss function .
remember that the additivity of the function is kept unchanged in the conjugate function , i.e. , @xmath184 .
revision of uncertainty set defined by vertex representation : : : suppose that the uncertainty set is described by . for @xmath139 , we define @xmath185-dimensional vectors @xmath186 and @xmath187 . for the convex function
@xmath188 , we define @xmath189 by @xmath190 the revised uncertainty set @xmath191,\,o\in\{p , n\}$ ] is defined as @xmath192= \bigg\ { \sum_{i\in{m_o}}\alpha_i\x_i\,:\ , \sum_{i\in{m_o}}\alpha_i=1,\,\alpha_i\geq 0,\,i\in{m_o},\ , \frac{1}{m}\sum_{i\in{m_o}}\bar{\ell}^*(\alpha_im)\leq{c } \bigg\}. \end{aligned}\ ] ] revision of uncertainty set defined by level - set representation : : : suppose that the uncertainty set is described by and that the mean of the input vector @xmath5 conditioned on the positive ( resp .
negative ) label is given as @xmath193 .
the null vector is denoted as @xmath194 .
we define the function @xmath195 by @xmath196 the revised uncertainty set @xmath191 , o\in\{p , n\}$ ] is defined as @xmath192=\bigg\ { \sum_{i\in{m_o}}\alpha_i\x_i\,:\ , \sum_{i\in{m_o}}\alpha_i=1,\,\alpha_i\geq 0,\,i\in{m_o},\ , \frac{1}{m}\sum_{i\in{m_o}}\bar{\ell}^*(\alpha_im)\leq{c},\ , \bigg\}. \end{aligned}\ ] ] we apply the parallel shift of training samples so as to be @xmath197 or @xmath198 .
we explain the reason why the revised uncertainty set is defined as above . in the revision ,
the uncertainty set is kept unchanged , when the function @xmath199 is described in the additive form .
the precise description is presented in the following theorem .
[ theorem : conservation_law ] let @xmath200 be convex functions , and @xmath201 be the function defined by for given @xmath202 and @xmath203 .
suppose that @xmath204 is a closed , convex , proper function such that @xmath205 and @xmath206 for @xmath207 hold . 1 .
suppose that the equality @xmath208 holds for all non - negative @xmath209 .
then , the equality @xmath210 holds .
2 . suppose that the equality @xmath211 holds for all @xmath212 .
then , the equality @xmath210 holds .
we prove the first statement . from the definition of @xmath201 and the assumption on @xmath213 , the equality @xmath214 holds for @xmath207 .
suppose @xmath215 .
the assumption on @xmath202 and @xmath203 leads to @xmath216 .
hence , we have @xmath217 .
the second statement of the theorem is straightforward .
theorem [ theorem : conservation_law ] implies that the transformation of @xmath199 to @xmath218 is a projection onto the set of functions with the additive form .
in addition , the second statement of theorem [ theorem : conservation_law ] denotes that the projection is uniquely determined when we impose the condition that the values on the diagonal @xmath219 are unchanged .
next , we explain the validity of the formula .
we want to find a function @xmath220 such that @xmath221 is close to @xmath222 in some sense .
we substitute @xmath223 into @xmath224 . in the large sample limit , @xmath225 is approximated by @xmath226 .
suppose that @xmath227 is represented as @xmath228 .
then , we obtain . for the revised uncertainty sets @xmath229 $ ] and @xmath230 $ ] , the corresponding primal problem of @xmath231,\ \z_n\in\bar{\ucal}_n[c_n ] \label{eqn : revised - uncertainty - problem}\end{aligned}\ ] ] is given as @xmath232 the revision of the uncertainty sets leads to the empirical mean of the revised loss function @xmath233 .
when we study statistical properties of the estimator given by the optimal solution of , we can apply the standard theoretical tools , since the objective in the primal expression is described by the empirical mean of the revised loss functions .
we show some examples to illustrate how the revision of the uncertainty set works .
[ example : quad - uncertainty - set - to - loss ] let @xmath234 be the convex function @xmath235 , where @xmath236 is a positive definite matrix .
the revised function defined by is given as @xmath237 for @xmath212 .
then , we have @xmath238 when both @xmath239 and @xmath240 are the identity matrix , the equality @xmath241 holds .
let @xmath242 be @xmath243 .
then , the revised uncertainty set is given as @xmath244 & = \bigg\ { \sum_{i\in{m_o}}\alpha_i\x_i\ , : \sum_{i\in{m_o}}\alpha_i=1,\,\alpha_i\geq0\,(i\in{m_o } ) , \sum_{i\in{m_o}}\alpha_i^2 \leq \frac{cm}{k } \bigg\}. \end{aligned}\ ] ] for @xmath139 , let @xmath140 and @xmath141 be the empirical mean and the empirical covariance matrix , @xmath245 if @xmath141 is invertible , we have @xmath246 = \bigg\ { \z\in\conv\{\x_i : i\in{m_o}\}\,:\ , ( \z-\bar{\x}_o)^t\widehat{\sigma}_o^{-1}(\z-\bar{\x}_o)\leq \frac{c{}mm_o}{k } \bigg\}. \end{aligned}\ ] ] in the learning algorithm based on the revised uncertainty set , the estimator is obtained by solving @xmath247,\,\z_n\in\bar{\ucal}_n[c_n ] \\ & \longleftrightarrow \min_{c_p , c_n,\z_p,\z_n } c_p+c_n+\frac{m^2\lambda}{4k}\|\z_p-\z_n\|\ \
\st\ \z_p\in\bar{\ucal}_p\bigg[\frac{4c_p k}{m^2}\bigg],\ , \z_n\in\bar{\ucal}_n\bigg[\frac{4c_n k}{m^2}\bigg ] .
\end{aligned}\ ] ] the corresponding primal expression is given as @xmath248 [ example : quad - levelset - uncertainty - set - to - loss ] we define @xmath249 for @xmath139 by @xmath250 where @xmath168 is the mean vector of the input vector @xmath5 conditioned on each label and @xmath236 is a positive definite matrix . in practice ,
the mean vector is estimated by using a prior knowledge which is independent of the training samples @xmath251 .
suppose that @xmath252 .
then , for @xmath212 , the revision of leads to @xmath253 where @xmath254 and @xmath255 are constant numbers .
thus , we have @xmath246 & = \bigg\ { \sum_{i\in{m_o}}\alpha_i\x_i \,:\ , \sum_{i\in{m_o}}\alpha_i=1,\,\alpha_i\geq0\ , ( i\in{m_o}),\ , \sum_{i\in{m_o}}\alpha_i^2\leq \frac{c - b_1}{mb_2 } \bigg\}\\ & = \bigg\ { \z\in\conv\{\x_i : i\in{m_o}\ } \,:\ , ( \z-\bar{\x}_o)^t\widehat{\sigma}_o^{-1}(\z-\bar{\x}_o)\leq m_o\cdot\frac{c - b_1}{mb_2 } \bigg\ } , \end{aligned}\ ] ] where @xmath140 and @xmath141 are the estimators of the mean vector and the covariance matrix based on training samples @xmath170 .
the corresponding loss function is obtained in the same way as example [ example : quad - uncertainty - set - to - loss ] . figure [ fig : revised - ellipsoidal_uncertainty - set ]
illustrates an example of the revision of the uncertainty set . in the left panel
, the uncertainty set does not match the distribution of the training samples .
the revised uncertainty set in the right panel seems to well approximate the dispersal of the training samples . [ cols="^,^ " , ]
in this paper , we studied the relation between the loss function approach and the uncertainty set approach in binary classification problems .
we showed that these two approaches are connected to each other by the conjugate property based on the legendre transformation .
given a loss function , there exists a corresponding parametrized uncertainty set . in general , however , uncertainty set does not correspond to the empirical loss function .
we presented a way of revising the uncertainty set such that there exists an empirical loss function .
then , we proposed a modified maximum - margin algorithm based on the parametrized uncertainty set .
we proved the statistical consistency of the learning algorithm .
numerical experiments showed that the revision of the uncertainty set often improves the prediction accuracy of the classifier . in our proof of the statistical consistency ,
the hinge loss used in @xmath0-svm is excluded .
@xcite proved the statistical consistency of @xmath0-svm with a nice choice of the regularization parameter .
we are currently investigating the relaxation of the assumptions of our theoretical result so as to include the hinge loss function and other popular loss functions such as the logistic loss .
as for the statistical modeling , the relation between the loss function approach and the uncertainty set approach can be a useful tool . in optimization and control theory ,
the modeling based on the uncertainty set is frequently applied to the real - world data ; see the modeling in robust optimization and related works @xcite .
we believe that the learning algorithm with the revision of the uncertainty set can bridge a gap between statistical modeling based on some intuition and nice statistical properties of the estimated classifiers .
tk was partially supported by grant - in - aid for young scientists ( 20700251 ) . at
was partially supported by grant - in - aid for young scientists ( 23710174 ) .
ts was partially supported by mext kakenhi 22700289 and the aihara project , the first program from jsps , initiated by cstp .
first , we prove the existence of an optimal solution . according to the standard argument on the kernel estimator
, we can restrict the function part @xmath32 to be the form of @xmath256 then , the problem is reduced to the finite - dimensional problem , @xmath257 let @xmath258 be the objective function of .
let us define @xmath259 be the linear subspace in @xmath260 spanned by the column vectors of the gram matrix @xmath261 .
we can impose the constraint @xmath262 , since the orthogonal complement of @xmath259 does not affect the objective and the constraint in .
we see that assumption [ assump : universal_kernel ] and the reproducing property yield the inequality @xmath263 . due to this inequality and the assumptions on the function @xmath94
, the objective function @xmath258 is bounded below by @xmath264 hence , for any real number @xmath57 , the inclusion relation @xmath265 holds .
note that the vector @xmath266 satisfying @xmath267 and @xmath268 is restricted to a compact subset in @xmath260 .
we shall prove that the subset is compact , if they are not empty .
we see that the two sets above are closed subsets , since both @xmath269 and @xmath270 are continuous . by the variable change from @xmath271 to @xmath272 ,
@xmath273 is transformed to the convex function @xmath274 defined by @xmath275 the subgradient of @xmath102 diverges to infinity , when @xmath276 tends to infinity .
in addition , @xmath102 is a non - decreasing and non - negative function .
then , we have @xmath277 the same limit holds for @xmath278 .
hence , the level set of @xmath274 is closed and bounded , i.e. , compact . as a result ,
the level set of @xmath273 is also compact .
therefore , the subset is also compact in @xmath279 .
this implies that has an optimal solution .
next , we prove the duality between and .
since has an optimal solution , the problem with the slack variables @xmath280 , @xmath281 also has an optimal solution and the finite optimal value .
in addition , the above problem clearly satisfies the slater condition ( * ? ? ? * assumption 6.2.4 ) . indeed , at the feasible solution , @xmath282 and @xmath283 , the constraint inequalities are all inactive for positive @xmath91 .
hence , proposition 6.4.3 in @xcite ensures that the min - max theorem holds , i.e. , there is no duality gap .
then , in the same way as , we obtain with the uncertainty set as the dual problem of .
we show proofs of lemmas in section [ subsec : convergence+to+optimal_expected_loss ] .
let @xmath284 be the subset @xmath285 , then we have @xmath286 .
due to the non - negativity of the loss function @xmath94 , we have @xmath287 for given @xmath288 satisfying @xmath289 , we define the function @xmath290 by @xmath291 we derive a lower bound @xmath292 . since @xmath102 is a finite - valued convex function on @xmath293 , the subdifferential @xmath294 is given as @xmath295 formulas of the subdifferential are presented in theorem 23.8 and theorem 23.9 of @xcite .
we prove that there exist @xmath296 and @xmath297 such that @xmath298 holds .
since the second condition in assumption [ assumption : expectedloss_consistency ] holds for the convex function @xmath94 , the union @xmath299 includes all the positive real numbers .
hence , there exist @xmath300 and @xmath301 satisfying @xmath302 and @xmath303 .
then , for @xmath304 , the null vector is an element of @xmath305 .
since @xmath290 is convex in @xmath306 , the minimum value of @xmath290 is attained at @xmath307 .
define @xmath308 as a real number satisfying @xmath309 since @xmath289 is assumed , both @xmath300 and @xmath301 are less than @xmath308 due to the monotonicity of the subdifferential .
then , the inequality @xmath310 holds for all @xmath311 and all @xmath288 such that @xmath312 .
hence , for any measurable function @xmath19 and @xmath313 , we have @xmath314 as a result , we have @xmath315 .
corollary 5.29 of @xcite ensures that the equality @xmath316:f\in\hcal\}=\inf\{\ebb[\ell(\rho - yf(x))]:f\in{}l_0\ } \end{aligned}\ ] ] holds for any @xmath317 .
thus , we have @xmath318 for any @xmath317
. then , the equality @xmath319 holds . under assumption
[ assump : non - deterministic - assumption ] and assumption [ assumption : expectedloss_consistency ] , we have @xmath320 due to lemma [ lemma : risk_boundedness ] .
then , for any @xmath321 , there exist @xmath322 and @xmath323 such that @xmath324 and @xmath325 hold . for all @xmath326 we have @xmath327 on the other hand
, it is clear that the inequality @xmath328 holds .
hence , eq . holds . under assumption
[ assump : non - deterministic - assumption ] , the label probabilities , @xmath329 and @xmath330 , are positive .
we assume that the inequalities @xmath331 hold . applying chernoff bound
, we see that there exists a positive constant @xmath332 depending only on the marginal probability of the label such that holds with the probability higher than @xmath333 .
lemma [ lemma : existence_opt_sol ] ensures that the problem has optimal solutions @xmath334 .
the first inequality in , i.e. , @xmath335 , is clearly satisfied .
then , we have @xmath336 from the reproducing property of the rkhss . the definition of the estimator and the non - negativity of @xmath94 yield that @xmath337
then , we have @xmath338 next , we consider the optimality condition of @xmath339 . according to the calculus of subdifferential introduced in section 23 of @xcite , the derivative of the objective function with respect to @xmath340 leads to an optimality condition , @xmath341 the monotonicity and non - negativity of the subdifferential and the bound of @xmath20 lead to @xmath342 the above expression means that there exist numbers in the subdifferential such that the inequality holds , where @xmath343 denotes the @xmath129-fold sum of the set @xmath344 . let @xmath345 be a real number satisfying @xmath346 , i.e. , all elements in @xmath347 are greater than @xmath348 .
then , @xmath349 should be less than @xmath345 . in the same way , for @xmath350 satisfying @xmath351
, we have @xmath352 .
the existence of @xmath345 and @xmath350 is guaranteed by assumption [ assumption : expectedloss_consistency ] .
hence , the inequalities @xmath353 hold , in which @xmath354 is used in the second inequality .
define @xmath355 as a real number such that @xmath356 inequalities in lead to @xmath357 hence , we can choose @xmath355 satisfying @xmath358 .
suppose that @xmath359 holds for @xmath360 .
then , the inequalities @xmath361 hold with the probability higher than @xmath333 for @xmath360 . by choosing an appropriate positive constant @xmath362
, we obtain .
since @xmath363 holds for @xmath23 such that @xmath364 , we have the following inequality @xmath365 in the same way as the proof of lemma 3.4 in @xcite , hoeffding s inequality leads to the upper bound .
is the direct conclusion of and .
lemma [ lemma : convergence_regularized_exp_loss ] assures that , for any @xmath366 , there exists sufficiently large @xmath367 such that @xmath368 holds for all @xmath369 .
thus , there exist @xmath370 and @xmath371 such that @xmath372 and @xmath373 hold for @xmath374 . due to the law of large numbers
, the inequality @xmath375 holds with high probability , say @xmath376 , for @xmath377 . the boundedness property in lemma [ lemma : estimator_bound ] leads to @xmath378 for @xmath379 .
in addition , by the uniform bound shown in lemma [ lemma : uniform_convergence_loss ] , the inequality @xmath380 holds with probability @xmath381 .
hence , the probability such that the inequality @xmath382 holds is higher than @xmath383 for @xmath379 .
let @xmath384 be @xmath385 .
then , for any @xmath366 , the following inequalities hold with probability higher than @xmath386 for @xmath387 , @xmath388 the second inequality above is given as @xmath389 for a fixed @xmath340 such that @xmath390 , the loss function @xmath391 is classification - calibrated @xcite , since @xmath392 holds .
hence @xmath393 in assumption [ assumption : loss_bayesrisk_consistency ] satisfies @xmath394 , @xmath395 for @xmath396 , and @xmath393 is continuous and strictly increasing in @xmath397 $ ] .
in addition , for all @xmath23 and @xmath65 , the inequality @xmath398-
\inf_{f\in\hcal , b\in\rbb}\ebb[\ell(\rho - y(f(x)+b ) ) ] \end{aligned}\ ] ] holds .
details are presented in theorem 1 and theorem 2 of @xcite .
here we used the equality @xmath399:f\in\hcal , b\in\rbb\ } = \inf\{\ebb[\ell(\rho - y(f(x)+b))]:f\in{}l_0,b\in\rbb\ } , \end{aligned}\ ] ] which is shown in corollary 5.29 of @xcite .
hence , we have @xmath400- \inf_{f\in\hcal , b\in\rbb}\ebb[\ell(\widehat{\rho}-y(f(x)+b))]\\ & = \rcal(\widehat{f}+\widehat{b},\widehat{\rho } ) - \inf_{f\in\hcal , b\in\rbb}\rcal(f+b,\widehat{\rho } ) , \end{aligned}\ ] ] since @xmath401 holds due to .
we assumed that @xmath402 converges to @xmath403 in probability .
then , for any @xmath321 , the inequality @xmath404 holds with high probability for sufficiently large @xmath405 .
thus , @xmath406 converges to zero in probability .
the inequality @xmath407 and the assumption on the function @xmath408 ensure that @xmath409 converges to @xmath38 in probability , when @xmath405 tends to infinity . as a result , for any @xmath366 , @xmath410 holds with probability higher than @xmath411 with respect to the probability distribution of @xmath412 , where @xmath413 satisfies @xmath414 for any @xmath366 .
next , we study the relation between @xmath415 and @xmath416
. the sample size of @xmath417 is @xmath418 . for any fixed @xmath23
, we define the set of 0 - 1 valued functions , @xmath419 .
the vc - dimension of @xmath420 equals to one . indeed , for two distinct points @xmath421 such that @xmath422 , the event such that @xmath423 and @xmath424 is impossible .
hence , for any @xmath321 and any @xmath23 , the inequality @xmath425 holds with probability higher than @xmath426 with respect to the joint probability of training sample @xmath417 .
note that @xmath427 depends only on @xmath418 , @xmath428 and the vc - dimension of @xmath420 .
thus , @xmath429 is independent of the choice of @xmath23 .
remember that @xmath415 depends only on the data set @xmath412 .
due to the law of large numbers , the inequality @xmath430 holds with probability higher than @xmath431 with respect to the probability distribution of @xmath417 conditioned on @xmath412 .
since the 0 - 1 loss is bounded , it is possible to choose @xmath432 independent of @xmath433 . from the uniform convergence property
, the following inequality also holds @xmath434 with probability higher than @xmath426 with respect to the probability distribution of @xmath417 conditioned on the observation of @xmath412 .
in addition , we have @xmath435 given the training samples @xmath412 satisfying , the inequalities @xmath436 hold with probability higher than @xmath437 with respect to the probability distribution of @xmath417 conditioned on the observation of @xmath412 .
hence , as for the conditional probability , we have @xmath438 remember that @xmath432 and @xmath427 do not depend on @xmath412 .
hence , as for the joint probability of @xmath412 and @xmath417 , we have @xmath439 the above inequality implies that @xmath440 converges to @xmath38 in probability , when @xmath405 and @xmath418 tend to infinity .
for @xmath441 and @xmath442 , we can directly confirm that the lemma holds . in the following , we assume @xmath443 and @xmath390 .
we consider the following optimization problem involved in @xmath393 , @xmath444 the objective function is a finite - valued convex function on @xmath293 , and diverges to infinity when @xmath276 tends to @xmath445 .
hence , there exists an optimal solution .
let @xmath446 be an optimal solution of .
the optimality condition is given as @xmath447 we assumed that both @xmath448 and @xmath449 are positive and that @xmath450 holds .
hence , both @xmath451 and @xmath452 should not be zero .
indeed , if one of them is equal to zero , the other is also zero .
hence , we have @xmath453 and @xmath454 .
these inequalities contradict @xmath455 .
then , we have @xmath456 and @xmath457 , i.e. , @xmath458 . in addition , we have @xmath459 since @xmath460 holds on @xmath461 , the second derivative of the objective in satisfies the positivity condition , @xmath462 for all @xmath276 such that @xmath463 and @xmath464 .
therefore , @xmath465 is uniquely determined . for a fixed @xmath466
, the optimal solution can be described as the function of @xmath340 , i.e. , @xmath467 . by the implicit function theorem , @xmath468 is continuously differentiable with respect to @xmath340
. then , the derivative of @xmath393 is given as @xmath469 the convexity of @xmath470 for @xmath471 leads to @xmath472 hence , we have @xmath473 for @xmath450 and @xmath443 . as a result
, we see that @xmath393 is non - decreasing as the function of @xmath340 .
we use the result of @xcite . for a fixed @xmath340 ,
the function @xmath474 is continuous for @xmath475 , and the convexity of @xmath94 leads to the non - negativity of @xmath474 .
moreover , the convexity and the non - negativity of @xmath102 lead to @xmath476 for @xmath477 and @xmath390 , where @xmath478 and @xmath479 are positive for @xmath480 .
the above inequality and the continuity of @xmath481 ensure that there exists @xmath276 satisfying @xmath482 for all @xmath483 such that @xmath484 .
we define the inverse function @xmath485 by @xmath486 for @xmath484 . for a fixed @xmath390 ,
the loss function @xmath391 is classification - calibrated @xcite .
hence , lemma 3 in @xcite leads to the inequality @xmath487 for @xmath484 .
define @xmath488 by @xmath489 from the definition of @xmath490 , @xmath491 is well - defined for all @xmath492 . since @xmath493 holds , we have @xmath494 .
in addition , @xmath479 is non - decreasing as the function of @xmath340 .
thus , we have @xmath495 for all @xmath390 and @xmath496 .
then , we can choose @xmath497 it is straightforward to confirm that the conditions of assumption [ assumption : loss_bayesrisk_consistency ] are satisfied . d. j. crisp and c. j. c. burges . a geometric interpretation of @xmath0-svm classifiers . in s.
a. solla , t. k. leen , and k .-
mller , editors , _ advances in neural information processing systems 12 _ , pages 244250 .
mit press , 2000 .
j. s. nath and c. bhattacharyya .
maximum margin classifiers with specified false positive and false negative error rates . in c.
apte , b. liu , s. parthasarathy , and d. skillicorn , editors , _ proceedings of the seventh siam international conference on data mining _ , pages 3546 .
siam , 2007 . | in binary classification problems , mainly two approaches have been proposed ; one is loss function approach and the other is uncertainty set approach . the loss function approach is applied to major learning algorithms such as support vector machine ( svm ) and boosting methods .
the loss function represents the penalty of the decision function on the training samples . in the learning algorithm , the empirical mean of the loss function
is minimized to obtain the classifier . against a backdrop of the development of mathematical programming , nowadays learning algorithms based on loss functions
are widely applied to real - world data analysis . in addition
, statistical properties of such learning algorithms are well - understood based on a lots of theoretical works .
on the other hand , the learning method using the so - called uncertainty set is used in hard - margin svm , mini - max probability machine ( mpm ) and maximum margin mpm . in the learning algorithm , firstly , the uncertainty set is defined for each binary label based on the training samples .
then , the best separating hyperplane between the two uncertainty sets is employed as the decision function .
this is regarded as an extension of the maximum - margin approach .
the uncertainty set approach has been studied as an application of robust optimization in the field of mathematical programming .
the statistical properties of learning algorithms with uncertainty sets have not been intensively studied . in this paper
, we consider the relation between the above two approaches .
we point out that the uncertainty set is described by using the level set of the conjugate of the loss function . based on such relation , we study statistical properties of learning algorithms using uncertainty sets . |
about twenty five years ago the possibility of neutrinos interacting by exchange of very light neutral scalars to produce interesting cosmological and astrophysical effects was considered by kawasaki , murayama and yanagida @xcite , and malaney , starkman and tremaine@xcite .
three of us later investigated a similar system , in which the scalar particle has a mass of order 1/a.u . , which leads to neutrino clustering@xcite .
following the observation of the re - acceleration of the expansion of the universe @xcite , which led to the concept of dark energy ( for a recent review see ref .
fardon , nelson and weiner @xcite noted that the inferred energy density of dark energy , of order ( 2.4 mev)@xmath10 @xcite , was , remarkably , of the order of the experimentally inferred value of neutrino masses .
they also utilized the concept of a scalar field interacting with the neutrinos to obtain the desired effect .
it is worth emphasizing that negative pressure is not a strange feature in physics , in that any self bound system has an equilibrium density , and will have a higher energy per particle as the density is decreased , and thus a negative pressure in this region . because this feature exists in our neutrino clustering model , we realized that it could produce negative values of @xmath0 and was thus relevant to the dark energy problem .
we provided some early comments @xcite , and then showed how it led to a system with @xmath3 in a restricted range of the development of the universe @xcite , with @xmath11 at very early times , and @xmath12 near the present . in this paper
we will be elaborating our model of dark energy . to obtain the energy density of the neutral fermion
scalar field system as a function of @xmath2 one needs to make assumptions about parameters .
typically @xmath0 has a minimum as a function of @xmath2 , and the position of the minimum may be moved by a choice of parameters , but a rapid variation of @xmath0 between the minimum and the present is characteristic of our model .
this lends support to attempts to probe the @xmath2 dependence of dark energy in more detail .
it is worth noting that mota et al@xcite have recently introduced the possibility of neutrino clustering in the neutrino scalar field model to obtain a dependence of the dark energy density on @xmath2 .
our work shows that a variation of @xmath0 and the dark energy density @xmath13 with epoch can be obtained in a homogeneous model . after outlining the model
, we display our results for @xmath0 as a function of the density of the neutral fermions , and then discuss how to choose the parameters of the model to obtain results for @xmath0 as a function of @xmath2 , and discuss the implications of the results .
following ref .
@xcite , the equation of motion for a neutral fermion field , @xmath14 , interacting with a scalar field , @xmath15 , is @xmath16\phi & = & g\overline{\psi}\psi \label{phi}\\ \left[i{/\!\!\!\!\!\partial } - m_n^{(0)}\right]\psi & = & -g \phi \psi,\label{psi}\end{aligned}\ ] ] with @xmath17 .
the nonlinear scalar selfcouplings are omitted here , even though they are required to exist by field theoretic selfconsistency @xcite , as they may consistently be assumed to be sufficiently weak as to be totally irrelevant .
the parameter @xmath18 is the renormalized vacuum mass of the isolated neutral fermion , and takes into account the contributions from the electroweak theory , as well as contributions from the vacuum expectation value of the new scalar field @xmath15 .
we look for solutions of these equations in infinite matter which are static and translationally invariant . in the freidman equation description of the evolution of the universe we use the adiabatic approximation , assuming that the scalar field expectation value takes the value determined by the fermion density at the appropriate time
. equations ( [ phi ] , [ psi ] ) then lead to an effective mass for the neutral fermion of @xmath19 these equations are simply the equations of quantum hadrodynamics @xcite . as in @xcite , we act with these operator equations on state which is a filled fermi sea of neutral fermions , with a fermion number density @xmath20 and fermi momentum @xmath21 , related , for majorana particles ( which we assume henceforth ) , by @xmath22 .
we introduce the parameter @xmath23 and the variables @xmath24 , @xmath25 , @xmath26 , and @xmath27 .
equation ( [ eq : effm ] ) becomes an equation for @xmath28 : @xmath29,\label{eq : yx}\ ] ] the total energy of the system , @xmath30 , with @xmath31 the number of fermions , and @xmath32 .
\label{epf}\ ] ] as @xmath33 goes from @xmath34 to @xmath35 , @xmath28 varies monotonically from @xmath36 to @xmath34 , behaving as @xmath37 for small @xmath33 , and @xmath38 for large @xmath33 . for the fermion system to be bound ,
the minimum of @xmath39 must be less than 1 .
we find that for sufficiently large @xmath40 the neutral fermion scalar field system is bound @xcite , and the total energy displays the characteristic behaviour of a self - bound system noted above , suggesting that we should see negative pressures in this system .
we now go on to compute the equation of state parameter .
the equation of state , or the equation for the pressure , @xmath41 , as a function of the other variables of the system , is represented through the equation of state parameter , @xmath0 , which is defined by the proportionality between @xmath41 and @xmath13 , the energy density @xmath42 as is well known , for radiation , @xmath43 , for cold non - interacting matter , @xmath44 , and , for a cosmological constant , @xmath45 .
any @xmath46 describes a system with negative pressure . using the equations of the friedman - lematre - robertson - walker ( flrw ) metric , with a scale parameter @xmath47 , the equation of state parameter
is shown to satisfy @xmath48 in these equations @xmath13 is the _ total _ energy density .
assuming for this paper that the only source of this energy is the coupled fermion - scalar system , then @xmath49 and we obtain @xmath50 this is the basic equation for computing @xmath0 in our model , and it gives @xmath51 it follows immediately that @xmath52 in our model . vs. log(@xmath33 ) for 8 values of @xmath40.,scaledwidth=75.0% ] from equations ( [ eq : yx],[eq : w ] ) , we can compute @xmath0 as a function of @xmath33 , given @xmath40 .
figure [ fig : wvslogxf ] shows some results using a log scale for @xmath33 .
note that @xmath0 approaches close to @xmath1 as the density decreases ( remember the direction of decreasing density is the direction of expansion of the universe ) and then departs sharply towards zero , at large @xmath33 , it is clear that @xmath0 approaches @xmath53 as it should for a relativistic gas of fermions .
the value goes to zero at zero density . for large values of @xmath40
the minimum value of @xmath0 is very close to @xmath1 and is reached near the value @xmath54 which can be obtained from an analytic approximation . more precisely , and empirically @xmath55 . . the minimum value of @xmath0 , the point @xmath56 at which this is achieved , and the approximate location @xmath57 [ cols="^,^,^,^",options="header " , ] the minimum value of @xmath0 approaches @xmath1 more closely as @xmath40 increases , and the low density recovery of @xmath0 to zero becomes steeper as @xmath40 increases .
it is clear from these results that we have a model in which the value of @xmath0 is a function of the scale parameter @xmath47 or the red - shift @xmath2 , i.e. that we have an epoch dependent dark energy .
but these results are in terms of dimensionless parameters and variables . to connect to the real world we must link these dimensionless numbers to the dimensionful numbers which characterize it . as was shown in ref .
( @xcite ) , there are only a few , weak constraints on the actual parameter values . moreover ,
very large values of @xmath40 are possible even for very small values of @xmath58 if the range of the scalar is very large , corresponding to very small values of @xmath59 . even if long - ranged , such weak interactions between fermions , especially neutrinos or those outside of the standard model altogether ( such as the lsp ) are exceptionally difficult to constrain by any laboratory experiments .
the energy density of dark energy is quoted as @xmath60 @xcite if we define @xmath61 , then @xmath62 mev .
it is important to remember that the conventional value of the dark energy density is derived on the assumption that it is constant during the evolution of the universe , and this is not the case in our model .
pending a study of the development of the universe using our model we use the present estimate of @xmath13 and apply it to the region @xmath63 to estimate the relevant parameters in our model .
@xmath64 to use this equation to estimate @xmath65 , first note that near @xmath3 , for large @xmath40 , @xmath33 is small and we can use the appropriate approximations to get @xmath66 where , in the last equation , we use @xmath67 at the minimum .
@xmath40 determines the relationship between @xmath65 and @xmath68 , @xmath69^{1/4 } m_e$ ] .
we choose @xmath40 , taking care that the implied values of @xmath58 and @xmath59 are reasonable . for @xmath70 , @xmath71 mev , in the neutrino range , and for @xmath72 , @xmath73 gev , in the range expected for neutralinos .
what are the implications for the other parameters ?
one can immediately estimate the density of the neutral fermions as @xmath74 @xmath75 and @xmath76 @xmath75 for these two cases .
because we are assuming that the neutral fermions form a homogeneous background , it is appropriate to set the range of the scalar field to be the scale parameter of the universe at the relevant time , which we will take to be @xmath77 , _ i.e. _ @xmath78 with this value of @xmath59 we can obtain the implied value of @xmath79 : @xmath80 and @xmath81 in these two cases .
even the largest coupling is far too weak to be constrained by terrestrial experiments .
an alternative way to proceed would be to assume a present density of the neutral fermions , @xmath82 , and thus a present value of the fermi momentum , @xmath83 , at a scale parameter @xmath84 , use @xmath85 and @xmath86 .
equation ( [ e1 ] ) then gives @xmath87 to give an explicit example , assume that the present density of neutral fermions is 100 @xmath75 , characteristic of neutrinos , and extrapolate this to a density of 800 @xmath75 at @xmath77 .
this gives a value @xmath88ev , and @xmath89 , @xmath90 , and the even smaller coupling constant , @xmath91 . at this value of @xmath40 ,
the minimum value of @xmath0 is @xmath92 , at the @xmath93 level from the value @xmath94 of the essence supernova survey@xcite , so even these parameters are not excluded .
it is impossible to make further progress without finding additional ways to constrain the parameters . given their extreme values the most promising approach is to use additional theoretical input , which we leave as a challenge for future work , by us and others .
now that we have some parameters we can convert the determination of @xmath0 as a function of the dimensionless parameter @xmath33 to a dependence of @xmath0 on the red shift @xmath2 . to illustrate the results we have selected the values @xmath95 and @xmath96 for @xmath40 , and set the minimum of @xmath0 to occur at @xmath4 .
these results are illustrated in figure [ fig : wvz6 ] .
vs. @xmath2 for @xmath70 , @xmath97 and @xmath98.,scaledwidth=75.0% ] note that there is a slow variation of @xmath0 with @xmath2 at epochs earlier than @xmath99 at which the minimum occurs , but a rapid change for @xmath2 between the present epoch ( @xmath6 ) and that of the minimum . with the present uncertainty in data we can choose to place
the value of @xmath99 at any value of @xmath2 between , say @xmath100 and @xmath101 without being in conflict with the present data , and have chosen to show the results at @xmath102 as illustrative results .
in an ideal future world one could imagine that @xmath103 is indeed observed to have a minimum value at some @xmath99 , and the the value of @xmath104 , close to but greater than @xmath1 is accurately known .
the fact that the minimum value of @xmath0 is near @xmath1 shows that @xmath105 , and the actual value of @xmath104 , if known accurately enough , would determine the value of @xmath40 . knowing the value of @xmath40 allows us to predict a value of @xmath13 at the minimum @xmath0 , which provides a good test on the model
. then a selection of parameters can place that minimum near an appropriate @xmath2 value , @xmath99 .
our prediction of a characteristic @xmath2 dependence of @xmath0 and @xmath13 leaves us vulnerable to developments in the precision of measurements and the extension of observations to larger and ( with more difficultly ) to smaller @xmath2 , and we await with interest the results of the proposed experiments , such as those described in ref .
this work was carried out in part under the auspices of the national nuclear security administration of the u.s .
department of energy at los alamos national laboratory under contract no .
de - ac52 - 06na25396 and supported in part by the australian research council .
m. kawasaki , h. murayama and t. yanagida , mod .
lett . a * 7 * , 563 ( 1992 ) .
r. a. malaney , g. d. starkman and s. tremaine , phys . rev .
d * 51 * , 324 ( 1995 ) .
g. j. stephenson , jr .
, j. t. goldman and b. h. j. mckellar , int .
j. mod .
a * 13 * , 2765 ( 1998 ) [ arxiv : hep - ph/9603392 ] .
a. g. riess _ et al .
_ [ supernova search team collaboration ] , astron .
j. * 116 * , 1009 ( 1998 ) [ arxiv : astro - ph/9805201 ] . s. perlmutter _ et al .
_ [ supernova cosmology project collaboration ] , astrophys .
j. * 517 * , 565 ( 1999 ) [ arxiv : astro - ph/9812133 ] .
j. frieman , m. turner and d. huterer , ann .
astrophys .
* 46 * , 385 ( 2008 ) [ arxiv:0803.0982 [ astro - ph ] ] .
r. fardon , a. e. nelson and n. weiner , jcap * 0410 * , 005 ( 2004 ) [ arxiv : astro - ph/0309800 ] . b. h. j. mckellar , m. garbutt , j. t. goldman and g. j. stephenson , jr .
a * 19 * , 1155 ( 2004 ) .
t. goldman , g. j. .
stephenson , jr .
, p. m. alsing and b. h. j. mckellar , arxiv:0905.4308 [ hep - ph ] .
d. f. mota , v. pettorino , g. robbers and c. wetterich , phys .
b * 663 * , 160 ( 2008 ) [ arxiv:0802.1515 [ astro - ph ] ] . v. pettorino , d. f. mota , g. robbers and c. wetterich , aip conf . proc . * 1115 * , 291 ( 2009 ) [ arxiv:0901.1239 [ astro - ph ] ] .
b. d. serot and j. d. walecka , adv .
* 16 * , 1 ( 1986 ) .
w. m. wood - vasey _ et al .
_ [ essence collaboration ] , astrophys .
j. * 666 * , 694 ( 2007 ) [ arxiv : astro - ph/0701041 ] . | we present a model in which the parameter @xmath0 approaches @xmath1 near a particular value of @xmath2 , and has significant negative values in a restricted range of @xmath2 .
for example , one can have @xmath3 near @xmath4 , and @xmath5 from @xmath6 to @xmath7 , and for @xmath8 .
the ingredients of the model are neutral fermions ( which may be neutrinos , neutralinos , etc ) which are very weakly coupled to a light scalar field .
this model emphasises the importance of the proposed studies of the properties of dark energy into the region @xmath9 .
la - ur-09 - 05030 address = school of physics , university of melbourne , victoria 3010 , australia .
address = theoretical division , los alamos national laboratory , + los alamos , new mexico , 87545 , usa address = dept . of physics & astronomy , university of new mexico , + albuquerque ,
new mexico 87131 , usa address = dept . of physics & astronomy , university of new mexico , + albuquerque ,
new mexico 87131 , usa |
in this article , we relate two problems , one from classical number theory , and one from lattice point enumeration in convex bodies . motivated by a conjecture of ehrhart @xcite and a result of xu @xcite , we study linear systems of diophantine equations with a single parameter . to be more precise ,
we suppose that the coefficients of our system are given by polynomial functions in a variable @xmath2 , and also that the number of solutions @xmath4 in nonnegative integers for any given value of @xmath2 is finite .
we are interested in the behavior of the function @xmath4 , and in particular , we prove that @xmath4 is * eventually a quasi polynomial * , i.e. , there exists some period @xmath5 and polynomials @xmath6 for @xmath7 such that for @xmath8 , the number of solutions for @xmath9 is given by @xmath10 .
the other side of our problem can be stated in a similar fashion : suppose that @xmath3 is a convex polytope whose vertices are given by rational functions in @xmath2 .
then the number of integer points inside of @xmath3 , as a function of @xmath2 , enjoys the same properties as that of @xmath11 as above .
we now describe in more detail some examples and the statements of our results . as a warmup to our result ,
we begin with two examples .
the first is a result of popoviciu .
let @xmath12 and @xmath13 be relatively prime positive integers .
we wish to find a formula for the number of nonnegative integer solutions @xmath14 to the equation @xmath15 . for a real number @xmath16 , let @xmath17 denote the greatest integer less than or equal to @xmath16 , and define @xmath18 to be the fractional part of @xmath16
. then the number of such solutions is given by the formula @xmath19 where @xmath20 and @xmath21 satisfy @xmath22 and @xmath23 .
see ( * ? ? ?
* chapter 1 ) for a proof .
in particular , this function is a quasi - polynomial in @xmath2 . for the second example which is a generalization of the first example , consider the number of solutions @xmath24 to the matrix equation @xmath25 where the @xmath26 and @xmath27 are fixed positive integers and @xmath28 for @xmath29 .
write @xmath30 .
we assume that @xmath31 , so that there exist integers ( not unique ) @xmath32 such that @xmath33 now define two regions @xmath34 for @xmath29 .
then if @xmath35 is in the positive span of the columns of the matrix in , there exist popoviciu - like formulas for the number of solutions of which depend only on whether @xmath36 or @xmath37 , and the numbers @xmath38 .
see section [ section : example1 ] for the precise statement . in particular , one can replace the @xmath26 , @xmath27 , and @xmath39 by polynomials in @xmath2 in such a way that for all values of @xmath2 , the condition @xmath31 holds .
for a concrete example , consider the system @xmath40 then for @xmath41 , we have that @xmath42 so that for these values of @xmath2 , there exists a quasi - polynomial that counts the number of solutions @xmath43 .
given these examples , we are ready to state our general theorem .
we denote by @xmath44 the set of functions @xmath45 which are eventually quasi - polynomial .
[ ehrhartconj ] let @xmath46 be an @xmath47 matrix , and @xmath48 be a column vector of length @xmath49 , such that their entries are integer coefficient polynomials in @xmath2 . if @xmath4 denotes the number of nonnegative integer vectors @xmath16 satisfying @xmath50 ( assuming that these values are finite ) , then @xmath51 .
this theorem generalizes the conjecture ( * ? ? ?
* exercise 4.12 ) .
see @xcite for some verified cases of a conjectural multivariable analogue , which we state here .
let @xmath52 be some subset .
we say that a function @xmath53 is a * multivariate quasi - polynomial * if there exists a finite index sublattice @xmath54 such that @xmath11 is a polynomial function on each coset of @xmath55 intersected with @xmath56 .
let @xmath57 be an @xmath58 matrix and @xmath59 be a column vector of length @xmath49 , such that all entries are linear functions in @xmath60 with integer coefficients such that for all @xmath60 , the number of nonnegative integer solutions @xmath16 to @xmath61 is finite .
then there exist finitely many polyhedral regions @xmath62 covering @xmath63 such that @xmath11 is a multivariate quasi - polynomial when restricted to each @xmath64 .
we first recall a classical theorem due to pick .
let @xmath65 be a convex polygon with integral vertices .
if @xmath66 , @xmath67 , and @xmath68 denote the area of @xmath0 , the number of integer points in the interior of @xmath0 , and the number of integer points on the boundary of @xmath0 , respectively , then one has the equation @xmath69 now let us examine what happens with dilates of @xmath0 : define @xmath70 .
then of course @xmath71 and @xmath72 whenever @xmath2 is a positive integer , so we can write @xmath73 or equivalently , @xmath74 which is a polynomial in @xmath2 .
the following theorem of ehrhart says that this is always the case independent of the dimension , and we can even relax the integral vertex condition to rational vertices : [ ehrharttheorem ] let @xmath75 be a polytope with rational vertices . then the function @xmath76 is a quasi - polynomial of degree @xmath77 .
furthermore , if @xmath78 is an integer such that @xmath79 has integral vertices , then @xmath78 is a period of @xmath80 . in particular , if @xmath0 has integral vertices , then @xmath80 is a polynomial
. see ( * ? ? ?
* theorem 4.6.25 ) or ( * ? ? ?
* theorem 3.23 ) .
the function @xmath81 is called the * ehrhart quasi - polynomial * of @xmath0 .
one can see this as saying that if the vertices of @xmath0 are @xmath82 , then the vertices of @xmath83 are given by the linear functions @xmath84 .
we generalize this as [ polytope ] given polynomials @xmath85 $ ] for @xmath86 and @xmath87 , let @xmath2 be a positive integer such that @xmath88 for all @xmath89 .
this is satisfied by @xmath2 sufficiently large , so we can define a rational polytope @xmath90 , where @xmath91
. then @xmath92 .
we call the function @xmath93 a * generalized ehrhart polynomial*. in section [ equi ] , we explain the equivalence of the two problems just mentioned , then prove our main result theorem [ ehrhartconj ] in section [ section : mainproof ] .
the proof gives us an algorithm to compute these generalized ehrhart polynomials , but it could be very complicated in practice . for computational reasons ,
we introduce the notions of generalized division and generalized gcd for the ring @xmath94 $ ] in section [ generalizeddivision ] . in a sense ,
these generalize the usual notions of division and gcd for the ring of integers @xmath95 .
the methods and algorithms are quite similar , but are more involved due to technical complications .
the proofs for the correctness of the algorithms are given in this section .
as an application of these tools , in section [ section : application ] , we will describe explicit computations of some special generalized ehrhart polynomials .
we thank richard stanley for useful discussions and for reading previous drafts of this article .
we also thank an anonymous referee for helping to improve the quality and readability of the paper .
sheng chen was sponsored by project 11001064 supported by national natural science foundation of china .
steven sam was supported by an nsf graduate fellowship and an ndseg fellowship .
as we shall see , the two problems of the diophantine equations and lattice point enumeration are closely intertwined . in this section ,
we want to show that theorem [ ehrhartconj ] is equivalent to theorem [ polytope ] . before this , let us see the equivalence of theorem [ polytope ] with the following result . for notation , if @xmath16 and @xmath96 are vectors , then @xmath97 if @xmath98 for all @xmath99 .
[ ineq ] for @xmath100 , define a rational polytope @xmath101 , where @xmath102 is an @xmath103 matrix , and @xmath104 is an @xmath105 column vector , both of whose entries are integer coefficient polynomials
. then @xmath92 .
notice that the difference of theorem [ polytope ] and theorem [ ineq ] is that one defines a polytope by its vertices and the other by hyperplanes .
so we will show their equivalence by presenting a generalized version of the algorithm connecting `` vertex description '' and `` hyperplane description '' of a polytope .
the connection is based on the fact that we can compare two rational functions @xmath4 and @xmath106 when @xmath2 is sufficiently large .
for example , if @xmath107 and @xmath108 , then @xmath109 for all @xmath110 , we denote this by @xmath111 ( `` even '' being shorthand for `` eventually '' ) . therefore , given a point and a hyperplane , we can test their relative position . to be precise ,
let @xmath112 be a point where the @xmath113 are rational functions and let @xmath114 be a hyperplane where all the @xmath115 are polynomials of @xmath2
. then exactly one of the following will be true : @xmath116 given this , we can make the following definition .
we say that two points @xmath117 and @xmath118 * lie * ( resp .
, * weakly lie * ) * on the same side of @xmath119 * if @xmath120 ( resp . , @xmath121 ) .
going from the `` vertex description '' to the `` hyperplane description '' : given all vertices of a polytope @xmath3 , whose coordinates are all rational functions of @xmath2 , we want to get its `` hyperplane description '' for @xmath100 .
let @xmath122 be a hyperplane defined by a subset of vertices .
if all vertices lie weakly on one side of @xmath122 , we will keep it together with @xmath123 , or @xmath124 or @xmath125 indicating the relative position of this hyperplane and the polytope .
we can get all the hyperplanes defining the polytope by this procedure .
going from the `` hyperplane description '' to the `` vertex description '' : let @xmath126 be a polytope , where @xmath102 is an @xmath103 matrix , and @xmath104 is an @xmath105 column vector , both of whose entries are integer coefficient polynomials .
we want to find its vertex description .
let @xmath127 be the linear functionals defined by the rows of @xmath128 .
so we can rewrite @xmath3 as @xmath129 the vertices of @xmath3 can be obtained as follows . for every @xmath130-subset @xmath131 , if the equations @xmath132 are linearly independent for @xmath100 , and their intersection is nonempty , then it consists of a single point , which we denote by @xmath133 . if @xmath134 for all @xmath135 , then @xmath136 is a vertex of @xmath3 , and all vertices
are obtained in this way .
we claim that the subsets @xmath137 for which @xmath133 is a vertex remains constant if we take @xmath2 sufficiently large .
first , the notion of being linearly independent equations can be tested by showing that at least one of the @xmath138 minors of the rows of @xmath128 indexed by @xmath137 does not vanish .
since these minors are all polynomial functions , they can only have finitely many roots unless they are identically zero . hence taking @xmath41
, we can assume that @xmath139 is either always linearly dependent or always linearly independent .
similarly , the sign of @xmath140 is determined by the sign of a polynomial , and hence is constant for @xmath41 .
we can easily transform an inequality to an equality by introducing some slack variables and we can also represent an equality @xmath141 by two inequalities @xmath142 and @xmath143 .
so the main difference between the two theorems is that theorem [ ehrhartconj ] is counting nonnegative solutions while theorem [ ineq ] is counting all integral solutions .
but we can deal with this by adding constraints on each variable .
a more interesting connection between theorem [ ehrhartconj ] and theorem [ ineq ] is worth mentioning here .
first consider any fixed integer @xmath2 .
then the entries of @xmath46 and @xmath48 in the linear diophantine equations @xmath50 of theorem [ ehrhartconj ] all become integers . for an integer matrix
, we can calculate its smith normal form .
similarly , we can use a generalized smith normal form for matrices over @xmath44 to get a transformation from theorem [ ehrhartconj ] to theorem [ ineq ] . then given @xmath46 and @xmath48 , by theorem [ smith ]
, we can put @xmath46 into generalized smith normal form : @xmath144 for some matrix @xmath145 with nonzero entries only on its main diagonal , and unimodular matrices @xmath146 and @xmath128 .
then the equation @xmath50 can be rewritten as @xmath147 . set @xmath148 and @xmath149 . by the form of @xmath150
, we have a solution @xmath96 if and only if @xmath151 divides @xmath152 for @xmath153 , and for any given solution , the values @xmath154 can be arbitrary .
however , since @xmath155 , we need to require that @xmath156 , and any such @xmath96 gives a nonnegative solution @xmath16 to the original problem . simplifying @xmath157 , where @xmath158 , we get @xmath159 , where @xmath160 , @xmath161 and @xmath162 .
although @xmath163 and @xmath164 has entries in @xmath44 , we can assume that they are polynomials by dealing with each constituent of the quasi - polynomials separately .
so we reduce theorem [ ehrhartconj ] to theorem [ ineq ] .
consider the nonnegative integer solutions to @xmath165 write @xmath166 .
when @xmath167 , the smith normal form of @xmath168 is @xmath169 so the equation becomes @xmath170 set @xmath171 , so that we have @xmath172 and thus @xmath173 . by the condition @xmath174
, we require @xmath175 which gives us @xmath176 a one dimensional polytope .
so the number of solutions for @xmath167 is @xmath177 we can do the case @xmath178 similarly .
the smith normal form of @xmath179 is @xmath180 and @xmath181 , thus @xmath182 .
the proof of theorem [ smith ] is based on a theory of generalized division and gcd over the ring @xmath94 $ ] , which mainly says that for @xmath183 $ ] , the functions @xmath184 , @xmath185 , and @xmath186 lie in the ring @xmath44 . one interesting consequence of these results is that every finitely generated ideal in @xmath44 is principal , despite the fact that @xmath44 is not noetherian .
we developed this theory in order to appoach theorem [ ehrhartconj ] at first , but subsequently have found a proof that circumvents its use .
further details can be found in section [ generalizeddivision ] .
to prove theorem [ ehrhartconj ] , we will use a `` writing in base @xmath2 '' trick , to reduce equations with polynomial coefficients to linear functions . briefly , the idea of the following `` writing in base @xmath2 '' trick is as follows : given a linear diophantine equation @xmath187 with polynomial coefficients @xmath115 and @xmath188 , fix an integer @xmath2 , then the coefficients all become integers . now consider a solution @xmath189 with @xmath190 .
put the values of @xmath189 into the equation , then both sides become an integer .
then we use the fact that any integer has a unique representation in base @xmath2 ( @xmath2 is a fixed number ) , and compare the coefficient of each power of @xmath2 in both sides of the equation . finally , letting @xmath2 change
, we happen to have a uniform expression for both sides in base @xmath2 when @xmath2 is sufficiently large
. moreover , the coefficient of each power of @xmath2 in both sides of the equation are all linear functions of @xmath2 ( lemma [ repr ] and lemma [ reduce ] ) . then by lemma [ linear ] , we can reduce these equations with linear function coefficients to the case when we can apply ehrhart s theorem ( theorem [ ehrharttheorem ] ) to show that the number of solutions are quasi - polynomials of @xmath2 .
this completes the proof of theorem [ ehrhartconj ] .
[ repr ] given @xmath191 $ ] with @xmath192 for @xmath100 ( i.e. , @xmath193 has positive leading coefficient ) , there is a unique representation of @xmath117 in base @xmath2 : @xmath194 where @xmath195 is a linear function of @xmath2 such that for @xmath41 , @xmath196 for @xmath197 and @xmath198 .
we denote @xmath199 .
note that @xmath200 may not be equal to @xmath201 .
for example , @xmath202 is represented as @xmath203 with @xmath204 , @xmath205 , and @xmath206 .
let @xmath207 with integral coefficients and @xmath208 .
if the @xmath209 are all nonnegative , the proposition holds for @xmath210 , otherwise we can prove it by induction on @xmath211 , where @xmath99 is the smallest index such that @xmath212 .
suppose @xmath213 is the minimal number such that @xmath214 , then in the representation of @xmath117 , put @xmath215 for @xmath216 , @xmath217 . by induction , we have a representation for @xmath218 so we can add it to @xmath219 to get the desired representation for @xmath117 .
since for @xmath100 we have @xmath220 , by induction , we can make sure @xmath221 for @xmath100 . for a lower bound
, @xmath222 is sufficient , i.e. , for all @xmath223 , the desired unique representation is guaranteed to exist .
note that @xmath224 , since @xmath225 .
so this process will stop in finitely many steps .
uniqueness of this representation is clear .
[ reduce ] fix an integer @xmath226 and @xmath41 .
consider the set @xmath227 where @xmath228 ( as a usual polynomial ) and @xmath229 ( represented in base @xmath2 as in lemma [ repr ] ) , with @xmath230 and @xmath231 .
then @xmath232 is in bijection with a finite union of sets of the form @xmath233 where @xmath234 and @xmath235 is a linear form of @xmath16 with constant coefficients .
fix a sufficiently large positive integer @xmath2 . by our assumptions , for any point @xmath236 , we can write @xmath237 with @xmath238 .
the rest of the lemma is a direct `` base @xmath2 '' comparison starting from the lowest power to the highest power in the equation @xmath239 . to get a feel for the proof , we recommend that the reader look at example [ oneeq ] first .
we will use this as a running example to explain the steps of the proof .
first write @xmath240 in base @xmath2 , so that each @xmath241 is a linear function in @xmath2 . then we know that @xmath242 since we have fixed @xmath2 sufficiently large . going back to example [ oneeq ]
, we have @xmath243
so that @xmath244 , @xmath245 and @xmath246 , so that here `` sufficiently large '' means @xmath247 . now expand out the equation @xmath248 to get @xmath249 the constant term in base @xmath2 of gives us the equation @xmath250 since we have the bound @xmath251 for all @xmath99 , we can in fact say that the lhs is equal to @xmath252 where @xmath253 is an integer such that @xmath253 is strictly greater than the sum of the negative @xmath254 and strictly less than the sum of the positive @xmath254 ( and @xmath253 can also be 0 ) .
but note that @xmath253 only depends on the @xmath254 , so if our @xmath2 is sufficiently large , we may assume that @xmath255 .
going back to example [ oneeq ] , we have the equation @xmath256 so that @xmath257 .
so now we have finitely many cases for the value of @xmath253 to deal with .
going back to the equation , we can substitute our value of @xmath253 and compare linear coefficients to get @xmath258 again , since we know that @xmath251 and @xmath259 for all @xmath99 , we can say that the lhs is equal to @xmath260 where @xmath261 is an integer such that @xmath261 is greater than the sum of the negative @xmath254 and @xmath262 and less than the sum of the positive @xmath254 and @xmath262 ( and this is independent of @xmath253 because @xmath255 ) . going back to example [ oneeq ] ,
we get the equation @xmath263 so @xmath264 .
now we have finitely many cases of @xmath261 , and again we fix one .
we continue on in this way to get @xmath265 . at each point , we only had finitely many choices for the next @xmath266 , so at the end , we only have finitely many sequences @xmath267 . for a given @xmath268
, we see that the @xmath269 must satisfy finitely many equalities of the form @xmath270 where @xmath11 is a linear form in @xmath16 with constant coefficients and @xmath271 .
in example [ oneeq ] , for @xmath272 , we would have the equations @xmath273 along with the inequalities @xmath274 , and we have underlined the places where @xmath261 appears .
each such @xmath268 gives us a set of the form @xmath275 , and to finish we take the union of these sets over all sequences @xmath268 .
we will need one more prepatory lemma before doing the proof of theorem [ ehrhartconj ] .
[ linear]if @xmath3 is a polytope defined by inequalities of the form @xmath276 , where @xmath234 and @xmath235 is a linear form of @xmath16 with constant coefficients , then @xmath277 .
we first use the fact that the combinatorics of @xmath3 stabilizes for sufficiently large @xmath2 ( see section [ equi ] ) , say that the polytope has @xmath5 vertices @xmath278 .
also , note that the coefficients of the vertices of the polytope are linear functions in @xmath2 with rational coefficients .
let @xmath78 be the least common multiple of the denominators that appear in all of the coordinates .
then by ( * ? ? ?
* theorem 18.4 ) , the function sending @xmath279 to the number of integer points in @xmath3 is a polynomial if we restrict to a specific congruence class of @xmath2 modulo @xmath78 .
we can compose each of these functions with the polynomial @xmath280 to conclude that @xmath277
. we can also give a self - contained proof of lemma [ linear ] that only uses ehrhart s theorem ( theorem [ ehrharttheorem ] ) .
we will do induction on dimension first , the case of dimension 0 being trivial . an inequality of the form @xmath281 with @xmath282 will be called an inequality with constant term . secondly , we will do induction on the number of inequalities with constant term .
theorem [ ehrharttheorem ] gives the desired result when there are no inequalities with constant term .
in general , let @xmath3 be a polytope defined by @xmath281 plus other relations of the same form , which we will call @xmath283 .
let @xmath284 be the polytope with relations @xmath285 and @xmath283 .
if @xmath286 , then for @xmath287 , let @xmath288 be the polytope with relations @xmath289 and @xmath283 .
then @xmath290 so to show @xmath291 , it suffices to show that @xmath292 and @xmath293 for @xmath294 .
since @xmath284 has one less inequality with constant term , we know that @xmath292 by induction . as for @xmath288 , write @xmath295 and choose @xmath135 such that @xmath296 .
then the equation @xmath297 can be rewritten as @xmath298 making substitutions into the relations @xmath299 , each relation still has the form @xmath300 ( clear denominators if necessary to make sure all of the coefficients are integral ) , and we have eliminated the variable @xmath301 . if @xmath302 , then for @xmath303 , let @xmath288 be the polytope with relations @xmath304 and @xmath283
. then @xmath305 and we proceed as before .
hence @xmath293 by induction on dimension .
consider a system of linear diophantine equations @xmath306 with finitely many nonnegative integer solutions for each @xmath100 , as in theorem [ ehrhartconj ] .
let @xmath187 be any equation from the system .
the equations @xmath306 define a bounded polytope whose vertices are given by rational functions of @xmath2 as discussed in section [ equi ] , where the degrees of the rational functions ( difference between the degree of the numerator and the denominator ) are independent of @xmath2 .
we can express each point in this polytope as a convex combination of the vertices .
therefore , writing each variable @xmath307 in base @xmath2 , we can find some positive integer @xmath226 , which does not depend on @xmath2 , such that the coordinates of each point in the polytope are less than @xmath308 .
using lemma [ reduce ] , we can reduce each equation from @xmath306 into a new system like @xmath309 , with more variables than @xmath310 but all restrictions are of the form @xmath311 . applying lemma [ linear ]
finishes the proof .
we give an example for lemma [ linear ] . for @xmath2 a positive integer , let @xmath3 be the polygon defined by the inequalities @xmath312 , @xmath313 and @xmath314 .
then @xmath284 is defined by the inequalities @xmath312 , @xmath313 , and @xmath315 , and @xmath316 is defined by the inequalities @xmath312 , @xmath317 , and @xmath318 .
we can rewrite the equality as @xmath319 , and then the other inequalities become @xmath312 and @xmath320 .
we see that @xmath284 is the convex hull of the points @xmath321 , while @xmath316 is the interval @xmath322 $ ] .
the total number of integer points in @xmath284 and @xmath316 is given by the quasipolynomial @xmath323 its rational generating function is @xmath324 [ oneeq ] we give an example of lemma [ reduce ] .
consider nonnegative integer solutions for @xmath325 for any @xmath326 , @xmath327 is the expression in base @xmath2 .
now consider the left hand side .
writing @xmath328 in base @xmath2 , let @xmath329 , @xmath330 and @xmath331 with @xmath332 . in this case , @xmath333 , but we have further restrictions on the degrees of @xmath334 and @xmath335 in base @xmath2 coming from the coefficients @xmath336 .
then we have @xmath337 now we can write the left hand side in base @xmath2 with extra constraints on @xmath338 s .
we start with comparing the coefficient of @xmath339 in both sides .
we have the following three cases : @xmath340 @xmath341 and @xmath342 so in the language of proof of lemma [ reduce ] , we have @xmath257 .
we next consider @xmath343 term .
if @xmath16 satisfies @xmath344 for @xmath339 , @xmath345 , then the equation is reduced to @xmath346 now compare the @xmath343 terms .
we have five cases for each @xmath345 : @xmath347 where @xmath348 .
last , we compare the @xmath349 terms . note that since we assume @xmath100 , the @xmath339 term wo nt carry over to @xmath349 term , so the computation of @xmath349 term only depends on the term @xmath343 . if @xmath16 satisfies the @xmath261th condition for @xmath343 , the equation then becomes @xmath350 so for each @xmath351 , we have @xmath352 overall , we have the set @xmath353 is in bijection with the set @xmath354 such that @xmath16 satisfies the equations in the sets @xmath355 here we borrow the notation of matrix multiplication @xmath356 to represent intersection of sets @xmath357 and matrix summation @xmath358 to represent set union @xmath359 .
note that here all constrains @xmath360 on @xmath361 are in the form of @xmath362 , where @xmath234 and @xmath235 is a linear form of @xmath16 with constant coefficients .
an * integer - valued polynomial * , or a * numerical polynomial * is a function @xmath45 with a polynomial @xmath363 $ ] satisfying @xmath364 for all @xmath365 . for any numerical polynomial @xmath11 and a constant @xmath366 ,
it is not hard to see that @xmath367 .
we want to generalize this to @xmath368 for two numerical polynomials @xmath11 and @xmath369 . based on classical division ,
we have for all @xmath370 that @xmath371 , with @xmath372 , we define two integer - valued functions @xmath373 , called the * quotient * of @xmath4 and @xmath106 and @xmath374 , called the * remainder * of @xmath4 and @xmath106 .
we will show that @xmath375 by the generalized division algorithm .
this generalized division algorithm is based on the division for @xmath376 $ ] , except for two modifications .
first , in order to have @xmath377 for @xmath100 , we need to change the remainder condition of division in @xmath378 $ ] , as in [ algorithm [ divisionalg ] , lines [ rem1][rem2 ] ] .
for example , for @xmath379 and @xmath380 , then we should have @xmath381 , @xmath382 , but these do not satisfy @xmath383 nor @xmath384 for division in @xmath378 $ ] . secondly , in order to keep @xmath385 $ ] , we break @xmath16 into finitely many `` branches '' whenever we have noninteger coefficients , as in [ algorithm [ divisionalg ] , lines [ break1][break2 ] ] . for example , take @xmath386 and @xmath387 .
trying to do division in @xmath378 $ ] , leading coefficient of @xmath388 will not be an integer , so we break @xmath16 into 2 `` branches '' , since 2 is the leading coefficient of @xmath389 .
now we do division to @xmath390 and @xmath391 , when @xmath392 ; and @xmath393 and @xmath394 , when @xmath395 . for @xmath392 , we have @xmath396 and @xmath397 and for @xmath395 we have @xmath398 and @xmath399 .
we will show ( see theorem [ al1 ] ) that breaking @xmath16 into @xmath400 ( using notation in algorithm [ divisionalg ] ) `` branches '' is sufficient to conclude that @xmath401 $ ] , for @xmath402 .
@xmath403 $ ] .
@xmath404 @xmath405 and @xmath406 , @xmath407 $ ] , for @xmath408 such that @xmath409 output @xmath410 @xmath381 ; @xmath411 .
@xmath412 ; @xmath413 .
@xmath414 ; @xmath415 .
output @xmath416.[break1 ] set @xmath417 ; @xmath418 ; @xmath419 ; @xmath420.[start ] @xmath421 ; @xmath422;[degr-1 ] @xmath423 ; [ pi1 ] @xmath424 ; @xmath425 [ break2 ] @xmath426 ; @xmath427 ; [ pi2 ] @xmath428[ri2 ] @xmath425 @xmath424 [ rem1 ] output @xmath429 , @xmath406 ; @xmath430 ; @xmath431 [ rem2 ] note that in algorithm [ divisionalg ] , we define @xmath432 to be the leading coefficient of @xmath433 .
[ al1 ] algorithm [ divisionalg ] is correct . the algorithm will stop after finitely many steps ( @xmath434 is reduced by 1 each step in [ line [ degr-1 ] ] ) . for @xmath408 , and for @xmath435 , we claim the following two properties : 1 . [
1]@xmath436 $ ] .
[ 2]@xmath437 and @xmath438 , for @xmath100 and @xmath439 .
these two properties imply the correctness of the algorithm . by the algorithm ,
[ 2 ] is obvious , so we only need to show [ 1 ] .
note that in each step , we add to @xmath406 some @xmath421 or an integer @xmath104 ( see [ lines [ pi1 ] and [ pi2 ] ] ) , and subtract @xmath440 or @xmath441 from @xmath429 ( see [ lines [ degr-1 ] and [ ri2 ] ] ) , so it suffices to show @xmath442 $ ] . for every @xmath429 with degree @xmath443 in [ lines [ break1 ] - [ break2 ] ] , denote it by @xmath444 and its leading coefficient by @xmath445 .
so we need to prove that @xmath446 for all @xmath447 . to show this
, we use induction on @xmath448 to prove a stronger statement that @xmath449 for every monomial @xmath450 in @xmath444 . if @xmath451 , as in [ line [ start ] ] , @xmath452 and the claim holds by binomial expansion .
assume it still holds for some @xmath5 such that @xmath453 .
then for @xmath454 , we have @xmath455 . in [ line [ degr-1 ] ] , @xmath456 so we only need to show that @xmath457 also has the property that @xmath458 , for every monomial in @xmath440 .
this is true because @xmath449 in @xmath459 , and in @xmath460 we have @xmath461 by hypothesis and @xmath462 for @xmath463 , and thus @xmath464 .
we want to generalize algorithm [ divisionalg ] to @xmath44 .
first we need another expression for numerical polynomials .
[ numericalpoly ] let @xmath11 be a numerical polynomial , i.e. , @xmath45 with a polynomial @xmath363 $ ] satisfying @xmath364 for all @xmath365 .
then there exists @xmath405 , and @xmath465 $ ] for @xmath402 such that @xmath466 for all @xmath439 .
suppose @xmath467 $ ] for some integer @xmath405 . for @xmath402
, define @xmath468 . by the binomial expansion of @xmath469 , it is easy to see that this is a polynomial with integer coefficients .
similarly , for an element @xmath470 , there exists @xmath471 and @xmath472 $ ] ( @xmath473 ) , satisfying @xmath474 for @xmath100 and @xmath475 .
we call @xmath476 a * representation * of @xmath477 .
another simple property is that for @xmath478 and @xmath479 , we can extend to a common period @xmath480 and @xmath481 and @xmath482 .
@xmath483 @xmath405 and @xmath406 , @xmath484 $ ] , for @xmath408 such that @xmath485 find a common period @xmath486 for @xmath487 , such that @xmath488 and @xmath489 .
apply algorithm 1 to @xmath490 and @xmath491 , get @xmath492 , @xmath493 $ ] , @xmath494 . set @xmath495 , and get @xmath406 , @xmath407 $ ] , for @xmath408 from @xmath496 .
[ gcd ] let @xmath487 be numerical polynomials , or more generally @xmath497 . define @xmath498 .
then by the generalized division over @xmath44 ( algorithm [ qpdivision ] ) , we have @xmath499 .
we call @xmath130 the * generalized gcd * of @xmath11 and @xmath369 , and denote it by @xmath500
. apply algorithm [ qpdivision ] successively .
define @xmath501 to be some constituent polynomial of @xmath11 and @xmath502 to be some constituent polynomial of @xmath369 ( a * constituent polynomial * of a quasi - polynomial is a polynomial on a congruence class modulo its period ) .
in general , if @xmath503 , define @xmath504 to be some constituent polynomial of @xmath505 . we will prove that @xmath506 for some @xmath99 by induction on @xmath434 and @xmath432 ( recall that @xmath432 means the leading coefficient of @xmath433 . ) .
hence , since our choice of constituent polynomial was arbitrary , we will eventually have a remainder of 0 after finitely many steps .
hence the algorithm terminates after finitely many steps , or else there exists an infinite path in decision tree .
if @xmath507 and @xmath508 , the algorithm works the same as the classical euclidean algorithm for integers . if @xmath509 , then since @xmath510 for @xmath41 , we have three cases for the next step : 1 .
@xmath511 2 .
@xmath512 and @xmath513 .
after finitely many iterations , one will either have @xmath514 , or @xmath511 .
3 . @xmath512 and @xmath515 . in the next step , @xmath516 .
so the algorithm terminates after finitely many steps . tracing through a natural generalization of the euclidean algorithm
, we can get @xmath517 such that @xmath518 .
more generally for @xmath519 we have [ ggcd ] let @xmath519 , @xmath520 .
then @xmath521 and there exists @xmath522 with @xmath523 such that @xmath524 this implies , in particular , that every finitely generated ideal of @xmath44 is principal .
it is peculiar that in fact @xmath44 is not a noetherian ring .
for example , take @xmath525 to be the function such that @xmath526 if @xmath527 , and 0 otherwise
. then @xmath528 is not a finitely generated ideal .
recall that for any @xmath529 , there exist unimodular matrices @xmath530 , i.e. , determinant @xmath531 , such that @xmath532 with @xmath533 and @xmath78 is called the smith normal form of @xmath534 . by generalized division over @xmath44 ( algorithm [ qpdivision ] ) and corollary [ ggcd ]
, we can generalize the classical smith normal form theorem as follows .
[ smith ] for any @xmath535 , define a matrix function @xmath536 such that @xmath150 is the smith normal form of @xmath537 .
then we have @xmath538 and there exists @xmath539 , @xmath540 such that @xmath541 are unimodular for @xmath100 and @xmath542 .
we call this matrix function @xmath78 the * generalized smith normal form * of @xmath534 . before the proof ,
let us see a simple example for generalized gcd and generalized smith normal form .
let @xmath543 , @xmath544 .
denote @xmath545 , namely @xmath546 .
+ put @xmath547 and @xmath548 , @xmath549 , @xmath550 , @xmath551 , @xmath552 , @xmath553 , we have @xmath554 for @xmath555 , let @xmath556 $ ] , where @xmath557 and @xmath558 $ ] , where @xmath559 and @xmath560 $ ] , where @xmath561 we have @xmath562 is the smith normal form for @xmath46 for @xmath563 where @xmath564 and @xmath565 . the following proof is based on the proof for classical smith normal form in @xcite .
[ row]let @xmath566 with @xmath567 .
then there is a matrix @xmath568 with first row @xmath569 $ ] and determinant @xmath570 . by induction on @xmath571 .
for @xmath572 , trivial .
now suppose the lemma is true for @xmath573 , where @xmath574 , and let @xmath575 be a matrix with first row @xmath576 $ ] and determinant @xmath577 .
since @xmath578 by corollary [ ggcd ] , there exist @xmath579 such that @xmath580 .
put @xmath581 then we have @xmath568 with first row @xmath582 $ ] and @xmath583 .
write @xmath584 .
if @xmath585 , then switch rows and columns so that this is true . without loss of generality , we may assume that @xmath586 divides all elements in the column .
if not , let @xmath587 .
by corollary @xmath588 , there exist @xmath589 with @xmath523 such that @xmath590 . by lemma @xmath591
, there exists a matrix @xmath592 with first row @xmath593 $ ] and determinant @xmath594 .
so we have @xmath595 such that @xmath596 divides all elements in the column .
suppose @xmath586 divides all elements @xmath597 .
first , by the same process as above , we can assume that @xmath586 divides all elements in the first row and column . by elementary row and column operations ,
all the elements in the first row and column , other than the @xmath598 position , can be made zero .
if there is some element @xmath597 , @xmath599 is not divisibleby @xmath586 , add column @xmath135 to column @xmath594 and repeat the previous step to replace @xmath586 with one of its divisor which divides @xmath597 .
thus we must finally reach the stage where the element in the @xmath598 position divides every element of the matrix and all the other elements of the first row and column are zero . repeat the entire process with the submatrix obtained by deleting the first row and column .
we can see that the algorithm we give in the proof of theorem [ ehrhartconj ] is not very efficient . in this section , by the theory of generalized division developed in the previous section , we are able to compute some special cases more efficiently .
first recall the classical popoviciu formula .
let @xmath12 and @xmath13 be relatively prime positive integers .
then the number of nonnegative integer solutions @xmath14 to the equation @xmath15 is given by the formula .
now consider the equation @xmath600 where @xmath601 are integeral polynomials of @xmath2 . by the theory of generalized division , the above formula for the number of nonnegative integer solutions @xmath602 can be easily generalized to the following then as mentioned in the introduction
, we can consider the number of solutions @xmath24 to the matrix equation where the @xmath26 and @xmath27 are fixed positive integers and @xmath608 for @xmath29 .
write @xmath609 .
we assume that @xmath31 , so that there exist integers ( not unique ) @xmath32 such that @xmath33 now define two regions @xmath34 for @xmath29 .
then if @xmath35 is in the positive span of the columns of the matrix in , there exist the following popoviciu - like formulas ( * ? ? ?
* theorem 4.3 ) for @xmath610 , the number of solutions of .
notice that in the above formula , everything can be directly generalized to the ring @xmath94 $ ] using generalized division and generalized gcd .
in other words , we can replace the @xmath26 , @xmath27 , and @xmath39 by polynomials in @xmath2 in such a way that for all values of @xmath2 , the condition @xmath31 always holds .
for example , consider the system @xmath40 then for @xmath41 , we have that @xmath42 so the number of solutions @xmath43 is counted by the formula of @xmath610 for @xmath619 and it is easy to see that @xmath620 . for polytopes in dimension two , there is an efficient algorithm for computing the number of lattice points using continued fractions .
see ( * ? ? ?
* chapter 15 ) for the algorithm in the case of non - parametrized two dimensional polytopes . by the generalized division theory
, we can generalize this algorithm to the case of a polytope whose vertices are given by polynomials .
notice that the only difference is the generalization of continued fractions : for any @xmath621 $ ] , we expand @xmath622 as a continued fraction .
then by theorem [ gcd ] , for @xmath100 , the terms in the expansion are quasi - polynomials and the number of terms does not depend on @xmath2 . for @xmath625
, we have @xmath626 & n=2m\\ [ m-1;\,3,\,1,\,m-1 ] & n=2m-1 \end{cases}.\]]and for the cone generated by @xmath627 and @xmath628 , we have the following efficient algorithm of decomposition into prime cones .
matthias beck and sinai robins , _ computing the continuous discretely : integer - point enumeration in polyhedra _ , undergraduate texts in mathematics , springer , new york , 2007 , available for download from http://math.sfsu.edu/beck/ccd.html .
zhiqiang xu , an explicit formulation for two dimensional vector partition functions , _ integer points in polyhedra geometry , number theory , representation theory , algebra , optimization , statistics _ ,
contemporary mathematics * 452 * ( 2008 ) , 163178 . | let @xmath0 be a polytope with rational vertices .
a classical theorem of ehrhart states that the number of lattice points in the dilations @xmath1 is a quasi - polynomial in @xmath2 .
we generalize this theorem by allowing the vertices of @xmath3 to be arbitrary rational functions in @xmath2 . in this case
we prove that the number of lattice points in @xmath3 is a quasi - polynomial for @xmath2 sufficiently large .
our work was motivated by a conjecture of ehrhart on the number of solutions to parametrized linear diophantine equations whose coefficients are polynomials in @xmath2 , and we explain how these two problems are related . |
99 m. rivera , j. lightwave technol . * 13 * , 233 ( 1995 ) . | a widespread , intuitive and computationally inexpensive method to analyze light guidance through waveguide bends is by introducing an equivalent straight waveguide with refractive index profile modified to account for actual waveguide bend .
here we revise the commonly used equivalent - index formula , ending up with its simple extension that enables rigorous treatment of one- and two - dimensionally confined , uniformly bent waveguides , including tightly coiled microstructure fibers , curved ridge waveguides and ring microresonators .
we also show that such technique is applicable only to waveguides composed of isotropic or uniaxially anisotropic materials , with anisotropy axis directed perpendicular to the curvature plane . to understand and to predict light wave behavior at waveguide bends was of major importance and interest to integrated- and fibre - optics community from around 1970s and onwards .
today this interest is stimulated largely by two developments : ( i ) increasing the packaging density of integrated - optic circuits , backed by minimizing integrated waveguide bend radii while keeping bend losses at a tolerable level ; and ( ii ) the advent of photonic crystal fibers possessing quite complicated , high index - contrast dielectric profiles as compared to step- or graded - index fibers for which early theoretical methods to treat bend losses were developed .
numerical methods to simulate light propagation through waveguide bends , such as beam propagation method @xcite , the method of lines @xcite , as well as more general - purpose finite - element and finite - difference techniques , can be very demanding computationally : in one recent example @xcite , a 64-processors cluster was used in full - vector finite - element modelling .
no wonder one leans to analytic techniques to reduce computation burden whenever possible ; one such technique , perhaps the simplest and most widely used in modelling of microstructure fibre bends today @xcite , is equivalent - profile method @xcite .
it reduces dimensionality of the problem by replacing actual waveguide bend by a straight piece with refractive index profile @xmath0 where @xmath1 is dielectric permittivity in the bent waveguide cross - section , as measured in cartesian frame , @xmath2 is the bend radius pointing from the @xmath3-directed curvature axis to a ( somewhat arbitrary ) waveguide central plane , @xmath4 the distance from that plane . designed for elegant and inexpensive treatment of bends ,
though , formula ( [ standardalternativeepsilon ] ) grounds on approximations whose validity may often be questioned , as those of weak - guidance regime and small curvature ( @xmath5 ) , and originally was introduced for quite a specific case of non - magnetic , step - index , low index - contrast fibre . in this letter
we overcome all these limitations by an alternative , first - principle derivation of expressions for modified dielectric permittivity @xmath6 and magnetic permeability @xmath7 analogous to ( [ standardalternativeepsilon ] ) , but applicable to waveguides of arbitrary cross - section under the only assumption that bend radius is constant ( while not necessarily large ) .
following schouten @xcite and post @xcite , we start from the generally covariant maxwell equations @xmath8 } = -\dot b_{\lambda\nu } , & \quad & \partial_{[\kappa } b_{\lambda\nu ] } = 0 , \label{maxwellinv1}\\ \partial_{\nu } \mathfrak{h}^{\lambda\nu } = \dot\mathfrak{d}^\lambda + \mathfrak{j}^\lambda , & \quad & \partial_{\lambda}\mathfrak{d}^{\lambda } = \rho . \label{maxwellinv2}\end{aligned}\ ] ] here the square brackets denote alternation : @xmath9 } = \frac{1}{2!}(\cdot_{\lambda\nu } - \cdot_{\nu\lambda})$ ] , @xmath10 and @xmath11 are covariant electric vector and magnetic bivector fields coinciding with electric and magnetic three - vectors @xmath12 and @xmath13 in cartesian frame , @xmath14 and @xmath15 are contravariant magnetic bivector density and electric vector density of weight @xmath16 corresponding to @xmath17 and @xmath18 , @xmath19 and @xmath20 are electric current and charge densities , @xmath21 . with such transformation characteristics assigned to the electromagnetic field quantities , eqs .
( [ maxwellinv1 ] ) , ( [ maxwellinv2 ] ) are known to be form - invariant @xcite , i.e. , they do not change their form under arbitrary revertible coordinate transformations
. it might be practical to convert eqs .
( [ maxwellinv1 ] ) , ( [ maxwellinv2 ] ) with use of dual equivalents @xmath22 and @xmath23 , where @xmath24 and @xmath25 are pseudo ( hence tildes ) permutation fields equal to levi - civita symbol in any coordinate system , to the form directly reminiscent of maxwell equations in cartesian frame : @xmath26 while the constitutive relations are implied being @xmath27 here @xmath28 and @xmath29 are tensor densities of weight @xmath16 : @xmath30 as stipulated by transformation characteristics of the fields @xmath31 , @xmath32 , @xmath33 and @xmath34 .
we denote by @xmath35 the jacobian transformation matrix , and @xmath36 its determinant . in this formulation , geometry enters maxwell equations ( [ maxwellinvmod1 ] ) , ( [ maxwellinvmod2 ] ) through material fields ( [ materialconstantstransform ] ) exclusively ; since the form of ( [ maxwellinvmod1 ] ) , ( [ maxwellinvmod2 ] ) is precisely as if they were written in cartesian components , a multitude of analytic and numeric methods developed for rectangular cartesian frame apply .
furthermore , whenever @xmath28 and @xmath29 happen to be independent of one ( or more ) of the coordinates in a chosen coordinate system , that coordinate ( or those coordinates ) can be separated in the usual manner , like @xmath37 coordinate in the case of straight homogeneous waveguide .
now we show that in cylindrical coordinates @xmath38 , such ruling out of angular coordinate @xmath39 is the case for , in particular , isotropic guide of arbitrary cross - section , bent along @xmath39 , and specify eqs .
( [ materialconstantstransform ] ) for that case .
since isotropic media are defined as those possessing scalar ( though , in general , position - dependent ) permittivity @xmath1 and permeability @xmath40 , as referred to cartesian system , transformation rules ( [ materialconstantstransform ] ) reduce to @xmath41 owing to transformation behavior of the fundamental ( metric ) tensor @xmath42 ; its determinant @xmath43 insofar as @xmath44 in cartesian frame . equations ( [ isotropicmediacartesian ] ) resemble ` effective ' permittivity and permeability introduced in @xcite , albeit we would refrain from using the word ` effective ' , on the ground that ( [ isotropicmediacartesian ] ) are nothing but transformation rules for the permittivity and permeability of isotropic media . with transformation from orthogonal cartesian to cylindrical coordinates given by @xmath45 ( @xmath3 defines bending axis ) , one can find the cartesian - to - cylindrical transformation matrix for contravariant components @xmath46 and contravariant metric tensor @xmath47 @xmath48 . after substituting these expressions in eqs .
( [ isotropicmediacartesian ] ) , separating the @xmath39 variable in maxwell equations ( [ maxwellinvmod1 ] ) , ( [ maxwellinvmod2 ] ) and differentiating the @xmath49 multipliers with respect to @xmath39 , normalizing the propagation constant @xmath50 by @xmath2 in the usual manner and introducing shifted coordinate @xmath51 , one obtains @xmath52 ( we do not distinguish between co- and contravariant indices once particular coordinate system is chosen ) , and similarly for permeability components : @xmath53 note that expressions for @xmath54 and @xmath55 , multiplied by their magnetic counterparts , lead to refractive tensor components @xmath56 and @xmath57 precisely in line with standard formula ( [ standardalternativeepsilon ] ) in the case of non - magnetic media .
the @xmath58 and @xmath59 components manifestly differ from the rest , however ( see figure [ refractivetensorcomponents ] ) ; this difference becomes especially pronounced when departing from weakly guiding approximation , as soon as non - negligible @xmath37-components of electric and magnetic fields would ` probe ' the @xmath60 and @xmath61 profiles then .
of waveguide with bending , for diagonal refractive tensor components @xmath62 according to eqs .
( [ myalternativeisotropiceps ] ) , ( [ myalternativeisotropicmu]).,width=313 ] a benchmark example to illustrate the validity of formulae ( [ myalternativeisotropiceps ] ) , ( [ myalternativeisotropicmu ] ) is a problem of light propagation through a homogeneous slab waveguide bend that permits precise analytic treatment
see e.g. @xcite for an overview . in our approach , covariant maxwell curl equations , @xmath63 and @xmath64 ,
reduce to four independent first - order scalar equations then , grouped in two pairs : one for the @xmath65 , @xmath66 components ( te mode ) : @xmath67 where @xmath68 is the free - space wavenumber , @xmath69 the dimensionless mode index ; and another pair for @xmath70 , @xmath71 ( tm mode ) . to derive second - order equation for @xmath65 ( which can be regarded as an eigenproblem in @xmath72 )
, we differentiate ( [ ex ] ) with respect to @xmath73 , use ( [ hf ] ) for @xmath74 and @xmath75 to exclude @xmath66 .
this leads , in the regions of constant @xmath40 , to @xmath76 a bessel equation used customarily in exact analysis of bent slab waveguides @xcite , while an attempt to get similar equation with use of approximate equivalent - index formula ( [ standardalternativeepsilon ] ) fails .
the results of full - vector two - dimensional finite - difference frequency - domain ( fdfd ) modelling of confined modes in a high - contrast step - index optical fibre ( figure 2 ) demonstrate systematic discrepancy between approximate model ( [ standardalternativeepsilon ] ) and formulae ( [ myalternativeisotropiceps ] ) , ( [ myalternativeisotropicmu ] ) ; a discrepancy which can not be tolerated in realistic simulations of high index contrast ridge waveguides and holey fibers , ring microresonators , or sharp fibre bends such as those due to non - perfect alignment of fibre splices . of a step - index fibre of radius @xmath77 and dielectric index @xmath78 in the air background ,
calculated numerically with ( [ standardalternativeepsilon ] ) and ( [ myalternativeisotropiceps ] ) , ( [ myalternativeisotropicmu]).,width=313 ] a closer inspection of ( [ amatrix ] ) brings to conclusion that the only type of initial anisotropy in @xmath1 and @xmath40 which still permits ruling out the @xmath39 dependence in @xmath28 and @xmath29 transformed according to ( [ materialconstantstransform ] ) is that of uniaxial crystal with anisotropy axis pointing in the @xmath3 direction ( i.e. , along the bend axis ) : @xmath79 in that case , one gets instead of formulae ( [ myalternativeisotropiceps ] ) : @xmath80 and likewise for modified @xmath7 .
unfortunately , however , the case of anisotropy given by ( [ tensorialepsilonpermitted ] ) is not one encountered when modelling , e.g. , photonic bandgap fibre filled with liquid crystal ; for that and other examples of curved waveguides possessing anisotropy of some general kind , _ rigorous _ equivalent - profile formulation is not applicable and should be substantiated by more brute - force numeric or less stringent analytic techniques . in summary ,
equivalent - profile expressions ( [ myalternativeisotropiceps ] ) , ( [ myalternativeisotropicmu ] ) to treat waveguide bends have been presented . unlike conventional formula ( [ standardalternativeepsilon ] ) , they hold for arbitrarily tight bends , arbitrary high - contrast material profiles , and permit the inclusion of magnetic parts .
this offers three - dimensional modelling accuracy with those existing full - vectorial two - dimensional finite - difference or finite - element solvers which permit @xmath81 diagonal matrices for @xmath6 and @xmath7 be put in . eliminating the restraints on bend radius and refractive index contrast enables one to simulate also the whispering - gallery modes in spherical , toroidal and other types of optical microcavities .
the principal limitation of the equivalent - profile technique is that many cases of non - trivial anisotropy can not be treated rigorously in the manner above . |
in the paper , we examine @xmath19-smooth real functions which can be computed ( in some new but natural meaning which is rigorously defined below ) on finite automata , i.e. , on sequential machines that have only finite number of states .
we show that all these functions are affine and , moreover , that they can be expressed as complex functions @xmath29 and thus can be ascribed ( also in some natural rigorous meaning ) to matter waves from quantum theory .
a general problem of evaluation of real functions on abstract discrete machines naturally arose at the very moment the first digital computers had been invented .
there are a number of various mathematical statements of the problem which depend both on specific mathematical model of a digital computer ( the abstract machine ) and on the representation of reals in some ` digital ' form .
for instance , real number computations on turing machines constitute a core of theory of constructive reals and computable functions .
the theory demonstrates intensive development for during more than half a century , see e.g. @xcite and references therein .
sequential machines ( also known as mealy automata , or as finite - state letter - to - letter transducers ) are , speaking loosely , turing machines whose heads move only in one direction .
sequential machines are therefore less power computers compared to general turing machines ; however , a number of real world phenomena and processes can be modelled by sequential machines since the latter can be considered as ( non - autonomous ) discrete dynamical systems .
that is why the theory of functions computable by sequential machines , which constitutes a substantial part of automata theory , has numerous applications not only in mathematics itself ( e.g. , in real analysis , @xmath8-adic analysis , number theory , complexity theory , dynamics , etc . ) but also in computer science , physics , linguistics and in many other sciences , see e.g. monographs @xcite for details and references .
the paper was motivated by empirical data obtained during a research project related to an applied problem which assumed intensive computer experiments with automata modelling of various cryptographic primitives used in stream ciphers , hash functions , etc .
word transformations performed by the automata where visualised , namely , represented by points of the unit square @xmath30\times[0,1]$ ] in real plane @xmath3 so that coordinates of the points relate numerical ( radix ) representations of input words to the numerical representations of corresponding output words .
it was noticed that once the modelled system was finite - state , and once input words were taken sufficiently long , some linear structures ( looking like segments of straight lines and somewhat resembling pictures from a double - slit experiment in quantum physics , cf . figures [ fig : plot-16][fig : plot-17 ] and figure [ fig:2slit ] ) may appear in the graph , but more complicated structures like smooth curves of higher order had never been observed .
a particular aim of the paper is to give mathematical explanation of the phenomenon and to characterize these linear structures .
but during the research it became evident that the problem ( which actually is a question what smooth real functions can be modelled on finite automata ) has applications not only to cryptography ( see e.g. ( * ? ? ?
* chapter 11 ) ) but also may be related to mathematical formalism of quantum theory . as a matter of fact , the latter relation ( which we believe does exist ) can be regarded as a yet another answer to the following question discussed by a. khrennikov in a series of papers devoted to so - called prequantum classical statistical field theory , see e.g. @xcite : _ why mathematical formalism of quantum theory _ ( which is based on the theory of linear operators on hilbert spaces ) _ is essentially linear although a number of quantum phenomena demonstrate an extremely non - linear behavior ? _
thus the goal of the paper is twofold : * firstly , to characterize real functions which can be computed by finite automata ; and * secondly , to give ( using obtained description of the functions ) some mathematical reasoning why wave phenomena are inherent in quantum systems .
the major part of the paper focuses on real functions which can be computed by finite automata while the said mathematical reasoning is considered in a closing section which contains a discussion of possible applications of mathematical results of the paper to quantum theory .
we are not going to discuss cryptographic applications here ; they will be postponed to forthcoming papers . in the paper , by a general automaton ( whose set of states is not necessarily finite ) we mean a machine which performs letter - by - letter transformations of words over input alphabet into words over output alphabet : once a letter is feeded to the automaton , the automaton updates its current state ( which initially is fixed and so is the same for all input words ) to the next one and produces corresponding output letter .
both the next state and the output letter depend both on the current state and on the input letter .
therefore each letter of output word depends _ only _ on those letters of input word which have already been feeded to the automaton .
an input word is a finite sequence of letters ; the letters can naturally be ascribed to ` causes ' while letters of the corresponding output word can be regarded as effects .
` causality ' just means that effects depend only on causes that ` already have happened ' ; therefore an automaton is an adequate mathematical formalism for a specific manifestation of causality principle once we assume that there exist only finitely many causes and effects , cf .
, e.g.,@xcite .
when studying real functions that can be computed by an automaton @xmath0 whose input / output alphabets are @xmath31 ( where @xmath32 is an integer from @xmath33 ) most authors follow common approach which described in e.g. ( * ? ? ?
* chapter xiii , section 4 ) : they associate an infinite word @xmath34 over @xmath6 to a real number whose base-@xmath8 expansion is @xmath35 and consider a real function @xmath36 defined as follows : given @xmath37 $ ] , take its base-@xmath8 expansion @xmath38 ; then produce an infinite output sequence @xmath39 of @xmath0 by successfully feeding the automaton with the letters @xmath40 , @xmath41 , etc . , and put @xmath42 .
being feeded by infinite input sequence @xmath34 , the automaton @xmath0 produces a unique infinite output sequence @xmath39 ; therefore the function @xmath36 is well defined everywhere on the real closed unit interval ( segment ) @xmath43 $ ] with the exception of maybe a countable set @xmath44 $ ] of points ; namely , of those having two distinct base-@xmath8 expansions @xmath45 .
the point set @xmath46\}$ ] can be considered as a graph of the real function @xmath36 specified by the automaton @xmath0 ( every time , before being feeded by the very first letter of each infinite input word the automaton @xmath0 is assumed to be in a fixed state @xmath47 , the _ initial state _ ) .
indeed , @xmath48 is defined uniquely for @xmath49\setminus d$ ] and @xmath48 can be ascribed to at most two values for @xmath50 ; so @xmath36 can be treated a real function which is defined on the unit segment @xmath14 $ ] and has not more that a countable number points of discontinuity in @xmath14 $ ] . in the sequel
we refer @xmath51 as to the _ monna graph _ of the automaton @xmath0 , cf .
subsection [ ssec : plots ] . the said common approach ( and its various generalisations )
is utilised in numerous papers , see e.g. @xcite . speaking loosely , the common approach looks as if one feeds the automaton @xmath0 by a base-@xmath8 expansion of a real number @xmath37 $ ] so that _ leftmost digits are feeded to the automaton prior to rightmost ones _ and observes output as real numbers since the automaton outputs accordingly leftmost digits of the base-@xmath8 expansion of @xmath52 $ ] prior to rightmost ones thus ascribing to the automaton @xmath11 the real function @xmath36 .
we stress that the function @xmath36 is well defined almost everywhere on @xmath14 $ ] due to _ namely that order _ in which digits of base-@xmath8 expansion are feeded to ( and outputted from ) the automaton @xmath0 .
a crucial difference of the approach used in our paper from the mentioned one is that _ the order we feed digits to ( and read digits from ) the automaton is inverse _ : namely , 1 . given a real number @xmath37 $ ] , we represent @xmath53 via base-@xmath8 expansion @xmath54 ( we take both expansions if @xmath53 has two distinct ones ) ; 2 . from the base-@xmath8 expansion @xmath55 we derive corresponding sequence @xmath56 of words ; then 3 . feeding the automaton @xmath11 successively by the words
@xmath56 so that _ rightmost letters are feeded to @xmath0 prior to leftmost ones _ we obtain corresponding output word sequence @xmath57 ; 4 . to the output sequence we put into a correspondence the sequence @xmath58 of rational numbers
whose base-@xmath8 expansions are @xmath59 thus obtaining a point set @xmath60 in the real unit square @xmath30\times[0,1]$ ] ; after that 5 .
we consider the set @xmath61 of _ all _ cluster points of the sequence @xmath62 ; 6 .
finally , we specify a
_ real plot _ ( or , briefly , a plot ) of the automaton @xmath11 as a union @xmath63,y\in\mathcal f(x)}((x;y)\cup{\euscript}x(x))$ ] . in other words
, @xmath15 is a closure in the unit square @xmath64 of the union @xmath65 where @xmath66 is the _
@xmath67-th layer _ of the plot @xmath15 .
that is , the plot @xmath68 can be considered as a ` limit ' of the sequence of sets @xmath69 , the _
approximate plots _ at word length @xmath70 , while @xmath71 ( see more formal definitions in subsection [ ssec : plots ] ) .
note that according to automata 0 - 1 law ( cf .
* proposition 11.15 ) and @xcite ) the plot @xmath15 of arbitrary automaton @xmath0 can be of two kinds only : either @xmath72 or @xmath21 is a ( lebesgue ) measure-0 closed subset of @xmath3 .
moreover , if the number of states of the automaton @xmath0 is finite ( further in the paper these automata are referred to as _ finite _ ones ) , then the second case takes place . we stress crucial advantage of real plots over monna graphs : in a contrast to the monna graph @xmath51 , a real plot @xmath21 is capable of showing true long - term behavior of automaton @xmath0 ( i.e. , when @xmath0 is feeded by sufficiently long words ) rather than a short - term behaviour displayed by the monna graph @xmath73 since due to the very construction of the real plot the higher order ( i.e. , the most significant ) digits of the real number represented by the output word are formed by the latest outputted letters of the output word whereas the construction of the monna graph assumes that the higher order digits are formed by the earliest outputted letters .
this results in a drastically different appearances of the real plot and of the monna graph : real plot clearly demonstrates that corresponding automaton is ` ultimately linear ' ( that is , exhibits linear long - term behavior ) , cf .
figures [ fig : plot-16][fig : plot ] ; whereas the monna graph is incapable to reveal this important feature of the automaton , cf .
figure [ fig : monna ] .
this is the main reason why in the paper we focus on real plots of automata rather than on their monna graphs .
therefore when specifying a notion of computability of a real - valued function @xmath74 $ ] ( where @xmath75 $ ] ) on automata , at least two different approaches do exist : the first one is to speak of the case when the graph @xmath76 of the function @xmath77 lie completely in @xmath78 for some automaton @xmath0 while the second one is to consider the case when @xmath79 .
papers @xcite mentioned above basically deal with the computability in the first meaning whereas our s paper deals with the computability of the second kind . note that classes of real functions which are computable on finite automata are different depending on the meaning : for instance ,
the function @xmath80 ( where @xmath81 stands for the integral part of @xmath82 , i.e. , for the biggest integer not exceeding @xmath83 ) is not computable in the first meaning but is computable in the second one whilst the function @xmath84 is computable in the first meaning but is not computable in the second one . to the best of our knowledge , our approach ( which is based on real plots rather than on monna graphs ) was originally used in @xcite and was never considered before by other authors . in the sequel we refer real functions @xmath74 $ ] with domain @xmath75 $ ] as to _ finitely computable _ if there exists a finite automaton @xmath0 whose real plot contains the graph of the function @xmath85 ; i.e. , if @xmath86 . main result of our paper is theorem [ thm : main ] which characterizes all finitely computable @xmath19-functions @xmath85 defined on a sub - segment @xmath87 $ ] : the theorem yields that _ if a finitely computable function @xmath88 $ ] is twice differentiable and if its second derivative is continuous everywhere on @xmath89 then @xmath85 is necessarily affine of the form @xmath90 for suitable rational @xmath8-adic @xmath91 _ ( that is , for @xmath92 , @xmath93 which can be represented by irreducible fractions whose denominators are co - prime to @xmath8 ) .
moreover , this is true in @xmath94-dimensional case as well ( theorem [ thm : main - mult ] ) . in view of theorem [ thm : main ]
it is noteworthy that despite the classes of functions computable on finite automata are different depending on the meaning the computability is understood , nonetheless if a function @xmath95{\rightarrow}[0,1]$ ] is everywhere differentiable on @xmath96 and @xmath97 for some finite automaton @xmath11 with binary input / output alphabets then @xmath85 is necessarily affine , see @xcite . in @xcite it is shown that a similar assertion holds for multivariate continuously differentiable functions and arbitrary finite alphabets . therefore finite automata should be judged as rather ` weak computers ' in all meanings since only quite simple real functions can be evaluated on these devices . from this view , results of the current paper are some contribution to the theory of computable real functions .
it is worth mentioning right now that actually our proof reveals a basic reason why smooth functions which can be represented by finite automata are necessarily affine : this is because _ squaring can not be performed by a finite automaton _ ; that is , an automaton which , being feeded by a base-@xmath8 expansion of @xmath94 , outputs a base-@xmath8 expansion of @xmath98 for every positive integer @xmath94 , can not be finite ( the latter is a well - known fact from automata theory , see e.g. , ( * ? ? ?
* theorem 2.2.3 ) ) .
it is also worth mentioning that the question when @xmath99 is somewhat easier to handle than the question when @xmath79 .
indeed , in the first case once the automaton @xmath0 is feeded by an infinite word @xmath100 , the word is treated as a base-@xmath8 expansion of a unique real number @xmath101 $ ] , corresponding output of @xmath0 is also an infinite word @xmath102 which also is treated as a base-@xmath8 expansion of a unique real number @xmath103 $ ] .
this results in a unique point @xmath104 of the unit square @xmath105 in the first case ; whilst in the second case the automaton @xmath0 , being feeded by the infinite word @xmath100 , produces generally an infinite point set of a cardinality continuum : the set is a closure of the point set @xmath106 in @xmath64 . due to this reason during the proofs we have to use more complicated techniques from real analysis which in some cases we combine with methods of @xmath8-adic analysis .
therefore some proofs are involved ; but to make general idea of a proof as transparent as possible in the sequel we explain it in loose terms when appropriate .
last but not least : our approach reveals another important feature of smooth functions which can be computed on finite automata . from figure
[ fig : plot ] it can be clearly observed that limit points of the plot constitute a torus winding if one converts a unit square into torus by gluing together opposite sides of the square .
this is not occasional : our theorem [ thm : main ] yields that _ if the unit square @xmath64 is mapped onto a torus @xmath107 , the smooth curves from the plot become torus windings _ ; and these _ windings after being represented in cylindrical coordinates are described by complex - valued functions @xmath29 _ ( @xmath37 $ ] ) , see corollary [ cor : mult - add - compl ] .
but in quantum theory the latter exponential functions are ascribed to matter waves ( cf . , de broglie waves ) ; therefore , since automata can be considered as models for discrete casual systems , the _ results of our paper give some mathematical evidence that matter waves are inherent in quantum systems merely due to causality principle and discreteness of matter _ ( quantization ) .
we discuss these possible connections to physics in section [ sec : concl ] .
note that for not to overload the paper with extra calculations we consider only automata whose input and output alphabets consist of @xmath8 letters @xmath108 where @xmath32 is a prime number though our approach can be expanded to the case when @xmath8 is arbitrary integer greater than 1 ( and even to the case when @xmath8 is not necessarily an integer , see section [ sec : concl ] ) . for a prime @xmath8
, we naturally associate when necessary letters of the alphabet @xmath108 to residues modulo @xmath8 , i.e. , to elements of a finite field @xmath18 .
the paper is organized as follows : * in section [ sec : prelm ] we recall basic definitions as well as some ( mostly known ) facts from combinatorics of words , from automata theory , from @xmath8-adic analysis , and from knot theory .
also in this section we formally introduce the notion of real plot of automaton and examine its basic properties . * in section [ sec : aff ] we completely describe cluster points of real plots of finite autonomous automata and of finite affine automata : we show that the points constitute links of torus knots . * in section [ sec : fin - comp ]
we prove numerous ( mostly technical ) results on finitely computable functions ; that is , on real functions whose graphs lie in plots of finite automata . loosely speaking ,
in the section we ( rigorously ) develop techniques to examine real functions computed on finite automata as if the automata are feeded by base-@xmath8 expansions of real arguments of the functions so that less significant digits are feeded to automaton prior to more significant ones . *
section [ sec : main ] contains main results of the paper : we prove that once a finitely computable function is @xmath19-smooth than it is affine and may be associated to a finite collection of complex - valued functions @xmath109 , @xmath110 for suitable rational numbers @xmath91 which are @xmath8-adic integers .
we prove a multivariate version of the theorem as well . * in section [ sec : concl ] we discuss possible connections of the main results to informational interpretation of quantum theory .
we argue that the results show that wave function is a mathematical consequence of two basic assumptions which are causality principle and discreteness of matter : we show that using @xmath111-expansions of real numbers ( where @xmath112 and @xmath113 is small ) rather than base-@xmath8 expansions for positive integer @xmath32 , main results of the paper imply that a quantum system may be considered as a finite automaton which calculates functions @xmath114 ; but the functions are approximately equal to @xmath115 when @xmath116 since @xmath117 is small and thus @xmath118 ; moreover , @xmath112 implies that both input and output alphabets of the automaton must be necessarily binary , i.e. , @xmath119 .
therefore one may say that the automaton produces waves @xmath115 ( since variables @xmath120 may be regarded as ` position ' and ` time ' respectively ) from bits .
this may serve a mathematical evidence in favour of j. a. wheeler s _ it from bit _ doctrine which suggests that all things physical ( ` its ' ) are information - theoretic in origin ( ` from bits ' ) , @xcite .
technically the paper is a sort of interplay between real analysis and @xmath8-adic analysis ; but although real analysis is the tool we mostly use in proofs , in some important places we also use @xmath8-adic analysis to examine specific properties of automata maps since the maps actually are 1-lipschitz functions w.r.t .
@xmath8-adic metric .
this is why we first recall some facts about words over a finite alphabet , @xmath8-adic integers , and automata .
an _ alphabet _ is just a finite non - empty set @xmath6 ; further in the paper usually @xmath121 .
elements of @xmath6 elements are called _ symbols _ , or _ letters_. by the definition , a _ word of length @xmath94 over alphabet @xmath6 _ is a finite sequence ( stretching from right to left ) @xmath122 , where @xmath123 . the number @xmath94 is called the _ length _ of the word @xmath124 and is denoted via @xmath125 .
the _ empty word _
@xmath126 is a sequence of length 0 , that is , the one that contains no symbols .
given a word @xmath124 , any word @xmath127 , @xmath128 , is called a _ prefix _ of the word @xmath16 ; whereas any word @xmath129 , @xmath130 is called a _ suffix _ of the word @xmath16 .
every word @xmath131 where @xmath132 is called a _
subword _ of the word @xmath124 . given words
@xmath133 and @xmath134 , the _ concatenation _ @xmath135 is the following word ( of length @xmath136 ) : @xmath137 given a word @xmath16 , its @xmath138-times concatenation is denoted via @xmath139 : @xmath140 we denote via @xmath17 the set of all non - empty words over @xmath31 and via @xmath141 the set of all words including the empty word @xmath126 . in the sequel
the set of all @xmath94-letter words over the alphabet @xmath18 we denote as @xmath142 ; so @xmath143 .
to every word @xmath124 we put into the correspondence a non - negative integer @xmath144 .
thus @xmath145 maps the set @xmath17 of all non - empty finite words over the alphabet @xmath6 onto the set @xmath146 of all non - negative integers . we will also consider a map @xmath147 of the set @xmath17 into the real unit half - open interval @xmath148 ; the map @xmath147 is defined as follows : given @xmath149 , put @xmath150 we also use the notation @xmath151 for @xmath152 .
along with finite words we also consider ( left-)infinite words over the alphabet @xmath6 ; the ones are the infinite sequences of the form @xmath153 where @xmath154 , @xmath155 . for infinite words
the notion of a prefix and of a subword are defined in the same way as for finite words ; whilst suffix is not defined .
let an infinite word @xmath16 be eventually periodic , that is , let @xmath156 for @xmath157 ; then the subword @xmath158 is called a _ period _ of the word @xmath16 and the suffix @xmath159 is called the pre - period of the word @xmath16 .
note that a pre - period may be an empty word while a period can not .
we write the eventually periodic word @xmath16 as @xmath160 .
see @xcite for introduction to @xmath8-adic analysis or comprehensive monographs @xcite for further reading . fix a prime number @xmath8 and denote respectively via @xmath161 and @xmath162 the set of all positive rational integers and the ring of all rational integers . given @xmath163 ,
the _ @xmath8-adic absolute value _ of @xmath94 is @xmath164 , where @xmath165 is the largest power of @xmath8 which is a factor of @xmath94 ; so @xmath166 where @xmath167 is co - prime to @xmath8 . by putting @xmath168 , @xmath169 and @xmath170 for @xmath171 ,
@xmath172 we expand the @xmath8-adic absolute value to the whole field @xmath173 of rational numbers .
given an absolute value @xmath174 , we define a metric in a standard way : @xmath175 is a
_ @xmath8-adic metric _ on @xmath173 .
the field @xmath176 of _ @xmath8-adic numbers _ is a completion of the field @xmath173 of rational numbers w.r.t .
the @xmath8-adic metric while the ring @xmath177 of _ @xmath8-adic integers _ is a ring of integers of @xmath176 ; and the ring @xmath177 is a completion of @xmath178 w.r.t .
the @xmath8-adic metric .
the ring @xmath177 is compact w.r.t .
the @xmath8-adic metric : actually @xmath177 is a ball of radius 1 centered at 0 ; namely @xmath179 . balls in @xmath176 are _ clopen _ ; that is , both closed and open w.r.t .
the @xmath8-adic metric .
a @xmath8-adic number @xmath180 admits a unique _ @xmath8-adic canonical expansion _
@xmath181 where @xmath182 , @xmath183 , @xmath184 .
note that then any @xmath8-adic integer @xmath185 admits a unique representation @xmath186 for suitable @xmath187 .
the latter representation is called a _
canonical form _ ( or , a _
canonical representation _ ) of the @xmath8-adic integer @xmath185 ; the @xmath67-th coefficient @xmath188 of the expansion will be referred to as _
the @xmath67-th @xmath8-adic digit _ of @xmath189 and denoted via @xmath190 .
it is clear that once @xmath191 , the @xmath67-th @xmath8-adic digit @xmath192 of @xmath189 is just the @xmath67-th digit in the base-@xmath8 expansion of @xmath189 .
note also that a @xmath8-adic integer @xmath185 is a _ unity _ of @xmath177 ( i.e. , has a multiplicative inverse @xmath193 ) if and only if @xmath194 ; so any @xmath8-adic number @xmath195 has a unique representation of the form @xmath196 where @xmath197 is a unity .
the @xmath8-adic integers may be associated to infinite words over the alphabet @xmath198 as follows : given a @xmath8-adic integer @xmath185 , consider its canonical expansion @xmath199 ; then denote via @xmath200 the infinite word @xmath153 ( allowing some freedom of saying we will sometimes refer @xmath200 as to a _ base-@xmath8 expansion of @xmath185 _ ) .
vice versa , given a left - infinite word @xmath201 we denote via @xmath202 corresponding @xmath8-adic integer whose base-@xmath8 expansion is @xmath16 thus expanding the mapping @xmath145 defined in subsection [ ssec : word ] to the case of infinite words as well .
it is worth noticing here that addition and multiplication of @xmath8-adic integers can be performed by using the same school - textbook algorithms for addition / multiplication of non - negative integers represented via their base-@xmath8 expansions with the only difference : the algorithms are applied to infinite words that correspond to @xmath8-adic canonical forms of summands / multipliers rather than to a finite words which are base-@xmath8 expansions of summands / multipliers . given @xmath203 and a canonical expansion @xmath186 for @xmath185 , denote @xmath204 .
the mapping @xmath205 is a ring epimorphism of @xmath177 onto the residue ring @xmath206 ( under a natural representation of elements of the residue ring by the least non - negative residues @xmath207 ) .
the series in the right - hand side of the canonical form converges w.r.t .
the @xmath8-adic metric ; that is , the sequence of partial sums @xmath208 converges to @xmath189 w.r.t .
the @xmath8-adic metric : @xmath209 .
it is worth noticing here that arbitrary infinite series @xmath210 where @xmath211 converges in @xmath176 ( i.e. , w.r.t .
@xmath8-adic metric ) if and only if @xmath212 since @xmath8-adic metric is _ non - archimedean _ ; that is , it satisfies _ strong triangle inequality _
@xmath213 for all @xmath214 .
note that @xmath191 if and only if all but a finite number of coefficients @xmath188 in the canonical form are 0 while @xmath215 if and only if all but a finite number of @xmath188 are @xmath216 .
further we will need a special representation for _ @xmath8-adic integer rationals _ ; that is , for those rational numbers @xmath189 which at the same time are @xmath8-adic integers , i.e. , for @xmath217 .
note that @xmath217 if and only if @xmath189 can be represented by an irreducible fraction @xmath218 , @xmath219 where @xmath220 is co - prime to @xmath8 . the following proposition is well known , cf . ,
e.g. , ( * ? ? ?
* theorem 10 ) : [ prop : p - adic - rat ] a @xmath8-adic integer @xmath189 is rational if and only if the sequence of coefficients of its canonical form is eventually periodic : @xmath221 for suitable @xmath222 , @xmath223 , @xmath224 . in other words ,
once a @xmath8-adic integer @xmath189 is represented in its canonical form , @xmath225 , the corresponding infinite word @xmath226 is eventually periodic : @xmath227 .
it is clear that given @xmath217 , both @xmath228 and @xmath229 are not unique : for instance , @xmath230 where @xmath231 .
but once both pre - periodic and periodic parts ( the prefix @xmath232 and the word @xmath233 ) are taken the shortest possible , both the _ pre - period length _ @xmath228 and the _ period length _
@xmath229 are unique for a given @xmath8-adic rational integer @xmath217 ; we refer to @xmath232 and to @xmath234 as to _ pre - period _ of @xmath189 and _ period _ of @xmath189 accordingly .
given @xmath217 we mostly assume further that in the representation @xmath235 ( respectively , in eventually periodic infinite word @xmath236 that corresponds to @xmath189 ) @xmath228 is a pre - period length and @xmath229 is a period length . note that a pre - period may be an empty word ( i.e. , of length 0 ) while a period can not .
rational @xmath8-adic integers can also be represented as fractions of a special kind : [ prop : p - repr ] a @xmath8-adic integer @xmath185 is rational if and only if there exist @xmath224 , @xmath237 , @xmath238 such that @xmath239 indeed , @xmath217 if and only if @xmath189 is of the form ; therefore @xmath240 where @xmath241 is a base-@xmath8 expansion of the least non - negative residue @xmath242 of @xmath243 modulo @xmath244 .
[ note : inverse ] recall that @xmath245 , for every @xmath246 .
[ note : pq - frac - rep ] note that once in @xmath228 is a pre - period length and @xmath229 is a period length of @xmath217 , the representation is unique ; that is , the choice of @xmath247 and @xmath248 in is unique . in the sequel we often use base-@xmath8 expansions of @xmath8-adic rational integers reduced modulo 1 ( recall that if @xmath249 then by the definition @xmath250 ) along with their @xmath8-adic canonical forms . for reader
s convenience , we now summarize some facts on connections between these representations .
it is very well known that a base-@xmath8 expansion of a rational number is eventually periodic ; that is , given @xmath251 $ ] , the base-@xmath8 expansion for @xmath53 is @xmath252 where @xmath253 .
note that in the base-@xmath8 expansions of rational integers from @xmath14 $ ] we use _
right_-infinite words rather than left - infinite ones that correspond to canonical expansions of @xmath8-adic integers .
[ prop : r - p - repr - qz ] given @xmath217 , represent @xmath189 in the form ; then @xmath254 where @xmath255 for @xmath256 and @xmath257 is the least non - negative residue of @xmath228 modulo @xmath229 if @xmath258 or @xmath259 if otherwise . indeed , by note [ note : inverse ] , @xmath260 in @xmath177 ; so @xmath261 where @xmath262 and @xmath263 .
therefore @xmath264 but @xmath265 in @xmath266 ; so @xmath267 and thus @xmath268 .
now just note that @xmath269 for @xmath270 , @xmath271 ; so @xmath272 and therefore @xmath273 where @xmath274 for @xmath275 .
combining with proposition [ prop : p - repr ] we see that all real numbers whose base-@xmath8 expansions are purely periodic must lie in @xmath276 ; therefore the following criterion is true : [ cor : r - repr ] a real number @xmath53 is in @xmath276 if and only if base-@xmath8 expansion of @xmath277 is purely periodic : @xmath278 for suitable @xmath279 .
the following corollary expresses base-@xmath8 expansion of a @xmath8-adic rational integer via its representation in the form given by proposition [ prop : p - repr ] : [ cor : r - p - repr - qz ] once a @xmath8-adic rational integer @xmath217 is represented in the form as of proposition [ prop : p - repr ] then @xmath280 where @xmath281 . indeed , under notation of proposition [ prop : p - repr ] , @xmath282 and the result follows since @xmath265 in @xmath266 .
now we can find a period length of @xmath217 provided @xmath189 is represented as an irreducible fraction @xmath218 , where @xmath283 , @xmath284 .
[ prop : mult - per ] once a @xmath8-adic rational integer @xmath285 is represented as an irreducible fraction @xmath218 , and if @xmath286 , then the period length @xmath229 of @xmath189 is equal to the multiplicative order of @xmath8 modulo @xmath220 .
note that the multiplicative order @xmath287 of @xmath8 modulo @xmath220 is the smallest positive integer such that @xmath288 .
indeed , @xmath289 for a suitable @xmath290 ; so @xmath291 . on the other hand , if @xmath292 for some @xmath293 then @xmath294 and thus @xmath295 since @xmath83 is co - prime to @xmath220 ( as the fraction @xmath296 is supposed to be irreducible ) .
now , from corollary [ cor : r - p - repr - qz ] it immediately follows that @xmath297 once @xmath229 is a period length of @xmath189 and that @xmath229 is the smallest positive integer with that property .
finally we conclude that @xmath298 .
now given @xmath284 , @xmath220 co - prime to @xmath8 , we denote via @xmath299 the multiplicative order of @xmath8 modulo @xmath220 if @xmath286 or put @xmath300 once @xmath301 . then @xmath299 is the period length of @xmath217 once @xmath189 is represented as an irreducible fraction @xmath218 where @xmath283 and @xmath284 . note that we consider here only infinite words that correspond to @xmath8-adic rational integers ; thus to , e.g. , 0 there corresponds a word @xmath302 ( so a period of 0 is 0 and a pre - period is empty ) and the respective base-@xmath8 expansion of 0 is @xmath303 . also , @xmath304 , the corresponding infinite word is @xmath305 ; therefore @xmath306 is a pre - period of 1 , @xmath307 is a period of 1 , and the representation of 1 in the form is @xmath308 .
let @xmath309 ; then @xmath310 is a canonical 2-adic expansion of @xmath311 ; so the corresponding infinite binary word is @xmath312 .
therefore the period length of @xmath311 is 2 ( and note that the multiplicative order of @xmath313 modulo @xmath314 is indeed @xmath313 ) , the period is @xmath315 , the pre - period is @xmath306 .
also , @xmath316 and @xmath317 once @xmath311 is represented in the form of proposition [ prop : p - repr ] ; @xmath318 is a base-2 expansion of @xmath311 , cf .
proposition [ prop : r - p - repr - qz ] and corollary [ cor : r - p - repr - qz ] . here
we remind some basic facts from automata theory ( see e.g. monographs @xcite ) . by the definition ,
a ( non - initial ) automaton is a 5-tuple @xmath319 where @xmath320 is a finite set , the _ input alphabet _ ; @xmath321 is a finite set , the _ output alphabet _ ; @xmath322 is a non - empty ( possibly , infinite ) set of _ states _ ; @xmath323 is a _ state transition function _ ; @xmath324 is an _
output function_. an automaton where both input alphabet @xmath320 and output alphabet @xmath321 are non - empty is called a _ transducer _ , see e.g. @xcite . the _ initial automaton _
@xmath325 is an automaton @xmath0 where one state @xmath326 is fixed ; it is called the _ initial state_. we stress that the definition of an initial automaton @xmath327 is nearly the same as the one of _ mealy automaton _
( see e.g. @xcite ) with the only important difference : the set of states @xmath322 of @xmath327 is _ not necessarily finite_. note also that in literature the automata we consider in the paper are also referred to as _
( letter - to - letter ) transducers _ ; in the sequel we use terms ` automaton ' and ` transducer ' as synonyms . given an input word @xmath328 over the alphabet @xmath320
, an initial transducer @xmath325 transforms @xmath16 to output word @xmath329 over the output alphabet @xmath321 as follows ( cf . figure [ fig : transd - sc ] ) : initially the transducer @xmath327 is at the state @xmath47 ; accepting the input symbol @xmath330 , the transducer outputs the symbol @xmath331 and reaches the state @xmath332 ; then the transducer accepts the next input symbol @xmath333 , reaches the state @xmath334 , outputs @xmath335 , and the routine repeats . this way
the transducer @xmath336 defines a mapping @xmath337 of the set @xmath338 of all @xmath94-letter words over the input alphabet @xmath320 to the set @xmath339 of all @xmath94-letter words over the output alphabet @xmath321 ; thus @xmath0 defines a map of the set @xmath340 of all non - empty words over the alphabet @xmath320 to the set @xmath341 of all non - empty words over the alphabet @xmath321 .
we will denote the latter map by the same symbol @xmath342 ( or by @xmath343 if we want to stress what initial state is meant ) , and when it is clear from the context what alphabet @xmath6 is meant we use notation @xmath17 rather than @xmath344 .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ( 1,2.5)(20,13 ) ( 12,10)1 ( 12,4)1 ( 12,11)(12,12 ) ( 18,7)(13,7 ) ( 6,10)(11,10 ) ( 6,4)(11,4 ) ( 4,7)(6,7 ) ( 6,4)(6,10 ) ( 18,12)(18,7 ) ( 12,12)(18,12 ) ( 12,8)(12,9 ) ( 12,3)(12,2 ) ( 12,2)(18,2 ) ( 12,5)(12,6 ) ( 11,6)(13,8 ) [ 180](12.4,7)@xmath345 [ 180](3.5,7)@xmath346 [ 90](12,9.7)@xmath347 [ 90](12,2.7)@xmath348 [ 0](11.7,11.3)@xmath349 [ 0](11.7,12.7)state transition [ 0](0,8)input [ 0](11.5,2.5)@xmath350 [ 0](17.5,2)@xmath351 [ 0](18,3)output _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ throughout the paper , ` automaton ' mostly stands for ` initial automaton ' ; we make corresponding remarks if not .
further in the paper we mostly consider transducers .
furthermore , throughout the paper we consider _ reachable transducers _ only ; that is , _ we assume that all states of the initial transducer @xmath327 are _ reachable from _ the initial state @xmath47 _ : given @xmath352 , there exists input word @xmath16 over alphabet @xmath320 such that after the word @xmath16 has been feeded to the automaton @xmath327 , the automaton reaches the state @xmath242 .
a reachable transducer is called _ finite _ if its set @xmath322 of states is finite , and transducer is called _ infinite _ if otherwise . to the initial automaton @xmath327 we put into a correspondence a family @xmath353 of all _ sub - automata _ @xmath354 , @xmath352 , where @xmath355 is the set of all states that are reachable from the state @xmath242 and @xmath356 are respective restrictions of the state transition and output functions @xmath357 on @xmath358 .
a sub - automaton @xmath359 is called _ proper _ if the set @xmath360 of all its states is a proper subset of @xmath322 .
a sub - automaton @xmath359 is called _ minimal _ if it contains no proper sub - automata .
it is obvious that a finite sub - automaton is minimal if and only if every its state is reachable from any other its state .
the set of all states of a minimal sub - automaton of the automaton @xmath0 is called an _ ergodic component _ of the ( set of all states ) of the automaton @xmath0 .
it is clear that once the automaton is in a state that belongs to an ergodic component , all its further states will also be in the same ergodic component .
therefore all states of a finite automaton are of two types only : the _ transient states _ which belong to no ergodic component , and _ ergodic states _ which belong to ergodic components .
it is clear that the set of all ergodic states is a disjoint union of ergodic components .
note that we use the term ` minimal automaton ' in a different meaning compared to the one used in automata theory , see , e.g. , @xcite : our terminology here is from the theory of markov chains , see , e.g. , @xcite ( since to the graph of state transitions of every automaton there corresponds a markov chain ) .
hereinafter in the paper the word ` automaton ' stands for a letter - to - letter initial transducer whose input and output alphabet consists of @xmath8 symbols , and we mostly assume that @xmath8 is a prime .
thus , for every @xmath361 the automaton @xmath362 maps @xmath94-letter words over @xmath18 to @xmath94-letter words over @xmath18 according to the procedure described above , cf .
figure [ fig : transd - sc ] . given two such automata @xmath363 and @xmath364 , their _ sequential composition _ ( or briefly , a _ composition _ ) @xmath365 can be defined in a natural way via sending output of the automaton @xmath0 to input of the automaton @xmath366 so that the mapping @xmath367 the automaton @xmath368 performs is just a composite mapping @xmath369 ( cf . any of monographs @xcite for exact definition and further facts mentioned in the subsection ) . note that a composition of finite automata is a finite automaton . in a similar manner one can consider automata with multiply inputs / outputs ; these can be also treated as automata whose input / output alphabets are cartesian powers of @xmath18 : for instance , and automaton with @xmath4 inputs and @xmath94 outputs over alphabet @xmath18 can be considered as an automaton with a single input over the alphabet @xmath370 and a single output over the alphabet @xmath371 . moreover , as the letters of the alphabet @xmath372 are in a one - to - one correspondence with residues modulo @xmath373 ; the automaton with @xmath4 inputs and @xmath94 outputs can be considered ( if necessary ) as an automaton with a single input over the alphabet @xmath374 and a single output over alphabet @xmath206 .
is an automaton with 2 inputs and 1 output which compositions of automata with multiple inputs / outputs can also be naturally defined : for instance , given automata @xmath375 , @xmath376 , and @xmath377 with @xmath378 inputs and @xmath379 outputs respectively , in the case when @xmath380 one can consider a composition of these automata by connecting every output of automata @xmath375 and @xmath376 to some input of the automaton @xmath381 so that every input of the automaton @xmath381 is connected to a unique output which belongs either to @xmath375 or to @xmath376 but not to the both . this way one obtains various compositions of automata @xmath375 and @xmath376 , with the automaton @xmath377 , and either of these compositions is an automaton with @xmath382 inputs and @xmath383 outputs . moreover , either of the compositions is a finite automaton if all three automata @xmath375 , @xmath376 , @xmath377 are finite .
automata can be considered as ( generally ) non - autonomous dynamical systems on different configuration spaces ( e.g. , @xmath142 , @xmath17 , etc . ) ; the system is autonomous when neither the state transition function @xmath322 nor the output function @xmath321 depend on input ; in this case the automaton is called _ autonomous _ as well . for purposes of the paper
it is convenient to consider automata with input / output alphabets @xmath384 as dynamical systems on the space @xmath177 of @xmath8-adic integers , i.e. , to relate an automaton @xmath0 to a special map @xmath385 . in the next subsection
we recall some facts about the map @xmath386 .
we identify @xmath94-letter words over @xmath18 with non - negative integers in a natural way : given an @xmath94-letter word @xmath387 ( i.e. , @xmath388 for @xmath389 ) , we consider @xmath16 as a base-@xmath8 expansion of the number @xmath390 . in turn , the latter number can be considered as an element of the residue ring @xmath206 modulo @xmath391 .
we denote via @xmath392 an inverse mapping to @xmath145 .
the mapping @xmath392 is a bijection of the set @xmath393 onto the set @xmath142 of all @xmath94-letter words over @xmath18 .
as the set @xmath207 is the set of all non - negative residues modulo @xmath391 , to _ every automaton @xmath394 there corresponds a map @xmath395 from @xmath206 to @xmath206 _
, for every @xmath361 .
namely , for @xmath396 put @xmath397 , where @xmath342 is a word transformation of @xmath142 performed by the automaton @xmath0 , cf .
subsection [ ssec : auto ] .
speaking less formally , the mapping @xmath395 can be defined as follows : given @xmath398 , consider a base-@xmath8 expansion of @xmath228 , read it as a @xmath94-letter word over @xmath1 ( put additional zeroes on higher order positions if necessary ) and then feed the word to the automaton so that letters that are on lower order positions ( ` less significant digits ' ) are feeded prior to ones on higher order positions ( ` more significant digits ' ) . then read the corresponding output @xmath94-letter word as a base-@xmath8 expansion of a number from @xmath399 keeping the same order , i.e. when the earliest outputted letters correspond to lowest order digits in the base-@xmath8 expansion .
we stress the following determinative property of the mapping @xmath395 which follows directly from the definition : given @xmath400 , _ whenever @xmath401 for some @xmath402 then necessarily @xmath403_. this implication may be re - stated in terms of @xmath8-adic metric as follows : latexmath:[\[\label{eq : n - compat } furthermost , _ every automaton @xmath336 defines a mapping @xmath405 from @xmath177 to @xmath177 _ which can be specified in a manner similar to the one of the mapping @xmath395 : given an infinite word @xmath406 ( that is , an infinite sequence ) over @xmath18 we consider a @xmath8-adic integer whose @xmath8-adic canonical expansion is @xmath407 ; so , by the definition , for every @xmath185 we put @xmath408 where @xmath409 , @xmath410 , and @xmath192 is the @xmath67-th @xmath8-adic digit of @xmath189 ; that is , the @xmath67-th term coefficient in the @xmath8-adic canonical representation of @xmath189 : @xmath411 , @xmath412 ( see subsection [ ssec : p - adic ] ) .
the so defined map @xmath405 is called the _ automaton function _ ( or , the _ automaton map _ ) of the automaton @xmath0 .
note that from it follows that @xmath413 where @xmath414 is a map from the @xmath415-th cartesian power @xmath416 of @xmath18 into @xmath18 .
more formally , given @xmath185 , define @xmath417 as follows : consider a sequence @xmath418 and a corresponding sequence @xmath419 ; then , as the sequence @xmath418 converges to @xmath189 w.r.t .
@xmath8-adic metric ( cf . subsection [ ssec : p - adic ] ) , the sequence @xmath419 in view also converges w.r.t .
the @xmath8-adic metric ( since the latter sequence is fundamental and @xmath177 is closed in @xmath176 which is a complete metric space ) .
now we just put @xmath417 to be a limit point of the sequence @xmath419 .
thus , the mapping @xmath420 is a well - defined function with domain @xmath177 and values in @xmath177 ; by the function @xmath420 satisfies lipschitz condition with a constant 1 w.r.t .
@xmath8-adic metric .
the point is that _ the class of all automata functions that correspond to automata with @xmath8-letter input / output alphabets coincides with the class of all maps from @xmath177 to @xmath177 that satisfy the @xmath8-adic lipschitz condition with a constant 1 _ ( the 1-lipschitz maps , for brevity ) , cf .
, e.g. , @xcite .
we note that the claim can also be derived from a more general result on asynchronous automata ( * ? ? ?
* proposition 3.7 ) ; for @xmath309 the claim was proved in @xcite .
further we need more detailed information about _ finite automata functions _ , that is , about functions @xmath421 where @xmath363 is a finite automaton ( i.e. , with a finite set @xmath322 of states ) .
it is well known ( cf .
previous subsection [ ssec : auto ] ) that _ the class of finite automata functions is closed w.r.t .
composition of functions and a sum of functions _ : once @xmath422 are finite automata functions , either of mappings @xmath423 and @xmath424 @xmath425 is a finite automaton function . another important property of finite automata functions is that _ any finite automaton function maps @xmath276 into itself_. in view of , the latter property is just a re - statement of a a well - known property of finite automata which yields that any finite automaton feeded by an eventually periodic sequence outputs an eventually periodic sequence , cf . , e.g. , ( * ? ? ?
* corollary 2.6.9 ) , ( * ? ? ?
* chapter xiii , theorem 2.2 . ) .
since further we often use that property of finite automata , we state it as a lemma for future references : [ le : per ] if a finite automaton @xmath0 is being feeded by a left - infinite periodic word @xmath426 , where @xmath427 is a finite non - empty word , then the corresponding output left - infinite word is eventually periodic ; i.e. , it is of the form @xmath428 , where @xmath429 , @xmath430 . to put it in other words , if a finite automaton is being feeded by an eventually periodic finite word @xmath431 , where @xmath427 , @xmath432 , and @xmath402 is sufficiently large , then the output word is of the form @xmath433 , where @xmath434 , @xmath429 , @xmath435 and @xmath228 is either empty or a prefix of @xmath436 : @xmath437 for a suitable @xmath438 .
therefore the output word is of the form @xmath439 , where @xmath440 is a cyclically shifted word @xmath436 . to study finite automata functions
it is convenient sometimes to represent 1-lipschitz maps from @xmath177 to @xmath177 as special convergent @xmath8-adic series , the _ van der put series_. details about the latter series may be found in , e.g. , @xcite ; here we only briefly recall some basic facts .
given a continuous function @xmath441 , there exists a unique sequence @xmath442 of @xmath8-adic integers such that @xmath443 for all @xmath444 , where @xmath445 and @xmath446 if @xmath447 ; @xmath94 is uniquely defined by the inequality @xmath448 otherwise .
the right side series in is called the _ van der put series _ of the function @xmath449 . note that the sequence @xmath450 of _ van der put coefficients _ of the function @xmath449 tends @xmath8-adically to @xmath307 as @xmath451 , and the series converges uniformly on @xmath177 .
vice versa , if a sequence @xmath450 of @xmath8-adic integers tends @xmath8-adically to @xmath307 as @xmath451 , then the the series in the right part of converges uniformly on @xmath177 and thus define a continuous function @xmath452 .
the number @xmath94 in the definition of @xmath453 has a very natural meaning ; it is just the number of digits in a base-@xmath8 expansion of @xmath454 : @xmath455 therefore @xmath456 for all @xmath454 ( that is why we assume @xmath457 ) .
note that coefficients @xmath458 are related to the values of the function @xmath449 in the following way : let @xmath459 be a base-@xmath8 expansion for @xmath4 , i.e. , @xmath460 , @xmath461 and @xmath462 , then @xmath463 it worth noticing also that @xmath464 is merely a characteristic function of the ball @xmath465 of radius @xmath466 centered at @xmath454 : @xmath467 [ thm : vdp - comp ] a function @xmath441 is 1-lipschitz if and only if @xmath449 can be represented as @xmath468 where @xmath469 for @xmath470 by using the van der put series it is possible to determine whether a mapping @xmath471 is an automaton function of a finite automaton .
we first remind some notions and facts from the theory of automata sequences following @xcite .
an infinite sequence @xmath472 over a finite alphabet @xmath6 , @xmath473 , is called _ @xmath8-automatic _ if there exists a finite transducer @xmath474 such that for all @xmath475 , if @xmath476 is feeded by the word @xmath477 which is a base-@xmath8 expansion of @xmath478 , @xmath479 if @xmath480 , then the @xmath138-th output symbol of @xmath476 is @xmath481 ; or , in other words , such that @xmath482 for all @xmath483 , where @xmath484 and @xmath485 stands for the @xmath138-th digit in the base-@xmath486 expansion of @xmath223 .
a _ @xmath8-kernel _ of the sequence @xmath487 is a set @xmath488 of all subsequences @xmath489 , @xmath470 , @xmath490 .
[ thm : p - ker ] a sequence @xmath487 is @xmath8-automatic if and only if its @xmath8-kernel is finite .
[ thm : fin - auto ] let a 1-lipschitz function @xmath471 be represented by van der put series .
the function @xmath449 is a finite automaton function if and only if the following conditions hold simultaneously : 1 .
all coefficients @xmath491 , @xmath470 , constitute a finite subset @xmath492 , and 2 .
the @xmath8-kernel of the sequence @xmath493 is finite .
condition ( ii ) of the theorem is equivalent to the condition that the sequence @xmath493 is @xmath8-automatic , cf .
theorem [ thm : p - ker ] .
criteria to determine if an automaton function is finite which are based on expansions other than van der put are also known , cf .
@xcite . in literature ,
automata with multiple inputs and outputs over the same alphabet are also studied .
we remark that in the case when the alphabet is @xmath18 , the automata can be considered as automata whose input / output alphabets are cartesian powers @xmath371 and @xmath370 , for suitable @xmath494 . for these automata a theory similar to that of automata with a single input / output can be developed : corresponding automata function are then 1-lipshitz mappings from @xmath495 to @xmath496 w.r.t .
@xmath8-adic metrics .
recall that @xmath8-adic absolute value on @xmath497 is defined as follows : given @xmath498 , put @xmath499 . the so defined absolute value ( and the corresponding metric ) are non - archimedean as well .
the main theorem of the paper holds ( after a proper re - statement ) for these automata as well , see theorem [ thm : main - mult ] .
it is worth recalling here a well - known fact ( which also can be proved by using theorem [ thm : fin - auto ] ) that _ addition of two @xmath8-adic integers can be performed by a finite automaton with two inputs and one output _ : actually the automaton just finds successively ( digit after digit ) the sum by a standard addition - with - carry algorithm which is used to find a sum of two non - negative integers represented by base-@xmath8 expansions thus calculating the sum with arbitrarily high accuracy w.r.t .
the @xmath8-adic metric . on the contrary ,
_ no finite automaton can perform multiplication of two arbitrary @xmath8-adic integers _ since it is well known that no finite automaton can calculate a base-@xmath8 expansion of a square of an arbitrary non - negative integer given a base-@xmath8 expansion of the latter , cf . ,
e.g. , ( * ? ? ?
* theorem 2.2.3 ) . from these remarks combined with theorem [ thm : fin - auto ]
the following properties of finite automata functions can be deduced : [ prop : fin - auto ] let @xmath500 be finite automata , let @xmath501 be @xmath8-adic rational integers .
then the following is true : 1 . the mapping @xmath502 of @xmath177 into @xmath177 is a finite automaton function ; 2 . a composite function @xmath503 , @xmath504 ,
is a finite automaton function ; 3 .
a constant function @xmath505 is a finite automaton function if and only if @xmath506 ; 4 .
an affine mapping @xmath507 is a finite automaton function if and only if @xmath508 .
note that the van der put expansion of the constant function @xmath509 is @xmath510 while the van der put expansion for the identity function @xmath511 is @xmath512 where @xmath513 stands for the @xmath514-th digit in the base-@xmath8 expansion of @xmath67 .
now all statements of the proposition follow immediately from theorem [ thm : fin - auto ] and the above mentioned facts from finite automata theory .
note that the statement of proposition [ prop : fin - auto ] is known : for instance , it can be deduced from the old work @xcite of a. g. lunts . to our best knowledge ,
lunts was the first who revealed connections between automata theory and @xmath8-adic analysis .
it is worth noticing that lunts defines automata functions in a slightly different way than we do : in his work , an automaton function is a 1-lipschitz function @xmath515 such that @xmath516 for all @xmath195 .
also , lunts methods of proofs are completely different form the ones of proposition [ prop : fin - auto ] .
unfortunately , most automata theorists seem to be unaware of the paper @xcite since it was never translated into english and even was never reviewed by mathematical reviews . concluding the subsection
, we remark that in literature ( finite ) automata functions are also known under names of _
( bounded ) determinate functions _ , or _
( bounded ) deterministic functions _ , cf .
, e.g. , @xcite .
further in the paper we consider special representation of automata functions by point sets of real and complex spaces .
as we have already mentioned in previous section , several representations of this sort were considered : via the so - called limit sets ( see e.g. @xcite ) , via the monna graphs ( see e.g. @xcite ) and via real plots which were originally introduced in ( * ? ? ?
* chapter 11 ) . in the paper we focus on real plots
; however we will start this subsection with saying few words about monna graphs since in some meaning they are counterpart of real plots ; and we will not touch limit sets at all since they are standing somewhat apart .
the monna graphs are based on the monna s representation of @xmath8-adic integers via real numbers of the unit closed segment @xmath14 $ ] originally suggested by monna in @xcite : given a canonical expansion @xmath186 of @xmath8-adic integer @xmath185 ( cf .
subsection [ ssec : p - adic ] ) , consider a real number @xmath517\subset{\mathbb r}$ ] .
it is clear that @xmath518 maps @xmath177 onto @xmath14 $ ] , however , @xmath518 is not bijective : the only points from the open interval @xmath96 that have more than one ( actually , exactly two ) pre - image w.r.t .
@xmath518 are rational numbers of the form @xmath519 where @xmath520 for some @xmath521 since @xmath522 where @xmath523 for all @xmath524 , @xmath525 for all @xmath526 and @xmath527 .
we can naturally associate the segment @xmath14 $ ] ( or a half - open interval @xmath148 ) to the real circle @xmath528 by _ reducing @xmath14 $ ] modulo 1 _ ; that is , by taking fractional parts of reals from @xmath14 $ ] : @xmath529{\mathbin{\mathsf{mod}}}1 $ ] . then in a similar manner we may consider a mapping of @xmath177 onto @xmath528 ; we will denote the mapping also via @xmath518 since there is no risk of misunderstanding .
note that w.r.t .
the latter mapping the point @xmath530 has exactly two pre - images since @xmath531 in @xmath528 .
now , given an automaton @xmath336 , we define the _ monna graph _ of @xmath0 as follows : let @xmath532 be a corresponding automaton function , cf . subsection [ ssec : a - map ] ( that is , @xmath471 is a 1-lipschitz function w.r.t .
@xmath8-adic metric ) .
then the monna graph @xmath533 ( or , which is the same , of the automaton function @xmath449 ) is the point set @xmath534 .
note that we can consider the monna graph when convenient either as a subset of the unit real square @xmath64 , a cartesian square of a unit segment , @xmath30\times[0,1]$ ] , or as a subset of a 2-dimensional real torus @xmath535 , a cartesian square of a real unit circle @xmath528 .
a monna graph can be considered as a graph of a real function @xmath536 defined on @xmath14 $ ] and valuated in @xmath14 $ ] .
indeed , given a point @xmath37 $ ] , which is not of the form , there is a unique @xmath185 such that @xmath537 .
therefore , @xmath538 is well defined at @xmath53 since there exists a unique @xmath539 $ ] such that @xmath540 ; so we just put @xmath541 .
once @xmath53 is of the form , then there exist exactly two @xmath542 , @xmath543 such that @xmath544 . as @xmath545 is not necessarily equal to @xmath546 , then @xmath538 may be not well defined at @xmath53 : one have to assign to @xmath547 both @xmath548 and @xmath549 which may happen to be non - equal . to make @xmath538 well defined on @xmath14 $ ] a usual way
is to admit only representations of one ( of two ) types for @xmath53 of the form ; say , only those with finitely many non - zero terms , cf .
, e.g. , @xcite . in this case
the function @xmath538 becomes well - defined everywhere on @xmath14 $ ] and having points of discontinuity at maybe the points of type only .
a typical monna graph of the function @xmath538 looks like the one represented by figure [ fig : monna ] .
now we are going to introduce a notion of the _ real plot _ of an automaton function ; the latter notion is somehow ` dual ' to the notion of monna graph . given an automaton @xmath363 , we associate to an @xmath4-letter non - empty word @xmath5 over the alphabet @xmath18 a rational number @xmath550 whose base-@xmath8 expansion is @xmath551 then to every @xmath4-letter input word @xmath10 of the automaton @xmath11 and to the respective @xmath4-letter output word @xmath12 ( rightmost letters are feeded to / outputted from the automaton prior to leftmost ones ) there corresponds a point @xmath13 of the real unit square @xmath64 ; then we define @xmath15 as a closure in @xmath3 of the point set @xmath13 where @xmath16 ranges over the set @xmath17 of all finite non - empty words over the alphabet @xmath18 . given an automaton function @xmath552 define a set @xmath553 of points of the real plane @xmath3 as follows : for @xmath554 denote @xmath555 a point set in a unit real square @xmath30\times[0,1]$ ] and take a union @xmath556 ; then @xmath557 is a closure ( in topology of @xmath3 ) of the set @xmath558 .
note that if @xmath559 is a @xmath8-adic canonical expansion of @xmath185 then @xmath560 , c.f .
; so @xmath561
. moreover , @xmath562 , see further note [ note : plot - auto - in ] .
[ def : plot - auto ] given an automaton @xmath0 , we call a _ plot of the automaton _
@xmath0 the set @xmath15 .
we call a _ limit plot _ of the automaton @xmath0 the point set @xmath563 which is defined as follows : a point @xmath564 lies in @xmath563 if and only if there exist @xmath185 and a strictly increasing infinite sequence @xmath565 of numbers from @xmath566 such that simultaneously @xmath567 [ note : plot - auto ] further in the paper we consider @xmath563 ( as well as @xmath21 and @xmath568 ) either as a subset of the unit square @xmath569 or as a subset of the unit torus @xmath570 when appropriate . note that when considering the plot on the unit torus we reduce coordinates of the points modulo 1 , that is , we just ` glue together ' 0 and 1 of the unit segment @xmath571 thus transforming it into the unit circle @xmath528 ( whose points we usually identify with the points of the half - open segment @xmath148 via a natural one - to - one correspondence , say , @xmath572 ) . also ,
sometimes we consider @xmath563 ( as well as @xmath21 and @xmath568 ) as a subset of the cylinder @xmath573 or of the cylinder @xmath574 by reducing modulo 1 either @xmath575- or @xmath53-coordinate respectively . we denote the corresponding plot via @xmath576 by using the subscript @xmath577 and we omit the subscript when it is clear ( or when it is no difference ) on which of the surfaces the plot is considered .
we take a moment to recall some well - known topological notions and to introduce some notation . in the sequel , given a subset @xmath347 of a topological ( in particular , metric ) space @xmath578 which satisfies the hausdorff axiom
we denote via @xmath579 the set of all accumulation points of @xmath347 .
recall that the point @xmath580 is called an _ accumulation point of _ @xmath581 once every neighborhood of @xmath53 contains infinitely many points from @xmath347 ; and a point @xmath582 is called _ isolated point of _ @xmath347 ( or , the point isolated from @xmath347 ; or , the point isolated w.r.t .
@xmath347 ) once there exists a neighborhood @xmath583 such that @xmath584 contains no points from @xmath347 other than ( maybe ) @xmath575 .
we may omit the subscript and use notation @xmath585 when it is clear from the context what metric space is meant .
we also write @xmath586 ( or briefly @xmath587 , or @xmath588 for the set of all limit points of the sequence @xmath589 over @xmath578 . recall that a point @xmath590 is called a limit ( or , cluster ) point of the sequence @xmath591 if every neighbourhood of @xmath53 contains infinitely many members of the sequence @xmath591 ; that is , given any neighborhood @xmath584 of @xmath53 , the number of @xmath67 such that @xmath592 is infinite ( note that the very @xmath592 are not assumed to be pairwise distinct points of @xmath578 ; some , or even all of them may be identical ) .
note that in topology the terms ` accumulation point of a set ' and ` limit point of a set ' are used as synonyms ; however to avoid possible misunderstanding we reserve the term ` limit point ' only for sequences while for sets we use the term ` accumulation point ' .
[ note : plot - auto - in ] the definition of @xmath15 immediately implies that @xmath593 if and only if there exists a sequence @xmath594 of finite non - empty words @xmath595 such that @xmath596 for all @xmath412 and @xmath597 , @xmath598 .
note that once @xmath599 then there exists a sequence @xmath594 of words such that the sequence @xmath600 of their lengths is strictly increasing : one just may take @xmath601 , cf . and subsection [ ssec : a - map ] . therefore @xmath602 .
moreover , from definition [ def : plot - auto ] it readily follows that @xmath603 since given a finite non - empty word @xmath16 and taking any @xmath185 such that the prefix of the corresponding infinite word is @xmath16 ( i.e. , such that @xmath604 ) we see that @xmath605 .
this implies that @xmath606 since @xmath607 ; so in the sequel we do not differ automata plots from the plots of automata functions and use both @xmath608 and @xmath553 as notation for the plot of the automaton @xmath11 .
also we may use notation @xmath609 along with @xmath610 to denote the limit plot of the automaton @xmath0 .
we stress here once again a crucial difference in the construction of plots and of monna graphs of automata : given a canonical expansion of @xmath8-adic integer @xmath559 we put into a correspondence to @xmath189 a _ single _ real number @xmath611 while constructing monna graphs ; whereas in the construction of plots we put into a correspondence to @xmath189 a _ whole set of all limit points _ of the sequence @xmath612 , and the latter set may not consist of a single point ; moreover , ` usually ' the set never consists of a single point since with a probability 1 the set is a whole segment @xmath14 $ ] . therefore to study structure of plots we need to work with sets of all limit points of ( usually non - convergent ) sequences rather than with limits of convergent sequences as in the case of monna maps .
[ prop : no - iso ] let @xmath0 be an arbitrary automaton ; then @xmath563 contains no points isolated w.r.t . @xmath613 .
let @xmath599 be a point isolated w.r.t .
@xmath614 . as @xmath599 ,
let @xmath615 be a @xmath8-adic canonical representation of the @xmath8-adic integer @xmath185 mentioned in definition [ def : plot - auto ] ; and let @xmath616 be a @xmath8-adic canonical expansion of the @xmath8-adic integer @xmath617 .
then as the point @xmath104 is isolated , there exists @xmath618 such that @xmath619 and @xmath620 for all @xmath621 , cf .
( if otherwise , the point @xmath622 is not isolated w.r.t .
@xmath614 ) . put @xmath623 , @xmath624 ; then @xmath625 for all @xmath626 .
we claim that then necessarily both @xmath627 and @xmath628 ( whence both @xmath629 and @xmath630 ) .
indeed , as the sequence @xmath631 is infinite and strictly increasing , then taking @xmath632 in we conclude that necessarily @xmath633 for all @xmath634 .
therefore , taking @xmath635 large enough so that @xmath636 ( which is always possible since the sequence @xmath637 is strictly increasing ) we see that @xmath638 and thus @xmath639 for all @xmath155 since @xmath640 for all @xmath626 by .
but this implies that @xmath627 ( whence @xmath629 ) . the same argument combined with shows that @xmath641 and @xmath630 .
consider now an automaton @xmath366 whose automaton function @xmath642 is defined as follows : given a @xmath8-adic canonical representation @xmath615 , let @xmath643 and @xmath644 for @xmath645 .
such an automaton @xmath366 exists since the so defined function @xmath642 satisfies and thus is 1-lipschitz , cf .
subsection [ ssec : a - map ] .
actually the automaton being feeded by the input word @xmath646 just put @xmath306 as the first output letter and put @xmath647 for the @xmath514-th output letter for @xmath645 where @xmath648 is the output word of the automaton @xmath0 feeded by the input word @xmath646 ; that is , @xmath366 outputs @xmath649 being feeded by @xmath646 . from [ eq : plot ] and definition [ def : plot - auto ] it immediately follows that @xmath650 and that a point @xmath564 is an isolated point of @xmath651 if and only if it is an isolated point of @xmath614 .
but by the claim we have proved above , once @xmath104 is an isolated point of @xmath651 , then necessarily @xmath652 .
but the first letter of any output word of automaton @xmath366 is 1 by the construction of @xmath653 ; thus @xmath654 and so @xmath655 .
from the claim we have proved at the beginning of the proof it follows now that @xmath656 can not contain isolated points of @xmath651 ; thus @xmath563 can not contain isolated points of @xmath614 .
note that proposition [ prop : no - iso ] only states that @xmath610 contains no points isolated from @xmath614 , but of course @xmath610 may contain isolated points w.r.t . itself .
for instance , let @xmath11 be a @xmath8-adic odometer ; that is , @xmath657 ( the automaton @xmath0 may be taken a finite then ) .
then the point @xmath658 is an isolated point of @xmath659 w.r.t .
@xmath659 ; however @xmath660 contains no points isolated w.r.t . @xmath660 .
[ thm : ap = lp ] if automaton @xmath0 is finite and minimal then @xmath661 . by proposition [ prop : no - iso ] ,
@xmath662 ; we need to prove that the inverse inclusion also holds .
let @xmath663 ; then there exists a sequence @xmath664 of @xmath8-adic integers and a sequence @xmath665 of integers from @xmath566 such that all the points @xmath666 are pairwise distinct and @xmath667 we may assume that the sequence @xmath665 is increasing since otherwise in the point sequence @xmath668 there are only finitely many pairwise distinct points .
moreover , we may assume that the sequence @xmath665 is strictly increasing ; we consider corresponding infinite point subsequence of @xmath668 if otherwise .
so we see that there exists an infinite sequence of words @xmath669 of strictly increasing lengths @xmath670 such that @xmath671 that is , there exists a sequence @xmath672 of words @xmath673 of strictly increasing lengths @xmath674 such that @xmath675 . from here
it follows that once @xmath67 is sufficiently large ( say , once @xmath676 ) then @xmath677 for a suitable @xmath678 . by the same reason , @xmath679 for a suitable
@xmath680 once @xmath67 is large enough ( say , once @xmath681 ) , etc .
moreover , we may assume that the sequence @xmath682 is strictly increasing .
therefore , @xmath683 . applying a similar argument to the sequence @xmath684 ( @xmath412 ) we conclude that there exists a strictly increasing sequence @xmath685 such that @xmath686 once @xmath687 and therefore @xmath688 .
moreover , by the construction of the sequences @xmath682 and @xmath685 we may assume that @xmath689 for all @xmath690 .
thus we have shown that @xmath691 where @xmath692 and @xmath693 , ( @xmath694 ) .
let @xmath695 be a state the automaton @xmath0 reaches after being feeded by the input word @xmath696 ( note that @xmath697 , the initial state , once @xmath696 is empty word ) . as the number of states of @xmath11 is finite , at least one state @xmath352 repeats in the sequence @xmath698 infinitely many times .
therefore @xmath699 denote @xmath700 , @xmath701 , ( @xmath694 ) . as every state of the automaton @xmath0 is reachable from the initial state @xmath47 ,
there exists a word @xmath702 such that the if the automaton @xmath0 ( which is initially at the state @xmath47 ) has been feeded by the word @xmath703 , then @xmath0 outputs the word @xmath704 and reaches the state @xmath242 .
thus @xmath705 , and the automaton @xmath0 after being feeded by the word @xmath706 reaches the state @xmath707 .
as the automaton @xmath11 is minimal , there exists a word @xmath708 such that once the automaton @xmath709 has been feeded by the word @xmath710 , the automaton reaches the state @xmath242 .
now being feeded by the word @xmath711 , the automaton @xmath712 outputs the word @xmath713 and reaches a state @xmath714 . by the minimality of @xmath0 ,
there exists a word @xmath715 such that after @xmath716 has been feeded by the word @xmath717 , the automaton reaches the state @xmath242 .
now after @xmath718 has been feeded by the word @xmath719 , the automaton @xmath718 reaches the state @xmath720 , and we can find a word @xmath721 in a manner similar to that of described .
now being feeded by the so constructed left - infinite word @xmath722 , the automaton @xmath0 outputs the left - infinite word @xmath723 where @xmath724 , @xmath725 .
now consider @xmath8-adic integers @xmath726 and @xmath727 which correspond to infinite words @xmath722 and @xmath723 accordingly ; that is , @xmath728 and @xmath729 .
then , by the construction we have that @xmath730 , and from it follows that @xmath731 where @xmath732 .
as the sequence @xmath733 is strictly increasing , from it follows now that @xmath599 in view of definition [ def : plot - auto ] .
it is well known ( see e.g. ( * ? ? ?
* ch.2 , exercise 2 ) ) that the set of all accumulation points of a hausdorff topological space ( the _ derived set _ of the space ) is a closed subset of the space . from theorem [ thm :
ap = lp ] it follows that once a finite automaton is minimal then its limit plot is a derived set of its plot ( whence , closed ) : [ cor : ap = lp ] let an automaton @xmath0 be finite and minimal ; then the set @xmath610 is a derived set of @xmath15 and therefore is closed in @xmath3 .
a point @xmath564 belongs to @xmath563 if and only if there exists a sequence @xmath734 of finite non - empty words of strictly increasing lengths @xmath735 such that the sequence @xmath736 tends to @xmath53 and the corresponding sequence @xmath737 tends to @xmath575 as @xmath738 , where @xmath739 are respective output words of the automaton @xmath0 that correspond to input words @xmath740 .
we stress once again that words @xmath741 are feeded to the automaton @xmath11 from right to left ; i.e. the letter @xmath742 is feeded to @xmath0 first , then the letter @xmath40 is feeded to @xmath0 , etc . by the definition ,
the set @xmath743 is a derived set of @xmath15 ; whence by theorem [ thm : ap = lp ] the set @xmath610 is a derived ( thus , closed ) set of @xmath2 .
the necessity of conditions of the corollary follows immediately from definition [ def : plot - auto ] since once @xmath744 then there exist a @xmath8-adic integer @xmath745 and a strictly increasing sequence @xmath746 over @xmath566 such that holds ; that is , we just put @xmath747 and @xmath748 , where @xmath449 is an automaton function of @xmath0 , cf . note
[ note : plot - auto - in ] . to prove sufficiency of the conditions note that the conditions just yield that there exists an infinite sequence of words @xmath673 of strictly increasing lengths @xmath670 such that hold .
the argument that follows of the proof of theorem [ thm : ap = lp ] now proves the sufficiency .
it is worth noticing here that the limit plot of a finite minimal automaton does not depend on what state of the automaton is taken as initial : [ note : ap = lp ] if @xmath749 are states of a finite minimal automaton @xmath0 , @xmath750 , then @xmath751 . indeed , due to the minimality
, every state of @xmath0 is reachable from any other state of @xmath0 . therefore if @xmath752 then by definition [ def : plot - auto ] there exist @xmath185 and a strictly increasing infinite sequence @xmath565 of numbers from @xmath566 such that holds . by the minimality of @xmath0
, there exists a finite word @xmath16 of length @xmath753 such that after the automaton @xmath359 has been feeded by @xmath16 , it reaches the state @xmath229 .
now by substituting in definition [ def : plot - auto ] @xmath754 for @xmath189 and @xmath755 for @xmath565 we see that holds and therefore @xmath756 .
using an idea similar to that of note [ note : ap = lp ] it can be easily demonstrated that if @xmath366 is a sub - automaton of @xmath0 then @xmath757 since every state of the automaton @xmath0 is reachable from its initial state : [ note : sub - auto ] let @xmath758 be a sub - automaton of the automaton @xmath0 . as the initial state @xmath242 of the automaton
@xmath366 is reachable from the initial state @xmath47 of the automaton @xmath0 , from the definition of the respective sets it immediately follows that @xmath759 , @xmath760 , and @xmath761 .
the following useful lemma is a sort of a counter - part of lemma [ le : per ] in terms of points from @xmath563 rather than in terms of words .
[ le : qint - lp ] given a finite automaton @xmath0 and a point @xmath762 , if @xmath744 for some @xmath249 then @xmath763 . as @xmath599
then there exist @xmath185 and a strictly increasing sequence @xmath764 over @xmath566 such that holds .
therefore there exists an infinite sequence of words @xmath765 of strictly increasing lengths @xmath670 such that hold simultaneously . now repeating for the case @xmath766 the argument that follows in the proof of theorem [ thm :
ap = lp ] we conclude that hold in our case as well ( note that nowhere in the mentioned argument from the proof of theorem [ thm : ap = lp ] we used that @xmath0 is minimal ) .
moreover , in the notation of the argument , there exists a strictly increasing sequence @xmath682 over @xmath566 such that @xmath767 but @xmath768 do not depend on @xmath67 since in our case @xmath769 ( where @xmath770 ) for all @xmath67 ; therefore @xmath771 for all @xmath772 . as @xmath683 ( cf . ) and @xmath762 then the right - infinite word @xmath773 must be purely periodic ( cf .
corollary [ cor : r - repr ] ) with a period @xmath774 of length @xmath775 : that is , @xmath776 . now in put @xmath777 ; then for every @xmath246 we have that @xmath778 for all @xmath779 . now denote via @xmath780 the state the automaton @xmath0 reaches after have been feeded by the word @xmath781 .
fix @xmath246 and denote @xmath782 a state which occurs in the sequence @xmath783 infinitely often ; due to the finiteness of the automaton @xmath0 such state exists .
denote @xmath784 the smallest @xmath779 such that @xmath785 ( therefore @xmath786 ) . and
again due to the finiteness of the automaton @xmath0 in the sequence @xmath787 some state ( say , @xmath242 ) occurs infinitely often .
let @xmath788 be the corresponding infinite ( thus , strictly increasing ) subsequence , i.e. , @xmath789 ; then as the sequence @xmath682 is strictly increasing and as @xmath786 , in the sequence @xmath790 there exists a strictly increasing subsequence , say @xmath791 ( note that the sequence @xmath792 is also strictly increasing ) .
now from it follows that once being feeded successfully by purely periodic words @xmath793 for @xmath794 , @xmath795 , the automaton @xmath359 outputs the words @xmath796 .
now by combining lemma [ le : per ] with corollary [ cor : r - repr ] we conclude that if @xmath197 is such that @xmath797 then @xmath798 and @xmath799 . yet one more property of automata plots is their invariance with respect to @xmath8-shifts .
that is , _ given a point @xmath800 , take base-@xmath8 expansions @xmath801 , @xmath802 of coordinates @xmath803 ; then @xmath804 .
_ to put it in other words , the following proposition is true : [ prop : p - shift ] for an arbitrary automaton @xmath0 , if @xmath805 resp .
, @xmath806 then @xmath807 resp . , @xmath808 .
the first statement follows immediately from note [ note : plot - auto - in ] since once @xmath809 as @xmath738 then necessarily @xmath810 as @xmath738 . to prove the second statement ,
let @xmath811 be an automaton function of the automaton @xmath0 .
as @xmath599 , there exists @xmath185 and a strictly increasing sequence @xmath665 over @xmath566 such that @xmath812 and @xmath813 , cf .
definition [ def : plot - auto ] .
therefore @xmath814 as @xmath815 once @xmath816 is a @xmath8-adic canonical representation for @xmath185 . by the same reason , @xmath817
. therefore @xmath808 by definition [ def : plot - auto ] .
it is known that the plot @xmath818 of the automaton @xmath0 can be of two types only ; namely , _ given an automaton @xmath0 , the set @xmath15 either coincides with the whole unit square @xmath64 or @xmath15 is nowhere dense in @xmath105 _ : being closed in @xmath3 , the set @xmath15 is measurable w.r.t .
lebesgue measure on @xmath3 , and the measure of @xmath15 is 1 if and only if @xmath819 and is 0 if otherwise : the later assertion is a statement of automata 0 - 1 law , cf .
* proposition 11.15 ) and @xcite .
moreover , _ once an automaton @xmath0 is finite , the measure of @xmath68 is 0 and @xmath15 is nowhere dense in @xmath105 _ ( cf . op .
therefore , _ plots of finite automata are lebesgue measure 0 nowhere dense closed subsets of the unit square @xmath64 _ ; thus they can not contain sets of positive measure , but they may contain lines .
the goal of the paper is to prove that if @xmath0 is a finite automaton then smooth curves which lies completely in @xmath68 ( thus in @xmath563 , cf .
further theorem [ thm : main ] ) can only be straight lines .
moreover , we will prove that _ if finite automata plots are considered as subsets of the unit torus @xmath820 in @xmath821 then smooth curves lying in the plots can only be torus windings_. for this purpose we will need some extra information ( which follows ) about torus knots .
further in the paper we will need only few concepts concerning torus knots theory ; details may be found in numerous books on knot theory , see e.g. @xcite . for our purposes
it is enough to recall only two notions , the knot and the link .
recall that a knot is a smooth embedding of a circle @xmath528 into @xmath821 and a link is a smooth embedding of several disjoint circles in @xmath821 , cf .
we will consider only special types of knots and links , namely , torus knots and torus links .
informally , a torus knot is a smooth closed curve without intersections which lies completely in the surface of a torus @xmath822 , and a link ( of torus knots ) is a collection of ( possibly knotted ) torus knots , see e.g. ( * ? ? ?
* section 26 ) for formal definitions .
we also need a notion of a cable of torus .
formally , a cable of torus is any geodesic on torus . recall that geodesics on torus @xmath820 are images of straight lines in @xmath3 under the mapping @xmath823 of @xmath3 onto @xmath824 , cf . ,
e.g. , ( * ? ? ?
* section 5.4 ) .
[ def : cable ] a _ cable of the torus _ is an image of a straight line in @xmath3 under the map @xmath825 of the euclidean plain @xmath3 onto the 2-dimensional real torus @xmath826 .
if the line is defined by the equation @xmath827 we say that @xmath83 is a _ slope _ of the cable @xmath828 .
we denote via @xmath829 a cable which corresponds to the line @xmath830 , the _ meridian _ , and say that the slope is @xmath831 in this case .
cables @xmath832 of slope 0 ( i.e. , the ones that correspond to straight lines @xmath833 ) are called _
parallels_. in dynamics , cables of torus @xmath820 are viewed as orbits of _ linear flows on torus _ ; that is , of dynamical systems on @xmath820 defined by a pair of differential equations of the form @xmath834 on @xmath820 , whence , by a pair of parametric equations @xmath835 in cartesian coordinates , cf .
e.g. ( * ? ? ?
* subsection 4.2.3 ) .
[ note : cable ] it is well known that a cable defined by the straight line @xmath827 is dense in @xmath820 if and only if @xmath836 and the slope @xmath837 is irrational , see e.g. ( * ? ? ?
* proposition 4.2.8 ) or ( * ? ? ?
* section 5.4 ) . given a cartesian coordinate system @xmath838 of @xmath821 ,
a torus can be obtained by rotation around @xmath839-axis of a circle which lies in the plain @xmath840 .
if a radius of the circle is @xmath228 and the circle is centered at a point lying in @xmath841-axis at a distance @xmath842 from the origin , then in cylindrical coordinates @xmath843 of @xmath821 ( where @xmath844 is a radius - vector in cartesian coordinate system @xmath845 , @xmath846 is an angle of the radius - vector in coordinates @xmath845 , @xmath189 is a @xmath839-coordinate in cartesian coordinate system @xmath838 ) the torus is defined by the equation @xmath847 and a cable ( with a rational slope @xmath848 where @xmath849 and @xmath850 ) of the torus is defined by the system of parametric equations ( with parameter @xmath851 ) of the form @xmath852=\left [ \begin{array}{c } r+r\cos \left ( \frac{\alpha}{\beta}t+\omega \right ) \\
t\\ r\sin \left ( \frac{\alpha}{\beta}t+\omega \right ) \end{array } \right],\ ; t\in{\mathbb r}. \ ] ] the cable defined by the above equations winds @xmath111 times around @xmath839-axis and @xmath853 times around a circle in the interior of the torus ( the sign of @xmath854 determines whether the rotation is clockwise or counter - clockwise ) , see for an example of the corresponding torus knot figures [ fig : lin - r ] and [ fig : lin - t ] where @xmath855 and @xmath856 . letting @xmath857 in the above equations
take a finite number of values we get an example of torus link , see e.g. figures [ fig : link - square ] and [ fig : link - torus ] which illustrate a link consisting of a pair of torus knots whose slopes are @xmath858 .
note that figures [ fig:2link - square ] and [ fig:2link - torus ] illustrate a union of two distinct torus links ( of two and of three knots respectively ) rather than a single torus link of 5 knots . finally , due to the above representation of a torus link in the form of equations in cylindrical coordinates ,
we naturally associate the torus link consisting of @xmath70 cables with slopes @xmath848 to a family of complex - valued functions @xmath859 of real variable @xmath851 @xmath860 where @xmath67 stands for imaginary unit @xmath861 : @xmath862 . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ( 0,0)(15,15 ) ( -4,0)(11,15 ) ( -4,10)(-1,15 ) ( -4,5)(2,15 ) ( -4,0)(5,15 ) ( -1,0)(8,15 ) ( 2,0)(11,15 ) ( 5,0)(11,10 ) ( 8,0)(11,5 ) _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ , scaledwidth=92.0% ]
in this section we completely describe limit plots of finite automata maps of the forms @xmath509 ( constant maps ) , @xmath863 ( linear maps ) and @xmath864 ( affine maps ) , where @xmath865 are some ( suitable ) @xmath8-adic integers and the variable @xmath189 takes values in @xmath177 .
recall that an automaton @xmath325 is called _ autonomous _ once neither its state update function @xmath347 nor its output function @xmath348 depend on input ; i.e. , when @xmath866 ( @xmath412 ) , cf .
[ fig : transd - sc ] .
it is clear that an autonomous automaton function is a constant ; however a limit plot of this function is not necessarily a straight line .
for instance , the limit plot of a constant @xmath867 is the whole unit square @xmath64 once @xmath868 where the infinite word @xmath869 over @xmath18 is such that every non - empty finite word @xmath870 over @xmath18 occurs as a subword in @xmath436 ; that is , if there exist a finite word @xmath871 and an infinite word @xmath242 over @xmath18 such that @xmath436 is a concatenation of @xmath871 , @xmath16 and @xmath242 : @xmath872 , cf .
@xcite .
on the other hand , once an autonomous automaton @xmath0 is finite , the corresponding infinite output word must necessarily be eventually periodic .
that is , @xmath873 for suitable @xmath874 ; therefore a finite autonomous automaton function is a rational constant , i.e. , @xmath506 , cf .
propositions [ prop : p - adic - rat ] and [ prop : fin - auto ]
. furthermore , the numbers that correspond to ( sufficiently long ) finite output words are then all the form @xmath875 for @xmath876 .
consequently , the limit plot of the automaton ( in @xmath3 ) consists of @xmath229 pairwise parallel straight lines which correspond to the numbers @xmath877 where @xmath876 , cf .
subsection [ ssec : plots ] ; or ( which is the same ) to the numbers @xmath878 .
that is , all the lines from the limit plot are @xmath879 , @xmath690 , for any line @xmath880 belonging to the limit plot ; thus the number of lines in the limit plot does not exceed @xmath229 . respectively , being considered as a point set on the torus @xmath820 , the limit plot consists of not more than @xmath229 parallels , cf .
, e.g. , figures [ fig : parallel - square ] and [ fig : parallel - torus ] .
now we present a more formal argument and derive a little bit more information about the number of lines in the limit plot .
given @xmath881 , represent @xmath882 as an irreducible fraction @xmath883 for suitable @xmath283 , @xmath284 .
note that @xmath884 since @xmath885 .
denote @xmath886 since @xmath887 , by proposition [ prop : p - adic - rat ] a @xmath8-adic canonical form of @xmath296 is @xmath888 for suitable @xmath889 , or , in other words , the infinite word that corresponds to @xmath296 is @xmath890 .
then from proposition [ prop : r - p - repr - qz ] it follows that @xmath891 where @xmath892 , @xmath893 , and @xmath257 is the least non - negative residue of @xmath228 modulo @xmath229 if @xmath258 or @xmath259 if otherwise . from here in view of
we deduce that @xmath894 and thus @xmath895 where @xmath896 ( cf .
proposition [ prop : p - repr ] and corollary [ cor : r - p - repr - qz ] ) .
now we can suppose that @xmath229 is a period length of the rational @xmath8-adic integer @xmath887 ( cf .
subsection [ ssec : p - adic ] ) ; then in view of proposition [ prop : mult - per ] we conclude that @xmath897 since @xmath898note that @xmath899 except of the case when @xmath900 and @xmath16 is a single - letter word that consists of the only letter @xmath216 ( in the latter case @xmath901 and thus @xmath902 ) .
similarly , @xmath903 except of the case when @xmath904 and thus @xmath905 ( so @xmath906 and @xmath907 ) . but this case happens if and only if @xmath296 for some @xmath908 ; i.e. , when @xmath909 .
we now summarize all these considerations in a proposition : [ prop : const ] let @xmath910 be an automaton function of a finite automaton @xmath0 ; then @xmath911 is a disjoint union of @xmath229 parallels @xmath912 , @xmath913 , and @xmath229 is a period length of @xmath882 . [
note : const ] in conditions of proposition [ prop : const ] the constant @xmath881 can be represented as an irreducible fraction @xmath883 where @xmath283 , @xmath284 , @xmath884 ( we put @xmath301 and @xmath914 if @xmath915 ) . then the limit plot @xmath911 is a torus link that consists of @xmath916 trivial torus cables ( parallels ) with slopes @xmath307 ; to the link there corresponds a collection of @xmath229 complex constants ( which are @xmath220-th roots of 1 ) @xmath917 where @xmath67 stands for imaginary unit @xmath918 : @xmath862 ( cf .
subsection [ ssec : knot ] ) .
being considered in the unit real square @xmath64 , the limit plot @xmath610 is a collection of @xmath916 segments of straight lines @xmath919 that cross @xmath64 , where @xmath920 here @xmath921 , @xmath922 , and a base @xmath8-expansion of @xmath436 is @xmath923 ( cf .
proposition [ prop : p - repr ] ) ; @xmath924 for @xmath925 .
in other words , all the constants @xmath926 are of the form @xmath927 where @xmath871 runs trough all cyclic shifts of the word @xmath928 ; that is , @xmath929 . if @xmath882 is represented in a @xmath8-adic canonical form rather than in a form of proposition [ prop : p - repr ] , then all the lines of the limit plot can be represented as @xmath930 note that we may omit @xmath931 in and in in all cases but the case when simultaneously the length @xmath229 of the period is 1 and @xmath932 ( respectively , @xmath933 ) ; but @xmath934 in that case and therefore @xmath935 . the following property of the set @xmath936 will be used in further proofs : [ cor : const ] given @xmath937 , the following alternative holds : either @xmath938 or @xmath939 .
the result is clear enough since the numbers that constitute @xmath936 are exactly all numbers whose base-@xmath8 expansions are of the form @xmath940 where @xmath436 runs through all cyclic shifts of the finite word @xmath16 which is the ( shortest ) period of @xmath941 , cf . note [ note : const ] ; nonetheless we give a formal proof which follows .
given @xmath942 , @xmath943 , represented as irreducible fractions @xmath944 whose denominators @xmath945 are co - prime to @xmath8 , let @xmath946 ; then @xmath947 for suitable @xmath948 . if @xmath949 then @xmath950 and thus @xmath938
let @xmath951 , then @xmath952 ; so since @xmath953 we conclude that @xmath954 for a suitable @xmath955 and @xmath956 such that @xmath957 . therefore necessarily @xmath958 where @xmath959 since @xmath960 .
but then we conclude that @xmath961 and therefore @xmath962 , @xmath963 where @xmath964 .
hence @xmath965 ; the inverse inclusion also holds since @xmath966 for any @xmath881 by .
e.g. , . [ ex : const ] let @xmath309 and @xmath967 . then @xmath968 and the limit plot consists of 3 lines .
the binary infinite word that corresponds to the 2-adic canonical representation of @xmath969 is @xmath970 , so the period of @xmath969 is @xmath971 , the pre - period is @xmath315 , and @xmath972 .
therefore the tree lines of the limit plot are : @xmath973 , @xmath974 , @xmath975 .
the limit plot ( on the unit square and on the torus ) is illustrated by figures [ fig : parallel - square ] and [ fig : parallel - torus ] accordingly .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ( 0,0)(15,15 ) ( -4.8,0)(13.2,18 ) ( -4.8,0.1).3 ( -4,12.857)(11,12.857 ) ( -4,12.143)(11,12.143 ) ( -4,6.43)(11,6.43 ) _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ , scaledwidth=92.0% ] in this subsection we consider limit plots of linear maps @xmath976 ( @xmath185 ) which are finite automaton functions . by proposition [ prop : fin - auto ]
, the latter takes place if and and only if @xmath506 . [ prop : mult - p ] given @xmath506 , represent @xmath977 , where @xmath283 , @xmath284 , @xmath978 are coprime , @xmath884 .
if @xmath0 is an automaton such that @xmath979 @xmath504 then @xmath980 since @xmath981 = c(c,0)@xmath982c@xmath983t^2@xmath984(a ) for every @xmath506 the automaton @xmath0 may be taken a finite . by proposition
[ prop : fin - auto],the map @xmath976 on @xmath177 is an automaton function of a finite automaton if and only if @xmath506 . given @xmath985 , take @xmath185 such that @xmath986 for a suitable strictly increasing sequence @xmath987 .
as @xmath506 , then @xmath988 for suitable @xmath989 , @xmath224 , @xmath990 , by proposition [ prop : p - repr ] .
if @xmath258 , then considering residues of @xmath670 modulo @xmath229 we see that at least one residue ( say , @xmath991 ) occurs in the sequence @xmath992 infinitely many times .
therefore @xmath993 for a respective strictly increasing sequence @xmath994 .
the latter equality trivially holds when @xmath900 : one just takes @xmath995 and @xmath996 .
so further we assume that @xmath997 , @xmath410 .
for @xmath410 we have that @xmath998 as @xmath999 and @xmath997 , we have that @xmath1000 note that the argument of @xmath1001 in the right - hand side of is negative once @xmath67 is sufficiently large ; therefore once @xmath67 is large enough then @xmath1002 for a suitable @xmath1003 which does not depend on @xmath67 ( actually it is not difficult to see that @xmath1004 ) .
thus , @xmath1005 firstly we note that given @xmath1006 , @xmath1007 @xmath1008 as @xmath244 is a factor of @xmath1009 .
secondly , put @xmath1010 , then @xmath1011 and @xmath1012 so in @xmath1013 and @xmath1014 ( recall that @xmath997 where @xmath991 ) .
let @xmath1015 be a base-@xmath1016 representation of @xmath1017 ( that is , @xmath1018 ) ; then by combining and with we get @xmath1019 where @xmath1020 stands for a @xmath1016-weight of a natural number , that is , the sum of digits of the number in its base-@xmath1016 representation ; i.e. , @xmath1021 .
therefore every limit point of the sequence @xmath1022 is of the form @xmath1023 for a suitable @xmath1024 .
we claim that , on the other hand , given @xmath985 and @xmath1025 ( that is , @xmath1026 lying in the ideal @xmath1027 of the residue ring @xmath1028 generated by @xmath871 ) there exists @xmath185 and a strictly increasing sequence @xmath1029 such that @xmath1030 indeed , take @xmath185 and @xmath997 as above ; then all limit points of the sequence @xmath1022 are of the form for , say , @xmath1031 . if @xmath1032 for some @xmath1033 , then there is nothing to prove ; if @xmath1034 for all @xmath1033 then we tweak @xmath189 as follows .
as the point of the form for @xmath1035 is a limit point of the sequence @xmath1022 then @xmath1036 occurs in the sequence @xmath1037 infinitely many times ( cf . )
; so some @xmath1038 such that @xmath1039 occurs in the sequence @xmath1040 infinitely many times : @xmath1041 for @xmath1042 ( @xmath1043 ) .
as @xmath1044 , then @xmath1045 for a suitable @xmath1046 .
now put @xmath1047 , where @xmath1048 ; then @xmath1049 .
but @xmath1050 ; so finally we conclude by that @xmath1051 thus we have shown that @xmath1052 but the right - hand side in is a cable of torus with slope @xmath247 since @xmath1053 indeed , if @xmath1054 for some @xmath1055 then @xmath1056 and @xmath1057 , and follows .
this concludes the proof .
[ exmp : lin ] take @xmath309 and @xmath1058 .
figures [ fig : lin - r ] and [ fig : lin - t ] illustrate the limit plot of the function @xmath1059 in @xmath64 and in @xmath820 respectively . in this subsection we combine the above two cases ( constant maps and linear maps ) into a single one to describe limit plots of finite automata whose functions
are affine , i.e. , of the form @xmath1060 ( @xmath185 ) .
it is evident that the limit plot should be a torus link consisting of several disjoint cables with slopes @xmath247 since the limit plot of the constant @xmath882 is a collection of parallels , cf .
propositions [ prop : mult - p ] and [ prop : const ]
. we will give a formal proof of this claim and find the number of knots in the link .
recall that by proposition [ prop : fin - auto ] the map @xmath1060 of @xmath177 into itself is an automaton function of some finite automaton if and only if @xmath1061 .
the following proposition shows that we do not alter the limit plot of the map once we replace @xmath882 by @xmath1062 for arbitrary @xmath1055 .
[ prop : add - p - md1 ] given @xmath1063 @xmath504 where @xmath1061 , denote @xmath1064 , @xmath1065 . then @xmath1066 . indeed , once @xmath1055 then @xmath1067 ; the limit is @xmath306 if and only if @xmath94 is negative since given a canonical @xmath8-adic representation @xmath1068 of a negative @xmath1055 , all @xmath520 if @xmath67 is large enough , cf .
subsection [ ssec : p - adic ] . therefore ( @xmath1069 for all @xmath1070 .
note that the map @xmath1071 from the statement of proposition [ prop : add - p - md1 ] is an automaton function for a suitable finite automaton @xmath1072 and @xmath650 , where @xmath11 is a finite automaton whose automaton function is @xmath449 .
now we describe limit plot of a special affine map with @xmath1073 .
[ lem : add - p ] given a finite automaton @xmath0 whose automaton function is @xmath1074 , the limit plot @xmath1075 is a link of a finite number of torus knots which are cables @xmath1076 where @xmath1077 is running over @xmath936 .
we will prove that once @xmath0 is a finite automaton such that @xmath1078 then @xmath1079 note that if @xmath1080 and @xmath913 then @xmath935 by proposition [ prop : const ] and there is nothing to prove .
so further we assume that @xmath913 and @xmath1081 . by proposition [ prop :
add - p - md1 ] we may assume that @xmath1082 then @xmath1083 for suitable @xmath1084 , cf .
proposition [ prop : p - repr ] ; that is , @xmath1085 , where @xmath1086 and therefore @xmath1087 as @xmath1088 in @xmath177 by note [ note : inverse ] .
therefore , in @xmath177 the rational number @xmath882 can be represented as @xmath1089 where @xmath1090 , @xmath1091 .
given @xmath985 take a sequence @xmath1092 , @xmath410 , and a strictly increasing sequence @xmath1093 , @xmath410 , such that @xmath1094 , @xmath1095 , and @xmath1096 for all @xmath410 .
this is always possible since if , e.g. , @xmath1097 for suitable @xmath1098 then one takes @xmath1099 where @xmath1100 and put @xmath1101 for @xmath410 .
considering a sequence @xmath1102 in @xmath177 , we see that @xmath1103 but @xmath1101 , @xmath1104 ; thus @xmath1105 if @xmath1106 , or @xmath1107 if @xmath1108 . from
and it follows that @xmath1109 by of note [ note : const ] .
thus we have proved that given @xmath985 and @xmath1110 , necessarily @xmath1111 ; so @xmath1112 for every @xmath1113 . on the other hand , given @xmath185 and a strictly increasing sequence @xmath987 , by combining and with of note [ note : const ] we conclude that all limit points of the sequence @xmath1114 , @xmath410 , are in @xmath1115 by an argument similar to the above one
. therefore , limit points of the sequence @xmath1116 , @xmath410 , are all of the form @xmath1117 , where @xmath53 is an limit point of the sequence @xmath1118 and @xmath1110 .
this proves that @xmath1119 and that is true .
now we are ready to prove the main claim of the section .
[ thm : mult - add - p ] given @xmath1120 , a map @xmath1121 of @xmath177 into itself is an automaton function of a finite automaton if and only if @xmath1061 .
given a finite automaton @xmath0 whose automaton function is @xmath1122 for @xmath1061 , represent @xmath1123 as irreducible fractions @xmath977 , @xmath1124 , where @xmath1125 , @xmath1126 and @xmath1127 ; then the limit plot @xmath1075 is a link of @xmath1128 torus knots , where @xmath1129 , and every knot of the link is a cable @xmath1130 for @xmath913 : @xmath1131 moreover , @xmath1132 for @xmath1133 if and only if @xmath1134 where @xmath1135 , @xmath943 , cf . .
[ note : mult - add - p ] once @xmath1136 , i.e. , once @xmath1137 , the congruence @xmath1138 holds trivially , @xmath1139 and the link consists of a single knot ; so in that case @xmath1140 for all @xmath1133 . the first statement of the theorem is already proved ,
see proposition [ prop : fin - auto ] . given @xmath1141 , we have that @xmath1142 for suitable @xmath989 , @xmath1143 , @xmath990 , @xmath1144 , by proposition [ prop : p - repr ] . note that we may assume that @xmath1145 since the set of all limit points of the sequence @xmath1146 is the same as that of the sequence @xmath1147 by proposition [ prop : add - p - md1 ] and the case @xmath915 is already considered , cf . proposition [ prop : mult - p ] .
now we will prove that @xmath1148 for @xmath913 .
as @xmath1149 , the canonical @xmath8-adic representation of @xmath882 is eventually periodic and the period length of @xmath882 is @xmath1150 , cf .
subsection [ ssec : p - adic ] .
now fix @xmath913 , take corresponding @xmath1151 and consider a sequence @xmath1152 @xmath1153 ; then @xmath1154 , cf .
the proof of lemma [ lem : add - p ] . given @xmath985 take @xmath185 and a sequence @xmath1155 as in the proof of proposition [ prop : mult - p ] ; so @xmath1156 . note that if @xmath1157 for some @xmath454 then @xmath1158 ; so the proof of proposition [ prop : mult - p ] remains valid if one substitutes @xmath1017 for @xmath189 and any strictly increasing subsequence @xmath1159 of the sequence @xmath1160 for the sequence @xmath637 .
we claim that for some @xmath454 there exist an increasing sequence @xmath1161 and a subsequence @xmath1162 of the sequence @xmath1163 such that @xmath1164 indeed , let @xmath1165 be the greatest common divisor of @xmath1150 and @xmath229 ; then @xmath1166 , @xmath1167 , @xmath1168 and @xmath1169 are co - prime .
as the infinite sequence @xmath1163 is strictly increasing , there exists @xmath1170 such that @xmath1171 for infinitely many @xmath1172 , say , for @xmath1173 .
put @xmath1174 ; @xmath1175 .
take the smallest @xmath1176 , @xmath483 , such that @xmath1177 , then put @xmath1178 and find @xmath1179 from the equation which now is equivalent to the equation @xmath1180 : as @xmath1181 for a suitable @xmath271 by the definition of @xmath1182 , one sees that @xmath1183 for @xmath1175 thus proving our claim .
we conclude now that given arbitrary @xmath249 and @xmath1110 there exist @xmath1184 and a sequence @xmath1185 such that @xmath1186 cf . , and proposition [ prop : const ]
. therefore , @xmath1187 and so the point @xmath1188 is in @xmath563 .
thus we have proved that @xmath1189 for every @xmath1110 . on the other hand ,
given arbitrary @xmath185 and arbitrary strictly increasing sequence @xmath987 , limit points of the point sequence @xmath1190 are all in @xmath1191 by proposition [ prop : mult - p ] whereas limit points of the sequence @xmath1114 are all in @xmath936 by proposition [ prop : const ] . therefore limit points of the point sequence @xmath1192 are all in @xmath1193 . finally we conclude that @xmath1194 ; or ( which is the same ) that @xmath1195 note that it may happen that @xmath1196 even if @xmath1197 ( and even @xmath1198 ) : for instance , shows that @xmath1199 for some @xmath1200 .
therefore to finish the proof we must now calculate the number of pairwise distinct cables @xmath1130 when @xmath913 . during the proof of proposition [ prop : mult - p ]
we have shown that ( in the notation of the proposition under the proof ) @xmath1201 for every @xmath1202 , cf .
equation and the text which follows it .
therefore @xmath1140 if @xmath1203 for some @xmath1202 .
the converse statement is also true : if @xmath1140 then @xmath1203 for some @xmath1202 . to prove this , for @xmath1204 let @xmath1205 be a set of all points where the cable @xmath1206 crosses zero meridian of the torus @xmath820 ; that is , @xmath1207 where @xmath1208 , @xmath1209 ; therefore by @xmath1210 finally , as @xmath1140 if and only if @xmath1211 since the both cables cross zero meridian at a same angle ( which is equal to @xmath1212 ) , this means that @xmath1140 if and only if @xmath1213 for some @xmath1214 , as claimed .
now we are able to calculate the number of torus knots ( cables ) which constitutes the link @xmath563 .
let for some @xmath1215 , ( @xmath1216 ) and @xmath1217 the following equality holds : @xmath1218 we see that @xmath1219 for suitable @xmath1220 by note [ note : const ] ( @xmath943 ) .
therefore is equivalent to the congruence @xmath1221 for a suitable @xmath1222 ; but the latter congruence in turn is equivalent to the congruence @xmath1223 where @xmath1224 , @xmath1225 , @xmath1226 ( we assume that @xmath1227 since the case @xmath1228 is trivial ) . from here it follows that @xmath1229 once @xmath1230 ; therefore necessarily @xmath1231 since @xmath1232 .
so @xmath1233 for a suitable @xmath1234 and thus is equivalent to the congruence @xmath1235 , and the latter congruence gives the value of @xmath514 ( modulo @xmath1236 ) so that is satisfied .
this means that when @xmath1237 , [ eq : congr - md1 ] holds if and only if @xmath1231 thus , if @xmath1237 ( that is , if @xmath1238 is not a factor of @xmath220 ) then the number of pairwise distinct torus knots in the link is @xmath1239 . in the remaining case when @xmath1136 ( i.e. , when @xmath1238 divides @xmath220 ) always holds : if @xmath1240 then we can take @xmath1241 to satisfy ; otherwise the left - hand side of just gives an expression for a unique residue @xmath514 modulo @xmath1236 ( which thus satisfies ) .
therefore the link consist of a unique cable ; so the number of pairwise distinct cables in the link is @xmath1242 in this case as well .
this concludes the proof .
[ note : mult - add - p-1 ] in conditions of theorem [ thm : mult - add - p ] note that @xmath1243 is the only case when the link @xmath563 consists of a single cable .
note also that from the proof of theorem [ thm : mult - add - p ] it is clear that if the number @xmath1244 of points in @xmath936 is 1 then the link necessarily consists of a single cable . by note [
note : const ] , @xmath1245 if and only if the period length of @xmath882 is 1 and therefore @xmath1246 for some @xmath1247 ) .
[ exmp : mult - add - p ] let @xmath309 and @xmath1248 .
then in conditions of theorem [ thm : mult - add - p ] we have that @xmath1249 and therefore the link consists of @xmath1250 cables with slopes @xmath1251 , cf .
figures [ fig : link - square ] and [ fig : link - torus ] . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ( 0,0)(15,15 ) ( -4.8,0)(13.2,18 ) ( -4.8,0.1).3 ( -4,14)(-2.4,15 ) ( -4,11)(2.6,15 ) ( -4,13)(-0.7,15 ) ( -4,5)(11,14.2 ) ( -4,4)(11,13.2 ) ( -4,7)(9,15 ) ( -4,8)(7.5,15 ) ( -4,10)(4.2,15 ) ( -4,2)(11,11.2 ) ( -4,1)(11,10.2 ) ( 2.5,0)(11,5.2 ) ( -2.4,0)(11,8.2 ) ( 9,0)(11,1.2 ) ( 4.1,0)(11,4.2 ) ( -0.75,0)(11,7.2 ) ( 7.4,0)(11,2.2 ) _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ , scaledwidth=92.0% ] [ cor : mult - add - compl ] there is a one - to - one correspondence between maps of the form @xmath1252 on @xmath177 ( where @xmath1253 ; @xmath1125 ; @xmath1126 ) and collections of @xmath1254 complex - valued exponential functions @xmath859 of real variable @xmath249 @xmath1255 here @xmath861 is imaginary unit and @xmath1256
. indeed , embedding the unit torus @xmath820 into a 3-dimensional euclidean space @xmath821 and using cylindrical coordinates as in note [ note : cable ] , in view of theorem [ thm : mult - add - p ] every knot from the link can be expressed in the form with @xmath1257 for @xmath1110 since @xmath1258 and @xmath1259 specifies position of the point where the knot crosses zero meridian of the torus ( i.e. , when @xmath1260 in ) . but
@xmath1124 and thus @xmath1261 by . as two such knots ( with accordingly @xmath1262 , @xmath943 )
coincide if an only if @xmath1263 by , i.e. , if and only if @xmath1264 .
but the latter congruence is equivalent to ; so finally the assertion follows from theorem [ thm : mult - add - p ] .
in this section we introduce central notion of the paper , the _ finite computability _ , and prove some technical results which will be needed further during the proof of main result of the paper , the affinity of finitely computable smooth functions , cf . further section [ sec : main ] .
[ def : eval - s ] a non - empty point set @xmath1265 ( @xmath1266 , @xmath1267 , @xmath1268 ) is called _ ( ultimately ) finitely computable _ ( or , ( ultimately ) computable by a finite automaton ) if there exists a finite automaton @xmath0 such that @xmath347 is a subset of @xmath1269 ( of @xmath1270 ) .
we say that the automaton @xmath0 _ ( ultimately ) computes _ the set @xmath347 ; and @xmath0 is called an _ ( ultimate ) computing automaton _ of the set @xmath347 . in most further cases given a real function @xmath1271 with the domain @xmath1272 by the _ graph of the function _
( on the torus @xmath23 ) we mean the point subset @xmath1273 .
however , given a function @xmath1274 where either @xmath44 $ ] or @xmath1275 and @xmath1150 is either @xmath14 $ ] or @xmath1276 , we call a graph @xmath1277 of the function @xmath85 the set @xmath1278 where either @xmath1279 if @xmath44 $ ] or @xmath1280 if @xmath1275 and accordingly either @xmath1281 if @xmath1282 $ ] or @xmath1283 if @xmath1284 . in the sequel
we always explain what is meant by @xmath1285 if this is not clear from the context .
also , we may omit the subscript @xmath89 when it is clear what is the domain .
[ def : eval ] given a real function @xmath1286 with domain @xmath1287 and an automaton @xmath11 , the function @xmath85 is called _ ( ultimately ) computable by @xmath0 at the point @xmath50 _ if @xmath1288 ( @xmath1289 ) . also , if either @xmath44 $ ] or @xmath1275 and @xmath1274 where either @xmath1282 $ ] or @xmath1290 we will say that @xmath0 _ ( ultimately ) computes @xmath85 at the point @xmath50 _ if @xmath1291 where either @xmath1279 if @xmath44 $ ] or @xmath1280 if @xmath1275 and accordingly either @xmath1281 if @xmath1282 $ ] or @xmath1283 if @xmath1284 ( cf . note [ note : plot - auto ] ) given a real function @xmath1286 with domain @xmath1287 , the function @xmath85 is called _ ( ultimately ) finitely computable _ ( or , _ ( ultimately ) computable by a finite automaton _ ) if there exists a finite automaton @xmath0 such that @xmath1292 ( @xmath1293 ) . the automaton @xmath0
( ultimately ) computes _ the function @xmath85 is called the _ ( ultimate ) computing automaton _ of the function @xmath85 . in a similar manner
we define these notions for the cases when @xmath1274 and @xmath89 , @xmath1150 are as above . in loose terms , when assigning a real - valued function @xmath1294{\rightarrow}[0,1]$ ] to automaton @xmath11 via monna map @xmath1295 ( cf .
subsection [ ssec : plots ] ) one feeds the automaton by a base-@xmath8-expansion of argument @xmath37 $ ] and considers the output as a base-@xmath8 expansion of @xmath1296 : a base-@xmath8 expansion specifies a unique right - infinite word in the alphabet @xmath18 and the automaton ` reads the word from head to tail ' , i.e. , is feeded by digits of the base-@xmath8 expansion from left to right
( i.e. , digits on more significant positions are feeded prior to digits on less significant positions ) ; and the output word specifies a base-@xmath8 expansion of a unique real number from @xmath14 $ ] . to examine functions computed by automata in the meaning of definition [ def : eval ] it would also be convenient to work with base-@xmath8 expansions of real numbers ; but the problem is that we need feed the automaton by a right - infinite word in the inverse order ` from tail ( which is at infinity ) to head ' : digits on less significant positions ( the rightmost ones ) should be feeded prior to digits on more significant positions ( the leftmost ones ) .
so straightforward inversion is impossible since it is unclear which letter should be the first when feeding the automaton this way ; thus output word is undefined and so is the real number whose base-@xmath8 expansion is the output word . in this subsection
we rigorously specify this inversion and develop some techniques needed in further proofs .
let a function @xmath1297 ( or @xmath88 $ ] ) whose domain @xmath89 is either a subset of a real unit circle @xmath528 or a subset of a unit segment @xmath14 $ ] be ultimately computable by a finite automaton @xmath336 ; that is , for any @xmath50 there exists @xmath1298 such that @xmath53 is a limit point of the sequence @xmath1299 and @xmath1300 is a limit point of the sequence @xmath1301 , where @xmath421 is automaton function of the automaton @xmath11 , cf .
definition [ def : eval ] and definition [ def : plot - auto ] . as said ,
further to examine finitely computable real functions it is however more convenient to work with automata maps as maps of reals into reals rather than to consider automata functions on @xmath8-adic integers and then represent @xmath1302 and @xmath1303 as limit points of the sequences @xmath1299 and @xmath1301 , respectively .
further in this subsection we are going to show that once @xmath50 and once @xmath1304 is a base-@xmath8 expansion of @xmath53 , we can find a state @xmath1305 of the automaton @xmath0 and a strictly increasing infinite sequence of indices @xmath746 such that the sequence @xmath1306 tends to @xmath1307 ( recall that @xmath1308 is an @xmath287-letter output word of the automaton @xmath359 whose initial state is @xmath242 once the automaton has been feeded by the @xmath287-letter input word @xmath1309 , cf .
subsection [ ssec : auto ] ) .
this means , loosely speaking , that once we feed the automaton @xmath359 with approximations @xmath1310 of @xmath53 , the automaton outputs the sequence of approximations @xmath1311 of @xmath1300 , and these sequences tend to @xmath53 and to @xmath1300 accordingly while @xmath1312 .
moreover , we will show that if the function @xmath85 is continuous then there exists a state @xmath352 such that all @xmath50 for which @xmath1313 constitute a dense subset in @xmath89 .
recall that given @xmath1314 , there exists a ( right-)infinite word @xmath1315 over @xmath1316 such that @xmath1317 the _ base-@xmath8 expansion _ of @xmath53 . if @xmath53 is not of the form @xmath1318 for some @xmath1319 , where @xmath1320 is the length of the base-@xmath8 expansion of @xmath483 ( recall that we put @xmath1321 , cf
. subsection [ ssec : a - map ] ) , @xmath1322 , then the right - infinite word @xmath1323 over @xmath1316 is uniquely defined ( and the corresponding @xmath53 is said to have a _ unique _
base-@xmath8 expansion ) ; else there are exactly two infinite words , @xmath1324 where @xmath1325 , such that @xmath1326 . in that case
@xmath53 is said to have a _
non - unique _ base-@xmath8 expansion ; the corresponding base-@xmath8 expansions are called _ right _ and _ left _ respectively . both 0 and 1
are assumed to have unique base-@xmath8 expansions since @xmath1327 , @xmath1328 ; so @xmath1329 , @xmath1330 . this way we define @xmath1331 for all @xmath37 $ ] ; and to @xmath1318 we will usually put into the correspondence both infinite words @xmath1332 and @xmath1333 if converse is not stated explicitly . the only difference in considering a unit circle @xmath528 rather than the unit segment
@xmath43 $ ] is that we identify 0 and 1 and thus have two representations for 0 , @xmath303 and @xmath1334 . given a finite word @xmath10 , we denote via @xmath1335 the ( right-)infinite word @xmath1336 and we put @xmath1337 ( note that then @xmath1338 ) . of course , @xmath1339
; but we use notation @xmath1340 if we want to stress that we deal with infinite base-@xmath8 expansion . to unify our notation
, we also may write @xmath1341 for a ( right-)infinite word @xmath1342 ; then @xmath1343 .
let @xmath1344 be a ( right-)infinite word over @xmath1 .
given an automaton @xmath0 with the initial state @xmath242 , we further denote via @xmath1345 the set of all limit points of the sequence @xmath1346 .
we may omit the subscript @xmath242 if it is clear from the context what is the initial state of the automaton .
given @xmath37 $ ] further @xmath1347 stands for @xmath1348 if @xmath53 admits a unique base-@xmath8 expansion , and @xmath1349 if the expansion is non - unique ( thus , if @xmath53 admits both left and right base-@xmath8 expansions ) .
we also consider @xmath1347 for @xmath1350 rather that for @xmath37 $ ] ; in that case we take for @xmath307 both base-@xmath8 expansions @xmath303 and @xmath1351 ( since @xmath1352 ) and reduce modulo 1 all limit points of all sequences @xmath1346 .
we further use the same symbol @xmath1347 independently of whether we consider @xmath37 $ ] or @xmath1350 ; we make special remarks when this may cause a confusion . we stress that @xmath1353 is a uniquely defined finite word whenever @xmath427 is a finite word ( and therefore @xmath1354 consists of a single number ) , but in the case when @xmath16 is an infinite word or @xmath16 is a real number from @xmath14 $ ] ( or @xmath1355 ) , the set @xmath1353 may contain more than one element .
given @xmath251 $ ] , in view of lemma [ le : per ] it is clear that if the automaton @xmath0 is finite then @xmath1356 $ ] since a real number is rational if and only if its base-@xmath8 expansion is eventually periodic .
the following propositions reveals some more details about @xmath1357 for a rational @xmath53 ; and especially for @xmath629 .
[ prop : auto - q ] if @xmath0 is finite , @xmath251 $ ] then @xmath1358 $ ] and @xmath1357 is a finite set .
moreover , if @xmath1359 $ ] then @xmath1360 $ ] . in particular , if @xmath1361 then @xmath1362 for suitable @xmath937 .
let a @xmath0-computable function @xmath1297 be defined on the domain @xmath1275 and continuous at @xmath1363 .
if the domain @xmath89 is open then there exists @xmath1364 such taht @xmath1365 ; and either @xmath1366 or @xmath1367 if the domain @xmath89 is half - open and @xmath53 is a boundary of @xmath89
. follows from lemma [ le : per ] and proposition [ prop : const ] ( see the proof of the latter )
. given @xmath251 $ ] , a base-@xmath8 expansion of @xmath53 is eventually periodic , cf .
: @xmath1368 . given @xmath53 , take @xmath1369 the smallest possible ( note that then the word @xmath1370 may be empty ) . from the definition of @xmath1357 it follows that @xmath1357 consists of all limit points of the sequences @xmath1371 , where @xmath1372 is large enough and @xmath1373 are all suffixes of the word @xmath1374 , for every right - infinite word @xmath16 which corresponds to a base-@xmath8 expansion of @xmath1375 . as the automaton @xmath336 is finite , the number of states it reaches after being feeded by either of words @xmath228 , where is finite ; say , these states are @xmath1376 . by the same reason , being feeded by the words @xmath1377 where @xmath1372 is large enough , the automata @xmath1378 output respectively words @xmath1379 , where the words @xmath1380 do not depend on @xmath1372 , each of the words @xmath1381 is either empty or a prefix of the respective word @xmath1382 , and all the output words @xmath1379 have the same length as the one of the input word @xmath1377 , cf .
lemma [ le : per ] .
that is , the output words of automata @xmath1383 are all of the form @xmath1384 , where @xmath440 stands for a cyclically shifted word @xmath436 , and after outputting the word @xmath1384 , each automaton @xmath1383 , @xmath1385 , reaches some of finitely many its states , say , @xmath1386 . after reaching respective state , the automaton is feeded by the word @xmath871 and outputs the corresponding output word @xmath1387 , @xmath1388
therefore , all limit points of the sequences @xmath1389 are of the form @xmath1390 where @xmath1391 runs through a ( sub)set of all cyclic shifts of the word @xmath1392 , @xmath1393 , @xmath1388 .
but there are only finite number of points of that form ; therefore , given a base-@xmath8 expansion of @xmath1394 , the set @xmath1395 is a union of a finite number points of the said form .
as every @xmath1396 has at most two base-@xmath8 expansions , this proves the first claim of the proposition .
if @xmath1359 $ ] , then base-@xmath8 expansion of @xmath53 is purely periodic by proposition [ prop : r - p - repr - qz ] : @xmath1397
. therefore once the automaton @xmath0 is being feeded by finite words of the form @xmath1398 , where @xmath1399 , @xmath229 is a suffix of @xmath16 or empty and @xmath486 is large enough , by lemma [ le : per ] the corresponding output word will be of the form @xmath1400 , and the number of different words @xmath1401 is finite since the number of different words @xmath229 is finite . applying the same argument as above
we conclude that all limit points of corresponding sequences are of the form @xmath1402 where @xmath436 runs through a finite number of finite words .
but by corollary [ cor : r - repr ] , all these points are in @xmath1403 $ ] .
this proves the second claim of the proposition .
to prove the final claim , we must consider both base-@xmath8 representations of zero point of the unit circle @xmath528 : @xmath1404 and @xmath1405 .
sending a left - infinite zero sequence to the automaton @xmath92 , the output sequence will be of the form @xmath1406 by suitable finite words @xmath1407 by lemma [ le : per ] ; so @xmath1408 consists of all points of the form @xmath940 , where @xmath436 runs through all cyclic shifts of the word @xmath1409 ; therefore @xmath1410 for a suitable @xmath1411 , cf .
note [ note : const ] . by the same reason ,
@xmath1412 for a suitable @xmath1413 . in the case
when the @xmath11-computable function @xmath1297 is continuous at @xmath307 and there exists an open neighborhood @xmath584 of @xmath307 such that @xmath1414 then necessarily @xmath1415 and @xmath1416 ; so @xmath946 and therefore @xmath938 by corollary [ cor : const ] .
if no such neighborhood @xmath584 exists then the domain is half - open and @xmath307 is a boundary point ; thus due to the continuity of @xmath85 at @xmath307 we see that either @xmath1415 or @xmath1416 and the conclusion follows .
[ cor : auto - q ] let @xmath0 be a finite automaton , let @xmath1417 , and let @xmath1418 ; then @xmath763 . if @xmath629 then @xmath1419 ; moreover , there exists @xmath763 such that @xmath1420 . as @xmath762 and @xmath985 then by corollary [ cor : r - p - repr - qz ]
the base-@xmath8 expansion is purely periodic ; that is , @xmath1375 , where @xmath1421 is a right - infinite periodic word : @xmath1422 for a suitable finite non - empty word @xmath1423 . as @xmath1424 then by note [ note : plot - auto - in ] there exists a sequence @xmath1425 of finite non - empty words such that @xmath1426 and simultaneously @xmath1427 .
let @xmath1428 ; then @xmath871 is not a 1-letter zero word : @xmath1429 . therefore since @xmath1426 , for all sufficiently large @xmath67 the words @xmath1392 must be of the form @xmath1430 where @xmath1431 increases unboundedly while @xmath738 .
therefore we may assume that the sequence @xmath1431 is strictly increasing ( we consider a strictly increasing subsequence of the sequence @xmath1432 if otherwise ) and that all @xmath1392 are of the form @xmath1433 .
let @xmath1434 be a state the automaton @xmath0 reaches after being feeded by the word @xmath1391 ( @xmath1435 if @xmath1436 is empty ) . as the automaton @xmath0 is finite , then there are only finitely many pairwise distinct @xmath1434 , say these are @xmath1437 .
now we consider automata @xmath1438 and apply the same argument as in the proof of the second statement of proposition [ prop : auto - q ] thus proving that @xmath763 .
the same argument can be applied for the case when @xmath629 but there exists an infinite sequence of words @xmath1392 whose lengths are increasing unboundedly while @xmath738 .
therefore the only rest case is now @xmath629 and once the word sequence @xmath1425 is such that @xmath1426 and lengths of @xmath1392 are bounded ; @xmath1439 for all @xmath155 .
but this just means that for all sufficiently large @xmath67 all the words @xmath1392 are @xmath1372-letter zero words : @xmath1440 for a suitable @xmath1441 .
but then @xmath1442 ; thus @xmath1419 . the last claim of the corollary trivially follows from proposition [ prop : auto - q ] since once @xmath1443 then @xmath1444 by the definition of @xmath15 .
the following definition introduces an important technical notion , the mark - up , which will be used in further proofs : [ def : mark - up ] given a function @xmath88 $ ] defined on the domain @xmath44 $ ] , an automaton @xmath0 , and a point @xmath50 consider a right - infinite word @xmath16 such that @xmath1375 , cf .
an infinite strictly increasing sequence @xmath1445 over @xmath399 is called an _ @xmath242-mark - up of the word @xmath1315 w.r.t . @xmath85 and @xmath0 _ ( or briefly a _ mark - up
_ when it is clear what @xmath85 , @xmath0 and @xmath242 are meant ) if there exists a state @xmath352 of the automaton @xmath0 such that @xmath1446 . given @xmath1447 , the mark - ups of @xmath53 are defined exactly in the same way .
note only that the point @xmath629 of @xmath528 coincides with the point @xmath1448 and @xmath1449 as @xmath1450 . in a similar way we define the mark - up
when @xmath1451 .
the following proposition shows , speaking loosely , that if a continuous real function is finitely computable , then all base-@xmath8 expansions of all its arguments can be marked - up : [ prop : mark - up ] given a continuous function @xmath1452 $ ] let @xmath86 for a suitable finite automaton @xmath11 .
then for every @xmath1453 and every infinite word @xmath16 such that @xmath1375 there exists an @xmath242-mark - up , for a suitable state @xmath1454 of the automaton @xmath0 .
the idea of the proof is as follows : once feeding a finite automaton @xmath11 by infinite sequence of finite words @xmath1455 over @xmath177 , the automaton reaches some of its states infinitely many times ; this state @xmath242 specifies a mark - up @xmath1456 since corresponding sequences of approximations @xmath1457 and @xmath1458 tend accordingly to @xmath1459 and to @xmath1300 due to the continuity of @xmath85 at @xmath53 .
now we prove the proposition rigorously .
firstly consider the case when @xmath1453 has a unique base-@xmath8 expansion , say , @xmath1460 , where @xmath1315 .
as @xmath1300 is @xmath0-computable , there exists a sequence @xmath1461 of finite non - empty words such that @xmath1462 , @xmath597 and @xmath1463 .
note that the sequence @xmath1464 is increasing since otherwise @xmath1465 for suitable @xmath1466 as @xmath597 and thus @xmath53 has a non - unique base-@xmath8 expansion .
we may assume that @xmath1467 is a strictly increasing sequence since otherwise we just take a suitable subsequence of @xmath594 .
moreover , by the same reason we may assume that @xmath1468 .
consider a word sequence @xmath1469 .
as @xmath1453 , there exists @xmath1470 such that @xmath1471 once @xmath1472 . without loss of generality
we may assume that @xmath1473 .
as @xmath1474 , for every @xmath483 there exists @xmath1475 , @xmath1476 , such that @xmath1477 provided @xmath1478 ; therefore as @xmath1479 , we conclude that @xmath1480 for @xmath1481 and suitable finite word @xmath1482 . given @xmath483 ,
let @xmath1483 be the smallest with the said property ; this way we obtain an increasing sequence @xmath1484 .
considering the subsequence @xmath1485 of the word sequence @xmath594 , we see that @xmath1486 for @xmath1487 and suitable finite non - empty words @xmath1488 , that @xmath1489 and that @xmath1490 .
now to the word sequence @xmath1491 we put into the correspondence the sequence @xmath1492 of states of the automaton @xmath0 , where @xmath1493 is the state the automaton @xmath0 reaches after being feeded by the input word @xmath1488 .
as the number of different states of @xmath0 is finite , in the sequence @xmath1494 at least one state , say @xmath242 , occurs in the sequence @xmath1494 infinitely many times ; say , for @xmath1495 ( @xmath1496 ) .
therefore @xmath1497 recall that @xmath337 where @xmath47 is the initial state of the automaton @xmath336 .
note that nowhere in the argument above we have used that @xmath85 is continuous . now consider the case when @xmath1465 for suitable @xmath1466 , @xmath1498 . in this case , let @xmath1499 be either of base-@xmath8 expansions @xmath1333 , @xmath1332 .
we will show that similarly to the case when a base-@xmath8 expansion is unique , in the case under consideration there also exist @xmath352 and sequences @xmath1500 , @xmath1501 over @xmath399 such that @xmath1502 for all @xmath1503 and both and hold .
for this purpose , consider arbitrary sequence @xmath1504 of right - infinite words over @xmath1316 which are not eventually periodic and such that @xmath1505 is a prefix of @xmath1506 for all @xmath412 .
then @xmath1507 and therefore all @xmath1508 once @xmath626 where @xmath1509 is large enough ( we may assume that @xmath1510 ; otherwise we consider a subsequence @xmath1511 rather than the whole sequence @xmath594 ) .
note that then @xmath1512 as @xmath85 is continuous on @xmath1513 ; therefore there exists a sequence @xmath1514 over @xmath566 such that for all @xmath412 the following inequality holds : latexmath:[\[\label{eq : lim-4 } @xmath1514 is strictly increasing ( if not , we consider a corresponding infinite subsequence of the sequence @xmath594 rather than the whole sequence ) .
consider now a word @xmath1516 from the above word sequence ( note that @xmath1517 for @xmath1518 ) . as
every @xmath1519 is a unique base-@xmath8 expansion of the corresponding real number from @xmath1513 , there exists a state @xmath1434 of the automaton @xmath0 and a strictly increasing sequence @xmath1520 of numbers from @xmath399 such that @xmath1521 cf .
the case we just have considered above at the beginning of the proof of the proposition . therefore , for any @xmath1503 there exists @xmath1522 such that @xmath1523 and there exists a strictly increasing sequence of @xmath138 such that the corresponding sequence of @xmath1524 is also strictly increasing in force of . without loss of generality
we may assume that the sequence @xmath1525 is strictly increasing ( otherwise we consider a corresponding subsequence of the sequence @xmath1526 ) .
as a total number of states of @xmath11 is finite , in the infinite sequence @xmath1527 at least one state , say @xmath242 , occurs infinitely many times .
we may assume that @xmath1528 for all @xmath155 ; otherwise we just consider respective subsequence of @xmath594 rather than the whole sequence .
as the sequence @xmath1529 is strictly increasing , all @xmath1530 once @xmath138 is large enough .
given @xmath155 , denote via @xmath1531 the smallest number such that @xmath1530 once @xmath1532 .
we again may assume that @xmath1533 ; if otherwise we will just consider the subsequence @xmath1534 rather than the whole sequence @xmath1529 .
then @xmath1535 for all @xmath138 where @xmath1536 is a non - empty finite word .
let @xmath1537 be a state the automaton @xmath359 reaches after being feeded by the input word @xmath1536 ( the latter state @xmath242 is defined above ) .
as the total number of states of @xmath359 is finite , in the sequence @xmath1538 at least one state , say @xmath1539 , occurs infinitely many times .
moreover , by the same reason at least one state , say @xmath1540 , occurs in the sequence @xmath1541 infinitely many times . and
again , without loss of generality we may assume that @xmath1542 for all @xmath1543 ; otherwise we consider corresponding subsequences of the sequences @xmath594 and @xmath1529 .
note that being feeded by the input word @xmath1535 , the automaton @xmath359 ( rightmost letters are feeded prior to leftmost ones ) , the automaton outputs a word of length @xmath1544 whose ( left ) suffix of length @xmath1545 is the word which outputs the automaton @xmath1546 if being feeded by the word @xmath1505 . therefore @xmath1547 ; but @xmath1548 since @xmath1499 is a base-@xmath8 expansion of @xmath53 .
this finally proves the proposition .
[ note : mark - up ] actually during the proof of proposition [ prop : mark - up ] we have shown that the following claim is true : _ let the function @xmath1549 be defined on an open neighbourhood @xmath1550 of a point @xmath53 , let @xmath85 be continuous at @xmath53 , and let @xmath86 for a suitable finite automaton @xmath0 ; then there exists a mark - up for every base-@xmath8 expansion of @xmath53 .
moreover , if @xmath1551{\rightarrow}[0,1]$ ] is a continuous function on the closed segment @xmath1552 $ ] then there exist an @xmath242-mark - up for right base-@xmath8 expansion of @xmath83 and for left base-@xmath8 expansion of @xmath220 . _
[ cor : mark - up ] in conditions of proposition [ prop : mark - up ] , if the automaton @xmath11 is minimal then @xmath1553 . follows immediately from corollary [ cor : ap = lp ] by the definition of mark - up .
the following proposition reduces examination of continuous functions computable by a finite automaton @xmath0 for the case when the function is defined on a segment for which there exists a state @xmath242 of the automaton such that the set of all points from the segment that has @xmath242-mark - ups , is dense in the segment ; and so values at these points completely specify the function on the segment .
[ prop : cover ] let @xmath1551{\rightarrow}[0,1]$ ] , @xmath1552\subset[0,1]$ ] , be a continuous function ; let @xmath86 for a suitable automaton @xmath0 whose set of states @xmath322 is finite . then @xmath1552 $ ] is a union of a countably many sub - segments @xmath1554\subset[a , b]$ ] , @xmath1555 , @xmath1556 having the following property : for every @xmath1556 there exists a state @xmath1557 , @xmath1558 , such that the set @xmath1559)$ ] of all points from @xmath1554 $ ] that have @xmath1560-mark - ups is dense in @xmath1554 $ ]
. for @xmath352 denote via @xmath1561 the set of all right - infinite words @xmath1421 such that @xmath1562 and an @xmath242-mark - up for @xmath16 exists ; put @xmath1563 . given @xmath352 such that @xmath1564 , let @xmath1565 be intersection of the closure @xmath1566 of @xmath1567 with @xmath1513 ; so @xmath1568 is closed in @xmath1513 w.r.t .
the induced topology on @xmath1513 and there are only finitely many pairwise distinct @xmath1568 ; say , these are @xmath1569 .
proposition [ prop : mark - up ] implies that @xmath1570 .
we argue that for some @xmath1569 their interiors @xmath1571 are not empty .
indeed , @xmath1572 for all @xmath1573 ; but from proposition [ prop : mark - up ] it follows that @xmath1574 $ ] and therefore @xmath1575 is dense in @xmath1552 $ ] as @xmath1552 $ ] is baire , cf .
e.g. ( * ? ? ?
* theorems 6.166.17 ) .
as some ( without loss of generality we may assume that all ) of the interiors @xmath1576 are non - empty , the interiors are countable unions of open intervals : @xmath1577 , @xmath1578 , @xmath1579
. therefore @xmath1580 $ ] .
this completes the proof as @xmath1574 $ ] .
[ cor : cover ] in conditions of proposition [ prop : cover ] , the segment @xmath1552 $ ] admits a countable covering by closed sub - segments @xmath1554\subset[a , b]$ ] such that the graph @xmath1581 of the restriction of the function @xmath85 to the sub - segment @xmath1554 $ ] lies in @xmath1582 for a suitable sub - automaton @xmath359 of the automaton @xmath0 , @xmath1583 . indeed , the closure of the point set @xmath1584)\}$ ] in @xmath3 is a graph of the restriction @xmath1585 of the function @xmath85 to the segment @xmath1554 $ ] as @xmath1585 is continuous on @xmath1554 $ ] . on the other hand
the closure must lie in @xmath1586 for @xmath1587 as @xmath1588 where @xmath1589 and @xmath1590 is an @xmath242-mark - up .
note also that @xmath1591 as every state of the automaton @xmath0 is reachable from its initial state @xmath47 ( since we consider reachable automata only , cf . subsection [ ssec : auto ] ) .
[ note : cover ] from the respective proofs it follows that both proposition [ prop : cover ] and corollary [ cor : cover ] remain true for a continuous function @xmath1551{\rightarrow}\mathbb s$ ] as well as for the case when @xmath1552\subset\mathbb s$ ] .
the following theorem shows that we may restrict our considerations of finitely computable continuous functions to the case when computing automata are minimal .
[ thm : eval - erg ] given a continuous function @xmath1551\to[0,1]$ ] , @xmath1552\subset[0,1]$ ] such that @xmath86 for a finite automaton @xmath0 , there exists a countable covering @xmath1592\subset[a , b]:j=1,2,\ldots ; a_j^\prime < b_j^\prime\}$ ] of the segment @xmath1552 $ ] such that for every @xmath514 the graph @xmath1581 of the restriction @xmath1585 of the function @xmath85 to the segment @xmath1554 $ ] lies in @xmath1593 for a suitable minimal sub - automaton @xmath1594 of @xmath0 , @xmath1595 .
the state @xmath1560 from proposition [ prop : cover ] is either ergodic or transient , see subsection [ ssec : auto ] .
we consider these two cases separately . * case 1 : * the state @xmath1560 is ergodic . as the set @xmath1559)$ ] from proposition [ prop : cover ]
is dense in @xmath1554 $ ] and @xmath1585 is continuous , every point @xmath1596 for @xmath1597 $ ] is a limit of a sequence @xmath1598 where @xmath1599 is a sequence of points from @xmath1559)$ ] and @xmath1599 tends to @xmath53 as @xmath67 tends to infinity : @xmath1600 but @xmath1601 where @xmath1506 is a right - infinite word for which there exists an @xmath1560-mark - up ( cf . the construction of the set @xmath1559)$ ] ) ; therefore from it follows now that there exists a sequence @xmath1602 of finite words @xmath1603 of strictly increasing lengths such that @xmath1604 indeed , the words @xmath1603 are ( left ) prefixes of words @xmath1605 that correspond to @xmath1560-mark - up ; that is , @xmath1603 are of the form @xmath1606 where the sequence @xmath1607 is the @xmath1560-mark - up of the word @xmath1608 .
now , as the state @xmath1560 is ergodic ( that is , @xmath1560 a state of a certain minimal sub - automaton , say @xmath1609 , of the automaton @xmath0 , cf .
subsection [ ssec : auto ] ) , then we just mimic the proof of theorem [ thm : ap = lp ] starting with and show that @xmath1610 .
* case 2 : * now let the state @xmath1560 from proposition [ prop : cover ] be not ergodic ( whence transient ) . thus there exists a finite word @xmath1611 such that after the automaton @xmath1612 has been feeded by the word @xmath436 ( rightmost letters are feeded to the automaton prior to leftmost ones ) , the automaton reaches some ergodic state ( say @xmath229 ) which is a state of a minimal sub - automaton @xmath1613 , see subsection [ ssec : auto ] .
note that then all words of the form @xmath1614 have the same property , for all @xmath430 : after being feeded by @xmath1614 , the automaton reaches some state from the set of states of @xmath1615 due to the minimality of @xmath1615 .
therefore , the set @xmath1616 of all @xmath8-adic integers whose base-@xmath8 expansions ( cf .
subsection [ ssec : p - adic ] ) are left - infinite words @xmath1421 such that if the automaton @xmath1617 while being feeded by the word @xmath16 reaches at a finite step some ergodic state ( that is , reaches a state which is a state of some minimal automaton of @xmath0 ) is a union of balls of non - zero radii in @xmath177 ; thus , the set @xmath1618 is an open subset in @xmath177 since every ball of a non - zero radius is open in @xmath177 w.r.t .
the @xmath8-adic topology , cf .
subsection [ ssec : p - adic ] .
hence the set @xmath1619 is a closed subset of @xmath177 ; and the set @xmath1620 consists of all @xmath8-adic integers such that if the automaton @xmath1621 is being feeded by a left - infinite word that is a base-@xmath8 expansion of some @xmath8-adic integer from @xmath1620 , the automaton @xmath0 never reaches an ergodic state .
let @xmath1622 be the set of all finite prefixes of words from @xmath1620 ; denote @xmath1623 a closure of the set @xmath1624 in @xmath266 .
_ the interior of @xmath1623 is empty _ ( therefore @xmath1625 is nowhere dense in @xmath14 $ ] ) . indeed , if not then @xmath1623 contains an open interval @xmath1626 .
take a finite non - empty word @xmath436 such that @xmath1627 .
as @xmath1628 then , given an arbitrary finite non - empty word @xmath1629 where @xmath1630 , there exists a sequence @xmath1631 of finite non - empty words @xmath1632 such that @xmath1633 ( recall that @xmath1634 is a concatenation of words @xmath436 and @xmath871 )
. therefore either @xmath1635 ( thus @xmath1636 by the construction of @xmath1622 ) or @xmath1637 contains an infinite subsequence of words of the form @xmath1638 where @xmath1639 , @xmath1640 ( recall that @xmath1641 is a word of length @xmath1642 all whose letters are @xmath216 ) ; hence by the construction of @xmath1622 there exists an infinite sequence @xmath1643 over @xmath1622 .
thus we conclude that once @xmath203 , the closure of @xmath1620 in @xmath177 ( thus , the very set @xmath1620 itself as it is closed ) must either contain @xmath94 or @xmath1644 ( recall that negative rational integers in @xmath177 are exactly that ones whose canonical @xmath8-adic expansions have only a finitely many terms with coefficients other than @xmath216 , cf .
subsection [ ssec : p - adic ] ) . but
this implies that @xmath1645 as the set @xmath1646 ( where @xmath1647 or @xmath1648 are taken in arbitrary order ) is dense in @xmath177 . on the other hand , by the construction
the set @xmath1620 consists of all @xmath8-adic integers such that if the automaton @xmath1621 is being feeded by a left - infinite word that is a base-@xmath8 expansion of some @xmath8-adic integer from @xmath1620 , the automaton @xmath0 never reaches an ergodic state ; therefore the equality @xmath1645 contradicts our assumption that @xmath1560 is transient ( since then there must exist a left - infinite word @xmath16 such that at a finite step the automaton @xmath1612 reaches an ergodic state if being feeded by @xmath16 ) .
this proves our claim .
denote now via @xmath1649 an automaton function of the automaton @xmath1612 ; and for @xmath554 put @xmath1650 a point set in the unit real square @xmath30\times[0,1]$ ] ; then take a union @xmath1651 ; denote via @xmath1652 a closure ( in topology of @xmath3 ) of the set @xmath1653 ( cf . ) .
denote via @xmath1585 a restriction of the function @xmath85 to @xmath1554 $ ] . as the function @xmath1585 is continuous on @xmath1554 $ ] ( cf .
corollary [ cor : cover ] ) and @xmath1654 , then necessarily @xmath1655 since the set @xmath1623 is nowhere dense in @xmath1554 $ ] by claim 1 . as the set @xmath322 of all states of the automaton
@xmath0 is finite , there are only finitely many ergodic components in @xmath322 ; say they are @xmath1656 .
given an ergodic component @xmath1657 @xmath1658 denote @xmath1659 where @xmath1660 is the smallest @xmath402 such that after the automaton @xmath1612 has been feeded by the word @xmath1661 , the automaton reaches a state from @xmath1657 . then the union @xmath1662 is disjoint and @xmath1663 is a closure of @xmath1664 by .
therefore @xmath1665 where @xmath1666 is a closure of @xmath1667 in @xmath64 ; hence @xmath1668 .
note that from the definition of @xmath15 ( cf .
subsection [ ssec : plots ] ) it follows that @xmath1669 where @xmath1594 is a minimal sub - automaton ( of the automaton @xmath0 ) whose set of states is @xmath1657 .
further , as the function @xmath1585 is continuous on @xmath1554 $ ] , the set @xmath1581 is closed in @xmath64 ; therefore the set @xmath1670 is closed in @xmath3 .
hence , the set @xmath1671{\colon}(x , g(x))\in\mathbf e_n\}$ ] is closed in @xmath266 and @xmath1554=\cup_{n=1}^m\mathbf r_n$ ] .
now by argument similar to that from the proof of proposition [ prop : cover ] we conclude that some of the interiors @xmath1672 must be non - empty and hence either of the non - empty interiors is a union of a countably many open intervals . by taking closures of the intervals
we see that @xmath1554 $ ] is a union of the closures , that is , @xmath1554 $ ] is a union of a countably many its closed sub - segments @xmath1673 $ ] @xmath1674 of non - zero lengths , and the graph of the restriction @xmath1675 of @xmath1585 to either of the sub - segments lies in @xmath1669 for a suitable @xmath1676 .
now we apply proposition [ prop : cover ] substituting @xmath1675 for @xmath85 and @xmath1673 $ ] for @xmath1552 $ ] ; but as every @xmath1560 from the statement of proposition [ prop : cover ] is now a state of the minimal sub - automaton @xmath1594 , we now are in conditions of case 1 . therefore @xmath1677 ; but @xmath1678 for all states @xmath749 of the automaton @xmath1679 due to the minimality of the automaton , cf . note [ note : ap = lp ] .
this finally proves the theorem .
[ note : eval - erg ] from the proof of theorem [ thm : eval - erg ] it follows that the theorem remains true for a continuous function @xmath1551{\rightarrow}\mathbb s$ ] as well as for the case when @xmath1552\subset\mathbb s$ ] .
the following proposition shows that we may if necessary consider only finitely computable continuous functions defined everywhere on the unit segment @xmath14 $ ] rather than on sub - segments of @xmath14 $ ] .
[ prop : ext ] if a continuous function @xmath1551{\rightarrow}\mathbb s$ ] , @xmath1552\subset[0,1]$ ] , is such that @xmath1680}(g)\subset \mathbf p(\mathfrak a)$ ] for a suitable finite automaton @xmath336 then for every @xmath1681 such that @xmath1682 and @xmath1683 $ ] the function @xmath1684 , where @xmath1685 , is continuous on @xmath14 $ ] , and @xmath1686}(g_d)\subset\mathbf p(\mathfrak a)$ ] . as
a base-@xmath8 expansion of @xmath248 is @xmath1687 then , given a base-@xmath8 expansion for @xmath1688 $ ] , a base-@xmath8 expansion for @xmath1689 is @xmath1690 and @xmath1691 $ ] for all right - infinite words @xmath773 ( thus , for all @xmath37 $ ] ) . therefore if @xmath1500 is a mark - up for @xmath1692 ( cf .
proposition [ prop : mark - up ] ) then @xmath1693 , where @xmath1694 , is an @xmath242-mark - up of the infinite word @xmath773 for a suitable state @xmath352 of the automaton @xmath336 w.r.t .
the function @xmath85 .
hence , @xmath1695 as @xmath1696 . by our assumption on reachability of the automaton @xmath0 ( cf .
subsection [ ssec : auto ] ) , there exists a finite word @xmath1697 such that the automaton @xmath0 being feeded by @xmath436 reaches the state @xmath242 and outputs the corresponding finite word @xmath1698 ; therefore the automaton @xmath336 being feeded by a concatenated finite word @xmath1699 outputs the concatenated finite word @xmath1700 . but
@xmath1701 and simultaneously @xmath1702 since the words @xmath1703 are finite and fixed ; therefore @xmath1704 for all @xmath37 $ ] in view of .
the function @xmath1705 is conjugated to a continuous function by a continuous map and therefore is also continuous : once @xmath1706 $ ] are such that @xmath1707 to ensure that @xmath1708 for a sufficiently large @xmath1003 then @xmath1709 .
[ cor : ext ] if a continuous function @xmath1551{\rightarrow}\mathbb s$ ] , @xmath1552\subset[0,1]$ ] , is such that @xmath1710 for a suitable finite automaton @xmath336 then for every @xmath1681 such that @xmath1682 and @xmath1711 * the function @xmath1712 is continuous on @xmath14 $ ] for all sufficiently large @xmath1713 , and * @xmath1686}(g_{d , m})\subset\mathbf p(\mathfrak a)$ ] .
indeed , in the proof of proposition [ prop : ext ] as a base-@xmath8 expansion for @xmath1685 just use @xmath1714 where @xmath1713 is large enough so that @xmath1715 $ ] . note that nowhere in the proof of the proposition we used that some of @xmath1716 are not zero .
[ note : ext ] corollary [ cor : ext ] shows that given any point @xmath1717 and a rational approximation @xmath1685 of @xmath1718 , the graph of the function @xmath85 on a sufficiently small closed neighbourhood @xmath1719 $ ] of the point @xmath1720 is similar to the graph of the function @xmath1721 on @xmath14 $ ] where @xmath1685 and @xmath635 is large enough . summarizing results of the current subsection
we may say that while considering a continuous function @xmath1551{\rightarrow}\mathbb s$ ] ( where @xmath1552\subset[0,1]$ ] or @xmath1552\subset\mathbb s$ ] ) whose graph @xmath1722 lies in @xmath15 for some finite automaton @xmath0 one can if necessary assume that the function is defined and continuous on @xmath14 $ ] ( or on @xmath528 except for maybe a single point ) , the automaton @xmath0 is minimal , the function @xmath85 is ultimately computable by @xmath0 and that for some state @xmath242 of @xmath11 the set of all points from @xmath14 $ ] which have base-@xmath8 expansions admitting @xmath242-mark - ups is dense in @xmath14 $ ] ( respectively , in @xmath528 ) .
it is clear that a composition of finitely computable continuous functions should be a finitely computable continuous function .
the following proposition states this formally and gives some extra information about the graph of a composite finitely computable function .
[ prop : comp ] let @xmath1552,[c , d]\subset[0,1]$ ] and let @xmath1551{\rightarrow}[0,1]$ ] , @xmath1723{\rightarrow}[0,1]$ ] be two continuous functions such that @xmath1724)\subset[c , d]$ ] and there exist finite automata @xmath0 and @xmath366 such that @xmath1680}(g)\subset\mathbf p(\mathfrak a)$ ] , @xmath1725}(f)\subset\mathbf p(\mathfrak b)$ ] .
then there exists a covering @xmath1592\subset[a , b]{\colon}j\in j\}$ ] such that if @xmath1726 is a restriction of the composite function @xmath1727 to the sub - interval @xmath1554 $ ] then @xmath1728}(h_j)\subset\mathbf p(\mathfrak c_j)$ ] for every @xmath1729 , where @xmath1730 is a sequential composition of the automaton @xmath1731 with the automaton @xmath1732 and @xmath1733 are suitable states of the automata @xmath0 , @xmath366 accordingly . by note
[ note : mark - up ] , for every right - infinite word @xmath1734 such that @xmath1735 there exists a mark - up ( w.r.t . some state @xmath242 of the finite automaton @xmath11 ) @xmath1445 ; i.e. , @xmath1736 , where @xmath1737 ; and if @xmath1738 ( respectively , @xmath830 ) then the mark - up exists at least for right ( respectively , left ) base-@xmath8 expansion . by the same reason , for @xmath1739 , where @xmath1740 , there exists a mark - up @xmath1741 ( w.r.t . some state @xmath229 of the finite automaton @xmath366 ) such that @xmath1742 , where @xmath1743 .
now for @xmath454 denote @xmath1744 , consider the sequence @xmath1745 and let @xmath1746 be a strictly increasing subsequence of @xmath1745 .
denote @xmath1747 the state the automaton @xmath11 reaches after being feeded by the word @xmath1748 ; put @xmath1749 if the latter word is empty . as the automaton @xmath11 is finite
, there is a state , say @xmath1540 , that occurs in the sequence @xmath1750 infinitely often .
then the sequence @xmath1751 is a mark - up of the word @xmath16 w.r.t .
the automaton @xmath1752 , and simultaneously the same sequence is a mark - up of the word @xmath871 w.r.t . the automaton @xmath1753 .
therefore @xmath1754 , where @xmath1755 is a sequential composition of automata @xmath1756 and @xmath1753 .
by corollary [ cor : cover ] , the segment @xmath1724)$ ] can be covered by a countably many segments @xmath1757 $ ] , @xmath402 where for every @xmath138 there exists a state @xmath1758 of the automaton @xmath366 such that the set of all points from @xmath1757 $ ] whose base-@xmath8 expansions ( w.r.t . the function @xmath449 ) admit @xmath1758-mark - ups is dense in @xmath1757 $ ] . given a real number @xmath1759 $ ] and its base-@xmath8 expansion , in view of proposition [ prop : mark - up ] there exists a @xmath1758-mark - up of the base-@xmath8-expansion .
having this mark - up and by acting as above , we , given @xmath1760 find corresponding state @xmath1761 of the automaton @xmath0 and construct a strictly increasing sequence over @xmath566 such that the sequence is simultaneously a mark - up for @xmath575 ( w.r.t . @xmath1758 and the function @xmath449 ) and for @xmath53 ( w.r.t . @xmath1761 and the function @xmath85 ) .
let @xmath1762 be all pairwise distinct states of the automaton @xmath0 that satisfy the following condition : for every @xmath1763 such that @xmath1764 $ ] , @xmath1765 there exists an @xmath1766-mark - up ( for suitable @xmath1767 ) such that the mark - up is a mark - up both for @xmath16 ( w.r.t .
@xmath1766 and @xmath85 ) and for @xmath871 ( w.r.t .
@xmath1758 and @xmath449 ) simultaneously . for @xmath1767 denote via @xmath1768 the set of all infinite words @xmath1421 such that there exists an @xmath242-mark - up which is a mark - up both for @xmath16 and for @xmath871 simultaneously ; then proceeding in the same way as in the proof of proposition [ prop : cover ] we conclude that there exists @xmath1769 and a closed subinterval @xmath1719 $ ] such that @xmath1770 $ ] is dense in @xmath1719 $ ] .
but then @xmath1771 is dense in @xmath1772)$ ] and @xmath1773 is dense in @xmath1774))$ ] as @xmath85 is continuous on @xmath1719 $ ] and @xmath449 is continuous on @xmath1772)$ ] .
therefore for a finite automaton @xmath1775 which is a sequential composition of the automata @xmath1776 and @xmath1777 we have that the graph of the restriction @xmath1026 of the function @xmath1727 to @xmath1719 $ ] lies in @xmath1778 .
[ note : comp ] by arguing as in the proof of proposition [ prop : comp ] the following can be shown : _
let @xmath1552\subset[0,1]$ ] , let @xmath1551{\rightarrow}\mathbb s$ ] , @xmath1779{\rightarrow}\mathbb s$ ] be two continuous functions , and let there exist finite automata @xmath0 and @xmath366 such that @xmath1680}(g)\subset\mathbf p(\mathfrak a)$ ] , @xmath1680}(f)\subset\mathbf p(\mathfrak b)$ ] .
then there exists a covering @xmath1592\subset[a , b]{\colon}j\in j\}$ ] such that if @xmath1726 is a restriction of the function @xmath1780 to the sub - interval @xmath1554 $ ] then @xmath1728}(h_j)\subset\mathbf p(\mathfrak c_j)$ ] for every @xmath1729 , where @xmath1730 is a sum of the automaton @xmath1731 with the automaton @xmath1732 and @xmath1733 are suitable states of the automata @xmath0 , @xmath366 accordingly .
_ here by the sum of automata @xmath0 and @xmath366 we mean a sequential composition of the automata by automaton which has two inputs and a single output and performs addition of @xmath8-adic integers .
the latter automaton is finite , see subsection [ ssec : a - map ] and proposition [ prop : fin - auto ] . note also that we may assume that both @xmath449 and @xmath85 are defined on an arc of @xmath528 rather than on @xmath1552 $ ] .
[ cor : sum ] given @xmath1781 and continuous finitely computable functions @xmath1782{\rightarrow}\mathbb s$ ] , there exists a covering @xmath1592\subset[a , b]{\colon}j\in j\}$ ] such that the function @xmath1783 is finitely computable on every @xmath1784 $ ] . comparing theorem [ thm : eval - erg ] with proposition [ prop : comp ]
we see that in the class of continuous functions there is no big difference between finite computability and ultimate finite computability since given a finitely computable continuous function on a segment there exists a covering of the segment by sub - segments such that the function is ultimately finitely computable on either of the sub - segments .
in this section we prove that a graph of any @xmath19-smooth finitely computable function @xmath1551{\rightarrow}\mathbb s$ ] , @xmath1552\subset[0,1)$ ] , lies ( under a natural association of the half - open interval @xmath148 with the unit circle @xmath528 ) on a torus winding with a @xmath8-adic rational slope ; and if @xmath0 is a finite automaton that computes @xmath85 then necessarily the graph of the automaton contains the whole winding .
moreover , we prove a generalization of this theorem for multivariate functions . to make further proofs ( which are somewhat involved ) more transparent we begin with a brief ( and no too rigorous ) outline of their general underlying idea . given @xmath85 as above , fix @xmath1785 $ ] ; then for @xmath1786 $ ] and all sufficiently large @xmath287 from the differentiability of @xmath85 it follows that @xmath1787 , where @xmath1788 and @xmath229 is a map from @xmath399 to @xmath399 such that @xmath1789 faster than @xmath1790 while @xmath1791 . once @xmath1026 is fixed ( say , @xmath1792 ) then the above equality for large @xmath287 implies ( in view of proposition [ prop : comp ] and corollary [ cor : sum ] ) that there exists a finite automaton @xmath1793 which computes @xmath1794 being feeded by an infinite sequence of zero words whose lengths increase unboundedly , i.e. , @xmath1795 : this is because , speaking loosely , the error term @xmath1796 makes no perturbations of the infinite output sequence due to the fast growth of @xmath1797 .
but then necessarily @xmath1798 by proposition [ prop : auto - q ] .
further lemma [ le : der ] proves this fact rigorously .
we then ( see lemma [ le : der2 ] below ) play similar trick with the second derivative @xmath1799 : as @xmath85 is two times differentiable and @xmath1798 , the function @xmath1800 of argument @xmath1801 $ ] is also a @xmath19-smooth finitely computable function for every @xmath1802 . as @xmath1803 , @xmath1804 , we have ( for all sufficiently large @xmath287 ) that @xmath1805 where @xmath1806 , @xmath1807 , and @xmath16 is a map from @xmath399 to @xmath399 such that @xmath1808 as @xmath1791 . from here in a way similar to that of above we deduce that @xmath1809 .
but then , if @xmath1810 , the argument means that there exists a finite automaton which performs squaring @xmath1811 of every @xmath1786 $ ] with arbitrarily high accuracy . however as it is well known ( cf . subsection [ ssec : auto ] ) no finite automaton can do such squaring ; so necessarily @xmath1812 for all @xmath1785 $ ] .
but the set of these @xmath53 is dense in @xmath1552 $ ] ; therefore @xmath1812 for all @xmath1813 $ ] as @xmath1814 is continuous on @xmath1552 $ ] .
hence @xmath85 must be affine : @xmath1815 for all @xmath1801 $ ] .
note that then necessarily @xmath1816 since @xmath1817 and @xmath85 is finitely computable , cf .
proposition [ prop : auto - q ] .
after that by proposition [ prop : ext ] we can ` stretch ' the graph of the function @xmath85 from @xmath1552 $ ] to the whole unit circle @xmath528 and thus finally obtain a whole cable which lies in the plot of the finite automaton which calculates @xmath85 .
but then by theorem [ thm : mult - add - p ] the plot must contain the whole link of torus windings ; and the graph @xmath1680}(g)$ ] must lie completely on some of these windings . the number of links is finite since every link corresponds to some minimal sub - automaton ( see subsection [ ssec : auto ] and theorem [ thm : eval - erg ] ) of the automaton which computes @xmath85 ; and the number of minimal sub - automata of a finite automaton is clearly a finite .
finally , every such link corresponds to a finite family of complex - valued exponential functions of the form @xmath1818 , @xmath1819 , for suitable @xmath1781 as shown in corollary [ cor : mult - add - compl ] . figures [ fig:2link - square ] and [ fig:2link - torus ] illustrate how the graphs of @xmath19-functions from the plots of finite automata look like .
now we proceed with rigorous assertions and proofs . here
we show that @xmath19-smooth finitely computable functions defined on @xmath1552\subset[0,1)$ ] and valuated in @xmath148 are only affine ones .
once we associate the half - open interval @xmath148 with a unit circle @xmath528 under a natural bijection we may consider graphs of the functions as subsets on a surface of the unit torus @xmath1820 .
we show that then the graphs lie only on cables of the torus @xmath23 , and the slopes of the cables must be @xmath8-adic rational integers ( i.e. , must lie in @xmath276 ) , see subsection [ ssec : knot ] for definitions of torus knots , cables of torus , and links of knots .
[ thm : main ] consider a finite automaton @xmath0 and a continuous function @xmath85 with domain @xmath1552\subset[0,1)$ ] , valuated in @xmath1821 .
let @xmath1822 , let @xmath85 be two times differentiable on @xmath1552 $ ] , and let the second derivative @xmath1814 of @xmath85 be continuous on @xmath1552 $ ] .
then there exist @xmath1823 such that @xmath1824 for all @xmath1813 $ ] ; moreover , the graph @xmath1680}(g)$ ] of the function @xmath85 lies completely in the cable @xmath1825 and @xmath1826 for all @xmath1827 . given a finite automaton @xmath0 ,
there are no more than a finite number of pairwise distinct cables @xmath1828 of the unit torus @xmath23 such that @xmath1829 .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ( 0,0)(15,15 ) ( -4.8,0)(13.2,18 ) ( -4.8,0.1).3 ( -4,10)(1,0 ) ( 1,15)(8.5,0 ) ( 8.5,15)(11,10 ) ( -4,5)(-1.5,0 ) ( -1.5,15)(6,0 ) ( 6,15)(11,5 ) ( -4,1.71)(11,10.71 ) ( -4,0.43)(11,9.43 ) ( -4,0.86)(11,9.86 ) ( -4,4.71)(11,13.71 ) ( -4,3.43)(11,12.43 ) ( -4,3.86)(11,12.86 ) ( -4,7.71)(8.14,15 ) ( -4,6.43)(10.29,15 ) ( -4,6.86)(9.3,15 ) ( -4,10.71)(3.14,15 ) ( -4,9.43)(5.29,15 ) ( -4,9.86)(4.3,15 ) ( -4,13.71)(-1.86,15 ) ( -4,12.43)(0.29,15 ) ( -4,12.86)(-0.3,15 ) ( -1.86,0)(11,7.71 ) ( 0.29,0)(11,6.43 ) ( -0.3,0)(11,6.86 ) ( 3.14,0)(11,4.71 ) ( 5.29,0)(11,3.43 ) ( 4.3,0)(11,3.86 ) ( 8.14,0)(11,1.71 ) ( 10.29,0)(11,.43 ) ( 9.3,0)(11,.86 ) _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ in @xmath821 .
the plot consists of two torus links ; the links consist of 2 and of 3 knots accordingly.,scaledwidth=92.0% ] [ le : der ] consider a finite automaton @xmath0 and a continuous function @xmath85 with domain @xmath1552\subset[0,1)$ ] valuated in @xmath1821 .
let @xmath1822 and let @xmath85 be differentiable at the point @xmath1830 where @xmath483 .
then @xmath1798 .
[ le : der2 ] under conditions of theorem [ thm : main ] let @xmath53 be the same as in the statement of lemma [ le : der ] ; then @xmath1812 . under conditions of the lemma , the right base-@xmath8 expansion of @xmath53 is @xmath1831 , for suitable @xmath1832 .
we claim that @xmath1833 .
indeed , as @xmath85 is continuous and as @xmath1834 where @xmath1835 , there exists an @xmath242-mark - up of the right - infinite word @xmath1836 w.r.t .
some state @xmath352 ( cf .
note [ note : mark - up ] ) .
that is , there exists a strictly increasing sequence @xmath764 over @xmath566 such that for the infinite sequence of words @xmath1837 of strictly increasing lengths @xmath670 ( where @xmath1838 and @xmath1372 is large enough so that @xmath1839 ) the following is true : @xmath1840 therefore , the mark - up @xmath1841 of the zero right - infinite word @xmath1842 is such that ( in the notation of proposition [ prop : ext ] ) the following equalities hold simultaneously @xmath1843 now by combining proposition [ prop : ext ] ( or corollary [ cor : ext ] if necessary ) and proposition [ prop : auto - q ] we conclude that @xmath1844 where @xmath1845 ; therefore @xmath1833 as @xmath1846 . note that if @xmath1847 $ ] then we apply corollary [ cor : ext ] rather than proposition [ prop : ext ] and use @xmath1721 instead of @xmath1705 and @xmath635 instead of @xmath4 here and after .
take @xmath434 ; then by differentiability of @xmath85 , for all @xmath1848 and all sufficiently large @xmath690 we can represent @xmath1849 as @xmath1850 where @xmath1851 , @xmath1788 and @xmath229 is a map from @xmath399 to @xmath399 such that @xmath1789 faster than @xmath1790 while @xmath1791 .
that is , for all sufficiently large @xmath287 we may represent @xmath1797 as @xmath1852 , where @xmath16 is a map from @xmath399 to @xmath399 such that @xmath1808 as @xmath1791 .
further , by proposition [ prop : ext ] , the function @xmath1853 is continuous on @xmath14 $ ] and @xmath1686}(\tilde g)\subset\mathbf{p}(\mathfrak a)$ ] . from here by combining proposition [ prop : comp ] and theorem [ thm : mult - add - p ] we conclude that there exists a finite automaton @xmath368 such that the graph @xmath1686}(\bar g)$ ] of the function @xmath1854 lies completely in @xmath1855 .
indeed , as @xmath1833 then by the graph of the continuous function @xmath1856 on @xmath14 $ ] lies completely in @xmath1857 for a finite automaton @xmath366 whose automaton function is @xmath1858 , @xmath504 , and therefore the composite function @xmath1859 is finitely computable on @xmath14 $ ] , cf .
proposition [ prop : comp ] . we proceed with this in mind .
we see from that for all sufficiently large @xmath287 and all @xmath1026 the following is true : @xmath1860 now for the rest of the proof we take ( and fix ) @xmath1861 .
let @xmath1862 be a base-@xmath8 expansion of @xmath1863 ; so @xmath1864 is a base-@xmath8 expansion of @xmath1865 .
we may assume that @xmath1866 since if otherwise we consider the function @xmath1867 which satisfies conditions of the lemma as @xmath85 satisfies these conditions .
if there exists two different base-@xmath8 expansions for @xmath1863 we will consider only one of these . recall that these expansions are of the form @xmath1868 and @xmath1869 where @xmath1870 , @xmath1871 and @xmath1872 . now ,
if the function @xmath1873 is non - negative for an infinite number of @xmath434 , then we take the first of the base-@xmath8 expansions ; and we take the second one in the opposite case .
we claim that in all cases mentioned above there exists a strictly increasing sequence @xmath1874 of @xmath434 such that , speaking loosely , the term @xmath1875 has no affect on higher order digits of the base-@xmath8 expansion of the right - hand part of . in the case when @xmath1863 admits only one base-@xmath8 expansion this follows from the fact that @xmath1876 tends to 0 faster than @xmath1877 as we may take for @xmath1874 all sufficiently large @xmath287 . in the case
when @xmath1863 admits two base-@xmath8 expansions the claim is also true since we consider a right base-@xmath8 expansion @xmath1878 and assume that the function @xmath1873 in is non - negative for an infinite number of @xmath434 : in that case we take for @xmath1874 all sufficiently large @xmath287 such that @xmath1879 . when @xmath1863 admits two base-@xmath8 expansions and the function @xmath1873 is non - negative only for a finite number of @xmath434 , we consider a left base-@xmath8 expansion @xmath1880 where @xmath1870 , @xmath1871 and @xmath1872 .
then there exists infinitely many @xmath690 such that @xmath1881 , and we take for @xmath1874 all these sufficiently large @xmath287 .
in other words , if we take @xmath1882 , substitute @xmath1883 to @xmath1859 and apply then we get @xmath1884 where @xmath1885 for @xmath1886 , @xmath1887 ( note that @xmath1888 depends on @xmath287 ) .
further , by note [ note : mark - up ] we conclude now that given a right - infinite word @xmath436 and @xmath434 there exists an @xmath242-mark - up of the word @xmath1889 w.r.t .
the function @xmath1890 where @xmath242 is a suitable ( depending on @xmath436 and @xmath287 ) state of a finite automaton @xmath368 which is a sequential composition of the automaton @xmath0 with the automaton @xmath1072 this means in particular that given an infinite word @xmath1891 , for any @xmath1882 there exists a mark - up ( w.r.t . a suitable state @xmath1892 of the automaton @xmath1893 , cf .
proposition [ prop : mark - up ] ) @xmath1894 of the word @xmath1895 . as a total number of states of the automaton
@xmath368 is finite , at least one state , say @xmath1540 , in the sequence @xmath1896 occurs infinitely many times . denote @xmath1897 ( then @xmath1755 is a finite automaton as well ) and consider an infinite strictly increasing sequence @xmath1898 . given a term @xmath1899 of the latter sequence @xmath1900 take the smallest @xmath1503 such that @xmath1901 ; denote via @xmath1902 the state the automaton @xmath1755 reaches after being feeded by the word @xmath1903 . as the number of states of the automaton @xmath1755 is finite , in the infinite sequence @xmath1904 at least one term ,
say @xmath1905 , occurs infinitely many times .
consider an infinite sequence @xmath1906 such that @xmath1907 and consider an automaton @xmath1908 ( whence the latter automaton is finite also ) .
if the automaton @xmath1908 is being feeded by the word @xmath1909 then the automaton outputs the word @xmath1910 ; therefore being feeded by the word @xmath1911 the automaton outputs the word @xmath1912 of length @xmath1913 .
as @xmath1808 while @xmath1791 we may assume without loss of generality that the sequence @xmath1914 is strictly increasing ( since if otherwise we consider a subsequence @xmath1915 of the sequence @xmath1916 such that the sequence @xmath1917 ) .
now by mimic the proof of proposition [ prop : auto - q ] we show that the sequence @xmath1918 has only finitely many limit points and all these limit points are in @xmath276 . but from it follows that @xmath1919 ; therefore @xmath1920 and thus @xmath1921 .
let @xmath53 be as in the statement of lemma [ le : der ] ;
i.e. , let the right base-@xmath8 expansion of @xmath53 be as in the proof of lemma [ le : der ]
. then @xmath1922 by lemma [ le : der ] .
consider the function @xmath1800 of argument @xmath1801 $ ] where @xmath1802 ; then @xmath1923 is two times differentiable on @xmath1552 $ ] and @xmath1803 , @xmath1804 for all @xmath1801 $ ] .
as @xmath1923 is continuous on @xmath1552 $ ] , the constant @xmath247 may be taken so that @xmath1924 $ ] for all @xmath436 from a sufficiently small closed neighborhood @xmath1925 $ ] of @xmath53 .
we are going to prove that @xmath1926 for all @xmath1813 $ ] .
note that if @xmath1927 for some @xmath1813 $ ] then @xmath1928 does not change its sign on a sufficiently small neighborhood @xmath1929 $ ] of @xmath53 .
indeed , if not , then there exist two infinite sequences , @xmath1930 and @xmath1931 such that all the terms of either sequence are pairwise distinct , @xmath1932 , and @xmath1933 , @xmath1934 for all @xmath155 . but as @xmath1935 ( since @xmath1928 is continuous at @xmath53 ) then necessarily @xmath1926 ; but this contradicts our assumption that @xmath1927 .
we therefore may assume that @xmath1936 for all @xmath1937 $ ] ; otherwise consider the function @xmath1938 rather than @xmath1923 .
finally by corollary [ cor : sum ] we conclude that @xmath1923 is finitely computable on a sufficiently small closed neighbourhood @xmath1939 $ ] of @xmath53 .
further we use @xmath85 for @xmath1923 and @xmath1552 $ ] for @xmath584 without risk of misunderstanding .
thus we have : 1 .
@xmath85 is finitely computable on @xmath1552\ni x$ ] ; 2 .
@xmath85 is two times differentiable on @xmath1552 $ ] ; 3 .
@xmath1814 is continuous on @xmath1552 $ ] ; 4 .
@xmath1940 on @xmath1552 $ ] ; 5 .
@xmath1941 ; now , since @xmath85 is two times differentiable on @xmath1552 $ ] , for all @xmath1942 and all sufficiently large @xmath690 we can represent @xmath1943 as @xmath1944 where @xmath1945 stands for @xmath1946 , @xmath1788 , @xmath1807 , and @xmath16 is a map from @xmath399 to @xmath399 such that @xmath1808 as @xmath1791 .
we prove that by mimic of the respective part of the proof of lemma [ le : der ] .
firstly we show that @xmath1833 as in the proof of lemma [ le : der ] ; thus considering a function @xmath1948 we see that @xmath1949 for all sufficiently large @xmath287 .
let @xmath1862 be a base-@xmath8 expansion of @xmath1950 ( we may take either of the expansions if there exist two different ones ) ; so @xmath1951 is a base-@xmath8 expansion of @xmath1945 .
take @xmath1952 ; then for all sufficiently large @xmath287 the base-@xmath8 expansion of the right - hand part of is of the form @xmath1953 where @xmath1885 for @xmath1954 depend on @xmath287 .
the function @xmath1955 of argument @xmath189 is continuous and finitely computable on @xmath14 $ ] by a finite automaton @xmath368 . now considering an infinite word @xmath1956 with the corresponding mark - up we prove in the same way as in lemma [ le : der ] that the corresponding sequence of finite output words of the automaton @xmath368 is a sequence of initial finite sub - words of the infinite word @xmath1957 and
then deduce as in the proof of lemma [ le : der ] that @xmath1947 ( note that given @xmath1958 we may always take @xmath4 so that @xmath1959 without altering the value of @xmath53 just by multiplying both numerator and denominator by a suitable power of @xmath8 ) .
now we prove that @xmath1960 .
assume that @xmath1961 ; that is , that @xmath1962
. let @xmath1963 , @xmath237 , be a @xmath8-adic absolute value of @xmath1945 ; therefore @xmath1964 , where @xmath885 , @xmath882 is a unity of @xmath177 , cf .
subsection [ ssec : p - adic ] . by claim 1 , @xmath1947 ,
so necessarily @xmath1965 and @xmath1966 ; whence @xmath881 and thus @xmath1967 as @xmath882 is a unity .
therefore the function @xmath1968 of argument @xmath539 $ ] is finitely computable on @xmath14 $ ] ( say , by the automaton @xmath366 ) : this follows from propositions [ prop : ext ] , [ prop : comp ] , and corollary [ cor : sum ] .
now by we conclude that @xmath1969 where @xmath1970 ( we may assume that the inequality is true just by taking @xmath4 sufficiently large by multiplying both numerator and denominator by a suitable power of @xmath8 and thus without altering the value of @xmath53 ) , @xmath1806 , @xmath1971 , and @xmath711 is a map from @xmath399 to @xmath399 such that @xmath1972 as @xmath1791 .
further , given @xmath1973 and a word @xmath1974 and taking @xmath1975 with @xmath1976 we have that @xmath1977 .
let @xmath1978 be corresponding mark - up of the infinite word @xmath1979 ( the latter infinite word corresponds to @xmath1975 once @xmath1976 ) .
take @xmath228 the smallest so that @xmath1980 ; denote @xmath1981 the state the automaton @xmath366 reaches after being feeded by the word @xmath1982 . as the number of states of the automaton @xmath366 is finite ,
in the sequence @xmath1983 at least one state , say @xmath1984 , occurs infinitely many times .
consider a strictly increasing sequence @xmath1985 ; then once @xmath1986 , the automaton @xmath1987 , being feeded by the word @xmath1988 , outputs the word @xmath1989 . from it follows that @xmath1990 .
therefore @xmath1991 where @xmath1992 ( in other words , @xmath1993 is a base-@xmath8 expansion of the square of the number whose base-@xmath8 expansion is @xmath1994 ) . thus necessarily
@xmath1995 for all sufficiently large @xmath514 . but
@xmath1977 for all @xmath287 ; thus by proposition [ prop : p - shift ] , @xmath1996 for all @xmath1214 .
therefore for all sufficiently large @xmath514 ( such that @xmath1997 ) we have that @xmath1998 since @xmath1976 . in other words , as the automaton @xmath1987 , being feeded by the word @xmath1988 , outputs the word @xmath1999 once @xmath514 is sufficiently large , @xmath1997 , the automaton @xmath2000 , being feeded by the zero word @xmath2001 , outputs the word @xmath1993 .
this means that , given an arbitrary number @xmath1470 whose base-@xmath8 expansion ( where higher order digits might be 0 ) is of length @xmath94 , @xmath94 being sufficiently large , there exists a state @xmath242 of the finite automaton @xmath366 such that the automaton @xmath2002 , being feeded by a zero sequence of length @xmath2003 , outputs a word ( of length @xmath2003 ) that is a base-@xmath8 expansion of @xmath2004 . but
this is not possible since it is well known that squaring is not possible by a finite automaton ( cf.,e.g . ,
* theorem 2.2.3 ) or ( * ? ? ?
* proposition 7.1.6 ) ) ; however , a short proof follows .
as the automaton @xmath366 is finite , then there are only finitely many sub - automata @xmath2005 . but any finite automaton , being feeded by a sufficiently long zero word @xmath2006 outputs the word of the form @xmath2007 , where @xmath2008 , @xmath2009 , @xmath2010 , and the words @xmath2011 are completely determined by the finite automaton
, @xmath2012 is a right prefix of @xmath2013 , cf .
lemma [ le : per ] . but given finitely many words @xmath2014 of that sort , @xmath2015 , there exist infinitely many words @xmath1993 which are base-@xmath8 expansions of squares of numbers from @xmath399 and which are not of the form @xmath2016 , @xmath2017 , @xmath2018 .
the contradiction proves that @xmath1960 ; therefore , @xmath1812 .
we already have proved that @xmath1812 if @xmath1830 , where @xmath483 , @xmath1682 .
but the set of all these @xmath53 is dense in @xmath1552 $ ] , so , as the second derivative @xmath1814 is is continuous on @xmath1552 $ ] by the condition of the theorem under proof , @xmath1814 must vanish everywhere on @xmath1552 $ ] ; therefore , @xmath2019 . but
this implies that there exist @xmath2020 such that @xmath90 for all @xmath1813 $ ] and @xmath2021 for all @xmath1813 $ ] . from lemma
[ le : der ] it follows now that @xmath2022 . now taking an arbitrary number @xmath2023 we see that @xmath2024 by proposition [ prop : auto - q ] ; hence @xmath2025 must be also in @xmath276 as @xmath2026 .
now we will prove that @xmath1825 . to begin with
, we note that by theorem [ thm : eval - erg ] there exists a minimal sub - automaton @xmath1615 and a segment @xmath1719\subset[a , b]$ ] such that @xmath2027}(g)\subset\mathbf{lp}(\mathfrak a^\prime)$ ] .
taking @xmath2028 as in the statement of proposition [ prop : ext ] , we conclude that the graph @xmath1686}(g_d)$ ] of the function @xmath2029 on @xmath14 $ ] lies completely in @xmath2030 ; thus @xmath1686}(g_d)=\{(x;g_d(x)){\colon}x\in[0,1]\}\subset \mathbf{lp}(\mathfrak a^\prime)$ ] by corollary [ cor : mark - up ] .
@xmath2031 ; that is , @xmath2032 . as @xmath2033
then @xmath2034 for suitable @xmath2035 , @xmath2036 . now given arbitrary @xmath2037
we take @xmath94 and @xmath4 so that @xmath2038 satisfies conditions of proposition [ prop : ext ] ( that is , @xmath2032 ) and @xmath2039 for a suitable @xmath2040 and conclude that @xmath2041\}=\{(x;(ax+ar+p^mb){\mathbin{\mathsf{mod}}}1){\colon}x\in[0,1]\ } \subset \mathbf{lp}(\mathfrak a^\prime)\ ] ] given arbitrary @xmath2037 , the above inclusion holds for all sufficiently large @xmath4 ; therefore due the structure of @xmath2042 ( cf .
subsection [ ssec : const ] ) the following inclusion holds for every @xmath2037 and every @xmath2043 : @xmath2044\ } \subset\mathbf{lp}(\mathfrak a^\prime).\ ] ] but @xmath2045\}=\{((x{\mathbin{\mathsf{mod}}}1;(ax+b^\prime){\mathbin{\mathsf{mod}}}1){\colon}x\in{\mathbb r}\ } = \mathbf c(a , b^\prime)$ ] ; therefore we have shown that @xmath2046 for all @xmath2047 .
that is , @xmath2048 contains the whole link of cables @xmath2049 for all @xmath2050 ( i.e. , contains @xmath2051 where @xmath2052 , @xmath185 , cf . theorem
[ thm : mult - add - p ] ) and @xmath1680}(g)$ ] lies completely in a suitable cable of the link .
this proves the first claim of theorem [ thm : main ] since @xmath2053 , cf .
note [ note : sub - auto ] . to prove the second claim , given a finite automaton @xmath11 consider all cables @xmath1828 such that @xmath2054 ; whence by the first claim of the theorem all these cables lie in @xmath563 .
moreover , as we have shown during the proof of the first claim of the theorem , for either of the cables @xmath1828 there exists a minimal sub - automaton @xmath2055 of the automaton @xmath0 such that @xmath2056 .
the cables cross zero meridian @xmath2057 of the torus @xmath820 only when @xmath2058 ; therefore the point set @xmath2059 of all the points where the cables cross zero meridian consists of the points of the form @xmath2060 where @xmath2061 and @xmath2062 contains all the points of the form @xmath2063 where @xmath93 are constant terms of the cables . as @xmath2064 is a finite set ( cf .
proposition [ prop : auto - q ] ) , there are no more than a finite number of pairwise distinct numbers @xmath2065 ( note that cables with equal slopes whose constant terms are congruent modulo 1 coincide ) .
now taking @xmath2066 we see that all the points of the form @xmath2067 of the cables belong to zero meridian and therefore to the finite set @xmath2068 ; hence , there exist no more than a finite number of pairwise distinct numbers @xmath2069 where @xmath575 ranges over rational integers @xmath178 and @xmath92 are slopes of the cables from @xmath2070 .
thus if there exists an infinite number of cables in @xmath21 then there exists a minimal sub - automaton @xmath1615 of the automaton @xmath0 such that @xmath2071 contains an infinite number of cables of the form @xmath2072 with @xmath91 fixed and @xmath2073 ranging through an infinite subset @xmath2074 of @xmath178 so that @xmath2075 are all equal one to another .
therefore for the rest of the proof we may ( and will ) assume that the automaton @xmath0 is minimal . by the first claim of the theorem , @xmath2033 ;
so there exists a unique representation of @xmath92 in the form @xmath2076 where @xmath224 is a period length of @xmath92 , @xmath237 , and @xmath238 , cf .
proposition [ prop : p - repr ] and note [ note : pq - frac - rep ] .
therefore @xmath2074 must contain an infinite subset of numbers from the coset @xmath2077 for a suitable @xmath2078 since @xmath2079 implies @xmath2080 , i.e. , @xmath2081 .
thus from the assumption that there are infinitely many pairwise distinct cables in @xmath21 it follows that then in @xmath563 there exist infinitely many cables of the form @xmath2082 with @xmath2083 fixed ( @xmath2084 , @xmath2085 ) and @xmath89 running through an infinite subset @xmath2086 . by considering @xmath2087 ( and the corresponding finite automaton ) if necessary we may assume that @xmath2088 is an infinite subset of @xmath566 .
therefore , @xmath2088 constitutes a strictly increasing sequence @xmath2089 of natural numbers .
now take arbitrary @xmath2090 and consider a sequence @xmath2091 . as the sequence @xmath2092 is strictly increasing , @xmath2093
; therefore @xmath2094 .
therefore @xmath2095 as @xmath2096 and @xmath563 is closed in @xmath820 , cf .
corollary [ cor : ap = lp ] .
thus we have proved that zero meridian @xmath2097 of the torus @xmath820 lies completely in @xmath563 . on the other hand , if @xmath2098 then @xmath1443 by definitions of @xmath563 and @xmath2099 , see subsections [ ssec : plots ] and [ ssec : mark - up ]
; but there are only finitely many points in @xmath2099 by proposition [ prop : auto - q ] .
the contradiction proves the second claim of the theorem . in this subsection
we are going to extend theorem [ thm : main ] for the case of finite automata with multiply inputs / outputs .
note that actually an automaton over alphabet @xmath198 with @xmath4 inputs and @xmath94 outputs can be considered as a letter - to - letter transducer with a single input over the alphabet @xmath2100 and a single output over the alphabet @xmath2101 ; therefore the plot of that automaton is a closed subset of the unit square @xmath64 .
we however are going to consider plots of automata of that sort as subsets of multidimensional unit hypercube @xmath2102
. therefore automata functions of such automata are 1-lipschitz mappings from @xmath496 to @xmath495 , see subsection [ ssec : a - map ] ; and vice versa , every 1-lipschitz mapping from @xmath2103 is an automaton function of a suitable automaton @xmath0 with @xmath4 inputs and @xmath94 outputs over the alphabet @xmath18 .
note that @xmath2104 where @xmath2105 ( @xmath2106 ) is 1-lipschitz and therefore is an automaton function of an automaton with @xmath4 inputs and a single output .
now we re - state our definition of a ( limit ) plot for that case of automata with @xmath4 inputs and @xmath94 outputs .
[ def : plot - auto - mult ] given an automaton function @xmath2107 define a set @xmath2108 of points of @xmath2109 as follows : for @xmath554 denote @xmath2110 a point set in a unit real hypercube @xmath2102 ; here given @xmath2111 we put @xmath2112 then take a union @xmath2113 and denote via @xmath2114 a closure ( in topology of @xmath2115 ) of the set @xmath2116
. given an automaton @xmath0 , we call a _ plot of the automaton _
@xmath0 the set @xmath15 .
we call a _ limit plot _ of the automaton @xmath0 the point set @xmath563 which is defined as follows : a point @xmath2117 lies in @xmath563 if and only if there exist @xmath2118 and a strictly increasing infinite sequence @xmath565 of numbers from @xmath566 such that simultaneously @xmath2119 to put it in other words , at every step a letter - to - letter transducer @xmath0 ( which has @xmath4 inputs and @xmath94 outputs over a @xmath8-symbol alphabet @xmath18 ) * obtains a vector @xmath2120 ( each @xmath67-th letter @xmath2121 is sent accordingly to the @xmath67-th input of the automaton , @xmath2122 ) , * outputs a vector @xmath2123 ( each @xmath514-th output of the automaton outputs accordingly the letter @xmath2124 , @xmath2125 ) which depends both on the current state and on the input vector @xmath487 , * reaches the next state ( which depends both on @xmath2126 and on the current state )
. then the routine repeats .
therefore after @xmath138 steps the automaton @xmath11 transforms the input @xmath4-tuple @xmath2127 of @xmath138-letter words @xmath2128 ( @xmath2122 ) into the output @xmath94-tuple @xmath2129 of @xmath138-letter words @xmath2130 ( @xmath2131 ) . for @xmath2132
running over all @xmath4-tuples of @xmath138-letter words , @xmath554 we consider the set @xmath2133 of all points @xmath2134 ; here @xmath2135 stands for @xmath2136 where @xmath2137 are @xmath138-letter words
. then we define @xmath2 as a closure in @xmath2115 of the set @xmath2133 . following the lines of note [ note : plot - auto - in ]
it can be shown that @xmath2138 .
we stress that @xmath0 is a synchronous letter - to - letter transducer ; that is why in the definition of the plot all @xmath4 input words as well as corresponding @xmath94 output words of the automaton must have pairwise equal lengths . given a real function @xmath2139 with the domain @xmath2140 , by the _ graph of the function _ ( on the torus @xmath2141 ) we mean the point subset @xmath2142 .
note that if @xmath2143 then @xmath2144 stands for @xmath2145 .
[ thm : main - mult ] let @xmath0 be a finite automaton over the alphabet @xmath1316 , let @xmath0 have @xmath4 inputs and @xmath94 outputs , and let @xmath2146= [ a_1,b_1]\times\cdots\times[a_m , b_m]{\rightarrow}[0,1)^n$ ] ( where @xmath2147\subset[0,1)$ ] , @xmath2148{\rightarrow}[0,1)$ ] , @xmath2122 ) be a two times differentiable function such that all its second partial derivatives are continuous on @xmath2149 $ ] . if @xmath2150 then there exist an @xmath2151 matrix @xmath2152 and a vector @xmath2153 such that @xmath2154 , @xmath2155 ( @xmath2122 ; @xmath2131 ) and @xmath2156 for all @xmath2157 $ ] .
there are not more than a finitely many @xmath2158 and @xmath2159 such that @xmath2154 , @xmath2155 ( @xmath2122 ; @xmath2131 ) and @xmath2160}((\mathbf x \mathbf a+\mathbf b){\mathbin{\mathsf{mod}}}1)\subset\mathbf p(\mathfrak a)$ ] for some @xmath2161\subset[0,1)^m$ ] ; moreover , if @xmath2160}(\mathbf x \mathbf a+\mathbf b)\subset\mathbf p(\mathfrak a)$ ] for some @xmath2161\subset[0,1)^m$ ] then @xmath2162 .
let @xmath2163 be automaton function of the automaton @xmath0 .
having @xmath2164 and @xmath2165 fixed , take arbitrary numbers @xmath2166 $ ] , @xmath2167 , consider the map @xmath2168 and the function @xmath2169 .
as @xmath2166 $ ] and @xmath2170\subset[0,1)$ ] then the map @xmath2171 is a finite automaton function : actually the corresponding automaton @xmath2172 is a sequential composition of the automaton @xmath0 with autonomous automata @xmath2173 which produce accordingly purely periodic output words @xmath2174 ( cf .
corollary [ cor : r - p - repr - qz ] ) and feed accordingly @xmath138-th inputs ( @xmath2175 ) of the automaton @xmath0 while the output of the automaton @xmath2172 is the @xmath514-th output of the automaton @xmath0 .
we assert that @xmath2176 } ( \bar g_{ij})\subset\mathbf p(\bar{\mathfrak a}_{ij})$ ] .
to prove the claim , firstly note that by corollary [ cor : r - p - repr - qz ] , for every @xmath2175 we have that @xmath2177 where @xmath2178 is a period length of @xmath2179 ( see subsection [ ssec : p - adic ] ) .
let @xmath1150 be the least common multiple of all @xmath2180 ; then @xmath2181 for all @xmath2175 .
denote the right - infinite purely periodic word @xmath2182 via @xmath2183 for suitable @xmath2184 .
take arbitrary @xmath2185 $ ] and put @xmath2186 $ ] ; then @xmath2187 .
let @xmath1304 be a base-@xmath8 expansion of @xmath53 ( the word @xmath2188 is right - infinite ) ; then from the definition of the plot it follows that there exists a strictly increasing sequence @xmath2189 over @xmath566 such that @xmath2190 where @xmath2191 see remarks which follow definition [ def : plot - auto - mult ] above .
moreover , since base-@xmath8 of all @xmath2179 are unique , the arguing like in the first part of the proof of proposition [ prop : mark - up ] we conclude that there exists a state @xmath242 of the automaton @xmath0 and a strictly increasing sequence @xmath2192 over @xmath566 such that @xmath2193 where @xmath718 is the automaton which differs from @xmath0 only maybe by the initial state ( which is @xmath242 rather than @xmath47 ) .
now recall that @xmath2194 for all @xmath2175 ; so given @xmath434 let @xmath2195 be the largest such that @xmath2196 and @xmath2197 for some ( thus , for all ) @xmath2175 . since all the words @xmath2198 are periodic with a period of length @xmath1150 such @xmath2199 exists for all sufficiently large @xmath2200 .
denote via @xmath2201 the state the automaton @xmath359 reaches after being feeded ( via respective inputs ) by words @xmath2202 ( @xmath2175 ) and @xmath2203 . by the finiteness of the automaton , in the sequence @xmath2204 at least one state ,
say @xmath2205 , occurs infinitely many times ; therefore from it follows that @xmath2206 where @xmath2207 is the automaton which differs from @xmath0 only maybe by the initial state ( which is @xmath2205 rather than @xmath47 ) .
note that @xmath1150 is a divisor of @xmath2199 by the construction of @xmath2199 since all the words @xmath2198 are periodic with a period of length @xmath1150 . by the definition of the plot
we conclude that prove our claim .
thus the function @xmath2208 satisfies all conditions of theorem [ thm : main ] ; therefore the second derivative of @xmath2208 is zero . but this means by the construction of @xmath2209 that every second partial derivative @xmath2210 is zero for all @xmath2166 $ ] ( @xmath2175 ) and all @xmath2185 $ ] .
as @xmath2211 $ ] is dense in @xmath2170 $ ] for all @xmath2175 we conclude that @xmath2212 everywhere on @xmath2213 $ ] and for all @xmath2131 , @xmath2122 .
now we are going to prove that @xmath2214 vanishes everywhere on @xmath2213 $ ] and for all @xmath2131 , @xmath2215 , @xmath2216 ( without loss of generality , let @xmath2217 in what follows ) .
assume that the opposite is true , that is , that there exist @xmath2218 and a point @xmath2219 $ ] such that @xmath2220 .
then due to the continuity of second partial derivatives of the function @xmath77 , by using the argument similar to that from the beginning of the proof of lemma [ le : der2 ] we conclude that there exist a point ( which without risk of misunderstanding we denote by the same symbol @xmath2221 ) in @xmath2222 and a neighborhood @xmath584 of that point such that @xmath2223 everywhere on @xmath584 . therefore we always may take @xmath2166 $ ] ; @xmath2224 ; @xmath634 and @xmath2225 such that the point @xmath2226 lies in @xmath584 for all @xmath2227 , @xmath2228 and all @xmath2229 . arguing like in the proof of proposition [ prop : ext ] we see that the following inclusion holds : @xmath2230\right\}\subset\mathbf
p(\mathfrak a),\ ] ] where @xmath2231 ( we reduce all coordinates modulo 1 ) .
consider a finite automaton @xmath2232 which is obtained by ` gluing together ' the @xmath67-th and the @xmath229-th inputs of the automaton @xmath0 while feeding the rest @xmath138-th inputs with infinite words @xmath2233 ; that is , the automaton function of the automaton @xmath2232 is @xmath2234 where @xmath2235 , @xmath2236 . by argument
similar to that for the case @xmath2237 ( see the proof of the claim above ) we conclude that the automaton @xmath2232 is finite and that the graph of the function @xmath2238 ^ 2{\rightarrow}[0,1)$ ] when @xmath1026 is running through @xmath148 lies in @xmath2239 .
but on the other hand we have that @xmath2240 since @xmath2241 by what we have already proved above . but
this is a contradiction to theorem [ thm : main ] since the function @xmath2242 of argument @xmath1026 satisfies all conditions of the theorem and has a non - zero second derivative .
thus we have proved that under conditions of theorem [ thm : main - mult ] the function @xmath77 must be affine : @xmath2243 for all @xmath2157 $ ] .
now fix arbitrary @xmath2164 , @xmath2244 , and @xmath2245\cap{\mathbb z}_p\cap{\mathbb q}$ ] for @xmath2106 , @xmath2246 ; consider the function @xmath2208 and the automaton @xmath2172 as in the beginning of the proof of theorem [ thm : main - mult ] . then from the affinity of the function @xmath77 it follows that @xmath2247 . since @xmath2176}(\bar g_{ij})\subset\mathbf p(\bar{\mathfrak a}_{ij})$ ] by the claim above , theorem [ thm : main ] implies that @xmath2248 .
further , arguing like in the proof of proposition [ prop : ext ] we conclude that for suitable @xmath634 and @xmath2249 the graph @xmath1686^m}(h(\mathbf v))$ ] of the function @xmath2250 lies completely in @xmath15 .
now considering the function @xmath2251 and the corresponding automaton @xmath2252 as above for @xmath2253 , @xmath2106 , @xmath2246 , @xmath2254 , we conclude by theorem [ thm : main ] that there are only finitely many @xmath2255 ; whence finitely many @xmath2158 .
if for some of these @xmath2158 there were infinitely many @xmath2256 such that @xmath1686^m}(h(\mathbf v))\subset\mathbf p(\mathfrak a)$ ] then for some @xmath2165 there were infinitely many pairwise distinct @xmath2257 . but given arbitrary @xmath2245\cap{\mathbb z}_p\cap{\mathbb q}$ ] for @xmath2258 and considering corresponding automata @xmath2259 for various @xmath2260-tuples @xmath2261 ( cf .
the beginning of the proof of theorem [ thm : main - mult ] ) , from the construction of @xmath2259 it follows ( cf .
the proof of the claim ) that there are only finitely many these automata @xmath2259 since the automaton @xmath0 is finite . therefore applying theorem [ thm : main ] to every automaton @xmath2259
we finally conclude that there are only finitely many @xmath2257 ; a contradiction to our assumption .
therefore there are only finitely many pairwise distinct functions @xmath2262 as above . now by mimic the respective part of the proof of the first assertion of theorem [ thm : main ] we conclude that given an @xmath2263-matrix @xmath2158 and a vector @xmath2264 over @xmath276 such that the graph of the function @xmath2265 on @xmath2161\subset[0,1]^m$ ] lies completely in @xmath2 then necessarily @xmath2162 . [ note : main - alph ] an automaton with a single input and a single output over respective alphabets @xmath2101 and @xmath2266 , @xmath2267 , can be considered as an automaton with @xmath94 inputs and @xmath138 outputs over an alphabet @xmath1316 and therefore theorem [ thm : main - mult ] can be applied to automata of that sort as well .
now we are going to outline possible relations of main results of preceding section to quantum theory leaving apart applications to cryptography ( the latter are subject of future paper ) .
although further physical interpretation of the results is highly speculative , it reveals deep analogies between automata and quantum systems and thus worth a short discussion to explain a direction in which it is reasonable to develop the results in order to derive some physically meaningful assertions ( and maybe models ) from mathematical theorems of the paper .
we start with some remarks on what is ` physical law ' .
let us ( somewhat naively ) think of a physical law as of mathematical correspondence between quantities which express impacts a physical system is exposed to and quantities which express responses the system exhibits .
suppose for simplicity that both impacts and responses are scalars . as the measured experimental values of physical quantities are rational numbers ( since there is no possibility to obtain during measurements an exact value of irrational number , cf .
@xcite ) the result of measurements are points in @xmath3 , the experimental points . to find a particular physical law one seeks for a correspondence between cluster points ( w.r.t .
the metrics in @xmath266 ) of experimental values and tries to draw an experimental curve .
the latter curve is a ( piecewise ) smooth curve ( the @xmath19-smoothness is common ) which is the best approximation of the set of the experimental points .
a physical law is then a curve which approximate with the highest achievable accuracy ( w.r.t .
metric in @xmath3 ) the experimental curves obtained during series of measurements .
let physical quantities which correspond to impacts and reactions be quantized ; i.e , let they take only values ( measured in suitable units and properly normalized ) , say , @xmath108 , where @xmath32 is an integer .
then , once the system is exposed to a sequence of @xmath138 of impacts , it produces corresponding sequence of @xmath138 reactions .
every impact changes current state of the system to a new one ; therefore provided the systems is causal , both the next state and the reaction ( effect ) depends only on impacts ( causes ) the system has already been exposed to ; so an automaton @xmath0 is an adequate model of the system .
every finite sequence @xmath2268 of impacts / reactions corresponds to a base-@xmath8 expansion of natural number @xmath2269 to which after normalization there corresponds a rational number @xmath2270 .
every measurement is a sequence of interactions @xmath2268 of the measurement instrument with the system , and if the accuracy of the instrument is not better than @xmath2271 , then the result of a single measurement lies within the segment @xmath2272 $ ] . assuming that @xmath2273 we see that even if the system before every measurement has been prepared in a fixed state @xmath47 ( the initial state of the automaton ) during a single measurement the system @xmath327 will be exposed to a random sequences of impacts @xmath2274 which switches the system to a new state @xmath2275 ; so actually as a result of the measurement due to its limited accuracy we obtain an experimental point @xmath2276 where @xmath2277 is the output of the automaton @xmath2278 ( whose initial state is @xmath2275 ) feeded by the sequence @xmath2279 .
theorem [ thm : main ] shows that if the number of states of the system @xmath11 is much less than the length of input sequence of impacts then experimental curves necessarily tend to straight lines ( or torus windings , under a natural map of the unit square onto a torus ) , cf .
figures [ fig : plot-16 ] , [ fig : plot-17 ] , and [ fig : plot ] . this may be judged as linearity of corresponding physical law and , what is even more important , the way experimental points are clustering on the unit square is very much alike to that of the points where electrons hit target screen in a double - slit experiment , cf .
figures [ fig : plot-16][fig : plot-17 ] and figure [ fig:2slit ] .
we are not going here to discuss further parallels of the computer experiments with automata and behaviour of quantum systems such as analogies between transition and ergodic states of automata and mixed and pure quantum states respectively , or probabilities of markov chain related to an automaton and probabilities in quantum systems , etc . : although we believe that the analogies are not external but reflect deep relations between quantum systems and automata , the issues are far from the subject of the paper and that s why the discussion is postponed to further relevant papers . here
we briefly touch only an interesting analogy between smooth curves in plots of finite automata and matter waves of quantum theory . by theorem [ thm : main ] , the smooth curves from the plot of a finite automaton @xmath0 can be described by families of complex - valued exponential functions of the form @xmath1818 , @xmath1819 , for suitable @xmath1781 , cf .
corollary [ cor : mult - add - compl ] .
the wave function of a particle is of the form @xmath2280 where @xmath4 is momentum , @xmath53 position , @xmath857 angular frequency , and @xmath247 is a complex amplitude . comparing the two expressions
we see that @xmath373 may serve as a time for the automaton @xmath0 since multiplication by @xmath373 is a @xmath138-step shift of a base-@xmath8 expansion of a number . but can we someway associate it to physical time @xmath229 of quantum theory ? in what follows
we argue that yes , there is a natural way to do this .
let us forget for a moment that @xmath8 is a positive integer and suppose that @xmath2281 where @xmath2282 is a small real number ; then @xmath2283 and if @xmath117 is a small time interval which is out of accuracy of measurements ( e.g. , let @xmath117 be planck time which is approximately @xmath2284 s. ) .
therefore the torus link @xmath1818 , @xmath1819 can be approximately described by @xmath2285 , @xmath2286 since it is reasonable to assume that @xmath2287 is just a time @xmath229 as @xmath117 is a small time interval , a time quantum , the planck time .
but @xmath2288 is a wave function of a particle with momentum @xmath92 , angular frequency @xmath2289 and amplitude @xmath2290 .
is this mathematically correct to substitute @xmath2291 for @xmath8 in our reasoning ?
yes , this is correct ; but to explain why this is correct we need to recall a notion of _ @xmath111-expansion _ of real number .
the @xmath111-expansions are radix expansions in non - integer bases ; they were first introduced more than half - century ago , see @xcite , and now @xmath111-expansions are a substantial part of dynamics , see e.g. survey @xcite . following @xcite , given @xmath2292 and @xmath2293 , @xmath2294 we call a sequence @xmath2295 over the alphabet @xmath2296 a
_ @xmath111-expansion _ of @xmath53 once @xmath2297 for suitable @xmath2298 .
note that sometimes the term @xmath111-expansion is used in a narrower meaning , when the ` digits ' @xmath2299 are obtained by the so - called ` greedy algorithm ' only , cf .
* section 7.2 ) but this is not important at the moment : in what follows we just sketch the way how the results of current paper can be modified to handle the case of @xmath111-expansions rather than the case of base-@xmath8 expansions only .
we leave details and rigorous proofs for further paper . from the definition
we see that the notion of @xmath111-expansion is a generalization of the notion of base-@xmath8 expansion : it is clear that for @xmath2300 the @xmath111-expansion of @xmath53 is just base-@xmath8 expansion of @xmath53 , and that is why both @xmath111-expansions and base-@xmath8 expansions share some common properties . for instance , given @xmath111-expansion of reals it is possible to perform arithmetic operations with reals in a way similar to that of school - textbook algorithms for base-@xmath8 expansions of reals .
however , differences between base-@xmath8 expansions and @xmath111-expansions should also be taken into the account since when @xmath111 is not an integer , a @xmath111-expansion of a real number is generally not unique ; moreover a real number may have a continuum of different @xmath111-expansions for @xmath111 fixed .
nonetheless , we can perform arithmetic operations with numbers represented by @xmath111-expansions , i.e. , with words over the alphabet @xmath2301 . these operations for some non - integer @xmath111
may be represented by finite automata as well .
for instance , if @xmath23022 $ ] then arithmetic operations with numbers represented by @xmath23032$]-expansions @xmath153 and @xmath648 ( which are binary words over the alphabet @xmath119 since @xmath23042\rfloor=1 $ ] ) can be performed in a manner similar to that when one applies school - textbook algorithms for base-@xmath8 expansions , with the only difference : a ` carry ' from @xmath67-th position should be added to @xmath2305-th position ; e.g. for @xmath2306 we have that @xmath2307 while in the case @xmath2308 we have that @xmath2309 . note that @xmath2310 , @xmath2311 ( and thus @xmath2312 ) when @xmath2306 ; and @xmath2310 , @xmath2313 when @xmath2308 .
when an automaton @xmath0 proceeds a word ( or , a corresponding system reacts to impacts ) it just evaluates step - by - step a @xmath8-adic 1-lipschitz function @xmath385 ( cf .
subsection [ ssec : a - map ] ) , and no @xmath111 appears at this moment .
but we need to specify @xmath111 when we ` visualize ' the function @xmath386 in @xmath3 : to every word @xmath2314 over the alphabet @xmath1 we put into the correspondence a point @xmath2315 ; thus to every pair of input / output words of the automaton there corresponds a point in the unit square @xmath64(or , on the unit torus @xmath2316 ) .
we then take a closure of all these points and obtain a _
@xmath111-plot _ of the automaton @xmath0 in a way similar to that when we constructed a plot of the automaton ( which corresponds to the case when @xmath2300 ) , cf .
definition [ def : plot - auto ] .
we then consider smooth curves in the @xmath111-plots of finite automata , in particular , the curves which correspond to affine automata functions @xmath2317 . to these functions there correspond torus windings which can be expressed in a form of complex - valued functions @xmath2318 , @xmath2319 , @xmath249 ; and these functions can by approximated with arbitrarily high accuracy by functions @xmath2285 , @xmath2320 , just by taking @xmath2294 sufficiently close to 1 .
moreover , the case when @xmath111 is close to 1 is the only case when approximations are of the form of wave functions . but
this means that the _ corresponding automata must necessarily be binary _ ; i.e. , their input / output alphabets are @xmath2321 .
so these automata ( which are just models of causal discrete systems ) indeed produce waves , the _ its _ , _ from bits_. from this view , main results of the current paper may be considered as a contribution to informational interpretation of quantum theory , namely , to j. a. wheeler s _ it from bit _ doctrine which suggests that all things physical ( ` its ' ) are information - theoretic in origin ( ` from bits ' ) , @xcite : we have given some evidence above that this is indeed so regarding particular ` its ' , the matter waves .
we stress once again that our conclusion is based on the following assumptions only : a quantum system is causal and discrete , whence is an automaton ; and the number of states of the automaton is finite .
d. dubischar , v. m. gundlach , o. steinkamp , and a. khrennikov .
the interference phenomenon , memory effects in the equipment and random dynamical systems over the fields of @xmath8-adic numbers .
, 114(4):373382 , 1999 .
n. sidorov .
arithmetic dynamics . in s.
bezuglyi and s. kolyada , editors , _ topics in dynamics and ergodic theory _ ,
volume 310 of _ london math .
lecture note series _ , pages 145189 .
cambridge university press , cambridge , 2003 .
j. vuillemin .
finite digital synchronous circuits are characterized by 2-algebraic truth tables . in _ advances in computing science - asian 2000
_ , volume 1961 of _ lecture notes in computer science _ , pages 17 , 2000
. john a. wheeler .
information , physics , quantum : the search for links . in w.
h. zurek , editor , _ complexity , entropy , and the physics of information _ ,
pages 309336 , redwood city , calif . , 1990 .
addison - wesley pub . | given an automaton ( a letter - to - letter transducer ) @xmath0 whose input and output alphabets are @xmath1 , one visualizes word transformations performed by @xmath0 by a point set @xmath2 of real plane @xmath3 as follows : to an @xmath4-letter non - empty word @xmath5 over the alphabet @xmath6 put into the correspondence a rational number @xmath7 whose base-@xmath8 expansion is @xmath9 ; then to every @xmath4-letter input word @xmath10 of the automaton @xmath11 and to the respective @xmath4-letter output word @xmath12 ( rightmost letters are feeded to / outputted from the automaton prior to leftmost ones ) there corresponds a point @xmath13 of the real unit square @xmath14 ^ 2 $ ] ; denote @xmath15 a closure ( in the topology of @xmath3 ) of the point set @xmath13 where @xmath16 ranges over the set @xmath17 of all non - empty words over the alphabet @xmath18 . for a finite - state automaton @xmath0
, it is shown that once some points of @xmath15 constitute a smooth ( of a class @xmath19 ) curve in @xmath20 , the curve is a segment of a straight line with a rational slope ; and there are only finitely many straight lines whose segments are in @xmath21 . moreover ,
when identifying @xmath15 with a subset of a 2-dimensional torus @xmath22 ( under a natural mapping of the real unit square @xmath14 ^ 2 $ ] onto @xmath23 ) the smooth curves from @xmath15 constitute a collection of torus windings . in cylindrical coordinates either of the windings can be ascribed to a complex - valued function @xmath24 @xmath25 for suitable rational @xmath26 . since @xmath27 is a standard expression for a matter wave in quantum theory ( where @xmath28 ) , and since transducers can be regarded as a mathematical formalization for causal discrete systems , the main result of the paper might serve as a mathematical reasoning why wave phenomena are inherent in quantum systems : this is because of causality principle and the discreteness of matter . |
commonly , the sliding current of an incommensurate charge density wave ( cdw ) is driven by the electric field @xmath3 in the direction @xmath4 of chains , which is originated by the applied voltage @xmath5 and mediated by the normal current density @xmath6 .
new experiments @xcite on mesa junctions in a high magnetic field ( hmf ) @xcite show , for the first time , that the cdw can be driven also by the hall voltage @xmath7 originated by the normal current passing in a transverse direction , fig.[fig : scheme+mesa ] .
the observations are quite counterintuitive with respect to the common notion . in a narrow channel cut
transversely to the chains , the conductivity drops below the threshold rather than rising , shapiro steps are quantized in voltage rather than in current , etc . facing these challenges
, we shall discuss here the hall effect in the deformable cdw at the quantum limit of normal carriers .
our other studies show that the regime of classically strong magnetic fields ( before installing the quantum hall regime at lowest lls ) can not reproduce the complex of experimental dependencies on b and sizes and the scale of the universal josephson ratio for shapiro steps . as a minimal model of the basic experimental geometry shown in fig.[fig : scheme+mesa ] ( left panel ) , we shall consider a hall bar with the current and the magnetic field applied orthogonally in the interchain directions @xmath8 .
small pockets of electrons can be present in the otherwise gapped cdw state . in a layered system at hmf
the carriers are concentrated at one or a few lowest lls forming fractionally ( @xmath9 ) filled qh state .
as always , the lorentz force pushes electrons to one edge @xmath10 of the bar forming the slightly over / under charged boundary layers which give rise to the hall voltage which strength @xmath11 compensates for the lorentz force .
but now , the gigantic polarizability of the cdw reduces the electric field allowing for so strong redistributions of electronic densities which are unthinkable in usual circumstances .
e.g. at low temperature t<1k , the experimental @xmath12 forces all carriers to completely fill a fraction @xmath1 of the chain length , thus forming there the integer qh state , while leaving the fraction @xmath2 of the chain length unoccupied .
in a ( quasi)-2d system at a hmf the electrons form essentially quantum states .
but with transitions among the lls being frozen , the kinetic energy leaves the game , then the density is determined classically by the electric potential @xmath13 . at edges of usual qh setups ,
the density distribution follows almost slavishly the strong coulomb potential which has been build up already before the hmf was applied ; it comes from pronounced finite size effects for the 2d electron gas embedded to the 3d geometry of the device , see @xcite .
but for electrons in pockets upon the cdw , deformations of the cdw phase are able to neutralize any charge so that electrons are allowed to form themselves highly inhomogeneous distributions . here
they are originated intrinsically , to provide the electric field opposing the lorentz force . at low @xmath14 , that extents to separation of the mean carriers density @xmath15 among a fully depleted - @xmath16 region where @xmath17 and an overfilled one - @xmath18 where @xmath19 , i.e. between the two , both nonconducting , regions where the same ll is either full or empty .
of the maximal density in the iqh state and a fully depleted region @xmath20 .
phase slips should take place at the boundary in the sliding regime .
@xmath21 is the bias transverse normal current . @xmath22 and @xmath23 are the compensating normal and collective currents in the chains direction.,title="fig:",scaledwidth=40.0%,height=188 ] of the maximal density in the iqh state and a fully depleted region @xmath20 .
phase slips should take place at the boundary in the sliding regime .
@xmath21 is the bias transverse normal current . @xmath22 and @xmath23 are the compensating normal and collective currents in the chains direction.,title="fig:",scaledwidth=40.0%,height=188 ] we consider a stack of plain rectangles of the size @xmath24 , each composed of @xmath25 chains with a spacing @xmath26 .
each chain segment bears a few @xmath27 electrons . in the hmf
they are supposed to occupy the lowest ll which can acquire as much as @xmath28 electrons .
the mean ll filling is @xmath29 where @xmath30 is the magnetic length .
the profile for the density of carriers @xmath31 is @xmath32 where @xmath33 is the fermi distribution function at a given t and we have assumed that the ll energy @xmath34 is included to the definition of the fermi potential energy @xmath35 .
the level @xmath35 is determined at the fixed number of particles by the condition @xmath36 the in - plane current density in the direction @xmath37 is given by properties of the landau eigenfunctions in crossed electric and magnetic fields : @xmath38 integration over @xmath4 gives the overall @xmath37 current as a function of only values of the boundary potentials : @xmath39 + 1 } { \exp[(\phi_{f}-\phi(l))/t]+1 } \label{j , v0,v1}\ ] ] in the limit @xmath40 and if all states are filled ( @xmath41 which requires that both @xmath42 ) , then we recover the standard result of the iqhe @xmath43 . in the limit @xmath40 for the partial filling @xmath0 ,
the states are all filled over a segment @xmath44 and all are depleted beyond , i.e. @xmath45 , @xmath46 , @xmath47 ; then @xmath48 depends explicitly only on the potential at one boundary . but
this is only a part of the total voltage : the electric potential keeps growing also in the depleted region @xmath49 .
the solution of the elementary electrostatic problem yields the value of @xmath50 at the depleted boundary .
finally the mean current density reproduces the standard result for the classical hall conductance @xmath51 : @xmath52 we have got that , in spite of the iqh filling of the populated area , because the division line between the areas is displaced with changing @xmath53 . by now , there was no voltage drop in @xmath37 direction which can be only the dissipative @xmath54 with a negligible @xmath55 . for that , the longitudinal conductivities @xmath56 , @xmath57 should be small with respect to the transverse one @xmath58 . in conventional qh devices
it comes from localization of carriers ( as the consequence of absence of their kinetic energy ) at irregular potential of dopants . for electrons in nbse3 , with its @xmath59 hierarchy , at least the interchain conductivity @xmath57 is very small which ensures the condition @xmath60 for domination of the transverse ( @xmath61 ) currents @xmath62 and of the lorentz force . allowing the cdw to start sliding brings about the normal counter - current @xmath63 .
the lorentz force component appears in @xmath37 direction giving rise to the additional voltage @xmath64 .
its magnitude is given by the universal law similar to ( [ j , v0,v1 ] ) with interchanged indices @xmath4 and @xmath37 ( and the opposite sign ) .
for the current per chain we get ( the second line in ( [ delta - vy ] ) is written for @xmath65 ) : @xmath66 here @xmath67 is the mean interchain voltage drop ; literarily , it is concentrated only over the filled stripe @xmath68 and it is zero elsewhere while actually the source and the drain edges are expected to be equipotential .
following the experience @xcite from the traditional iqhe , that can be corrected by a consistent @xmath69 solution for the distribution of the carries density and the electric field .
by now , we have imposed the perturbation from the @xmath4 currents upon the already solved case of the bias @xmath37 current .
the last step is to remind that the cdw countercurrent , per chain , is @xmath70 and it requires for phase slips to annihilate with the normal current @xmath71 at @xmath72 .
let each phase slip absorbs / releases @xmath73 electrons per chain , then the repetition frequency is @xmath74 we arrive at the josephson relation with an additional factor @xmath73 ; according to experiments , with a good accuracy @xmath75 which provocatively coincides with the number of electrons per chain .
it can happen that each phase slip swallows all electrons at the enriched end to deliver them at the depleted end .
such a strong event can not be localized only near the ends and the whole dynamical problem needs to be modeled .
recall some relevant properties of the cdw in nbse3 .
the most important feature is the gigantic dielectric constant @xmath76 which source is the cdw polarization .
there are several sources in the literature : @xmath77 at t=50k @xcite ; @xmath78 at t=42k , and @xmath79 at t=20k @xcite ; @xmath80 at t=22k @xcite
. we can extrapolate that to lower @xmath81 with a common wisdom ( while not verified at low @xmath14 ) that @xmath82 which itself is known to be @xmath83 with @xmath8415.2k @xcite . taking this law , or just points from plots in @xcite
, we expect a 20 times reduction from @xmath85 to @xmath86 and 3 times from @xmath87 ; then we get the low @xmath14 saturation at @xmath88 .
notice that @xmath89 corresponds to the pinning length @xmath90 , then there will be no bulk pinning at all for our length scales .
now we summarize the facts related to normal carriers in nbse3 at hmf .
hereafter all numerical estimations will be done for @xmath91 .
zeeman splitting is @xmath92 , but still small to fully polarize the pocket of electrons which partial fermi energy is @xmath93 , then @xmath94 of electrons are available to form spinless pairs for an exchange with the cdw condensate . \2 .
magnetic length @xmath95 is very short in comparison with the chain length @xmath96 .
there are two types @xmath97 of carriers : extrinsic @xmath98 which are electrons and intrinsic @xmath99 which are holes , with effective masses @xmath100 in the @xmath101 plane .
@xmath102 are the cyclotron frequencies where @xmath103 for electrons and an expected bigger value for lighter holes which are for sure in the truly quantum regime , since b>1 t @xcite .
we shall suppose that bandwidths in the heaviest @xmath104 direction are smaller than @xmath105 which statement is marginal for electrons and may be ensured for holes
. then electrons or at least holes form a 2d state within every @xmath101 plane .
the filling factor @xmath1 of one ll is @xmath106 .
then only one - two lls are necessary to accommodate the present @xmath107 electrons per chain segment and only a small fraction of one ll is sufficient for expected @xmath108 holes . precisely for these numbers : for electrons , the @xmath16 ll will be full and the @xmath109 ll will be filled with a fraction @xmath110 ; for holes , the @xmath16 ll will be filled with the fraction @xmath111 .
we shall proceed with a minimalistic model which concentrates upon only one type of carriers at a zero ll ; the spin splitting is not taken into account .
we shall ignore the special low t states : the fractional qhe at special rational values of @xmath112 and the wigner crystallization ( its source , the long range coulomb repulsion , is strongly suppressed in presence of the cdw ) .
electronic interaction will be taken into account only via the self - consistent slowly varying coulomb potential @xmath113 .
this hartry approximation was used effectively @xcite in theory of inhomogeneous iqhe near gates boundaries in semiconductors .
we shall use the parameters as : chain length @xmath114 , pinning length @xmath115 ( corresponding to @xmath116 ) , temperature @xmath117 , magnetic field @xmath91 , number of particles @xmath118 per @xmath119 ( like for holes ) , filling factor of the ll as @xmath120 and @xmath121 .
we tested @xmath122 and @xmath123 corresponding to the threshold voltage and to the maximal one in experiments .
for static distributions , the analytic treatment is possible at t=0 which may be applicable at the lower experimental limit @xmath124 .
but for higher @xmath125 , we should use numerical solutions .
the @xmath65 solution confirms the full charge segregation which sets in when @xmath126 the critical value @xmath127 at @xmath128 is the energy of coulomb interaction between the over- and under charged areas .
plots in figs.[fig : x - plots ] show the numerically calculated distributions . .
the parameters are @xmath129 , @xmath130 ; left panel : @xmath131 , @xmath132 ; right panel : @xmath133 , @xmath134,title="fig:",scaledwidth=40.0%,height=188 ] .
the parameters are @xmath129 , @xmath130 ; left panel : @xmath131 , @xmath132 ; right panel : @xmath133 , @xmath134,title="fig:",scaledwidth=40.0%,height=151 ] @xmath135 cyan lines with fillings give distributions of occupation numbers - the fermi function @xmath136 . at @xmath137
the full depletion appears at the high - voltage end @xmath138 when @xmath139 is low ( cf . at @xmath65 the depletion @xmath140 would spread over @xmath141 ) ; the iqh limit @xmath142 is reached only near @xmath10 . at the elevated @xmath143 and smaller @xmath134 the population
is spread over the whole length , while it is strongly suppressed from @xmath144 at @xmath10 to @xmath145 at @xmath146 .
@xmath135 blue lines give distributions of @xmath147 .
they follow the rise of both @xmath148 and @xmath149 towards @xmath10 . but the current can rise also towards depleted edge @xmath146 where the high field can overcome the drop of the density .
@xmath135 black lines give the monotonic potential @xmath150 , with @xmath151 rising near both ends and suppressed , together with the current , in the middle .
@xmath135 the brown line gives the cdw stress which is also the electrochemical potential @xmath152 of the condensate .
the shape of @xmath152 is similar to that of @xmath113 but with a discrepancy @xmath153 which changes sign near the middle .
that means the two reservoirs of electrons are not stable with respect to exchange of particles among them , provided the phase slips are allowed @xcite
. then the cdw will give up its periods near the depletion region where @xmath154 and will absorb pairs near the overpopulated end where @xmath155 .
the discrepancy @xmath156 is largest at the left end as we see it clearly at the plot for @xmath120 , @xmath157 where @xmath158 reaches @xmath159 compatible with expectations for the phase slip voltage .
numerical results were obtained from the following equations .
the one for the phase is the balance of the stress gradient and the pinning force @xmath160 : @xmath161 the potential satisfies the poisson eq . with the total charge density per chain @xmath162 @xmath163 + q/\pi=0 \ , \
\label{pot } \\ \nonumber\ ] ] where @xmath164 , @xmath165 , @xmath166 . here
@xmath167 is the cdw wave number shift , @xmath168 is the background compensating charge density . with so small @xmath169 , eq .
( [ pot ] ) is reduced to the constraint of the local electroneutrality @xmath170 + q=0 \label{en}\ ] ] eqs .
( [ lin - pin],[en ] ) were solved numerically with boundary conditions here the first condition ensures the total electroneutrality while the second one fixes the hall voltage . we extend the above modeling in two respects .
we shall model the dependence @xmath171 in such a way that the linear law ( [ lin - pin ] ) sharply drops to zero at @xmath172 - the threshold phase displacement @xmath173 of depinning : @xmath174 this picture corresponds to a conventional first threshold field which looks to be always valid for nbse3 , while a more complicated two - step depinning is observed at low t in gapful cdws , see @xcite .
we shall look for the time dependent evolution starting from the initial moment when the current @xmath21 is turned on .
we shall not limit the cdw current , given by the phase velocity @xmath175 , to vanish at the boundary .
we shall suppose that the phase slips concentrated in narrow boundary layers @xcite will do this job .
in the future , it should be possible to incorporate that to the model .
now , the eq .
( [ lin - pin ] ) is generalized as @xmath176 eqs .
( [ mu - c ] ) , together with ( [ en ] ) where @xmath177 , were solved numerically with the paired boundary condition for the electric field and separate conditions for potentials .
@xmath178 , @xmath179 , @xmath180 where the first condition ensures the total electroneutrality while others turn on the boundary voltage values gradually , over a short time @xmath181 .
actually @xmath182 and @xmath183 are not fixed separately ; rather the given current determines the relation ( [ j , v0,v1 ] ) among them .
notice that for low @xmath14 when both @xmath184 it fixes only @xmath185 leaving @xmath186 to be evaluated .
finally , we exclude @xmath150 from eqs .
( [ en],[mu - c ] ) to arrive at the eq . for the phase alone : @xmath187 with @xmath188 .
remarkably , such a nonlinear nonlocal partial - differential equation yields reliably a complicated spacio - temporal behavior described below .
the results presented in fig.[fig : plots(x , t ) ] were obtained for @xmath189 , @xmath190 , @xmath191 , @xmath192 , @xmath193 .
also we put @xmath194 which fixes the time scale .
the left panel of fig.[fig : plots(x , t ) ] shows a plot of the electric field @xmath195 .
it demonstrates two domain walls of the propagating depinning which collide and annihilate giving rise to a stationary sliding depinned state with essential but smooth variations over the length .
the plot at the right panel shows that the phase velocity is not as sensitive to the passing walls but gives a strong , very narrow in time peak at their collision .
the interpretation is that electric fields ( created by electrons redistributions near the boundaries ) exceed the threshold field while the bulk is still below that .
the depinning penetrates into the pinned region by propagating sharp fronts of the electric field .
an overshoot will be noticed if we plot also the distribution @xmath196 : behind the fronts the occupation numbers are flat at 1 ( the transient iqhe state ) or zero ( the full depletion ) .
but later on the smooth profiles are established . by now
, we studied only the single run ( after switching on of the boundary potentials ) assuming a source and a drain for the cdw current by phase slips .
the implementation to the hall bar and particularly to observed oscillations is anticipated to be the following , recall sec.2 .
the cdw sliding is arrested at the boundaries by conversion of particles between the condensate and the normal states which job is made by phase slips .
the rate is governed by difference of chemical potentials of the two reservoirs ( notice the discrepancy between black and brown lines in plots of the fig.[fig : x - plots ] ) .
we expect that each impulse of the phase slip will reduce the conditions to the bare ones and the development similar to the above modeling will repeat itself sequentially .
it seems that emergence of depinning waves in our modeling may be of an significant insight to other , more conventional experimental situations on sliding cdws .
in the bulk of sliding cdws , the normal @xmath197 and the collective @xmath198 currents flow in the same direction while regulating their participation ratio depending on the geometry and the background @xcite . a curious question may be posed : if the partial currents can flow to opposite directions ? even more , if they can be compensated to a limit when the total current is zero ?
it happens to be possible indeed in new circumstances when the on - chain currents are driven by the hall voltage in a geometry with the closed circuit in a narrow @xmath4 direction of the hall bar , as described in this article and most probably already seen in the experiment @xcite .
actually , the old experiments @xcite on sliding under the temperature gradient alone should also be understood as the case of compensating currents .
the picture of counter - currents ( the collective and the normal ones ) closing the loop by annihilating at the hall bar boundaries looks rather fantastic and even suspicious as a _
perpetuum mobile_. still it is feasible , as well forgotten fundamentals of the conventional hall effect in semiconductors @xcite can tell us . in this old science
there is a silently supposed and well forgotten feature of processes with more than one kind of carriers : presence of partial hall currents for electrons and holes ; the currents are compensated so that only the total one is zero . there , the currents of electrons and holes should be compared with the normal and the collective currents of the cdw . in both cases
the equilibration among the carriers is hindered : by necessity of the energy relaxation across the gap in semiconductors or of making phase slips in the cdw .
the cdw has a drastic influence already in the stationary regime , before the onset of sliding .
recall the theory @xcite of inhomogeneous iqhe near gates boundaries in semiconductors . in those cases
the stripe structure of electronic density appeared in response to the built - in electrostatic field spread from the gate boundary which is present already at @xmath199 . in our case , the inhomogeneous distribution of concentration of electrons is caused internally by the hall voltage .
any spread density profile deviating from the background charge of donors and the gate would be prohibited in a 2d electron gas in semiconductors because of the very high cost of coulomb energy . in our case ,
the deformations of the cdw background allows to reduce the field by the gigantic dielectric constant @xmath76 .
the redistribution of charge caused by the hall voltage alone , up to segregation into the overcharged fully filed iqh segment and the undercharged fully depleted one , is unthinkable in usual circumstance of the qhe .
latyshev , et al , presentation at ecrys-2014 .
orlov , et al .
, jetp lett .
* 87 * ( 2008 ) 433 .
latyshev , et al , physica b , * 404 * ( 2009 ) 399 .
macdonald , et al , phys .
b * 28 * ( 1983 ) 3648 . d.b .
chklovskii , et al , phys .
b * 47 * ( 1993 ) 12593 . c. wexler and d. thouless phys . rev .
b * 49 * ( 1994 ) 4815 . i.m .
ruzin , phys .
b , * 47 * ( 1993 ) 15 727 . j. mccarten , et al , phys .
b * 46 * ( 1992 ) 4456 . g. gruner , et al , phys . rev .
, * 45 * ( 1980 ) 935 . k. seeger , solid state comm . , * 75 * ( 1990 ) 155
i. latyshev , a.p .
orlov , and p. monceau , jetp lett .
, * 93 * ( 2011 ) 101 .
coleman , et al , phys .
b * 41 * ( 1990 ) 480 . a.a .
sinchenko , et al , j. phys .
: cond . mat .
* 21 * ( 2009 ) 435601 .
h. requardt , et al , phys .
* 80 * ( 1998 ) 5631 .
s. brazovskii and n. kirova , j. physique iv , * 9 * ( 1999 ) 139 .
s. brazovskii and t. nattermann , advances in physics , * 53 * ( 2004 ) 177 , ch.13 .
gorkov , jetp lett .
* 38 * ( 1983 ) 87 .
beyermann , et al , phys .
* 56 * ( 1986 ) 1489 .
r. landauer and j. swanson , phys . rev .
* 91 * ( 1953 ) 555 . | we suggest a theory for a deformable and sliding charge density wave ( cdw ) in the hall bar geometry for the quantum limit when the carriers in remnant small pockets are concentrated at lowest landau levels ( ll ) forming a fractionally ( @xmath0 ) filled quantum hall state .
the gigantic polarizability of the cdw allows for a strong redistribution of electronic densities up to a complete charge segregation when all carriers occupy , with the maximum filling , a fraction @xmath1 of the chain length - thus forming the integer quantum hall state , while leaving the fraction @xmath2 of the chain length unoccupied .
the electric field in charged regions easily exceeds the pinning threshold of the cdw , then the depinning propagates into the nominally pinned central region via sharp domain walls .
resulting picture is that of compensated collective and normal pulsing counter - currents driven by the hall voltage .
this scenario is illustrated by numerical modeling for nonstationary distributions of the current and the electric field .
this picture can interpret experiments in mesa - junctions showing depinning by the hall voltage and the generation of voltage - controlled high frequency oscillations ( yu.i .
latyshev , p. monceau , a.a .
sinchenko , et al , presented at ecrys-2011 , unpublished ) .
_ after proceedings of ecrys 2014 _ * keywords : * cdw , qhe , shapiro steps , nbn , mesa junction , nbse3 |
cyg x-1 is a high - mass x - ray binary with the orbital period of @xmath3 d ( gies & bolton 1982 ) , containing a black hole ( bolton 1972 ; webster & murdin 1972 ) and the o9.7 iab supergiant hde 226868 primary ( walborn 1973 ) .
although the primary nearly fills the roche lobe , accretion is via a stellar wind ( gies & bolton 1986a ; gies et al .
observations of optical lines indicate that the wind departs from the spherical symmetry by being focused towards the compact object .
still , only a few percent of the wind is accreted .
the orbital period has been detected in all energy bands where the source was detected .
x - ray modulation in the hard spectral state has the form of broad intensity dips accompanied by spectral hardening centred on the superior conjunction of the x - ray source , i.e. , when it is behind the primary ( e.g. , priedhorsky , brandt & lund 1995 ; wen et al . 1999 ;
lachowicz et al . 2006 , hereafter l06 ) .
the dip is a result of bound - free absorption by the wind ( wen et al .
periodic changes in equivalent widths of uv absorption lines are due to the orbital motion of a highly ionized , x - ray irradiated , region of the stellar wind . in the superior conjunction ,
the equivalent widths achieve the maximal values ( treves et al .
1980 ) . in the visual band ,
the orbital period is seen in radial velocity variations of absorption and emission lines ( e.g. , bolton 1975 ) and in ellipsoidal variations ( walker 1972 ) .
the radio emission of cyg x-1 is most likely due to the synchrotron process in a jet , whose velocity has been claimed to be mildly relativistic , @xmath4 ( stirling et al .
2001 , hereafter s01 ; gleissner et al .
it has a flat ( in the energy flux , @xmath5 ) spectrum at @xmath620 mjy in the x - ray hard spectral state ( see , e.g. , zdziarski et al .
2002 for a review of the x - ray states of cyg x-1 ) , with no detection of either low or high frequency cut - off between 2.2 and 220 ghz ( fender et al .
2000 ) .
orbital periodicity of the radio emission was discovered by han ( 1993 ) .
however , more insight in it was possible thanks to long term radio monitoring of cyg x-1 by the ryle telescope at the mullard radio astronomy observatory at 15 ghz and the green bank interferometer at 2.25 and 8.3 ghz ( pooley , fender & brocksopp 1999 ; l06 ) .
it was found that the total modulation depth _ in the hard state _ increases with the frequency , from @xmath70.03 at 2.25 ghz to @xmath70.3 at 15 ghz , and that there is a time lag of the minimum of the modulation with respect to the superior conjunction decreasing with the frequency , from @xmath70.3 of the orbital period at 2.25 ghz to @xmath70.15 of it at 15 ghz . in the soft state , the orbital modulation takes place as well ( l06 ) , but constraints on it are much weaker , due to the much lower on average but flaring flux ( fender et al .
2006 ; l06 ) .
we would like to point out that an orbital periodicity of radio emission is a rare phenomenon in x - ray binaries .
it appears to be seen in only two other systems , lsi @xmath8 ( taylor & gregory 1982 ) and cir x-1 ( whelan et al .
1977 ) . both of them are of rather different nature than cyg x-1 .
although their compact objects orbit around stars undergoing heavy mass loss ( as in cyg x-1 ) , the orbits are wide and highly eccentric . during the periastron passage ,
the accretion rate onto the compact object is greatly enhanced and both the x - ray and radio fluxes increase ( e.g. , coe et al . 1983 ; marti & paredes 1995 ; haynes , lerche & murdin 1980 ) . on the other hand , this is certainly not the case in cyg x-1 , in which the orbit is circular ( brocksopp et al .
1999b ) , and the orbital separation is similar to the diameter of the primary .
thus , the nature of the orbital modulation of its radio emission has to be different . the process likely to be responsible for the modulation of the radio emission has been identified by brocksopp , fender & pooley ( 2002 , hereafter bfp02 ) .
they proposed it is free - free absorption in the stellar wind of the primary .
it is then analogous to the x - ray modulation , except for the absorption being free - free rather than bound - free .
unfortunately , although the work of bfp02 generally explains the origin of the radio modulation , it is plagued by a number of errors ( as noticed by l06 ) . due to those errors ,
their results are generally incorrect .
for example , errors in the equations and the numerical code used lead to the result that the fractional depth of the modulation _ decreases _ with the increasing inclination .
this result is rather counterintuitive , as the system is fully symmetric for a face - on observer and most asymmetric for an edge - on one , and it is indeed incorrect .
then , the absorption in the wind outside the cylinder defined by the orbit was underestimated by @xmath740 orders of magnitude .
furthermore , they assumed a uniform and rather low wind temperature , of 5000 k. in reality , the wind is compton - heated on the side of the x - ray source to temperatures about two orders of magnitude higher ( strongly reducing the free - free opacity with respect to the treatment of bfp02 ) . also , the jet model of bfp02 was relatively simplified , assuming that emission at a given frequency is emitted entirely at a single point along a linear jet , and it did not explain the phase lags . in the present work ,
we consider models of the jet including its radial emission profile as well as its bending .
we consider only the hard - state data . in modelling the stellar wind , we take into account its irradiation by the uv and x - ray photons , and propagation during the orbital motion of the primary .
we model both the observed depth of the modulation as a function of frequency and the corresponding lags of the minima of the folded radio lightcurves with respect to the spectroscopic superior conjunction .
this yields constraints on the jet velocity and emission profiles of the radio emission .
.the average hard - state modulation depth and the phase shift [ defined in equation ( [ moddepth ] ) and calculated using the fitted model fluxes , see fig .
[ data ] ] , from l06 .
the uncertainties are @xmath9 . [
cols="<,<,<,<",options="header " , ] [ values ]
we first present the main observational results on the orbital variability of the radio emission from cyg x-1 , in order for this work to be self - contained .
[ data ] shows the folded and averaged lightcurves from the ryle and green bank radio telescopes in the hard state ( l06 ) .
the 15 ghz lightcurve has the best statistical quality as well the highest modulation depth .
as found in l06 , its fitting requires three harmonics . on the other hand ,
it is sufficient to use a single sinusoidal dependence at 2.25 ghz and 8.3 ghz .
the fractional modulation depth is defined as in l06 , @xmath10 where @xmath11 and @xmath12 are the maximum and minimum fluxes in the averaged and folded lightcurves , observed at the phases of @xmath13 and @xmath14 , respectively , and the phase lags , given by @xmath14 . here , @xmath15 corresponds to the spectroscopic superior conjunction of the x - ray source , and @xmath16 is within the 01 interval .
table [ values ] gives the values of @xmath17 and @xmath14 of l06 .
they constrain the parameters of the binary and the jet .
, with a fast jet .
( note that the counter - jet has not yet been observed . )
the centre of mass is located within the primary .
the observer is in the @xmath18 plane .
, width=313 ]
a schematic picture of the binary including the jet is shown in fig.[system ] .
the centre of our coordinate frame is in the centre of mass , and the orbit is in the @xmath19 plane .
the observer is in the @xmath18 plane .
we note that if the the jet is precessing ( which may explain the observed superorbital modulation , see section [ super ] below ) , the jet axis will be inclined with respect to the binary plane . in our modelling of the orbital modulation
below , we neglect this effect since the precession amplitude is only poorly constrained ( e.g. , l06 ) and the superorbital modulation may in principle be alternatively explained by an intrinsic cycle of the supergiant wind .
the inclination , @xmath20 , remains relatively uncertain . on the basis of analysis of absorption lines , gies & bolton ( 1986a ) estimated @xmath21 . on the other hand , the polarimetric measurements of dolan & tapia ( 1989 ) yield @xmath22@xmath23 . in our study
, we consider the inclination range of 30@xmath2460@xmath24 .
we assume the black hole and the primary masses of @xmath25 and @xmath26 ( zikowski 2005 ) , respectively , and a circular orbit .
the radius of the primary is taken as @xmath27 cm ( gies et al .
then , the separation corresponds to @xmath28 cm , and the orbital velocity is @xmath29 cm s@xmath1 and @xmath30 cm s@xmath1 for the black hole and the primary , respectively
. we hereafter use @xmath31 as the unit , as it gives a measure of the asymmetry in the photon paths during the orbital motion .
the distance is assumed to be @xmath32 kpc ( zikowski 2005 ) .
k ( the dot - dashed curve ) and a hard state x - ray spectrum from the vicinity of the black hole .
the x - ray spectrum is from frontera et al .
the dots , long dashes and short dashes correspond to the soft x - ray excess , the main hard x - ray comptonization component , and the fe k@xmath33 line , respectively .
, width=313 ] we assume the wind to be smooth , spherically symmetric and radiatively driven .
we follow schaerer ( 1996 ) and santolaya - rey , puls & herrero ( 1997 ) in assuming a simple wind stratification at its base . the inner part , which we do not consider here , is the stellar atmosphere . the outer part
, i.e. , the proper wind , is parametrized by the velocity law , @xmath34 where @xmath35 is the terminal wind velocity , @xmath36 , \label{r0}\ ] ] and the lower boundary of the wind at @xmath37 is chosen to correspond to the isothermal sound speed , @xmath38 where @xmath39 is the boltzmann constant , @xmath40 is the hydrogen atom mass , @xmath41 is the mean molecular weight , and @xmath42 is the temperature of the atmosphere , respectively .
the particle number density at @xmath43 is related to the mass - loss rate , @xmath44 , and and @xmath45 via the continuity equation , @xmath46 note that equations ( [ velocity][n_r ] ) apply to a star at rest .
below , we apply those equations taking into account the orbital motion , with @xmath47 measuring the distance travelled by a wind element ( see below for details ) .
we adopt @xmath48 yr@xmath1 ( gies et al.2003 ) , @xmath49 cm s@xmath1 , @xmath50 ( gies & bolton 1986b ) , and @xmath51 where @xmath52 is the effective temperature taken as @xmath53 k ( zikowski 2005 ) .
this temperature at the assumed stellar radius yields @xmath54 erg s@xmath1 . for calculating @xmath41
, we assume the wind to be ionized with the abundances for an o9.7 iab star ( lamers & leitherer 1993 ) , which yields @xmath55 .
we allow the wind to extend out to the infinity , which assumption has a negligible effect on its absorption .
we note that , with the exception of a very dense wind region close to the stellar surface , the wind is optically thin for both x - ray and uv photons .
the resulting optically - thin approximation allows us to greatly simplify calculations of the wind structure . in this approximation ,
the radiation field at a distance @xmath56 away from a source of the luminosity , @xmath57 , is determined only by the geometrical dilution , @xmath58 , with no radiative transfer calculations being necessary . moreover , at each point in the cloud with the number density , @xmath59 , the state of the gas for a given spectral shape depends solely on the ionization parameter , @xmath60 ( tarter , tucker & salpeter 1969 ) .
thus , we can apply the results of a single model to a number of conditions with the same value of @xmath61 .
based on the results of kallman & mccray ( 1982 ) , we apply the optically - thin approximation for @xmath62 .
( solid curve ) , the stellar @xmath63 k blackbody ( dotted curve ) , and the hard - state x - ray spectrum ( dashed curve ) .
, width=313 ] in order to calculate the ionization and temperature structure of the wind , we use the xstar photoionization code ( bautista & kallman 2001 ) .
it calculates the state of the gas surrounding an ionizing radiation source .
the gas is assumed to have a constant density but using the optically thin scaling law allows us to apply the results to the optically - thin part of the wind .
the illuminating source is assumed to have the spectrum equal to the sum of the hard - state spectrum of cyg x-1 as modelled by frontera et al .
( 2001 ) and the primary blackbody spectrum of @xmath64 k , shown in fig.[spectrum ] .
our model spectrum is relatively similar to that of the model # 5 of kallman & mccray ( 1982 ) .
we stress that in spite of the primary having the luminosity about two orders of magnitude higher than that of the x - ray source , it is necessary to include both sources in the calculations .
in particular , the gas temperature reaches the compton temperature , @xmath65 , in the limit of full ionization , @xmath66 ( e.g. , begelman , mckee & shields 1983 ) , corresponding to the balance of the compton gains and losses only .
although the denominator is almost completely dominated by the blackbody emission , the dominant contribution to the numerator is from near the high - energy cutoff of the x - ray spectrum ( in general , it is the case for any spectrum harder than @xmath67 ) .
the compton temperature of the x - ray component alone is as high as @xmath68 k ( see fig . [ xitemp ] ) .
then , the addition of the stellar blackbody increases the compton cooling [ the denominator of equation ( [ t_c ] ) ] and reduces @xmath69 approximately by the ratio between the two luminosities , i.e. , by two orders of magnitude , to @xmath70 k. in the plane of the orbit ( the phase of @xmath15 corresponds to the observer on the left - hand side ) .
the corresponding values of @xmath71 are given by the solid curve in fig .
[ xitemp ] .
the length unit is the separation , @xmath31 .
the hatched area covers the primary and the wind shadowed by it from the influence of the ionizing source .
the small dot represents the position of the black hole .
the slight asymmetry of the contours with respect to the line joining the star centres results from their ( clockwise ) orbital motion and the finite wind velocity .
, width=313 ] the stellar blackbody emission has @xmath72 erg s@xmath1 , whereas the x - ray luminosity is only @xmath73 erg s@xmath1 ( see also the data compilation in zdziarski et al.2002 ) .
we note that when calculating the ionization parameter , @xmath61 [ equation ( [ xi ] ) ] , xstar uses only the luminosity in the 13.6 ev13.6 kev range , equal to @xmath74 erg s@xmath1 in our case . however , the full spectrum is used in calculations of the temperature structure and the distribution of different ions .
the dependencies of @xmath75 for the entire spectrum of fig . [ spectrum ] and its two components
separately are shown in fig .
[ xitemp ] .
we assume that the entire radiation source is located at the position of the black hole ( and thus @xmath56 is the distance from the black - hole center ) .
this approximation allows us to use xstar with a single input spectrum of fig .
[ spectrum ] and for a range of the gas density .
although it introduces relatively large errors close to the binary plane , it reproduces well the actual temperature of the gas away from the plane , which region is important for calculating absorption of the jet emission . note that the gas density in the the inertial frame changes with the binary motion . in this frame
, the wind density at a given point has to be calculated taking into account both the past motion of the primary and the motion of the wind .
however , we neglect the effects of the orbital motion and rotation on the initial wind velocity .
we also neglect the dynamics of the wind , e.g. , effects of gravity and radiation .
we then calculate the wind ionization structure , and the resulting temperature distribution . the distribution of @xmath76 in the orbital plane is shown in fig .
[ ionmap ] . because of a high density close to the surface of the primary , the optically - thin approximation breaks down there .
however , this region does not contribute to absorption of the radio emission , which originates at relatively large distances from the primary . though the region shadowed by the primary does not contribute to that absorption
, we also calculate its temperature , based the dotted curve in fig .
[ xitemp ] . in the vicinity of the black hole , @xmath77 , the wind is completely ionized and its temperature equals @xmath65 .
note that since the wind reaches its terminal velocity far away from the orbital plane , @xmath78 .
there , @xmath61 reaches a constant , @xmath79 in our case , which corresponds to @xmath80 k. above the orbital plane .
the phase of @xmath81 corresponds to the observer on the left - hand side .
the image of the primary lags by @xmath82 with respect to its actual position , shown by the short - dash circle , due to the stellar motion and the finite wind velocity .
the solid and long - dash circles give the orbits of the black hole and the primary , respectively .
the arrow shows the direction of the orbital motion .
the wind shadowed from the x - ray irradiation by the primary is seen on the left.,width=321 ] the free - free absorption coefficient ( in cgs units ) is ( rybicki & lightman 1979 ) , @xmath83 where @xmath84 and @xmath85 is the number density of electrons and ions , respectively .
the velocity averaged gaunt factor , @xmath86 , is approximated as @xmath87 ( leitherer et al .
1995 ) . for simplicity , we hereafter use @xmath88 . combining equations
( [ velocity][r0 ] ) , ( [ n_r ] ) and ( [ alpha][gaunt ] ) , and using the numerical values assumed above , we obtain , @xmath89 in equations ( [ alpha]-[alpha2 ] ) , @xmath71 is in k , @xmath43 is in cm and @xmath90 in hz .
we integrate the absorption coefficient along the line of sight from the place of radio emission at each phase to the observer , which gives us the optical depth and , in turn , the observed flux , @xmath91 . as a result of the stellar motion and the finite wind velocity , the image of the primary reflected in the wind density , or the absorption coefficient ,
is phase - lagged with respect to the actual position of the primary at that time .
we illustrate this effect in fig .
[ absorptionmap ] .
it shows the absorption coefficient at the height of @xmath92 above the orbital plane .
we see a shift of the stellar image with respect to the actual primary position . the effect is visible here because the velocity of the wind is larger than the orbital velocity by only one order of magnitude , with the wind travelling the distance of @xmath93 away from the stellar surface during one orbital period .
this phase lag of the primary increases with the height above the orbital plane , and will by itself cause a phase shift in the orbital modulation of the radio emission with respect to the actual ephemeris . we take it into account in our calculations ; however , we find it can not by itself explain the observed phase lags .
we also see the shadow wind in fig .
[ absorptionmap ] . because of its relatively low temperature , it has higher free - free optical depth than the rest of the wind .
we assume that the radio emission originates in the jet .
we neglect its transverse size and consider only its longitudinal structure .
we assume that the geometry of the jet is directly defined by the initial condition at its base , namely the initial vertical velocity , @xmath94 , the acceleration , @xmath95 , and the perpendicular instantaneous orbital velocity , @xmath96 , where @xmath97 is the orbital radius of the black hole .
( here , we do not take into account a possible jet precession and thus @xmath94 and @xmath95 have only vertical components . )
we assume a continuous jet moving away from the place of the ejection .
the trajectory of a jet element has the parametric form , @xmath98 @xmath99 @xmath100 where @xmath101 is the time elapsed from the ejection .
if the jet is fast , it remains nearly perpendicular to the orbital plane , as shown schematically in fig.[system ] . for a slower jet ,
the relative contribution from the orbital velocity is important , and it bends , i.e. , its position at a given radius lags behind the black hole .
the projection of the motion of a single blob in the jet onto the orbital plane forms a spiral .
[ curvedjet ] shows two examples for @xmath102 and two different values of @xmath95 . and a low and a high acceleration , ( a ) @xmath103 cm s@xmath104 , ( b ) @xmath105 cm s@xmath104 , respectively .
the black hole orbit is shown in the orbital plane , and three circles above it show the jet orbit at the height of @xmath106 10 , 20 and @xmath107.,width=321 ] we then consider the jet emissivity profile .
the simplest one corresponds to discrete emission , where each point of the jet emits a ( single ) different frequency ( which approximation was used by bfp02 ) . as a more realistic parametrization of continuous jet emission
, we use a double - power - law profile , @xmath108 where @xmath109 is the radio flux emitted at @xmath110 and @xmath111 , @xmath112 , @xmath113 ( @xmath114 ) and @xmath115 ( @xmath116 ) are @xmath90-dependent parameters . the variable @xmath110 gives the distance along the jet from its base .
hence , only in the case of the fast , straight , jet , @xmath117 .
if we consider only relative emission profiles , without specifying the absolute value of the emitted flux , we can set @xmath111 to the value normalizing the profile to unity at the maximum .
the maximum flux is emitted at @xmath118 and @xmath119 and @xmath120 far above and far below the maximum , respectively .
the profile normalized to unity is given by @xmath121
and 0.5 . the optical paths at these two phases differ from each other both inside and outside the cylinder defined by the black hole orbit .
, width=321 ] orbital modulation is produced when the source of the radio emission travels through the wind and thus it is observed through different optical depths , as illustrated in fig .
[ lineofsight ] .
the depth of the modulation , its phase shift and the average transmitted flux depend on the spatial profile of the radio emission , the jet velocity profile , and the inclination , which dependencies we investigate . in general , as it is apparent from fig .
[ lineofsight ] , the difference between the lines of sight is not confined to the space inside the cylinder defined by the black hole orbit . because of a strong dependence of the absorption coefficient on the distance from the primary , @xmath122 , a small difference in the line of sight can strongly affect the observed modulation . it is therefore not correct to neglect the differences in the absorption outside the orbital cylinder ( as it was done in bfp02 ) . we first consider a simple model with a fast jet , perpendicular to the orbital plane .
we compare the observed emission from points along the jet at @xmath15 and 0.5 .
we show the results in fig .
[ absorptionprofile ] .
we see that the strength of the modulation is close to unity near the base of the jet , and then decreases fast with the height . however , the intrinsic jet emission is very strongly attenuated in the regions where the jet modulation is close to unity . excluding these region ,
the fast decrease of the modulation strength with the height implies that the spatial emissivity profile at a given frequency has to be narrow in order to yield a substantial modulation .
these considerations point to the relative narrowness of the parameter space where strong radio modulation by wind absorption can be achieved , and may explain why cyg x-1 is the only binary showing radio modulation due to that process .
we then stress the importance of properly accounting for the system geometry , see fig .
[ lightcurves ] .
we plot here three normalized lightcurves for models with three different system geometries .
first ( dots ) , the position of the primary is fixed at the centre with the black hole revolving around it on a circular orbit , and the absorption is confined to the cylinder defined by the black hole orbit , as assumed by bfp02 .
second ( dashes ) , we keep the former assumption but take into account the entire absorption .
third ( the solid curve ) , we consider the actual case with the stars orbiting around the centre of mass . in all cases ,
we assume the emission from a single point of the straight jet .
we see that the shape of the dotted curve disagrees with the observed lightcurve , in particular , there is no absorption at @xmath1230.75 .
then , accounting for the orbital motion of the both stars ( solid curve ) results in shallower modulation as compared to the case with the fixed companion .
the phase shift of the minimum of the solid curve is due to the apparent displacement of the companion , see section [ wind ] and fig.[absorptionmap ] . .
the transmitted fluxes are shown for both @xmath81 ( lower curve in pair ) and @xmath124 .
each curve is marked with the corresponding frequency in ghz .
the curves take into account the binary motion .
, width=321 ] and @xmath125 .
the dotted curve assumes the fixed primary and the wind absorption model as in bfp02 .
the dashed curve is for the fixed companion but properly accounting for the absorption .
the solid curve is for our model , with the stars orbiting around the centre of mass .
the visible phase lag in this curve is due to the phase lag of the image of the primary ( see fig .
[ absorptionmap ] ) , and it is @xmath126 ( corresponding to @xmath127 ) . , width=321 ] @xmath128 space allowed by the observed modulation depths ( table [ values ] ) , assuming discrete emission of a vertical jet .
the black dots give the values assumed in fig .
[ estimates ] below .
, width=321 ] the difference between the minimum and maximum optical depth , and thus the modulation depth for emission from a single point , depends on the height above the orbital plane , @xmath128 , and the inclination , @xmath20 .
we show this relationship in fig .
[ modulationmap ] for the observed modulation depths ( table [ values ] ) .
for example , @xmath129 at 15 ghz at @xmath125 corresponds to point emitting at @xmath130 .
the allowed emission regions at different frequencies are distinctly different , with lower frequencies emitted higher . for each @xmath90 , the region below the respective shaded area
is occupied by points whose transmitted flux has higher modulation depth with respect to observed ones .
emission from points above the shaded area is modulated with lower amplitude . with 1@xmath131 error bars , fitted with the @xmath132 theoretical relationship ( solid curve ) and the best fit for @xmath133 ( @xmath134 ; dashed curve ) .
, width=321 ] and @xmath90 based on equation ( [ fit2 ] ) , for four values of @xmath20 .
the heavy and light curves correspond to the modulation depth and the fractional transmitted flux , respectively.,width=321 ] we then compare our estimated emission distributions with the @xmath135 relationship obtained by blandford & knigl ( 1979 ) .
this relation follows from the position of the ( photospheric ) height where the jet becomes optically thin for synchrotron self - absorption .
the three points shown in fig.[modulationmap ] have been fitted ( including the errors on @xmath128 estimated from fig .
[ modulationmap ] ) with @xmath136 , where @xmath137 , and @xmath138 is the jet power in relativistic electrons and magnetic fields ( blandford & knigl 1979 ) . the best - fit relation , @xmath139 shown in fig .
[ estimates ] , does not agree with our model ( based on the observed modulation depths ) for 2.25 and 8.3 ghz within the 1@xmath131 error ranges .
however , it does agree with the data within their 90 per cent confidence intervals .
we also plot the best - fit curve for a general power law dependence , for which we find , @xmath140 thus , the best - fit power - law index is only slightly lower than that given by blandford & knigl ( 1979 ) .
the relationships at other inclinations than @xmath141 are slightly different , but generally the heights of the 2.25 ghz and 8.3 ghz emission is slightly less than that predicted by equation ( [ fit ] ) .
we then obtained the positions and modulation depths for other frequencies extrapolating the observed modulation depths using equation ( [ fit2 ] ) , see fig . [ inclination ] .
the curvature of the modulation depth at the low - frequency end results from changes of the apparent orbital separation related to the displacement of the primary image due to the motion of both stars and the finite wind velocity ( see fig .
[ absorptionmap ] ) . at @xmath142 ghz , @xmath143 , i.e.
, the emission at the minimum is then very strongly absorbed .
the phase - averaged fractional transmitted flux above 40 ghz is @xmath144 .
we also see that an increase of @xmath20 increases the modulation depth .
ghz , @xmath125 , and @xmath145 ( dashed curves ) and @xmath15 ( dotted curves ) .
the red , blue , green and magenta curves correspond to ( @xmath146 , @xmath113 , @xmath115 ) equal to @xmath147 , @xmath148 , @xmath149 , @xmath150 , respectively.,width=321 ] as an approximation to realistic jet emission , we consider double - power - law profiles , equation ( [ double ] ) .
this form of the spatial distribution can closely reproduce detailed jet models , e.g. , those of hjellming & johnston ( 1988 ) and kaiser ( 2006 ) . in fig .
[ profiles ] we show examples of that profile for a number of parameters together with the same profiles transmitted through the wind at the phases of @xmath81 and @xmath124 .
the difference between the transmitted fluxes integrated over the jet length at those two phases gives ( approximately , as the true minima have some phase shifts ) the modulation depth .
we can see that the higher the point of emission along the jet , the lower the difference between the two transmitted profiles , and hence the lower modulation .
the low-@xmath128 ( @xmath151 ) branch of the profiles is almost completely absorbed and thus the shape of the transmitted profiles and the modulation depth is only weakly dependent on @xmath113 .
however , the fractional transmitted flux ( i.e , with respect to the unobserved intrinsic emission ) does significantly depend on @xmath113 .
we have then created contour maps in the location width plane .
the former is represented by @xmath112 , determining the profile peak [ equation ( [ zmax ] ) ] , and the latter , by @xmath152 , which is approximately [ see equation ( [ double_norm ] ) ] , inversely proportional to its width .
fig.[brokeninf1 ] shows two maps corresponding to profiles with significant ( @xmath153 ) and small ( @xmath154 ) contributions to the intrinsic emission from below the profile maximum .
the hatched areas represent the profiles with the observed modulation depth ( table [ values ] ) at different @xmath90 and @xmath20 . as it can be inferred from fig .
[ brokeninf1 ] , broader profiles correspond to lower @xmath128 and higher @xmath20 .
also profiles emitted at lower @xmath128 have lower fractional transmitted fluxes . on the other hand
, narrow profiles have to be emitted at higher @xmath128 , and have the value of @xmath112 approximately equal to the corresponding location of point emission profiles shown in fig .
[ modulationmap ] . the areas of the parameter space allowed by the observations are smaller for @xmath155 because these profiles are narrower , and the corresponding modulation depth quickly decreases with the increasing height of the jet .
these profiles are also shifted towards lower values of @xmath112 than the @xmath156 profiles .
these effects are more prominent at lower frequencies .
the phase shifts we are able to obtain for the double - power - law profile along a straight jet are within about @xmath157 ( due to the effects of the finite wind velocity coupled with the orbital motion ) .
they are not sufficient to explain the observed phase lags ( table [ values ] ) . , vs.@xmath152 of the double - power - law emissivity profile allowed by the observed modulation depths ( table [ values ] ) , for different values of @xmath90 and @xmath20 and for ( a ) @xmath156 and ( b ) @xmath158 .
, width=321 ] here , we consider the jet bending due to its initial ( binary ) motion together with the black hole , see section [ jet ] . as it can be seen from fig.[curvedjet ] , a curved jet bends outside the cylinder defined by the black hole motion .
thus , the amplitude of the optical depth variations is higher than in the case of the vertical jet , which will then increase the strength of the modulation .
also , the resulting lag behind the compact object will lead to a lag in the phase of the maximum absorption ( i.e. , the minimum flux ) , as observed .
in general , the flux phase profile depends on the inclination , @xmath20 , the profile parameters ( @xmath159 , @xmath160 , @xmath161 ) , and on the velocity profile , given by @xmath94 and @xmath95 .
then , in order to reduce the parameter space , we consider here only two velocity profiles , one with @xmath162 and @xmath163 ( corresponding to acceleration only at the jet base ) , and one with @xmath164 and a constant @xmath165 .
in addition to the constraints from @xmath17 , we now also take into account the constraints from the phase lag of the minimum flux , @xmath166 ( table [ values ] ) .
first , we find that _ no _ jet with @xmath162 and @xmath163 can satisfy the observational constraints . at most
, the observational constraints can be satisfied at only one of the observed frequencies .
then , we consider the case with @xmath164 and constant @xmath165 , which we find can easily satisfy those constraints .
[ brokenmap ] shows the ranges of the parameters allowed by the data for a selection of values of @xmath95 .
we show the results for @xmath156 and 10 , analogous to fig .
[ brokeninf1 ] for the case of the straight jet .
the coloured regions are those where both of the constraints are satisfied for a given frequency .
such regions exist for each of the observed frequencies and given an appropriate choice of @xmath95 [ cm s@xmath104 ] in the range of @xmath168@xmath169 for @xmath170 , @xmath168@xmath171 for @xmath125 , @xmath172@xmath173 for @xmath174 and @xmath172@xmath175 for @xmath176 .
the values of @xmath95 used in fig .
[ brokenmap ] correspond approximately to the middle of these ranges .
it is , of course , possible that the jet acceleration is not constant , but as this is not required by the data , we do not consider such models .
we also show the corresponding fractional transmitted fluxes . they are allowed to be rather large , typically @xmath1770.5 for broad profiles ( @xmath156 ) and @xmath178 for narrow profiles ( @xmath158 ) .
the acceleration is relatively slow and the jet will be curved , with the curvature somewhere in between those shown in fig .
[ curvedjet](a ) and ( b ) . at the distance of @xmath179 ( @xmath180 cm ) , around which the 2.25 ghz radiation is emitted in our model ,
a jet with @xmath181 cm s@xmath104 will reach @xmath182 cm s@xmath1 during @xmath183 s. at this point , it will lag behind the black hole by @xmath184 and it will be at the distance from the orbital axis of 4.5 times the radius of the black - hole orbit .
there are a few essential features of our model , necessary to account for the observed orbital modulation depths and the phase lags .
one is the wind heating due to the irradiation by the x - rays .
this naturally creates a strong temperature anisotropy , and consequently strong anisotropy of the free - free absorption coefficient .
though the x - ray luminosity is @xmath185 times lower than the uv luminosity of the primary , compton heating is proportional to the _ product _ of the irradiating energy flux and the photon energy .
the latter is @xmath186 times higher for the x - rays than for the uv , and consequently , the x - ray heating is @xmath185 times higher than the uv heating .
we also point out that the range of the temperatures of the wind obtained by us , @xmath187@xmath188 k , is in full agreement with that found from _
chandra _ soft x - ray observations of cyg x-1 by miller et al.(2005 ) . for test purposes
, we have also run models with a uniform wind temperature equal @xmath189 k. they required strong fine - tuning in order to achieve the observed modulation depths , as well as very low resulting transmitted fraction of the intrinsic jet flux , @xmath190 .
inclusion of x - ray heating resolves those problems .
another essential feature is the required low velocity of the bulk of the mass of the jet , necessary for the jet bending required to account for the observed phase lags of the flux minima ( with respect to the superior conjunction of the black hole ) .
the velocity is nonrelativistic within the jet region constrained by the data .
the jet bending results then from the initial jet velocity in the binary plane ( due to the orbital motion of the black hole ) being comparable to its vertical velocity component .
we have searched for other ways to account for the observed phase lags , and have found none . in particular , the lags associated with the finite wind velocity together with the motion of the primary star can not account for the observations .
we note that the modulation depth increases with @xmath90 in the 2.2515 ghz range .
our model reproduces this increase .
not unexpectedly , it predicts the depth of the orbital modulation to increase further at @xmath191 ghz .
it would be of high interest to test this prediction observationally .
we note that there have been rather few observations at @xmath191 ghz so far ( fender et al .
2000 ; note some errors in the observation dates given there ) , insufficient for that test .
a related issue is the shape of the _ intrinsic _ jet spectrum .
the observed average , absorbed , spectrum , is approximately given by @xmath192 with @xmath193 in the @xmath194200 ghz range ( fender et al .
the intrinsic spectrum of the jet is also likely to be a power law , in general with a value of @xmath33 allowed to be different than 0 . however , if the intrinsic index is substantially different from 0 , then obtaining the @xmath193 power law would require a specific power - law absorption law .
we can obtain it with our model .
however , given the relatively large number of the free parameters compared to the relatively few observational constraints , the resulting model constraints would be rather loose . on the other hand , the most natural way to obtain the observed
@xmath193 is with the transmission fraction through the wind being high , close to 1 . as we show in fig .
[ brokenmap ] , there is a relatively large parameter space with that fraction being high , @xmath195 , and thus this is indeed possible .
the radio maps obtained from observations at 8.4 ghz ( s01 ) show the jet to form an extended feature .
the flux reaches the maximum at the core of an angular size of @xmath196 mas , and then it decreases over @xmath197 mas away from it . for @xmath20 from @xmath198 to @xmath199 , 3 and 15
mas correspond to ( 29@xmath200@xmath201 cm and ( 140@xmath202@xmath203 cm , respectively .
our constraints are for the distances along the jet of @xmath204 at 2.25 ghz and @xmath205 at 8.3 ghz .
thus , they apply only to the core of the jet .
still , we can compare the predictions of our model with the radio maps of s01 at larger distances using the double - power - law profiles .
we find that the observed extended structure of the jet is substantially more luminous than the predictions of our simple models .
the propagation of the jet beyond the core may be affected by various effects , e.g. , its moving away from the stellar wind region , and this issue is beyond the scope of the present work .
we note that the 2.2515 ghz monitoring data used by us are integrated over the entire observed source .
the emission at a given frequency from outside of the jet core is very unlikely to be modulated by the stellar wind .
thus , the fractional modulation of the emission close to the jet base has to be actually somewhat higher than that used by us , in order to account for the constant part emitted outside the core . according to the results of s01 ,
the 8.4 ghz flux from the extended part is less than or similar to that of the core .
then , the fractional modulation of the core emission should be higher by a factor @xmath206 than that averaged over the entire structure , but the values of the phase lags are not affected .
it is possible to reproduce the core emission profiles with such a modulation depth in our model .
the required profiles are located to the left of the 8.3 ghz region in fig .
[ brokenmap ] , which introduces only a small change in the value of the acceleration , reducing it only by at most 10 per cent .
then , the results of s01 show that the 15.4 ghz emission is mostly in the core of the extend of @xmath206 mas , with only @xmath207 per cent of the emission outside the core .
this is approximately compatible with our spatial emission profiles .
then , no radio maps are available at @xmath194 ghz .
our models yield the jet acceleration of @xmath208@xmath209 cm s@xmath104 . at @xmath179 ,
the jet achieves the velocity of @xmath210@xmath211 cm s@xmath1 , and @xmath212 times that at @xmath213 , i.e. , @xmath214 in both cases . on the other hand , s01 has estimated the jet velocity to be mildly relativistic based on the lack of the observed counter jet being explained by the relativistic doppler beaming
. the results of s01 combined with the constraints of gleissner et al .
( 2004 ) ( based on the lack of radio x - ray correlations on m
h time scales ) yield the jet velocity of @xmath215@xmath216 .
this requires an acceleration @xmath217 times higher than that estimated by us .
we note , however , that our results do not constrain at all the jet region at @xmath218 .
thus , the acceleration there can be much higher , e.g. , due to the stellar wind density being much lower than in the inner jet region .
another argument for the relativistic bulk velocity of the jet electrons emitting synchrotron radiation is the presence of the superorbital ( i.e. , on a time scale much longer than the orbital period ) modulation .
the modulation , with the period of @xmath219 d , is present in both radio and x - rays ( e.g. , brocksopp et al . 1999a ; l06 ) .
the standard ( though not proven ) explanation of that effect in cyg x-1 ( and a number of other x - ray binaries ) is precession ( e.g. , larwood 1998 ; l06 ) .
then , the jet precession can cause the flux variability at this period if the jet is at least mildly relativistic .
the radio data on the superorbital modulation are compatible with this picture as the modulation depth at the three monitored frequencies , 2.25 , 8.3 and 15 ghz , is compatible with being constant , and there are no significant phase lags .
we note , however , that those data do not require the jet velocity to be relativistic in the core , as the superorbital modulation can be caused by electrons outside it .
also , the required velocity can be as low as @xmath220 if the ( unknown ) precession amplitude is @xmath221 . on the other hand , we can provide here an explanation for the superorbital radio modulation without invoking a relativistic jet velocity .
namely , a precessing jet moves with respect to the wind structure ( which is defined only by the binary orbit and does not precess ) .
the line of sight from a given point in the jet , and thus the amount of the orbit - averaged free - free absorption in the wind , will then change with the superorbital phase ( without phase lags ) .
however , the modulation depth is frequency - dependent in this model . in principle , it is possible to find parameters of the system yielding the same modulation depth from 2 to 15 ghz , but it requires certain fine - tuning .
as we have discussed above , the observed phase lags of the orbital radio modulation require that the bulk of the jet mass is nonrelativistic within the region where most of the radio emission takes place . on the other hand , the non - detection of the counter jet and the superorbital radio modulation point to ( though not completely require ) relativistic bulk motion of most of the synchrotron - emitting electrons .
this apparent contradiction may be reconciled provided that the jet contains two flows at different velocities , e.g. , an outer slow part and a fast inner part . such a model for radio - emitting x - ray binaries
has been recently proposed by ferreira et al .
( 2006 ) . it is based on an analogous , two - flow , model for extragalactic radio jets of sol , pelletier & asso ( 1989 ) . the main contribution to the total jet power
is an electron - proton plasma that is either nonrelativistic in x - ray binaries or mildly relativistic in agn jets .
this plasma is generated by an mhd process in a large - scale magnetic field anchored in the accretion disc .
the second flow is composed of relativistic e@xmath222 pairs , and it carries only a small fraction of the total power .
it is produced in the inner regions of the mhd flow , where the pairs are created and then accelerated .
the outer flow confines the inner one . in the agn case ,
the outer slow flow is identified with kpc - scale jets , fuelling the radio lobes and hotspots , and the fast inner flow is identified with pc - scale jets and the superluminal motion ( sol et al .
we note that the specific model of ferreira et al .
( 2006 ) postulates that the threshold for pair production , forming the inner fast jet , is reached only above the accretion rates corresponding to the hard spectral state .
a massive disk outflow ( linked with the super - eddington mass transfer rate ) has been proposed to exist in the x - ray binary ss 433 ( zitter , calvani & dodoricio 1991 ; fabrika 1993 , 2004 ) .
the outflow , with the velocity of @xmath223 cm s@xmath1 , is radiatively driven from outer parts of the accretion disc .
its direction is determined by the tilt of the disc and is subject to both precessional and nutational motions of the outer disc .
begelman , king & pringle ( 2006 ) then postulated that the outflow is massive enough to deflect the relativistic jet launched from the vicinity of the accreting object and to align it with the outflow axis .
in the case of cyg x-1 , we may identify the slow flow with the one needed to explain the radio phase lags , and the fast one , with that at a relativistic motion explaining the invisibility of the counter jet and the superorbital modulation .
however , this identification requires modifications of the above models to allow for the existence of the two flows already at lower accretion rates , corresponding to the hard spectral state .
we further note that our postulated slow ( and dark ) flow is very likely to be identical to that found by gallo et al .
( 2005 ) to be required to power the radio lobes of cyg x-1 .
the kinetic power of that flow was found by gallo et al .
( 2005 ) to be comparable to the x - ray luminosity . on the other hand , heinz ( 2006 ) found that the kinetic power carried by fields and radio - emitting electrons in the visible jet of cyg x-1 was much smaller .
this is in agreement with the picture in which the visible ( fast ) flow is confined by the dark ( slow ) one .
also , heinz ( 2006 ) estimated the velocity of the dark flow powering the radio lobes of cyg x-1 as @xmath224 cm s@xmath1 .
this is in agreement with the velocity constrained by the radio modulation , see section [ general ] above .
cyg x-3 is an example of another high - mass x - ray binary with the companion emitting strong wind and with very strong orbital modulation of x - rays , thought to be caused by absorption and scattering in the wind .
it also has strong radio emission , most likely originating from a jet of that system .
still , no orbital modulation of the radio emission is seen ( hjalmarsdotter et al .
2004 ) . to explain this lack of modulation
, we first point out that cyg x-3 is a much more compact system than cyg x-1 . the orbital period of the former is only 4.8 hr . cherepashchuk & moffat ( 1994 ) have estimated its orbital separation as @xmath225@xmath226 cm , an order of magnitude less than that of cyg x-1 .
the degree of asymmetry of the wind at a given height , @xmath128 , is proportional to @xmath227 .
thus , if the jet of cyg x-3 emits a given radio frequency at the same height as cyg x-1 , its orbital modulation will be still much lower .
furthermore , the companion star in cyg x-3 is a wolf - rayet star , which mass loss rate , estimated as @xmath228@xmath229 yr@xmath1 ( ogley , bell burnell & fender 2001 ) , is much higher than that of the ob star in cyg x-1 .
thus , the radio emission originating relatively close to the jet base is likely to be completely free - free absorbed in cyg x-3 .
the high mass - loss rate coupled with the small separation also makes the wind optically thick to x - rays , not allowing us to use our optically - thin model ( section [ wind ] ) . the height from which a radio frequency is predominantly observed is then roughly at @xmath230 ( blandford & knigl 1979 ) .
the power of the jet is , in turn , @xmath231 , with @xmath2321 ( e.g. , migliari & fender 2006 ) . the x - ray luminosity of cyg x-3 is @xmath233100 ) times that of cyg x-1 ( e.g. , gallo , fender & pooley 2003 ; vilhu et al .
this consideration also points out to the height of radio emission in cyg x-3 being significantly higher than that in cyg x-1 .
consequently , the wind at the radio - emitting region in cyg x-3 is likely to be almost completely axially symmetric , resulting in no orbital modulation in the radio range .
our main conclusions are as follows .
we find the observational data on the orbital modulation of the radio emission from cyg x-1 fully consistent with the emission originating in a jet .
as postulated before , the modulation is caused by free - free absorption in the wind from the companion , with the line of sight changing during the orbital motion .
we find a crucial role of x - ray irradiation of the wind in enhancing its axial asymmetry with respect to the jet axis .
the irradiation strongly increases the wind temperature , up to the compton temperature of @xmath70 k , in the vicinity of the x - ray source , which then strongly reduces the free - free absorption on the side of the black hole .
it also increases the overall wind temperature to @xmath234 k ( except for the vicinity of the companion ) , which is about one order of magnitude above the star temperature .
this reduces the constant component of the free - free absorption and allows the transmitted fraction of the radio emission to be close to unity .
consequently , the intrinsic jet emission is likely to be close to the observed flat radio spectrum ( fender et al .
the wind temperatures calculated by us agree with those inferred from the _ chandra _ data ( miller et al .
2005 ) .
another crucial requirement following from our study is the velocity of the bulk of the mass of the jet being nonrelativistic , @xmath235 cm s@xmath1 , in the jet core ( within @xmath236 cm ) .
it follows from the relatively large phase lags of the absorption maxima with respect to the spectroscopic zero phase .
the only explanation of these lags appears to be a kinematic bending of the jet caused by the initial jet velocity in the binary plane ( equal to the orbital velocity of the black hole ) being comparable ( though still substantially lower ) to the jet vertical velocity in the region where most of the 215 ghz emission is produced . also , the data imply that the above velocity is reached following an approximately constant acceleration within the core , while a rapid , instantaneous - like acceleration is ruled out
. however , the jet may then become mildly relativistic outside of the core , as implied by the lack of a detectable counter jet ( s01 ) .
also , the synchrotron - emitting electrons may have a relativistic bulk motion already in the core if the jet is composed of two components , a slow , heavy , dark outer flow , and a relativistic , light , and bright inner flow ( as proposed by several authors in order to explain other observational features of jets ) .
we also consider the observed superorbital modulation of the radio emission , with the period of @xmath2 d. that modulation can be caused by the doppler effect in a relativistic precessing jet .
however , we find we can also explain that modulation by the free - free absorption varying due to the jet precession .
when the precessing jet is inclined towards the observer , the average absorption of its emission is weaker .
we construct specific models constraining the distribution of the radio emission at a given frequency along the jet .
the lower the frequency , the higher up it is emitted , and thus it is less modulated by the absorption , as observed . within the 90 per cent confidence intervals , our results are compatible with the @xmath132 law of blandford & knigl ( 1979 ) .
our results are also compatible with the plausible values of the inclination of cyg x-1 , @xmath221@xmath198 . for a given model , the higher the inclination , the deeper the modulation .
a prediction from our models is that the modulation depth should increase with frequency also at @xmath237 ghz .
thus , future monitoring at those frequencies would be of great importance for our understanding of the jet structure .
we also consider the issue why there is no observed radio modulation in cyg x-3 , which system is also powered by wind accretion and which shows strong orbital modulation of its x - ray emission .
we find a major reason for the lack of radio modulation is its orbital separation being an order of magnitude lower than in cyg x-1 , thus strongly reducing the wind asymmetry seen from the jet . also , it has much higher both the wind rate and the luminosity , which both factors imply radio emission take place at higher heights than in cyg x-1 ( further reducing any asymmetry of the wind seen from the jet ) . finally , we point out that our treatment of the jet and the wind has still neglected their dynamical interactions .
we also treat jet emission only in the core ( where it is modulated ) , and do not model its observed extended component .
these issues can be addressed in future work .
we thank m. sikora for valuable discussions , r. goosmann , j. poutanen and the referee for useful comments , c. brocksopp for discussion of her results and the numerical code used in bfp02 , and p. lachowicz for providing us with the data shown in fig . 4 of l06 .
this research has been supported by the grants 1p03d01827 , 1p03d01128 and 4t12e04727 . | we present results of our detailed theoretical study of the observed orbital modulation of the radio emission in cyg x-1 .
the modulation occurs due to free - free absorption in the wind from the companion star varying with the orbital phase , and our results put strong constraints on the spatial distribution of the jet radio emission at the frequencies of 215 ghz . a crucial role in enhancing the asymmetry of the wind absorption suffered by the jet emission is played by the irradiation by x - rays emitted in the vicinity of the black hole .
this increases the wind temperature by more than order of magnitude with respect to that of the wind of an isolated supergiant .
the observed phase lags of the minima of the radio emission with respect to the spectroscopic zero phase strongly imply the bulk of the mass of the jet is nonrelativistic ( @xmath0 cm s@xmath1 ) within the jet core .
the jet can , however , become relativistic outside the core .
also , the jet can have a two - component structure , being slow on the outside and fast inside , in which case its synchrotron - emitting part may be relativistic already in the core .
we also consider the observed superorbital modulation of the radio emission ( with the period of @xmath2 d ) and find it can be explained by a jet precession both causing variable wind absorption and changing the jet doppler factor .
finally , we consider the case of cyg x-3 , and show that its lack of observable orbital radio modulation ( in spite of strong modulation of x - rays ) is explained by that system being both much more compact and much more luminous than cyg x-1 .
[ firstpage ] acceleration of particles binaries : general radio continuum : stars stars : individual : cyg x-1 stars : individual : hde 226868 stars : individual : cyg x-3 |
most low - mass main sequence stars are members of binary or multiple systems ( duquennoy & mayor 1991 , fischer & marcy 1992 ) . the binary frequency of pre - main - sequence stars is also high , and probably higher than for main - sequence stars ( mathieu , walter & myers 1989 ; simon et al .
1992 ; ghez , neugebauer & matthews 1993 ; leinert 1993 ; reipurth & zinnecker 1993 ) . in the last decade , interest in these systems has increased significantly with the discovery that many t - tauri and other pre - main - sequence binary stars , possess both circumstellar and circumbinary disks ( for a review see mathieu 1994 ) .
observations of binary systems suggest the existence of disk material around one or both stars , as is inferred from observations of excess radiation at infrared to millimeter wavelengths , polarization , and both balmer and forbidden emission lines ( mathieu 2000 ) .
some extrasolar planets have been found to orbit stars that have a stellar companion , e.g. , 16 cygni b , @xmath1 bootis , and 55 @xmath2 cancri ( butler et al .
1997 ; cochran et al .
1997 ) , confirming that planets can form in binary star systems .
thus the study of stellar disks in binary systems , as well as the possibility of stable orbits , is a key element for better understanding stellar and planetary formation .
it is currently believed that multiple stellar systems result from fragmentation , which produces mainly eccentric binaries ( bonnell & bastien 1992 , bate 1997 ; bate & bonnell 1997 ) , in particular wide stellar systems with separations @xmath3 10 @xmath4 ( bate , bonnell & bromm 2002 ) .
although main - sequence binary systems typically have eccentric orbits ( duquennoy & mayor 1991 ) , most theoretical studies have focused on binaries in near - circular orbits .
extensive and very good theoretical work has been done here ( lubow & shu 1975 ; paczyski 1977 ; rudak & paczyski 1981 ; papaloizou & pringle 1977 ; bonnell & bastien 1992 , bate 1997 ; bate & bonnell 1997 ) .
only a few observed pre - main sequence binaries have known orbital elements ( pogodin et al .
2004 ; schaefer et al . 2003 ; masciadri & raga 2002 ) and information on the accompanying disks is even rarer ( kastner 2004 ; jensen et al .
2004 ; nielbock 2003 ; liu 2003 ; rodriguez et al .
1998 , and references therein ) . from the observational point of view , measuring the size of a circumstellar disk is quite difficult , since these are typically @xmath5 across .
modelers also have a hard task , since dynamical effects like resonances can introduce very fine and complicated structure . in circular binaries , where the gravitational potential is fixed in a uniformly - rotating frame , and an energy - like integral is conserved along particle orbits , the position and strength of resonances can be calculated . in the interesting eccentric case , we lack a conserved integral , and it is not clear how to proceed .
one of the most important studies to date is by artymowicz & lubow ( 1994 ) , who compute analytically the position of orbital resonances ( between 1:4 and 1:3 ) in the circumstellar disks , and approximate the sizes of disks by computing the radius where resonant and viscous torques balance each other .
their results , however , depend strongly on the badly - constrained viscosity parameter . the complexity of the problem has prompted some researchers to use direct @xmath6-body simulations to study the possible existence of stable orbits for planets around eccentric orbits ( david et al .
2003 and references therein ) .
this approach , however , is computationally expensive when accuracy is required .
others have used hydrodynamic simulations ( foulkes et al .
2004 ; guerrero , garca - berro , isern 2004 ; lanzafame 2003 and references therein ) .
unfortunately these simulations depend sensitively on the unknown viscosity as well , and are quite expensive computationally . in this work
we have opted for a simpler approach , analogous to using the structure of periodic orbits in a circular binary , to predict the gas flow .
the path followed by a gas parcel in a stable disk around a star must not intersect itself , or the path of a neighboring parcel ( in the case of planets , the paths may cross and we consider this in the application to a planetary system ) . using a test particle method , we probe the orbital structure of binaries of various eccentricities and mass ratios , and identify families of _ stable invariant loops_. these consist of closed contours in configuration space , such that as each point initially in them evolves forward in time in the binary potential , the contour changes its shape but comes back to its original form and position when the potential has completed a period ( maciejewski & sparke 1997 , 2000 ) . when invariant loops do not cross , they may be filled with gas ;
if the gas loses or gains angular momentum it may drift through a sequence of invariant loops .
although further restrictions may limit the regions occupied by gas in an eccentric binary , our study offers a firm first survey of the regions where gaseous disks and planets can not exist around stars in eccentric binary systems .
in section [ whatisloop ] we review the concept of an _ invariant loop _ , and describe our orbit integration method and the strategy used to find invariant loops .
the application to the circular binary as a test case is presented in section [ ecc0 ] . the application to the general case of arbitrary orbital eccentricity is shown in section [ eccgeneral ] . in section [ observations ]
we apply this study to observations of l1551 irs5 and @xmath0 centauri .
conclusions are presented in section [ conclusions ] .
a time - periodic potential in the ( two dimensional ) orbital plane of an eccentric binary , is mathematically equivalent to a 3-d system with an autonomous hamiltonian , but with the addition of time ( in our case , the binary phase ) and the hamiltonian as two extra dimensions in phase space ( e.g. lichtenberg & lieberman 1992 , section 1.2 ) .
if we restrict ourselves to orbits that lie within the orbital plane of the binary , the extended phase space will have 6 dimensions and regular orbits will lie on a 3-dimensional hypersurface .
the motion of a regular orbit is then multiply periodic ( arnold , 1984 ) with three frequencies , one of which is the oscillation frequency of the potential .
if we examine the system at fixed orbital phases , we slice through the extended phase space at a fixed position on the time axis , and the resulting projection of a regular orbit will lie on a 2-dimensional hypersurface , densely filling an area on the orbital plane .
an additional integral of motion would confine an orbit to lie on a 1-dimensional curve .
every time the system comes back to the initial orbital phase , the particle that follows this orbit will land on the same 1-dimensional curve : an _ invariant loop_. if we were to `` paint '' in a series of different colors all particles that initially lie on such a curve , we would see each follow its own path as the binary stars move around each other , distorting the initial shape ; but as the binary returns to the initial orbital phase , the painted curve would come back to the same initial locus , and although each particle would be at a different point along the curve , the original order of the colors would be conserved .
stable invariant loops represent the generalization to periodically time - varying potentials of the stable closed orbits that form the ` backbone ' of the orbital structure in time - independent potentials .
particle orbits that start near a stable invariant loop will remain trapped close to it , exploring a nearby region in phase space .
gas in a low - viscosity regime trying to settle down in a quasi - static configuration will converge on the non - intersecting stable invariant loops .
we write the equations of motion for the binary star system in term of the eccentric anomaly @xmath7 ( e.g. goldstein 2002 , section 3.7 ) .
we use units where the gravitational constant @xmath8 , the binary semi - major axis @xmath9 , and its total mass @xmath10 are set to unity so that , the binary period is @xmath11 , and its frequency @xmath12 .
the separation of the two stars at time @xmath13 , measured from periastron where the azimuthal angle @xmath14 , is given by the radius @xmath15 , @xmath16 the binary eccentricity , defined as @xmath17 where @xmath9 , and @xmath18 are the semimajor and semiminor axes , and the mass ratio @xmath19 are the only free parameters .
we used an adams integrator ( from the nag fortran library ) to follow the motion of a test particle moving in the orbital plane of the two stars .
kepler s equation ( [ eq1 ] ) is solved within a tolerance of @xmath20 . in the circular case ,
the jacobi energy of the test particle ( computed as a diagnostic for the quality of the numerical integration ) is conserved within one part in @xmath21 per binary period .
the equations of motion of the test particle are solved in an inertial reference frame using cartesian coordinates , with their origin at the center of mass of the binary .
all test particle trajectories are launched when the binary is at periastron , with the two components lying on the @xmath22-axis .
we search initially for families of loops that are symmetric about this line , so we launch test particles at various points along the @xmath22-axis and perpendicular to it .
we then store the particle s position and velocity every time the stars return to periastron .
the computation is halted if the particle runs away , moving further than 10 times the semimajor axis ( @xmath9 ) from the center of mass , or if it comes within a distance of either star that results in a high number of force computations , generally due to close approaches to the stars . in this manner , we obtain orbits around each star ( circumstellar ) , orbits around both stars ( circumbinary ) , runaway orbits , and orbits that are captured by one of the stars . among all the possible orbits , we are seeking a very special type : the _ invariant loops _ , for which the successive phase - space coordinates of our test particle fall on a one - dimensional curve . to find them , we examine the iterates in some two - dimensional subspace , such as the @xmath23 plane .
we plot the positions of the test particle at each complete binary period , and compute their dispersion along the radial direction for those that lie within a sector that spans a small angle ( @xmath24 ) about the x axis . repeating the integration with the same starting @xmath22-value , we adjust the launch velocity @xmath25 to minimize this dispersion . in each panel of figure [ fig.seekloop ] , we show iterates from three particles that orbit around the primary star , and three around the secondary .
as @xmath25 is adjusted , the iterates converge towards the loops in panel d. effectively , our method finds only stable loops .
particles launched near an unstable loop would rapidly diverge from it , and we would not see the convergence illustrated in figure [ fig.seekloop ] . very near the stars , the stable loops are close to stable circular orbits in the potential of that star alone .
we exploit this fact to find families of circumstellar loops : we start by launching a particle close to either star , with a speed appropriate to the circular orbit .
once we have found a loop , we move the starting point in small steps away from the star , and use the loops we have just found to predict the next starting speed .
the process ends when we can find no more loops with larger starting distances . to map out a family of circumbinary loops , we begin at large radii , where the loops are close to circular orbits about the center of mass ; we continue inwards until no more loops are found .
-axis . in panels a ) to d ) , the launch speeds approach those for stable loops , and the dispersion of the iterates decreases until the loop is found.,width=317 ]
in the limit that the binary s eccentricity goes to zero , the invariant loops should be exactly the closed circumstellar and circumbinary orbits ( periodic orbits ) for the circular system .
we computed several orbits for cases with different mass ratios , to compare with other theoretical work .
we took arbitrary points of several computed loops as initial conditions to calculate the orbit in the non - inertial reference frame , corotating with the line that joins the stars .
the resultant orbits indeed close on themselves , showing that for the circular binary , invariant loops are none other than the familiar periodic orbits . in figure [ fig.periodic ] we show
the invariant circumstellar loops found for a circular binary .
the jacobi constant is conserved within @xmath20 . in the top panel
are plotted the non - intersecting circumstellar loops , which could be populated by gas particles . in the bottom panel of this figure
we show an example of the intersecting loops that start mostly when the disks loops approach to the roche lobe . these kind of loops ,
although stable , can not be populated by gas particles due to the intersection with the inner disk loops which would induce shocks dissipating first the less attached gas particles settle down in the intersecting loops . the dotted circles ( aimed with straight arrows ) represent a good fit to the radius of the roche lobe as calculated by the analytic approximation by eggleton ( 1983 ) . for star @xmath26 , the roche radius @xmath27 is given by @xmath28 note that the circumstellar periodic orbits are not circular , but are elongated perpendicular to the line joining the stars .
orbits very close to the roche lobe change their shape dramatically crossing in some cases this frontier ( e.g. circumprimary disk with @xmath29 ) and producing intersections with the inner orbits . in an eccentric binary ,
the circumstellar disks tend to be also elongated in the same sense . in high eccentricity cases ( @xmath30 ) , disks are nearly circular . using our method to search for invariant loops ,
we have computed the limiting radii of circumstellar and circumbinary disks for mass ratios @xmath31 ranging from the jupiter - sun system ( @xmath32 ) to the equal - mass case @xmath33 .
the limiting radii are selected as the loops overlap or when no more loops are found ( the more common case when @xmath34 ) . in figure
[ fig.radiuse0 ] , we present the disk sizes at periastron as a function of the mass fraction . for our limiting loops , we measured the disk extent in both directions along the @xmath22-axis joining the two stars .
the nearly - vertical continuous lines marked at the top with l1 , l2 , l3 are the lagrange points .
the circumstellar disks are confined within the lagrange points , while the circumbinary disk is approximately centered on the origin , the mass center of the binary .
the filled squares are the limiting radii found by rudak & paczyski ( 1981 ) , who approximate the gas streamlines in pressureless accretion disks by non - intersecting simple periodic orbits .
they compute in this way several disk sizes allowed by this potential .
the results of our computations for disk sizes in the circular case are in very good agreement . .
the lagrange points @xmath35 , @xmath36 , and @xmath37 are indicated , and the position of the stars is shown by near - vertical lines with a star at the top and bottom of each one .
the horizontal lines show the actual mass fractions where the computations were done . with filled squares we mark for comparison some examples of the calculated radii from rudak & paczyski ( 1981).,width=317 ]
unlike the zero eccentricity case , where particles see a simple time - steady potential , we now have a variation of the potential in time ; in this manner particles will see a changing situation in any rotating reference frame .
the phase space increases its dimensionality including now time as a canonical variable . in the same manner as for the circular case , we have constructed figures showing disk sizes for eccentricities up to @xmath38 , for a given mass ratio @xmath31 . in figure [ fig.q0.1 ]
we show the regions accessible to circumstellar or circumbinary disks for a binary with a fixed mass ratio of @xmath39 as the eccentricity grows .
@xmath35 , @xmath36 , and @xmath37 are the lagrangian points in the pericenter of the system multiplied by @xmath40 .
the circumstellar disks become rapidly smaller as the binary grows more eccentric , while the circumstellar disk recedes .
additionally , in table [ avrads1 ] , we present the averaged radii of the circumstellar disks . in table [ diskcomps2l ] we compare the averaged radius of the circumsecondary and circumprimary disks , respectively , with the lagrangian radius computed at the pericenter multiplied by @xmath40 .
for all mass ratios and @xmath41 , the size of the disk around the secondary star is well predicted as @xmath42 of the lagrange radius times @xmath40 .
we have fitted a power law in both @xmath43 and @xmath31 to the calculated radii , for eccentricities in the range [ 0.00 , 0.9 ] , and @xmath31 in the range [ 0.01 , 0.99 ] , obtaining the following relation for the size of the circumstellar disks : @xmath44 where @xmath45 is eggleton s estimate for the roche lobe average radius ( equation [ eq.roche ] ) . in table
[ diskcomps2lr ] we have calculated the ratio between this quantity and the actual radii of the circumstellar disks computed ; our formula is accurate to @xmath46 . in table
[ lrcb ] we show the left and right radii of the circumbinary disk measured from the center of mass .
a change in the mass fraction leaves the size and shape of the circumbinary disk almost unaffected . on the other hand ,
a slight increase in the eccentricity will result in a noticeable change as can be appreciated from figure [ fig.q0.1 ] ; the circumbinary disk is now off - center with respect to the center of mass ; it almost preserves a constant distance from the closest approach of either star : see figures [ fig.q0.1_apo ] and [ fig.cbperiapo ] .
( disks radii are computed at periastron ) but with orbital eccentricity as ordinate for the case q=0.1 .
the position of the stars is shown by near - vertical lines with a star at the top of each one .
the horizontal lines show the actual eccentricities where the computations were done .
the dashed lines marked as @xmath35 , @xmath36 , and @xmath37 are the lagrangian points for the circular binary , multiplied by @xmath40.,width=317 ] we have calculated the approximate position of the closest resonances at the edge of the circumstellar and circumbinary disks in all the computed cases presented . for the circular case , for mass ratios
@xmath47 all disks finish approximately at their 1:3 resonance . for an extreme mass ratio , as with the sun - jupiter case , the circumbinary and circumprimary disks can extend further , to the 1:2 resonance .
as the eccentricity increases , the disk is truncated at successively higher order resonances . for @xmath48
the closest resonances to the end of the circumstellar disks are 1:5 or 1:6 . for a higher eccentricity ( @xmath49 ) the closest resonances are of even higher order ( 1:8 to 1:20 ) . for the circumbinary disk at eccentricities , @xmath50 ,
the closest resonances to the inner boundary are 1:4 or 1:5 . , but at apastron : boundaries of the circumstellar disks and the inner edge of the circumbinary disk , for mass ratio @xmath39 , as a function of eccentricity @xmath43 .
dashed lines indicate the inner boundary of the circumbinary disk in figure [ fig.q0.1 ] : showing that the position of this disk changes little with binary phase.,width=317 ] , @xmath51 .
dotted curves in the center show the stars trajectories ; and small curves surrounding the stars ( which are pointed by triangles ) show the extent of circumstellar disks.,width=317 ] [ cols="<,<,>,>,>,>,<,>,<,>",options="header " , ] proper motions do not give clear information about the orbit , but we can use the disk sizes to set bounds on the eccentricity and the mass ratio of the system . for any fixed eccentricity , table [ tab.l1551 ]
shows that for a mass ratio @xmath29 , or @xmath52 , the circumsecondary disk should be roughly 60% smaller than the circumprimary disk .
since we see two nearly equal disks , this implies @xmath53 for this system .
if we assume that is a circular orbit , the circumstellar disks could both extend to 13.5 @xmath4 if the stellar masses are equal , or 9.5 and 15.3 @xmath4 for circumsecondary and circumprimary disks respectively , if @xmath29 . the circumbinary disk gap would then end about 90 @xmath4 from the center of mass of the system .
the measured disks are only approximately 10 @xmath4 in radius , which might be probably due simply to the fact that the emission weakens , make it difficult to detect , however we have also explored the possibility that the reduced size of the disks is due to the fact that the binary orbit is slightly eccentric , in this manner we have constructed a range of possibilities for this system . for @xmath54 ,
table [ tab.l1551 ] shows that the disks should be truncated short of their observed extent of 10 @xmath4 ; so we conclude that the eccentricity of the system is in the interval @xmath55 $ ] . the extreme case with @xmath29 , and @xmath56 , would produce very different disk sizes ( @xmath57 ) , so we discard this possibility also .
thus the system parameters are constrained approximately by @xmath58 . as one of the alternatives to form planets in accretion disks arised the `` planetesimal theory '' :
planets are formed in circumstellar dust disks , where colliding dust grains accrete into planetesimals approximately of 1 - 10 km of diameter ( safronov 1969 ; lissauer 1993 ) .
as the sizes of particles increase and gravitational attraction becomes more important , the cross - section for accretion also increases , as more and more particles collide .
this effect , combined with dynamical friction , leads to `` runaway growth '' : large bodies accrete more efficiently than small bodies , becoming planetary embryos that increase their masses through collisions until most of the material is accreted or dispersed by these new protoplanets ( greenberg et al . 1978 ; wetherhill & stuart 1989 ) .
assuming that planets are initally formed in accretion disks , invariant loops can give us also a good idea of what possibilities a binary system has to harbor stable planets , and the maximum radius of a protoplanetary disk that can give rise to them .
@xmath0 centauri is composed of a binary star consisting on a g2 star with 1.1 m@xmath59 ( @xmath0 centauri a ) and a k1 star with 0.91 m@xmath59 ( @xmath0 centauri b ) and a third component ( @xmath0 centauri c or proxima centauri ) , which is thought to orbit this pair but at a very large distance ( 12,000 @xmath4 ) . the binary system has an eccentricity of 0.52 and a semimajor axis of 23.4 @xmath4 ( see 1893 , heintz 1982 ) . this system and the stability of planets around them has been extensively studied .
wiegert & holman ( 1997 ) used direct numerical integration to find that planets orbiting prograde in the plane of @xmath0 centauri a and b within 3 @xmath4 of either star can be stable for several million years , as can planets in circumbinary orbit more than 70 @xmath4 from the center of mass of the system .
marziari & scholl ( 2000 ) studied the evolution of planetesimals perturbed by gas drag in a disk in the plane of the binary system , finding that planetesimals are able to accrete one another within 2 @xmath4 of @xmath0 cen a. quintana et al .
( 2002 ) follow the growth of planetary embryos under the gravitational forces of the binary system using a symplectic n - body accretion algorithm , finding stable planets in prograde orbits up to 2.5 @xmath4 from either star , and showing how for high inclinations of the orbits , with respect to the plane of the binary , they become unstable . centauri .
ellipses at the center mark the stellar trajectories .
smaller curves around the two three - pointed stars mark the limit of the circumstellar disks.,width=317 ] .,width=317 ] centauri circumprimary disk constructed with invariant loops .
dotted curves show stellar trajectories .
numbers show the ordered phases , as in figure [ fig.diskevol].,width=317 ] circumbinary and circumstellar disks constructed with invariant loops applied to the case of @xmath0 centauri are presented in figures [ fig.totacen ] and [ fig.zoomacen ] .
we calculated a size of 3 @xmath4 for the circumstellar disk ( or potential protoplanetary disk ) around @xmath0 centauri a , and 2.3 @xmath4 for @xmath0 centauri b. the circumbinary gap , measured from the center of mass , should extend to 80 @xmath4 . in figure [ fig.evolacen ]
we show the evolution of the circumprimary disk ( @xmath0 centauri a ) with the phase .
the disk changes its shape and slightly its size as it evolves in time . while it is unlikely that such small changes in shape could be observed directly , the crowding of orbits at phases 1 , 2 , and 5 are likely to lead to increased heating and inflow when the disk contains gas
with the concept of _ invariant loop _ , we have extended the possibilities for orbital studies in accretion or protoplanetary disks of binary stellar systems , with no restriction on the mass ratio ( @xmath31 ) or eccentricity ( @xmath43 ) of the binary . in this manner
the method represents an extension to the periodic - orbit analysis in the well - known circular case . sweeping the parameter space , we are able to address the limits and possibilities for gaseous circumstellar and circumbinry disks from the point of view of the `` pure '' potential exerted by the binary . in the case of accretion disks
, this approximation would closely represent the low viscosity regime of the gas . although the studies presented in this work were restricted to the plane of the disks , this technique
can also be applied to orbits out of the plane of the binary orbit .
compared with high - resolution hydrodynamic simulations the method is computationally cheap and fast .
we find that the size of the circumstellar and circumbinary disks depends strongly on the eccentricity .
the average radius of the circumstellar disks is approximately 40@xmath60 of the lagrangian radius calculated at the pericenter of the binary orbit .
equation [ eq.rfloop ] provides a more accurate analytic approximation to the disk size . the inner radius of the circumbinary disks ( the _ gap _ radius ) is practically independent of the mass ratio . for the circular binary
, the circumbinary disk ends close to the 1:3 resonance for non - extreme mass ratios ( @xmath47 ) . for higher eccentricities ,
the inner edges of the circumbinary disks appear to be truncated close to higher order resonances ( 1:4 , to 1:6 ) .
even a slight increase in eccentricity for a given mass ratio will cause the gap to grow to much larger radii .
the inner edge of a circumbinary disk is not centered at the center of mass ; instead particles in the disks appear to pass at near - equal distances from the apocenter positions of the two stars . in the temporal evolution ,
the circumbinary disks show almost no change with binary phase , while the circumstellar disks change their shape and size only slightly .
the average radius changes by only @xmath61 at different phases of the binary .
the largest or smallest radius can be reached at any phase , i.e. , there is no preference for larger radii at any specific phase . near the outer edge ,
the shape of the disks become less circular in general and slightly off - centered , by up to 5% the change of lopsidedness with the phase is less than @xmath62 of the lopsidedness measured at the pericenter of the binary . for the circumstellar disks the maximum ellipticity is in most cases reached at pericenter , where the semimajor axes of the circumstellar disks are perpendicular to the line that joins the stars .
the ellipticity at the pericenter is higher in the cases with lower eccentricities ( @xmath63)and it is generally below @xmath64 . for larger eccentricities
the ellipticity of the circumstellar disks is @xmath65 .
the maximum change of ellipticity with time is @xmath66 of the ellipticity measured at the pericenter of the binaries . for the circumbinary disks the maximum ellipticity is generally reached when the stars are at the apocenter and
is @xmath67 .
the maximum change in the ellipticity for the circumbinary disks with phase is @xmath61 of the ellipticity at the pericenter . as an application to observations we selected two specific objects : l1551-ir5 and @xmath0 centauri . assuming that the line that joins the stars rests approximately in the plane of the sky , the eccentricity of the system in l1551-ir5 is @xmath68 and the mass ratio is @xmath69 .
the circumbinary inner edge disk should lie approximately at 90 @xmath4 if @xmath70 , and in the interval 100 - 150 @xmath4 for the extreme case @xmath56 . in the case of @xmath0 centauri , we calculated the zones for planets around each star and around the binary .
we find that for @xmath0 cen a , the last stable loop reaches 3 @xmath4 , while for @xmath0 cen b it is at 2.3 @xmath4 . beyond the gap ,
the innermost invariant loops start at a distance of 80 @xmath4 from the center of mass . every day
, disks are being discovered in more binary ( and multiple ) stellar systems .
theoretical studies of these systems can help to restrict their geometrical characteristics , and may also help to constrain some other unknown physical characteristics like viscosity .
we thank luis felipe rodrguez for supplying figure [ fig.l1551 ] .
l. s. and b. p. thank nasa and the space telescope science institute for support through grant hst - ar-09522.01-a .
l. a. is grateful to the department of astronomy of the university of wisconsin - madison for a fruitful sabbatical stay where this work got started .
he also acknowledges support from dgapa / unam through grant in113403 .
b. p. thanks mexico s conacyt for a postdoctoral fellowship , and nasa for support through grant nag5 - 10823 .
pogodin , m. a. , miroshnichenko , a. s. , tarasov , a. e. , mitskevich , m. p. , chountonov , g. a. , klochkova , v. g. , yushkin , m. v. , manset , n. , bjorkman , k. s. , morrison , n. d. , wisniewski , j. p. 2004
, a&a , 417 , 715 | we explore test particle orbits in the orbital plane of eccentric stellar binary systems , searching for `` invariant loops '' : closed curves that change shape periodically as a function of binary orbital phase as the test particles in them move under the stars gravity .
stable invariant loops play the same role in this periodically - varying potential as stable periodic orbits do in stationary potentials ; in particular , when dissipation is weak , gas will most likely follow the non - intersecting loops , while nearby particle orbits librate around them .
we use this method to set bounds on the sizes of disks around the stars , and on the gap between those and the inner edge of a possible circumbinary disk .
gas dynamics may impose further restrictions , but our study sets upper bounds for the size of circumstellar disks , and a lower bound for the inner radius of a circumbinary disk .
we find that circumstellar disks are sharply reduced as the binary s eccentricity grows .
for the disk around the secondary star , the tidal ( jacobi ) radius calculated for circular orbits at the periastron radius , gives a good estimate of the maximum size .
disks change in size and shape only marginally with the binary phase , with no strong preference to increase or decrease at any particular phase .
the circumstellar disks in particular can be quite asymmetric .
we compare our results with other numerical and theoretical results and with observations of the @xmath0 centauri and l1551 systems , finding very good agreement .
the calculated changes in the shapes and crowding of the circumstellar orbits can be used to predict how the disk luminosity and mass inflow should vary with binary phase .
[ firstpage ] circumstellar matter , disks binary : stars . |
the driving mechanism of collimated outflows or jets with relativistic speeds which are observed in active galactic nuclei ( agns ) , gamma - ray bursts , and galactic microquasars is one of the major problems in astrophysics .
a most widely discussed model is based on blandford - znajek ( bz ) process , the electromagnetic energy extraction from a rotating black hole ( bh ) along magnetic field lines threading it @xcite .
this process produces poynting - dominated outflows , which may be collimated by the pressure of the surrounding medium such as the accretion disk and the disk wind ( e.g. * ? ? ?
* ; * ? ? ?
the particles in the outflow can be gradually accelerated depending on the geometrical structure ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , which is consistent with the recent observational implications from the radio jet in the giant elliptical galaxy m87 @xcite ( see also * ? ? ?
bz process was proposed by a pioneering paper of @xcite , who found steady , axisymmetric , force - free solutions of kerr bh magnetosphere in the slow rotation limit where the outward angular momentum ( am ) and poynting fluxes are non - zero along the field lines threading the event horizon .
this was followed by demonstrations of analytical and numerical magnetohydrodynamic ( mhd ) solutions ( e.g. * ? ? ?
* ; * ? ? ?
? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and other force - free solutions ( e.g. * ? ? ?
? * ; * ? ? ?
* ; * ? ? ?
* ) . however , the physical mechanism how the fluxes are created in the electromagnetically - dominated plasma has not been clearly explained .
in contrast , the origin of pulsar winds is identified definitely with the rotation of the matter - dominated central star .
the rotation velocity of the matters of the star @xmath0 and the magnetic field threading the star @xmath1 provide the electromotive force @xmath2 on charges , maintaining the electric field @xmath3 and the poloidal electric currents ( with the toroidal magnetic field @xmath4 ) which form the outward poynting flux @xmath5 in the magnetosphere . as its feedback , the rotation of the stellar matters slows down @xcite ( see also a review in * ? ? ?
* hereafter tt14 ) .
bz process , working in the electromagnetically - dominated bh magnetosphere , does not include any matter - dominated region in which the poloidal magnetic field is anchored .
one should also note that the toroidal magnetic field can not be produced in vacuum just by the rotation of the space - time @xcite .
then how the electric field and currents forming the am and poynting fluxes are created and how the bh rotational energy is reduced in bz process are not simple problems and have been still matters of debate .
see recent discussions in @xcite ( hereafter k09 ) and @xcite . among the numerous calculations , the force - free numerical simulations performed by @xcite ( hereafter k04 ) are most insightful for investigating the essential points on the origin of the fluxes .
tt14 extended the arguments in k04 and k09 and analytically showed that for open magnetic field lines threading the ergosphere in the steady , axisymmetric kerr bh magnetosphere , the situation of no electric potential difference with no poloidal electric current ( i.e. no outward am or poynting flux ) can not be maintained .
the origin of the electric potential differences is ascribed to the ergosphere .
it was also shown that for the open field lines threading _ the equatorial plane in the ergosphere _ , the poloidal currents are driven by electric field ( perpendicular to the magnetic field ) stronger than the magnetic field in the ergosphere , where the force - free condition is violated ( see also section [ sec : equat ] below ) . in this paper
, we mainly discuss the field lines threading the event horizon .
some theorists consider that the membrane paradigm @xcite is useful for effectively understanding the production mechanism of the am and poynting fluxes for such field lines ( e.g. * ? ? ?
* ; * ? ? ?
this interprets the condition at the horizon as a boundary condition @xcite and the horizon as a rotating conductor which creates the potential differences and drives the electric currents in an analogy with the unipolar induction for pulsar winds explained above .
however , the horizon does not actively affect its exterior , but just passively absorbs particles and waves @xcite .
the condition at the horizon should be interpreted as a regularity condition ( * ? ? ?
the mechanism of producing the steady am and poynting fluxes has to work outside the horizon , making the physical quantities consistent with the regularity condition .
for such a causal flux production , some other theorists proposed a scenario that certain types of negative energies ( as measured in the coordinate basis , i.e. as measured at infinity ) created outside the horizon flows towards the horizon , resulting in the positive outward energy flux .
this is an analogy with the mechanical penrose process , in which the rotational energy of a bh is extracted by making it absorb negative - energy particles @xcite .
however , mhd simulations demonstrate that no regions of negative particle energy are seen in the steady state @xcite , although a transient inflow of negative particle energy is possible as a feedback from generation of an outward mhd wave @xcite .
the role of _ negative electromagnetic energy density _ in the steady state has been discussed recently ( k09 ; * ? ? ?
* ; * ? ? ?
* ) , although the concept of ` advection of the steady field ' is ambiguous .
below we show that the sign of the electromagnetic energy density depends on the coordinates , and thus the negative field energy is not physically essential ( see section [ sec : negative_em ] below ) .
we argue that the causal production mechanism of the electromagnetic am and poynting fluxes can not be fully understood by investigating only the steady - state structure .
we examine a time - dependent process evolving towards the steady state with an analytical toy model to clarify how the steady outward fluxes are created . in order to find the essential physics ,
our analysis is performed both in the boyer - lindquist ( bl ) and the kerr - schild ( ks ) coordinate systems .
most of the previous analytical studies used the bl coordinates ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
? * ) ( but see * ? ? ?
* ) , most of the recent numerical simulations used the ks coordinates ( e.g. k04 ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) ( but see e.g. * ? ? ? * ; * ? ? ? * ) , and both of them focused on the steady - state structure .
our new analytical studies of time - dependent process in the bl and ks coordinates will be highly helpful for understanding physics in bz process .
this paper is organized as follows . in section [ sec : formulation ] , we explain our formulation of general relativistic electrodynamics , set generic assumptions for kerr bh magnetosphere , and review the recent analytical understandings given by k04 and tt14 .
section [ sec : equat ] concentrates on the field lines threading the equatorial plane , for which we show the flux production mechanism and the role of the negative energy of particles . in section [ sec : horizon ] , we explain differences between the equatorial plane and the horizon , and then we focus on the field lines threading the horizon , discussing differences of the electromagnetic structures as seen in the bl and ks coordinates and the role of the negative electromagnetic energy density . in section [ sec : unsteady ] , we discuss the time - dependent process towards the steady state . section [ sec : conclusion ] is devoted to conclusion .
the space - time metric can be generally written as @xmath7 where @xmath8 is called the lapse function , @xmath9 the shift vector and @xmath10 the three - dimensional metric tensor of the space - like hypersurfaces .
those hypersurfaces are regarded as the absolute space at different instants of time @xmath11 ( cf . * ? ? ?
we focus on kerr space - time with fixed bh mass @xmath12 and angular momentum @xmath13 .
( the electromagnetic field with outward fluxes which we consider below is a test field for kerr space - time . )
we adopt the units of @xmath14 and @xmath15 , for which the gravitational radius @xmath16 .
we use the dimensionless spin parameter @xmath17 .
kerr space - time has two symmetries , i.e. @xmath18 .
these correspond to the existence of the killing vector fields @xmath19 and @xmath20 . in the coordinates @xmath21 , @xmath22 the event horizon , where @xmath23 , is located at @xmath24 .
the ergosphere is the region @xmath25 , where the killing vector @xmath19 is space - like , @xmath26 . at infinity , this space - time asymptotes to the flat one .
the local fiducial observer ( fidos ; * ? ? ?
* ; * ? ? ?
* ) , whose world line is perpendicular to the absolute space , is described by the coordinate four - velocity @xmath27 the am of this observer is @xmath28 , and thus fido is also a zero am observer ( zamo ; * ? ? ?
note that the fido frame is not inertial , but it can be used as a convenient orthonormal basis to investigate the local physics @xcite .
it should also be confirmed that the fidos are time - like , physical observers ( i.e. @xmath29 )
. in the bl coordinates , one has the following non - zero metric components : @xmath30 where @xmath31 bl fidos rotate in the same direction as the bh with the coordinate angular velocity @xmath32 the bl coordinates have the well - known coordinate singularity ( @xmath33 and @xmath34 , where @xmath35 ) at the horizon .
the bl fidos are physical observers only outside the horizon .
the ks coordinates have no coordinate singularity at the event horizon .
the coordinates @xmath11 and @xmath36 are different from those in the bl coordinates .
the non - zero metric components in this coordinate system are : @xmath37 where @xmath38 ( k04 ; * ? ? ?
the ks spatial coordinates are no longer orthogonal ( @xmath39 ) . from the space - time symmetries ,
@xmath40 are the same in the bl and ks coordinates .
we should note that the ks fidos are different from the bl fidos ( k04 ) .
we follow the definitions and formulations of k04 for electrodynamics in kerr space - time ( except for keeping @xmath41 in maxwell equations ) , in a similar way to tt14 ( see also k09 , references therein , and * ? ? ?
the covariant maxwell equations @xmath42 and @xmath43 are reduced to @xmath44 @xmath45 where @xmath46 and @xmath47 denote @xmath48 and @xmath49 , respectively , and @xmath50 is the levi - civita pseudo - tensor of the absolute space .
the condition of zero electric and magnetic susceptibilities for general fully - ionized plasmas leads to the following constitutive equations , @xmath51 @xmath52 where @xmath53 denotes @xmath54 . at infinity , @xmath55 and @xmath56 , so that @xmath57 and @xmath58 . here
@xmath59 and @xmath60 are the electric and magnetic fields as measured by the fidos , while @xmath61 and @xmath62 are the electric and magnetic fields in the coordinate basis , where @xmath63 .
the current @xmath64 is related to the current as measured by the fidos , @xmath65 , as @xmath66 see appendix [ app : current ] on the relation between convective current and particle velocity .
the covariant energy - momentum equation of the electromagnetic field @xmath67 gives us the am equation as @xmath68 and the energy equation as @xmath69 where @xmath70 denotes @xmath71 , @xmath72 , @xmath73 is the am density , @xmath74 is the am flux ( @xmath75 ) , @xmath76 is the energy density , and @xmath77 is the poynting flux ( @xmath78 ) .
we study the axisymmetric electromagnetic field in kerr space - time which is filled with a plasma .
( the steadiness of the field is partly discarded in section [ sec : unsteady ] . )
we put the additional assumptions similarly to tt14 : ( 1 ) the poloidal @xmath1 field produced by the external currents ( whose distribution is symmetric with respect to the equatorial plane ) is threading the ergosphere .
we call the field lines threading the ergosphere ` ergospheric field lines ' .
( 2 ) the plasma in the bh magnetosphere is dilute and collisionless , but its number density is high enough to screen the electric field along the @xmath1 field lines , i.e. @xmath79 the energy density of the particles is much smaller than that of the electromagnetic fields .
( 3 ) the gravitational force is negligible compared with the lorentz force . ( the gravitational force overwhelms the lorentz force in a region very close to the event horizon @xcite , but
the physical condition in that region hardly affects its exterior . ) the condition @xmath80 and equation ( [ eq : relation_e ] ) lead to @xmath81 .
_ in the steady state , _ we have @xmath82 , which means that @xmath3 is a potential field , and the axisymmetry leads to @xmath83
. then one can write @xmath84 substituting this equation into @xmath82 , one obtains @xmath85 that is , @xmath86 is constant along each @xmath1 field line .
the @xmath3 field is also described by @xmath87 in terms of the magnetic flux function @xmath88 , so that each @xmath1 field line is equipotential and @xmath86 corresponds to the potential difference between the field lines . in the steady , axisymmetric state ,
the equations ( [ eq : am ] ) and ( [ eq : energy ] ) are reduced to @xmath89 @xmath90 where the subscript p denotes the poloidal component . here
one sees that the poloidal am and poynting fluxes are described by @xmath91 respectively .
it should be noted that @xmath92 and @xmath93 are the same in the bl and ks coordinates ( k04 ) .
tt14 shows that the condition @xmath94 is inevitable for the ergospheric field lines in the steady , axisymmetric state ( see also k04 ; k09 ) .
furthermore , for the ergospheric field lines crossing the outer light surface ( see section [ sec : basic_motion ] ) , the condition @xmath95 has to be maintained , i.e. the poloidal am and poynting fluxes are steadily non - zero ( tt14 ) .
the following discussion in this paper focuses on how their values are causally determined and the role of the negative energies as measured in the coordinate basis . under the assumption ( 2 ) for the magnetospheric plasma stated in section [ sec : setup ] , the force - free condition @xmath96 ( or @xmath97 ) is satisfied when @xmath98 ( e.g. k04 ; tt14 ; see also appendix [ app : current ] )
. then equation ( [ eq : am2 ] ) indicates @xmath99 equations ( [ eq : am2 ] ) and ( [ eq : energy2 ] ) mean that no am or energy is exchanged between the particles and the electromagnetic fields
. in this case , in the bl coordinates , the particles drift in the azimuthal direction with angular velocity @xmath86 when @xmath100 ( tt14 ) .
the light surfaces are thus defined as where the four - velocity of a particle with the coordinate angular velocity @xmath86 becomes null , i.e. @xmath101 , where @xmath102 there can be two light surfaces ; the outer light surface ( outside which @xmath103 ) and the inner light surface ( inside which @xmath103 ) . in the case of @xmath104 , @xmath105 at the outer light surface , and @xmath106 at the inner light surface , which is located in the ergosphere ( * ? ? ?
* k04 ; tt14 ) .
the condition @xmath107 is satisfied when the outward poynting flux is non - zero either for the field lines threading the horizon @xcite or the ergospheric field lines threading the equatorial plane ( tt14 ) .
note that @xmath108 is a scalar , so that the location of each light surface is the same in the bl and ks coordinates .
if @xmath109 is realized , the cross - field current flows , i.e. @xmath110 ( tt14 )
. the force - free condition is violated , and @xmath111 varies along a field line .
the mechanism of @xmath86 and @xmath111 being determined along the ergospheric field lines threading the equatorial plane has been already clarified ( k04 ; tt14 ) . generally in the bl
coordinates , one has from equations ( [ eq : relation_e ] ) , ( [ eq : relation_h ] ) , and ( [ eq : omegaf ] ) @xmath112 the key point is that @xmath113 on the equatorial plane because of the symmetry .
therefore , the region of @xmath114 ( i.e. inside the inner light surface , which is within the ergosphere ) can satisfy the condition @xmath109 around the equatorial plane , driving the poloidal current to flow across the field lines . (
note that @xmath115 is a scalar , so that one has @xmath109 also in the ks coordinates . )
this leads to @xmath116 outside the region where @xmath109 , which we call ` current crossing region ' .
the value of @xmath86 will be regulated so that the current crossing region is finite ( see figure 4 of tt14 ) , and thus it is expected to depend on the microphysics in the ergosphere .
the values of @xmath86 and @xmath111 will be determined by the conditions around the equatorial plane and at infinity . in the current crossing region
, @xmath117 is in the opposite direction of @xmath3 , i.e. @xmath118 , as seen in the bl coordinates ( see figure 3 of tt14 ) .
this leads to @xmath119 and @xmath120 , which generate the poloidal electromagnetic am and poynting fluxes ( see equations [ eq : am2 ] and [ eq : energy2 ] ) .
( we confirm that @xmath118 also in the ks coordinates in appendix [ app : b2d2 ] . )
all the ergospheric field lines crossing the outer light surface have @xmath94 and @xmath121 for the northern hemisphere ( @xmath122 for the southern hemisphere ) , while ` the last ergospheric field line ' , which passes the equatorial plane at @xmath123 , has @xmath124 .
this means that the current flows inward along the ergospheric field lines and outward along the last ergospheric field line ( see figure [ fig : summary ] below ) .
correspondingly , the current crossing region extends over @xmath125 .
such a poloidal current structure prevents the bh from charging up .
equations ( [ eq : am2 ] ) and ( [ eq : energy2 ] ) imply that the particles in the current crossing region lose their ams and energies by the feedback , @xmath126 and @xmath127 , from the production of the electromagnetic am and poynting fluxes .
we find that this feedback can make the particles have negative energy as measured in the coordinate basis .
the production of the particle negative energy can be explained by showing the particle motions in the local orthonormal basis carried with the bl fidos , in which the equation of a particle motion with four - velocity @xmath128 , three - velocity @xmath129 , charge @xmath130 , and mass @xmath131 is written as @xmath132 where @xmath133 denotes the vector component in respect of the fido s orthonormal basis ( * ? ? ?
* tt14 ) . in this basis
one can investigate local , instantaneous particle motions under the lorentz force as special relativistic dynamics .
( the fido frame is not inertial and a particle feels the gravitational force , although we neglect it compared with the lorentz force , based on the assumption ( 3 ) set in section [ sec : setup ] . )
the am and energy per mass of a particle as measured in the coordinate basis are @xmath134 u^t = ( \alpha + \sqrt{\gamma_{\varphi\varphi}}\omega \hat{v}^\varphi ) \hat{u}^t , \label{eq : energy_p}\end{aligned}\ ] ] where we have used @xmath135 ( cf .
( upper three lines with positive values ) and @xmath136 ( lower three lines with negative values ) of the posively charged particle in the fixed bl fido s orthonormal basis as functions of time normalized by gyration time scale @xmath137 . the solid , dashed , and dot - dashed lines are calculation results for @xmath138 and @xmath139 , respectively .
the initial conditions are @xmath140 . ] near the equatorial plane , the @xmath141 field is approximately perpendicular to that plane , because @xmath142 at that plane , and then the @xmath143 field is radial in that plane ( see figure [ fig : motion ] ) .
the motion of a test particle can be easily solved in such fields @xcite . for the case of @xmath144 which we focus on
, the positively ( negatively ) charged particles are accelerated in the directions of @xmath143 ( @xmath145 ) and @xmath146 . in figure
[ fig : motcal ] , we show the calculation results for @xmath147 and @xmath139 , where we fix the basis and assume that the electromagnetic fields are uniform . for @xmath148 ( i.e. @xmath149 ) in particular ,
the particles are strongly accelerated in the direction of @xmath146 , and then one obtains @xmath150 in several tens of gyro radius scale @xmath151 ( not normalized by the gravitational radius ) . as a consequence , from equations ( [ eq : am_p ] ) and ( [ eq : energy_p ] ) , one has @xmath152 in the ergosphere , where @xmath153 . for @xmath109 , @xmath136 does not approach @xmath154 , so that @xmath155 near the boundary of the ergosphere where @xmath156 .
however , @xmath157 for @xmath158 implies that @xmath159 can be realized near the horizon .
here we emphasize that @xmath160 and @xmath161 are scalars , and thus @xmath162 and @xmath159 also in the ks coordinates . for typical agn jets
, @xmath163 is expected to be @xmath164 orders of magnitude smaller than @xmath165 ( cf . * ? ? ?
* ) , so that the distance which a particle travels until it achieves the asymptotic azimuthal velocity is tiny compared to the size of the ergosphere .
this justifies our calculations of the particle motion in the fixed orthonormal basis with the uniform electromagnetic fields , and the asymptotic velocities can be interpreted as the local velocities of the test particles . since the current crossing region
is bounded at @xmath166 , the positively charged particles do not cross the last ergospheric field line and will gyrate around this field line . when they emerge out of the ergosphere
, they contribute to the current flowing outward along the last ergospheric field line ( see figure [ fig : summary ] ) .
the particles outside the ergosphere generally have positive energies .
we argue that bz process for the ergospheric field lines threading the equatorial plane is similar to the mechanical penrose process , in which the rotational energy of a bh is extracted as mechanical energy by making the bh absorb negative - energy particles @xcite . for simplicity ,
let us consider the positively and negatively charged particles in the geometrically thin current crossing region as a one - fluid .
the energy equation for this fluid in the steady state is written as @xmath167 where @xmath168 is the energy - momentum tensor of the fluid .
the boundary condition at @xmath123 is @xmath169 .
therefore , the solutions of equation ( [ eq : particleeom ] ) should be @xmath170 in the current crossing region .
then one has @xmath171 , where @xmath172 and @xmath173 are the comoving mass density and the four - velocity of the fluid , respectively .
since all the particles may have negative energy , it is reasonable to estimate @xmath174 this means that the current crossing region generates the inflow of the negative - energy fluid and the outward poynting flux , which appears to be a similar process to the mechanical penrose process . as a result
, the bh loses its rotational energy by the poloidal particle energy flux @xmath175 .
we summarize our argument in figure [ fig : summary ] ( see section [ sec : horizon ] for the field lines threading the horizon ) .
( _ open arrows _ ) , the poloidal poynting flux @xmath176 ( _ filled arrows _ ) , and the poloidal particle energy flux @xmath177 near the equatorial plane ( i.e. the inflow of the particle negative energies ; _ striped arrow _ ) in the steady state in the northern hemisphere in the ks coordinates .
the bh loses its rotational energy directly by @xmath176 along the field lines threading the horizon ( see sections [ sec : horizon ] and [ sec : unsteady ] ) and by @xmath177 near the equatorial plane which is associated with @xmath176 along the field lines threading the equatorial plane in the ergosphere ( see section [ sec : equat ] ) . ]
however , it is too simple to treat the charged particles in the current crossing region as a one - fluid , since the average velocities of the positively and negatively charged particles should be different .
furthermore , figure [ fig : motcal ] is just the result of the test particle calculations .
more detailed studies of the plasma particle motions are required to confirm whether the condition of equation ( [ eq : penrose ] ) is realistic in the current crossing region .
mhd numerical simulations treat the energy of particles ( while force - free simulations not ) , so that they can observe the negative particle energy in principle .
however , the mhd simulation results of the dilute kerr bh magnetosphere with cylindrical magnetic field at the far zone in @xcite do not show any negative particle energy in the steady state .
this is just due to the disappearance of the ergospheric field lines threading the equatorial plane , although the reason of this disappearance has not been identified .
such behavior is also seen in the three dimensional mhd simulations including the dense accretion flow @xcite(but see * ? ? ? * ) .
in contrast to the equatorial plane where @xmath113 from the symmetry , one generally has @xmath116 at the horizon .
thus the above argument on the field lines threading the equatorial plane is not applicable for the field lines threading the horizon . at the horizon
, the regularity condition should be satisfied @xcite : @xmath178 for the bz split - monopole solution @xcite as an example , in which @xmath179 , @xmath180 , and @xmath181 ( to the zeroth order of @xmath182 ) , so that one has @xmath183 therefore the force - free condition is satisfied at the horizon .
we confirm this fact more generally in the ks coordinates . from the calculation shown in appendix [ app :
b2d2 ] , we obtain @xmath184 ( b^r)^2 , \label{eq : b2d2}\end{aligned}\ ] ] where @xmath185 is generally satisfied ( see equation [ eq : identity2 ] ) . for the bz split - monopole solution as an example , in which @xmath186 and @xmath187 at the northern hemisphere ( see section [ sec : structure ] below ) , @xmath188 , and @xmath189 , one has @xmath190 in the region where @xmath191 .
we see that @xmath190 is generally satisfied where @xmath192 is weak , @xmath193 , and @xmath191 .
( note that for the field lines threading the equatorial plane , @xmath192 is the dominant field component near that plane , where @xmath194 can be realized . )
therefore , for the field lines threading the horizon , the force - free condition can be satisfied , and then no poloidal current is driven to flow across the field lines in the steady state .
no am or energy is transferred from the particles to the electromagnetic fields .
these properties clearly indicate that the flux production mechanism for the field lines threading the horizon is different from that for the field lines threading the equatorial plane . here
we focus on the electromagnetic structure of the bz split - monopole solution , and show that some properties are measured differently in the bl and ks coordinates .
this analysis is useful for finding the essential physics in bz process for the field lines threading the horizon , which should be independent of the adopted coordinate systems .
the split - monopole field is given by @xmath195 which satisfies @xmath196 .
( note that @xmath197 for @xmath198 . ) in the bl coordinates , one has @xmath199 as is well known , @xmath200 changes its sign at the point where @xmath201 ( see figure [ fig : structure ] , _ left _ ) .
we can see that @xmath202 and @xmath200 diverge as @xmath158 , while @xmath203 is finite , and one has @xmath204 near the horizon .
on the other hand , in the ks coordinates , one has @xmath205 equation ( [ eq : bphi_ks ] ) is derived by rewriting @xmath111 in equation ( [ eq : relation_h ] ) with equation ( [ eq : relation_e ] ) and ( [ eq : identity3 ] ) ( k04 ) . the regularity condition at the horizon ( @xmath35 ) for the steady flow to pass with no diverging physical quantities is given by @xmath206 which is equivalent to equation ( [ eq : regularity_bl ] ) .
we calculate @xmath207 and @xmath200 from @xmath208 towards infinity for small values of @xmath182 and @xmath209 , and find that @xmath187 , that @xmath200 does not change its sign ( see figure [ fig : structure ] , _ right _ ) , and that @xmath210 near the horizon .
that is , the @xmath117 field in the ks coordinates is not only so weak that it can not drive the cross - field current but also it does not change its direction , i.e. @xmath211 in the whole region .
this situation is in stark contrast to the field lines threading the equatorial plane , for which the cross - field current is driven by the strong @xmath117 field with @xmath118 . in the bl
coordinates the point where @xmath201 and @xmath212 appears special , and it was considered as a key in some previous analytical discussions ( e.g. * ? ? ?
however , the electromagnetic quantities are clearly continuous or seamless in the ks coordinates , as shown in figure [ fig : structure ] .
below we generically consider the cases in which the force - free condition is satisfied along the field lines threading the horizon ( see section [ sec : ff_horizon ] ) . in those cases ,
an essential point is that the outward am and poynting fluxes , @xmath213 and @xmath214 , are seamless along each field line from the event horizon to infinity in the steady state ( from equation [ eq : forcefree ] ) , with no transfer of am and energy from the particles .
this situation is described in figure [ fig : summary ] .
now we discuss how @xmath215 and @xmath176 are created along the field lines threading the horizon .
blandford & znajek @xcite show that @xmath86 and @xmath111 in the steady state are determined _ mathematically _ from equations ( [ eq : omegaf ] ) and ( [ eq : forcefree ] ) with the conditions at the horizon and at infinity ( see also k04 ) .
this mathematics and the seamless property shown above may lead to an incorrect consideration that the fluxes are created at the horizon .
the conditions at the horizon and at infinity are not boundary conditions but regularity conditions ( * ? ? ?
* k04 ) , as stated in section [ sec : intro ] .
the place where the fluxes are created must not be the horizon , but outside the horizon .
we note that the non - zero outward am and poynting fluxes at the horizon in the ks coordinates do not violate causality , because the steady fluxes carry no information .
it should be also noted that the steady poynting flux @xmath216 is not a product of a certain energy density and its advection speed like steady particle energy flux @xmath217 ( see section [ sec : negative_em ] for a related discussion ) .
the poynting flux is just a result of the currents flowing in the plasma with the potential differences .
consequently , the issue on the field lines threading the horizon is well defined as how is the steady current structure causally built ? "
we consider that this issue may not be resolved by investigating only the steady - state structure .
the phenomena at the horizon should be a result from those having occurred outside the horizon in the _ prior _ times @xmath11 . in section [ sec : unsteady ] , we address this issue by discussing a time - dependent state evolving towards the steady state .
lasota et al .
@xcite and koide & baba @xcite argue that the outward poynting flux is mediated by ` inflow of the negative electromagnetic energy ' ( see also k09 ) .
although this interpretation analogous to the mechanical penrose process looks attractive for causal production of the poynting flux , it is difficult to consider _ the flow of the steady field _ ( rather than waves ) .
furthermore , we find that the sign of the electromagnetic energy density depends on the coordinates
. in the bl coordinates , the electromagnetic am and energy densities can be written down by ( k09 ) @xmath218.\end{aligned}\ ] ] thus @xmath219 and @xmath220 is negative ( and diverges ) near the horizon when @xmath221 .
this condition is satisfied in the bz split - monopole solution . on the other hand , in the ks
coordinates , the calculations shown in appendix [ app : b2d2 ] lead to @xmath222 b^\theta b_\theta + \delta \sin^2\theta ( b^\varphi)^2 \nonumber \\ & & - 2a\sin^2\theta b^r b^\varphi + \left[1+\omega_{\rm f}^2 ( \varrho^2 + 2r)\sin^2\theta \right ] ( b^r)^2 .
\label{eq : e_ks}\end{aligned}\ ] ] in the bz split - monopole solution as an example , in which @xmath188 and @xmath187 in the northern hemisphere , one has @xmath223 this condition is generally valid when @xmath192 is weak and @xmath193 .
note that @xmath224 depend on the coordinates , while @xmath225 and @xmath226 are scalars .
the concept of the negative electromagnetic energy density depends on the coordinates , and thus it is not physically essential . where @xmath227 and @xmath228 is the four - velocity of the bl fido . ]
as stated in section [ sec : problem ] , we address the issue how the steady poloidal current structure is built causally , by discussing a time - dependent state evolving towards the steady state . in the steady state
, the plasma has the inner and outer light surfaces ( see section [ sec : basic_motion ] ) .
the particles flow in across the inner light surface and flow out across the outer light surface . therefore , new particles have to keep injected between the two light surfaces , as discussed in many literatures ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
in this paper we have assumed that the plasma particles keep injected from outside the magnetosphere through electron - positron pair creation by collisions of two photons @xcite and/or diffusion of high - energy hadrons @xcite , but sufficient for satisfying @xmath80 @xcite . ] , and that those particles maintain @xmath80 and carry the currents .
now let us first consider a _
vacuum _ in kerr space - time , and then begin the continuous injection of force - free plasma particles between the two light surfaces as a gedankenexperiment . the inflow ( outflow ) will fill the vacuum near the horizon ( at infinity ) . simultaneously we will see a process building the poloidal current structure .
hereafter we will call the ( inflow + outflow ) region filled with the force - free plasma ` force - free region ' .
figure [ fig : unsteady ] is a schematic picture of this process focusing on the inflow .
we show the space - time diagrams of the inner and outer boundaries of the force - free region in the bl and ks coordinates in figure [ fig : spacetime ] , in which the radial light signals are represented by the small arrows .
the outflow continues to propagate into the vacuum , i.e. the radius of the outer boundary @xmath198 for @xmath229 . in the bl
coordinates , the inflow also continues to propagate towards the horizon , @xmath158 for @xmath229 . in the ks
coordinates , the inflow can pass the horizon in a finite time of @xmath230 . in both of the coordinates ,
when the inner boundary approaches the horizon , the outward signal from it becomes slower and slower and it can hardly affect the force - free region .
this will lead to the steady state .
although such a time - dependent state should be analyzed numerically , we use a toy model to qualitatively illustrate the process of building the poloidal current structure .
this model assumes that ( 1 ) @xmath231 is fixed to be split - monopole @xmath232 in the whole region , and that ( 2 ) the kerr bh magnetosphere is separated into the force - free region and the vacuum by geometrically thin boundaries moving radially . for further simplicity , ( 3 )
we assume that the force - free region and the vacuum have their steady - state structures , but the values of the physical quantities , particularly @xmath86 and @xmath111 , keep updated as determined by the varying conditions of the inner and outer boundaries .
some of these assumptions would be violated in realistic experiments .
nevertheless we consider that our toy model is useful to suggest the key points for resolving the issue on the causality in the coordinate basis ( section [ sec : flux_production ] ) , which also allows us to understand how the steady state is maintained ( section [ sec : discussion ] ) . the electromagnetic quantities in the force - free region
are given as follows .
the condition @xmath80 and @xmath82 lead to @xmath233 where @xmath234 hereafter we will put the subscript and superscript ` ff ' on the quantities in the force - free region . equations ( [ eq : relation_e ] ) and ( [ eq : relation_h ] ) give us @xmath235 @xmath236 equation @xmath237 and the force - free condition lead to @xmath238 @xmath239 these two equations imply that @xmath240 .
we focus on the northern hemisphere , where @xmath241 and @xmath242 .
the current flowing outward @xmath243 , which prevents the bh from charging up , is assumed to be concentrated on the equatorial plane .
the poloidal am and poynting fluxes are @xmath244 which satisfy @xmath245 and @xmath246 . in the vacuum region , one has @xmath247 .
equations @xmath82 and @xmath248 lead to @xmath249 which indicates @xmath250 hereafter we will put the subscript and superscript ` vac ' on the quantities in the vacuum region .
let us focus on the inner boundary of the force - free ( inflow ) region , and derive the conditions on the boundary , i.e. the junction conditions between the force - free and vacuum regions .
the similar analysis can be done for the outer boundary . for equation
@xmath251 we substitute @xmath252 where @xmath253 and @xmath254 are the heaviside step function and the dirac delta function , respectively , and @xmath255 where @xmath256 and @xmath257 are the initial radius and the velocity of the boundary .
the location of the boundary is represented by @xmath258 .
we have introduced @xmath259 in equation ( [ eq : jr_bl ] ) , i.e. possible contribution to @xmath260 from moving surface charges at the boundary .
the assumption ( 3 ) stated in the first part of this section implies that the timescale for the quantities in the force - free and vacuum regions becoming adjusted for steady - state structure is much smaller than the timescale of the boundary propagation .
we focus on the latter timescale , considering that only @xmath261 depends on @xmath11 in equation ( [ eq : ampere_bl ] ) .
then we have @xmath262 taking account of equation ( [ eq : ampere_ff_bl ] ) , we obtain @xmath263 which implies that the surface charge density on the boundary @xmath264 .
this can be confirmed by integrating @xmath265 over the infinitesimally thin ( in the @xmath266 direction ) region enclosing the small area on the boundary and taking account of @xmath267 . for equation
@xmath268 we substitute @xmath269 and equation ( [ eq : hphi_bl ] ) .
we have introduced @xmath270 , possible contribution to @xmath271 from the surface current flowing on the boundary .
then we have @xmath272 which leads to @xmath273 the last one of maxwell equations nontrivial for the present problem is @xmath274 for which we substitute @xmath275 then we have @xmath276 = 0 .
\label{eq : faraday_bl2}\end{aligned}\ ] ] integrating equation ( [ eq : faraday_bl2 ] ) over @xmath277 and take a limit of @xmath278 , the last term vanishes , and we obtain @xmath279 where we have used equation ( [ eq : relation_e ] ) for the last equality . eliminating @xmath280 from equations ( [ eq : v1_bl ] ) and ( [ eq : v2_bl ] ) leads to @xmath281 here we take the minus sign , since we have assumed that the inner boundary keeps moving inward . in section [ sec : consistency ] , we will confirm that this assumption is consistent with the electromagnetic structure which we found .
let us consider the case of @xmath282 .
then we have @xmath283 and @xmath284 \right|_{r=0}. \label{eq : znajek_bl}\end{aligned}\ ] ] substituting @xmath285 for equation ( [ eq : metric ] ) , we find @xmath286 which has to be @xmath287 .
this means that the four - velocity of the boundary is null . in reality , however , the particles at the boundary can not propagate with this speed , and thus one can conclude @xmath288 i.e. , the cross - field current must flow on the boundary .
note that equation ( [ eq : znajek_bl ] ) with @xmath289 becomes equivalent to the regularity condition at the horizon ( equation [ eq : regularity_bl ] ) .
in our toy model of the time - dependent state , we have not taken into account equations of the particle motions , using the force - free approximation for the force - free region , but we have assumed that the inner boundary keeps moving inward , i.e. @xmath290 . here
we examine the direction of the lorentz force exerted on the particles at the boundary , and confirm that it is consistent with the assumption of @xmath291 .
it is reasonable that the force - free approximation is not applicable for the boundary between the force - free and vacuum regions , and indeed we have seen that the cross - field current flows there , @xmath292 .
the particle number density @xmath293 of the force - free region is high enough to screen the electric field along the @xmath1 field lines , i.e. @xmath294 .
we may even assume that @xmath295 , where @xmath296 is the charge density of the force - free region , and then the distribution of @xmath293 is not directly related to that of @xmath296 .
on the other hand , @xmath297 approaches zero at the boundary towards the vacuum region , where @xmath298 is not valid , and non - zero surface charge density @xmath299 just implies non - zero surface mass density @xmath300 .
thus we can write the equation of the particle motions in the @xmath266 direction as @xmath301 = f^r_{~~\nu } i^\nu$ ] and the continuity equation as @xmath302 = 0 $ ] , where @xmath303 is the mass density of the force - free region .
we combine these two equations , use @xmath304 , and integrate the equation over @xmath305 ( i.e. , keep the components including @xmath254 as done in section [ sec : innerb ] ) to have @xmath306 at the boundary @xmath307 and @xmath308 , and then the lorentz force will be much stronger than the gravitational and inertial forces . we neglect the latter forces as in section [ sec : equa_eom ] , so that the first term in the left - hand side of equation ( [ eq : eom_boundary ] ) can be rewritten as @xmath309 . in equation ( [ eq : eom_boundary ] ) , @xmath310 should have a value between @xmath294 and @xmath311 , and @xmath312 between @xmath313 and @xmath314 .
we also found @xmath292 .
these mean that the right - hand side of equation ( [ eq : eom_boundary ] ) is negative , i.e. , the lorentz force exerted on the boundary is in the direction of @xmath315 .
the second term in the left - hand side of equation ( [ eq : eom_boundary ] ) represents momentum change of the boundary layer due to its mass exhange with the force - free region , and this term is zero when @xmath316 .
in the other case , we have @xmath317 since the boundary and the force - free region do not separate . in our toy model , the particles are continuously injected between the two light surfaces , and the particles flow outward across the outer light surface and flow inward across the inner light surface . for the inflowing force - free region , the continuity equation @xmath318 and
its assumed steady axisymmetric structure mean @xmath319 . if @xmath320 , the acceleration @xmath309 is negative , and we continue to have @xmath321 and then @xmath290 .
the acceleration could be positive when @xmath322 , but this case means @xmath290
. the acceleration can not be positive while @xmath323 .
therefore , the inner boundary keeps @xmath290 .
we can also confirm @xmath323 for the outer boundary .
equations ( [ eq : etar_bl ] ) , ( [ eq : v1_bl ] ) , and ( [ eq : v2_bl ] ) are valid with changes @xmath324 and @xmath325 , and thus one obtains the conditions @xmath326 and @xmath327 .
these indicate @xmath328 .
equation ( [ eq : eom_boundary ] ) is valid with change @xmath329 .
the same argument as for the inner boundary leads to the conclusion @xmath330 . in order to check the consistency of our model
more rigorously , fully time - dependent numerical calculations without the assumptions that we set are required , but they are beyond the scope of this paper .
since @xmath291 for the inner boundary and @xmath331 , equation ( [ eq : v2_bl ] ) means @xmath332 the electromagnetic am density is given as @xmath333 .
then equation ( [ eq : dshift_bl ] ) indicates @xmath334 that is , the inner boundary of the force - free region converts the vacuum with larger am density into the force - free plasma with smaller am density .
now equation ( [ eq : am ] ) can be written as @xmath335 substituting @xmath336 and equations ( [ eq : hphi_bl ] ) and ( [ eq : jtheta_bl ] ) for equation ( [ eq : am_bl ] ) , we obtain @xmath337_{r=0}. \label{eq : amconv_bl}\ ] ] taking account of equations ( [ eq : cfcurrent_bl ] ) and ( [ eq : amineq_bl ] ) , we find that the electromagnetic am flux in the force - free region is produced by the conversion of the electromagnetic am density from the vacuum to the force - free plasma through the boundary and the torque of the cross - field current at the boundary . equation ( [ eq : amconv_bl ] ) can also be derived from equation ( [ eq : v1_bl ] ) .
these equations also mean that @xmath338 is produced by the displacement current @xmath339 and the cross - field current @xmath340 .
none of these two contributions appears in the steady state ( see sections [ sec : ff_horizon ] and [ sec : structure ] ) .
equation ( [ eq : energy ] ) can be reduced to @xmath341 substituting @xmath342 and equations ( [ eq : hphi_bl ] ) , ( [ eq : jr_bl ] ) , and ( [ eq : jtheta_bl ] ) for equation ( [ eq : energy_bl ] ) , we obtain @xmath343_{r=0}. \label{eq : energyconv_bl}\ ] ] by using the expressions @xmath344 we find that @xmath345 satisfy equation ( [ eq : energyconv_bl ] ) . in equation ( [ eq : energyconv_bl ] ) , the term @xmath346 .
one can see that the poynting flux in the force - free region is produced by the electromagnetic energy conversion @xmath347 and the work of the cross - field current @xmath348 .
we can obtain the same conclusions as above in the ks coordinates , where the calculations are complicated compared to those in the bl coordinates due to @xmath39 .
equations having different shapes from those in the bl coordinates are @xmath349 for the force - free region , and @xmath350 for the vacuum region . from equations ( [ eq : ampere_bl ] ) , ( [ eq :
ampere2_bl ] ) , and ( [ eq : faraday_bl ] ) , we obtain @xmath351 @xmath352_{r=0 } , \label{eq : v1_ks2}\end{aligned}\ ] ] @xmath353 where we have used equations ( [ eq : hff_ks ] ) and ( [ eq : hvac_ks ] ) to derive equation ( [ eq : v1_ks2 ] ) .
eliminating @xmath280 from equations ( [ eq : v1_ks2 ] ) and ( [ eq : v2_ks2 ] ) leads to @xmath354 the sign of the first term in the right - hand side is not determined in this analysis .
however , we have assumed @xmath290 , and will confirm that the assumption @xmath290 is consistent with the electromagnetic structure .
if @xmath282 , we have @xmath355 and @xmath356 substituting @xmath285 for equation ( [ eq : metric ] ) , we find @xmath357 which has to be @xmath287 , indicating that the four - velocity of the boundary is null . because this velocity can not be realized , we can conclude @xmath358 .
the force - free region always satisfies equation ( [ eq : bphi_ks ] ) and has no diverging quantities , and thus the regularity condition at the horizon ( equation [ eq : regularity_ks ] ) automatically becomes satisfied after the boundary crosses the horizon .
we have shown that @xmath292 in the bl coordinates .
since @xmath359 is the same in the bl and ks coordinates , we have @xmath360 also in the ks coordinates .
this is consistent with our assumption that the outward flowing current is concentrated on the equatorial plane ( see the texts below equation [ eq : ampere_ff_bl ] ) . by substituting equation ( [ eq : vmax_ks ] ) for equation ( [ eq : metric ] )
we find that @xmath361 is required for the time - like propagation @xmath362 .
then we have @xmath363 . equation ( [ eq : v1_ks ] )
shows that @xmath338 is produced by the displacement current and the cross - field current at the boundary . if the former is dominant , @xmath364 is realized so that @xmath365 .
this means @xmath366 .
the production of the electromagnetic am flux can also be understood by the same equation as equation ( [ eq : amconv_bl ] ) in the bl coordinates . in this case ,
equation ( [ eq : v2_ks2 ] ) with @xmath367 leads to @xmath368 , which means that the minus sign should be taken in equation ( [ eq : vmax_ks ] ) .
the production of the poynting flux can also be understood by equation ( [ eq : energyconv_bl ] ) similarly to the case in the bl coordinates . here
the electromagnetic energy densities are @xmath369 where we should note that @xmath370 is different from @xmath371 , since @xmath372 .
we confirmed that equation ( [ eq : e0_bl ] ) satisfies equation ( [ eq : energyconv_bl ] ) also in the ks coordinates .
we confirm that the assumption @xmath290 is consistent with the electromagnetic structure which we found .
the equation of motion is written down as @xmath373 we should have @xmath374 by the same argument as in the bl coordinates , and we have @xmath375 .
thus the lorentz force is in the @xmath315 direction while @xmath290 , and will overwhelm the gravitational and inertial forces . the inner boundary layer starts with @xmath290 and it is reasonable that it changes its velocity continuously . then @xmath323 must not be realized because the acceleration can not be positive around @xmath376 .
that is , the inner boundary layer keeps @xmath291 .
as shown above , @xmath338 and @xmath86 , or the electromagnetic am and poynting fluxes , are created at the inner boundary which propagates towards the horizon .
this is a causal mechanism of the flux production as measured in the coordinate basis .
after the inner boundary becomes very close to the horizon in the bl coordinates or it crosses the horizon in the ks coordinates , it does not affect the exterior , and @xmath338 and @xmath86 are fixed to be consistent with the regularity condition at the horizon ( and at infinity ) .
this implies that _ no source of @xmath338 and @xmath86 or the am and poynting fluxes is required in the steady state . _
the steady poloidal currents are just flowing along the field lines without crossing them , and no force is required to drive the currents in the steady state , partly because the force - free plasma is assumed to have no resistivity .
this situation is essentially different from that in a steady pulsar wind , in which the electromotive force @xmath2 drives the cross - field current in the rotating star , and the fluxes definitely have the electromagnetic sources , i.e. @xmath377 and @xmath378 ( see section [ sec : intro ] and tt14 ) . as a result
, we see that the bh loses its rotational energy directly by @xmath176 along the field lines threading the horizon , as described in figure [ fig : summary ] .
the plasma may have a finite resistivity in more realistic bh magnetosphere or in the numerical simulations . in this case a certain force is required to maintain the steady - state currents . k04 and k09 suggest that a weak @xmath117 field component _ parallel to the @xmath1 field line _ is induced and drives the currents in the steady state without violating the regularity condition significantly .
the inner boundary approaching the horizon as seen in the bl coordinates looks similar to the stretched horizon in the membrane paradigm at first sight .
however , they are essentially different . in the membrane paradigm
, @xmath338 is produced by the fictitious cross - field current flowing on the stretched horizon with joule dissipation .
on the other hand , we have shown that @xmath338 is produced not only by the cross - field current but also by the displacement current .
we should note that mechanism of driving the cross - field current on the inner boundary might be different from that near the equatorial plane which is discussed in section [ sec : equat ] . in the latter case , @xmath109
can be realized due to the property of the ergosphere , which drives the cross - field current . on the other hand , the mechanism of driving the cross - field current between the force - free and vacuum regions ( and its relation to the property of the ergosphere )
may not be understood in our toy model , where @xmath80 could be violated there .
more studies on the plasma physics as measured by the fidos would be required .
we have generically discussed the axisymmetric kerr bh magnetosphere in which a collisionless plasma satisfies @xmath80 ( i.e. there is no gap in the plasma region ) , and clarified the causal production mechanism of the electromagnetic am and poynting fluxes ( i.e. @xmath111 and @xmath86 ) along the ergospheric field lines crossing the outer light surfaces and the role of the negative energies as measured in the coordinate basis .
our conclusion is the following . for the field lines threading the equatorial plane ,
as shown in k04 and tt14 , @xmath111 is produced by the cross - field current flowing in the region where @xmath109 near the equatorial plane , and @xmath86 will be regulated so that the current crossing region is finite . in this paper
, we have shown that the particles in that region can have negative am and negative energy as measured in the coordinate basis by a feedback from the flux production , and shown by using the one - fluid approximation that those particles flow towards the horizon ( see figure [ fig : summary ] ) .
thus bz process for these field lines appears to be a similar process to the mechanical penrose process .
we have also compared our arguments to the recent mhd numerical simulation results briefly in section [ sec : mhdnumerical ] . for the field lines threading the horizon , the structure of the outward electromagnetic am and poynting fluxes ( or the poloidal currents and the electric potential differences )
must not be created by the horizon , but must be a result from phenomena having occurred outside the horizon in the prior coordinate times . to illustrate this concept ,
we have built a toy model of a time - dependent state in which the force - free plasma injected continuously between the two light surfaces is filling a vacuum ( see figure [ fig : unsteady ] ) . as a result , we have seen that the fluxes are produced by the contributions from the displacement current and the cross - field current at the in - going boundary ( see equations [ eq : amconv_bl ] and [ eq : energyconv_bl ] ) . in the steady state
, the in - going boundary does not affect the force - free region , and the fluxes are maintained without any electromagnetic source ( if the resistivity is negligible ) . @xmath111 and @xmath86 are maintained to be consistent with the regularity condition at the horizon and at infinity .
the force - free condition is satisfied along the field lines threading the horizon in the steady state , and then conversion of the am and energy from the particles is negligible .
thus we support the mathematical treatments of @xcite and @xcite for determining @xmath111 and @xmath86 of the steady - state force - free plasma .
we have shown that the concept of the inflow of negative electromagnetic energy along the field lines threading the horizon is not physically essential .
the steady outward poynting flux should be interpreted just as a result of the currents flowing in the plasma with the electric potential differences .
the outward poynting flux at the horizon in the ks coordinates does not violate causality , because the steady fluxes carry no information .
the bh loses its rotational energy directly by this outward poynting flux without being mediated by any infalling negative - energy objects , as described in figure [ fig : summary ] .
finally , we should emphasize that our analysis is based on several assumptions ( see sections [ subsec : magnetosphere ] , [ sec : penrose ] , and [ sec : unsteady ] ) .
our arguments for the particle motions near the equatorial plane and in the toy model are required to be justified by numerical simulations . as for the electromagnetic field ,
the principal assumption is @xmath80 in the steady state .
it is still debated whether this condition is satisfied in the steady state at the boundary between the inflow and the outflow ( i.e. at @xmath379 in our toy model ) @xcite and even in the whole bh ( or pulsar ) magnetosphere @xcite .
this issue is closely related to radiation physics , which should be resolved for validating theories on bz process by observations @xcite .
we thank the referee for his / her useful comments . a part of this paper
has already been presented by k.t . in the workshop relativistic jets : creation , dynamics , and internal physics " held at krakow , 20 - 24 april 2015 .
thanks its organizers for the wonderful hospitality and its participants for stimulating him to improve the discussion in this paper . k.t .
also thanks t. harada , s. koide , y. kojima , k. nakao , and y. sekiguchi for useful discussions .
this work is partly supported by jsps grants - in - aid for scientific research 15h05437 and also by jst grant `` building of consortia for the development of human resources in science and technology . ''
99 r.d .
blandford and r.l .
znajek , mon . not .
r. astron .
soc . , 176 , 465 ( 1977 ) .
mckinney , mon . not .
r. astron .
soc . , 368 , 1561 ( 2006 ) .
a. tchekhovskoy , r. narayan , and j.c .
mckinney , mon . not .
r. astron .
, 418 , l79 ( 2011 ) .
komissarov , n. vlahakis , a. knigl , and m.v .
barkov , mon . not .
r. astron .
, 394 , 1182 ( 2009 ) y. lyubarsky , astrophys . j. , 698 , 1570 ( 2009 ) .
k. toma and f. takahara , prog .
phys . , 2013 , 3e02 ( 2013 ) . k. asada , m. nakamura , a. doi , h. nagai , and m. inoue , astrophys . j. , 781 , l2 ( 2014 ) .
m. kino , f. takahara , k. hada , k. akiyama , h. nagai , and b.w .
sohn , astrophys .
j. , 803 , 30 ( 2015 ) .
o. porth and s.s .
komissarov , mon . not .
r. astron .
soc . , 452 , 1089 ( 2015 ) .
beskin and i.v .
kusnetsova , nuovo cimento b , 115 , 795 ( 2000 ) .
s. koide , k. shibata , t. kudoh , and d.l .
meier , science , 295 , 1688 ( 2002 ) .
komissarov , mon . not .
r. astron .
soc . , 359 , 801 ( 2005 ) .
mckinney , a. tchekhovskoy , and r.d .
blandford , mon . not .
r. astron .
soc . , 423 , 3083 ( 2012 ) .
komissarov , mon . not .
r. astron .
soc . , 350 , 427 ( 2004 ) ( k04 ) . m. ruiz , c. palenzuela , f. galeazzi , and c. bona , mon . not .
r. astron .
soc . , 423 , 1300 ( 2012 ) . i. contopoulos , d. kazanas , and d.b .
papadopoulos , astrophys .
j. , 765 , 113 ( 2013 ) .
beskin and a.a .
zheltoukhov , astronomy letters , 39 , 215 ( 2013 ) .
p. goldreich and w.h .
julian , astrophys .
j. , 157 , 869 ( 1969 ) .
k. toma and f. takahara , mon . not .
r. astron .
soc . , 442 , 2855 ( 2014 ) ( tt14 ) .
wald , phys .
d , 10 , 1680 ( 1974 ) .
komissarov , j. korean phys .
, 54 , 2503 ( 2009 ) ( k09 ) .
beskin , phys .
uspekhi , 53 , 1199 ( 2010 ) .
thorne , r.h .
price , and d.a .
macdonald , the membrane paradigm .
yale univ .
press , new haven , ct ( 1986 ) .
penna , phys .
d , 89 , 4057 ( 2014 ) .
penna , r. narayan , and a. sadowski , mon . not .
r. astron .
, 436 , 3741 ( 2013 ) .
znajek , mon . not .
r. astron .
soc . , 179 , 457 ( 1977 ) .
b. punsly and f.v .
coroniti , phys .
d , 40 , 3834 ( 1989 ) .
r. penrose , nuovo .
, 1 , 252 ( 1969 ) . j.m .
bardeen , w.h . press , and s.a .
teukolsky , astrophys .
j. , 178 , 347 ( 1972 ) .
m. bejger , t. piran , m. abramowicz , and f. hakanson , phys .
, 109 , 121101 ( 2012 ) .
lasota , e. gourgoulhon , m. abramowicz , a. tchekhovskoy , and r. narayan , phys .
d , 89 , 024041 ( 2014 ) .
s. koide and t. baba , astrophys .
j. , 792 , 88 ( 2014 ) .
m. takahashi , s. nitta , y. tatematsu , and a. tomimatsu , astrophys .
j. , 363 , 206 ( 1990 ) . g. menon and c.d .
dermer , astrophys .
j. , 635 , 1197 ( 2005 ) .
i. okamoto , publ .
japan , 58 , 1047 ( 2006 ) .
pu , m. nakamura , k. hirotani , y. mizuno , k. wu , and k. asada , astrophys .
j. , 801 , 56 ( 2015 ) j.c .
mckinney and c.f .
gammie , astrophys .
j. , 611 , 977 ( 2004 ) .
k. beckwith , j.f .
hawley , and j.h .
krolik , astrophys .
j. , 678 , 1180 ( 2008 ) .
b. carter , in hawking s. w. , israel w. , eds , general relativity : an einstein centenary survey .
cambridge univ . press , cambridge ,
( 1979 ) .
t. uchida , phys .
e , 56 , 2181 ( 1997 ) .
gralla and t. jacobson , mon . not .
r. astron .
, 445 , 2500 ( 2014 ) .
h. yang , f. zhang , and l. lehner , phys .
d , 91 , 4055 ( 2015 ) .
misner , k.s .
thorne , and j.a .
wheeler , gravitation ( san francisco : freeman ) ( 1973 ) .
b. punsly , black hole gravitohydromagnetics , 2nd edition .
springer , berlin ( 2008 ) .
landau and e.m .
lifshitz , the classical theory of fields , course of theoretical physics , vol .
2 ( new york : pergamon ) ( 1975 ) .
j. kuijpers , h.u .
frey , and l. fletcher , space sci .
, 188 , 3 ( 2015 ) k. toma and f. takahara , astrophys .
j. , 754 , 148 ( 2012 ) .
b. punsly , in contopoulos i. et al . , eds , the formation and disruption of black hole jets . springer , p. 149
( 2015 ) .
broderick and a. tchekhovskoy , astrophys .
j. , 809 , i d . 97
( 2015 ) .
a. levinson and f. rieger , astrophys .
j. , 730 , 123 ( 2011 ) .
m. mocibrodzka , c.f .
gammie , j.c .
dolence , and h. shiokawa , astrophys .
j. , 735 , 9 ( 2011 ) .
kimura , k. toma , and f. takahara , astrophys .
j. , 791 , 100 ( 2014 ) .
kimura , k. murase , and k. toma , astrophys .
j. , 806 , 159 ( 2015 ) .
s. bogovalov and k. tsinganos , astrophys .
j. , 305 , 211 ( 1999 ) . s.s .
komissarov , mon . not .
r. astron .
soc . , 350 , 1431 ( 2004 ) .
beskin and r.r .
rafikov , mon . not .
r. astron .
soc . , 313 , 433 ( 2000 ) s.a .
petrova , mon . not .
r. astron .
soc . , 446 , 2243 ( 2015 ) y. kojima ,
mon . not .
r. astron .
soc . , 454 , 3902 ( 2015 )
the relation between convective current and velocity of particles is summarized as follows .
let us consider the case in which the positively and negatively charged particles have the same velocity @xmath129 as measured in the coordinate basis .
generalization to other cases is easy .
the local physics as measured by the fidos indicates @xmath380 where we note that @xmath381 is a scalar . each spatial component in respect of the bl fido s orthonormal basis
can be rewritten as ( tt14 ) @xmath382 ( the form of the @xmath383 component is the same as the @xmath266 component . )
note that @xmath384 is a four - vector , while the particle velocity is @xmath385 in terms of the four - velocity @xmath386
. then we can write @xmath387 this relation is valid also in the ks coordinates .
equation ( [ eq : relation_j ] ) leads to @xmath388 as an example , the velocity of the @xmath389 drift is ( tt14 ) @xmath390 then we have the drift current as @xmath391 under the assumptions in section [ sec : setup ] , the current is generally @xmath392 for @xmath98 , where @xmath393 is a scalar factor .
this corresponds to the force - free condition , @xmath394 which is equivalent to @xmath395 .
here we explain how equations ( [ eq : b2d2 ] ) , ( [ eq : l_ks ] ) , and ( [ eq : e_ks ] ) are derived , and examine the sign of @xmath396 in the ks coordinates .
the following identities are useful for such calculations : @xmath397 from equations ( [ eq : relation_e ] ) and ( [ eq : omegaf ] ) , generally one has @xmath398 .
\label{eq : d2cal}\ ] ] in the ks coordinates , this equation is reduced to @xmath399 where @xmath400 by using equation ( [ eq : f ] ) and the above identities , we derive equation ( [ eq : b2d2 ] ) . the electromagnetic am density is written by using equations ( [ eq : relation_e ] ) and ( [ eq : omegaf ] ) as @xmath401 .
\label{eq : app_l1}\end{aligned}\ ] ] in the ks coordinates , one has @xmath402 which can be straightforwardly rewritten as equation ( [ eq : l_ks ] ) .
the electromagnetic energy density is written by using equations ( [ eq : relation_e ] ) and ( [ eq : relation_h ] ) as @xmath403 then by using equation ( [ eq : d2cal ] ) , one has @xmath404 which can be straightforwardly rewritten as equations ( [ eq : e_ks ] ) . for the field lines threading the equatorial plane , @xmath405 can be realized near that plane , where @xmath406 , @xmath113 [ which lead to @xmath407 by equation ( [ eq : bphi_ks ] ) ] , and @xmath408 ( tt14 ) .
let us confirm @xmath118 in the ks coordinates .
generally this quantity can be calculated by using equations ( [ eq : app_l1 ] ) and ( [ eq : app_l2 ] ) as @xmath409.\end{aligned}\ ] ] the conditions @xmath406 and @xmath408 lead to @xmath118 .
such a @xmath117 field drives the poloidal current to flow in the direction of @xmath410 . | blandford - znajek process , the steady electromagnetic energy extraction from a rotating black hole ( bh ) , is widely believed to work for driving relativistic jets in active galactic nuclei , gamma - ray bursts and galactic microquasars , although it is still under debate how the poynting flux is causally produced and how the rotational energy of the bh is reduced .
we generically discuss the kerr bh magnetosphere filled with a collisionless plasma screening the electric field along the magnetic field , extending the arguments of komissarov and our previous paper , and propose a new picture for resolving the issues . for the magnetic field lines threading the equatorial plane in the ergosphere
, we find that the inflow of particles with negative energy as measured in the coordinate basis is generated near that plane as a feedback from the poynting flux production , which appears to be a similar process to the mechanical penrose process .
for the field lines threading the event horizon , we first show that the concept of the steady inflow of negative electromagnetic energy is not physically essential , partly because the sign of the electromagnetic energy density depends on the coordinates .
then we build an analytical toy model of a time - dependent process both in the boyer - lindquist and kerr - schild coordinate systems in which the force - free plasma injected continuously is filling a vacuum , and suggest that the structure of the steady outward poynting flux is causally constructed by the displacement current and the cross - field current at the in - going boundary between the plasma and the vacuum . in the steady state , the poynting flux is maintained without any electromagnetic source . |
quark - gluon plasma ( qgp ) is a deconfined state of quarks and gluons , and is currently a subject of much theoretical and experimental research .
qgp is produced in heavy - ion collisions , e.g. au@xmath0au collisions at relativistic heavy - ion collider ( rhic ) or pb@xmath0pb collisions at the large hadron collider ( lhc ) experiments .
some of the signatures of qgp include heavy quarkonium ( @xmath1 or @xmath4 ) suppression , collective flow and photon / dilepton production , etc .
usually , heavy quarkonium is suppressed to a much larger extent , if qgp is formed @xcite , in the heavy - ion collision experiments . however , at the lhc , @xmath1 yield may actually increase due to recombination @xcite , which would obviously reduce the effective suppression .
for proton - lead ( p@xmath0pb ) collisions , the heavy quarkonium yield may be explained using cold nuclear matter ( cnm ) effects itself @xcite .
however , these do not necessarily prove beyond reasonable doubt that qgp is not formed at p@xmath0pb collisions .
if one were to look at peripheral collisions in heavy ion collisions @xcite , it can be seen that most of the suppression mechanisms yield much smaller additional suppression though qgp is present .
the central p@xmath0pb collisions have comparable energy , number of nuclear participants ( @xmath5 ) , and number of binary collisions ( @xmath6 ) as extreme peripheral heavy - ion collisions .
the question then arises as to what would be a defining signature for p@xmath0pb collisions which indicates whether or not the qgp is formed in such a collision . in this work ,
we attempt to explore the yield enhancement of the charmonium state @xmath2 w.r.t .
@xmath1 , at high @xmath3 , as a possible indication of the presence of qgp .
one possible explanation for yield enhancement is secondary recombination of @xmath7 and @xmath8 pair .
however , secondary recombination is unlikely to be a reason for @xmath2 enhancement at high @xmath3 in the case of p@xmath0pb collision .
this is because , both theoretical and experimental data indicate that recombination decreases at high @xmath3 @xcite .
secondly , secondary recombination depends quadratically on the number of @xmath7 and @xmath8 pairs @xcite , which would be very less for p@xmath0pb collision .
we argue that a possible reason for @xmath2 enhancement could be due to the gluon induced excitation of @xmath9 to @xmath2 .
we further argue that , in a medium of equilibrated gluon distribution , this gluon induced excitation increases with @xmath3 of @xmath1 .
this can be understood in the following way .
when the gluon medium achieves equilibrium , it would follow the bose - einstein distribution , which results in the concentration of gluons in the low energy regime .
the gluon density would then decrease exponentially with gluon energy , @xmath10 .
the mass difference between @xmath2 and @xmath9 is about @xmath11 gev , which needs to be provided by the gluon .
but in a bose - einstein distribution , @xmath12 , at a temperature of around @xmath13 mev ( dissociation temperature of @xmath2 ) , most of the gluons would have much smaller energy than @xmath11 gev .
however , for @xmath1 with large @xmath3 , there would be a blue - shift in the gluon energy in the @xmath1 frame of reference , in the forward direction , and a red - shift in the backward direction .
we represent this doppler shift as @xmath14 , where @xmath15 is the relative velocity between the medium and @xmath1 , and @xmath16 is the scattering angle between @xmath15 and incoming gluon momentum .
the subsequent bose - einstein distribution , then becomes @xmath17 .
the modified bose - einstein , @xmath18 , leads to an exponential increase in the availability of gluons with energy around @xmath11 gev , leading to a significant number of @xmath1 getting excited to the @xmath2 state .
the above effect is described in sec .
[ sec : results](see figure [ fig : doppler ] ) .
thus , one would qualitatively expect that gluon induced excitation from @xmath9 to @xmath2 would increase with @xmath3 .
we explore this analytically in this work , and compare the results with preliminary atlas data @xcite at @xmath19 tev .
the suppression effects are also included for the purpose of comparing with experimental data .
the organization of the rest of the paper is as follows . in sec .
[ sec : nrqcd ] , the @xmath20 transition cross - section , @xmath21 , and the excitation rate , @xmath22 , are calculated .
this is followed by sec .
[ sec : results ] , where the results are shown and compared with the experimental data . finally , in sec .
[ sec : conclusion ] , we draw our final conclusions .
in this section , the cross - section @xmath21 and the excitation rate @xmath23 are calculated using the potential non - relativistic qcd ( pnrqcd ) formulation .
the relevant part of the pnrqcd lagrangian used for the calculation is given by @xmath24 the variables , @xmath25 , @xmath26 , @xmath27 and @xmath28 , refer to the singlet , octet , gluon s chromoelectric field and coupling constant respectively .
figure [ fig:1s_2s ] depicts the corresponding feynman s diagram for the process .
state to @xmath29 state.,width=302,height=302 ] [ sec : nrqcd ] the notation and variables used are described as follows .
the center - of - mass coordinates are : * @xmath30 ; * @xmath31 the relative motion coordinates are : * @xmath32 ; * @xmath33 .
we denote the singlet @xmath34 and @xmath29 wavefunctions as @xmath35 and @xmath36 , respectively .
the octet wavefunction = @xmath37 = @xmath38 , with @xmath39 being the energy of the octet state and @xmath40 being the generators of @xmath41 in the adjoint representation .
@xmath42 , @xmath43 , @xmath44 , @xmath45 and @xmath46 refer to the 3-momentum of the incoming @xmath1 , outgoing @xmath2 , octet , incoming and outgoing gluons respectively .
the variables , @xmath47 , @xmath48 and @xmath49 , are the masses of the @xmath1 , @xmath2 and octet states respectively , while , @xmath50 , @xmath51 and @xmath52 refer to the relative momentum between the @xmath53 pair comprising these particles .
@xmath54 is the energy corresponding to @xmath55 , with @xmath56 .
let @xmath57 be the overall color factor and coupling constant , and the propagator be @xmath58 .
then , @xmath59.\end{aligned}\ ] ] the coupling constant , @xmath60 the amplitude @xmath61 would then be given by @xmath62 replacing the summation @xmath63 , by an integral @xmath64 , since the octet energies are continuous , and finally performing all the integrals : @xmath65 where @xmath66 , an arbitrary phase factor @xmath67 , and @xmath68 .
in the @xmath1 rest frame , @xmath69 , and thus @xmath70 with , @xmath71 .
the @xmath20 transition cross - section , @xmath21 , is given by @xmath72\\ \delta^3(p_2 + k_2 - k_1 ) \delta \left ( k_{01 } - \frac{p_2 ^ 2}{2m_{2s } } - k_{02 } - \delta m \right).\end{aligned}\ ] ] in the @xmath1 rest frame , @xmath73 .
we decompose @xmath46 and @xmath43 into parallel and perpendicular components to @xmath45 , and approximating @xmath74 , we get @xmath75 substituting @xmath76 and @xmath77 , with @xmath78 being the angle between @xmath45 and @xmath46 , and evaluating the @xmath79 integral , we finally obtain @xmath80 d\alpha,\end{aligned}\ ] ] with @xmath81 .
one can see that there is a pole when @xmath82 .
this pole is unphysical , and occurs when @xmath45 is large .
in other words , this implies that the above formulation is invalid for very large values of @xmath45 .
the rate constant , @xmath22 , is obtained by integrating the @xmath20 transition cross - section , @xmath21 , with the gluon density function , @xmath83 , where @xmath84 is the number of gluon degrees of freedom = 16 . in other words , @xmath85 @xmath86de_g .
\end{aligned}\ ] ]
we begin by discussing the effect of modified bose - einstein distribution @xmath18 on the gluon density available for exciting @xmath9 to @xmath2 . we depict @xmath87 , w.r.t .
@xmath16 , for various values of @xmath15 in fig .
[ fig : doppler ] . we have plotted @xmath88 , instead of @xmath18 , since the @xmath89 term appears in the integrand of the expression for the rate constant , @xmath90 ( eq . [ eq : rateconstant ] ) . in the blue shifted forward region ,
i.e. , @xmath91 , there is a substantial increase in the gluon density , as @xmath15 increases .
this gives an indication as to why gluo - excitation of @xmath1 to @xmath2 should increase with @xmath3 .
we now discuss the simulation results .
the @xmath9 and @xmath2 wavefunction is obtained by solving the schrdinger equation with the potential @xcite : @xmath92 where @xmath93 is the debye mass given by @xmath94 and @xmath95 is the relative displacement between the @xmath96 pair . here
@xmath97 , number of flavors ; @xmath98 ; @xmath99 gev@xmath100 .
@xmath101 , where we have taken @xmath102 .
this value of alpha , along with @xmath98 , gives the dissociation temperatures close to the values @xmath103 mev for @xmath1 and @xmath13 mev for @xmath2 @xcite .
the octet wave function , @xmath104 , is obtained by solving the schrdinger s equation with potential , @xmath105 @xcite .
we assume the critical temperature , @xmath106 , to be @xmath107 mev , and the charm quark mass to be @xmath108 gev . with these values , we determine @xmath22 .
the centrality and time dependence of the temperature of the qgp medium is taken as @xcite @xmath109 the values for @xmath5 and @xmath110 have been obtained from @xcite , and we have assumed @xmath111 for the most central bin to be 2.5 fm .
figure [ fig : psi_inc ] depicts the fractional increase in @xmath2 yield as a function of @xmath3 .
we see that this increases with @xmath3 .
apart from the transition of @xmath9 to @xmath2 , there would be multiple factors contributing to the @xmath2 yield .
these include cnm effects , suppression effects due to debye color screening , etc .
absorption due to cnm effects can be expected to be negligible at the lhc @xcite .
the other cnm effects , namely cronin effect and shadowing , being initial state effects , would be expected to be the same between @xmath1 and @xmath2 .
hence , we take the yield of @xmath1 as a measure of cnm . for ease of reference ,
the @xmath1 yield in the p@xmath0pb collision is also plotted in figs .
[ fig : psi_yield ] and [ fig : psi_yield_cs ] . in fig .
[ fig : psi_yield ] , we compare the @xmath2 yield with the experimental data .
the @xmath2 yield includes the @xmath9 yield as a measure of cnm .
the regions in the plot , where the experimental yield of @xmath2 is close to one , is a probable indication of suppression effects . to include suppression , we model the mechanism of suppression due to debye color screening .
we follow the formulation outlined in @xcite . in fig .
[ fig : psi_yield_cs ] , the final @xmath2 yield after including the suppression due to color screening is depicted .
the @xmath3 dependence of color screening depicted in fig .
[ fig : psi_yield_cs ] , is after taking into account the effect of the radial medium velocity , @xmath112 .
relative to the medium , the effective momentum of @xmath2 , would be lorentz transformed according to @xmath113 .
this lorentz transformed relative momentum is used for calculation of suppression due to debye color screening .
the medium velocity is taken as @xmath114 based on @xcite . the @xmath2 yield incorporating @xmath9 to @xmath2 transition
is depicted using red curve .
the net yield of the same after additionally including the suppression due to color screening is shown by the green curve in fig .
[ fig : psi_yield_cs ] .
it can be seen here that , with suppression included , the experimental data is better captured .
the results show that the @xmath2 suppression is relatively high at low @xmath3 , while gluo - excitation of @xmath9 to @xmath2 is high at high @xmath3 .
a note on the effect of @xmath1 suppression is due at this point .
we expect the temperature of the qgp , if it is formed , to be not very high .
the @xmath1 dissociation temperature is about @xmath103 mev , which would almost always be above the qgp temperature formed in p@xmath0pb collisions .
thus , @xmath1 will not undergo any suppression .
simulation data ( not shown in this paper ) , indicate the same . at this point
, it is worthwhile to discuss about what happens to a @xmath3 integrated @xmath2 yield .
since most of the @xmath2 production would be expected to be in the low @xmath3 region , the net @xmath2 yield would be dominated by what occurs in the low @xmath3 region , which is suppression . in peripheral p@xmath0pb collisions , where both suppression and gluon induced enhancement is expected to be low , there would be no enhancement or suppression w.r.t .
it is not even clear , whether qgp is formed in peripheral p@xmath0pb collisions .
another question that arises is whether such a phenomenon would be seen in pb@xmath0pb collision . in a pb@xmath0pb collision ,
the temperatures of the medium are much higher , and can go upto @xmath115 or @xmath116 mev . at temperatures above @xmath13 mev , @xmath2 can not exist .
any @xmath2 present would only dissociate .
the gluon induced enhancement to @xmath2 can not happen during this period of the qgp . during the short time , when the temperature of the qgp reduces to below @xmath13 mev , this phenomenon may happen , but may get overshadowed by the suppression phenomenon that has been happening throughout the lifetime of the qgp . however , in extreme peripheral collisions , where the temperatures are much lower during the qgp lifetime , there could be a possibility of gluon induced @xmath2 enhancement at high @xmath3 .
the authors are currently unaware of @xmath3 dependent lhc data , for pb@xmath0pb collisions , in the extreme peripheral region .
in conclusion , we have attempted to explain the transverse momentum dependence of @xmath1 suppression data observed in @xmath117 collision at the lhc energy , over a wide span of transverse momentum .
we have found a differential enhancement of @xmath2 w.r.t .
@xmath9 at higher transverse momentum using gluon induced @xmath34 to @xmath29 transition approach , which qualitatively agrees with the preliminary atlas experimental data .
we have also included the effect of suppression via the color screening mechanism . the combined result seems to corroborate better with experimental atlas data .
we expect that the enhancement of @xmath9 to @xmath2 , as an increasing function of @xmath3 , if confirmed experimentally , can be a more definitive evidence for the presence of qgp in p@xmath0pb collision .
m. mishra is grateful to the department of science and technology ( dst ) , new delhi for financial assistance .
captain r. singh is grateful to bits pilani , pilani for the financial assistance .
99 t. matsui and h. satz , phys .
b * 178 * , 416 ( 1986 ) .
m. c. chu and t. matsui , phys .
d * 37 * , 1851 ( 1988 ) .
m. mishra , c. p. singh , v. j. menon and ritesh kumar dubey , phys .
b * 656 * , 45 ( 2007 ) .
s. ganesh and m. mishra , phys . rev .
c * 88 * , 044908 ( 2013 ) .
s. ganesh and m. mishra , phys .
c * 91 * , 034901 ( 2015 ) . rishi sharma and ivan vitev , phys .
c * 87 * , 044905 ( 2013 ) .
p. k. srivastava , m. mishra and c. p. singh , phys .
c * 87 * , 034903 ( 2013 ) .
s. ganesh and m. mishra , nucl .
a. * 947 * , 38 ( 2016 ) .
a. andronic , p braun - munzinger , k. redlich and j. stachel , j. of phys .
g , * 38 * , 124081 , ( 2011 ) .
yunpeng liu , zhen qu , nu xu and pengfei zhuang , phys .
b * 678 * ( 2009 ) ; arxiv:0907.2723v1 [ nucl - th ] ( 2009 ) .
xingbo zhao and ralf rapp , nucl .
phys . a * 859 * ( 2011 ) .
yunpeng liu , che ming ko and taesoo song , phys .
b , * 728 * , 437 - 442 , ( 2014 ) .
alice collaboration , phys .
b * 734 * , 314 - 327 ( 2014 ) .
thews , aip conf .
* 631 * ( 2002 ) 490 - 524 ; arxiv : hep - ph/02061179v1 ( 2002 ) captain r. singh , p. k. srivastava , s. ganesh , m. mishra , phys .
c * 92 * , 034916 ( 2015 ) .
atlas collaboration , atlas - conf-2015 - 023 ( 2015 ) .
m. laine , o. philipsen , m. tassler , p. romatschke , j. high energy phys .
* 03 * , 054 ( 2007 ) . c. y. wong , phys .
d * 60 * , 114025 ( 1999 ) .
alice collaboration , phys .
c , * 91 * , 064905 ( 2015 ) .
r. vogt , phys .
c * 81 * , 044903 ( 2010 ) ; arxiv:1003.3497v1 [ hep - ph ] ( 2010 ) . | proton - nucleus collisions serve as an important baseline for the understanding and interpretation of the nucleus - nucleus collisions .
these collisions have been employed to characterize the cold nuclear matter effects at sps and rhic energies for the past several years , as it was thought that quark - gluon plasma ( qgp ) is not formed in such collisions . however , at the large hadron collider ( lhc ) , it is not yet clear whether qgp is formed during proton - lead ( p@xmath0pb ) collisions . in this work , we have derived an expression for gluon induced excitation of @xmath1 to @xmath2 , using pnrqcd , and show that the relative enhancement of @xmath2 vis a vis @xmath1 , especially at high @xmath3 , gives an indication that the qgp is indeed formed in p@xmath0pb collisions at the lhc .
0.5 cm _ keywords _ : qgp , @xmath2 enhancement , pnrqcd .
+ _ pacs numbers _ : 12.38.mh , 12.39.pn , 11.25.wx , 11.15.-q |
four - dimensional @xmath0 supersymmetric type ii orientifolds ( @xmath5@xcite and references therein ) provide a class of consistent open - string solutions which in turn could shed light on the physics of strongly coupled heterotic string models . from a more phenomenological viewpoint
, they also provide a natural setup to localized gauge interactions in lower dimensional sub - manifolds of space - time , namely the d - branes , and lead to brane - world models .
unfortunately , the constraints ( arising from ramond - ramond ( rr ) tadpole cancellation conditions ) on four - dimensional models are rather restrictive in the supersymmetric case , and lead to relatively unrealistic gauge sectors and matter contents in the simplest constructions . among the several discrete or continuous deformations of the simplest models that have been explored , often mainly motivated by the search for standard model like theories
, we may recall * blowing - up of orientifold singularities @xcite : the resulting models are not described by a free world - sheet conformal field theory ( cft ) , hence the space - time spectrum can only be computed using field theory techniques . in compact setups ,
the @xmath6 orientifold of @xcite is the only example in which this analysis has been performed @xcite ( it is interesting to point out that results in d - brane models differ in nature from those of perturbative heterotic orbifolds @xcite ) . *
locating the branes at different point in the internal orbifold / orientifold space ( see for instance @xcite ) . in a t - dual picture
these possibilities correspond to the turning of continuous or discrete wilson lines .
explicit examples of wilson lines have been extensively considered , see for instance @xcite for discrete wilson lines , and @xcite for continuous ones . for partially successful attempts at supersymmetric
model building using wilson lines see _ e.g. _ , @xcite .
* introduction of discrete values for the ( neveu - schwarz - neveu - schwarz ) ns - ns @xmath7 field @xcite
. the novel feature of this discrete deformation is that the rank of the gauge group is reduced .
this can also be understood in the t - dual picture as tilting the tori , thereby requiring fewer orientifold planes ( see appendix [ tadpoles ] ) and hence fewer d - branes . *
introduction of gauge fluxes in the d - brane world - volumes ( see @xcite for an early discussion ) . in the supersymmetric context
this has been explored in six - dimensional examples in @xcite . in a t - dual version ,
models with gauge fluxes correspond to type ii orientifolds with d - branes intersecting at angles .
supersymmetric models in six and four dimensions @xcite have been constructed in the situation with d6-branes parallel to the o6-planes in the model .
such constructions however lead to non - chiral models .
( in six dimensions , chiral supersymmetric models with d - branes not parallel with the o - planes have been obtained in @xcite , and implicitly in @xcite ) .
recently it has been realized that the rr tadpoles cancellation conditions turn out to be much less constraining if one gives up the requirement of supersymmetry .
this observation in @xcite was exploited in the context of type iib orientifold model building in _
e.g. _ , @xcite .
it has also allowed for a construction of type iia orientifold and orbifold models with d - branes at supersymmetry breaking angles and realistic theories , with the standard model gauge group ( or simple extensions thereof ) and three quark - lepton families @xcite ( for other proposals for realistic model - building using d - branes , without a specific string construction or in a non - compact set - up , see @xcite ) . despite this remarkable success ,
non - supersymmetric models have more complicated dynamics than supersymmetric ones , hence are less understood .
in particular , tree level flat directions are generically lifted by quantum corrections , leading to involved stabilization problems . also , the models contain uncancelled ns - ns tadpoles which force to redefine the background geometry @xcite , as is obvious , for example , from the existence of non - zero cosmological constant in the models . for these and other reasons ,
it is more reassuring to restrict to supersymmetric model building at the stringy level . however , even for such models , the eventual supersymmetry breaking required in any model attempting to describe realistic low - energy physics , will lead to these or analogous omnipresent issues , like the cosmological constant problem .
the purpose of this paper is explore the construction of four - dimensional @xmath0 supersymmetric type iia orientifolds with d6-branes intersecting at angles , and leading to chiral gauge sectors .
the simplest models satisfying those requirements are orientifolds of toroidal type iia orbifolds @xmath8 or @xmath9 , with d6-branes not parallel to the o6-planes . in this paper
we focus on the @xmath10 orbifold , for which the general pattern of the chiral spectrum is simple enough .
extension to other orientifolds should be more involved , but otherwise straightforward
. we succeeded in constructing the first @xmath0 supersymmetric model with standard model gauge group and three quark - lepton families in this setup .
supersymmetric d3-brane realistic model in section 4.3 in @xcite in a different context . ]
( the letter version that summarizes the key results for this model appeared in @xcite . ) beyond the structure of the minimal supersymmetric standard model ( mssm ) , the model contains some additional gauge factors , right - handed neutrinos , a chiral set of fields with exotic standard model gauge quantum numbers , and diverse vector - like multiplets . despite its lack of fully realistic features
, it provides the first construction of phenomenologically appealing supersymmetric compactifications in the setup of intersecting brane worlds .
moreover , a particularly nice feature of such construction is that supersymmetry avoids the hierarchy problem generically present in the ( otherwise realistic ) models with d6-branes in @xcite .
namely , our models will have a relatively large string scale ( close to the 4d planck scale ) and not very large internal dimensions ( @xmath11 at the largest ) .
we also discuss a number of interesting general results involved in the construction .
we describe the cancellation or previously unnoticed mixed @xmath12-gravitational anomalies , present in some orientifold models ( even without orbifold projection ) .
they are canceled by a green - schwarz mechanism mediated by untwisted rr closed string fields , similar to that in @xcite for mixed non - abelian anomalies .
the models under consideration , in a t - dual version , correspond to chiral supersymmetric versions of the @xmath13 type iib orientifold , with d9-branes with magnetic fluxes .
we show that such models are related to the familiar non - chiral @xmath13 orientifold in @xcite by the t - dual of a four - dimensional version of the small instanton transitions @xcite .
these transitions correspond , in the original picture , to recombinations of d6-branes wrapped on intersecting cycles .
this provides a explicit picture of the transitions , which allows to reproduce interesting phenomena .
in particular , we explicitly construct transitions where a toy model with standard model - like gauge group changes the number of generations , in a manner reminiscent of the chirality changing phase transitions in @xcite . finally
, the models upon consideration are supersymmetric only for specific choices of the untwisted ( complex structure ) moduli .
namely , the condition to preserve supersymmetry @xcite is that the different d6-branes and o6-planes are related by rotations in @xmath14 .
this implies certain constraints on the angles among objects , in the three complex planes in @xmath15 . for
fixed wrapped three - cycles , they imply constraints on the untwisted complex structure moduli , so that supersymmetric solutions exist generically only at isolated points in moduli space .
we briefly discuss the interesting question of the dominant supersymmetry breaking effects upon small departures from the supersymmetric points , and of supersymmetry restoration by vacuum restabilization .
the paper is organized as follows . in section
[ rules ] we describe the construction of @xmath13 orientifolds with d6-branes at angles , and discuss their spectrum , independently of supersymmetry . in section [ cons ] we formulate the conditions to preserve 4d @xmath0 supersymmetry .
we present several explicit examples , including a four - family standard model like theory , and a four - family @xmath16 grand unified theory ( gut ) model .
the construction of three - family models turns out to be very constrained , but we succeed in building a model with sm gauge group ( times additional factors ) and three quark - lepton families ( plus additional exotic and vector - like matter ) .
the requirement of three families demands introducing tilted complex structure in one two - torus , and supersymmetry requires choosing specific ratios for the radii in the remaining two - tori . in section
[ small ] we discuss how our chiral supersymmetric models are related to the familiar non - chiral @xmath13 orientifold @xcite . in a t - dual version
they are connected through small instanton transitions . in the picture of branes at angles the transition corresponds to a recombination of 3-cycles
on which the d6-branes wrap .
such processes can mediate phase transitions changing the number of chiral families , as we illustrate in a toy construction with the standard model gauge group . in section [ ssb ]
we briefly discuss the supersymmetry breaking effect in the open string sector when the ratios of radii in the two - tori are chosen slightly away from the supersymmetry - preserving values . these closed string moduli couple as fayet - iliopoulos ( fi ) terms ,
hence they generate d - term supersymmetry breaking terms whose magnitude is related to a deviation of the untwisted moduli away from their supersymmetric values . finally , in section [ gtwo ] we comment on more formal applications of our constructions , which provide examples whose lift to m - theory corresponds to compact 7-dimensional spaces admitting @xmath3 holonomy metrics , and leading to chiral four - dimensional gauge theories .
section [ conc ] contains our final comments .
in this section we provide the rules to construct consistent orientifolds , and to obtain the spectrum of massless states .
we state these rules independently of supersymmetry , so they are valid for non - supersymmetric model building as well .
our starting point is type iia theory on @xmath1 , with generators @xmath17 , @xmath18 associated to the twists @xmath19 and @xmath20 , hence acting as & : & ( z_1,z_2,z_3 ) ( -z_1,-z_2,z_3 ) + & : & ( z_1,z_2,z_3 ) ( z_1,-z_2,-z_3 ) where @xmath21 are complex coordinates in the @xmath15 . for simplicity , we consider the case of factorizable @xmath15 .
we mod out this theory by the orientifold action @xmath22 , where @xmath23 is world - sheet parity , and @xmath24 acts by r : ( z_1,z_2,z_3 ) ( z_1,z_2,z_3 ) the model contains four kinds of o6-planes , associated to the actions of @xmath22 , @xmath25 , @xmath26 , @xmath27 , as shown in figure [ orient ]
. we will not be specially interested in the closed string sector , which can anyway be computed using standard techniques . for the case of rectangular two - tori , it is as in @xcite , by t - duality . in order to cancel the corresponding rr crosscap tadpoles , we introduce d6-branes wrapped on three - cycles , which we consider factorized , namely obtained as the product of one - cycles in each of the three two - tori . specifically , we consider @xmath28 stacks of @xmath29 d6-branes , @xmath30 , wrapped on the @xmath31 cycle in the @xmath32 two - torus .
we also need to include the images of these under the elements in the orientifold group .
assuming for simplicity that our two - tori are rectangular ( extension to tilted two - tori is easy , and is discussed below ) , we include @xmath29 d6-branes with wrapping numbers @xmath33 .
for branes on top of the o6-planes we also count branes and their images independently . for future convenience , we define the homology class of the corresponding 3-cycles by = _
i=1 ^ 3 ( n_a^i [ a_i ] + m_a^i [ b_i ] ) and analogously for @xmath34 $ ] .
we also define the homology classes of the cycles wrapped by the o6-planes , which for rectangular tori read = [ a_1 ] & ; & [ _ r ] = - [ a_1 ] = - [ b_1 ] & ; & [ _ r ] = - [ b_1 ] [ defosix ] and we define @xmath35=[\pi_{\omega r } ] + [ \pi_{\omega r \theta } ] + [ \pi_{\omega r\omega}]+[\pi_{\omega r\theta\omega}]$ ] .
concerning the orbifold projections , let us focus on the case where the branes pass through fixed points of the orbifold actions , hence @xmath17 and @xmath18 map each stack of branes to itself ( and with the same 3-cycle orientation ) .
extension to other cases is simple , and briefly mentioned in section iiia . to specify the action of the different actions on the chan - paton indices of the branes , for each stack of d6@xmath36-branes , and
their @xmath22 images , denoted d6@xmath37-branes , we introduce the chan - paton actions _ , a & = & ( i _ n_a/2,-i _ n_a/2 ; -i _ n_a/2,i _ n_a/2 ) + + _ , a & = & + + _ r , a & = & the actions for the orbifold group form a projective representation , which corresponds to the choice of closed string sector usually known as without discrete torsion .
the models are constrained by rr tadpole cancellation conditions .
orientifolds by @xmath22 action do not generate twisted crosscaps @xcite , hence the twisted disk tadpoles should vanish . the simplest way to accomplish
this is to choose traceless chan - paton matrices , as done above .
cancellation of untwisted rr tadpoles simply requires the cancellation of d6-brane and o6-plane charge , namely _
a n_a [ _ a ] + _ a n_a [ _ a ] + ( -4)8 [ _ o6 ] = 0 [ homosum ] it is further discussed in appendix a , and leads to the constraints ( [ tadpole1 ] ) .
such choices define a consistent model , for which the resulting spectrum is discussed in the following .
the results for branes at generic angles are shown in table [ matter ] .
let us consider the @xmath38 sector ( strings stretched within a single stack of d6@xmath36-branes ) which is invariant under @xmath17 , @xmath18 , and which is exchanged with @xmath39 by the action of @xmath22 .
for the gauge group , the @xmath17 projection breaks @xmath40 to @xmath41 , and the further @xmath18 projection identifies both factors , leaving @xmath42 . concerning the matter multiplets , we obtain three adjoint @xmath0 chiral multiplets .
this sector is however not @xmath43 since the superpotential for the adjoints @xmath44 , @xmath45 and @xmath46 reads w=(_1 _ 2 _ 3 + _
1_3_2 ) instead of the @xmath43 commutator structure .
this agrees with the result in @xcite in the t - dual ( mirror ) picture . for branes parallel to some o6-plane the projections are identical to @xcite , leading to a @xmath47 group with three @xmath0 chiral multiplets in the two - index antisymmetric representation ( in our ` antisymmetric ' of symplectic factors we also include the singlet )
. the @xmath48 sector , strings stretched between d6@xmath36- and d6@xmath49-branes , is invariant as a whole under the orbifold projections , and is mapped to the @xmath50 sector by @xmath22 . the matter content before any projection would be given by @xmath51 chiral fermions in the bifundamental @xmath52 , where i_ab=[_a]=_i=1 ^ 3 ( n_a^i m_b^i - m_a^i n_b^i ) is the intersection number of the wrapped cycles , and the sign of @xmath51 denotes the chirality of the corresponding fermion ( our default convention is that negative intersection numbers correspond to left - handed fermions ) . for supersymmetric intersections , additional massless scalars
complete the corresponding supermultiplet . for non - supersymmetric intersections ,
the masses for light scalars are as in @xcite . in principle , in performing the orbifold quotient one needs to take into account the orbifold action on the intersection point .
the final result however turns out to be rather insensitive to this subtlety , as opposed to the six - dimensional case @xcite . for an intersection point fixed under @xmath17 and @xmath18 ,
the orbifold projections reduce the matter content to a bifundamental @xmath53{6.5pt}{0.4pt}}\rule[6.5pt]{0.4pt}{0.4pt}\hskip-0.4pt\rule{0.4pt}{6.5pt}}}}_a,{\overline{{\raisebox{-.5pt}{{\hbox{\rule{0.4pt}{6.5pt}\hskip-0.4pt\rule{6.5pt}{0.4pt}\hskip-6.5pt\rule[6.5pt]{6.5pt}{0.4pt}}\rule[6.5pt]{0.4pt}{0.4pt}\hskip-0.4pt\rule{0.4pt}{6.5pt}}}}}}_b)$ ] of @xmath54 . for intersection points exchanged by the orbifold actions , for instance two points fixed under @xmath18 but exchanged under @xmath17
, we should consider just one point ( the other being merely its @xmath17-image ) and not impose the @xmath17 projection .
the resulting fields are _ two _ bifundamentals . due to this compensation ,
the total number of fields in the @xmath55 sector is simply @xmath51 chiral fermions in the @xmath53{6.5pt}{0.4pt}}\rule[6.5pt]{0.4pt}{0.4pt}\hskip-0.4pt\rule{0.4pt}{6.5pt}}}}_a,{\overline{{\raisebox{-.5pt}{{\hbox{\rule{0.4pt}{6.5pt}\hskip-0.4pt\rule{6.5pt}{0.4pt}\hskip-6.5pt\rule[6.5pt]{6.5pt}{0.4pt}}\rule[6.5pt]{0.4pt}{0.4pt}\hskip-0.4pt\rule{0.4pt}{6.5pt}}}}}}_b)$ ] of @xmath56 ( plus scalars , which fill out supermultiplets in the supersymmetric case ) .
a similar effect takes place in @xmath57 sector , for @xmath58 , where the final matter content is given by @xmath59 chiral fermions in the bifundamental @xmath53{6.5pt}{0.4pt}}\rule[6.5pt]{0.4pt}{0.4pt}\hskip-0.4pt\rule{0.4pt}{6.5pt}}}}_a,{\raisebox{-.5pt}{{\hbox{\rule{0.4pt}{6.5pt}\hskip-0.4pt\rule{6.5pt}{0.4pt}\hskip-6.5pt\rule[6.5pt]{6.5pt}{0.4pt}}\rule[6.5pt]{0.4pt}{0.4pt}\hskip-0.4pt\rule{0.4pt}{6.5pt}}}}_b)$ ] .
finally , let us consider the @xmath60 sector . in this case
, the orbifold action on the intersection point turns out to be crucial . at an intersection point with angle @xmath61 in the @xmath32 two - torus states
are labeled , in the bosonized formulation ( see @xcite ) , by a vector @xmath62 , where @xmath63 in the ns , r sectors , respectively , and @xmath64 due to the gso projection , and @xmath65 corresponds to respective left - handed and right - handed fermions ( in our default convention ) . for an intersection point invariant under @xmath17 and @xmath18 , the eigenvalues of such state under @xmath17 , @xmath18 and @xmath22 are @xmath66 , @xmath67 , and @xmath68 , where recall @xmath69 , @xmath70 are the twist vectors .
the projections on the chan - paton factors are & = & e^2i rv _ , 6 _ , 6^-1 + & = & e^2i rw _ , 6 _ , 6^-1 + & = & - _
r^-1 before the orientifold projection , one gets a chiral fermion in the bifundamental @xmath71 , regardless of the @xmath17 , @xmath18 eigenvalues of the state .
the orientifold projection , however , distinguishes the different cases and leads to a two - index antisymmetric representation of @xmath42 , except for states with @xmath17 and @xmath18 eigenvalue @xmath72 , where it yields a two - index symmetric representation .
now consider points not fixed under some orbifold element , say two points fixed under @xmath18 , and exchanged by @xmath17 .
then one simply keeps one point , and does not impose the @xmath18 projection .
equivalently , one considers all possible eigenvalues for @xmath18 , and applies the above rule to read off whether the symmetric or the antisymmetric representation survives .
it is possible to give a closed formula for the precise chiral matter content in this sector , which basically follows from cancellation of non - abelian anomalies .
the final result for the net number of symmetric and antisymmetric representations in the @xmath73 sector is n_-0.4pt & = & -12 ( i_aa - i_a , o6 ) + n_-6.9pt & = & -12 ( i_aa + i_a , o6 ) [ aaprime ] where @xmath74 is the number of tilted tori and @xmath75\cdot [ \pi_a']$ ] , @xmath76\cdot[\pi_{o6}]$ ] .
notice that the definitions ( [ defosix ] ) should be modified in the obvious way for tilted tori .
the result ( [ aaprime ] ) is then general . )
implies that @xmath77 .
the first two members , for @xmath78 give ( minus two times ) the @xmath79 anomaly due to @xmath55 and @xmath80 fields .
the rest , which equals @xmath81 must correspond to ( minus two times ) the anomaly in the @xmath73 sector , which is given by @xmath82{6.5pt}{0.4pt}}\rule[6.5pt]{0.4pt}{0.4pt}\hskip-0.4pt\rule{0.4pt}{6.5pt}}}\hskip-0.4pt \raisebox{-.5pt}{{\hbox{\rule{0.4pt}{6.5pt}\hskip-0.4pt\rule{6.5pt}{0.4pt}\hskip-6.5pt\rule[6.5pt]{6.5pt}{0.4pt}}\rule[6.5pt]{0.4pt}{0.4pt}\hskip-0.4pt\rule{0.4pt}{6.5pt } } } } } ( n_a/2 + 4 ) + n_{{\raisebox{-3.5pt}{{\hbox{\rule{0.4pt}{6.5pt}\hskip-0.4pt\rule{6.5pt}{0.4pt}\hskip-6.5pt\rule[6.5pt]{6.5pt}{0.4pt}}\rule[6.5pt]{0.4pt}{0.4pt}\hskip-0.4pt\rule{0.4pt}{6.5pt}}}\hskip-6.9pt \raisebox{3pt}{{\hbox{\rule{0.4pt}{6.5pt}\hskip-0.4pt\rule{6.5pt}{0.4pt}\hskip-6.5pt\rule[6.5pt]{6.5pt}{0.4pt}}\rule[6.5pt]{0.4pt}{0.4pt}\hskip-0.4pt\rule{0.4pt}{6.5pt } } } } } ( n_a/2 - 4))$ ] . equating both for arbitrary @xmath29 yields ( [ aaprime ] ) . ] clearly , getting the non - chiral piece requires the full computation of the spectrum .
.general spectrum on d6-branes at generic angles ( namely , not parallel to any o6-plane in all three tori ) .
the spectrum is valid for tilted tori .
the models may contain additional non - chiral pieces in the @xmath73 sector and in @xmath55 , @xmath80 sectors with zero intersection , if the relevant branes overlap .
in supersymmetric situations , scalars combine with the fermions given above to form chiral supermultiplets .
[ matter ] [ cols="^,>",options="header " , ] let us consider the @xmath13 models , with rectangular two - tori , described in table [ transit ] . consider the initial model , where d6-brane wrapping numbers are given by the first six rows in the table . in the t - dual picture the model is understood in terms of instanto bundles as follows .
consider the model without magnetic fluxes in @xcite , with group @xmath83 , the four factors arising from d9- and d5@xmath84-branes .
consider dissolving 8 d5@xmath85- and 4 d5@xmath86-branes ( and images ) as instantons within a @xmath87 sub - group of @xmath88 . in the decomposition usp(16)_9 u(4)u(2)usp(4 )
[ split ] the two @xmath12 generators which acquire non - zero flux correspond to the @xmath12 s within @xmath89 and @xmath90 , respectively .
the surviving group from the d9-branes is the commutant of the gauge background , namely @xmath91 .
the final spectrum can be computed directly in string theory in the flux or angle picture , or in field theory using the index theorem ( see @xcite for a discussion ) .
the chiral spectrum is & `` u(4)''u(2 ) usp(12)usp(4)usp(8 ) usp(16 ) & + & 2(4,2;1,1,1,1 ) + ( 4,1;12,1,1,1 ) + ( 4,1;1,1,1,16 ) + & + & ( 1,2;1,1,8,1 ) + ( 1,2;1,1,1,16 ) & where the symplectic groups arise from ( a ) d9-branes without flux , and ( b ) d5@xmath84-branes not dissolved as fat instantons .
the quotation marks for @xmath89 denote that we are actually interested in splitting it as @xmath92 ( by separating the d6-branes , or equivalently by introducing d9-brane wilson lines ) .
the resulting model corresponds to a two - family standard - like model .
one can even obtain a sensible hyper - charge by splitting the @xmath93 symplectic factors as @xmath94 so as to generate additional @xmath12 s .
in fact the linear combination q_y = 16 q_3 -12 q_1 + 12 ( q_12 + q_4 + q_8 ) [ hyp1 ] ( where @xmath95 , @xmath96 arise from @xmath97 , and @xmath98 , @xmath99 , @xmath100 arise as diagonal combinations of the @xmath12 s from symplectic factors ) is automatically anomaly - free and massless , and plays the role of hyper - charge in the above model .
the net chiral content with respect to standard model interactions is 2(3,2)_1/6 + 2(3,1)_1/3 + 2(3,1)_-2/3 + 2(1,2)_-1/2 + 2(1,1)_1 + 2(1,1)_0 namely two standard quark - lepton generation ( plus right handed neutrino ) . in the angle picture ,
let us consider the 4 d6-branes along @xmath101 ( and their @xmath22 images ) combine with the 12 along @xmath102 ( and images ) to give 4 d6-branes along @xmath103 ( and images ) .
this process is possible since the total homology class is conserved , and is triggered by vev s for scalars in strings stretching between the stacks involved .
the final model has d6-branes with wrapping numbers given by the first four and the last rows in table [ transit ] .
let us describe the process in the t - dual picture .
the final model corresponds , in terms of the underlying @xmath83 , to the following .
consider dissolving 8 d5@xmath85- and 16 d5@xmath86-branes ( and images ) as instantons in a @xmath104 sub - group corresponding to the splitting ( [ split ] ) .
the resulting flux structure corresponds to the final model .
hence the transition connecting both models amounts to dissolving 12 additional d5@xmath86-branes ( and images ) as instantons within the same @xmath12 sub - group where the previous 8 were already dissolved .
in particular this shows that the unbroken d9-brane gauge group has identical generators in both cases .
however the chiral fermion content may differ , due to the presence of additional flux modifying the index of the dirac operator in the internal space ( or the intersection number in the angle picture ) .
the final spectrum can be computed with string theory or index theory techniques , and reads & `` u(4)''u(2 ) usp(4)usp(8 ) usp(16 ) & + & 6(4,2;1,1,1,1 ) + 2(4,2;1,1,1,1 ) + ( 4,1;1,1,16 ) + & + & 2(1,2;1,1,8,1 ) + 2(1,2;1,1,1,16 ) & upon splitting @xmath105 the model corresponds to an eight - family standard model . in this case hyper - charge can be obtained by splitting the symplectic factors into @xmath12 s and considering the anomaly - free and massless combination q_y = 16 q_3 -12 q_1 + 12 ( q_4 + q_8 ) the chiral spectrum with respect to standard model gauge interactions is 8(3,2)_1/6 + 8(3,1)_1/3 + 8(3,1)_-2/3 + 8(1,2)_-1/2 + 8(1,1)_1 + 8(1,1)_0
namely , there are eight standard quark - lepton families .
hence the process of dissolving additional d5-branes in the gauge bundle on the d9-branes leads to phase transitions changing the number of chiral families .
a minor difficulty in the above transition is that the hyper - charge generator is not the same in both theories .
this can be avoided by removing the @xmath98 term in ( [ hyp1 ] ) , although this yields to exotic hyper - charges in the initial model . leaving these subtle points aside
, we would like to emphasize that it is remarkable that one can describe quite explicitly these transition in relatively realistic models .
we have succeeded in showing that small instanton transitions can mediate changes in the number of families in a model . moreover
, the t - dual interpretation of these processes as recombination of cycles provides a useful tool in analyzing these transitions in the type ii orientifold setup .
we believe these techniques can be useful in the study interesting phenomena in a simple geometric setup , and hence are complementary to other realizations of these transitions @xcite .
as mentioned above , the general models we have considered are supersymmetric for specific choices of the untwisted moduli @xmath106 . in this section
we briefly consider the main supersymmetry breaking effects on the open string sector as one moves away from the special supersymmetric values for @xmath106 . on general grounds ,
the complex structure moduli @xmath106 are expected to couple open string modes in our d6-branes on 3-cycles ( a - branes ) as fayet - iliopoulos ( fi ) terms .
this is mirror to the statement that khler moduli couple to b - branes as fi - terms , a familiar situation for d - branes at singularities @xcite .
( note that related techniques have been employed in the blowing - up procedure of the type iib orientifold singularities @xcite ) .
this has appeared in a related context in @xcite , and in situations where there are d - branes with ( @xmath107 or equivalently magnetic field ) fluxes in @xcite . following the latter ,
we expect the corresponding fi - terms to be proportional to the deviation from the supersymmetric situation , in particular we expect the terms in the effective @xmath108 action _ a d^4x ( ^1_a + ^2_a + ^3_a ) d_a [ fi ] where @xmath109 is understood as a function of @xmath106 for fixed wrapping numbers @xmath31 . for instance , for square tori _ a^i = _ i. the fi - term vanishes for the supersymmetric situation @xmath110 , hence in general it is proportional to the deviation from this case .
it is easy to see that this term reproduces the leading order splitting between scalar and fermion masses , as one would obtain from the string computation .
namely , in @xmath55 sector , chiral fermions remain massless at tree level , while their scalar partners obtain a mass proportional to @xmath111 . in supersymmetric models , the familiar arguments in @xcite
relate the existence of green - schwarz anomaly cancellation mechanism with the existence of fi terms , controlled by the partners of the fields mediating the gs interactions .
their precise determination also requires knowledge about their khler potential , which should be easily determined for the untwisted complex structure moduli in our models .
however , we skip the derivation of the fi terms , and briefly discuss the physics arising from eq.([fi ] ) .
the turning of fi terms when untwisted moduli are shifted from the supersymmetric values actually does not automatically imply breaking of supersymmetry . as is familiar in heterotic constructions
, some scalar fields may acquire vev s so as to make the d - term vanish .
hence , supersymmetry would be restored in the shifted vacuum .
an important difference with respect to heterotic models is that the fi - terms are not related to the dilaton , and can be tuned at will by tuning the untwisted moduli .
the physics behind this process is that the original d6-brane configuration is no longer supersymmetric for the new choice of untwisted moduli ( since the angles are changed ) , so some intersecting d6-branes recombine into a smooth 3-cycle which is supersymmetric .
this recombination is described by the vev acquired by certain scalar fields at intersections .
it is an interesting question whether supersymmetry can always be restored in this fashion . despite the lack of a general argument
, we strongly suspect that this is the case , at least in compact models .
notice that however non - compact models allow for supersymmetry breaking by this fi - term mechanism ( see @xcite for discussion ) . in order to provide a simple illustrative example ,
let us consider the string - gut model in section [ gut ] , where there is only one relevant set of angles , namely those formed by the @xmath112 branes with the horizontal axes , denoted by @xmath61 henceforth .
there are two relevant kinds of scalars , those arising at the intersections between the @xmath112 branes and their images , @xmath113 , and between the former and the @xmath114 branes , @xmath115 .
the corresponding chiral multiplets carry opposite charges with respect to the single @xmath12 symmetry in the model , hence the d - term has the schematic structure d = 2 |_aa|^2 - |_ab|^2 + ( _ 1+_2+_3 ) hence , for deformations such that @xmath116 , the fields @xmath115 become tachyonic and acquire a vev , restoring supersymmetry .
this corresponds to recombining the @xmath114 and the @xmath112 branes . for @xmath117
, it is @xmath113 which acquire vev s to restore supersymmetry , triggering the recombination of the @xmath112 branes and their images
. it would be interesting to explore these processes in more detail , both in their effective field theory and in their geometric description .
in this section we briefly outline a different ( more formal ) aspect of our models .
recently there has been a lot of interest in the study of the dynamics of m - theory on 7-dimensional manifolds @xmath118 admitting a @xmath3 holonomy metrics @xcite .
the interest stems from the fact that such compactifications lead to four - dimensional @xmath0 supersymmetric field theories , with gauge interactions determined by the singularity structure of @xmath118 .
moreover , such constructions have provided a geometric interpretation @xcite of the duality between type iia configurations with d6-branes on the special lagrangian 3-cycle in the deformed conifold , and type iia on the resolved conifold , without d6-branes but with rr 2-form fluxes @xcite .
these results have been extended in diverse directions ( see , _
e.g. _ , @xcite ) , and suggest interesting connections with gauge theory dynamics and string duality .
topological manifolds admitting a @xmath3 metric are not easy to characterize , as opposed to , _
e.g. _ , spaces admitting a @xmath119 holonomy metric , which can be characterized by a topological condition ( the vanishing of the first chern class ) .
the explicit construction of @xmath3 metrics is difficult , and has only been achieved in a few non - compact examples constructed in @xcite and the more recent generalizations in @xcite .
( for applications to regular configurations of m - theory with @xmath0 supersymmery , see _ e.g. _ , @xcite and references therein . )
however , string theory duality provides a simple strategy to obtain topological spaces which admit a @xmath3 metric , without constructing it explicitly .
basically , any type iia configuration preserving @xmath108 @xmath0 supersymmetry , and including at most d6-branes and o6-planes , will lift to an m - theory compactification on a @xmath3 manifold ( see @xcite for nice discussions ) .
the topological information can be used to obtain interesting qualitative features of these theories . in this respect , the @xmath108 @xmath0
supersymmetric type iia orientifolds with d6-branes and o6-planes at angles , studied in @xcite and in this paper , correspond to m - theory compactifications on @xmath3 manifolds . in configurations where the rr 7-form charges are locally cancelled ( namely , 2 d6-branes and 2 images on top of each o6-plane in the configuration ) , the m - theory lift is remarkably simple .
the m - theory circle is constant over the base space @xmath120 , leading to a total variety @xmath121 , where the @xmath122 flips the coordinate parametrizing the m - theory circle , and acts on @xmath120 as an antiholomorphic involution ( hence changing the holomorphic 3-form to its conjugate ) .
this is analogous to the discussion in @xcite .
unfortunately , such models lead generically to non - chiral spectra , in the sense that even though one obtains chiral multiplets , they arise in real representations of the gauge group . on the other hand ,
configurations with d6-branes away or not fully aligned with the o6-planes would led to more involved m - theory lifts .
however , they are of great interest since , as shown in our general constructions , they lead to chiral gauge theories . certainly , there exist simpler lines of attack to obtain chirality out of @xmath3 singular spaces . in particular , the simplest type iia supersymmetric configuration leading to chiral fermions is simply two intersecting d6-branes in flat space , related by an @xmath14 rotation .
its m - theory lift would correspond to a rigid 7-dimensional @xmath3 singularity , and is the basic building block for engineering chiral theories using @xmath3 geometries ; the chiral multiplet arises from m2-brane wrapping a collapsed two - cycle . in this sense , our configurations are more complicated , and what they provide is a consistent embedding of this building block singularity in a compact setup .
we expect that the generic class of models described here may exhibit some interesting phenomena in this context , in particular the existence of non - perturbative equivalences among seemingly different models , which nonetheless share the same m - theory lift , in analogy with @xcite .
on the other hand the type iia transitions in which intersecting d6-branes recombine ( the t - dual of the small instanton transitions ) would have interesting m - theory descriptions , in which the topology of the @xmath3 space changes . it would be interesting to explore possible connections of such process with @xcite .
we hope that our explicit constructions may provide a useful laboratory to probe new ideas in this exciting development .
in this paper , we constructed four - dimensional @xmath0 supersymmetric type ii orientifolds with branes at angles .
we provided the first d - brane construction of a three - family @xmath0 supersymmetric vacuum solution with the standard model gauge group @xmath123 as part of the gauge structure .
we have also illustrated other possibilities for model building by constructing a supersymmetric gut model with ( four ) chiral families and adjoint higgs multiplets , the first example of its kind in the orientifold setup .
although we have not discussed here , it is quite straightforward to construct other extensions of the standard model , such as the left - right symmetric models ( _ e.g. _ , the pati - salam type ) . even though our models are explicit string realizations of the brane - world scenario , they generically require a high string scale . in models with intersecting d6-branes ,
the experimental bounds on masses of kaluza - klein replicas of standard model gauge bosons imply that the internal dimension can not be large ( since there is no dimension transverse to _ all _ standard model branes ) .
hence , a large planck mass can be generated only from a large string scale , and not from a large volume .
specifically , one obtains g_ym^2 m_p^(4d ) = m_s where @xmath124 is the volume of the cycle wrapped by the corresponding brane , and @xmath125 is the total internal volume . moreover
, large anisotropies in the internal space would generically reflect in different gauge couplings for different gauge group factors . for nearly isotropic configurations ,
the string scale is of the order of the planck scale .
there is however more freedom than in the traditional heterotic approach , and it could be used to lower the string scale to , _
e.g. _ , @xmath126 gev , a certainly desirable choice for gut models . another interesting feature of this class of models with branes at angles is the structure of the yukawa couplings . since the quarks , leptons and higgs fields are located at different intersections of the branes , the yukawa couplings @xmath127 are generically exponentially suppressed by the area @xmath128 of the string world - sheet stretching between the locations of these fields ( measured in string units ) @xcite , _
i.e. _ , @xmath129 these exponential factors may provide an interesting geometrical explanation for the observed fermion masses . in order for the yukawa couplings not to be negligibly small
, the area of the string world - sheet ( which is typically the compactification scale ) can not be much larger than the string length , so internal dimensions should be of the order of the string scale .
note that in addition to the above yukawa couplings , there are generically couplings among @xmath38 , @xmath55 and @xmath130 fields , and among @xmath38 fields .
these couplings are not exponentially suppressed .
therefore , in studying the resulting phenomenology , one should examine whether they may pose phenomenologically challenges to this general class of models involving branes at angles .
hopefully , they might be useful , _ e.g. _ , to get rid of the unwanted non - chiral matter in the @xmath38 sector , even though we have no concrete proposal in this respect .
we also note in passing that unlike the recent models in @xcite , our models have absolute proton stability due to symmetries ( as discussed in @xcite ) . the basic model building rules that we have constructed , allow for the exploration of a potentially large class of supersymmetric models .
in particular , to simplify the conditions from supersymmetry , we have mainly restricted our search to d6-branes at angles of the form @xmath131 , @xmath132 or @xmath133 . it is quite remarkable that within this restricted class of models , there exist three - family standard - like models .
the three - family model is however not fully realistic , as it contains extra vector - like multiplets as well as exotic chiral matter .
nevertheless , it is possible that variants of this ( or other ) model(s ) may eliminate these additional states , and lead to solutions closer to the standard model ( in particular , it would be interesting to reproduce the very economical spectrum structure in @xcite ) .
clearly , a detailed search of realistic models deserves further investigation . in this regard ,
it is interesting to note that supersymmetric d6-branes with three non - trivial angles , say @xmath134 , contribute to some tadpole conditions ( but not all ! ) with the same sign as that of the o6-planes .
this implies that even though the configuration is supersymmetric , the branes can carry negative rr charges under some rr forms in the dimensional reduction on the internal space .
this allows for more flexibility in satisfying the tadpole conditions , and may give more room for embedding of the standard model .
the general class of models with branes at angles are connected to familiar orientifolds by ( the t - dual of ) small instanton transitions .
for instance , the three - family model we presented is connected by such a transition to the non - chiral @xmath135 orientifold in @xcite . in the picture of d6-branes at angles
, the transition amounts to a recombination of the intersecting cycles , resulting in d6-branes not fully aligned with the o6-planes , and leading to chiral matter .
the recombination description provides a simple setup to analyze these transitions , and their field theory interpretation in detail .
we have also discussed the main physical effects when the untwisted moduli are chosen away from the supersymmetric point .
the model develops fi terms proportional to the deviation from the supersymmetric situation .
the corresponding non - zero d - terms in general force some scalars to acquire vev s and restore supersymmetry at the restabilized vacuum .
this is the field theory counterpart of a process in which the intersecting d6-branes , non - supersymmetric in the new situation , recombine and wrap a new 3-cycle which is supersymmetric in the new , deformed , complex structure . as this general class of supersymmetric orientifold models
involve only d6-branes and o6-planes , their m - theory lift correspond to compactifications on purely geometrical background , in fact , a compact , singular 7-manifold with @xmath3 holonomy .
given the recent interest in m - theory compactifications on such spaces , we expect our general class of orientifold models may lead to new insights into the construction of spaces with special holonomy leading to four - dimensional gauge theories with chiral fermions , and into new physical phenomena in such compactifications .
there are many promising avenues to explore in supersymmetric orientifolds with d6-branes at angles , both from the phenomenological and the theoretical viewpoints .
we hope our results here have provided the first steps in some of these directions .
we thank gerardo aldazabal , savas dimopoulos , jens erler , gary gibbons , jaume gomis , luis ibez , paul langacker , hong l , chris pope , raul rabadn and edward witten for useful discussions .
g.s . thanks ralph blumenhagen and boris krs for email correspondence , and matt strassler for sharing his insights on generation - changing transitions .
we would like to thank the theory division at cern ( m.c . and g.s .
) , camtp , university of maribor , slovenia ( m.c . )
amsterdam workshop on string theory ( m.c . ) and benasque workshop ( m.c . and
a.m.u . ) for hospitality during the course of the work
thanks m. gonzlez for kind encouragement and support .
this work was supported in part by u.s .
department of energy grant no .
doe - ey-76 - 02 - 3071 ( m.c . ) , in part by the class of 1965 endowed term chair ( m.c . ) , upenn sas dean s funds ( g.s . ) and the nato linkage grant 97061 ( m.c . ) .
in this appendix we derive the tadpole cancellation conditions for the @xmath10 orientifold model of the type discussed in the main text .
the results can be generalized in a rather straightforward way to other types of @xmath136 orientifold models . as obtained in @xcite
, @xmath22 orientifolds of type iia toroidal orbifolds do not contain twisted crosscap tadpoles .
correspondingly , d - branes wrapped on factorized three - cycles on the six - torus ( i.e. wrapped on one - cycles in each complex plane ) do not generate twisted disk tadpoles .
therefore , only cancellation of untwisted rr tadpoles should be imposed , namely , the consistency conditions are the same as for a set of o6-planes and d6-branes ( and their images ) in the six - torus .
this is simply gauss law , the cancellation of the total charge under the rr 7-form .
such charges are proportional to the homology class of the wrapped three - cycles .
let us consider the case of rectangular two - tori , and denote @xmath137 $ ] , @xmath138 $ ] the @xmath139 and @xmath140 homology one - cycles in the @xmath32 two - torus .
the o6-planes fixed under the orientifold actions @xmath22 , @xmath25 , @xmath26 , @xmath27 , shown in figure [ orient ] , carry an overall charge proportional to -48 ( [ a_1 ] - [ b_1 ] - [ a_1 ] - [ b_1 ] ) [ chor ] where the @xmath141 is the charge of a single o6-plane ( in d6-brane charge units , and as counted in the covering space ) , and 8 is the number of o6-planes of each kind .
= 3.5 cm we would like to cancel this charge by introducing sets of @xmath29 d6-branes wrapped on the three - cycle defined by wrapping numbers @xmath31 , and their orientifold images , with wrapping numbers @xmath33 .
the total d6-brane charge is _ a n_a _
i=1 ^ 3 ( n_a^i [ a_i ] + m_a^i [ b_i ] ) + _ a n_a _
i=1 ^ 3 ( n_a^i [ a_i ] - m_a^i [ b_i ] ) [ chdb ] the requirement that charges ( [ chor ] ) , ( [ chdb ] ) add up to zero yields the rr tadpole constraints _ a n_a n_a^1 n_a^2 n_a^3 - 16 & = & 0 + _ a n_a n_a^1 m_a^2 m_a^3 + 16 & = & 0 + _ a n_a m_a^1 n_a^2 m_a^3 + 16 & = & 0 + _ a n_a m_a^1 m_a^2 n_a^3 + 16 & = & 0 [ tadpole1 ] the procedure carries over in the same spirit for the case with some tilted tori .
for instance , consider only the third torus is tilted , = 3.5 cm the total o6-plane rr charge is -4 4 \ { [ a_1](2[a_3]-[b_3 ] ) & - & [ b_1](2[a_3]-[b_3 ] ) + & - & [ a_1]- [ b_a ] } where we have taken into account that due to the tilt there are only four o6-planes of each kind , see figure [ tiltorient ]
. the charge of sets of @xmath29 branes with wrapping numbers @xmath31 , and their images , with wrapping numbers @xmath142 are _ a n_a _
i=1 ^ 3 ( n_a^i [ a_i ] + m_a^i [ b_i ] ) + _ a n_a _
i=1 ^ 3 ( n_a^i [ a_i ] - ( m_a^i+n_a^i ) [ b_i ] ) cancellation of rr charges leads to _ a n_a n_a^1 n_a^2 n_a^3 - 16 & = & 0 + _ a n_a n_a^1 m_a^2 m_a^3 + 8 & = & 0 + _ a n_a m_a^1 n_a^2 m_a^3 + 8 & = & 0 + _ a n_a m_a^1 m_a^2 n_a^3 + 16 & = & 0 [ tadpole2 ] where we have defined @xmath143 .
the general rule is to modify tadpole conditions involving the @xmath144 if the @xmath32 two - torus is tilted .
the modification is simply to replace @xmath145 , and cut by half the corresponding crosscap contribution .
we first consider the case of @xmath22 orientifolds of type iia on @xmath15 @xcite , and turn to the case with additional @xmath13 orbifold projections towards the end .
consider the @xmath22 orientifold of type iia on @xmath15 .
for simplicity we center on a six - torus factorizable into three rectangular two - tori .
other cases are worked out analogously . in order to cancel the tadpoles
, we introduce sets of @xmath29 d6-branes wrapped on 3-cycles in the homology class @xmath146 $ ] , defined by the wrapping numbers @xmath31 .
we also introduce their @xmath22 images , namely @xmath29 branes on cycles @xmath34 $ ] , defined by wrapping numbers @xmath33 . for convenience , we define @xmath35 $ ] the homology class associated to the 3-cycles with wrapping numbers @xmath139 along each two - torus
. the rr tadpole cancellation conditions amount to cancellation of rr charge in homology , namely _ a n_a [ _ a ] + _ a
n_a [ _ a ] - 32 [ _ o6 ] = 0 the spectrum of chiral fermions is obtained from @xcite , and for branes on generic cycles reads as discussed in @xcite in the absence of the orientifold projections , the mixed @xmath148-@xmath149 anomaly does not vanish in general , rather it is proportional to a_ab = 12 n_a [ _ a ] ( [ _ b]+ [ _ b ] ) in the toroidal models in @xcite the mixed @xmath148-gravitational anomalies vanishes automatically .
this is not true in general for @xmath150 orientifolded models , where the anomaly is proportional to a_a^grav = 24 n_a [ _ a]hence , for any model with branes intersecting with the o6-plane the anomaly does not vanish , which is the case for the examples in @xcite ( notice that in the orientifold models in @xcite gravitational anomalies vanish due to the specific choice of d6-branes , which never intersect the o6-planes ) .
these anomalies are canceled by a green - schwarz mechanism mediated by untwisted rr fields , as discussed in @xcite for mixed non - abelian anomalies .
our treatment of the gravitational anomalies below is novel . expanding the chern - simons couplings for the d6-branes ( and the images ) and o6-planes ( see _ e.g. _ , @xcite ) _ d6 c e^f ; _ o6 c we obtain the following relevant interactions & 12 _ d6_a c_3(f_a f_a ) & ; _ d6_a c_5 f_a + & - _ d6_a c_3(r r ) & ; ( -4)12 _ o6 c_3(r r ) [ couplsix ] operating as in @xcite , we introduce two dual basis of homology 3-cycles , @xmath151 \}$ ] , @xmath152 \}$ ] , satisfying @xmath153 \cdot [ \sigma_j ] = \delta_{ij}$ ] , and introduce the expansions & [ _ a ] = _ i r_ai [ _ i ] & ; = _ i p_ai [ _ i ] + & [ _ o6 ] = _ i r_i [ _ i ] & ; = _ i p_i [ _ i ] we define the untwisted rr fields @xmath154 } c_3 \quad ; \quad b_2^i = \int_{[\sigma_i ] } c_5 $ ] , hodge duals in four dimensions . the four - dimensional couplings read & 12 _ i p_ai _ m_4 _ i ( f_a f_a ) & ; n_a _ i r_ai _ m_4 b_2^i f_a + & - n_a _ i p_ai _ m_4 _ i ( r r ) & ; 12 ( -32 ) _ i p_i _ m_4 _ i ( r r ) these couplings can be combined in gs diagrams where @xmath148 couples to the @xmath32 untwisted field , which then couples to either two non - abelian gauge bosons or two gravitons , and hence may cancel both kinds of mixed anomalies
. the coefficients of these amplitudes , taking into account the coupling from @xmath22 image branes is , for the mixed non - abelian anomaly 12 _ i n_a ( r_aip_bi + r_aip_bi- r_aip_bi - r_aip_bi ) = n_a [ _ a]+ n_a [ _ a]for the mixed gravitational anomaly we have -_i n_a n_b ( r_aip_bi + r_aip_bi - r_aip_bi - r_aip_bi ) + ( -32)/2 _ i n_a ( r_ai p_i - r_ai p_i + = n_a [ _ a](-2_b n_b [ _ b ] - 2_b n_b [ _ b ] + ( -32 ) [ _ o6 ] ) = -2 48n_a [ _ a]where we have used the tadpole cancellation conditions .
the final expressions , modulo numerical factors not computed carefully , have precisely the form required to cancel the residual anomaly . the cancellation of mixed non - abelian and mixed gravitational anomalies in the @xmath13 orbifold model works analogously .
in fact , considering the chiral spectrum given in table [ matter ] , one can reproduce step by step the above computation and reach analogous results . namely the anomalies are canceled by a green - schwarz mechanism mediated by untwisted rr fields , whose couplings to gauge bosons and gravitons follow from the chern - simons interactions for d6-branes and o6-planes .
it is also clear that an analogous mechanism will be at work in models with other orbifold groups .
m. bianchi , g. pradisi and a. sagnotti , nucl .
b * 376 * , 365 ( 1992 ) ; m. bianchi , nucl .
b * 528 * , 73 ( 1998 ) e. witten , jhep * 9802 * , 006 ( 1998 ) ; z. kakushadze , g. shiu and s .- h .
h. tye , phys .
d * 58 * , 086001 ( 1998 ) . c. bachas , hep - th/9503030 .
g. aldazabal , a. font , l. e. ibez and a. m. uranga , nucl . phys .
b * 452 * ( 1995 ) 3 , hep - th/9410206 ; nucl .
b * 465 * ( 1996 ) 34 , hep - th/9508033 ; + j. erler , nucl . phys .
b * 475 * ( 1996 ) 597 , hep - th/9602032 ; + z. kakushadze and s. h. henry tye , phys .
d * 55 * ( 1997 ) 7878 , hep - th/9610106 ; phys .
d * 55 * ( 1997 ) 7896 , hep - th/9701057 ; z. kakushadze , g. shiu , s. h. tye and y. vtorov - karevsky , int . j. mod .
a * 13 * , 2551 ( 1998 ) , hep - th/9710149 and references therein .
m. cveti , g. w. gibbons , h. l and c. n. pope , hep - th/0101096 ; m. cveti , h. l and c. n. pope , hep - th/0105096 ; m. cveti , g. w. gibbons , j. t. liu , h. l " u and c. n. pope , hep - th/0106162 ; for a review , see : m. cveti , g. w. gibbons , h. l " u and c. n. pope , hep - th/0106177 . | we construct @xmath0 supersymmetric four - dimensional orientifolds of type iia on @xmath1 with d6-branes intersecting at angles .
the use of d6-branes not fully aligned with the o6-planes in the model allows for a construction of many supersymmetric models with chiral matter , including those with the standard model and grand unified gauge groups .
we perform a search for realistic gauge sectors , and construct the first example of a supersymmetric type ii orientifold with @xmath2 gauge group and three quark - lepton families .
in addition to the supersymmetric standard model content , the model contains right - handed neutrinos , a ( chiral but anomaly - free ) set of exotic multiplets , and diverse vector - like multiplets .
the general class of these constructions are related to familiar type ii orientifolds by small instanton transitions , which in some cases change the number of generations , as discussed in specific models .
these constructions are supersymmetric only for special choices of untwisted moduli .
we briefly discuss the supersymmetry breaking effects away from that point .
the m - theory lift of this general class of supersymmetric orientifold models should correspond to purely geometrical backgrounds admitting a singular @xmath3 holonomy metric and leading to four - dimensional m - theory vacua with chiral fermions .
epsf.tex # 1#2#3nucl .
phys .
b * # 1 * ( 19#2 ) # 3 # 1#2#3phys .
lett .
b * # 1 * ( 19#2 ) # 3 # 1#2#3phys . lett . *
# 1*b ( 19#2 ) # 3 # 1#2#3phys .
rev .
d * # 1 * ( 19#2 ) # 3 # 1#2#3phys .
rev .
lett . *
# 1 * ( 19#2 ) # 3 # 1#2#3phys . rep . * # 1 * c ( 19#2 ) # 3 # 1#2#3mod .
phys .
lett . *
# 1 * ( 19#2 ) # 3 # 1 # 1 2 2z2@xmath4 # 1 c i # 1 # 1 # 1^#1 # 1_#1 # 1/#1 /#1#1 #
1#1 # 1#1 # 1#1 # 1 # 1 # 1 # 1#1| # 1| # 1 # 1#1 # 1| # 1| # 1 # 1 # 1 # 1 # 1 # 1#2#1#2 # 1#2#1#2 # 1#2#3 ^ 2 # 1#2 # 3 # 1#2^#1 # 1 # 1 |
financial returns are known to be non - gaussian and exhibit fat - tailed distribution @xcite-@xcite .
the fat tail relates to intermittency an unexpected high probability of large price changes , which is of utmost importance for risk analysis .
the recent development of high - frequency data bases makes it possible to study the intermittent market dynamics on time scales of less than a day @xcite-@xcite . using foreign exchange ( fx ) intraday data , mller et al .
@xcite showed that there is a net flow of information from long to short timescales , i.e. , the behavior of long - term traders influences the behavior of short - term traders .
motivated by this hierarchical structure , ghashghaie et al .
@xcite have discussed analogies between the market dynamics and hydrodynamic turbulence @xcite , and claimed that the information cascade in time hierarchy exists in a fx market , which corresponds to the energy cascade in space hierarchy in a three - dimensional turbulent flow .
these studies have stimulated further investigations on similarities and differences in statistical properties of the fluctuations in the economic data and turbulence @xcite-@xcite .
differences have also emerged .
mantegra and stanley @xcite and arneodo et al .
@xcite pointed out that the time evolution , or equivalently the power spectrum is different for the price difference ( nearly white spectrum ) and the velocity difference ( @xmath2 spectrum , i.e. , @xmath3-law for the spectrum of the velocity ) .
moreover , from a parallel analysis of the price change data with time delay and the velocity difference data with time delay ( equivalent to the velocity difference data with spatial separation under the taylor hypothesis @xcite ) , it was shown that the time evolution of the second moment and the shape of the probability density function ( pdf ) , i.e. , the deviation from gaussian pdf are different in these two stochastic processes @xcite . on the other hand , non - gaussian character in fully developed turbulence
@xcite has been linked with the nonextensive statistical physics @xcite-@xcite . as dynamical foundation of nonextensive statistics ,
beck recently proposed a new model describing hydrodynamic turbulence @xcite .
the velocity difference @xmath4 of two points in a turbulent flow with the spatial separation @xmath5 is described by brownian motion ( an overdamped langevin equation @xcite ) in a power - law potential . assuming a @xmath0-distribution for the inverse temperature
, he obtained a tsallis distribution @xcite for @xmath4 .
however , if we take into account the almost uncorrelated behavior of the price change @xcite-@xcite , the picture by means of the brownian motion seems to be more appropriate for the market data rather than turbulence . moreover , the description by the langevin equation is able to relate the pdf of the price change to that of the volatility , a quantity known as a measure of the risk in the market
. thus we applied the model to fx market dynamics by employing the correspondence by ghashghaie et al .
we substitute the fx price difference @xmath6 with the time delay @xmath7 for the velocity difference @xmath4 with the spatial separation @xmath5 .
beck s model for turbulence then reads @xmath8 where @xmath9 is a constant , and @xmath10 is gaussian white noise corresponding to the temperature @xmath11 , satisfying @xmath12 .
the ` force ' @xmath13 is assumed to be obtained by a power - law potential @xmath14 with an exponent @xmath15 , where @xmath16 and @xmath17 is a positive constant .
that is , the system is subject to a restoring force proportional to the power of the price difference , @xmath18 besides the random force @xmath10 .
especially when @xmath19 , the restoring force is linear to @xmath20 . under a constant temperature @xmath11 , eq .
( [ dzdt ] ) leads to a stationary ( i.e. , thermal equilibrium ) distribution of @xmath20 as @xmath21 where @xmath22 is the inverse temperature @xcite .
the ` local ' variance of @xmath20 , which is defined for a fixed value of @xmath23 , is obtained from the conditional probability in eq .
( [ pzbeta ] ) as @xmath24 we define the volatility by the square root of the local variance of @xmath20 ( see eq .
( [ v ] ) ) .
when @xmath25 and @xmath26 , @xmath27 coincides with the temperature and the conditional pdf reduces to gaussian , while for @xmath28 , @xmath29 is proportional to @xmath11 .
let us assume that the ` temperature ' of the fx market is not constant and fluctuates in larger time scales , and @xmath23 is , just as in beck s model for turbulence , @xmath0-distributed with degree @xmath30 : @xmath31 where @xmath32 is the gamma function , @xmath33 is the average of the fluctuating @xmath23 and @xmath30 relates to the relative variance of @xmath23 : @xmath34 equation ( [ fbeta ] ) implies that the local variance @xmath35 fluctuates with the distribution @xmath36 .
the conditional probability in eq .
( [ pzbeta ] ) together with eq .
( [ fbeta ] ) yields a tsallis - type distribution @xcite for the ultimate pdf of @xmath20 : @xmath37 where tsallis nonextensivity parameter @xmath38 is defined by @xmath39 which satisfies @xmath40 because of @xmath16 and @xmath41 . since @xmath42 , the distribution of @xmath20 exhibits power - law tails for large @xmath20 : @xmath43 .
hence , the @xmath44th moment @xmath45 converges only for @xmath46 . in the limit of @xmath47 , @xmath48 in eq .
( [ pz ] ) reduces to the canonical distribution of extensive statistical mechanics : @xmath49 .
we have applied the present model to the same fx market data set as used in ref .
@xcite ( provided by olsen and associates @xcite which consists of 1 472 241 bid - ask quotes for us dollar - german mark exchange rates during the period october 92 - september 93 ) .
the volatility is often estimated by the standard deviation of the price change in an appropriate time window @xcite . employing this definition , we express the volatility in terms of the local standard deviation of the price change @xmath50 as here
the window size has been chosen as @xmath52 . since @xmath53 corresponds to @xmath23 ( see eq .
( [ z2beta ] ) ) which is @xmath0-distributed , @xmath30 can be explicitly obtained from the relative variance of @xmath23 using eq .
( [ beta ] ) .
thus , there is only one adjustable parameter among @xmath54 , because we have another relation , eq .
( [ q ] ) . in other words ,
the pdf , @xmath48 of the price change and the pdf , @xmath55 of the inverse power @xmath56 of the volatility are determined simultaneously once the value of @xmath57 has been specified .
the pdf s with time delay @xmath7 varying from five minutes up to approximately six hours are displayed in fig.[tsallis ] together with theoretical curves obtained from eq .
( [ pz ] ) . as the time scale @xmath7 increases , @xmath30 increases , while @xmath57 and @xmath38 decrease .
the nonextensivity parameter @xmath38 tends to the extensive limit : @xmath58 as @xmath7 increases . using the same parameter values @xmath54 , the pdf s of @xmath23
are compared in fig .
[ x2 ] . the average ` temperature ' in the market @xmath59 increases with @xmath7 since @xmath60 is positive .
( we obtained the scaling exponent @xcite @xmath61 , which is larger than 2/3 obtained for turbulence . ) in contrast , the fluctuation of the temperature increases with decreasing @xmath7 because the variance of the inverse temperature is proportional to @xmath62 .
the smaller values of @xmath30 imply the stronger intermittency which occurs in small time scales .
the intermittent character in the price change can be seen as a fat tail of @xmath48 in fig .
[ tsallis ] . also in fig .
[ x2 ] , the peak of @xmath55 shifts to smaller @xmath23 as @xmath7 decreases ( from d to a in fig .
[ x2 ] ) , which means relatively high temperatures are realized more frequently in short time scales . it should be noted that the pdf in eq .
( [ pz ] ) with @xmath19 reduces to student s @xmath63-distribution @xcite , which has often been used to characterize the fat tails @xcite .
when @xmath19 , there is no adjustable parameter because @xmath30 is decided from eq .
( [ beta ] ) .
the fx market system is then subject to a restoring force linear to the price change and the volatility is proportional to the temperature .
we have found that the pdfs of @xmath20 and @xmath23 reproduce , although very roughly , the olsen and associates data points even if @xmath57 is fixed at @xmath19 .
however , adjusting the parameter @xmath57 improved the line shape of @xmath48 in a range close to @xmath64 .
the data points in fig .
[ tsallis ] exhibit a cusp at @xmath64 for large time scales @xmath7 , which implies a singularity of the second derivative of the pdf at @xmath64 .
the larger reduction of @xmath57 from unity leads to the stronger singularity .
( note that the factor @xmath65 arises from @xmath66 . )
thus the better fitting for large @xmath7 was obtained from a reduced value of @xmath57 .
the trend of decrease in @xmath57 with increasing @xmath7 was observed for the turbulent flow @xcite as well .
however , the deviation from @xmath67 is much smaller than the present case and the cusp is invisible .
ghashghaie et al .
@xcite have used a model for turbulence by castaing et al .
@xcite , in which a log - normal distribution has been assumed for the local standard deviation of the price change .
the present model reduces to the model by ghashghaie et al . if the stochastic process given in eq .
( [ dzdt ] ) is assumed with @xmath67 ( the local variance of @xmath20 is then proportional to @xmath11 ) and the @xmath1-distribution for @xmath23 ( the inverse of the local variance of @xmath20 ) is replaced by the log - normal distribution for @xmath68 ( the local standard deviation of @xmath20 ) .
although no analytic expression like eq .
( [ pz ] ) for @xmath48 is obtained , a similar qualitative explanation can be applied to their model : the volatility ( or equivalently , the square root of the temperature ) fluctuates slowly with a log - normal distribution , and the smaller time scale corresponds to the larger variance of the logarithm of the volatility .
( the variance is denoted by @xmath69 in ref .
however , the power - law behavior of the tail of the volatility distribution @xcite can be better described by the @xmath1-distribution for the inverse of the variance .
we have proposed the stochastic process described by eq .
( [ dzdt ] ) for fx market dynamics in small time scales .
in fact , eq . ( [ dzdt ] ) is the simplest stochastic process which can realize the thermal equilibrium distribution , eq .
( [ pzbeta ] ) in the power - law potential .
more realistic and more complicated processes that assure convergence to eq .
( [ pzbeta ] ) at local temperatures might be possible . mantegna and stanley
have proposed a different stochastic model of the price change , which is described by a truncated levy flight ( tlf ) @xcite .
the model , yielding approximately a stable distribution , well reproduces the self - similar property of pdf at different time scales : @xmath70 1 min to 1 000 min .
however , the parameters @xmath57 and @xmath71 characterizing the stable distribution fluctuate for larger ( monthly ) time scales @xcite , where @xmath71 gives a measure of the volatility .
in other words , the ultimate distribution ( let us denote it @xmath72 ) should be obtained , like eq .
( [ pz ] ) , from the weighted average over these parameters .
a difference between @xmath72 and @xmath48 is that @xmath72 has no cusp at @xmath64 : @xmath73 for @xmath74 , whereas @xmath75 , and the present fx date set indeed exhibits a cusp as seen in fig .
[ tsallis ] .
finally , a fundamental question has been left open : how to derive theoretically the increasing trend of fat - tailed character with decreasing time scale , i.e. , @xmath7-dependence of the parameters ( @xmath76 ) . an attempt to derive the non - gaussian fat - tailed character in small time scales
was made by friedlich et al .
recently @xcite .
they derived a multiplicative langevin equation from a fokker - planck equation and showed that the equation becomes more multiplicative and hence fat - tailed as @xmath7 decreases . clarifying the relation between their multiplicative langevin equation and eq .
( [ dzdt ] ) ( the latter being rather simple although including the fluctuating temperature ) should be important as well as interesting .
fa65 e. f. fama , j. business 38 ( 1965 ) 34 .
ms00 r. n. mantegna , h. e. stanley , an introduction to econophysics : correlations and complexity in finance , cambridge univ .
press , cambridge , 2000 .
bg01 l. bauwens , p. giot , econometric modelling of stock market intraday activity , kluwer acad .
publishers , boston , 2001 .
gbp96 s. ghashghaie , w. breymann , j. peinke , p. talkner , y. dodge , nature 381 ( 1996 ) 767 .
ms96 r. n. mantegna , h. e. stanley , nature 383 ( 1996 ) 587 .
abc96 a. arneodo , j. -p .
bouchaud , r. cont , j. -f .
muzy , m. potters , d. sornette , cond - mat /9607120 .
ms97 r. n. mantegna , h. e. stanley , physica a 239 ( 1997 ) 255 .
ams98 a. arneodo , j. -f .
muzy , d. sornette , eur .
j. b2 ( 1998 ) 277 .
fpr00 r. friedrich , j. peinke , ch .
renner , phys .
84 ( 2000 ) 5224 .
bgt00 w. breymann , s. ghashghaie , p. talkner , int . j. theor
finance 3 ( 2000 ) 357 .
gof97 c. a. e. goodhart , m. ohara , j. emp .
finance 4 ( 1997 ) 73 .
mdd97 u. a. mller , m. m. , dacorogna , r. d. dav , r. b. olsen , o. v. pictet , j. e. von weizscker , j. emp .
finance 4 ( 1997 ) 213 .
ms95 r. n. mantegna , h. e. stanley , nature 376 ( 1995 ) 46 .
lgc99 y. liu , p. gopikrishnan , p. cizeau , m. meyer , c. -k .
peng , h. e. stanley , phys .
e 60 ( 1999 ) 1390 . hkk01 r. huisman , k. g. koedijk , c. j. m. kool , f. palm , j. business and economic statistics 19 ( 2001 ) 208 .
fr95 u. frisch , turbulence : the legacy of a. n. kolmogorov , cambridge univ .
press , cambridge , 1995 .
cgh90 b. castaing , y. gagne , e. j. hopfinger , physica d 46 ( 1990 ) 177 .
ts88 c. tsallis , j. stat .
52 ( 1988 ) 479 .
ww00 g. wilk , z. wodarczyk , phys .
84 ( 2000 ) 2770 .
aa01 t. arimitsu , n. arimitsu , prog .
( 2001 ) 355 .
be01 c. beck , phys .
87 ( 2001 ) 180601 .
be01a c. beck , physica a 295 ( 2001 ) 195 .
bls01 c. beck , g. s. lewis , h. l. swinney , phys .
63 e ( 2001 ) 035303 .
ri89 h. risken , the fokker - planck equation : methods of solution and applications , 2nd edn , springer - verlag , berlin , 1989 .
oa93 high frequency data in finance 1993 , olsen and associates , zurich . | we present a model of financial markets originally proposed for a turbulent flow , as a dynamic basis of its intermittent behavior .
time evolution of the price change is assumed to be described by brownian motion in a power - law potential , where the ` temperature ' fluctuates slowly .
the model generally yields a fat - tailed distribution of the price change .
specifically a tsallis distribution is obtained if the inverse temperature is @xmath0-distributed , which qualitatively agrees with intraday data of foreign exchange market . the so - called ` volatility ' , a quantity indicating the risk or activity in financial markets , corresponds to the temperature of markets and its fluctuation leads to intermittency . , foreign exchange market , volatility , tsallis distribution , @xmath1-distribution , brownian motion |
the whitham equations are a collection of quasi - linear hyperbolic systems of the form @xcite,@xcite,@xcite @xmath7 with the ordering @xmath8 . for a given @xmath9 the system ( [ whithamg ] )
is called @xmath9-phase whitham equations . because of the diagonal form of systems ( [ whithamg ] ) the dependent variables @xmath8 are called riemann invariants . for @xmath0 the speeds @xmath10 , @xmath11 , depend through @xmath12 on complete hyperelliptic integrals of genus @xmath9 . for this reason
the @xmath9-phase system is also called genus @xmath9 system . the zero - phase whitham equation has the form @xmath13 where we use the notation @xmath14 . in this paper
we study the initial value problem of the whitham equations for monotone increasing smooth ( @xmath15 ) initial data @xmath16 , where the range of @xmath17 , @xmath18 , is the interval @xmath19 and @xmath20 .
the initial value problem consists of the following .
we consider the evolution on the @xmath21 plane of the initial curve @xmath16 according to the zero - phase equation ( [ burgereq ] ) .
the solution @xmath22 of ( [ burgereq ] ) with the initial data @xmath17 is given by the characteristic equation @xmath23 where @xmath24 is the inverse function of @xmath17 .
the solution @xmath22 in ( [ zp ] ) is globally well defined only for @xmath25 , where @xmath26 $ ] is the time of gradient catastrophe of ( [ zp ] ) . near the point of gradient catastrophe and for a short time
@xmath27 , the evolving curve is given by a multivalued function with three branches @xmath28 , which evolve according to the one - phase whitham equations .
outside the multivalued region the solution is given by the zero - phase solution @xmath22 defined in ( [ zp ] ) .
on the phase transition boundary the zero - phase solution and the one - phase solution are attached @xmath29-smoothly ( see fig[ivp ] ) .
since the whitham equations are hyperbolic , other points of gradient catastrophe can appear in the branches @xmath30 themselves or in @xmath22 . in general , for @xmath27 ,
the evolving curve is given by a multivalued function with and odd number of branches @xmath31 , @xmath32 .
these branches evolve according to the @xmath9-phase whitham equations .
the @xmath9-phase solutions for _ different @xmath9 _ must be glued together in order to produce a @xmath29-smooth curve in the @xmath33 plane evolving smoothly with @xmath34 ( see fig .
[ leeds4 ] ) . the initial value problem of the whitham equations is to determine , for almost all @xmath35 and @xmath6 , the phase @xmath36 and the corresponding branches @xmath37 from the initial data @xmath38 .
for example for the initial data @xmath39 , @xmath40 , the solution of the whitham equations is of genus at most equal to one @xcite@xcite .
it is of genus one inside the cusp @xmath41 @xcite ; it is of genus zero outside this cusp .
the point @xmath42 is the point of gradient catastrophe of the zero - phase solution .
the curve @xmath43 , @xmath27 , describes the locus of points where @xmath44 .
the curve @xmath45 , @xmath27 , describes the locus of points where @xmath46 . for generic initial data it is not known whether the genus of the solution of the whitham equations is bounded
we say that the cauchy problem for the whitham equations has a global solution if the genus @xmath47 for all @xmath6 and @xmath5 . using the geometric - hamiltonian structure @xcite of the whitham equations ,
tsarev @xcite showed that these equations can be locally integrated by a generalization of the method of characteristic .
namely he proved that if the functions @xmath48 , @xmath49 , solve the linear over - determined system @xmath50,\quad i , j=1,2,\dots,2g+1,\;\;i\neq j,\ ] ] where @xmath51 , are the speeds in ( [ whithamg ] ) , then the solution @xmath52 of the so called hodograph transform @xmath53 satisfies system ( [ whithamg ] ) .
conversely , any solution @xmath54 of ( [ whithamg ] ) can be obtained in this way .
tsarev theorem relies on two factors : \a ) the existence of a solution of the linear over - determined system ( [ tsarev0 ] ) ; \b ) the existence of a real solutions @xmath37 of the hodograph transform ( [ ch0 ] ) . in this paper
we solve completely problem a ) , namely we build for any smooth monotone increasing initial data @xmath38 the solution of system ( [ tsarev0 ] ) for any @xmath32 .
the solution @xmath55 , @xmath49 , of genus @xmath9 satisfies some natural boundary conditions which guarantee its uniqueness .
regarding problem b ) we characterize initial data such that the solution of the whitham equations exists only for @xmath3 where @xmath56 is a positive integer .
namely we show that if the initial data satisfies the condition @xmath57 for all real @xmath58 belonging to the domain of @xmath59 except one point , then the solution of the whitham equations has genus at most @xmath56 for any @xmath6 and @xmath5 . for @xmath60
this result has already been proved by tian @xcite .
the investigation of the initial value problem of the whitham equations was initiated by gurevich and pitaevskii @xcite . in the case
@xmath61 they solved the initial value problem of system ( [ whithamg ] ) for step - like initial data and studied numerically the case of cubic initial data .
the initial value problem of the whitham equations was deeply studied by tian . in @xcite
he constructed the general solution of the tsarev system ( [ tsarev0 ] ) for @xmath62 and for smooth monotone increasing initial data .
he also proved the solvability of the hodograph transform ( [ ch0 ] ) for @xmath62 .
in @xcite he obtained the solution of the tsarev system for @xmath63 for polynomial initial data .
the equations ( [ whithamg ] ) were found by whitham @xcite in the single phase case @xmath64 and more generally by flaschka , forest and mclaughlin @xcite in the multi - phase case .
the whitham equations were also found in @xcite when studying the zero dispersion limit of the korteweg de vries equation .
the hyperbolic nature of the equations was found by levermore @xcite .
this paper is organized as follows . in sec .
2 we give some background about abelian differentials on hyperelliptic riemann surfaces . in sec .
3 we describe the whitham equations . in sec .
4 we build the solution of the tsarev system ( [ tsarev0 ] ) for smooth monotone increasing initial data . in sec .
5 we show that under the hypothesis ( [ derivative ] ) the hodograph transform is solvable for @xmath3 . in sec .
6 we draw the conclusions .
let @xmath65 be the hyperelliptic riemann surface of genus @xmath32 .
we shall use the standard representation of @xmath66 as a two - sheeted covering of @xmath67 with cuts along the intervals @xmath68,\quad k=1,\dots , g+1,\quad u_{2g+2}=-\infty\,.\ ] ] we choose the basis @xmath69 of the homology group @xmath70 so that @xmath71 lies fully on the upper sheet and encircles clockwise the interval @xmath72 $ ] , @xmath73 , while @xmath74 emerges on the upper sheet on the cut @xmath72 $ ] , passes anti - clockwise to the lower sheet trough the cut @xmath75 $ ] and return to the initial point through the lower sheet .
the one - forms that are analytic on the closed riemann surface @xmath66 except for a finite number of points are called abelian differentials .
we define on @xmath66 the following differentials @xcite : \1 ) the canonical basis of holomorphic one - forms or abelian differentials of the first kind @xmath76 : @xmath77 the constants @xmath78 are uniquely determined by the normalization conditions @xmath79 we remark that an holomorphic differential having all its @xmath80-periods equal to zero is identically zero @xcite .
\2 ) the set @xmath81 , @xmath82 , @xmath32 , of abelian differentials of the second kind with a pole of order @xmath83 at infinity , with asymptotic behavior @xmath84dr
\quad \mbox{for large}\;\ ; r\ ] ] and normalized by the condition @xmath85 we use the notation @xmath86 in literature the differentials @xmath87 and @xmath88 are called quasi - momentum and quasi - energy respectively @xcite .
the explicit formula for the differentials @xmath89 , @xmath82 , is given by the expression [ b6 ] ^g_k(r)=dr , p^g_k ( r)= r^g+k+a^k_1 r^g+k-1+a^k_2 r^g+k-2 +a^k_g+k , where the coefficients @xmath90 , @xmath91 , @xmath92 are uniquely determined by ( [ sigma ] ) and ( [ norm2 ] ) .
\3 ) the abelian differential of the third kind @xmath93 with first order poles at the points @xmath94 and @xmath95 with residues @xmath96 respectively .
its periods are normalized by the relation @xmath97 let @xmath98 and @xmath99 be two abelian differentials on the riemann surface @xmath66 . if all the residues of @xmath98 and @xmath99 are equal to zero , then the integrals @xmath100 and @xmath101 do not have logarithm singularities on @xmath66 .
if the differential @xmath98 has non zero residue , then its integral has logarithm singularities .
let @xmath102 be the path connecting the singular points of @xmath100 .
we have the following relation . @xmath103 + \int_s\delta(d^{-1}\om_1)\om_2=2\pi i \sum_{\cs_g - s } { \operatornamewithlimits{res } } [ ( d^{-1}\om_1)\om_2]\,,\ ] ] where @xmath104 is the difference of the values of @xmath100 on the two sides of the cut @xmath102 and the quantity @xmath105 $ ] is the sum of the residues of the differential @xmath106 on the cut surface @xmath107 .
this formula is known as the riemann bilinear period relation @xcite .
assuming @xmath108 and @xmath109 in ( [ firstr ] ) we obtain @xmath110 assuming @xmath111 and @xmath112 in ( [ firstr ] ) we obtain @xmath113 differentiating with respect to @xmath114 and @xmath115 the above expression we obtain the identity @xmath116=d_p[{\omega_{pp_0}}(q)]\,,\ ] ] where @xmath117 and @xmath118 denote differentiation with respect to @xmath115 and @xmath114 respectively . from the expression ( [ int3 ] ) it follows that @xmath119 is a many - value analytic function of the variable @xmath115 .
the many - value character of @xmath119 as a function of @xmath115 can be described by the equations @xmath120=0\,,\quad
\int_{\beta_k}d_q[{\omega_{qq_0}}(r)]= 2\pi i \phi_k(r),\quad k=1,\dots , g\,,\ ] ] in the following we mainly use the normalized differential @xmath121 which has simple poles at the points @xmath122 with residue @xmath96 respectively .
the differential @xmath121 is explicitly given by the expression @xmath123 where @xmath124 , @xmath125 , is the normalized basis of holomorphic differentials .
@xmath121 as a function of @xmath126 , is an abelian integral having poles of first order at the points @xmath127 .
the periods of this integral are obtained from the relations ( [ mvp ] ) @xmath128=0,\quad \int_{\beta_j}d_z[{\omega^g_z(r)}]= 4\pi i \phi_k(r)\,,\quad j=1,\dots g\,.\ ] ] we apply the riemann bilinear relation ( [ firstr ] ) to the differentials @xmath129 and @xmath121 getting @xmath130 [ { \omega^g_z(r)}d^{-1}\sg^g_m(r ) ] , \quad m=0,\dots , g , \\ = & -\dfrac{4}{2m+1}\left(-\mu(z)\e_{mg}+\sum_{k=1}^g\sum_{j=1}^{g } \gamma^k_{j } \gamma_{m+1-j } \int_{\alpha_k}\dfrac{dt}{\mu(t)}\dfrac{\mu(z)}{t - z}\right)\ , .
\end{split}\ ] ] in the above formula @xmath131 for @xmath132 and zero otherwise , the coefficients @xmath133 have been defined in ( [ holo ] ) and the @xmath134 s are the coefficients of the expansion for @xmath135 of @xmath136 we define @xmath137 for @xmath138 . inverting ( [ norm ] ) and introducing the quantities
@xmath139 we obtain @xmath140 where the @xmath141 s are the coefficients of the expansion for @xmath142 of @xmath143 using ( [ norm1 ] ) the differential @xmath121 turns out to be given by the relation @xmath144 we remark that @xmath121 is a multivalued function of @xmath126 and it is regular at infinity . from the relation ( [ norm1 ] ) we obtain the identity which will be useful later @xmath145 the next proposition is also important for our subsequent considerations .
the abelian differentials of the second kind @xmath146 , @xmath147 , defined in ( [ sigma ] ) satisfy the relations @xmath148=-\dfrac{1}{2k+1}d_r{\operatornamewithlimits{res}_{z=\infty}}\left [ \dorz\,z^{k+\frac{1}{2}}\right]\,,\ ] ] where @xmath121 has been defined in ( [ ck ] ) , @xmath149 is the normalized abelian differential of the third kind with simple poles at the points @xmath127 with residue @xmath96 respectively and @xmath150 denotes differentiation with respect to @xmath151 .
* proof*:the differential @xmath152 $ ] is normalized because @xmath121 is a normalized differential . furthermore @xmath153
= r^{k-\frac{1}{2}}dr+ o(r^{-\frac{3}{2}})dr\quad \mbox { for } \;r\ra \infty.\ ] ] therefore @xmath154 $ ] coincides with the normalized abelian differential of the second kind @xmath146 . for proving the second equality in ( [ maini ] ) we consider the integral in the @xmath126 variable @xmath155 where @xmath156 is a close contour around the point at infinity . substituting the identity @xmath157 in the right hand side of the above relation we obtain the second relation in ( [ maini ] ) .
@xmath158 .2 cm
the speeds @xmath10 of the @xmath9-phase whitham equations ( [ whithamg ] ) are given by the ratio @xcite,@xcite : @xmath159 where @xmath87 and @xmath88 have been defined in ( [ dp ] ) .
in the case @xmath160 @xmath161 so that one obtains the zero - phase whitham equation ( [ burgereq ] ) . for monotone increasing smooth initial data @xmath38 , the solution of the zero - phase equation ( [ burgereq ] )
is obtained by the method of characteristic @xcite and is given by the expression @xmath162 the zero - phase solution is globally well defined only for @xmath163 where @xmath26 $ ] is the time of gradient catastrophe of the solution ( [ zerop ] ) .
the breaking is cause by an inflection point in the initial data .
for @xmath164 we expect to have single , double and higher phase solutions . for higher genus the whitham equations
can be locally integrated using a generalization of the characteristic equation ( [ zerop ] ) .
we have the following theorem of tsarev @xcite [ tsarevt ] if @xmath55 , @xmath91 , solves the linear over - determined system @xmath165,\quad i , j=1,2,\dots,2g+1,\;\;i\neq j,\ ] ] where @xmath166 then the solution @xmath167 of the hodograph transformation @xmath168 satisfies system ( [ whithamg ] ) .
conversely , any solution @xmath169 of ( [ whithamg ] ) can be obtained in this way . to guarantee that the @xmath170phase solutions for different @xmath9 are attached continuously ,
the following natural boundary conditions must be imposed on @xmath171 , @xmath49 . when @xmath172 , @xmath173 , @xmath174 and for @xmath175 [ b2 ] w_i^g(u_1, ,u_l , u_l, ,u_2g+1)= w_i^g-1(u_1, ,_l,_l, ,u_2g+1 )
. the sup - scripts @xmath9 and @xmath2 in the @xmath176 s specify the corresponding genus and the hat denotes the variable that have been dropped . when @xmath64 and @xmath177 we have that [ b3 ] lll & w_1(u_1,u_3,u_3)=f(u_1 ) + & w_2(u_1,u_3,u_3)=w_3(u_1,u_3,u_3 ) , where @xmath24 is the initial data .
similar conditions hold true when @xmath178 , namely [ b4 ] lll & w_3(u_1,u_1,u_3)=f(u_3 ) + & w_1(u_1,u_1,u_3)=w_2(u_1,u_1,u_3 ) .
we remark that the @xmath179 s satisfy the boundary conditions ( [ b1]-[b2 ] ) and for @xmath64 we have @xcite @xmath180 and @xmath181 the solution of the boundary value problem ( [ tsarev ] ) , ( [ b1]-[b4 ] ) has been obtained in @xcite for monotone increasing analytic initial data of the form @xmath182 where we assume that only a finite number of @xmath183 is different from zero . for such initial data the @xmath55 s which satisfy ( [ tsarev ] ) and
the boundary conditions ( [ b1]-[b4 ] ) are given by the expression @xcite @xmath184 the differential @xmath185 in ( [ wi1 ] ) is given by @xmath186 and the differentials @xmath187 , @xmath188 have been defined in ( [ b6 ] ) .
the solution ( [ ch ] ) of the @xmath9-phase whitham equations can also be written in an equivalent algebro - geometric form @xcite namely @xmath189 where @xmath190 , @xmath191 and @xmath185 have been defined in ( [ dp ] ) and ( [ ds ] ) respectively . the solution @xmath192 of the @xmath9-phase whitham equations ( [ whithamg ] ) is implicitly defined as a function of @xmath6 and @xmath34 by the equations ( [ ch ] ) or ( [ b20 ] ) .
the solution is uniquely defined only for those @xmath6 and @xmath34 such that the functions @xmath193 are real and @xmath194 , @xmath49 , are not vanishing .
one of the problems in the theory of the whitham equations is to determine when ( [ ch ] ) or ( [ b20 ] ) are solvable for real @xmath12 as functions of @xmath6 and @xmath34 .
this problem has been solved by tian for @xmath61 .
@xcite [ tian ] consider a monotone increasing initial data @xmath38 .
suppose that @xmath195 is the inflection point of @xmath24 that causes the breaking of the zero - phase solution ( [ burgereq ] ) at @xmath196 .
let be @xmath197 in a deleted neighborhood of @xmath198 .
then the one - phase whitham equations has a solution @xmath199 within a cusp in the @xmath200 plane for a short time after the breaking time of the zero - phase solution .
furthermore this solution satisfies the boundary conditions ( [ b3 ] ) and ( [ b4 ] ) on the cusp . if the initial data satisfies the condition @xmath197 for all @xmath58 except @xmath198 , then the solution of the whitham equations exists for all @xmath35 .
the solution is of genus one inside the cusp @xmath201 , @xmath27 , where @xmath202 are two real functions satisfying the condition @xmath203 .
the solution is of genus zero outside the cusp @xmath201 , @xmath27 .
in this section we build the solution of the boundary value problem ( [ tsarev ] ) , ( [ b1]-[b4 ] ) for monotone smooth initial data and we show that the solution obtained is unique .
we consider initial data of the form @xmath38 where @xmath24 is a monotone increasing function .
the domain of @xmath59 is the interval @xmath19 where @xmath20 , and the range of @xmath59 is the real line @xmath204 . in order to obtain the solution of the boundary value problem ( [ tsarev ] ) , ( [ b1]-[b4 ] ) we need the following technical lemma .
the differential @xmath205 defined in ( [ ds ] ) can be written in the form @xmath206 where @xmath207 , \quad q_k({{\vec{u}}})=-{\operatornamewithlimits{res}_{z=\infty}}\left[\dfrac{z^{g - k}\cf(z)dz}{2\mu(z)}\right],\ ; k=1,\dots g,\ ] ] @xmath208 @xmath209 and the polynomials @xmath210 , @xmath211 , have been defined in ( [ b6 ] ) , the @xmath212 s have been defined in ( [ gt ] ) . * proof*:using the second identity in ( [ maini ] ) ,
we rewrite the differential @xmath205 defined in ( [ ds ] ) in the form @xmath213\right)\,,\ ] ] where @xmath149 has been defined in ( [ ck ] ) and @xmath214 is the abel transform defined in ( [ abel ] ) of the analytic initial data ( [ b18 ] ) . the identity ( [ ds2 ] ) can be checked straightforward . using the explicit expression of @xmath149 in ( [ dob ] )
we obtain @xmath215 where @xmath216 and @xmath217 have been defined in ( [ psia ] ) . from ( [ psia ] )
we get the relations @xmath218 and for @xmath160 we define @xmath14 and @xmath219 using ( [ psira ] ) we transform the expression for @xmath205 in ( [ ds31 ] ) to the form ( [ ds3 ] ) .
@xmath158 .2 cm the relation ( [ ds3 ] ) enables us to write the quantities @xmath220 in ( [ wi1 ] ) in the form @xmath221.\ ] ] observe that in the formula ( [ wia ] ) all the information on the initial data is contained in the functions @xmath217 .
the functions @xmath222 , @xmath125 , solve the linear over - determined system @xcite [ qka ] rcl 2(u_i - u_j)&= & - , i , j=1 ,
, 2g+1 , + & & + q_k(_2g+1)&= & u^-k+ ( u^g - f_a^(k-1)(u ) ) , where @xmath223 is the @xmath224 derivative of the polynomial initial data @xmath225 .
the function @xmath216 in ( [ psia ] ) satisfies a similar linear over - determined system .
[ theot ] let be @xmath24 a smooth monotone increasing function with domain @xmath19 , @xmath20 . if @xmath226 , @xmath227 , is the symmetric solution of the linear over - determined system [ qk ] \ {
lll 2(u_i - u_j)= - , ij , i , j=1 , , 2g+1,g>0 & & + & & + q_k(_2g+1)=f_k(u ) & & + & & + f_k(u)= u^-k+ ( u^g - f^(k-1)(u ) ) , & & . with the ordering @xmath8 , then @xmath228 , \quad i=1,\dots , 2g+1,\ ] ] solves the boundary value problem ( [ tsarev ] ) , ( [ b1]-[b4 ] ) .
conversely every solution of ( [ tsarev ] ) , ( [ b1]-[b4 ] ) can be obtained in this way . before proving the theorem
we show how to obtain the solution of system ( [ qk ] ) for generic smooth initial data .
we follow the procedure in @xcite .
we start with the following lemma .
@xcite [ texist ] the system [ exist ] \ { cll 2(z - y)p_zy&=&p_z - p_y,>0 + p(z , z)&=&s(z ) .
has , for any smooth initial data @xmath229 one and only one solution .
moreover , the solution can be written explicitly @xmath230 where @xmath231 using the above lemma , the linear over - determined systems ( [ qk ] ) can be integrated for any smooth initial data in the following way .
suppose that @xmath232 is a solution of ( [ qk ] ) .
clearly @xmath233 satisfies rcl 2(u_1-u_2g+1)&= & -2 g + & & + a_k(u , u)&=&f_k(u ) which by lemma [ texist ] implies that @xmath234 for each fixed @xmath235 the function @xmath236 satisfies [ qb ] cll 2(u_1-u_2g)&= & -(2g-1 ) + & & + b_k(u , u , u_2g+1)&=&a_k(u ,
u_2g+1 ) using again lemma [ texist ] we obtain @xmath237 going on in the process of integration we obtain the solution @xmath238 of the boundary value problem ( [ qk ] ) namely @xmath239 where @xmath240 and @xmath241 has been defined in ( [ crho ] ) . when the initial data is of the form ( [ b18 ] ) , the expression ( [ fqk ] ) for the @xmath217 s is equivalent to ( [ psia ] ) .
the boundary value problem ( [ qk ] ) has one and only one solution .
the solution is symmetric and is given by ( [ fqk ] ) .
* proof*:uniqueness follows from lemma [ texist ] and the argument previous to ( [ fqk ] ) .
the boundary condition ( [ qk ] ) is clearly satisfied .
the symmetry follows from the construction . indeed in the process of integration we can interchange the role of any of the variable @xmath242 . @xmath158
.2 cm we have the following relations . the functions @xmath243 and the solutions @xmath217 , @xmath125 , of the boundary value problem ( [ qk ] ) satisfy the following relations . [ relations ] cll _
u f_k(u)&=&f_k+1(u)+u _ u f_k+1(u),k=1 , , g-1,g>0 + & & + _ u_iq_k()&=&q_k+1()+u_i _
u_i q_k+1 ( ) i=1. ,2g+1,k=1, ,g-1g>0 .
* proof of theorem [ theot ] ( first main theorem ) . *
we consider the non trivial case where @xmath244 and @xmath245 , @xmath246 .
the proof consists of three parts .
* a)*_the @xmath55 s defined in ( [ wi ] ) satisfy ( [ tsarev ] ) .
_ using the definition of @xmath55 in ( [ wi ] ) we have the following relation @xmath247 the following identities hold @xcite @xmath248 @xmath249 where @xmath179 has been defined in ( [ b9 ] ) and @xmath250 has been defined in ( [ sigma ] ) . using the definition of @xmath141 in ( [ gt ] ) it is easy to verify that @xmath251 applying repeatedly the relations ( [ relations ] ) we obtain the following expression for @xmath252 : @xmath253 from ( [ qk ] ) and ( [ x1]-[x4 ] ) we can write @xmath254 in ( [ aa1 ] ) in the form @xmath255 simplifying we obtain @xmath256 adding and subtracting the quantity @xmath257 to ( [ aa2 ] ) , we can reduce it to the form @xmath258=\partial_{u_j}q_g({{\vec{u}}})\left(\dfrac{2}{\lb_i-\lb_j}\dfrac{\partial \lb_i}{\partial u_j}\dfrac{\prod_{l=1,l\neq j}^{2g+1}(u_j - u_l)}{p_0^g(u_j ) } \right.\\ -&\left.\dfrac{\prod_{l=1,l\neq i}^{2g+1}(u_i - u_l)}{(u_i - u_j)p_0^g(u_i ) } + \sum_{l=1}^g(2l-1)\dfrac{p_{l-1}^g(u_i)}{p_0^g(u_i ) } \sum_{k = l}^{g}\gt_{k - l}u_j^{g - k}\right ) \end{split}\ ] ] the term in parenthesis in the right hand side of ( [ aa3 ] ) does not depend on the initial data @xmath24 .
it is identically zero for the analytic initial data ( [ b18 ] ) because in such case the @xmath176 s satisfy ( [ tsarev ] ) @xcite .
therefore we can conclude that @xmath259=0\ ] ] for any smooth monotone increasing initial data @xmath260 .
* b)*_the @xmath55 s satisfy the boundary conditions ( [ b1]-[b4 ] ) .
_ in the following we use the sup - script @xmath9 to denote the corresponding genus of the quantities we are referring to .
we have the following relations .
when @xmath261 for @xmath262 , the @xmath141 s defined in ( [ gt ] ) satisfy @xmath263 and the @xmath217 s defined in ( [ fqk ] ) satisfy @xmath264 for @xmath265 we have that @xmath266 which follows from ( [ relations ] ) .
when @xmath261 the polynomials @xmath267 s defined in ( [ b6 ] ) satisfy the relation @xcite @xmath268 using the relations ( [ z1]-[z4 ] ) we have that for @xmath269 , @xmath270 , @xmath271 which reduces to the form @xmath272 using ( [ mu ] ) the term in parenthesis in the right hand side of ( [ z5 ] ) turns out to be identically zero .
therefore the boundary conditions ( [ b2 ] ) are satisfied for any smooth monotone increasing initial data . since the functions @xmath217 s in ( [ fqk ] ) and the @xmath273 s defined in ( [ gt ] ) are symmetric with respect to @xmath274 we immediately deduce from ( [ z4 ] ) that @xmath275 so that the boundary conditions ( [ b1 ] ) are satisfied ( for a more detailed analysis of this limit see section [ transition ] ) .
when @xmath64 we deduce from ( [ pp ] ) @xmath276 and from ( [ z3]-[z4 ] ) @xmath277 from ( [ qk ] ) and ( [ fqk ] ) we get the relation @xmath278 so that @xmath279 an analogous result can be obtain when @xmath177 , so that the boundary conditions ( [ b3]-[b4 ] ) are satisfied .
0.3 cm * c ) * _ uniqueness .
_ we prove that when @xmath280 then @xmath281 for all @xmath282 , for @xmath283 and for any @xmath284 .
the proof is by induction on @xmath9 . for @xmath160
the statement is satisfied .
for @xmath64 we repeat the arguments of @xcite .
we fix @xmath285 and we consider the equation ( [ tsarev ] ) with boundary condition ( [ b3]-[b4 ] ) , namely rcl & = & a_13[w_1-w_3 ] + & & + & = & a_31[w_3-w_1 ] + w_1(u_1,u_2,u_2)&=&f(u_1)0 + w_3(u_2,u_2,u_3)&=&f(u_3)0
. we can regard the above equations as a first - order linear ordinary differential equations with non homogeneous term .
integrating them we obtain a couple integral equation . by standard contraction mapping method
it can be shown that when @xmath280 this system has only zero solution , i.e. @xmath286 for @xmath287 satisfying @xmath288 .
because of the arbitrariness of @xmath285 , @xmath289 and @xmath290 vanish as a function of @xmath291 and therefore , by ( [ tsarev ] ) so does @xmath292 .
now we suppose the theorem true for genus @xmath2 and we proof it for genus @xmath9 .
we fix @xmath293 and we consider the equation ( [ tsarev ] ) for @xmath294 and @xmath295 with boundary condition ( [ b1]-[b2 ] ) , namely rll w^g_1&=&a_1(2g+1)[w^g_1-w^g_2g+1 ] + w^g_2g+1&=&a_(2g+1)1[w^g_2g+1-w^g_1 ] + w^g_1(u_1,u_2, ,u_2g , u_2g)&= & w^g-1_1(u_1,u_2, ,u_2g-1,_2g,_2g)0 + w^g_2g+1(u_2,u_2, ,u_2g , u_2g+1)&= & w^g-1_2g+1(_2,_2,u_3, ,u_2g , u_2g+1)0 . repeating the arguments developed for genus @xmath64 we may conclude that @xmath296 , for arbitrary @xmath282 . we then repeat the above argument fixing @xmath297 and considering the equations ( [ tsarev ] ) for @xmath298 and @xmath299 , namely rll w^g_2&=&a_2(2g)[w^g_2-w^g_2 g ] + w^g_2g&=&a_(2g)2[w^g_2g - w^g_2 ] + w^g_2(u_1,u_2, ,u_2g-1,u_2g+1,u_2g+1)&= & w^g-1_2(u_1,u_2, ,u_2g-1,_2g+1,_2g+1)0 + w^g_2g(u_1,u_1,u_3, ,u_2g , u_2g+1)&= & w^g-1_2g(_1,_1,u_3, ,u_2g , u_2g+1)0 . it can be easily shown that also @xmath300 for arbitrary @xmath282 . repeating these arguments other @xmath301 times
we conclude that @xmath281 for @xmath302 and for arbitrary @xmath282 . applying
( [ tsarev ] ) and the boundary conditions ( [ b1])-([b2 ] ) we can prove that also @xmath303 is identically zero .
the theorem is then proved.@xmath158 .2 cm in the next section we consider the problem of reality of the solution of the hodograph transform ( [ ch ] ) .
the boundary conditions ( [ b1]-[b4 ] ) guarantee the @xmath29-smoothness of the solution of the whitham equations .
indeed it can be proved that the @xmath6 derivatives @xmath304 , @xmath305 , are continuous on the phase transition boundaries .
in this section we give an upper bound to the genus of the solution of the whitham equations for the monotone increasing smooth initial data @xmath38 . [ main ] if the monotone increasing smooth initial data @xmath24 satisfies the condition @xmath306 for all @xmath307 except at one point , then the solution of the whitham equations ( [ whithamg ] ) has genus at most @xmath56 .
* proof*:the solution of the whitham equations ( [ whithamg ] ) for different @xmath32 determines a decomposition of the @xmath200 plane , @xmath5 , into a number of domains @xmath308 with g=0,1,2 , ( see fig . [ bbb ] ) . to the inner part of each domain
@xmath308 it corresponds the @xmath9-phase solution @xmath31 of the whitham equations ( [ whithamg ] ) .
the common boundaries of the domains @xmath308 , @xmath32 , and @xmath309 , @xmath310 , are the points of phase transition between the @xmath9-phase solution and the @xmath311-phase solution .
we will show that each domain @xmath308 , @xmath3 , does not have a common boundary with any of the domains @xmath312 , @xmath313 . since the set of domains @xmath314 is not empty because @xmath315 the set @xmath316 must be empty .
indeed on the contrary the @xmath200 plane , @xmath5 , which is a connected set , would be split into a number of domains whose union forms a disconnected set . before determining the boundaries of the domains @xmath308 , @xmath32
, we need to study more in detail the hodograph transform ( [ ch ] ) .
[ zerophi ] let us consider the polynomial @xmath317 where @xmath318 and @xmath319 , @xmath0 , is given by the expression @xmath320 with the polynomials @xmath321 , @xmath211 , defined in ( [ b6 ] ) and the functions @xmath217 , @xmath322 , defined in ( [ qk ] ) .
then the hodograph transform ( [ ch ] ) is equivalent , for @xmath0 , to the equation @xmath323 * proof*:we observe that the @xmath55 s defined in ( [ wi ] ) are given by the ratio @xmath324 , @xmath49 , where @xmath319 is the polynomial defined in ( [ r2 g ] ) . hence we can write the hodograph transform ( [ ch ] ) in the form @xmath325_{r = u_i}=0 , \quad i=1,\dots,2g+1 . \;\;\ ] ] for @xmath0 , @xmath326 is a polynomial of degree @xmath327 and because of ( [ sol3 ] ) it must have at least @xmath328 real zeros .
therefore it is identically zero .
hence for @xmath329 ( [ zero ] ) is equivalent to ( [ sol3 ] ) and ( [ ch ] ) .
@xmath158 .2 cm in the following analysis we give some conditions for the existence of a real solution @xmath330 of the hodograph transform ( [ ch ] ) . for the purpose let us consider the function @xmath216 which solves the boundary value problem [ bpsi ] \
{ lll & & ^g(r;)- ^g(r;)=2(u_i - u_j ) ^g(r;),ij , i , j=1, 2g+1 + & & ^g(r;)-2 ^g(r;)=2(r - u_j ) ^g(r;),j=1, 2g+1 + & & ^g(r;)=
f^(g)(r ) . where @xmath331 is the @xmath9th derivative of the smooth monotone increasing initial data @xmath24 . from the results of sec .
[ slos ] we are able to integrate ( [ bpsi ] ) obtaining @xmath332 where @xmath333 and the @xmath241 s have been defined in ( [ crho ] ) .
the function @xmath216 is symmetric with respect to the variables @xmath334 .
for the initial data ( [ b18 ] ) , the expression of @xmath216 in ( [ psi ] ) is equivalent to ( [ psia ] ) .
[ zphi ] let us consider the function @xmath335 where @xmath216 has been defined in ( [ psi ] ) .
if @xmath37 satisfy the whitham equations ( [ whithamg ] ) then the function @xmath336 has , in the @xmath151 variable , at least one real zero in each of the intervals @xmath337 , @xmath0
. * proof*:if the functions @xmath37 satisfy the whitham equations then by proposition [ zerophi ] the polynomial @xmath338 .
therefore @xmath339 in the third equality of ( [ ppp ] ) we have used the fact that @xmath340 because of the normalization conditions ( [ norm2 ] ) .
the function @xmath216 satisfies the relations @xmath341 which can be easily obtained from ( [ psi ] ) . using ( [ psir ] )
we can rewrite the last term in ( [ ppp ] ) in the form @xmath342 where the last equality has been obtained integrating by parts . using the definition of @xmath336 in ( [ phi ] )
we rewrite the above relation in the form @xmath343 relation ( [ normphi ] ) is satisfied only if the function @xmath336 changes sign at least once in each of the intervals @xmath344 , @xmath125 .
@xmath158 .2 cm in the following we are going to determine the equations which describe , on the @xmath345 plane , the boundary between the domains @xmath308 and @xmath346 .
the boundary between the domains @xmath308 and @xmath346 represents a singular behavior in the solution of the @xmath9 or @xmath347-phase equations . as we have shown in the example on fig [ ivp ] , the boundary between the domains @xmath348 and @xmath349 is described by the curves @xmath350 where @xmath46 and @xmath44 respectively and by the point @xmath351 , @xmath352 of gradient catastrophe of the zero phase solution . in a similar way the generic boundary between the domains @xmath308 and @xmath346 is described by \a ) the curves @xmath353 where two riemann invariants of the @xmath347-phase solution coalesce ; \b ) the point of gradient catastrophe @xmath354 , @xmath355 of the @xmath9-phase solution , namely the point where one of the @xmath328 riemann invariants has a vertical inflection point .
the equations determining the point of gradient catastrophe of the @xmath9-phase solution can also be obtained considering the limit of the @xmath1 phase solution when three riemann invariants coalesce . to treat case a ) we consider the riemann surface @xmath356 of genus @xmath1 given by the equations @xmath357 @xmath358 where @xmath359 , and @xmath360 .
the riemann invariants are the @xmath361 variables @xmath362 , @xmath363 .
we suppose @xmath364 .
the hodograph transform ( [ ch ] ) for these @xmath361 variables has two different behavior for @xmath365 when @xmath366 belongs to one of the bands @xmath367 or gaps @xmath368 we call _ leading edge _ of the phase transition boundary the case in which @xmath366 belongs to the bands .
we call _ trailing edge _ of the phase transition boundary the case in which @xmath366 belongs to the gaps ( [ gap ] ) .
[ lt ] the leading edge of the phase transition boundary between the @xmath9-phase solution and the @xmath347-phase solution is described by the system [ y1 ] \ { lll ^g(v;)-6t_g0=0 & & + _ v^g(v;)=0 & & + x=_r = u_i&&,i=1, ,2g+1 , g0 + .
where @xmath369 , the function @xmath370 for @xmath160 and zero otherwise , the function @xmath336 has been defined in ( [ phi ] ) and the polynomial @xmath371 has been defined in ( [ r2 g ] ) .
we assume @xmath372 and @xmath373 , @xmath49 , on the solution of ( [ y1 ] ) . system ( [ y1 ] ) is a system of @xmath361 equations in @xmath374 unknowns @xmath375 and @xmath8 .
if system ( [ y1 ] ) is uniquely solvable for real @xmath376 and @xmath8 as a function of @xmath5 , then a phase transition between the @xmath9-phase solution and the @xmath347-phase solution occurs .
the curve @xmath377 describes on the @xmath345 plane the boundary between the domains @xmath308 and @xmath346 associated to the leading edge .
the conditions @xmath372 and @xmath373 , @xmath49 , on the solution of ( [ y1 ] ) exclude higher order degeneracy in the transition .
indeed in such case it can be proved that in a neighborhood of the solution @xmath378 , @xmath379 , @xmath380 of ( [ y1 ] ) the @xmath347-phase solution is uniquely defined .
[ tt ] the trailing edge of the phase transition boundary between the @xmath9-phase solution and the @xmath347-phase solution is described by the solution of the system [ t1 ] \ { lll ^g(v;)-6t_g0=0 & & + & & + _ v^u_2j-1(^g(r;)-6t_g0)(r)dr=0 & & + & & + x=_r = u_i , i=1, ,2g+1,g>0 + . where @xmath381 and the function @xmath336 has been defined in ( [ phi ] ) .
we assume @xmath382 and @xmath373 , @xmath49 , on the solution of ( [ t1 ] ) .
if system ( [ t1 ] ) is uniquely solvable for real @xmath376 and @xmath8 as a function of @xmath5 , then a phase transition between the @xmath9-phase solution and the @xmath347-phase solution occurs .
the curve @xmath383 describes on the @xmath345 plane the boundary between the domains @xmath308 and @xmath346 associated to the trailing edge .
the following theorem enables one to determine a point of gradient catastrophe of the @xmath9-phase solution .
this point can be obtained either as a limit of the @xmath347 phase solution when three riemann invariants coalesce or imposing a vertical inflection point on the @xmath9-phase solution .
[ cc ] let us consider the system [ cce ] \ { lll _
r^g(r;)|_r = u_l=0 & & + ^g(u_l;)-6t_g0=0 & & + x=_r = u_i , i=1,
,2g+1 , g0 + .
where @xmath370 for @xmath160 and zero otherwise , the function @xmath336 has been defined in ( [ phi ] ) .
when system ( [ cce ] ) is uniquely solvable for @xmath354 , @xmath384 and @xmath385 , then the @xmath9-phase solution has a point of gradient catastrophe on the @xmath386 branch , @xmath387 . to avoid higher order degeneracies in the transition
we impose the condition @xmath388 indeed we can prove that the above condition guarantees that the genus of solution of the whitham equations increases at most by one in the neighborhood of the point of gradient catastrophe .
therefore it is legitimate to consider the point of gradient catastrophe that solve ( [ cce ] ) and satisfies ( [ third ] ) as a point of the boundary between the domains @xmath308 and @xmath346 . the condition ( [ third ] ) in not essential in the genus @xmath160 case as illustrated in theorem [ tian ] .
we observe that both systems ( [ y1 ] ) and ( [ t1 ] ) in the limit @xmath389 , @xmath390 , coincide with system ( [ cce ] ) .
theorems [ lt ] , [ tt ] and [ cc ] characterize all types of boundaries between the @xmath9-phase solution and the @xmath347-phase solution , namely the leading edge , the trailing edge and the points of gradient catastrophe .
we will prove these theorems in the next section .
.3 cm * example 3.1 * for @xmath391 , @xmath392 , the solution of the whitham equations has genus at most equal to one @xcite,@xcite . on the @xmath200 , @xmath5 ,
plane we have only the domains @xmath348 and @xmath349 .
the one phase solution is defined within the cusp @xmath393 , where @xmath394 are two real constants and @xmath35 .
the point @xmath395 is the point of gradient catastrophe of the zero - phase solution .
the constants @xmath396 and @xmath397 can be computed explicitly . from ( [ psi ] ) and ( [ phi ] ) we obtain @xmath398 on the leading edge where @xmath399 and @xmath400 , @xmath401 , ( [ y1 ] ) is equivalent to the equations @xmath402 on the trailing edge where @xmath14 , @xmath403 , @xmath404 , system ( [ t1 ] ) is equivalent , for @xmath160 to @xmath405 equations ( [ ll ] ) and ( [ t00 ] ) have already been obtained in @xcite . solving system ( [ ll ] ) for @xmath34 when @xmath406 , @xmath392 we obtain @xmath396 @xcite,@xcite x_-(k)=-6(2z_-(k)-1 ) ^ , k=3,5,7 , , where @xmath407 is the unique real solution of @xmath408 . here
@xmath409 is the hypergeometric series .
the quantity @xmath397 is obtained from ( [ t00 ] ) x_+(k)=2(2z_+(k)-3)^ , k=3,5,7 , , where the number @xmath410 is the unique real solution of the equation @xmath411 .
we give some numerical values : @xmath412 .
0.3 cm theorems [ lt ] , [ tt ] and [ cc ] can be generalized to the case of multiple phase transitions .
[ tmultiple ] the transition boundary between the domains @xmath308 , @xmath32 and @xmath309 , @xmath413 , having @xmath414 leading edges and @xmath415 trailing edges and @xmath416 points of gradient catastrophes , @xmath417 , @xmath418 , is described by the solution of the system [ multiple ] \ { lll _
v_k^g(v_k;)=0,k=1 , , n_1 & & + ^g(v_k;)-6t_g0=0 , k=1 , , n_1 & & + ^g(y_l;)-6t_g0=0,l=1, ,n_2 & & + _ y_l^u_2j_l-1(^g(r;)-6t_g0)(r)dr=0,l=1 , ,
n_2 & & + _ r^g(r;)|_r = u_j_m=0,m=1, ,n_3 & & + ^g(u_j_m;)-6t_g0=0m=1, ,n_3 & & + & & + x=_r = u_i , i=1, ,2g+1 , + . where @xmath419 , @xmath420 , @xmath421 and @xmath422 , @xmath423 , @xmath424 .
the transition between the @xmath9-phase solution and the @xmath425phase solution , @xmath413 , is highly non generic ( cfr .
fig [ multiple ] ) . indeed system ( [ multiple ] ) is a systems of @xmath426 equations in @xmath427 unknowns @xmath428 , @xmath429 , @xmath12 , @xmath6 and @xmath5 . therefore if system ( [ multiple ] )
admits a real solution , @xmath430 variables are functions of all the others .
theorem [ tmultiple ] includes all the degenerate cases that have been excluded in theorems [ lt ] , [ tt ] and [ cc ] . for example
let us consider a transition with a double - leading edge described by the variables @xmath431 , @xmath432 and @xmath8 which solve ( [ multiple ] ) with @xmath433 , @xmath434 . in the limit @xmath435
we get a single degenerate leading edge .
the variables @xmath366 , @xmath8 satisfy ( [ lt ] ) and the equations @xmath436 and @xmath437 .
therefore we regard such degenerate single leading edge as a point of the boundary of the domains @xmath308 and @xmath438 . [ propo ] if the domains @xmath308 and @xmath346 have a common boundary , then the function @xmath439 has at least @xmath440 real zeros in the @xmath151 variable for @xmath37 satisfying ( [ zero ] ) and for @xmath35 . * proof * : the domains @xmath308 and the @xmath346 have a common boundary if one of the systems ( [ y1 ] ) , ( [ t1 ] ) or ( [ cce ] ) is solvable for some @xmath35 .
we first consider the leading edge . for @xmath160
the statement is obvious from system ( [ y1 ] ) . for @xmath0 system ( [ y1 ] ) can have a solution if the function @xmath336 has a double zero at @xmath441 when @xmath366 belongs to the bands ( [ band ] ) . combining this observation with proposition [ zphi ]
we immediately obtain the statement .
as regarding the trailing edge , the theorem is obvious for @xmath160 . for @xmath0 and @xmath442 system ( [ t1 ] ) can be satisfied if the function @xmath336 has at least two zeros in the interval @xmath443 . combining this observation with proposition [ zphi ]
we obtain the statement .
when @xmath444 , @xmath445 , ( [ normphi ] ) and ( [ t1 ] ) are satisfied if the function @xmath336 has at least three real zeros in the interval @xmath446 , @xmath445 , and changes sign at least once in each of the intervals @xmath344 , @xmath125 , and @xmath447
. therefore @xmath336 has at least @xmath440 real zeros at the trailing edge .
if the point of the boundary between the domains @xmath308 and the @xmath346 corresponds to a point of gradient catastrophe of the @xmath9-phase solution the statement is obvious from system ( [ cce ] ) and proposition [ zphi ] .
therefore , on the phase transition boundary between the domains @xmath308 and @xmath346 the function @xmath336 has at least @xmath440 real zeros in the @xmath151 variable , @xmath448 , for @xmath31 and for @xmath35 . .
@xmath158 .2 cm proposition [ propo ] can be generalized to multiple phase transitions .
[ propom ] on the phase transition boundary between the domains @xmath308 and the @xmath309 the function @xmath449 has at least @xmath450 real zeros in the @xmath151 variable for @xmath31 and for @xmath35 .
the proof is analogous to that of proposition [ propo ] .
[ ngzero ] if the smooth initial data @xmath38 satisfies ( [ condition ] ) , then the function @xmath451 has at most @xmath452 real zeros ( counting multiplicity ) in the @xmath151 variable for @xmath453 and for any real @xmath6 , @xmath5 and @xmath282 .
* proof*:for proving the lemma we need the following elementary result .
if the real smooth function @xmath454 satisfies the condition @xmath455 for all @xmath151 belonging to the domain of @xmath456 , then @xmath454 has at most @xmath457 real zero . using ( [ condition ] ) , ( [ bpsi ] ) and ( [ phi ] )
we obtain @xmath458 for @xmath0 , for any real @xmath151 and for any fixed real @xmath192 belonging to the interval @xmath19 .
therefore @xmath336 has at most @xmath452 real zeros in the @xmath151 variable for @xmath453 and for any fixed real @xmath459 .
@xmath158 .2 cm [ nzero ] if the smooth monotone increasing initial data @xmath38 satisfies ( [ condition ] ) , then the function @xmath460 has at most @xmath461 real zeros ( counting multiplicity ) in the @xmath151 variable , for any @xmath5 and @xmath462 , where @xmath195 is the unique solution of the equation @xmath463 . for what @xmath464
is concerned we have two possibilities : either it has at most @xmath461 real zeros ( counting multiplicity ) in the @xmath151 variable for any @xmath5 and some of these zeros are distinct , or it has at most two real zeros in the @xmath151 variable for any @xmath35 except for @xmath465 where @xmath466 is a zero of multiplicity higher than @xmath461 . * proof*:since @xmath467 and because of ( [ condition ] ) we have two possibilities : either @xmath468 has at most @xmath56 simple minima and some of these minima are distinct , or @xmath468 is a non negative convex function with a minimum at @xmath198 , where @xmath195 is the unique solution of the equation @xmath463 . from the above considerations the lemma easily follows .
@xmath158 .2 cm we remark that when @xmath468 is a convex function , we can apply tian s theorem [ tian ] . indeed in such case
we have @xmath197 for all @xmath462 and therefore the solution of the whitham equations has genus at most one .
.2 cm we continue the proof of the second main theorem . in
the following we exclude the case in which @xmath468 is a convex function . from proposition [ propom ]
we deduce that when the domains @xmath308 and @xmath469 , @xmath470 , have a common boundary the function @xmath471 defined in ( [ phi ] ) has at least @xmath472 real zeros for some values of @xmath6 and @xmath34 . from lemma [ ngzero ] the function @xmath473 defined in ( [ phi ] )
has at most @xmath452 real zeros for @xmath474 when the initial data @xmath24 satisfies ( [ condition ] ) .
therefore @xmath475 this shows that the set of domains @xmath476 does not have common boundaries with any of the domains in the set @xmath477 .
the second main theorem is proved .
@xmath158 .2 cm in this subsection we prove theorems [ lt ] , [ tt ] , [ cc ] and ( [ tmultiple ] ) . in the following
we denote with a sup - script the genus of the quantity we are referring to . whenever we omit the sup - script we are referring to genus @xmath9 quantity .
we denote with @xmath478 the normalized abelian differential of the second kind with pole at infinity of order @xmath83 defined on the surface @xmath356 of genus @xmath1
@xmath357 @xmath479 where @xmath359 , and @xmath360 .
we define the polynomials @xmath480 the expression of the polynomial in ( [ r2 g ] ) for the riemann invariants @xmath481 , @xmath482 and @xmath8 reads @xmath483 where @xmath484 , @xmath485 and @xmath486 , @xmath188 . we will sometimes omit the explicit dependence of @xmath487 , @xmath488 and @xmath489 on the parameters @xmath366 and @xmath490 .
* proof of theorem [ lt]*. we write the @xmath347-phase solution ( [ ch ] ) for the variables @xmath481 , @xmath482 , @xmath363 in the form [ wig1 ] \ { lll 0&=&(- ) , + & & + x&=&_r=_2 , + & & + x&=&_r = u_i , i=1, ,2g+1 + .
the reason to write the hodograph transform ( [ ch ] ) in the form ( [ wig1 ] ) is that system ( [ wig1 ] ) is non degenerate at the phase transition @xmath491 .
we show that system ( [ wig1 ] ) reduces to system ( [ y1 ] ) when @xmath491 or @xmath492 .
it is clear from theorem [ theot ] and proposition [ zerophi ] that the last @xmath328 equations in ( [ wig1 ] ) in the limit @xmath491 reduce to those of system ( [ y1 ] ) . for computing the limit @xmath493 of the first two equations in ( [ wig1 ] )
we first need to study the behavior of the differentials @xmath487 , @xmath82 , when @xmath494 .
when @xmath366 belongs to one of the bands ( [ band ] ) , the expansion of @xmath487 for @xmath365 reads @xcite @xmath495 where @xmath496 is the normalized differential of the second kind defined on @xmath66 with pole at infinity of order @xmath83 , @xmath497 and @xmath498 .
the differential @xmath499 is the normalized abelian differential of the third kind with poles at the points @xmath500 with residue @xmath501 respectively .
the explicit expression of @xmath499 has been given in ( [ ck ] ) the differential @xmath502 has a pole at @xmath441 of order @xmath503 and zero residue . from ( [ dplead ] ) we can get the expansion of the polynomial @xmath504 , namely @xmath505 where @xmath250 has been defined in ( [ b6 ] ) and the @xmath506 s have been defined in ( [ norm1 ] ) . from ( [ dpleadp ] )
we can evaluate the following @xmath507 to evaluate the first two equations in ( [ wig1 ] ) at the point of phase transition we need also the following relations @xmath508 where the function @xmath216 has been defined in ( [ psi ] ) . from ( [ z1]-[z2 ] ) , ( [ plead ] ) and ( [ qpsi ] )
we obtain @xmath509 from ( [ mu ] ) , ( [ psir ] ) and using the definition of @xmath336 in ( [ phi ] ) the above relation can be written in the form @xmath510 where @xmath511 has been defined in ( [ r2 g ] ) .
we need to consider also the quantity @xmath512_{\tilde{u}_1=
\tilde{u}_2}= 8\left.\left(\sqrt\d\dfrac{\mu^2(v+\sqrt\d)}{p_0^{g+1}(v+\sqrt\d,\d)}\partial_{\tilde{u}_1}\partial_{\tilde{u}_2}q^{g+1}_{g+1}(\tilde{u}_1,\tilde{u}_2,{{\vec{u}}})\right)\right|_{\d=0}\\ + & 2\left.\left(\left(\dfrac{\mu^2(v+\sqrt\d)}{p_0^{g+1}(v+\sqrt\d,\d)}-\dfrac{\mu^2(v-\sqrt\d)}{p_0^{g+1}(v-\sqrt\d,\d)}\right)\partial_{\tilde{u}_2}q^{g+1}_{g+1}(\tilde{u}_1,\tilde{u}_2,{{\vec{u}}})\right)\right|_{\d=0}\\ + & \sum_{k=1}^{g}q^g_k({{\vec{u } } } ) \sum_{l=1}^k(2l-1 ) \tilde{\gamma}^{g}_{k - l}({{\vec{u}}})\left.\left(\dfrac{1}{2\sqrt\d}\left(\dfrac{p^{g+1}_{l-1}(v+\sqrt\d,\d ) } { p^{g+1}_0(v+\sqrt\d,\d)}-\dfrac{p^{g+1}_{l-1}(v-\sqrt\d,\d ) } { p^{g+1}_0(v-\sqrt\d,\d)}\right)\right ) \right|_{\d=0}\\ + & q_{g+1}^{g+1}(v , v,{{\vec{u}}})\sum_{l=1}^{g+1}(2l-1 ) \tilde{\gamma}^{g}_{g+1-l}({{\vec{u}}})\left.\left(\dfrac{1}{2\sqrt\d}\left(\dfrac{p^{g+1}_{l-1}(v+\sqrt\d,\d)}{p^{g+1}_0(v+\sqrt\d,\d)}-\dfrac{p^{g+1}_{l-1}(v-\sqrt\d,\d ) } { p^{g+1}_0(v-\sqrt\d,\d)}\right)\right ) \right|_{\d=0 } \end{split}\ ] ] where we have used ( [ z1]-[z2 ] ) to obtain the right hand side . using ( [ mu ] ) ,
( [ psir ] ) , ( [ plead ] ) and ( [ qpsi ] ) the above reduces to the form @xmath513_{\tilde{u}_1= \tilde{u}_2}=\dfrac{\partial}{\partial_v}\left(\dfrac{2\mu^2(v ) \phi^g(v)+ r^{g}(v)}{p^{g}_0(v)}\right).\ ] ] from ( [ plead ] ) , ( [ h1 ] ) and ( [ h2 ] ) , the system ( [ wig1 ] ) on the point of phase transition @xmath514 reads [ wg1 ] \ { lll 0&= & ( ) + & & + x&= & + & & + x&=&_r = u_i , i=1,
,2g+1 + . from proposition [ zerop ]
the last @xmath328 equations in ( [ wg1 ] ) are equivalent to the condition @xmath515 therefore system ( [ wg1 ] ) is equivalent to system ( [ y1 ] ) for @xmath0 . for @xmath160 substituting ( [ zerodiff ] ) in ( [ wg1 ] ) we obtain [ y00 ] \ { lll 0= _ v(-6tu+f(u)+2(v - u)(_0(v , u)-6 t ) ) & & + & & + x=-6tu+f(u)+2(v - u)(_0(v , u)-6 t ) & & + & & + x=-6tu+f(u ) & & + . where @xmath14 and @xmath24 is the initial data .
it is clear that ( [ y00 ] ) is equivalent to ( [ y1 ] ) for @xmath160 .
@xmath158 .2 cm * proof of theorem [ tt]*. for getting the equations determining the trailing edge we have to repeat all the above calculation in a slightly different way .
we write the @xmath347-phase solution for the variables @xmath481 , @xmath482 , @xmath363 in the form [ ug1 ] \ { lll 0&=&(12 t ( -)- + ) , + & & + x&=&_r=_2 , + & & + x&=&_r = u_i , i=1, ,2g+1 + .
the reason to write the hodograph transform in the form ( [ ug1 ] ) is that system ( [ ug1 ] ) is non degenerate at the phase transition @xmath514 when @xmath516 . from theorem [ theot ] and proposition [ zerophi ]
we deduce that the last @xmath328 equations in ( [ ug1 ] ) in the limit @xmath491 reduce to those of system ( [ t1 ] ) .
next we investigate the behavior of the abelian differentials of the second kind @xmath487 , @xmath82 , in the limit @xmath365 , when @xmath366 belongs to one of the gaps ( [ gap ] ) .
we have that @xcite @xmath517 where @xmath518 has been defined in ( [ ck ] ) and @xmath500 . from ( [ dptrail ] ) we can obtain the expansion for the polynomial @xmath519 , namely @xmath520 so that @xmath521 using ( [ r2 g ] ) , ( [ z1]-[z2 ] ) and ( [ ptrail ] ) we have that @xmath522 adding and subtracting @xmath523 to ( [ temp ] ) and using ( [ mu ] ) and ( [ psir ] ) we obtain @xmath524 where @xmath336 and @xmath319 have been defined in ( [ phi ] ) and ( [ r2 g ] ) respectively . we observe that @xmath525 for all @xmath366 belonging to the gaps ( [ gap ] ) .
we need to consider also the quantity in the first equation of ( [ ug1 ] ) namely @xmath526 using ( [ mu ] ) , ( [ psir ] ) , ( [ ptrail ] ) and ( [ qpsi ] ) we simplify the above relation to the form @xmath527 using repeatedly ( [ psir ] ) we can write the above relation in the form @xmath528 where @xmath336 and @xmath319 have been defined in ( [ phi ] ) and ( [ r2 g ] ) respectively . from ( [ ptrail ] ) , ( [ d1 ] ) and ( [ d2 ] ) system ( [ ug1 ] ) can be reduced to the form [ ug2 ] \ { lll 0= 2 ^ 2(v)^g(v;)+z^g(v)+ _ q^-(v)^q^+(v)d & & + & & + 0=_q^-(v)^q^+(v)d & & + & & + x=_r = u_i , i=1,
,2g+1 & & + . where the polynomial @xmath529 has been defined in ( [ zeta ] ) . from proposition [ zerop ]
the last @xmath328 equations in ( [ ug2 ] ) are equivalent to the condition @xmath530 therefore system ( [ ug2 ] ) is equivalent for @xmath0 to the system [ ug3 ] \ { lll 0=^g(v ; ) & & + & & + 0=_q^-(v)^q^+(v)()^g ( ; ) d & & + & & + x=_r = u_i , i=1, ,2g+1 .
& & + . because of ( [ normphi ] ) , when @xmath531 , we can split the integral of the second equation of ( [ ug3 ] ) in the form @xmath532 therefore ( [ ug3 ] ) is equivalent to ( [ t1 ] ) for @xmath0 . for @xmath160 substituting ( [ zerodiff ] ) in ( [ ug2 ] )
it is easy to check that we obtain a system equivalent to ( [ t1 ] ) .
@xmath158 .2 cm 0.5 cm * proof of theorem [ cc]*. the equations describing the point of gradient catastrophe of the @xmath9-phase solution can be obtained either considering the limit of the @xmath533phase solution when three riemann invariants coalesce , or supposing that one of the @xmath328 distinct riemann invariants of the @xmath9-phase solution has a vertical inflection point for @xmath35 . for proving theorem [ cc ]
we follow the latter possibility . on the solution of ( [ ch ] )
@xmath534 @xcite therefore a point of gradient catastrophe of the @xmath9-phase solution is determined by the system [ cc1 ] \ { lll _ u_l(-_lt+w_l)=0 & & + & & + ( _ u_l)^2(-_lt+w_l)=0 & & + & & + x=_r = u_i , i=1, ,2g+1 + . where @xmath387 .
we show that system ( [ cc1 ] ) is equivalent to system ( [ cce ] ) . for the purpose
we compute explicitly the derivative @xmath535 . from a generalization of a result in @xcite
we obtain : @xmath536 so that @xmath537 in the above relation we need to compute the derivative @xmath538 where @xmath539 . for computing the derivatives of the normalization constants
@xmath540 in @xmath541 we need the following proposition .
@xcite [ dubrovin ] let @xmath542 and @xmath543 two normalized abelian differentials on @xmath66 .
let be @xmath544 the local coordinate at infinity and @xmath545 define the bilinear product @xmath546 then @xmath547\dfrac{\om_1(r)\om_2(r)}{dr},\quad i=1,\dots , 2g+1,\ ] ] where @xmath548\dfrac{\om_1(r)\om_2(r)}{dr}$ ] is the residue of the differential @xmath549 evaluated at @xmath550 . applying the above proposition to @xmath551 and @xmath552 , @xmath553 we obtain after non trivial simplifications @xmath554 where the @xmath141 s have been defined in ( [ gt ] ) . from the above formula we obtain @xmath555 using the relations ( [ relations ] ) , ( [ dgt ] ) , ( [ partiall ] ) and ( [ partialp ] ) we simplify ( [ asd ] ) to the form @xmath556 where the polynomial @xmath529 has been defined in ( [ zeta ] ) . applying in the second term of ( [ aq ] ) the relations ( [ psir ] ) we can rewrite ( [ aq ] ) in the compact form @xmath557 where @xmath336 has been defined in ( [ phi ] ) . from proposition [ zerop ]
the last @xmath328 equations in ( [ cc1 ] ) are equivalent to the condition @xmath530 therefore we can simplify ( [ aw1 ] ) to the form @xmath558 when @xmath8 satisfy the @xmath9-phase whitham equations . as regarding the second derivative @xmath559 , from ( [ dd1 ] )
we obtain @xmath560 observing that @xmath561 we obtain the relation @xmath562 .
therefore @xmath563 from ( [ dd1]-[dd2 ] ) and the fact that @xmath564 for @xmath8 , it is clear that system ( [ cc1 ] ) is equivalent to system ( [ cce ] ) .
@xmath158 .2 cm we remark that to avoid higher order degeneracies in system ( [ cce ] ) , we impose the condition @xmath565 indeed we prove that the above condition guarantees that the genus of solution of the whitham equations increases at most by one in the neighborhood of the point of gradient catastrophe .
it is sufficient to show that the transition to genus @xmath440 does not occur . for the purpose
let us suppose @xmath566 even and let us consider the function @xmath567 by proposition [ zphi ] in order to have a genus @xmath440 solution in the neighborhood of the point of gradient catastrophe , the function @xmath568 has to change sign in each of the intervals @xmath569 and @xmath570 ( @xmath566 even ) for arbitrary small @xmath571 . because of ( [ secondd ] ) @xmath572 for sufficiently small @xmath571 and for @xmath573 .
therefore the transition to genus @xmath440 does not occur .
we can exclude transitions from genus @xmath9 to genus @xmath574 , @xmath575 , by perturbations arguments .
therefore it is legitimate to consider the point of gradient catastrophe that solve ( [ cce ] ) and satisfies ( [ terzo ] ) as a point of the boundary between the domains @xmath308 and @xmath346 .
we remark that in the genus @xmath160 case , the condition ( [ secondd ] ) is not essential as illustrated by theorem ( [ tian ] ) 0.3 cm * proof of theorem ( [ tmultiple ] ) * a phase transition may occur between the zero - phase solution and the two - phase solution .
as shown on figure [ figpt ] we can have a double leading edge , a trailing - leading edge and a double trailing edge .
there are also other types of boundary between the zero - phase solution and the two - phase solution which we call `` point of gradient catastrophe @xmath576 leading edge '' and `` point of gradient catastrophe @xmath576 trailing edge '' .
multiple transitions may also occur between the @xmath9-phase solution and the @xmath577-phase solution . in order to determine the systems which describe such phase transitions we consider the riemann surface @xmath578 of genus @xmath579 defined by @xmath580 @xmath581 where @xmath582 , @xmath583 , @xmath49 .
then we study the hodograph transform for the distinct variables @xmath584 , @xmath585 and @xmath12 in the _ independent _ limits @xmath586 and @xmath587 .
when @xmath588 and @xmath432 belong to the bands ( [ band ] ) , we are considering the double leading edge . repeating the calculations done for the single leading edge we can determine the equations which describe the phase transition for the double leading edge , namely [ doubleleadeq ] \ { lll _ v_1^g(v_1;)=0 & & + ^g(v_1;)=0 & & + _ v_2^g(v_2;)=0 & & + ^g(v_2;)=0 & & + x=_r = u_i&&,i=1, ,2g+1 , g>0 + . from ( [ y1]-[t1 ] )
analogous systems can be obtained for the trailing - leading edge and double trailing edge . for studying the phase transitions between the @xmath9-phase solution and the @xmath311-phase solution , @xmath310 ,
having @xmath414 leading edges and @xmath415 trailing edges , @xmath589 we consider the riemann surface @xmath590 @xmath591 here @xmath592 , @xmath593 , belongs to the bands ( [ band ] ) and @xmath594 , @xmath595 , belongs to the gaps ( [ gap ] ) , @xmath596 , @xmath593 , and @xmath597 , @xmath595 .
we study the hodograph transform ( [ ch ] ) for the variables @xmath598 , @xmath599 , @xmath600 , @xmath595 and @xmath12 in the _ independent _ limits @xmath601 , @xmath599 and @xmath602 @xmath595 . repeating the calculations done for proving theorem [ lt ] and theorem [ tt ] it is easy to show that system ( [ multiple ] )
describes the phase transition between the @xmath9-phase solution and the @xmath311-phase solution having @xmath414 leading edges , @xmath415 trailing edges and no points of gradient catastrophe , @xmath603 . if we suppose that the @xmath9-phase solution has also @xmath416 points of gradient catastrophe , @xmath417 , then combining theorem [ lt ] , theorem [ tt ] and theorem [ cc ] we obtain proposition [ propom ] .
@xmath158 .2 cm
in this work we have constructed , in implicit form , the @xmath9-phase solution of the whitham equations for monotone increasing smooth initial data @xmath38 .
the goal is obtained solving the tsarev system ( [ tsarev ] ) .
we have shown that the solution of the tsarev system which satisfies the natural boundary conditions ( [ b1]-[b4 ] ) is unique .
then we have investigated the conditions for the solvability of the hodograph transform ( [ ch ] ) . for the purpose
we have derived all the equations which describe a phase transition of the solution of the whitham equations .
studying when phase transitions occur we have been able to prove the second main result of this work .
namely we have shown that when the initial data satisfies ( [ condition ] ) , the solution of the whitham equations has genus @xmath3 for all @xmath6 and @xmath5 .
it is still an open problem to effectively determine , on the @xmath345 plane , the function @xmath604 from the generic initial data @xmath38 .
.5 cm * acknowledgments .
* i am indebted with professor boris dubrovin who posed me the problem of this work and gave me many hints to reach the solution .
i am grateful to professor sergei novikov for his suggestions during the preparation of the manuscript .
this work was partially support by a cnr grant 203.01.70 and by a grant of s. novikov .
b. dubrovin , _ lectures on 2-d topological field theory _ , lectures notes in math .
1620 , springer - verlag berlin , heidelberg and new york 1996 .
krichever , the method of averaging for two dimensional integrable equations , _ funct .
appl . _ * 22*:200 - 213 ( 1988 ) . | we study the cauchy problem for the whitham modulation equations for monotone increasing smooth initial data .
the whitham equations are a collection of one - dimensional quasi - linear hyperbolic systems .
this collection of systems is enumerated by the genus g=0,1,2, of the corresponding hyperelliptic riemann surface .
each of these systems can be integrated by the so called hodograph transform introduced by tsarev .
a key step in the integration process is the solution of the tsarev linear overdetermined system . for each @xmath0
, we construct the unique solution of the tsarev system , which matches the genus @xmath1 and @xmath2 solutions on the transition boundaries .
next we characterize initial data such that the solution of the whitham equations has genus @xmath3 , @xmath4 , for all real @xmath5 and @xmath6
. 0.7 cm |
spin crossover in condensed matter physics is a transformation of a system with one spin @xmath0 at each lattice site into another state with spin @xmath1 induced by some external parameter like strong magnetic field , high pressure etc .
it accompanies by the energy level @xmath2 and @xmath3 crossing , where @xmath4 is the local energy of the magnetic ion with spin @xmath5 ( @xmath6 ) .
recently spin crossovers in magnetic oxides have been found under high pressure in @xmath7 @xcite , @xmath8 @xcite , @xmath9 @xcite , @xmath10 @xcite . below the curie temperature of magnetic order spin crossover
is accompanied by the sharp change of the magnetization , nevertheless it may be observed in the paramagnetic state like in @xmath11 @xcite as the sharp change of the xes satellite / main peak intensity ratio with pressure increase .
energy levels crossing results in the lost of analyticity in the energy spectrum at the critical point ( in the thermodynamic limit ) @xcite . near
the critical point adiabaticity breaks down and non - equilibrium phenomena associated with the drastically grown quantum fluctuations can drive the system away from the ground state .
the final result depends on how fast the transition occurs .
if the quench process is sufficiently fast , large numbers of topological defects are created and the final state , being characterized by mixture of high and low spin phases , can be essentially different from that been obtained as result of slow evolution .
qualitatively , quench dynamics can be described by the kibble - zurek theory of nonequilibrium phase transitions @xcite . in this paper
we consider quench dynamics in spin crossover induced by time - dependent pressure .
the paper is organized as follows . in sec .
ii , a general model of spin crossover under high pressure is introduced . in sec .
iii , we study quench dynamics .
we consider two cases : a ) pressure is a linear function of time ; b ) pressure is defined by a pulse of a given shape .
we conclude in sec .
iv with a discussion of our results .
the multielectron ion in a crystal field has the energies of terms for @xmath12 configurations determined numerically by the tanabe - sugano diagrams @xcite as a solution of the eigenvalue problem .
simple analytical calculations of the low energy terms with different spin value that is sufficient to study spin crossover has been done recently @xcite .
the crystal field parameter increase linearly with pressure @xmath13 .
thus the multielectron energies for spin @xmath0 and @xmath1 ( @xmath2 and @xmath3 ) are also linear functions of @xmath13 . to distinguish two different spin states in the lattice we introduce the ising pseudospin states @xmath14 and @xmath15 for @xmath16 and @xmath17 , where @xmath18 runs over all sites in the lattice .
thus we neglect the spin degeneracy of the @xmath19 terms but capture the possibility of energy level crossing that is the essential part of the spin crossover .
then , in the basis @xmath20 , @xmath15 , the hamiltonian of the system can be written as follows @xmath21 where @xmath22 , @xmath23 , and @xmath24 , @xmath25 are the identity and pauli matrices , respectively ; the hamiltonian of interaction between the spins being @xmath26 .
the @xmath26 includes the isotropic heisenberg term with the exchange interaction @xmath27 between nearest spins and the anisotropic term @xmath28 .
the interatomic interaction @xmath27 is negligibly small in comparison with the interatomic hund s coupling ( ratio @xmath29 ) .
thus its contribution to the localized spin energy @xmath2 and @xmath3 due to the effective molecular field can be neglected .
nevertheless the exchange interaction plays very important role : it results in the long range order and synchronize each spin in the same quantum state providing a cooperative behavior of the spin system .
if it were the ferromagnetic interaction , each spin at t=0 would have the maximal projection @xmath30 with integer magnetic moment @xmath31 . in all examples given above
there is the antiferromagnetic interaction .
the ground state of the isotropic heisenberg antiferromagnet has non integer local magnetic moment due to the quantum spin fluctuations .
it is known that for large spin @xmath30 effect of quantum fluctuations is less important then for small spin , and for @xmath7 spin is @xmath32 .
moreover the magnetic anisotropy additionally suppresses quantum fluctuations .
for example , in @xmath7 the anisotropy field is @xmath33 @xcite and the measured value of the effective moment @xmath34 is very close to the calculated for @xmath35 value @xmath36 .
thus we conclude that due to anisotropy the magnetic moment at @xmath37 has integer value ( of course it is a property of the magnetic insulator that does not hold for itinerant magnets ) , and due to exchange interaction all spins are in the same quantum state .
so spin crossover at @xmath37 is the transition of the whole crystal from one magnetically ordered state to another .
nevertheless the criterium of the transition can be found from consideration of the single ion energies crossover due to space uniform cooperative magnetic order .
anisotropic relativistic interactions , for example a spin - orbital interaction , are also important because can mix different spin states inside single ion .
we consider spin crossover far from the thermodynamic phase transition in the paramagnetic phase , it allows us to simplify this interaction and substitute the effect of exchange with the effective mean field . this mean field is spatially uniform for the ferromagnetic insulator or two - sublattice for the antiferromagnet one .
examples given above @xcite correspond to the anti- or ferrimagnetics . in any case this mean field just renormalizes the interionic multielectron energies @xmath2 and @xmath3 , and is irrelevant to the crossover phenomenon .
another interaction that is smaller then the exchange one is given by relativistic anisotropy contribution to the @xmath26 .
for example a spin - orbital interaction can mix different spin states inside single ion , and it occurs to be important in our problem . in what follows we will consider the simplified spatially uniform model @xcite .
motivation for this simplification is as follows . despite that spin crossover
is related to many body system , the essential features of its dynamics can be described by effective , landau - zener type , hamiltonian @xcite .
the hamiltonian of our model is given by @xmath38 , where @xmath39 the energy spectrum is given by @xmath40 .
both @xmath41 and @xmath42 are pressure dependent .
further we assume that the spin excitation gap is given by @xmath43 the crossover takes place at the point @xmath44 when @xmath45 .
the spin - orbit coupling @xmath46 ( with @xmath47 ) mixes the different spin states , and it plays the role of quantum fluctuations in our ising pseudospin basis . the dimensionless spin gap , @xmath48 , plays a role of relative temperature @xmath49 near the critical point @xmath50 @xcite .
we consider time dependent schrdinger equation for the hamiltonian ( [ eqh2a ] ) assuming for simplicity that the pressure is the linear function of time , @xmath51 . inserting this expression into eq .
( [ eq1a ] ) , we obtain @xmath52 .
the parameter @xmath53 depends on @xmath54 and can be written as @xmath55 .
let @xmath56 and @xmath57 are eigenstates of the operator @xmath25 , so that @xmath58 and @xmath59 . expressing a generic state vector as @xmath60 and @xmath61 satisfy the schrdinger equation with the time - dependent hamiltonian in the landau - zener ( lz ) form ( in units @xmath62 ) @xmath63 where @xmath64
. in terms of dimensionless scaled time @xmath65 with @xmath66 , the landau - zener model is described by the hamiltonian @xmath67 where @xmath68 is the dimensionless coupling constant .
writing @xmath69 , one can recast the schrdinger equation ( [ lz ] ) as @xmath70 here the time @xmath71 runs from the initial time @xmath72 , corresponding to the initial pressure @xmath73 , to final @xmath74 , corresponding to @xmath75 at the end of quench ( @xmath76 ) .
further we assume that @xmath77 , then time @xmath71 can be extended to @xmath78 , and the problem becomes fully equivalent to the lz problem . the energy spectrum of the hamiltonian ( [ h2 ] )
is given by @xmath79 , and its instantaneous eigenvectors can be written as @xmath80 where @xmath81 .
the energy gap between the ground and excited states equals @xmath82 . from eq .
( [ r ] ) it follows that while the ground state behaves at @xmath83 as @xmath84 and @xmath85 , the excited state behaves as follows : @xmath86 and @xmath87 .
the state @xmath88 corresponds to the high spin ( hs ) of the system and @xmath89 corresponds to the low spin ( ls ) .
thus , if the system initially was in the hs state , at the end of the evolution its ground state corresponds to the ls .
tunnelling between the positive and negative energy eigenstates , leading to the mixture of hs and ls , happens in the neighbourhood of the critical point @xmath90 ( @xmath91 ) when @xmath92 @xcite .
we assume further that the evolution of the system starts at the moment of time @xmath93 ( @xmath94 ) from the ground state @xmath95 .
since @xmath47 , we have @xmath96 .
this yields the following initial conditions : @xmath97 and @xmath98 . at the end of evolution
we obtain , @xmath99 , while @xmath100 . in figs .
[ e1 ] [ p2a ] we present the results of numerical solution of the schrdinger equation . in fig .
[ e1 ] , [ e1a ] time evolution of the bloch vector , @xmath101 , is shown .
the motion begins at the south pole of the two - dimensional sphere @xmath102 and for @xmath103 ends at the north pole .
however for the choice of the parameter @xmath104 , numerical simulation shows that the bloch vector never reaches the north pole , which corresponds to the ls state .
this implies that at the end of evolution the quantum system does not remains in the ground state and its final state is the mixture of the lh and ls states . in fig .
[ p2a ] the probability @xmath105 of transition @xmath57 @xmath106 @xmath107 is depicted for various values of @xmath108 .
as can be seen , with decreasing of adiabadicity parameter @xmath108 the transition probability decreases as well .
its asymptotic behaviour is described by the lz formula ( [ p ] ) .
the exact solution of the eq.([eq1 ] ) is given in terms of the parabolic cylinder functions@xcite , @xmath109 , where @xmath110 .
assuming that initially the system was in the ground state , @xmath84 , we obtain the initial conditions as follows : @xmath111 and @xmath112 .
the probability of transition to the state @xmath56 at time @xmath71 is given by @xcite @xmath113 as can be shown , the condition @xmath114 may be recast to @xmath115 , where @xmath72 is the initial time . since @xmath116 , for @xmath117 the so - called _ weak - coupling asymptotic _
approximation may be applied @xcite .
the asymptotic of the transition probability is @xmath118 where @xmath119 in the adiabatic approximation the probability of the system to remain in the ground state may be described by @xmath120 the condition for adiabatic evolution , required by the adiabatic theorem , is @xmath121 @xcite .
the probability of remaining in the ground state at the end of evolution ( @xmath122 ) is given by @xmath123 .
for slow evolution we can use the lz formula @xcite to describe the probability of adiabatic evolution : @xmath124 in figs .
[ p2c ] , [ p2e ] the probability of the adiabatic transition ( red line ) and the results of the exact solutions ( blue line ) are depcted . as can be seen for @xmath125
there is a good agreement between exact solution and the adiabatic formula ( [ p2b ] ) .
for @xmath126 the disagreement takes place . using the so - called _ adiabatic - impulse _ ( ai ) approximation @xcite , qualitatively , the dynamics of the landau - zener model can be described by the kibble - zurek theory of nonequilibrium phase transitions @xcite .
the ai - approximation assumes that the whole evolution can be divided in three parts and up to the phase factor the wave @xmath127 function approximately can be described as @xmath128 : & & |u(\tau ) \rangle\approx |u_{-}(\tau ) \rangle \\
\tau \in [ - \hat \tau,\hat \tau ] : & & |u(\tau ) \rangle\approx |u_{-}(-
\hat \tau ) \rangle \\
\tau \in [ \hat \tau , + \infty ] : & & |\langle u(\tau ) |u_{-}(\tau ) \rangle|^2 = \rm const\end{aligned}\ ] ] where the time @xmath129 , introduced by zurek @xcite , is called the _ freeze - out time _ and define the instant when behaviour of the system changes from the adiabatic regime to an impulse one where its state is effectively frozen and then back from the impulse regime to the adiabatic one . if the evolution starts at moment @xmath130 from the ground state , the equation for determining @xmath131 reads @xmath132 ( for details of calculation see ref .
@xcite ) , and its solution is given by @xmath133 using the relation @xmath134 , we find that the change of the adiabatic regime to a non - adiabatic one occurs when the pressure is @xmath135 , and the non - adiabatic evolution becomes the adiabatic evolution again when the pressure increases up to @xmath136 .
from here we find that within the interval of pressure , @xmath137 , the behaviour of the system is described by the impulse regime . employing eq .
( [ lz1 ] ) , we approximate @xmath138 for fast ( @xmath139 ) and slow ( @xmath140 ) transitions as @xmath141 in the ai approximation the probability , @xmath142 , of finding the system in the excited state at @xmath143 can be calculated as follows @xcite : @xmath144 substituting @xmath131 from ( [ lz1 ] ) , we obtain @xmath145 where @xmath146 . for @xmath139 , from eq .
( [ p3a ] ) it follows @xmath147 . in the first order
this coincides with the result predicted by exact lz formula : @xmath148 . for the adiabatic evolution , @xmath121
, we obtain @xmath149 ( see fig .
[ pai ] . ) . as can be seen ,
the ai approximation is good enough for @xmath150 and in the limit @xmath121 .
_ comparison with experimental data .
_ to compare our theoretical finding with the experiments on spin crossover under the high pressure we use the data from refs . @xcite .
the typical value of the critical pressure is @xmath151 , and the rest of parameters are taken as follows : @xmath152 , @xmath153 and @xmath154 . the computation yields : @xmath155 and @xmath156 .
thus the spin crossover under slowly changed pressure realized in the cited experiments is a highly adiabatic process ( @xmath121 ) . using ( [ eq6a ] )
, we find the that the domain of non - adiabaticity is defined by @xmath157 .
the corresponding interval of time @xmath158 is @xmath159 . in this section
we study the quench dynamics in the spin system under time - dependent pressure , @xmath160 .
we assume that at the initial moment of time @xmath161 , and at the end of evolution @xmath162 .
further it is convenient to present @xmath160 as @xmath163 , where @xmath164 and @xmath165 .
one can observe that the first term in the hamiltonian ( [ eqh2a ] ) , yielding contribution to the total phase factor of the the wave function , does not affects the dynamics of the system and may be omitted .
setting for simplicity @xmath166 and using eq .
( [ eq1a ] ) , one can recast the time - dependent driving hamiltonian as follows : @xmath167 where @xmath168 . for a given @xmath169 ,
the crossover occurs at the critical point @xmath170 defined as @xmath171 . in what follows
we specify the pressure as a pulse with the shape determined by @xmath172 where @xmath173 is the pulse height . for @xmath174 ,
expanding @xmath160 near of the critical point up to the first order , we obtain the related lz problem : @xmath175 .
from here we obtain the corresponding lz adiabaticity parameter as @xmath176 .
applying the adiabatic theorem we find that condition for adiabatic evolution can be written as follows : @xmath177 next using eq .
( [ eq6 ] ) , we obtain @xmath178 where @xmath179 . in terms of the function @xmath180 defined as @xmath181 the condition of adiabaticity ( [ eq2 ] ) can be recast to the following : @xmath182 in fig .
[ p5 ] the function @xmath183 is depicted ( @xmath184 = 1 ) .
we observe that @xmath185 has a maximum in the critical point .
this is in agreement with the general observation on breaking - down of adiabaticity in the neighbourhood of the critical point .
for @xmath186 we obtain the following estimate @xmath187 where @xmath188 in fig .
[ p5a ] the function @xmath189 is depicted .
we observe that @xmath190 for @xmath191 and it is increasing function , @xmath192 for @xmath193 . in figs .
[ p4c ] - [ p4e ] we present the results of numerical calculations for different choice of parameter @xmath169 . in all calculations
the parameters were chosen as follows : @xmath194 , @xmath195 and @xmath196 .
this yields @xmath197 , and for @xmath179 , we obtain @xmath198 . fig .
[ p4c ] shows that for @xmath174 , independently of the initial conditions , the final state of the system is defined by the equal mixture of lh and ls states .
the reason for this phenomena is in growing quantum fluctuations which lead to the mixture of lh and ls states at the critical pressure @xmath44 . in figs .
[ p4f ] and [ p4e ] the dependence of transition probability on time and pressure , respectively , is depicted for different values of the dimensionless amplitude of the pulse , @xmath199 . in the interval of pressure , @xmath200 ,
the evolution is highly adiabatic , and for @xmath201 the system comes to the ls state with the minimal density of defects .
however with the increasing of pressure amplitude , @xmath173 , the probability to pass to the ls state is decreasing , and the system remains in the mixture of the hs ans ls states .
further increasing of the amplitude leads to the domination of the hs population over ls population .
we have analytically studied the properties of spin crossover under the high pressure .
we showed that at static loading and @xmath91 occupation numbers of both hs and ls states are equal at zero temperature ( @xmath37 ) , @xmath202 .
for @xmath203 we obtain @xmath204 and @xmath205 , while for @xmath206 one has @xmath207 and @xmath208 .
static transition at @xmath37 is a sharp quantum phase transition with geometric berry - like phase being the order parameter @xcite .
finite temperature removes singularity in the @xmath209 dependence .
thermal fluctuations between two states @xmath210 and @xmath56 results in a smooth crossover instead of the quantum phase transition at @xmath37 . to verify our theoretical predictions
we have performed numerical simulations .
[ p4f ] shows that the temporal quantum fluctuations have an effect similar to the thermal fluctuations .
small deviations of the shock wave amplitude @xmath211 from the unity results not in a sharp change of the probability @xmath212 either to zero or to unity but to a continuous deviation of the @xmath212 from the 0.5 value . at the same time for @xmath213 any initial distribution of the hs and ls states will end its evolution in the equilibrium state @xmath202 ( fig.[p4c ] ) .
this conclusion is valid for any choice of parameter @xmath214 .
another dynamical effect we would like to discuss is the adiabaticity violations near the critical pressure ( fig.[p4e ] ) .
this is a manifestation of the general kibble - zurek theory .
the results shown in fig.[p4e ] are contra intuitive at the first glimpse . the shock wave with larger amplitude has smaller final probability for spin crossover and larger probability to stay in the initial hs state .
to understand this effect one should note that the characteristic scale of the pressure increase is given by the factor @xmath215 for the wave with amplitude @xmath169 .
thus larger amplitude wave also is faster .
this research was supported by the president of russia grant nsh-1044.2012.2 , presidium of the russian academy of science project 2.16 , rfbr grant 1202 - 90410_a , a.i .
nesterov acknowledges the support from the conacyt , grant no . | in this paper we have studied analytically and numerically dynamics of spin crossover induced by time - dependent pressure .
we show that quasi static pressure , with a slow dependence on time , yields a spin crossover leading to transition from the state of quantum system with high spin ( hs ) to the low spin ( ls ) . however , a quench dynamics under shock - wave load is more complicated .
the final state of the system depends on the amplitude and pulse velocity , resulting in the mixture of the hs and ls states . |
data obtained from a physical system sometimes possess many characteristic length and time scales . in such cases , it is desirable to construct models that are effective for large - scale structures , whilst capturing small scales at the same time . modeling this type of data via diffusion type models may be well - suited in many cases .
thus , multiscale diffusion models have been used to describe the behavior of physical phenomena in scientific areas such as chemistry and biology @xcite , ocean - atmosphere sciences @xcite , finance and econometrics @xcite . in many of these problems ,
the noise is taken to be small because one may , for example , be interested in modeling ( a ) : rare transition events between equilibrium states of a rough energy landscape @xcite , or ( b ) : short time maturity asymptotics for fast mean reverting stochastic volatility models @xcite .
see also @xcite for a thorough discussion on different mathematical and statistical modeling aspects of perturbations of dynamical systems by small noise .
parameter estimation in multiscale models with small noise is a problem of great practical importance , due to their wide range of applications , but also of great difficulty , due to the different separating scales .
the goal of this paper is to develop a theoretical framework for the estimation of unknown parameters in a multiscale diffusion model with vanishing noise .
more specifically , let @xmath0 be given and consider the @xmath1-dimensional process @xmath2 satisfying the stochastic differential equation ( sde ) @xmath3
dt+\sqrt{\epsilon}\sigma\left ( x_{t}^{\epsilon},\frac{x_{t}^{\epsilon}}{\delta}\right ) dw_{t},\hspace{0.2cm}x_{0}^{\epsilon}=x_{0 } , \label{eq : ldpanda1}\ ] ] where @xmath4 as @xmath5 , @xmath6 is an unknown parameter and @xmath7 is a standard @xmath1-dimensional wiener process . the functions @xmath8 and @xmath9 are assumed to be smooth , in the sense of condition [ a : assumption1 ] , and periodic with period @xmath10 in every direction with respect to the second variable .
the rate of convergence of @xmath11 and @xmath12 to zero determines the type of equation that one obtains in the limit .
for example , if @xmath11 is of order 1 as @xmath12 goes to zero , then equation ( [ eq : ldpanda1 ] ) reduces to a deterministic ode that we obtain if we set @xmath12 equal to zero . on the other hand ,
if @xmath12 is of order 1 as @xmath11 goes to zero , then homogenization occurs and this results to an equation with homogenized coefficients .
when both parameters @xmath12 and @xmath11 go to zero together , then we need to consider three different regimes depending on how fast @xmath12 goes to zero relative to @xmath11 : @xmath13 we mention here that asymptotic problems for models like ( [ eq : ldpanda1 ] ) have a long history in the mathematical literature .
we refer the interested reader to classical manuscripts such as @xcite for averaging and homogenization results and to the more recent articles @xcite for large deviations results and @xcite for importance sampling results on related rare event estimation problems .
in ( [ eq : ldpanda1 ] ) we assume that the drift term , through the functions @xmath14 and @xmath15 , depends on a physical parameter @xmath16 . generally , from a statistical inference point of view , the main questions of interest are the following : 1 .
how can one estimate the fast oscillating parameter @xmath11 and the intensity of the noise @xmath12 ?
2 . how can one estimate the unknown parameter @xmath17 the first question is undoubtedly a quite difficult one and is not addressed in the current work ; see @xcite for some related results for specific equations and further references .
instead , we focus on the second question . thus , assuming that the regime of interaction between @xmath12 and @xmath11 is known , we want to estimate the unknown parameter @xmath16 at time @xmath18 , based on the continuously observed process @xmath19 up to this time . in order to do so
, we will follow the maximum likelihood method .
maximum likelihood estimation in multiscale diffusions with noise of order @xmath20 has been studied by different authors and under different settings , see for example @xcite .
we also refer the reader to the manuscripts @xcite for general results on statistical estimation for diffusion processes .
the novelty of the present paper stems from the fact that we address the problem of parameter estimation when both multiscale effects and small noise are present , for all three regimes in ( [ def : threepossibleregimes ] ) , which requires a different approach for the construction of maximum likelihood estimators . indeed , in @xcite , assuming that the noise is of order @xmath20 , the authors fit the data from the prelimit process to the log - likelihood function of the limiting process , i.e. , of the process to which @xmath21 converges to , as @xmath22 .
however , when the diffusion coefficient vanishes in the limit , the limiting process is no longer the solution of an sde , but of an ode ( see theorem [ t : lln ] ) , thus it is deterministic and does not have a well defined likelihood .
therefore , instead of working with the likelihood function of the limiting process , we work with the log - likelihood of the original multiscale model and we infer consistency and asymptotic normality ( under conditions as described below ) by studying its limit . in particular , under regime 1 with @xmath23 and under regimes @xmath24 and @xmath25 ( see ( [ def : threepossibleregimes ] ) ) , we prove that the maximum likelihood estimator ( mle ) is consistent and asymptotically normal under broad conditions .
the situation of regime @xmath26 with @xmath27 is more complicated , because the original log - likelihood function does not have a well defined limit as @xmath28 , due to the @xmath29 terms .
we address this issue by introducing a modified ( pseudo ) log - likelihood which is well defined in the limit .
it turns out that the resulting pseudo mle is not consistent , however its `` bias '' can be computed exactly .
this is a known problem in multiscale parameter estimation problems @xcite ; see section [ s : mle ] for some more details on this . in this article , by `` bias '' we mean the remainder term when we compute the limit of the estimator in probability , that is @xmath30 .
the reason why we use quotes is because bias is usually defined as the remainder of the @xmath31-limit of the estimator . under regime 1 with @xmath27
, we support our findings with a simulation study for a small noise diffusion in a two - scale potential field , a model of interest in the physical chemistry literature , @xcite . for this particular model
, we can construct an estimator that is consistent and normal .
the rest of the paper is organized as follows . in section [ s :
prelim ] , we establish the necessary notation and we present the main ingredients and assumptions needed in the sequel . in section [ s : mle ] we discuss the maximum likelihood estimation problem for all three regimes . for regimes 2 and 3 and
regime 1 when @xmath23 , we prove the consistency of the mle , studying the limit of the log - likelihood function , in section [ s : limitnglikelihood ] , whereas we prove a central limit theorem for the mle in section [ s : clt ] .
finally , in section [ s : examples ] we study a particularly interesting case for regime @xmath26 , when @xmath27 ; a small noise diffusion in a two - scale potential field , we prove a central limit theorem for the pseudo mle in this particular setup and we present a simulated study illustrating the theoretical findings .
we work with the canonical filtered probability space @xmath32 equipped with a filtration @xmath33 that satisfies the usual conditions , namely , @xmath33 is right continuous and @xmath34 contains all @xmath35-negligible sets . regarding the sde ( [ eq : ldpanda1 ] ) we impose the following condition . [ a : assumption1 ] 1 .
the parameter @xmath36 where @xmath37 is open , bounded and convex . also
, the coefficients @xmath8 are lipschitz continuous in @xmath16 .
2 . the functions @xmath38 are lipschitz continuous and bounded in both variables . moreover , they are periodic with period @xmath10 in the second variable in each direction . in the case of regime @xmath26
we additionally assume that they are @xmath39 in @xmath40 and @xmath41 in @xmath42 with all partial derivatives continuous and globally bounded in @xmath42 and @xmath43 .
the diffusion matrix @xmath44 is uniformly nondegenerate .
for notational convenience we define the operator @xmath45 , where for two matrices @xmath46,b=[b_{ij}]$ ] @xmath47 under regime @xmath26 , we also impose the following condition .
[ a : assumption2 ] consider the second order elliptic partial differential operator @xmath48 equipped with periodic boundary conditions in @xmath40 ( @xmath42 is being treated as a parameter here ) .
let @xmath49 be the unique invariant measure corresponding to the operator @xmath50 .
under regime 1 , we assume the standard centering condition ( see @xcite ) for the drift term @xmath51 : @xmath52 where @xmath53 denotes the @xmath1-dimensional torus . under conditions [ a : assumption1 ] and [ a : assumption2 ] ,
theorem 3.3.4 in @xcite guarantees that for each @xmath54 there is a unique , twice differentiable function @xmath55 that is one periodic in every direction in @xmath40 and which solves the following cell problem : @xmath56 we write @xmath57 . under regime @xmath25
, we also impose the following condition .
[ a : assumption3 ] under regime @xmath25 and for any @xmath58 and @xmath59 , we assume that the ordinary differential equation @xmath60 has a unique invariant measure that is lipschitz continuous in @xmath61 .
notice that the existence of a unique smooth invariant measure is immediately implied for regimes @xmath26 and @xmath24 due to condition [ a : assumption1 ] .
however , the situation is more complicated for regime @xmath25 , since the operator of interest is a first order operator , where clearly the non - degeneracy condition does not hold .
for example condition [ a : assumption3 ] certainly holds in dimension @xmath62 , when @xmath63 everywhere and it is sufficiently smooth . before stating the main results , we need additional notation and definitions .
we borrow some notation from @xcite and modify it to fit our needs .
[ def : threepossibleoperators ] for the three possible regimes @xmath64 defined in ( [ def : threepossibleregimes ] ) and for @xmath65 , let @xmath66 \cdot\nabla_{y}+\gamma\frac{1}{2}\sigma(x , y)\sigma(x , y)^{t}:\nabla_{y}\nabla_{y},\nonumber\\ \mathcal{l}_{x,\theta}^{3 } & = c_{\theta}(x , y ) \cdot\nabla _ { y}.\nonumber\end{aligned}\ ] ] for @xmath67 we let @xmath68 and for @xmath69 , @xmath70
. we also define for regime @xmath71 a function @xmath72 , @xmath64 , as follows .
[ def : threepossiblefunctions ] for the three possible regimes @xmath64 defined in ( [ def : threepossibleregimes ] ) and for @xmath73 , define @xmath74 by @xmath75 where @xmath57 is defined by ( [ eq : cellproblem ] ) and @xmath76 is the identity matrix .
based on the results in @xcite , we obtain the following theorem , which essentially is the law of large numbers for ( [ eq : ldpanda1 ] ) . given the results in @xcite , the additional steps required in order to prove this theorem are minimal ,
so we include a short proof in this section as well .
[ t : lln ] consider any @xmath77 and any @xmath0 .
assume condition [ a : assumption1 ] .
in addition , in regime 1 assume condition [ a : assumption2 ] and under regime 3 assume condition [ a : assumption3 ] .
then , for all @xmath58 and @xmath78 and for regime @xmath64 , we have @xmath79=0,\ ] ] where for regime @xmath71 , @xmath80 is the unique solution to the deterministic equation @xmath81 and @xmath82 is the invariant measure corresponding to the operator @xmath83 from definition [ def : threepossibleoperators ] . under our assumptions ,
theorem 2.8 in @xcite guarantees weak convergence of @xmath84 to @xmath85 in @xmath86)$ ] for any @xmath0 .
since , the limiting process @xmath87 is deterministic and weak convergence to constants implies convergence in probability , we obtain the claim of the theorem . also , due to our assumptions , the limiting ode s in ( [ eq : limitingode ] ) are well defined and have a unique solution in their corresponding regime .
assume that we observe the process @xmath19 in continuous time and denote by @xmath88 the data we obtain .
the log - likelihood function for estimating the parameter @xmath16 in the statistical model ( [ eq : ldpanda1 ] ) can be expressed as follows @xmath89 where we denote @xmath90 and for any positive definite matrix @xmath91 @xmath92 the notation used in ( [ eq : likelihoodfunction ] ) is slightly unusual and the brackets @xmath93 outside of the integral are the integrand variables .
this notation is chosen for presentation purposes only , since if we used the arguments to each function in the stochastic integrals , this would result in long and complicated - looking formulas . sometimes , we will omit the subscript @xmath91 if @xmath94 . essentially , we define the likelihood function as the radon - nikodym derivative @xmath95 where @xmath35 is the measure for ( [ eq : ldpanda1 ] ) and @xmath96 the measure for ( [ eq : ldpanda1 ] ) when the drift term is equal to zero . therefore ,
for fixed @xmath97 , we define the maximum likelihood estimator ( mle ) of @xmath16 to be @xmath98 the presence of the small parameters @xmath12 and @xmath11 complicate the estimation of @xmath16 significantly .
our approach is to find the limiting likelihood ( in the appropriate sense ) for each regime @xmath64 , that is @xmath99 then , we prove consistency and derive asymptotic properties of the mle @xmath100 , by studying properties of the prelimiting log - likelihood @xmath101 and of the limiting log - likelihood @xmath102 .
in particular , as we shall see in section [ s : averaging ] , based on the analysis of the log - likelihood function ( [ eq : likelihoodfunction ] ) we prove that the mle is a consistent estimator of the true value @xmath103 , under regime @xmath26 with @xmath23 and regimes @xmath24 and @xmath25 . under the same framework , we also prove , in section [ s : clt ] , that the mle @xmath100 is asymptotically normal .
on the other hand , as we shall see in section [ s : homogenization ] , things get more complicated under regime @xmath26 when @xmath27 . in this case
, the likelihood function ( [ eq : likelihoodfunction ] ) does not necessarily have a well defined limit due to the terms that are multiplied by @xmath104 ( recall that in this case @xmath105 as @xmath28 ) .
we choose to resolve this issue , by taking the limit in an appropriately re - scaled and centered version of the original log - likelihood ( a pseudo log - likelihood ) . under certain conditions , this pseudo log - likelihood approach overcomes the convergence issue and a well defined limit exists .
however , the pseudo maximum likelihood estimator is consistent , even though the `` bias '' is explicitly characterized .
the consistency issue of the maximum likelihood estimation in the presence of `` unbounded drift terms '' , such as the term @xmath106 with @xmath29 is well known in the literature . in the context of @xmath107 and @xmath22 , which corresponds to regime @xmath26
, the problem has also been studied in @xcite under different scenarios and conditions and it is shown there that the maximum likelihood estimator is not consistent and one may need to result in sub - sampling of the data at appropriate rates in order to produce consistent estimators . in the case that has been studied in @xcite the issue
was treated with appropriate sub - sampling of the data .
the article @xcite followed a semi - parametric approach assuming a special structure of the coefficients . in this work
we do not address the consistency issue .
nevertheless , we provide an explicit formula for the asymptotic error in the transformed log - likelihood function .
moreover , we apply our results to the case of small noise diffusion in a two - scale potential field , see section [ s : examples ] . in this case , even though , the original estimator is not consistent , we can construct a consistent estimator and also derive a central limit theorem for the proposed estimator .
we first study the limiting likelihood for regime 1 when @xmath23 and for regimes 2 , 3 and then the proposed pseudo limiting likelihood for regime 1 when @xmath27 . in this section , we consider the limit of the likelihood function @xmath101 , defined by ( [ eq : likelihoodfunction ] ) for regimes 1 when @xmath23 and for regimes @xmath24 and @xmath25 .
let us define the following functions [ def : threepossiblelikelihoods ] for @xmath112 , @xmath73 and for the three possible regimes @xmath64 defined in ( [ def : threepossibleregimes ] ) , define @xmath113 we then prove the following theorem [ t : likelihoodconvergence1 ] let the assumptions of theorem [ t : lln ] hold .
let @xmath114 be a sample path of ( [ eq : ldpanda1 ] ) at @xmath115 .
in the case of regime @xmath26 we assume that @xmath116 .
then , under regime @xmath64 , the sequence @xmath117 converges in @xmath118 probability , uniformly in @xmath58 to @xmath119 from definition [ def : threepossiblelikelihoods ] and where @xmath87 is the solution to the corresponding limiting ode from theorem [ t : lln ] .
in particular , for any @xmath78 @xmath120=0.\ ] ]
lastly , in each regime , the function @xmath121 is maximized at @xmath115 . here ,
@xmath122 , i.e. , the parameter value is @xmath115 .
however , throughout this section we use the compact notation @xmath85 instead of @xmath123 for presentation purposes only , in order to simplify the formulas .
since @xmath114 is a sample path of ( [ eq : ldpanda1 ] ) at @xmath115 , we get that @xmath124 where @xmath125 and @xmath126 then standard averaging principle for locally periodic diffusions , see chapter 3 of @xcite , and the fact that the corresponding invariant measure @xmath82 is continuous as a function of @xmath42 and theorem [ t : lln ] imply that for any @xmath127 @xmath128 moreover , the burkholder - davis - gundy inequality @xcite applied to the stochastic integral @xmath129 and condition [ a : assumption1 ] imply @xmath130 for some constant @xmath131 , uniformly in @xmath58 .
thus , the proof of the claimed convergence follows by chebyschev s inequality and the uniform convergence in @xmath58 .
the fact that the limit is maximized at @xmath115 is easily seen to hold by completing the square in the expressions for @xmath121 at definition [ def : threepossiblelikelihoods ] .
for example , in the case of regime 1 , it is easy to see that @xmath132 and thus the maximum is easily seen to be attained at @xmath115 .
similarly for regimes @xmath24 and @xmath25 . before we continue , we need to impose the following identifiability condition for the true value of the parameter @xmath16 .
[ cond : identifiability ] for all @xmath78 , @xmath133 let @xmath134 . under condition [ cond :
identifiability ] and the assumptions of theorem [ t : likelihoodconvergence1 ] , the mle sequence @xmath135 converges in @xmath118 probability to the true parameter .
in particular , for any @xmath78 we have @xmath136=0.\ ] ] for all @xmath78 , we have that @xmath137 \leq \mathbb{p}_{\theta_{0 } } \left [ \sup_{|u|>\eta } \left ( z_{u+\theta_{0}}^{\epsilon}(\mathcal{x}_{t } ) - z_{\theta_{0}}^{\epsilon}(\mathcal{x}_{t } ) \right ) \geq 0 \right]\\ \leq & \;\;\mathbb{p}_{\theta_{0 } } \biggl [ \sup_{|u|>\eta } \left ( \left ( z_{u+\theta_{0}}^{\epsilon}(\mathcal{x}_{t } ) - z_{\theta_{0}}^{\epsilon}(\mathcal{x}_{t } ) \right ) - \left ( \bar{z}_{u+\theta_{0}}^{i}\left(\bar{x}^{i}_{\cdot}\right ) - \bar{z}_{\theta_{0}}^{i}\left(\bar{x}^{i}_{\cdot}\right ) \right ) \right ) \\ &
\qquad \qquad \qquad \qquad \qquad \qquad \qquad \qquad \qquad \geq -\sup_{|u|>\eta } \left(\bar{z}^{i}_{\theta_{0}+u } \left(\bar{x}^{i}_{\cdot}\right ) - \bar{z}^{i}_{\theta_{0}}\left(\bar{x}^{i}_{\cdot}\right)\right ) \biggr].\end{aligned}\ ] ] condition [ cond : identifiability ] gives that @xmath138 \\ \leq & \;\;\mathbb{p}_{\theta_{0 } } \left [ \sup_{|u|>\eta } \left ( \left ( z_{u+\theta_{0}}^{\epsilon}(\mathcal{x}_{t } ) - \bar{z}_{u+\theta_{0}}^{i}\left(\bar{x}^{i}_{\cdot}\right ) \right ) - \left ( z_{\theta_{0}}^{\epsilon}(\mathcal{x}_{t } ) - \bar{z}_{\theta_{0}}^{i}\left(\bar{x}^{i}_{\cdot}\right ) \right ) \right ) \geq \eta > 0 \right].\end{aligned}\ ] ] therefore , by conditioning on @xmath139 we have @xmath140 \leq \;\ ; \mathbb{p}_{\theta_{0 } } \left[\sup_{|u|>\eta } \left ( z_{u+\theta_{0}}^{\epsilon } - \bar{z}_{u+\theta_{0}}^{i}\right ) \geq
\frac{1}{2}\eta > 0 \right ] + \mathbb{p}_{\theta_{0 } } \left [ \left| z_{\theta_{0}}^{\epsilon } - \bar{z}_{\theta_{0}}^{i}\right| \geq \frac{1}{2}\eta > 0 \right].\ ] ] the result follows by the uniform convergence of theorem [ t : likelihoodconvergence1 ] . in the case of regime @xmath26 with @xmath27 ,
the situation is more involved because the limit of the log - likelihood @xmath101 by ( [ eq : likelihoodfunction ] ) is not well defined .
this is due to the @xmath104 and @xmath142 terms that appear in the expression of @xmath101 .
this leads us to re - parameterize the log - likelihood , so that it will have a well defined limit .
however , we need to re - parameterize the log - likelihood in such a way so that the limiting expression will coincide with the expression of section [ s : averaging ] for @xmath23 and at the same time maintain tractability and simplicity .
let us denote by @xmath143 the log - likelihood function ( [ eq : likelihoodfunction ] ) with @xmath23 .
we define the modified log - likelihood function @xmath144 to characterize the limit , we first need to define several quantities .
but first we impose an additional assumption .
[ a : assumption4 ] let the coefficients @xmath145 and @xmath146 be such that @xmath147 for all @xmath148 and for all @xmath59 . for example , condition [ a : assumption4 ] is trivially satisfied under condition [ a : assumption2 ] if the coefficients @xmath15 and @xmath146 are independent of @xmath149 ( see also remark [ r : remarkregime1 ] below ) .
then , we can consider the auxiliary partial differential equation @xmath150 under condition [ a : assumption4 ] , this poisson equation has a unique bounded , periodic in @xmath40 and smooth solution ( see theorem 3.3.4 of @xcite ) . in order to emphasize the dependence of @xmath151 on @xmath152 , we shall often write @xmath153 . next , we define @xmath154 and @xmath155 for the limiting distribution we prove the following theorem [ t : likelihoodconvergence2 ] let conditions [ a : assumption1 ] , [ a : assumption2 ] and [ a : assumption4 ] hold and consider regime @xmath26 .
let @xmath114 be a sample path of ( [ eq : ldpanda1 ] ) at @xmath115 .
then , the sequence @xmath156 , as defined by ( [ eq : regime1b ] ) , converges in @xmath118 probability , uniformly in @xmath58 to @xmath157 , where @xmath158 in particular , for any @xmath78 @xmath159=0.\ ] ] before proceeding with the proof of the theorem , we make two remarks . when @xmath23 we get that the `` bias '' @xmath160 ( since in this case @xmath161 ) , and we get back the result of theorem [ t : likelihoodconvergence1 ] .
the term @xmath162 is maximized at @xmath115 as in theorem [ t : likelihoodconvergence1 ] . however
, this is not true in general for @xmath163 .
this implies that maximum likelihood in general fails for regime @xmath26 .
[ r : remarkregime1 ] when condition [ a : assumption4 ] is not satisfied , the situation is more complicated . using the modified log - likelihood ( [ eq : regime1b ] ) , condition [ a : assumption4 ] is necessary in order for ( [ eq : poissonequation ] ) to have a solution .
this follows by fredholm alternative as in theorem 3.3.4 of @xcite .
the use of the poisson equation ( [ eq : poissonequation ] ) is an essential tool in the proof of theorem [ t : likelihoodconvergence2 ] .
there does not seem to be an obvious way to reparameterize the likelihood in such a way that it will have a well defined limit and at the same time maintain tractability .
however , as we shall see in section [ s : examples ] , theorem [ t : likelihoodconvergence2 ] covers one of the cases of interest which is the first order langevin equation with a two scale potential . to be more precise
, it covers the case of a small noise diffusion in two - scale potentials of the form ( [ eq : ldpanda1 ] ) with @xmath164 , @xmath165 and @xmath166constant . after some term rearrangement , we get @xmath167\left(x_{s},\frac{x_{s}}{\delta}\right)ds\nonumber\\ & & \hspace{0.2cm}+\frac{\epsilon}{\delta}\int_{0}^{t}\left < b_{\theta_{0}},c_{\theta}\right>_{\alpha}\left(x_{s},\frac{x_{s}}{\delta}\right)ds\nonumber\\ & & \hspace{0.2cm}+\int_{0}^{t}\left[\left < c_{\theta},c_{\theta_{0}}\right>_{\alpha}-\frac{1}{2}\left\vert c_{\theta}\right\vert_{\alpha}^{2}\right]\left(x_{s},\frac{x_{s}}{\delta}\right)ds+\nonumber\\ & & \hspace{0.2cm}+\frac{\delta}{\epsilon}\int_{0}^{t}\left[\left < b_{\theta},c_{\theta_{0}}\right>_{\alpha}+\left < b_{\theta_{0}},c_{\theta}\right>_{\alpha}-\left < b_{\theta},c_{\theta}\right>_{\alpha}\right]\left(x_{s},\frac{x_{s}}{\delta}\right)ds\nonumber\\ & & \hspace{0.2cm}+\left(\frac{\delta}{\epsilon}\right)^{2}\int_{0}^{t}\left[\left < c_{\theta},c_{\theta_{0}}\right>_{\alpha}-\frac{1}{2}\left\vert c_{\theta}\right\vert_{\alpha}^{2}\right]\left(x_{s},\frac{x_{s}}{\delta}\right)ds+ \nonumber\\ & & + \hspace{0.2cm}\sqrt{\epsilon}\left[\frac{\delta}{\epsilon}\int_{0}^{t}\left < b_{\theta},\sigma dw_{s}\right>_{\alpha}\left(x_{s},\frac{x_{s}}{\delta}\right)+\left(\left(\frac{\delta}{\epsilon}\right)^{2}+1\right)\int_{0}^{t}\left < c_{\theta},\sigma dw_{s}\right>_{\alpha}\left(x_{s},\frac{x_{s}}{\delta}\right)\right]\nonumber\\ & = & k_{1}^{\epsilon}+ \frac{\epsilon}{\delta}k_{2}^{\epsilon } + k_{3}^{\epsilon}+\frac{\delta}{\epsilon}k_{4}^{\epsilon}+\left(\frac{\delta}{\epsilon}\right)^{2}k_{5}^{\epsilon}+\sqrt{\epsilon}m^{\epsilon}_{t}. \label{eq : likelihood1}\end{aligned}\ ] ] we study the limiting behavior of the terms in the right hand side of ( [ eq : likelihood1 ] ) .
it is relatively easy to see that the @xmath168 converges to zero in the p - th mean for every @xmath127 .
moreover , the quadratic variation of the stochastic integral @xmath169 in ( [ eq : likelihood1 ] ) has a well defined limit in p - th mean , which together with the fact that it is multiplied by @xmath170 , gives us that this term on the right hand side of ( [ eq : likelihood1 ] ) converges to zero in p - th mean . therefore it remains to study the terms @xmath171 , @xmath172 and @xmath173 . by standard averaging principle for locally periodic diffusions
, it can be seen that @xmath174 converges in @xmath118 probability , uniformly in @xmath58 to @xmath175 ; see for example @xcite .
lastly , we need to study the term @xmath172 . for this purpose
we apply it formula to @xmath176 that satisfies ( [ eq : poissonequation ] ) with @xmath177 to get @xmath178dt\nonumber\\ & & + \left[\frac{\epsilon}{\delta}\left < b_{\theta_{0}},\nabla_{x}\phi \right>+\left
< c_{\theta_{0}},\nabla_{x}\phi\right>+ \frac{\epsilon}{2}\sigma\sigma^{t}:\nabla_{x}\nabla_{x}\phi+\frac{\epsilon}{\delta}\sigma\sigma^{t}:\nabla_{x}\nabla_{y}\phi\right]dt\nonumber\\ & & + \frac{\sqrt{\epsilon}}{\delta}\left<\nabla_{y}\phi,\sigma dw_{t}\right>+\sqrt{\epsilon}\left<\nabla_{x}\phi,\sigma dw_{t}\right>.\label{eq : itoformula1}\end{aligned}\ ] ] hence , recalling that @xmath151 satisfies ( [ eq : poissonequation ] ) , which has a unique , periodic in @xmath40 , bounded and smooth solution due to condition [ a : assumption4 ] , we obtain @xmath179\left(x_{s}^{\epsilon},\frac{x_{s}^{\epsilon}}{\delta}\right)ds\nonumber\\ & & + \sqrt{\epsilon}\int_{0}^{t}\left<\nabla_{y}\phi,\sigma dw_{s}\right>\left(x_{s}^{\epsilon},\frac{x_{s}^{\epsilon}}{\delta}\right)+\sqrt{\epsilon}\delta\int_{0}^{t}\left<\nabla_{x}\phi,\sigma dw_{s}\right>\left(x_{s}^{\epsilon},\frac{x_{s}^{\epsilon}}{\delta}\right)\nonumber\\ & & + \int_{0}^{t}\left <
c_{\theta_{0}},\nabla_{y}\phi\right>\left(x_{s}^{\epsilon},\frac{x_{s}^{\epsilon}}{\delta}\right ) ds.\label{eq : itoformula2}\end{aligned}\ ] ] from this statement the result follows immediately since the last term @xmath180 converges in @xmath118 probability , uniformly in @xmath58 , to @xmath181 .
the rest of the terms on the right hand side of the last display converge to zero in @xmath118 probability , uniformly in @xmath58 , due to the boundedness of @xmath151 and its derivatives and condition [ a : assumption1 ] .
this concludes the proof of the theorem .
in this section we state and prove a central limit theorem ( clt ) for the maximum likelihood estimator @xmath100 of @xmath16 in the case of subsection [ s : averaging ] .
the main structural assumption is that under regime 1 we have that @xmath182 . for notational convenience and without loss of generality , we then consider that @xmath182 for all three regimes
. for regime 2 when @xmath183 one essentially just replaces in the final formula , the function @xmath184 , by the function @xmath185 .
so , without loss of generality , let us assume that @xmath186 for all three regimes .
we define the normed log - likelihood ratio @xmath187 with @xmath35 probability @xmath26 we can write that @xmath188 for notational convenience , we also define the quantities @xmath189 the fisher information matrix is defined to be @xmath190 to this end we recall that the invariant measure @xmath191 has , under our assumptions , a smooth , uniformly bounded away from zero density , which is also periodic in the @xmath192variable ( theorem 3.3.4 and section 3.6.2 in @xcite ) for regimes 1 and 2 and condition 2.3 for regime 3
. we will denote by @xmath193 the density of @xmath82 , namely @xmath194 . in this section
the following condition is imposed .
[ a : assumptionclt ] 1 .
the function @xmath184 is twice continuously differentiable in @xmath16 with bounded derivatives .
the fisher information matrix @xmath195 is positive definite uniformly in @xmath196 , i.e. there exists @xmath197 such that @xmath198 3 .
the vector process @xmath199\right\}$ ] is continuous in probability , uniformly on @xmath58 in @xmath200 $ ] on @xmath16 and on @xmath201 in the point @xmath115 .
the vector valued function @xmath202 is lipschitz continuous in @xmath42 with a lipschitz constant that is uniformly bounded in @xmath203 .
we then have the following theorem .
[ t : clt ] let the conditions of theorem [ t : lln ] and condition [ a : assumptionclt ] hold .
consider regime @xmath64 and let @xmath100 be the maximum likelihood estimator of @xmath16 .
then , uniformly on compacts @xmath204 we have that in distribution under @xmath35 , the following central limit result holds @xmath205^{1/2}\left(\hat{\theta}^{\epsilon}-\theta\right)\rightarrow n(0,i).\ ] ] moreover , the mle has converging moments for all @xmath206 , i.e. , @xmath207 where @xmath208 is a standard @xmath209 random vector .
the proof of this theorem follows by theorem 1.6 in chapter @xmath26 of @xcite ) .
the lemmas [ l : clt_condition1 ] , [ l : clt_condition2 ] and [ l : clt_condition3 ] prove that the conditions of that theorem hold .
for notational convenience we omit writing the subscript @xmath71 , which denotes the particular regime under consideration .
[ l : clt_condition1 ] under the conditions of theorem [ t : clt ] , the family @xmath210 is uniformly asymptotically normal with normalizing matrix @xmath211 .
the proof of this lemma is presented in the appendix .
[ l : clt_condition2 ] under the conditions of theorem [ t : clt ] , there exists @xmath212 and a constant @xmath213 such that for every @xmath214 and compact @xmath215 @xmath216 the proof of this lemma is presented in the appendix .
[ l : clt_condition3 ] under the conditions of theorem [ t : clt ] , and for any @xmath217 and compact @xmath215 there exists a function @xmath218 with the property @xmath219 such that @xmath220 the proof of this lemma is presented in the appendix .
a particular model of interest is the first order langevin equation @xmath221 where @xmath222 is some potential function and @xmath223 the diffusion constant .
we are particularly interested in the case where the potential function @xmath222 is composed of a large - scale smooth part and a fast oscillating part of smaller magnitude : @xmath224 thus the equation of interest can be written as @xmath225 dt+\sqrt{\epsilon}\sqrt{2d}dw_{t},\hspace{0.2cm}x_{0}^{\epsilon } = x_{0 } , \label{eq : langevinequation2}\ ] ] an example of such a potential is given in figure 1 .
[ f : figure1 ] for the potential function drawn in figure [ f : figure1 ] , the unkown parameter @xmath16 corresponds to the curvature of @xmath226 around the equilibrium point .
we are interested in the statistical estimation problem for the parameter @xmath16 in the case of regime @xmath26 , i.e. , when @xmath29 .
in subsection [ ss : langevinequationmle ] we study the estimation problem for @xmath16 based on the methodology described in subsection [ s : homogenization ] . in subsection [ ss : langevinequationclt ] we study the corresponding central limit theorem . in subsection
[ ss : langevinequationsimulation ] we present a simulation study . to connect to our notation
let @xmath227 , @xmath228 , @xmath229 and we consider regime 1 . in this case
there is an explicit formula for the invariant density @xmath230 , which is the gibbs distribution @xmath231 moreover , it is easy to see that the centering conditions [ a : assumption2 ] and [ a : assumption4 ] hold .
notice that in this case the invariant measure does not depend neither on @xmath59 , nor on @xmath232 .
we also define @xmath233 we have the following proposition . [ p : langevin ] under the conditions and notation of theorem [ t : likelihoodconvergence2 ] we have that the error term is given by @xmath234 in the case @xmath227 , @xmath228 and @xmath229 , we notice that the solution @xmath151 to the poisson equation ( [ eq : poissonequation ] ) is related to the solution of the cell problem @xmath235 , ( [ eq : cellproblem ] ) , via the relation @xmath236 hence , we have that @xmath237 this concludes the proof of the proposition .
when we have a separable fluctuating part , i.e. @xmath238 , everything can be calculated explicitly .
we summarize the results in the following corollary .
this corollary also shows that in this case we can derive a consistent estimator @xmath103 in closed form .
[ c : corol ] assume @xmath239 and consider regime @xmath26 .
under the conditions and notation of theorem [ t : likelihoodconvergence2 ] , we have that the error term is given by @xmath240 where @xmath241 is the ( common ) period of the functions @xmath242 in the corresponding direction , @xmath243\ ] ] and for @xmath244 @xmath245 moreover , we have that @xmath246 .
furthermore , recall the mle @xmath247 .
if @xmath248 for both @xmath94 and @xmath249 , then @xmath250 converges in @xmath118 probability to @xmath103 , i.e. , @xmath251 is a consistent estimator of @xmath103 .
the separability assumption of @xmath252 gives us @xmath253 plugging that into ( [ eq : errorlangevingeneral ] ) we immediately get the simplified representation of the error term .
the second claim follows from hlder inequality .
indeed , it is easy to see that @xmath254 .
therefore , we obtain @xmath246 . next , we maximize the limiting log - likelihood function . by straightforward substitution to ( [ eq : maintermregime1general ] )
we see that @xmath255 we collect things together and write @xmath256 then , it is easy to see that this quantity is maximized for @xmath257 then , using theorem [ t : likelihoodconvergence2 ] we obtain the statement of the theorem . in this section ,
we prove a central limit for the maximum likelihood estimator of the first order langevin equation ( [ eq : langevinequation2 ] ) .
based on the modified log likelihood function ( i.e. , on ( [ eq : likelihood1 ] ) ) , the maximum likelihood estimator can be written @xmath258 some algebra manipulation in ( [ eq : modmle ] ) gives us @xmath259 we have the following theorem assume condition [ a : assumption1 ] .
consider the first order langevin equation ( [ eq : langevinequation2 ] ) and assume regime @xmath26 .
let @xmath100 be the maximum likelihood estimator of @xmath103 based on the modified log likelihood function @xmath260 .
then , we have that in distribution under @xmath118 , the following central limit result holds @xmath261 moreover , assuming @xmath239 , we also have that in @xmath118 probability @xmath262 which is ( [ eq : langevinequationestimator ] ) .
the first statement follows directly from the representation of the maximum likelihood estimator in ( [ eq : mle_langevin ] ) and the central limit theorem for stochastic integrals , see for example lemma 1.8 in chapter i of @xcite .
the second statement is as follows .
consider the unique , bounded and periodic in @xmath40 smooth solution of the auxiliary problem @xmath263 by applying it formula to @xmath264 , ( compare with ( [ eq : itoformula1 ] ) and ( [ eq : itoformula2 ] ) ) we get @xmath265 where the term @xmath266 converges to zero in probability as @xmath267 .
therefore , by substituting we obtain that @xmath268 then , as in proposition [ p : langevin ] and corollary [ c : corol ] we can solve the auxiliary pde ( [ eq : poissonequationlangevin ] ) in closed form and obtain the statement of the theorem .
we apply our results in the case when @xmath269 and @xmath270 . as we discussed in the previous sections , in this case we need to work with the modified likelihood , since @xmath27 . as we proved in proposition 6.1 due to the separability of @xmath271
, we can obtain a consistent estimator when properly normalized .
we start by simulating the model .
we use an euler discretization scheme for the multiscale diffusion as follows @xmath272 - \theta x_{t_{k}}^{\epsilon } \right\ } ( t_{k+1 } - t_{k } ) + \sqrt{\epsilon } \sqrt{2d } \left ( w_{t_{k+1 } } - w_{t_{k } } \right),\ ] ] where @xmath273 , @xmath274 is the number of simulated values . for the simulated data we choose @xmath275 and @xmath276 . from @xcite , we have that the euler scheme is bounded above by @xmath277 , where we denote by @xmath278 the discretization step . therefore ,
if we want an error of order 0.001 , we need to choose the discretization step @xmath278 to be equal to @xmath279 .
for the simulation procedure , we choose @xmath280 and @xmath281 .
the maximum likelihood procedure consists of constructing the pseudo log - likelihood function ( 4.9 ) .
more specifically , @xmath282 \\ & & - \theta\ ; \frac{1}{\sigma^{2 } } \left [ \int_{0}^{t } \nabla v(x_{s})dx_{s } \left ( 1 + \left(\frac{\delta}{\epsilon}\right)^{2 } \right ) + \frac{\delta}{\epsilon } \int_{0}^{t } \nabla q\left(\frac{x_{s}}{\delta}\right ) \nabla v(x_{s } ) ds \right ] + const,\end{aligned}\ ] ] where @xmath283 and @xmath284 is a quantity that is independent of the parameter @xmath16 .
the maximizer of this quantity computes as @xmath285}{- \int_{0}^{t } \left(\nabla v(x_{s})^{2 } ds\right ) \left ( \left(\frac{\delta}{\epsilon}\right)^{2 } + 1 \right ) } .\ ] ] although our model is continuous as well as our mle , in practice we obtain data in discrete time .
therefore , we need to discretize our estimator in order to implement it .
we directly discretize the stochastic integrals and we obtain @xmath286}{- \sum_{i=1}^{n-1 } \left(\nabla v(x_{s_i})^{2 } ( s_{i+1}-s_i)\right ) \left ( \left(\frac{\delta}{\epsilon}\right)^{2 } + 1 \right ) } .\ ] ] the consistent estimator @xmath287 will be the normalized @xmath288 .
the normalizing term equals @xmath289 , with @xmath10 and @xmath290 as defined in corollary ( [ c : corol ] ) .
it is important to mention here that we do not simplify the stochastic integral using it s lemma .
the reason is that in order to compute the estimator for @xmath16 we need to use just the observations we have available .
if we use it , then the integral with respect to brownian motion that appears contains a process ( the brownian motion ) that is not observed . using simulated data ,
we construct the mle for different values of the true parameter @xmath16 .
the results are summarized in table [ t : table ] , along with the corresponding 68% and 95% confidence intervals .
these are both empirical intervals meaning that we repeat the procedure ( simulation estimation ) several times ( m=100 ) .
then , we obtain the monte carlo estimator for @xmath16 as the average of all estimators , as well as the monte carlo standard deviation that we use in the construction of the intervals .
[ t : table ] .estimated values of @xmath16 and the corresponding empirical 68% and 95% confidence intervals for various true parameters @xmath16 . [ cols="^,^,^,^",options="header " , ] for @xmath291 , we plot ( figure [ f : figure2 ] ) the histogram of the empirical distribution that we obtain from the monte carlo procedure . for comparison , on the same graph we also plot the corresponding density curve of the theoretical asymptotic ( normal ) distribution with the appropriate variance as the one we computed in theorem 6.3 .
[ f : figure2 ] for the simulated dataset and the corresponding density ( theoretical ) curve from theorem 6.3.,width=10,height=8 ]
in this paper we studied the parameter estimation problem for diffusion processes with multiple scales and vanishing noise . under certain conditions , we derived consistent estimators and proved the related central limit theorems .
the theoretical results are supported by a simulation study of the first order langevin equation in a rough potential .
such results are useful when one is interested in parameter estimation of dynamical systems with more than one scales ( e.g. , in rough potentials ) perturbed by small noise .
let us denote @xmath292 , where @xmath211 .
we assume that @xmath293 belongs in a compact subset of @xmath37 , denoted by @xmath294 , and let @xmath295 as @xmath28 .
we start by rewriting the normed likelihood ratio as follows @xmath296ds -\frac{1}{2}\left(u_{\epsilon},u_{\epsilon}\right)\nonumber\\ & = & j^{\epsilon}_{1}(\theta_{\epsilon})+j^{\epsilon}_{2}(\theta_{\epsilon})+j^{\epsilon}_{3}(\theta_{\epsilon})+j^{\epsilon}_{4}.\nonumber\end{aligned}\ ] ] the last line of the previous computation is easily seen to hold by the following chain of identities @xmath297 which are applied for @xmath298 .
the goal is to prove that @xmath299 , where @xmath151 is distributed as normal @xmath209 and @xmath300 as @xmath28 in @xmath35 probability uniformly in @xmath58 .
this , will establish that the family @xmath210 is uniformly asymptotically normal with normalizing matrix @xmath211 , which then proves the lemma .
moreover , due to averaging and the law of large numbers result theorem [ t : lln ] , the definition of the fisher information matrix @xmath302 implies that @xmath303 converges in distribution with respect to @xmath35 , uniformly in @xmath304 , to @xmath305 where @xmath151 is distributed as @xmath209 , as @xmath28
. then we can write @xmath308\left(x^{\epsilon}_{s},\frac{x^{\epsilon}_{s}}{\delta}\right)\right\vert^{2}ds\nonumber\\ \leq\ ; & |i^{-1/2}(\theta_{\epsilon , u})u_{\epsilon}|^{2 } \sup_{\theta\in\tilde{\theta}}\sup_{|v|\leq c\sqrt{\epsilon } } \mathbb{e}\left|\int_{0}^{t}\left\vert\nabla_{\theta}c_{\theta+v}-\nabla_{\theta}c_{\theta}\right\vert^{2}_{\alpha}\left(x^{\epsilon}_{s},\frac{x^{\epsilon}_{s}}{\delta}\right)ds\right|\nonumber\\ \leq\ ; & c \sup_{\theta\in\tilde{\theta}}\sup_{|v|\leq c\sqrt{\epsilon } } \mathbb{e}\left|\int_{0}^{t}\left\vert\nabla_{\theta}c_{\theta+v}-\nabla_{\theta}c_{\theta}\right\vert^{2}_{\alpha}\left(x^{\epsilon}_{s},\frac{x^{\epsilon}_{s}}{\delta}\right)ds\right|\rightarrow 0 , \textrm { as } \epsilon\downarrow 0.\label{eq : j3_term1}\end{aligned}\ ] ] the last convergence is true due to the uniform continuity of @xmath309 in @xmath304 and tightness of @xmath310 . using it isometry , the last display implies that @xmath311
lastly , it remains to consider the term @xmath312 .
notice that standard averaging principle , the convergence of @xmath19 to @xmath313 as @xmath28 by theorem [ t : lln ] , and the continuous dependence of the involved functions on @xmath16 , imply that , @xmath314ds\right|\rightarrow 0 , \textrm { as } \epsilon\downarrow 0 .
\label{eq : j3_term2}\ ] ] by ( [ eq : j3_term1])-([eq : j3_term2 ] ) and the assumptions on the dependence on @xmath16 we obtain that @xmath315 the proof follows along the lines of lemma 2.3 in @xcite .
we review it here for completeness and mention the required modifications in order to account for the extra component of averaging .
let @xmath317 and define the interpolating point @xmath318\ ] ] by an absolutely continuous change of measure we have @xmath319 where , we have defined @xmath320 .
then , we write @xmath321 where @xmath322 thus , we obtain @xmath323d\ell\nonumber\\ & = \epsilon^{-m } ( 2m)^{-2 m } \int_{0}^{1}\mathbb{e}_{\theta_{2}}\left|\int_{0}^{t}\left<\nabla_{\theta}c_{\theta(\ell)}(\theta_{2}-\theta_{1}),\sigma dw_{s}\right>_{\alpha}\left(x^{\epsilon}_{s},\frac{x^{\epsilon}_{s}}{\delta}\right ) \right|^{2m}d\ell\nonumber\\ & \leq \epsilon^{-m } c_{m , t } \int_{0}^{1}\mathbb{e}_{\theta_{2}}\left[\int_{0}^{t}\left<\nabla_{\theta}c_{\theta(\ell)}\left(x^{\epsilon}_{s},\frac{x^{\epsilon}_{s}}{\delta}\right),(\theta_{2}-\theta_{1})\right>_{\alpha}^{2m}ds\right]d\ell\nonumber\\ & \leq \epsilon^{-m } c_{m , t } |\theta_{2}-\theta_{1}|^{2m}\sup_{\theta_{2},\theta\in\tilde{\theta}}\mathbb{e}_{\theta_{2}}\left[\int_{0}^{t}\left\vert\nabla_{\theta}c_{\theta}\right\vert_{\alpha}^{2m}\left(x^{\epsilon}_{s},\frac{x^{\epsilon}_{s}}{\delta}\right)ds\right]\nonumber\\ & \leq \epsilon^{-m } \phi^{2m}(\epsilon,\theta ) c_{m , t } |u_{2}-u_{1}|^{2m}\sup_{\theta_{2},\theta\in\tilde{\theta}}\mathbb{e}_{\theta_{2}}\left[\int_{0}^{t}\left\vert\nabla_{\theta}c_{\theta}\right\vert_{\alpha}^{2m}\left(x^{\epsilon}_{s},\frac{x^{\epsilon}_{s}}{\delta}\right)ds\right]\nonumber\\ & \leq i^{-m}(\theta ) c_{m , t } |u_{2}-u_{1}|^{2m}\sup_{\theta_{2},\theta\in\tilde{\theta}}\mathbb{e}_{\theta_{2}}\left[\int_{0}^{t}\left\vert\nabla_{\theta}c_{\theta}\right\vert_{\alpha}^{2m}\left(x^{\epsilon}_{s},\frac{x^{\epsilon}_{s}}{\delta}\right)ds\right]\nonumber\end{aligned}\ ] ] and the result follows by the assumed uniform boundedness of @xmath324 . in the absence of multiple scales , this is lemma 2.4 in @xcite .
here we provide the proof of the result with the additional component of multiple scales , which makes the analysis more involved .
for the sake of concreteness we only present the proof for the case of regime @xmath26 .
the required changes for the other regimes are minimal and are mentioned below at the appropriate place .
recall that @xmath211 and set @xmath325\ ] ] we can then write @xmath326\nonumber\\ & \leq \left(\mathbb{e}_{\theta}e^{-p_1\cdot \frac{p - q}{2}\int_{0}^{t}\left\vert \delta c_{\theta}\right\vert_{\alpha}^{2}\left(x^{\epsilon}_{s},\frac{x^{\epsilon}_{s}}{\delta}\right)ds}\right)^{1/p_{1}}\times \nonumber\\ & \quad\times \left(\mathbb{e}_{\theta}\left[e^{pp_{2}\int_{0}^{t}\left<\delta c_{\theta},\sigma dw_{s}\right>_{\alpha}\left(x^{\epsilon}_{s},\frac{x^{\epsilon}_{s}}{\delta}\right)-\frac{qp_{2}}{2}\int_{0}^{t}\left\vert \delta c_{\theta}\right\vert_{\alpha}^{2}\left(x^{\epsilon}_{s},\frac{x^{\epsilon}_{s}}{\delta}\right)ds}\right]\right)^{1/p_{2}}\end{aligned}\ ] ] choosing now @xmath327 , we have that @xmath328\leq 1\ ] ] setting @xmath329 , this implies @xmath330 so , the next step is to appropriately bound from above the term @xmath331 . at this point
, we recall the definition of @xmath332 from ( [ eq : definition_q ] ) and we write @xmath333 define the operator @xmath334 and for @xmath335 , let @xmath336 satisfy the auxiliary pde @xmath337 comparing with the case without the multiple scales , the additional difficulty here is the presence of the fast oscillating component , @xmath338 .
the consideration of the solution to this auxiliary pde , allows us to reduce the bound for the quantity at hand to a bound for a quantity that depends only on the slow component , @xmath19 .
notice that @xmath339 is the operator for regime @xmath26 defined in definition [ def : threepossibleoperators ] with @xmath23 . for regimes 2 and 3 ,
one would need to consider the solution to the pde governed by the corresponding operators from definition [ def : threepossibleoperators ] . since , @xmath340 fredholm alternative , theorem 3.3.4 of @xcite guarantees that there exists a unique , smooth , periodic in @xmath40 and bounded solution to the aforementioned auxiliary pde for @xmath151 .
the boundedness of @xmath37 and the imposed conditions on @xmath309 also guarantee that @xmath151 is bounded uniformly in @xmath341 .
let us apply it formula to @xmath176 with @xmath177 .
it formula gives an expression similar to ( [ eq : itoformula1 ] ) and after some term rearrangement , we get for @xmath304 that @xmath342ds= \int_{0}^{t}\mathcal{l}_{x^{\epsilon}_{s}}\phi\left(x^{\epsilon}_{s},\frac{x^{\epsilon}_{s}}{\delta}\right)ds\nonumber\\ & \quad= ( \delta^{2}/\epsilon)\left(\phi\left(x^{\epsilon}_{t},\frac{x^{\epsilon}_{t}}{\delta}\right)-\phi\left(x^{\epsilon}_{0},\frac{x^{\epsilon}_{0}}{\delta}\right)\right)\nonumber\\ & \quad -\int_{0}^{t}\left[\frac{\delta}{\epsilon}\left <
c_{\theta},\nabla_{y}\phi\right>+ \frac{\delta^{2}}{\epsilon}\left < c_{\theta},\nabla_{x}\phi\right > + \frac{\delta^{2}}{2}\sigma\sigma^{t}:\nabla_{x}\nabla_{x}\phi+\delta\sigma\sigma^{t}:\nabla_{x}\nabla_{y}\phi\right]\left(x^{\epsilon}_{s},\frac{x^{\epsilon}_{s}}{\delta}\right)ds\nonumber\\ & \quad - \frac{\delta}{\sqrt{\epsilon}}\int_{0}^{t}\left<\nabla_{y}\phi,\sigma dw_{s}\right>\left(x^{\epsilon}_{s},\frac{x^{\epsilon}_{s}}{\delta}\right)- \frac{\delta^{2}}{\sqrt{\epsilon}}\int_{0}^{t}\left<\nabla_{x}\phi,\sigma dw_{s}\right>\left(x^{\epsilon}_{s},\frac{x^{\epsilon}_{s}}{\delta}\right)\label{eq : pde_clt2}\end{aligned}\ ] ] due to the boundedness of the involved functions , the last display gives us the existence of a constant @xmath343 that may depend on @xmath294 ( but not on @xmath344 ) , such that @xmath345ds\right|\leq c \left(1+\sup_{t\in[0,t]}|w_{t}|\right ) \label{eq : crucialtermtobound1}\end{aligned}\ ] ] these computations , allow us to continue the right hand side of ( [ eq : crucialtermtobound ] ) as follows @xmath346ds\right]}\times\right.\nonumber\\ & \quad\qquad\left.\times e^{-\gamma\int_{0}^{t}\left(v , q_{\epsilon , v}^{1/2}\left(x^{\epsilon}_{s},\theta\right)\right)^{2}ds}\right\}\nonumber\\ & \leq \left(\mathbb{e}_{\theta}e^{-\gamma p_{3}\int_{0}^{t}\left(v , q_{\epsilon , v}^{1/2}\left(x^{\epsilon}_{s},\theta\right)\right)^{2}ds}\right)^{1/p_{3}}\times\nonumber\\ & \quad \times \left(\mathbb{e}_{\theta}e^{-\gamma q_{3}\left[\int_{0}^{t}\left[\left\vert \delta c_{\theta}\left(x^{\epsilon}_{s},\frac{x^{\epsilon}_{s}}{\delta}\right)\right\vert^{2}_{\alpha}-\left(v , q_{\epsilon , v}^{1/2}\left(x^{\epsilon}_{s},\theta\right)\right)^{2}\right]ds\right]}\right)^{1/q_{3}}\nonumber\\ & \leq \left(\mathbb{e}_{\theta}e^{-\gamma p_{3}\int_{0}^{t}\left(v , q_{\epsilon , v}^{1/2}\left(x^{\epsilon}_{s},\theta\right)\right)^{2}ds}\right)^{1/p_{3 } } \left(\mathbb{e}e^{\gamma c q_{3}\left(1+\sup_{t\in[0,t]}|w_{t}|\right)}\right)^{1/q_{3 } } \label{eq : crucialtermtobound2}\end{aligned}\ ] ] where , the first inequality in the last computation used hlder inequality with @xmath347 and the second inequality used ( [ eq : crucialtermtobound1 ] ) .
so , we now need to focus on the term @xmath348 .
define the vector valued function @xmath349 and notice that @xmath350 using the trivial inequality @xmath351 , applied with @xmath352 and @xmath353 we can write @xmath354 dy\end{aligned}\ ] ] hence , we obtain the bound @xmath355\label{eq : boundforterm_0}\end{aligned}\ ] ] so , as @xmath357 we will have @xmath358 the assumed uniform boundedness of @xmath359 , the fact that @xmath360 is a density and the lower bound from ( [ eq : boundforterm2.3_0 ] ) mean that there exist constants @xmath361 that may depend on @xmath294 such that @xmath362 moreover , by cauchy - schwartz inequality , we also have that @xmath363 } \left\vert x^{\epsilon}_{t}-\bar{x}_{t}\right\vert^{2}\label{eq : boundforterm2.3_2a}\end{aligned}\ ] ] to derive the inequality before the last one , we used the lipschitz continuity in @xmath42 of the function @xmath364 , with a lipschitz constant @xmath365 that may depend on @xmath294 .
to continue , we need to bound from above the quantity @xmath366 } \left\vert x^{\epsilon}_{t}-\bar{x}_{t}\right\vert^{2}$ ] . for this purpose , we set @xmath367 and write @xmath368ds+\int_{0}^{t}\left[\bar{c}_{\theta}\left(x^{\epsilon}_{s}\right)-\bar{c}_{\theta}\left(\bar{x}_{s}\right)\right]ds+\sqrt{\epsilon } \int_{0}^{t}\sigma\left(x^{\epsilon}_{s},\frac{x^{\epsilon}_{s}}{\delta}\right)dw_{s}\nonumber\end{aligned}\ ] ] thus , we obtain @xmath369ds\right\vert^{2}+\int_{0}^{t}\left\vert\bar{c}_{\theta}\left(x^{\epsilon}_{s}\right)-\bar{c}_{\theta}\left(\bar{x}_{s}\right)\right\vert^{2}ds\right.\nonumber\\ & \qquad\qquad\left.+\epsilon \left\vert\sigma\right\vert^{2}\sup_{s\in[0,t]}\left\vert w_{s}\right\vert^{2}\right\}\nonumber\\ & \leq 2^{3}\left\ { \left\vert \int_{0}^{t}\left[c_{\theta}\left(x^{\epsilon}_{s},\frac{x^{\epsilon}_{s}}{\delta}\right)-\bar{c}_{\theta}\left(x^{\epsilon}_{s}\right)\right]ds\right\vert^{2}+c_{5}\int_{0}^{t}\left\vert x^{\epsilon}_{s}-\bar{x}_{s}\right\vert^{2}ds+\epsilon \left\vert\sigma\right\vert^{2}\sup_{s\in[0,t]}\left\vert w_{s}\right\vert^{2 } \right\ } \label{eq : boundforterm2.3_2}\end{aligned}\ ] ] in the last inequality , we used the lipschitz continuity of @xmath370 with a lipschitz constant @xmath371 that may depend on @xmath294 .
let us explain now how the term @xmath372ds\right\vert^{2}$ ] can be treated . by considering the solution to an auxiliary pde problem analogous to ( [ eq : pde_clt ] ) with right hand side replaced by @xmath373 , we get ( similarly to ( [ eq : pde_clt2 ] ) ) that @xmath374ds\right\vert\leq c_{6}\left(1+\frac{\delta}{\sqrt{\epsilon } } \sup_{s\in[0,t]}\left\vert w_{s}\right\vert\right)\ ] ] for some constant @xmath375 that may depend on @xmath294 . thus putting things together , ( [ eq : boundforterm2.3_2 ] ) takes the form @xmath376}\left\vert w_{s}\right\vert^{2 } \right\ } \label{eq : boundforterm2.3_3}\end{aligned}\ ] ] and by grownwall inequality
, we can conclude that there exists a constant @xmath377 , that may depend on @xmath294 , such that @xmath378}\left\vert x^{\epsilon}_{t}-\bar{x}_{t}\right\vert & \leq c_{8}\sqrt{\epsilon+\frac{\delta^{2}}{\epsilon}}\sup_{t\in[0,t]}\left\vert w_{t}\right\vert \label{eq : boundforterm2.3_4a}\end{aligned}\ ] ] coming back to ( [ eq : boundforterm2.3_2a ] ) , we have obtained @xmath379}\left\vert w_{t}\right\vert\label{eq : boundforterm2.3_4}\end{aligned}\ ] ] set @xmath380 .
putting ( [ eq : boundforterm2.3_0 ] ) and ( [ eq : boundforterm2.3_4 ] ) together and recalling that @xmath335 , the bound ( [ eq : boundforterm_0 ] ) becomes @xmath381}\left\vert w_{t}\right\vert}\right]\nonumber\\ & \leq e^{-\gamma p_{3}c_{2}\left\vert u \right\vert^{2 } } \left [ 1 + 4\gamma p_{3 } c_{9 } \left\vert u\right\vert \sqrt{8\pi t } e^{8\gamma^{2}\left(p_{3}c_{9}t\right)^{2 } \left\vert u\right\vert^{2}}\right ] , \label{eq : boundforterm_1}\end{aligned}\ ] ] where the last inequality used lemma 1.14 by kutoyants , @xcite .
now , we have all the necessary ingredients in order to continue the bound of ( [ eq : crucialtermtobound ] ) . in particular , using ( [ eq : crucialtermtobound2 ] ) , ( [ eq : crucialtermtobound ] ) gives @xmath382}|w_{t}|\right)}\right)^{(q - p^{2})/(pq_{3})}\label{eq : crucialtermtobound4}\end{aligned}\ ] ] choosing @xmath383 such that @xmath384 and using the inequality @xmath385 , we then obtain from ( [ eq : boundforterm_1 ] ) @xmath386^{(q - p^{2})/(pp_{3})}\nonumber\\ & \leq e^{-\frac{c_{2}}{2}\left\vert u \right\vert^{2 } + \frac{q - p^{2}}{p}4\gamma c_{9 } \left\vert u\right\vert \sqrt{8\pi t } } \label{eq : crucialtermtobound5}\end{aligned}\ ] ] so , ( [ eq : crucialtermtobound4 ] ) and ( [ eq : crucialtermtobound5 ] ) give @xmath387}|w_{t}|\right)}\right)^{(q - p^{2})/(pq_{3})}\label{eq : crucialtermtobound6}\end{aligned}\ ] ] the right hand side of the last inequality defines our function @xmath388 , which certainly enjoys the property @xmath389 this concludes the proof of the lemma .
y. ait - sahalia , p.a .
mykland and l. zhang , ( 2005 ) , a tale of two time scales : determining integrated volatility with noise high - frequency data , _ journal of american statistical association _ , 100 , pp .
13941411 .
a. j. majda , c. franzke , and b. khouider , ( 2008 ) , an applied mathematics perspective on stochastic modelling for climate .
_ philosophical transactions of the royal society a : mathematical , physical and engineering sciences _ , 366 ( 1875 ) , pp . 24272453 . | we study the problem of parameter estimation for stochastic differential equations with small noise and fast oscillating parameters . depending on
how fast the intensity of the noise goes to zero relative to the homogenization parameter , we consider three different regimes .
for each regime , we construct the maximum likelihood estimator and we study its consistency and asymptotic normality properties . a simulation study for the first order langevin equation with a two scale potential
is also provided .
maximum likelihood estimation for small noise multiscale diffusions konstantinos spiliopoulos ^1^ , alexandra chronopoulou ^2^ ^1^department of mathematics & statistics , boston university + 111 cummington street , boston ma 02215 , e - mail : [email protected] ^2^department of statistics and applied probability , university of california , santa barbara + santa barbara , ca 93106 - 3110 , e - mail : [email protected] * keywords : * parameter estimation , central limit theorem , multiscale diffusions , dynamical systems , rough energy landscapes .
* msc : * 62m05 , 62m86 , 60f05 , 60g99 * acknowledgement : * the authors would like to thank the anonymous reviewer for pointing out a gap in the proof of theorem 5.1 in the original article , as well as all comments that lead to a significant improvement of the article .
k.s .
was partially supported , during revisions of this article , by the national science foundation ( dms 1312124 ) . |
forces between particles are governed by fields which themselves can be considered as composed of particles mediating the interaction by continuous exchange processes .
most prominent in the macroscopic world are electromagnetic fields and gravitational fields . for simplicity
we specialize to electromagnetic forces here , but the line of argument scetched out in the following is valid in general .
macroscopic bodies exert electromagnetic forces on one another whenever they are charged , but if they are neutral all electromagnetic forces apparently vanish .
casimir @xcite was the first to realize that this is not quite correct , because the electromagnetic field is fluctuating .
these fluctuations may be due to quantum fluctuations at zero temperature in vacuum or due to thermal fluctuations in a cavity which is in contact with a heat bath . in any case the fluctuation spectrum , i.e. , the energies which are associated with the eigenmodes of the system and the form of the eigenmodes themselves are manifestly influenced by the _ geometry _ of the system .
the geometry is given by an arrangement of surfaces which impose boundary conditions on the fluctuating field and thus determine its mode spectrum .
the free energy , which contains all information about the thermodynamic properties of the system , is essentially given by a sum over all modes and therefore the free energy will become geometry dependent . if , for example , two uncharged metallic bodies are placed at a certain distance in vacuum the free energy of the configuration depends on the shape of the bodies and the distance between them .
therefore , there will be an _ effective force _ acting between the bodies , which is given by the derivative of the free energy with respect to their distance .
note that this force is a direct consequence of the influence of the bodies on the electromagnetic fluctuation spectrum .
apart from the macroscopic length scales set by the geometry there are no other length scales in the system which limit the maximum wavelength of the fluctuations and therefore the force is governed by powers of the imposed length scales and scaling functions of their ratios , i.e. , the resulting force is _ long - ranged_. due to the history of their discovery @xcite these forces are now known as casimir forces and the influence of boundaries on the functional form of the free energy is known as the _
casimir effect_. it should also be noted that the presence of additional bodies in the above setup modifies the force between any two of them , i.e. , it is not possible to express the casimir effect as a sum of two - body contributions only .
there is a vast body of literature on the various aspects of the casimir effect in electromagnetism which are beyond the scope of this article . for a summary
we refer to review articles dedicated to these subjects @xcite and for a recent survey in a more general context we refer to chapter 3 of ref.@xcite . however , for the understanding of the mechanism and the interpretaion of the results it is instructive to demonstrate some of the fundamental physical concepts by a simple example , which we call _ classic _ here for its historic meaning .
let us assume that we have two parallel perfectly conducting plates placed at a distance @xmath0 in vacuum .
we further assume that the plates have an infinite lateral extension so that we consider the thermodynamic limit with respect to the surface area @xmath1 of the plates and only discuss the free energy @xmath2 per unit area , i.e. , we calculate @xmath3 .
note , that @xmath0 is the only macroscopic length scale in the problem .
the mode spectrum of the electromagnetic field in this parallel plate geometry is particularly simple . in three dimensions the wave vector @xmath4 consists of a lateral component @xmath5 which is unconstrained by the geometry and a discrete perpendicular component @xmath6 for @xmath7 due to the condition that the electric field vector at each of the metallic surfaces must be aligned with the surface normal .
a single mode is then characterized by @xmath8 and @xmath9 and its energy level spacing is given by @xmath10 the energy content @xmath11 of a particular mode is given by its occupation number @xmath12 in the form @xmath13 .
the free energy per unit area is then given by ( see also ref.@xcite for a recent reconsideration of the casimir effect at general temperature ) @xmath14,\ ] ] where @xmath15 and @xmath16 denote the boltzmann constant and the temperature , respectively .
an additional factor of two coming from the summation over the polarizations has already been included in eq.([f ] ) .
the integration over @xmath8 is carried out to an ultraviolett cutoff @xmath17 and the sum is truncated at some maximum mode number @xmath18 . the ultraviolett cutoff parameter @xmath17 is typically determined by the radius of the first brillouin zone of the plate material . if @xmath19 is the lattice constant of the material we identify @xmath20 .
the maximum mode number @xmath18 can be written as @xmath21 , where @xmath22 is also a microscopic length scale ( see below ) .
for @xmath23 only the first term in eq.([f ] ) remains and this is the first example studied by casimir @xcite .
here , we will discuss the high temperature limit @xmath24 , because this allows us to illustrate some aspects of the calculations involved within continuum models like , e.g. , the ginzburg - landau model on a rather elementary level .
for the more general case of layered dielectrica at finite temperature ( dispersion forces ) we refer to the classical literature @xcite and to ref.@xcite . to
leading order in @xmath25 we obtain @xmath26 note that the integral and the summation in eq.([fhight ] ) are only meaningful for finite cutoff parameters @xmath17 and @xmath18 . however , we only need the final result in the limits @xmath27 and @xmath28 . for simplicity
we identify @xmath29 to the order of magnitude which implies @xmath30 , i.e. , @xmath31 .
the integral in eq.([fhight ] ) is elementary and the resulting terms can be arranged as @xmath32 \right\}. \nonumber\end{aligned}\ ] ] the sum over the terms in the second line of eq.([fhtsum ] ) converges so that we can immediately perform the limit @xmath33 here .
we obtain @xmath34 where terms which _ vanish _ in the limit @xmath33 have already been dropped . in order to evaluate eq.([fht ] ) further for large @xmath18 and @xmath35 we employ stirlings formula and the series expansion of the logarithm . with @xmath36 and @xmath21
we obtain the final result @xmath37 + { \pi k_b t \over 8 a^2}\left[2 \ln \left({a \over b}\right ) + 1 \right ] - { k_b t \over l^2}\ { \zeta(3 ) \over 8 \pi } + \dots\ ] ] where @xmath38 is a special value of the riemann zeta function and the dots indicate contributions which are exponentially small in @xmath39 . all terms which vanish in the limit @xmath33 ( @xmath40 ) have consistently been dropped .
the decomposition of the free energy per unit area given by eq.([fhtfin ] ) is a special case of the general decomposition @xmath41 for a film with two surfaces of type @xmath19 and @xmath22 .
the leading contribution to @xmath42 is proportional to @xmath0 and it corresponds to the _ bulk _ contribution of the free energy . in our example
it is given by the radiation pressure @xmath43 between the plates .
the second contribution to eq.([fhtfin ] ) is independent of @xmath0 and it therefore corresponds to the sum of the surface free energies or _ surface tensions _
@xmath44 , where @xmath45 in the above example .
the third contribution varies as @xmath46 and it corresponds to the fluctuation induced long - ranged _
casimir interaction _ between the plates , which is the most prominent contribution to the _ finite - size _ part @xmath47 of the free energy in our example .
note that the casimir contribution is _ independent _ of the microscopic cutoff parameters @xmath19 and @xmath22 .
its absolute strength at a given distance @xmath0 and temperature @xmath16 is governed by the numerical constant @xmath48 which is usually called the _
casimir amplitude_. the casimir amplitude is negative here , so that the casimir force between the plates is _
attractive_. the casimir interaction can be obtained in very elegant ways known as zeta - function regularization , algebraic cutoff , or exponential cutoff schemes @xcite .
their equivalence with respect to the cutoff - independent casimir interaction has been explicitly shown for the example presented here @xcite .
for further details see also ref.@xcite .
the above example for the casimir effect appears to be very specific at first sight , but the functional form of the free energy given by eqs.([fhight ] ) and ( [ fhtfin ] ) is far more general than it seems .
in fact , the underlying mechanism which leads to fluctuation induced long - ranged forces only requires a fluctuating field with geometric restrictions and a macroscopic length scale @xmath0 imposed by the geometry as the only limiting factor for the wavelength of the fluctuations . any system which is at a
_ critical point _ also meets this requirement .
the fluctuating field in this case is given by the order parameter and each of the individual contributions to the free energy as given by eq.([fdecomp ] ) is a sum of a _
part and _ singular _ part which contains the critical behavior of the system .
right _ at _ the critical point the correlation length @xmath49 is infinite so that the distance @xmath0 between the system boundaries provides the only macroscopic length scale as required for the occurrence of long - ranged fluctuation induced forces .
above the critical point the correlation length is finite and therefore the ratio @xmath50 governs the range of these forces .
the existence of casimir forces in critical systems was anticipated by fisher and de gennes @xcite in the framework of the so - called distant wall corrections to critical profiles , where in many cases the casimir amplitudes govern the leading distant - wall correction term to the profile in the vicinity of one of the system boundaries . for details and an extendend list of further references
the reader is referred to chapter 4 of ref.@xcite and to ref.@xcite .
it is important to realize that the casimir amplitudes @xmath51 and the associated scaling functions @xmath52 that take the place of these amplitudes for _ finite _ @xmath53 ( see ref.@xcite and sec.2 ) are _ universal _ , i.e.
, they do not depend on microscopic details of the system under consideration .
note , however , that the precise form of the scaling functions @xmath54 depends on the _ definition _ of the correlation length @xmath49 . for systems with surfaces
the concept of universality classes raises the question of whether there is _ surface _ critical behavior and to what extent it is governed by _
universal surface critical exponents_. during the 1980 s this question was answered in favor of the general ideas of critical behavior and universality , i.e. , microscopic surface properties are indeed unimportant .
one only has to specify the type of _ boundary condition _ which the surface imposes on the order parameter . in this respect
there are only three fundamentally different _
surface universality classes _ @xcite . in particular
, the surface may enhance the order parameter with the result that the system undergoes a second order phase transition in presence of an already _ ordered _ surface ( extraordinary transition , e ) .
the surface may also suppress the order parameter with the result that the system undergoes a second order phase transition in presence of a _ disordered _ surface ( ordinary transition , o ) .
finally the surface and the bulk may order at the _ same _ temperature ( special transition , sb ) , so the critical point of the system is in fact a _
multicritical _ point .
if the spatial dimensionality @xmath55 of the system is high enough there are two options for the occurrence of surface order above the bulk critical temperature @xmath56
. the surface may order _ spontaneously _ at a certain critical temperature @xmath57 _ or _ the surface may be ordered _ externally _ by the presence of a _
surface field_. the bulk transition in presence of an externally ordered surface is called _ normal _ transition .
however , it has been shown recently by rigorous arguments that the normal and the extraordinary transitions only differ by corrections to scaling , so both belong to the extraordinary surface universality class @xcite .
surface critical behavior has already been extensively reviewed @xcite ( see also chapter 2 of ref.@xcite for a short summary ) , so we refrain from giving further details here .
the distinction between the surface universality classes is vital for the casimir forces , because the casimir amplitudes @xmath51 and the scaling functions @xmath54 depend on them .
the simplest boundary conditions apart from periodic ones are _ dirichlet _ boundary conditions which suppress the order parameter to zero at the surface .
a system with these boundary conditions provides a representation of the ordinary surface universality class .
the first systematic field - theoretic calculation of a casimir amplitude was done by symanzik @xcite for this case .
starting from eq.([fhight ] ) the above example essentially reproduces all neccessary steps for such a calculation at the one - loop level ( gaussian theory ) . in general , concepts of the field - theoretic renormalization group
are required which can not be described here .
the application of field theory to the critical behavior of finite systems is a field of ongoing research @xcite which has recently furnished unexpected results concerning the occurrence of _ nonuniversal _ critical finite - size behavior above the upper critical dimension @xcite . for reviews about the general concept of critical finite - size scaling the reader
is referred to ref.@xcite . above the critical temperature the range of the casimir force
is always limited by the correlation length , but below the critical temperature the situation may be different . in ising - like systems
the correlation length is also finite below @xmath56 and therefore the casimir forces have a finite range .
if the system has a continuous symmetry , however , _
goldstone modes _ cause correlation functions of the order parameter to remain long - ranged below @xmath56 .
the most prominent examples are xy and heisenberg ferromagnets which posess an @xmath58 symmetry with @xmath59 and @xmath60 , respectively , in contrast to the ising ferromagnet @xmath61 . in other words ,
the correlation length of continuous ferromagnets remains infinite below @xmath56 and therefore goldstone modes also give rise to fluctuation induced long - ranged forces between system boundaries .
fluids with this property are sometimes called _
correlated fluids_. the most important examples with respect to experimental realizations are liquid @xmath62he below the superfluid - normal transition @xcite and nematic liquid crystals in the nematically ordered phase @xcite , where fluctuations of the nematic director are responsible for the long - ranged nature of the casimir force . near the phase transition to the isotropic phase fluctuations of the degree of
nematic order and the degree of biaxiality generate _ short - ranged _ corrections to the casimir force @xcite . in summary ,
we have mentioned three options for the occurrence of fluctuation - induced long - ranged forces : the presence of long - ranged interactions ( e.g. , electromagnetism , see sec.1.a ) , the presence of critical fluctuations , and the presence of goldstone modes . in the following overview
we will only consider the second option , i.e. , systems in the vicinity of critical points .
in particular , recent progress in the theoretical understanding of critical casimir forces for all surface universality classes and especially for curved geometries will be presented .
special attention is also paid to the comparison of casimir amplitudes and corresponding scaling functions with computer simulations and experiments , which are still in progress at this time . due to the limited scope of this article other interesting developments in related areas
will not be described in any detail and an apology is made in advance to all authors whose work is not explicitly mentioned here .
the remainder of this article is organized according to the three main approaches to critical casimir forces , namely analytic theory , computer simulation , and experiments .
the analytic theory of the casimir effect in critical systems is based on the concept of finite - size scaling @xcite .
exact solutions of model systems in statistical mechanics give only limited access to the finite - size scaling functions , because they are mainly restricted to two dimensional systems . in @xmath63 dimensions
only the spherical model can be analyzed in a rigorous fashion @xcite which has recently been done with special regard to the film geometry for @xmath64 dimensions @xcite . despite their limitations
exact solutions provide valuable insight into the structure of the scaling functions and sometimes the results for @xmath65 can be used to improve estimates obtained by approximative methods for @xmath64 ( see sec.3 of ref.@xcite and below ) .
the concept of finite - size scaling is a natural extension of the principle of _ scale invariance _ to critical systems with geometric constraints on macroscopic length scales .
the principle of scale invariance itself may be viewed as a special case of the more general principle of _ conformal invariance _ ( see sec.3 of ref.@xcite and ref.@xcite ) .
conformal invariance implies the equivalence of systems with boundaries at @xmath66 if these systems can be mapped onto one another by a conformal transformation .
the principle of conformal invariance holds in any dimension , but it is particularly powerful for @xmath65 due to the exceptional large number of conformal mappings in this case ( large conformal group , see ref.@xcite ) . note that scale transformations are just very special conformal mappings . in the framework of conformal field theory
the _ stress tensor _ plays a key role @xcite .
here we only mention that the thermal average of the stress tensor yields the local casimir force in a critical system and therefore the stress tensor provides a very important tool in the analytic theory of the casimir effect .
in fact , most of the casimir amplitudes for @xmath65 have been obtained from conformal field theory rather than exact solutions ( see sec.3 of ref.@xcite and ref.@xcite ) .
many of the experimentally relevant results have been obtained from a field - theoretic analysis of the well - known ginzburg - landau hamiltonian @xmath67 with geometric constraints which can be decomposed according to @xmath68 .
the bulk contribution @xmath69 is given by @xmath70\ ] ] for systems characterized by a @xmath18-component order parameter @xmath71 confined to a volume @xmath72 , where @xmath73 characterize the ising , xy , and heisenberg universality class , respectively .
the parameters @xmath74 and @xmath75 correspond to the bare reduced temperature and external field . the physical ( renormalized ) reduced temperature and external field will be denoted by @xmath76 and @xmath77 in the following .
the surface contribution @xmath78 can be written as @xmath79,\ ] ] where @xmath80 and @xmath81 correspond to the surface enhancement and the surface field , respectively @xcite .
note , that the surface @xmath82 may consist of serveral disjoint parts .
the last contribution @xmath83 contains edge and curvature contributions to the hamiltonian @xmath67 which have first been considered in ref.@xcite within the framework of the renormalization group . for experiments the ordinary transition @xmath84 and the extraordinary transition ( e.g. , @xmath85 ,
see also ref.@xcite ) are the most important cases .
neither @xmath78 nor @xmath83 needs to be considered here in any detail , because their effect is completely contained in the boundary conditions for the order parameter @xmath86 .
we will therefore restrict the following discussion to the ordinary and the extraordinary surface universality class and to periodic boundary conditions .
the spherical model can be considered as the @xmath33 limit of @xmath58 symmetric classical spin models and it can also be expressed as the @xmath87 limit of eqs.([hb ] ) and ( [ hs ] ) .
we only summarize the most recent results here , for a brief overview the reader is referred to sec.2.2 of ref.@xcite and to ref.@xcite . in the presence of an external field @xmath77 and for sufficiently small values of the reduced temperature @xmath88 the _ singular _ contribution @xmath89 to the finite - size part @xmath47 of the free energy per unit area in a film geometry ( see eq.([fdecomp ] ) ) in @xmath55 dimensions can be cast into the scaling form @xcite @xmath90 near the critical point given by @xmath91 and @xmath92 , where @xmath93 indicates the combination of surface universality classes at the two surfaces . the critical exponents @xmath94 and @xmath95 characterize the temperature dependence of the correlation length @xmath96 , @xmath97 and the order parameter ( spontaneous magnetization ) @xmath98 , @xmath99 for @xmath100 , respectively .
the exponent @xmath101 characterizes the functional dependence of the magnetization @xmath102 on the external field @xmath77 for @xmath103 .
the form of the scaling arguments in eq.([df ] ) is imposed by the principle of scale invariance .
they can be obtained by observing that @xmath104 is equivalent to the first scaling argument and @xmath105 , where @xmath106 is the correlation length for finite field @xmath77 at @xmath103 , is equivalent to the second scaling argument . for nearest neighbor interactions for @xmath64
the critical exponents @xmath94 , @xmath95 , and @xmath101 of the spherical model are given by @xmath107 , @xmath108 , and @xmath109 .
the special value @xmath110 of the scaling function is the casimir amplitude . in units of @xmath111
the casimir force @xmath112 is characterized by the corresponding scaling form @xmath113 where the scaling function @xmath114 is given by @xmath115 the universal scaling functions @xmath116 and @xmath114 have been investigated recently for periodic boundary conditions @xmath117 @xcite . for @xmath92
the scaling functions @xmath118 and @xmath119 are both negative and increase monotonically with x , i.e. , unlike the scaling functions in ising - like systems they do not have a minimum in the vicinity of @xmath66 @xmath120 @xcite . for @xmath121 both scaling functions decay to zero exponentially , whereas for @xmath122 @xmath123 .
this behavior is due to the presence of _
goldstone modes _ in the spherical model below @xmath56 . for finite values of @xmath77 @xmath124 the scaling functions
@xmath125 and @xmath126 again decay exponentially as @xmath127 @xcite .
moreover , both scaling functions also decay exponentially for @xmath128 at @xmath129 .
the casimir amplitude @xmath130 can be obtained exactly @xcite and numerically it is very close the best availabe estimates for the ising model for @xmath64 ( see table [ dtable ] in sec.3 ) .
it has also been shown rigorously , that the scaling function @xmath125 is a monotonically increasing function of each of its arguments as long as the temperature @xmath16 is in the vicinity of @xmath56 @xcite .
however , the _ hypothesis _ that this statement is true for _ any _ nearest neighbor @xmath58 symmetric spin model for @xmath131 @xcite could not be substantiated so far ( see below ) .
the film geometry has also been reinvestigated for the ginzburg landau model for the case of the extraordinary surface universality class @xcite , which is of particular interest for experiments with critical binary liquid mixtures .
the scaling functions @xmath132 in zero external field have been determined within mean field theory for infinitely strong surface fields @xmath133 and @xmath134 which enclose an arbitrary angle @xmath135 between @xmath136 ( parallel surface fields ) and @xmath137 ( antiparallel surface fields ) . for ising - like systems only @xmath136 and @xmath137 can be realized and we refer to these cases as @xmath138 and @xmath139 .
the casimir amplitude is negative for @xmath136 and positive for @xmath137 , it changes sign at @xmath140 @xcite .
accordingly , the scaling function @xmath141 is negative and the scaling function @xmath142 is positive for all @xmath143 within mean field theory , but it seems very likely that this is also true beyond mean field theory . the functional form of @xmath141 and @xmath142 is illustrated in fig.[kx0 ] , where the normalization of ref.@xcite has been used .
note that both scaling functions take their extremal values at nonzero @xmath143 , which makes them qualitatively very similar to the corresponding scaling functions for an ising strip for @xmath65 , which can be solved exactly @xcite .
the one - loop corrections to the mean field behavior are very hard to otain and at present they only exist for the casimir amplitudes in the form of the @xmath144-expansion , where @xmath145 .
the numerical quality of the @xmath144-expansion when extrapolated to @xmath146 is very poor so that exact results for @xmath65 have been included in order to obtain an interpolation formula for the casimir amplitudes for @xmath64 .
the values for @xmath147 , @xmath148 , and @xmath149 obtained in this way agree reasonably well with monte - carlo estimates ( see sec .
3 ) @xcite .
the @xmath144-expansion for @xmath147 and @xmath148 has recently received some independent confirmation from local - functional methods @xcite which also provide reliable numerical estimates for @xmath64 ( see table [ dtable ] in sec .
apart from usual critical points , for which the upper critical dimension is @xmath150 , tricritical points in liquid mixtures with more than two components @xcite and in @xmath151he - @xmath62he mixtures ( see sec.6 of ref.@xcite ) also provide an opportunity for experimental tests of the casimir force .
a theoretically appealing feature of a tricritical point is , that its upper critical dimension is @xmath152 so that exact results for @xmath64 can be obtained essentially from a mean field or a gaussian theory .
if , for example , dirichlet boundary conditions are the correct ones for a @xmath151he - @xmath62he mixture in a film at the tricritical point , then the casimir amplitude @xmath153 given in eq.([fhtfin ] ) is also the right one for this system .
there is , however , some debate concerning the correct boundary conditions for tricritical @xmath151he - @xmath62he mixtures @xcite .
the result obtained for @xmath147 at a tricritical point for @xmath64 contains a logarithmic factor which is absent below the upper critical dimension @xmath154 and which introduces a dependence on a microcopic length scale into the casimir amplitude @xcite .
this depencence is very weak and @xmath147 at tricriticality is expected to be about seven times larger than the corresponding amplitude at a usual critical point @xcite . in view of experimental set - ups for ,
e.g. , atomic force microscopy it is desirable to consider other geometries than films , because two plates can not be kept parrallel accurately enough during force measurements .
curved geometries like a sphere in front of a planar wall or two spheres are much more convenient to control experimentally and are also much closer to reality in , e.g. , colloidal suspensions @xcite ( see also ref.@xcite ) . some theoretical effort has therefore been made on the investigation of these curved geometries , where conformal invariance considerations have proved to be a very powerful tool at the critical point @xcite .
if @xmath155 denotes the free energy of a critical fluid in which two spheres with radii @xmath156 and @xmath157 at a center - to - center distance @xmath158 are immersed , then the casimir interaction @xmath159 takes the scaling form @xcite @xmath160 where @xmath93 denotes the combination of surface universality classes and @xmath161 is the conformally invariant cross ratio @xmath162 note that the cases of two separate spheres in an unbounded critical medium and a single sphere inside a critical medium of spherical shape are conformally equivalent and are therefore governed by the same universal scaling function @xmath163 @xcite . for large @xmath156 and @xmath157 at fixed surface - to - surface distance @xmath164 one obtains from the limit of parallel plates @xcite @xmath165^{-(d-1)/2 } , \ ] ] where @xmath166 is the surface area of the unit sphere in @xmath55 dimensions and @xmath167 is the casimir amplitude for parallel plates . in the opposite limit @xmath168 the presence of the spheres can be taken into account by the small sphere expansion @xcite which yields @xmath169 where @xmath170 is the order parameter if both @xmath171 indicate the extraordinary surface universality class . in this case
the scaling exponent @xmath172 is the scaling exponent of the order parameter @xmath173 ( @xmath174 for the ising model for @xmath64 ) .
if only one of the surfaces is not characterized by the extraordinary surface universality class , the operator @xmath175 is given by the local energy density @xmath176 and @xmath172 is the corresponding scaling exponent @xmath177 ( @xmath178 for the ising model for @xmath64 ) .
the amplitudes @xmath179 and @xmath180 are the amplitudes of the critical profiles @xmath181 , @xmath182 of the operator @xmath175 in a seminifinite system bounded by a planar surface of type @xmath183 .
the amplitude @xmath184 is the amplitude of the @xmath185 correlation function in unbounded space .
although none of these amplitudes is universal individually , their combination in eq.([dfabss ] ) is universal and its value for various surface types is exactly known for the ising universality class for @xmath65 . in @xmath186 dimensions estimates can be calculated from a renormalization group analysis of the ginzburg landau model @xcite .
note , that eqs.([dfab ] ) , ( [ dfabpp ] ) , and ( [ dfabss ] ) only hold at the critical point .
the casimir interaction according to eq.([dfabss ] ) is in fact very long ranged . for the extraordinary surface universality class it decays about as slowly as the coulomb interaction . in all other cases the decay is faster , but it is still slower than the decay of , e.g. , dipolar interactions .
the full functional form of the scaling funcions @xmath187 , @xmath188 , @xmath189 , and @xmath190 has been calculated within mean field theory from the stress tensor in the concentric sphere geometry @xcite . as for the case of parallel plates @xmath191 and @xmath192 are negative ( attractive casimir force ) , whereas @xmath193 and @xmath194 are positive ( repulsive casimir force ) . the boundary conditions
@xmath195 , @xmath196 , and @xmath197 have been treated within the gaussian model , where @xmath198 and @xmath199 has been found . although the analytic information from mean field or gaussian theory is quite limited , the combination of these results with exact results for @xmath65 yields fairly reliable estimates for @xmath138 , @xmath200 , @xmath201 , and @xmath202 within the ising universality class in @xmath64 @xcite . higher order calculations beyond mean field or the gaussian approximation , respectively , for the concentric geometry are extremely demanding and results are not available . finally , we note that the sphere - planar wall ( spw ) geometry can also be obtained from the concentric geometry by a conformal mapping @xcite .
so far conformal invariance could be used to obtain the scaling functions of the casimir interaction for various geometries with spherical surfaces .
if the correlation length @xmath49 is finite , conformal invariance does no longer hold . moreover ,
if all length scales @xmath49 , @xmath203 , @xmath156 , and @xmath157 are comparable , small sphere expansions can not be made any longer and a new calculation is required for every geometry . in this case
even mean field results can only be obtained numerically @xcite .
so far , this has only been done in detail for the spw geometry @xmath204 with @xmath205 boundary conditions and at arbitrary temperature near the critical point for ising - like systems @xcite .
we restrict the discussion to the case @xmath206 , where the correlation length @xmath207 governs the decay of the order parameter correlation function in real space .
the casimir force can be cast into the scaling form @xcite @xmath208 where a corresponding scaling function @xmath209 governs the scaling behavior of the casimir force below @xmath56 .
the scaling functions are obtained from the mean field evaluation of the stress tensor which requires the knowledge of the order parameter profile within mean field theory .
the order parameter profile is obtained from a numerical solution of the euler - lagrange equation for eq.([hb ] ) in presence of parallel and infinite surface fields which dictate the boundary conditions .
the functional form of @xmath210 is illustrated in fig.[kspwplot ] . as for the case of parallel plates ,
the casimir force is attractive and takes its maximum value above @xmath56 .
the true position of the maximum is somewhat concealed in fig.[kspwplot ] due to the normalization factor @xmath211 , which is required in order to absorb the divergence of @xmath212 for @xmath213 . in this limit
the derjaguin approximation becomes valid , where the casimir force is represented as an integral over parallel plate contributions .
each of these `` parallel plates '' in the @xmath55 -dimensional spw geometry is an infinitesimal annulus of width @xmath214 and radius @xmath215 which is located on the surface of a paraboloid in order to approximate the sphere near the wall .
the distance of one of these annuli from the wall is then given by @xmath216 , where the integration is performed from @xmath217 to @xmath218 @xcite . note that this approximation is only valid for forces which decay sufficiently fast as @xmath219 .
the amplitude of the derjaguin approximation to the casimir force at @xmath66 is indicated by the open circle in fig.[kspwplot ] , where all solid lines meet .
the dashed line correponds to the small sphere expansion to leading order , where @xmath220 has been used instead of the correct choice @xmath221 , for which a factor @xmath222 is required for proper normalization @xcite . the presence of small external fields can be used to drive , e.g. , a critical binary liquid mixture slightly away from the critical concentration . within the small sphere expansion the casimir energy between two spheres ( colloidal particles ) turns out to be nonsymmetric with respect to deviations from the critical concentration so that the casimir force is enhanced when the concentration of the component preferentially adsorbed by the colloids is reduced @xcite .
this asymmetry is consistent with the asymmetry found experimentally in the flocculation phase diagram of colloidal suspensions @xcite .
the concentric sphere geometry for @xmath205 boundary conditions has also been considered at tricritical points @xcite , where the principle of conformal invariance can be used as well ( see ref.@xcite ) .
the expressions for the casimir forces in the different geometries are similar to the ones obtained for critical points @xcite apart from the logarithmic dependence on a microscopic length scale .
the scaling function of the casimir force depends on the conformally invariant cross ratio given by eq.([kappa ] ) . in the range of distances @xmath223 , where force measurements with the atomic force microscope appear to be feasable , both casimir and van der waals forces
are essentially governed by the parallel plate limit of the curved geometries studied here @xcite .
corresponding investigations of the casimir forces away from the tricritical point have apparently not been performed . finally , we mention that a diluted polymer solution may also serve as a critical medium which mediates long - ranged forces between , e.g. , colloidal particles @xcite
. systematic investigations , however , are still at an early stage and the description of these is beyond the scope of this article .
computer simulations of forces in liquid films have been performed in the past primarily with regard to the microscopic mechanisms of friction , adhesion , and lubrication , where both monte - carlo @xcite and molecular dynamics methods @xcite have been used ( see ref.@xcite for background material and more details ) . with regard to the casimir force in critical or correlated fluids
the situation is less satisfactory .
the computational effort involved in such calculations is substantial and consequently only very few monte - carlo studies of the critical casimir effect exist .
only rectangular geometries have been considered so far in @xmath64 , because the currently available system sizes do not provide sufficient resolution to investigate curved geometries .
the first systematic attempt to measure the casimir amplitudes of ising and potts models in a film geometry is based on a splitting procedure for lattice models _ at _ criticality @xcite .
the systems contain @xmath224 lattice sites , where an aspect ratio of @xmath225 turns out to be sufficient to approximate the film geometry . in the lateral directions
always boundary conditions are always applied .
a seam is introduced into the system hamiltonian , which continuously weakens existing bonds and simulaneously establishes new bonds until the lattice is cut in two halfs of size @xmath226 .
histograms taken in the seam energy give access to the change of the free energy as a function of the seam strength @xcite , which finally yields the total change of the free energy when the lattice is cut in two .
for periodic boundary conditions this method yields the casimir amplitude @xmath227 directly .
for other boundary conditions the knowledge of @xmath227 is required as an input information @xcite .
the method works very well for @xmath65 for critical potts models with @xmath228 , 3 , and 4 and has subsequently been applied to the ising model for @xmath64 with periodic boundary conditions @xcite and with surface fields @xcite .
a summary of the currently available estimates for the casimir amplitudes from various sources is displayed in table [ dtable ] which includes older migdal - kadanoff estimates taken from ref.@xcite .
apart from the well known numerical uncertainties regarding the extrapolation of the @xmath144-expansion and the migdal - kadanoff renormalization scheme the agreement between the estimates for each of the amplitudes is encouraging . especially for @xmath147 and @xmath148 , where the @xmath144-expansion and the migdal - kadanoff scheme are particularly unreliable , the other estimates are fairly consistent .
there are still some prospects to improve the migdal - kadanoff estimates also for these cases , but final results are not yet available @xcite
. it would also be desirable to obtain estimates for the casimir amplitudes from a field theoretic calculation for @xmath64 directly , but attempts in this direction have not yet been made .
the monte - carlo estimates for @xmath147 and @xmath148 are obtained by extrapolating the individual data to inifinite lattice size @xcite .
for @xmath148 this works rather well , but for @xmath147 substantial systematic uncertainties remain and additional data for larger systems are required to obtain a reliable extrapolation ( see figs.4 and 5 in ref.@xcite ) . at present local functional methods as set up in ref.@xcite seem to provide the most reliable estimates for @xmath147 and @xmath148 , because the dimensional dependence of these amplitudes appears to be captured rather well by the local free energy functional .
finally , we note that the casimir amplitudes may also be accessible by exploring the order parameter distribution at the critical point @xcite .
so far this method has only been used for fully finite cubic geometries , generalizations to other geometries have not yet been explored .
a great drawback of the monte - carlo method introduced in ref.@xcite is , that it can not be generalized to temperatures @xmath229 .
the method is based on the measurement of free energy differences , which correspond to linear combinations of the scaling functions at different scaling arguments for @xmath229 .
data of extremely high accuracy would be required to disentangle the individual contributions to the measured free energy difference .
an alternative approach is to mimic the complete wetting scenario ( see ref.@xcite ) near the critical end point of the demixing transition in a binary liquid mixture in a computer simulation @xcite .
the order parameter in this case is the concentration of the mixture ( @xmath230 , ising universality class ) rather than the density difference between liquid and gas , which is usually taken as the order parameter near the liquid - vapor critical point .
in fact , temperature and pressure are adjusted such that the mixture is in its vapor phase very close to liquid - vapor coexistence but _ far away _ from the liquid - vapor critical point .
the interplay between the interparticle potential and the interaction between the particles and an external wall ( substrate ) may lead to the formation of a macroscopic liquid wetting layer of thickness @xmath0 on the substrate at some temperature @xmath231 below the liquid - vapor critical point @xcite .
the problem in the preparation of such a complete wetting layer for a binary mixture is to find a system , i.e. , parameter values for a simulation , such that the critical end point of the demixing transitions is _ inside _ the complete wetting regime , where the macroscopic wetting layer remains stable ( see ref.@xcite for more background information on wetting transitions ) . the casimir effect associated with the critical demixing transition can then be studied in a liquid layer of thickness @xmath0 .
the suggestion to probe the casimir effect in complete wetting layers near critical end points was first made by nightingale and indekeu @xcite and was later worked out in more detail , as the first estimates for the scaling functions of the casimir force became available @xcite ( see also sec.6 of ref.@xcite ) .
the main objective of such a simulation , however , is to obtain more insight into the casimir effect in a more realistic off - lattice model with lennard - jones interactions , which would be the typical situation in an experiment @xcite .
simulations have been performed for a symmetric binary liquid characterized by the lennard - jones interparticle potential @xmath232\ ] ] for two paricle species @xmath233 , where the choices @xmath234 for the hard core parameters and @xmath235 , @xmath236 for the well depth parameters have been made .
note that with these choices the system is invariant under the species exchange @xmath237 and therefore the chemical potentials of both species have been set equal @xmath238 from the outset .
the simulations are performed in a box of size @xmath239 , where periodic boundary conditions are applied in the @xmath143 and @xmath240 directions and two hard walls are specified in the @xmath241 direction , one at @xmath242 and one at @xmath243 .
the wall at @xmath242 is characterized by the attractive potential @xmath244\ ] ] which acts _ equally _ on the particle species @xcite .
the interparticle interactions are truncated at @xmath245 , whereas no range cutoff is employed for @xmath246 .
the phase diagram of the system is spanned by the parameters @xmath247 , @xmath248 , and @xmath249 .
the system sizes used are @xmath250 , @xmath251 , and @xmath252 , where @xmath253 in all cases @xcite .
the parameters are chosen such that @xmath254 a complete wetting layer forms on the substrate and @xmath255 the endpoint of the critical demixing transitions is inside the complete wetting regime ( see above ) . in order to obtain a sufficiently thick wetting layer @xmath256 the undersaturation of the vapor @xmath257
must be tuned to about @xmath258 .
the data aquisition is strongly hamperd by large fluctuations of the liquid vapor interface position ( capillary waves ) which also lead to a substantial slowing down of the algorithm .
furthermore , the data are strongly affected by lateral finite - size effects , because the capillary wave fluctuations also limit the lateral system sizes attainable within reasonable computational effort @xcite .
the equilibrium thickness @xmath0 of the wetting layer ( see also sec.4 ) minimizes the effective interface potential @xmath259 @xcite as a function of the test layer thickness @xmath260 , which is a variational parameter in the spirit of mean - field theory .
the effective interface potential is given by @xmath261 where @xmath262 and @xmath263 denote the liquid and vapor densities , respectively .
the interfacial tensions @xmath264 between the wall @xmath265 and the liquid and @xmath266 between the liquid and the vapor do not depend on @xmath260 .
the last term @xmath267 contains the contributions of all interactions across the wetting layer and it depends on the boundary conditions . by construction
there are no surface fields acting on the model liquid which break the @xmath237 symmetry between the particle species , i.e. , @xmath268 ( see eq.([hs ] ) for @xmath230 ) . however , the wall potental given by eq.([vz ] ) acts like a _ negative _ surface enhancement @xmath80 ( see eq.([hs ] ) ) , which supports demixing near the surface , whereas the liquid - vapor interface acts as a free surface @xmath269 due to the internal @xmath237 symmetry of the model liquid .
the surface universality classes should thus be characterized by the combination @xmath270 .
@xmath271 @xmath267 is therefore used in the form @xcite @xmath272 where the hamaker constant @xmath273 governs the van der waals contribution to the interaction potential and the casimir amplitude @xmath274 ( see table [ dtable ] ) governs the casimir interaction for a symmetric liquid mixture @xcite . note that positive casimir amplitudes ( repulsion ) lead to a critical _ thickening _ of the wetting layer , whereas negative casimir amplitudes ( attraction ) lead to a critical _ thinning _ of the wetting layer . the last term in eq.([doml ] ) is governed by the casimir amplitude @xmath275 .
it provides an order - of - magnitude account of the aforementioned lateral finite - size effects which can be treated as a shift of the undersaturation @xmath276 . in the limit @xmath277 ( film geometry ) , the critical thickening of the wetting layer is given by @xcite @xmath278 where @xmath279 and @xmath280 is the equilibrium layer thickness outside the critical regime .
the measured film thickness @xmath281 versus temperature is shown in fig.[ltplot ] .
the critical thickening of the wetting layer @xmath282 for the largest system ( lowest curve in fig.[ltplot ] ) is of the order of 3% which is in rough agreement with eq.([lcl0 ] ) .
the apparent reduction of the critical thickening with incrasing lateral system size can be explained semiquantitatively by the lateral finite - size correction included in eq.([doml ] ) .
further studies of off - lattice models like this are certainly desirable , however , algorithmic improvements for the treatment of capillary waves will be indispensable for future progress .
the computer simulation of the complete wetting scenario is quite successful , but by its design it only gives an indirect account of the scaling functions @xmath283 of the casimir force . as already described in sec .
2 the most direct access to the casimir force is given by the thermal average of the _ stress tensor _ and it would therefore be most convenient to have a lattice expression for the stress tensor available for spin models .
such expressions can indeed be obtained and successfully used in monte - carlo simulations for lattice models for @xmath65 @xcite .
the basic idea behind the construction of the stress tensor is the same is in continuum theory : one calculates the response of the free energy to a nonconformal mapping of the system , e.g. , an _
anisotropic _ rescaling of the coupling constants . for example , one may choose @xmath284 , @xmath285 with @xmath286 and @xmath287 on a square lattice , such that the critical temperature does not depend on @xmath288 . for the ising model for @xmath65
this procedure yields , e.g. , the @xmath289 component of the lattice stress tensor at lattice site @xmath290 in the form @xcite @xmath291 where @xmath292 is the derivative of @xmath293 at the isotropic point @xmath294 .
the thermal averages @xmath295 of eq.([txx ] ) and @xmath296 of the stress tensor in conformal field theory are related by @xmath297 up to corrections to scaling , where @xmath135 is exactly known for the 2d ising model . for periodic boundary conditions in a strip geometry eq.([txx ]
) can be used to measure the casimir amplitude @xmath227 ( i.e. , the conformal anomaly number @xmath80 ) and two or more scaling dimensions for the ising model and other models if eq.([txx ] ) is generalized accordingly @xcite . for lattice models for @xmath65 conformal field theory
provides sufficient background information so that the desired quantities can be extracted from elaborate fit procedures @xcite .
although eq.([txx ] ) can be readily generalized to @xmath64 , additional information from conformal field theory , which is vital for the data interpretation for @xmath65 , is no longer available .
furthermore , @xmath295 still contains surface contributions for nonperiodic boundary conditions , because the surface tensions will depend on the anisotropy parameter @xmath288 even if the critical point does not .
however , for periodic boundary conditions @xmath295 is at least proportional to the casimir force and some preliminary studies for the xy model in an @xmath298 geometry for @xmath64 dimensions look promising @xcite , although high statistics is needed already for small systems .
experimental verifications of the casimir effect in critical liquids are exceedingly difficult , because data of high accuracy are required and both samples and apparatus must be prepared with great care . at present , two lines of approach
are considered , namely the wetting scenario scetched already in sec.3 and direct force measurements by atomic force microscopes ( afm ) . for a wetting experiment in the vicinity of a critical point a fluid
is required which posesses a critical end point on the liquid vapor coexistence line .
one option for this setup is provided by @xmath62he near its lower @xmath288 point @xcite . for this system the interaction part of the effective interface potential ( see eqs.([oml ] ) and ( [ doml ] ) ) must be modified according to the universality class of the @xmath288-transition in @xmath62he ( xy , @xmath59 ) .
the boundary conditions at the two interfaces of the wetting layer seem to be very well approximated by dirichlet boundary conditions ( o surface universality class ) .
this leads to @xcite @xmath299 where @xmath300 denotes the crossover length to retardation @xcite , and @xmath301 is the scaling function of the casimir potential for the ordinary surface universality class ( see eq.([df ] ) ) .
note that @xmath302 and @xmath303 depend on the dielectric properties of the adsorbate and the substrate . from eqs.([oml ] ) and ( [ domlhe ] ) one expects a critical _ thinning _ of the wetting layer thickness , because @xmath304 .
note that the second scaling argument of @xmath301 captures off - coexsistence effects due to the undersaturation @xmath276 of the @xmath62he vapor . at the @xmath288 point @xmath305
one has @xmath306 which results in a critical thinning of @xmath307 for standard substrates like , e.g. , copper @xcite . in the experimental setup
a stack of five copper capacitors is placed inside a cell which contains liquid @xmath62he at the bottom .
the surfaces of the capacitor provide the substrate potential ( see eq.([vz ] ) ) and their elevation @xmath77 in the gravitational field controls the undersaturation @xmath308 of the @xmath62he vapor .
the layer thickness is obtained from high precision measurements of the capacitance of each of the capacitors as a function of temperature .
the wetting behavior of @xmath62he is extremely sensitive to the surface morphology of the copper plates . in particular ,
microscopic scratches and dust particles lead to localized condensation of @xmath62he on the surface which results in an overestimation of the thickness .
moreover , surface roughness leads to an enhanced surface area which also increases the amount of liquid @xmath62he on the substrate . even with the most advanced polishing techniques these effects can not be avoided completely and therefore also the experimental verification of the dlp theory of dispersion forces @xcite remains a challenge .
nevertheless , the experimental data of the film thinning show a pronounced minimum well below @xmath309 which is given by the specific value @xmath310 of the first scaling argument @xmath311 in eq.([domlhe ] ) .
this value coincides with the minimum of the scaling function @xmath312 of the casimir force with respect to @xmath143 .
the experimental estimate of @xmath313 extracted from the data is displayed in fig.[kxplot ] , which does not show the expected data collapse for the scaling function @xmath314 .
on the contrary , a systematic trend in the data as function of the elevations @xmath77 of the capacitors is visible as shown in the inset of fig.[kxplot ] .
one possible explanation may be given by off - coexistence effects , which would require the second scaling variable @xmath315 in eq.([domlhe ] ) for data collapse .
if a linear dependence of @xmath314 on @xmath240 is assumed , the deviations from data collapse are indeed drastically reduced @xcite .
another option is provided by the introduction of a roughness correction factor as suggested in ref.@xcite , which leads to a similar improvement @xcite .
however , it is evident from fig.[kxplot ] , that there are no appreciable deviations from data collapse for @xmath316 , where a quantitative prediction for @xmath317 exists @xcite .
the comparison is displayed in fig.[kxscal ] which shows reasonable agreement between the data and the prediction .
finally , we note that a finite thinning effect remains visible for temperatures further below @xmath309 , as one would expect from the presence of goldstone modes @xcite .
the overall shape of @xmath314 is manifestly nonmonotonic in contrast to the recently stated monotonicity hypothesis for @xmath318 symmetric spin models @xcite . as second option for a wetting experiment in the vicinity of a critical end point
is provided by binary liquid mixtures near the critical end point @xmath319 ( see sec.3 ) of the line of second order demixing transitions @xcite ( see refs.@xcite for complete phase diagrams ) .
the physical situation is very much like @xmath62he near the lower @xmath288-point , except that both the bulk and surface universality classes are different here .
the second order demixing transition is characterized by a scalar order parameter ( concentration ) , so the system is in the ising @xmath61 universality class .
the substrate material as well as the liquid - vapor interface , which provide the boundaries of the system , usually show some preferential affinity for one of the two components of the mixture , i.e. , the concentration ( order parameter ) departs from its critical bulk value in the vicinity of the surfaces .
this phenomenon is known as critical adsorption ( see , e.g. , ref.@xcite ) and it is captured by the _ extraordinary
_ surface universality class . in the experiment
@xcite a molecularly smooth ( 100 ) si wafer ( n - type , phosphorous doping ) covered with a sio@xmath320 layer of @xmath321 thickness is used , which is suspended vertically inside a pyrex cell .
the elevation @xmath77 of the substrate above the bulk liquid , at which the wetting layer thickness is measured , controls the undersaturation of the vapor .
the reduction of temperature gradients for wetting agents other than superfluid @xmath62he is a quite demanding task and it substantially complicates the preparation of the cell and the sample @xcite . in this experiment two organic mixtures have been used , namely methanol + hexane ( mh ) and 2-methoxy - ethanol + methylcyclohexane ( mm ) . in mh
the methanol component is adsorbed at the si wafer , whereas hexane is adsorbed at the liquid - vapor interface of the wetting layer . in mm
the situation is similar : the 2-methoxy - ethanol is adsorbed at the si wafer , whereas the methylcyclohexane is adsorbed at the liquid - vapor interface .
the wetting layers of both mixtures are therefore characterized by the scaling functions @xmath322 of the casimir potential .
the interaction contribution @xmath267 to the effective interface potential @xmath259 in this case is assumed to be of the form @xcite @xmath323 where retardation and off - coexistence effects are neglected .
the exponential contribution to eq.([domlbl ] ) is due to the presence of the hard wall , which structures the adsorbed fluid over a molecular distance @xmath55 .
the critical temperature @xmath319 is about @xmath324 @xcite . at
the moment only mean field results @xcite and exact results for @xmath65 @xcite exist for the scaling function @xmath325 of the casimir force . in order to obtain reasonable estimates also for @xmath64
at least the one - loop corrections are required which only exist for @xmath326 at the moment ( see ref.@xcite and table [ dtable ] ) . from eq.([lcl0 ] ) and typical values for @xmath319 and the hamaker constant @xmath302 one expects a critical thickening @xmath327 of the wetting layer , when the estimate @xmath328 ( see table [ dtable ] ) is used .
as @xmath319 is approached from above a critical thickening of the wetting layer consistent with this expectation has been found in the experiment and the data for @xmath142 are indeed consistent with scaling @xcite . as function of the scaling variable @xmath329 the scaling function @xmath330
is shown in fig.[kpmscal ] . a comparison between @xmath331 _ at _ and away from the critical composition is shown in the inset for mh .
the shape of @xmath331 shown in fig.[kpmscal ] resembles that of the mean field estimate @xmath142 for @xmath332 shown in fig.[kx0 ] . however , the hamaker constant @xmath302 and therefore also the casimir amplitude @xmath148 , which have been extracted from the data , are much smaller than anticipated .
the reason for this discrepancy has not yet been fully understood .
one possible explanation could be that the parameters of the system are not in the complete wetting regime , i.e. , only _ partial _ rather than complete wetting @xcite is achieved .
further studies are currently under way .
as already mentioned in sec.3 the theoretical investigation of critical fluids in curved geometries is , inter alia , motivated by the prospects of measuring the casimir force directly with an afm . at the moment only exploratory results are available @xcite which have been obtained for the spw geometry in liquid crystals ( see refs .
@xcite ) . in this study
a temperature controlled afm has been used to measure the force between a sphere mounted on the cantilever tip of the afm and a planar wall immersed in an 8cb liquid crystal near the isotropic - nematic phase transition . above the transition in the isotropic phase an attractive force of the order of @xmath333 at a distance @xmath334 between the surface of the sphere @xmath335 and the wall is detected only when the two surfaces are _ moving apart_. this phenomenon is similar to the capillary force in afm microscopy and it is interpreted as the adsorption of a nematically ordered layer of the liquid crystal on the surface of the sensing probe @xcite . slightly above the transition to the nematically ordered phase
an additional attractive force of the order of @xmath336 is detected when the surfaces are _ approaching _ one another .
this additional force is conjectured to be the casimir force mediated by the onset of goldstone modes of the nematic director field in the ordered phase @xcite , where the boundary conditions are supplied by the type of anchoring of the nematic director on the surfaces @xcite .
further quantitative studies of casimir forces in critical and correlated liquids with this apparatus are certainly desirable .
finally , we note that the radiation pressure on a dielectric sphere in the evanescent field of totally reflected light has recently been measured using such afm techniques @xcite .
the theoretical knowledge about casimir forces in critical and correlated fluids which has been accumulated during the last 10 years has become so detailed , that the stage is set for experimental tests of various kinds .
wetting experiments near critical end points have already proved to be a powerful tool to accomplish this goal for quite a variety of fluids .
further studies in this direction are certainly highly desirable and the prospects for them are very good despite substantial experimental challenges one has to face . from the existing theoretical work on curved geometries it has also become clear , that the casimir forces in critical and correlated fluids are within reach of current afm designs .
preparation of the samples and temperature stabilization of the sample and the instrument again pose major challenges for afm force measurements , however , the prospects of probing the casimir effect quantitatively are also very good .
conversely , the experimental approaches to the casimir effect also pose new theoretical challenges .
the problem of substrate roughness in wetting experiments has already been mentioned above and for the case of quenched roughness theoretical results already exist @xcite . however , one of the boundaries of a wetting layer is a free liquid - vapor interface , which may undergo large scale fluctuations due to capillary waves .
what kind of corrections capillary waves as additional degrees of freedom impose on the critical casimir potential is an open question .
quantitative estimates of these corrections are not only important for experiments , they would also aid the data interpretation of computer simulations for critical wetting layers . to
what extent off - coexistence effects influence experimental and numerical wetting layer data is also a largely open problem . in this respect
the recently explored numerical access to the stress tensor of lattice models may prove particularly useful @xcite .
finally , it should be mentioned that improvements of existing theoretical or numerical estimates for the scaling function @xmath114 of the casimir force in particular for the extraordinary surface universality class in various geometries are still needed in order to extract the casimir effect from experimental data as reliably as possible .
although the history of the casimir effect goes back more than half a century it has remained an active field of research .
this article can therefore only provide a snapshot of current knowledge in this area rather than a completed picture .
if this presentation could finally help to trigger or direct new research work in this field , then its main purpose would be fulfilled .
the author gratefully acknowledges stimulating discussions with m. h. w. chan , s. dietrich , r. garcia , m. kardar , b. m. law , a. mukhopadhyay , and f. schlesener .
financial support for this work has been provided through the heisenberg program of the deutsche forschungsgemeinschaft , which is also gratefully acknowledged .
the author expresses his special thanks to m. h. w. chan and r. garcia for providing copies of figs.[kxplot ] and [ kxscal ] and to b. m. law and a. mukhopadhyay for providing a copy of fig.[kpmscal ] prior to publication .
99 h. b. g. casimir , proc .
akad . wet . *
51 * , 793 ( 1948 ) ; h. b. g. casimir and d. polder , phys . rev . * 73 * , 360 ( 1948 ) ; h. b. g. casimir , physica * 19 * , 846 ( 1953 ) ; c. i. sukenik , m. g. boshier , d. cho , v. sandoghdar , and e. a. hinds , phys . rev . lett . * 70 * , 560 ( 1993 ) ; s. k. lamoreaux , phys .
lett . * 78 * , 5 ( 1997 ) .
g. plunien , b. mller , and w. greiner , phys . rep . * 134 * , 87 ( 1986 ) ; v. m. mostepanenko and n. n. trunow , sov . phys .
usp . * 31 * , 965 ( 1988 ) and _ the casimir effect and its applications _ , clarendon press , oxford , 1997 ; e. elizalde and a. romeo , am .
j. phys . * 59 * , 711 ( 1991 ) ; p. w. milonni and m. - l. shih , contemp .
phys . * 33 * , 313 ( 1992 ) ; l. spruch , science * 272 * , 1452 ( 1996 ) .
m. krech , _ the casimir effect in critical systems _ , world scientific , singapore , 1994 .
k. kakazu and s. miyagi , prog
* 100 * , 687 ( 1998 ) .
e. m. lifshitz , sov .
jetp * 2 * , 73 ( 1956 ) ; l. e. dzyaloshinskii , e. m. lifshitz , and l. p. pitaevskii , adv .
* 10 * , 165 ( 1961 ) ; e. s. sabisky and c. h. anderson , phys .
rev . a * 7
* , 790 ( 1973 ) ; j. schwinger , l. l. deraad , and k. a. milton , ann . phys .
( n.y . ) * 115 * , 1 ( 1978 ) .
j. n. israelachvili , _ intermolecular and surface forces _ , academic , new york , 1992 .
e. elizalde and a. romeo , int .
a * 5 * , 1653 ( 1990 ) ; e. elizalde , j. math .
phys . * 31 * , 170 ( 1990 ) ; k. kirsten , j. phys .
a * 24 * , 3281 ( 1991 ) ; b. p. dolan and c. nash , commun .
phys . * 148 * , 139 ( 1992 ) ; s. leseduarte and a. romeo , europhys . lett . * 34 * , 79 ( 1996 ) ; annals of physics * 250 * , 448 ( 1996 ) . n. f. svaiter and b. f. svaiter , j. math . phys .
* 32 * , 175 ( 1990 ) ; n. f. svaiter and b. f. svaiter , j. phys . a * 25 * , 979 ( 1992 ) .
e. elizalde , j. phys .
a * 27 * , l299 ( 1994 ) .
m. e. fisher and p. - g. de gennes , c. r. acad .
paris b * 287 * , 207 ( 1978 ) .
e. eisenriegler , m. krech , and s. dietrich , phys .
e * 53 * , 14377 ( 1996 ) .
k. binder , in _ phase transitions and critical phenomena _ , edited by c. domb and j. l. lebowitz ( academic , london , 1983 ) , vol .
8 , p. 2 ; h. w. diehl , in _ phase transitions and critical phenomena _ , edited by c. domb and j. l. lebowitz ( academic , london , 1986 ) , vol .
10 , p. 76 ; h. w. diehl , int .
b * 11 * , 3503 ( 1997 ) ; h. w. diehl and m. shpot , nucl . phys . *
b 528 * , 595 ( 1998 ) .
t. w. burkhardt and h. w. diehl , phys .
b * 50 * , 3894 ( 1994 ) .
k. symanzik , nucl . phys . *
b190 * , 1 ( 1981 ) .
d. oconnor and c. r. stephens , phys .
lett . * 72 * , 506 ( 1994 ) ; f. freire , d. oconnor , and c. r. stephens , j. stat .
phys . * 74 * , 219 ( 1994 ) ; a. esser , v. dohm , and x. s. chen , physica a * 222 * , 355 ( 1995 ) ; a. esser , v. dohm , m. hermes , and j. s. wang , z. phys .
b * 97 * , 205 ( 1995 ) ; x. s. chen , v. dohm , and a. esser , j. phys .
( france ) i * 5 * , 205 ( 1995 ) ; f. freire , d. oconnor , and c. r. stephens , phys . rev .
e * 53 * , 189 ( 1996 ) ; x. s. chen , v. dohm , and a. l. talapov , physica a * 232 * , 375 ( 1996 ) ; x. s. chen , v. dohm , and n. schultka , phys .
lett . * 77 * , 3641 ( 1996 ) .
x. s. chen and v. dohm , physica a * 251 * , 439 ( 1998 ) ; eur .
j. b * 5 * , 529 ( 1998 ) .
m. n. barber , in _ phase transitions and critical phenomena _ , edited by c. domb and j. l. lebowitz ( academic , london , 1983 ) , vol .
8 , p. 145 ; v. privman , in _ finite size scaling and numerical simulation of statistical systems _ , edited by v. privman , ( world scientific , singapore , 1994 ) ; v. privman , p. c. hohenberg , and a. aharony , in _ phase transitions and critical phenomena _ , edited by c. domb and j. l. lebowitz ( academic , new york , 1991 ) , vol .
14 , p. 1 .
h. li and m. kardar , phys .
67 * , 3275 ( 1991 ) and phys .
a * 46 * , 6490 ( 1992 ) ; m. kardar , talk at the aps centennial meeting , atlanta , march 1999 .
a. adjari , l. peliti , and j. prost , phys .
lett . * 66 * , 1481 ( 1991 ) ; a. adjari , b. duplantier , d. hone , l. peliti , and j. prost , j. phys .
( france ) * 2 * , 487 ( 1992 ) ; m. l. lyra , m. kardar , and n. f. svaiter , phys .
rev e * 47 * , 3456 ( 1993 ) .
p. ziherl , s. zumer , and r. podgornik , braz .
journal of physics * 28 * , 267 ( 1998 ) ; p. ziherl , r. podgornik , and s. zumer , chem .
. lett . * 295 * , 99 ( 1998 ) .
s. allen and r. k. pathria , can .
j. phys . *
67 * , 952 ( 1989 ) ; can .
* 69 * , 753 ( 1991 ) ; j. g. brankov and n. s. tonchev , j , stat . phys . * 60 * , 519 ( 1990 ) ; j. g. brankov and d. m. danchev , j. math .
phys , * 32 * , 2543 ( 1991 ) ; j. stat . phys . * 71 * , 775 ( 1993 ) ; d. m. danchev , j. stat . phys .
* 73 * , 267 ( 1993 ) .
d. m. danchev , phys .
e * 53 * , 2104 ( 1996 ) .
d. m. danchev , phys .
e * 58 * , 1455 ( 1998 ) .
j. l. cardy in _ phase transitions and critical phenomena _ , edited by c. domb and j. l.
lebowitz ( academic , london , 1987 ) , vol .
11 , p. 55 ; c. itzykson and j .- m .
drouffe , _ statistical field theory _ , vol .
2 , ( cambridge university press , cambridge , 1992 ) .
d. m. mcavity and h. osborn , nucl .
b406 * , 655 ( 1993 ) .
m. krech , phys .
e * 56 * , 1642 ( 1997 ) .
r. evans and j. stecki , phys .
b * 49 * , 8842 ( 1994 ) .
z. borjan and p. j. upton , phys .
lett . * 81 * , 4911 ( 1998 ) .
u. ritschel and m. gerwinski , physica a * 243 * , 362 ( 1997 ) .
d. beysens and d. esteve , phys .
lett . * 54 * , 2123(1985 ) .
s. l. carnie , d. y. c. chan , and j. stankovich , j. coll .
* 165 * , 116 ( 1994 ) ; p. johansson and p. apell , phys . rev .
b * 56 * , 4159 ( 1997 ) .
t. w. burkhardt and e. eisenriegler , phys .
74 * , 3189 ( 1995 ) and references therein .
e. eisenriegler and u. ritschel , phys .
b * 51 * , 13717 ( 1995 ) .
a. hanke , f. schlesener , e. eisenriegler , and s. dietrich , phys .
lett . * 81 * , 1885 ( 1998 ) ; f. schlesener , diplomarbeit , universitt wuppertal , 1997 .
e. eisenriegler , a. hanke , and s. dietrich , phys .
e * 54 * , 1134 ( 1996 ) .
m. schoen , d. j. diester , and j. h. cushman , j. chem .
phys . * 100 * , 7707 ( 1994 ) . s. k. das , m. m. sharma , and r. s. schechter , j. phys .
chem . * 100*. 7122 ( 1996 ) .
m. krech and d. p. landau , phys .
e * 53 * , 4414 ( 1996 ) ; k. k. mon , phys .
rev . lett . * 54 * , 2671 ( 1985 ) ; phys .
b * 35 * , 3560 ( 1987 ) .
j. o. indekeu , m. p. nightingale , and w. v. wang , phys .
b * 34 * , 330 ( 1986 ) .
d. danchev , 1998 , private communication .
a. d. bruce , j. phys .
a. * 28 * , 3345 ( 1995 ) .
s. dietrich , in _ phase transitions and critical phenomena _ , edited by c. domb and j. l. lebowitz ( academic , london , 1988 ) , vol .
12 , p. 1 .
n. b. wilding and m. krech , phys .
e * 57 * , 5795 ( 1998 ) .
m. p. nightingale and j. o. indekeu , phys .
* 54 * , 1824 ( 1985 ) ; * 55 * , 1700 ( 1985 ) ; r. lipowski and u. seifert , phys .
b * 31 * , 4701 ( 1985 ) ; phys . rev
. lett . * 55 * , 1699 ( 1985 ) .
m. krech and s. dietrich , phys .
a * 46 * , 1886 ( 1992 ) ; 1922 ( 1992 ) .
p. j. m. bastiaansen and h. j. f. knops , phys .
e * 57 * , 3784 ( 1998 ) .
m. krech , talk at the aps centennial meeting , atlanta , march 1999 .
r. garcia and m. h. w. chan , phys .
lett . * 83 * , 1187 ( 1999 ) ; r. garcia , talk at the aps centennial meeting , atlanta , march 1999 .
e. cheng and m. w. cole , phys .
b * 38 * , 987 ( 1988 ) .
a. mukhopadhyay and b. m. law , phys .
lett . * 83 * , 772 ( 1999 ) ; b. m. law , talk at the aps centennial meeting , atlanta , march 1999 .
i. musevic , g. slak , and r. blinc , rev .
* 67 * , 2554 ( 1996 ) and _ abstract book of the @xmath337 international liquid crystal conference _ ,
kent , oh , p.91 ( 1996 ) .
m. vilfan , i. musevic , and m. copic , europhys . lett . * 43 * , 41 ( 1998 ) . | the current knowledge about fluctuation induced long - ranged forces is summarized .
reference is made in particular to fluids near critical points , for which some new insight has been obtained recently . where appropiate , results of analytic theory are compared with computer simulations and experiments . |
galaxy formation and evolution has been a very actively debated topic of observational cosmology in the last years .
only recently models and observations are converging toward a unified and coherent picture .
on one side the observations of the universe , with some controversies on the way , have finally agreed that half of the present - day stars was already in place at z @xmath5 1 and that the same mass deficit affects , within the measured accuracies , the high - mass end : basically , the galaxy stellar mass function evolves from the local determination to redshift one only by a normalization factor of about two .
this means that , both at high and at low masses , there is roughly a factor of two difference between the comoving number density at redshift one and the local value ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
measurements of the star formation rate density over cosmic times ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) are also in good agreement with half of the present - day stars being already born at redshift 1 . on the other side
the models , linking the hierarchical growth of dark matter structures to the observed galaxy properties by means of simplified prescriptions for the formation of baryonic systems , predict that galaxies form in a bottom - up fashion by following the cosmological destiny of dark matter halos .
massive galaxies , in these models , assemble most ( 50% ) of their stellar mass via merging at @xmath6 @xcite , but most of these accretion events are red mergers ( i.e. gas - free mergers that imply no induced star - formation ) , which are very difficult to detect from the observational point of view ( e.g. * ? ? ?
* ; * ? ? ?
. the latest realizations of these models ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) have been able to fully reproduce the galaxy stellar mass function up to high redshifts , thus reconciling the theoretical bottom - up assembly of dark matter halos to the claimed top - down assembly of galaxies ( e.g. * ? ? ?
* ) . in _ hierarchical _ models
star formation is supposed to take place only in disk structures and in gas - rich mergers ( see also @xcite for an up to date perspective ) .
this latter event would exhaust the residual gas and create a dynamically hot system : a bulge galaxy .
also dynamical instabilities and minor mergers are able to destroy the disk structures and finally create an elliptical galaxy .
detailed predictions in the literature do not exist but the general expectation is that the fraction of bulge dominated massive galaxies increases with time . the study of the evolution of the morphological mix in the galaxy stellar mass function is a decisive tool to put sensitive constraints on the models , because it could give unique insights in both the feedback and the merging processes in galaxy evolution ( see for a detailed discussion * ? ? ?
galaxies have been often classified , both at low and at high redshift , as `` red sequence '' and `` blue cloud '' objects based on their broad band colors @xcite .
however this kind of classification when used as a proxy for the galaxy morphology ( see e.g. @xcite ) , suffers from some obvious drawbacks : a disk galaxy populated by an old stellar population would be classified as an early type object , and vice versa ; a starburst galaxy , heavily obscured by dust , would be classified as a red , and _ dead _ , early
type object .
thanks to the advent of the advanced camera for surveys ( acs ; @xcite ) on board the hubble space telescope ( hst ) , it has become possible to investigate the optical rest - frame emission of galaxies up to redshift one and slightly beyond with sub - kiloparsec resolution and on relatively wide fields for hundreds of thousands of galaxies .
given the exponentially growing number of galaxies available to be morphologically classified it has become urgent for galaxy evolution studies to build up robust and automated tools for morphological and structural classification that could reliably substitute the visual classification ( see the discussion in @xcite on unavoidable drawbacks affecting the automated approach ) .
the efforts on the path to the automated image classification have essentially split in two main branches , parametric and nonparametric methods , each having its own advantages and drawbacks .
parametric methods are based on the idea that galaxy light profiles may be sufficiently well described by analytic formulas . indeed parametric profiles have been shown to well describe galaxy structural properties since many decades now .
the @xmath7 law @xcite describing almost perfectly luminous early - type galaxy light profiles , and the exponential law @xcite which instead well fits the light profile of disk - dominated galaxies , have been extensively used to derive galaxy physical quantities , such as the effective radius and surface brightness .
this has in fact allowed to formulate and study galaxy scaling relations , such as the tully - fisher @xcite or the fundamental plane @xcite , which have enormously boosted our knowledge and understanding of galaxy evolution and formation .
the @xmath7 and the exponential laws may be used in the same modeling to describe both the bulge and disk photometric components of a single galaxy ( see e.g. * ? ? ?
this approach is in fact doable only for very high signal to noise objects as the number of parameters involved becomes fairly large and hence the fitting itself heavily degenerate .
for this reason it has become very popular the use of the more simple and robust single component srsic ( 1968 ) model .
this is a convenient generalization of the @xmath7 profile with @xmath8 , where @xmath9 is called the srsic index .
the variation of the index @xmath9 allows to describe very different galaxy light profiles from the classical @xmath7 ( @xmath10 ) , to the late - type galaxy exponential profile ( @xmath11 ) , to the massive elliptical galaxy profiles ( @xmath12 ) and down the unresolved point - like gaussian profiles ( @xmath13 ) .
the srsic profile fitting , thanks to its flexibility and robustness , has become a preferred tool in the last years to classify galaxies in broad morphological classes ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , possibly combined to the `` bumpiness '' parameter as done in @xcite and @xcite . the main idea behind the nonparametric approach is to measure model independent quantities , such as the concentration , the asymmetry ( @xcite ) , the m@xmath14 and the gini parameters ( @xcite ) of the galaxy light distribution and to correlate these quantities to the visual galaxy appearance .
albeit being model independent , and hence conceptually more effective in describing the variegated galaxy morphology zoo , these classification schemes are not free from severe biases as addressed in @xcite and more recently in @xcite . having in mind to continue the study presented in this paper by investigating also the evolution of galaxy sizes , we opted for the parametric approach , that delivers both morphological classifications and structural parameters . in this context the cosmos
@xcite survey is the ultimate effort to cover a sensitive patch of the sky ( @xmath15 ) with hst , going factors of 10/100/1000 wider than previous efforts like the hdf - n @xcite , gems @xcite , the great observatories origins deep survey ( goods ; @xcite ) , the fors deep field ( fdf ; @xcite ) among many others .
the high resolution imaging has been complemented in cosmos by a state - of - the - art multiwavelength coverage all the way from the x - ray to the radio . by observing the universe in all its colors cosmos
is becoming a real breakthrough in observational cosmology . the need to tackle and understand the conspiracies of nature embedded in the universe variance makes of this wide and deep survey a unique and unprecedented tool for galaxy evolution studies . in this work
we rely on a deep and accurate ( i @xmath3 26.5 and ) photometric redshift catalog @xcite complemented with a morphologically classified sample ( f814 @xmath3 24 ) to study the contribution of galaxies of different morphologies to the redshift evolution of the stellar mass density , as well as its dependence on the environmental density , over an area of @xmath10.7 @xmath2 in the cosmos field .
@xcite , @xcite , @xcite and @xcite have carried out studies on the evolution of the galaxy stellar mass function split by morphological types , in a way qualitatively similar to the one we present here , but based either on shallower samples and visual morphological classification or on much smaller areas heavily affected by cosmic variance .
likely due to the small surveyed area , no one of the just mentioned studies defined environmental properties of galaxies in order to contrast the effect of local environment against cosmic time on morphological evolution .
this paper is organized as follows : in [ sec_data ] we discuss the ground - based dataset and the photometric redshifts on which this work is based , in [ mmll ] we discuss how galaxy stellar masses have been estimated , in 4 we present the acs data used in this work , the source extraction and the galaxy number counts , in
[ sec_sersic ] we briefly describe the quantitative morphological analysis , referring to a more detailed discussion in appendix a , in [ sec_mass ] we present the evolution of the morphological mass function and mass density , in
[ sec_env ] we discuss the effect of local environments on galaxy morphological evolution and finally in [ sec_conclusions ] we summarize our results and draw our conclusions . throughout this work
, we use ab magnitudes and adopt a @xmath16 cosmology with , , and .
the ground - based data used in this paper combine the _ ubvrizk _ cosmos dataset with proprietary imaging in the h band . a highly homogeneous multi - wavelength i - band selected catalog is used to derive accurate photometric redshifts . a full description of the nir data acquisition , data reduction and catalog assembling , and of the photometric redshifts estimation , is presented in @xcite .
here in the following , for the sake of clarity , we give only a brief summary .
our _ cleaned _ i - selected cataloggabasch / cosmos ] comprises 293 377 objects over an area of @xmath10.7 square degrees down to a limiting magnitude @xmath126.5 .
there is a good agreement between literature data and this catalog number counts up to the 25th magnitude . at the bright end ( i@xmath1721 )
the literature number counts are sensibly higher since most of the very bright objects are saturated in the subaru i - band images and thus not present in the final catalog . since objects brighter than i=21 are likely to be important for the determination of the high mass end of the galaxy stellar mass function at low and intermediate redshifts , we added those objects ( 1896 ) back to our catalog by complementing it with the cosmos legacy photometric catalog described in @xcite , where cfht i band photometry was used for this bright tail .
photometric redshifts were derived using the technique described in @xcite and @xcite . in short ,
the method consists of : i ) checking photometric zeropoints and , if necessary , determining photometric offsets by comparing theoretical and observed stellar _
locii _ , ii ) computing object fluxes in a fixed aperture ( @xmath18 ) from seeing matched images , iii ) determining a redshift probability function p(z ) for each object by matching the object s fluxes against a set of template spectra covering a wide range of stellar population ages and star - formation histories .
the additional use of the galex fuv and nuv bands @xcite allowed some photo - z degeneracies to be broken , and a final accuracy of to be reached .
figure 1 shows the redshift distribution of the subsample with available morphological information used in this work .
the morphological catalog , including stellar mass estimates , analysed in this work is made publicly available at the following address : http://www.aoc.nrao.edu/@xmath1mpannell/data.html .
the method we use to infer stellar masses from multi
wavelength photometry is described in detail in @xcite .
it is based on the comparison of the observed multi
color photometry with a grid of stellar population synthesis models produced with the @xcite code .
we parametrize the possible star formation histories ( sfhs ) by a two - components model , consisting of a main , smooth component described by an exponentially declining star formation rate @xmath19 , linearly combined with a secondary burst of star formation .
the main component timescale @xmath20 varies in @xmath21 $ ] gyr , and its metallicity is fixed to the solar metallicity .
the age of the main component , @xmath22 , is allowed to vary between 0.5 gyr and the age of the universe at the object s redshift .
the secondary burst of star formation is modeled as a 100 myr old constant star formation rate episode of solar metallicity .
we adopt a @xcite initial mass function for both components , with lower and upper mass cutoffs of 0.1 and 100 @xmath23 . additionally , adopting the calzetti et al .
( 1994 ) extinction , both the main component and the burst are allowed to exhibit a variable amount of attenuation by dust with @xmath24 @xmath25 $ ] and @xmath26 $ ] for the main component and the burst , respectively .
this takes into account the fact that young stars are found in dusty environments and that the star light from the galaxy as a whole may be reddened by a ( geometry dependent ) different amount .
we compare this set of models with multi color photometry of each object , computing the full likelihood distribution in the 5-dimensional parameter space ( @xmath27 ) , the likelihood of each model being @xmath28 .
to compute the likelihood distribution of mass to light ratios @xmath29 , we weight the @xmath29 of each model by its likelihood and marginalize over all parameters .
the uncertainty in @xmath29 is obtained from the width of this distribution . for young objects , with relatively high burst fractions ,
the width of the @xmath29 distribution is usually much wider in the optical than in nir bands , while for quiescent objects the width of the @xmath29 distribution is very similar in all bands . on average , the width of the likelihood distribution of @xmath29 at 68% confidence level is between @xmath30 and @xmath31 dex ( using the b band @xmath29 ) .
the uncertainty in mass has a weak dependence on mass itself ( increasing with lower @xmath32 photometry ) , and it mostly depends on the spectral type : quiescent galaxies have more tightly constrained masses than star - forming ones .
we estimated the galaxy stellar masses for all the objects in the i band selected catalog down to the limiting magnitude . in figure [ logmz ]
we show the distribution of galaxy stellar masses as a function of redshift for the morphological catalog used in this work . for comparison , and to show the impressive statistic available , we overplot the morphological catalogs used in @xcite from the fdf ( yellow ) and goods - s ( magenta ) fields . in figure [ rhoz ] we show the stellar mass density up to redshift 1 for objects with log m@xmath33 , which corresponds to the mass completeness value for the highest redshift considered .
dashed lines represent 1@xmath34 estimates of the local mass density from @xcite after applying the same mass cut .
the error bars represent the cosmic variance contribution , as estimated in @xcite , that is a factor 10 larger than statistical errors . assuming that the mass function evolves only in normalization over this redshift range
, one should add 0.28 dex to the plotted values to recover total mass densities .
the mass completeness at different redshifts was estimated as the mass of a maximally old stellar population having , at the redshift considered , an observed magnitude equal to the catalog completeness magnitude ( e.g. * ? ? ?
* ) . in table
[ ttmas ] we provide the mass completeness values as a function of redshift for the morphological catalog ( f814w @xmath3524 ) , as well as for the global photometric catalog ( i @xmath36 ) . .log
m@xmath37/m@xmath38 completeness versus redshift [ cols="^,^,^,^,^,^,^",options="header " , ]
we use the package gim2d @xcite to fit psf convolved @xcite profiles to the two - dimensional surface brightness distribution of each object , for all sources down to a limit of f814w=24 .
the psfs used to convolve the profiles were obtained for each individual tile by median stacking about 50 high s / n stars .
we tested our results by running also the galfit @xcite code on some of the cosmos acs patches .
the results from the two different codes are in excellent agreement , thus confirming both the robustness of our modelling and the choice of flux limits .
we further tested our results by running the gim2d code on simulated images .
this test was performed by adding to the real image , in a blank , pure sky region , fake objects with srsic profiles and structural parameters spanning a wide range of plausible values .
the whole procedure is then run exactly as for real objects : we find that the profile parameters of simulated objects are well recovered down to f814w=24 .
high srsic index objects have very extended wings that , depending on the total flux , can fall under the sky surface brightness , thus for these objects a lower srsic index , total flux and effective radii are usually recovered ( see also @xcite ) .
we point out that the same trend was found in @xcite with a different fitting code , thus confirming the `` physical bias '' as contrasted to a code failure . in the local universe a tight correlation between srsic index and luminosity / mass of giant elliptical galaxies ( see e.g. @xcite and @xcite ) has been robustly established . due to flux limits and mass completeness , morphological analysis of high redshift ellipticals
is mostly limited to the most massive galaxies which one would expect to have relatively high srsic index .
our simulations show that for such high srsic index objects at faint apparent magnitudes ( low s / n ) we would recover srsic indexes and effective radii biased toward lower values .
while we will deal in detail with the galaxy size evolution in a future work , we argue here that the strong size evolution of early - type galaxies with redshift , recently reported in the literature ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , could be partly due to such bias .
we try to cope with this `` biased '' srsic index by adopting the procedure described in the following .
we perform extensive monte - carlo simulations to take into account the srsic index uncertainty ( @xmath530% , estimated through simulations ) on our results .
one thousand simulations of the morphological catalog were generated .
objects with srsic index smaller than 2.5 were perturbed within a gaussian of sigma equal to the 30% of the measured value .
objects with srsic index larger than 2.5 were perturbed within only the positive side of the gaussian so that their value could only get larger than the measured ones .
we tried also to implement different solutions to this correction procedure and the final results were always in good agreement .
more details on the morphological analysis are given in the appendix a. in the following we describe the analysis aimed at quantitatively determine the morphological properties of the galaxy sample we use in this work .
the morphological analysis is based on the acs imaging of the cosmos field in the f814w passband , and was performed on all galaxies in our sample down to a limit of f814w=24 .
we explain below the details of the adopted procedure and the tests that were performed to assess the robustness of our results .
we use the package gim2d @xcite to fit psf convolved @xcite profiles to the two - dimensional surface brightness distribution of each galaxy : @xmath86}\ ] ] where @xmath87 is the effective radius of the galaxy , @xmath88 is the surface brightness at @xmath87 , @xmath9 is the power - law index ( srsic index ) , and @xmath89 is coupled to @xmath9 such that half of the total flux is always within @xmath87 . for @xmath90 , @xmath91 ; at low @xmath9 , @xmath92 flattens out towards 0 and is obtained by interpolation .
the adjustable form of the srsic law has the advantage of parameterizing , through the variable exponent @xmath9 , surface brightness distributions including the exponential radial fall - off of the light profile in bulgeless disks ( @xmath11 ) , and the classical de vaucouleur profile typical of elliptical galaxies ( @xmath10 ) . during the fitting process ,
the model is convolved with the user specified psf , and then compared to the input image .
when fitting a single srsic model , gim2d seeks the best fitting values for the following eight parameters : total flux , half - light radius , ellipticity , position angle , center of the galaxy , background level and value of the srsic index .
the accurate determination of the image psf is important for the proper determination of the morphological parameters .
the psfs used to convolve the profiles were obtained for each individual acs tile by stacking about 10 high s / n isolated stars .
although we did not expect the outcome of single srsic fits to depend strongly on small variations of the psf , we carried out some tests with varying psfs on a few cosmos tiles to assess the influence of the psf on the retrieved morphological parameters .
such tests were performed by using : i ) a single universal psf obtained by stacking all the cosmos high s / n isolated stars ; ii ) the psf we derived for the fdf in pannella et al .
( 2006 ) , which was observed in the same filter f814w and at almost the same depth as cosmos .
we found that the largest differences obtained in the output parameters were of order 35 % , meaning that single srsic fits are pretty robust against such psf variations .
we also tested the accuracy of the psf itself by fitting all the point like sources as selected with the sextractor catalog parameters .
specifically , we used the neural network star classifier ( class_star ) , the half light radius , and the fwhm , to identify point like sources . when gim2d fitting such point like sources , the vast majority but the saturated stars
has a best fit half
light radius smaller than 0.01 pixel , meaning that after the code convolves them with the provided psf they are indeed recovered as point like sources .
since the cosmos acs tiles are slightly overlapping , we took advantage of this feature to further test the reliability of our fitting procedure as well as to assess many other possible systematics , like for instance psf variation in time and over the acs field of view , or photometric and astrometric distorsion .
all sources appearing in two tiles were fitted separately in each tile , and the fit results were compared .
the outcome of this test is shown in figure [ nnrr ] , where we plot one against the other the gim2d output ( srsic index and half light radius ) coming from the two fits for all such objects ( @xmath5 8 000 ) .
an overall accuracy of @xmath93 and @xmath94 is obtained .
taking into account that in the tile borders only three out of four dithers are overlapping , and hence that the image rms is greater than in the central part of the image , one should consider these uncertainties as safe upper limits .
we further tested the accuracy of our analysis by running gim2d on simulated images .
this test was performed by adding 3000 fake objects , one at a time , to the real image , in a blank , pure sky region .
the simulated galaxies were generated with the gim2d task _ gsimul _
, by adding on the selected pure sky region objects with structural parameters uniformly distributed in specified ranges ( magnitude in [ 22.5 24 .
] , ellipticity in [ 0.1 0.9 ] , position angle in [ 0 90 ] , effective radius in [ 0 60 pixel ] , srsic index in [ 0.26 ] ) .
each single object is first added into the image , then it is extracted with the same sextractor configuration file used for the real images .
it is then analyzed from gim2d exactly as done for the real objects .
the input magnitudes are generally better recovered for objects with @xmath95 than for objects with @xmath96 , with a median @xmath97 m = 0.01 and 0.06 , respectively ( see left panel of figure [ mags ] ) .
high srsic index objects have very extended wings that , depending on the total flux , can fall under the sky surface brightness , thus for these objects a lower srsic index , total flux and effective radii are usually recovered ( see right panel of figure [ pann ] ) .
we find that the profile parameters of simulated objects are well recovered down to f814w=24 ( see figure [ eeemam ] ) .
the recovered ellipticity and position angle , as compared to the input ones , show no trends with magnitude and are generally very well recovered ( see left panel of figure [ eeemam ] for the ellipticity ) .
we tested our results by running also the galfit @xcite code on four of the cosmos acs patches .
galfit is a 2d galaxy fitting software package more recent than gim2d , and it was designed to extract many structural components from galaxy images .
it is very flexible and able to fit an arbitrary number of galaxies simultaneously on an image , making it an ideal tool to fit neighboring objects in a sensitive way and not being biased from external informations such as the sextractor segmentations used by gim2d . during the fitting process , as for gim2d ,
the model is convolved with the specified psf , and then compared to the input image .
[ gio ] shows a comparison between the two codes .
no systematics are present , the median differences for both the recovered srsic index and half light radius are very close to zero .
the scatter is about 20% for the srsic index and 10% for the half light radius .
one can argue that the cosmos field is not very crowded and that in more crowded fields one could see some systematic differences between the two codes .
still , a similar comparison on deep acs images of the galaxy cluster a1689 @xcite for a sample of bright ( i @xmath3 22.5 ) red - sequence early type galaxies , beside a few obvious cases for which galfit performed best , showed no significant systematics between the two codes ( see also @xcite for slightly different but still similar conclusions on the comparison between the two codes ) . of course
these considerations apply only to single srsic fitting , no simulations and tests were done concerning bulge / disk decomposition .
@xcite show that there is a good correlation between the average visual and automated classifications as parametrized by the morphological type @xmath39 , assigned according to the @xcite classification scheme , and the srsic index @xmath40 , respectively ( see figure 1 in that paper ) .
following @xcite , we split our sample , according to the srsic index value in three broad morphological classes : early - type ( n @xmath41 3.5 , which corresponds to bulge - dominated / elliptical galaxies ) , intermediate ( 2 @xmath17 n @xmath35 3.5 ) and late - type ( n @xmath35 2 , which corresponds to disk - dominated galaxies ) .
we restrict our analysis to @xmath42 to limit the bias introduced by the rest - frame emission band shifting , namely the effect of the morphological k - correction .
a number of studies in the last years have shown , although not in a conclusive and quantitative way , that this is indeed a fair assumption ( e.g. * ? ? ?
* ; * ? ? ?
we performed extensive monte - carlo simulations to take into account the effect of mass uncertainties ( @xmath43dex ) on our results .
one thousand simulations of the mass catalog were generated , perturbing each mass within a gaussian of sigma equal to its error . unless stated differently in the relevant figure captions , we use the median values of the monte - carlo simulations in all figures .
error bars take into account both poissonian errors on the median counts @xcite , and 16 - 84@xmath44 percentile values of each distribution . in figure [ nndd ]
we show the v / v@xmath45 corrected number density evolution split by morphological type , up to redshift 1.2 , for objects in stellar mass ranges log m@xmath37=[10.6 - 10.8 ] , [ 10.8 - 11 ] , [ 11 - 11.2 ] , and [ 11.2 - 12 ] , as indicated in the upper right corner of each panel . by multiplying the number densities by the average mass in the corresponding mass range , each panel can be interpreted as the evolution of the total mass densities contributed by objects in that stellar mass range .
the dotted vertical lines show the estimated redshift completeness for the relevant mass bin .
we find that early type objects always dominate the high mass end .
the lower the stellar mass , the more late type objects tend to dominate the relative contribution to the total number / mass density .
the same trend is true by looking at increasing redshift : the higher the redshift the more the relative contribution of early and late types gets closer .
for the lowest mass range [ 10.6 - 10.8 ] the transition from a bulge dominated to a disk dominated stellar mass budget happens around redshift 0.7 ; beyond that redshift late - types are always contributing more to the total mass budget . in the mass range log m@xmath37 = [ 10.8 - 11 ] the transition is occurring at @xmath46 . at each redshift , a _ transition mass _ can be identified where the transition from a bulge dominated to a disk dominated stellar mass budget takes place .
based on the very good statistics available in this work , we are able to confirm the main conclusions of pannella et al .
( 2006 ) : _ i ) _ the morphological mix at the high mass end evolves with redshift ; _ ii ) _ the transition mass increases with redshift .
at @xmath47 we find a value of the transition mass approximately consistent with the local value @xmath48m@xmath38 , as measured in @xcite by using the concentration parameter to discriminate between early and late type objects in a complete sample extracted from local surveys . at @xmath49
the late and early types contributions become comparable at @xmath50m@xmath38 . in a fixed mass range the number densities of late - types are consistent with being constant with redshift , while the early types follow the more general total declining . at a fixed redshift ,
the number density of early types of different masses is very similar ( in + - 0.1/0.2 dex ) , while the late type number densities move by one order of magnitude passing from the high to the low stellar mass range .
this naturally sets the evolution of the transition mass with redshift .
the slope of the redshift evolution of total number densities is consistent in all mass ranges , at least up to the common redshift completeness , meaning that the mass function in its high - mass end , i.e. at least down to @xmath51m@xmath38 , does not evolve in shape but only in normalization with redshift .
in figure [ nndd ] there is obvious scatter around the mean number density decline in all the mass bins .
however , an underdensity at [email protected] shows up quite strikingly .
the deficit of objects is also quite evident in the global redshift distribution of our morphological catalog , as shown in figure [ zzdist ] .
it gets more and more significant with increasing stellar mass .
the deficit of galaxies at this redshift is a factor of three for the highest mass objects , and less than a factor of two for objects in the lowest mass bin explored .
this differential deficit is in qualitative agreement with galaxy evolution dependence on the local density as expected in current galaxy formation models ( see @xcite and references therein ) : the more underdense is the environment the later massive objects will assemble because there is in general a time delay in starting star formation and also there are less small galaxy units to assemble the massive giant galaxies . from the morphological point of view , the evolution in this underdense redshift bin is _ delayed _ at all stellar masses as compared with the contiguous redshift bins . the early - type fraction with respect to both the total number density and
the late - type number density is much lower than those in the contiguous redshift bins , in agreement with the expectations from the morphology - density relation .
we also note that in the highest mass bin , in this underdense region there are no massive late - type galaxies .
we notice that the morphological mix in this underdense redshift bin is , at all masses probed , very close to the morphological mix at [email protected] , which has a comparable total number density of objects .
taking advantage of the good statistics available in this study , we wish to further check the importance and role of environment on galaxy morphological evolution . estimating the local volume density in a photometric redshift survey is not straightforward , and we try here to make it as simple and unbiased as possible .
first of all one has to decide which are the objects tracing the overdensities in a way not biased by the flux limited survey : a density - defining population ( ddp , @xcite ) of galaxies is needed .
the objects have to be bright enough to be seen in the whole redshift range considered . dealing with masses and morphologies
, one should also take care of the selection effects introduced by the different m / l ratios associated with different stellar populations .
basically , old stellar populations , which more often but not always reside in bulges , have much higher m / l ratios with respect to young stellar populations , often but not always populating disks . a pure luminosity cut even taking into account passive evolution would thus preferentially cut out of the sample high m / l objects ( early - types ) with respect to low
m / l objects ( late - types ) .
to avoid this kind of bias , we use as ddp all objects with masses down to the mass completeness value at the highest redshift we are exploring ( 7x10@xmath52 solar masses ) .
another issue is the relative scarcity of these objects ( being in the exponential cutoff of the mass function ) , so one has to avoid reaching the shot noise regime when calculating number densities . with this in mind
, we proceed as follows : we first fit the number density evolution with redshift for objects more massive than 7x10@xmath52 solar masses , obtaining in this way an expected value of the number density , @xmath53 , at each redshift z. we then split the entire redshift range in slices of @xmath54 ( corresponding to differential depths of about about 327 comoving mpc at redshift 0.5 and 244 at redshift 1 ) .
this slicing along the redshift axis turns out to be quite robust against the photo - z errors , in fact this corresponds to about 2 times the formal photo - z error at redshift 0.5 and 1.5 times at redshift 1 .
we then split the entire survey volume in cells of 2.8 comoving mpc by side at each redshift slice , that is the maximum contiguous angular area we have available in the lowest redshift bin .
finally , we calculate the comoving number density of each cell at the different redshifts and assign the cell galaxy population to one of three density classes : underdense ( @xmath55 ) , medium - dense ( @xmath56 ) and overdense(@xmath57 ) . the limits used here to define environmental classes are chosen to balance the needs to map sensibly different environments and to assign a relatively large number of objects to each of the three classes . in figure [ selct ]
we show the number density evolution for objects with log m@xmath58 10.8 .
the solid line , @xmath59 , shows the linear fit to the filled dots ( i.e. the number densities obtained over the whole field in redshift bins of @xmath60 = 0.2 ) .
the dashed line sets the 1.4@xmath61 limit above which we select the overdense regions .
below the dashed - dot line ( @xmath62 ) we select the underdense regions , while in between the two lines we select the medium - dense regions .
the empty squares overplotted show the number densities in all the cells we have split our survey volume .
we run 100 realizations of the procedure adopted to define environmental densities by randomizing the initial position of the grid in the sky and also by adopting different solutions for the redshift slicing of the total volume .
the outcome of this testing convinced us that the results we are going to show and discuss in the following are robust and not biased against the specific procedure . in figure [ mf_env ]
we show the galaxy stellar mass function as a function of redshift , split by morphology and for the different environmental densities as defined in the previous section .
each of the panels contains about 2000 objects .
vertical dotted lines set the estimated mass completeness at the different redshifts . for reference in all panels
are overplotted local estimates for the stellar mass function @xcite , which are fixed and independent from the environment .
they are only meant as a reference to look at differential evolution in different environments . in the lowest range explored , at @xmath64 , the mass completeness reaches masses as low as 7@xmath6510@xmath66m@xmath38 .
this enables us to explore the behavior of the slope of the mass function in different environments .
the slope of the mass function is marginally evolving from the lower to the higher density regions , becoming increasingly shallower and consistent with the local _ field _ one .
this trend is in fact in good agreement with what found in the local universe @xcite .
the transition mass moves from log m@xmath67 at high density to log m@xmath68 at low density . in the overdense and medium - dense environments
the shape of the mass function is in a very good agreement with the local determinations . in the overdense regions ,
this happening at all redshifts , a very massive bulge - dominated population seem to appear that overcomes the local _ field _ determination .
figure [ mf_env ] shows that the morphologically split mass function evolves along both the redshift and the density axes .
the evolution is differential though : the redshift evolution is much milder than the environmental one .
this is in fact a crucial point in interpreting the results since in almost all previous studies there has been no chance to split these two very different contributions .
this is better explained by the following examples : the total mass function , and the morphological mixtures are basically the same , up to the common mass completeness value , for overdense volumes at redshift 1 and medium - dense volumes at redshift 0.4 .
the transition mass is close to log m@xmath37 = 11 at redshift 1 in medium - dense volumes and it is already at that mass in the underdense environments at @xmath64 .
this shows how by analyzing small patches of the sky one could mix up different environments at different redshifts and hence get the wrong evolutionary paths for different morphologies . with the aim of trying to better quantify the morphological mix evolution as a function of redshift and environment
, we integrate the mass function over the common complete mass range in the three redshift bins and in the three different environments . in figure [ mdr ]
we show both the early - type fraction ( lower panel ) and the early - to - late type classification ratio ( upper panel ) at different redshifts and for different environments ( low cyan , medium
magenta , black high density ) . at all redshifts a morphology - density relation , as already reported in the literature ( e.g. * ? ? ?
* ) , is showing up : denser volumes contain also a higher relative fraction of bulge - dominated galaxies with respect to lower density volumes .
although it is difficult to compare with previous results given the fact that in our study for the first time , to our knowledge , a mass complete sample has been adopted , it is very reassuring that qualitatively we are in agreement with previous studies .
this confirms the robustness of our galaxy environment definition procedure . as a function of redshift
, in the medium and low density environments there is an increase of the early - type population with cosmic time which is more pronounced between redshift 1 and 0.7 and then flattens out to redshift 0.4 . in the high density environments the early - type fraction grows between redshift 1 and 0.7 , and then seems to turn - over down to redshift 0.4 .
a similar behavior has been actually found in @xcite . in our opinion
the sky patch , we are using here , might not cover enough cosmic volume at redshift around 0.4 to allow an exact environmental definition .
this shows up in figure 6 where all but two sub volumes are in fact assigned to the medium dense environment bin . in figure [ ages ]
we show the ages of massive early - type galaxies as a function of redshift and for different environments as obtained from the sed fitting procedure outlined above .
the redshift ranges considered are indicated in the top label of each panel .
we cut the total sample to the common completeness limit ( @xmath41 7@xmath6510@xmath52m@xmath69 ) .
the dotted , dashed and solid histograms represent the high , medium and low density environments , respectively . in each panel , the difference in billion years between the mean ages in the two most extreme environments ( the high and low density regions ) is indicated .
such differences deviate from zero at a two sigma level at all redshifts , and might suggest an earlier formation epoch of early - type galaxies in high density environments with respect to those living in a low density region .
the results found are in good agreement with recent studies ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . from figure [ ages ] three things
are worth noting : first , the histograms are quite similar to each other within each single panel ( as confirmed by a kolmogorov - smirnov test ) but for the oldest and youngest tails of the distribution which preferentially belong to the highest and the lowest density regions ; second , the width of the distributions is increasing with the cosmic clock , this being partially driven by the narrower range of ages allowed at earlier cosmic times ; third , in the lowest redshift bin a number of , relatively young , early - type objects are present which can not be the descendants of the higher redshifts early - type populations simply evolved by passive evolution .
we interpret these evidences as follows : i ) young massive early type galaxies are preferentially found in low density environments at all redshifts , ii ) the bulk of the early type stellar populations have similar characteristic ages independent of the environment ;
iii ) the oldest objects in the universe , or better saying the oldest stars in the universe , happen to belong to its highest density regions ; iv ) newly formed , or somehow rejuvenated , bulge - dominated galaxies are entering the sample with cosmic time .
this latter finding might explain why the difference in age seems to increase with the cosmic clock and reach the 2 gyrs quoted by thomas et al .
( 2005 ) in the local universe .
the estimate of stellar population ages suffers from some important degeneracies and has to be treated with _ a pinch of salt_. the sed modelling we have described in section 3 only considers stellar population models with solar metallicity .
this means we are not tackling the well known age - metallicity degeneracy : old stellar populations with low metallicities and younger stellar populations with high metallicities may well have very similar seds . only recently @xcite , but see also @xcite for different conclusions , reported on a mild observed correlation between metallicity and environment : massive galaxies in high - density environments are slightly more metal rich than galaxies belonging to low - density environments .
if such a correlation is indeed at work our results are biased and the correct differences in age would be even smaller than the ones reported in the panels of figure 9 .
we remind that our classification scheme does not allow any separation between ellipticals and early - type spirals . therefore , as happens in the local universe ( see e.g. * ? ? ?
* ) our early - type samples will be likely dominated by elliptical and s0 galaxies in overdense environments and by early - type spirals in low density regions . in figure [ ubcoll ]
we show the rest - frame u - b color as a function of mass , redshift , morphology and environment .
morphological classification is highlighted in different colors ( red = early , green = intermediate , blue = late ) .
contours are used to better show the different behavior of the three morphological classes .
the contours correspond to equally spaced levels between the minimum and maximum density of the data points of each morphological set .
the first three rows of panels starting from the top , show the high , medium and low density environments .
the bottom row shows instead the total population , _
i.e. _ all the different environments together .
redshift ranges are indicated , for each column , in the top label .
the red - sequence , mostly populated by early - types , in all environments becomes slightly redder with cosmic time because of the ageing of the stellar populations .
the evolution in the rest - frame u - b color is mild , as also shown for instance by @xcite up to [email protected] .
the red - sequence upper envelope reddens by @xmath70 mag in the redshift range [email protected] , in agreement with simple passive evolution expectations .
the bulge - dominated galaxy population is mostly confined to the red - sequence with some peculiarities dependent on environments .
at all redshifts , in higher density environments early - types tend to be more confined to the massive end of the red - sequence . on the other hand , in lower density environments early - types tend to populate the red - sequence down to lower masses ( as it is easier to see at low redshifts ) and reach into the blue - cloud .
this definitely points toward a difference in age between the bulge - dominated populations in different environments .
it is worth noting the evolution of the intermediate
type population : these are mainly spiral galaxies hosting fairly massive central bulges . at redshift 1 , intermediate type galaxies overlap in the color mass diagram with both the early and late populations in all environments .
as time goes by , this class of objects tend to leave the very high mass end of the red sequence in favour of bulge - dominated galaxies , and to drift toward the bluer late - type population .
this suggests that part of these intermediate type galaxies have moved to the early - type population .
the change of this class of objects with cosmic time correlates with the environmental density : at low redshifts , intermediate type objects in more dense environments tend to occupy the _ valley _ between the red sequence and the blue
cloud , while in low density environments they split in two populations , with the majority well overlapping the late - type population , and a minor sub population remaining in red - sequence together with the massive bulge - dominated galaxies .
the late - type population is confined at all redshifts and in all environments to the blue cloud .
given the mass completess limits it is definitely hard to make strong conclusions , but these data suggest the presence , possibly more pronounced at higher redshift and in lower density environments , of a population of massive disk - dominated galaxies reaching toward the red - sequence . in section 7.1 we have investigated the evolution with redshift of the morphology density relation ( see figure 8) . since figure 10 shows that at all redshifts early - types populate mainly the red sequence while late - type galaxies are mostly in the blue cloud , we can infer from our data that a color density relation is in place up to the highest redshift explored .
@xcite analysed a sample of 2900 galaxies with secure spectroscopic redshift determinations drawn from the vimos vlt deep survey ( vvds , @xcite ) in an area 4 times smaller than the one considered in this work and with @xmath532% sampling , slightly biased toward faint objects , of the whole galaxy population .
they obtain a very similar result to the one we present in this work up to redshift 1.2 , while claiming a dramatic evolution in the redshift bin 1.21.5 with blue star forming galaxies belonging preferentially to the highest density environments .
since they define environmental properties based on a volume limited sample , i.e. an absolute magnitude cut at all redshifts , which is biased against old stellar populations , it would be interesting to check whether this effect persists if defining environmental densities with a mass complete sample .
we postpone a more detailed investigation of this issue to a forthcoming paper . as of today , only a photometric reshift survey like cosmos , because of the high statistic available , may in fact allow such an investigation .
@xcite analysed a sample of 2041 galaxies with photometric redshifts within 0.61 @xmath35 z @xmath35 0.85 and covering an area of 270 arcmin@xmath72 , that is about 9 times smaller than the one we use here .
the accuracy of the their photometric redshifts is comparable to the one in this work ( see @xcite ) .
they perform a morphological classification up to i@xmath73 @xmath35 24 , thus the same flux limit we use in our analysis , using a set of nonparametric measurements of the galaxy light distribution @xcite .
also , following @xcite they estimate galaxy projected densities .
they do find a well defined morphology - density relation in place and also a clear separation between early and late types , in all environments , in the stellar mass vs. v - z observed color plane , with bulge - dominated galaxies sitting mainly in the red sequence and disk - dominated ones populating the blue cloud .
furthermore , they fit and discuss extensively the different slopes of the blue cloud ( late - type galaxies ) and the red sequence ( early - type galaxies ) distributions .
we see similar characteristics in our data but , in our opinion , this is an effect mostly due to the sample mass completeness . as it is clear by looking at the mid column of figure 10 , the pronounced slope of the late - type galaxies distribution , compared to the bulge - dominated galaxy population , is mostly drawn by the mass completeness : red objects with such low mass , if existing , would have an i band magnitude fainter than 24 .
nonetheless , their main results are in excellent agreement with the one presented in this work at [email protected] , in spite of the different ways used to estimate galaxy environmental properties .
one way to explore the contribution of star formation to the stellar mass growth in galaxies of different mass , is to study the redshift evolution of the specific sfr ( ssfr ) , which is defined as the sfr per unit stellar mass .
a number of works in the last years ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) have focused on the ssfr redshift evolution , consistently finding an increase of the mean ssfrs with redshift .
following @xcite , the sfr of individual objects can be estimated from the rest - frame uv luminosity as @xmath74 in units of m@xmath38yr@xmath75 , where the constant factor is computed for a salpeter imf .
the sfr of each galaxy is corrected by the dust attenuation obtained for the sed fitting using the extinction curve of @xcite . in figure [ ssfrl ]
we present the ssfr as a function of stellar mass and morphology , in three redshift ranges up to redshift 1.15 , and in different environments .
the figure shows contour isodensity levels drawn as in figure 10 . before discussing in some detail figure 11
, we would like to first discuss some evident similarities with figure 10 .
the rest - frame u - b color is bracketing the balmer break of a galaxy spectrum , hence is mostly sensitive to the age of the galaxy stellar population ( e.g. * ? ? ?
* ) . to convert an age in a star formation rate , one needs , to first order and neglecting the dust attenuation correction , to normalize it to the underlying continuum emission or to the galaxy stellar mass _ i.e. _ : @xmath76
taking this into account one can , again roughly speaking , link the two plots with the following conversion : @xmath77 . with this simple conversion in mind ,
it is easier to understand why there are so many similarities between the two figures .
in both figures the red and blue clouds are very well separated , at all redshifts and in all environments .
this is more pronounced in high density environments , while going to lower density regions there is clear trend of the low mass early types to head toward high ssfrs , or bluer u - b color , values .
these galaxies are young blue bulge - dominated galaxies with high star formation activity and stellar masses of order a few times 10@xmath78 solar masses .
the intermediate type galaxies fall in between , again confirming their nature of transition objects .
it is worth noting though that while they overlap with the low mass star forming late - type population at all redshifts and in all environments , they tend to leave the high mass end at progressively lower redshifts . coming more specifically to figure 11 , the location in the ssfr vs stellar mass plot of the maximum galaxy density of the early - type population , is increasing its star formation by not more than a factor three over the explored redshift range , while the blue cloud is moving by approximately a factor ten in the same redshift range .
the two clouds are pretty much segregated in mass : while late - types are always dominating at low masses , the early and intermediate type objects completely dominate the high mass tail .
this mass segregation gets definitely stronger in denser environments .
the upper envelope of the ssfr is running essentially parallel to lines of constant sfr , and it is shifting to higher sfrs with increasing redshift , as it was already noted in earlier work . note that this upper envelope is partly affected by a selection effect : heavily dust obscured starbursts can not be detected in our sample .
furthermore , it is evident that the most massive galaxies have the lowest amount of star formation per unit stellar mass up to the highest redshift probed , and hence they can not have formed the bulk of their stars in this redshift range .
this pushes the bulk of their star formation at earlier cosmic epochs , in agreement with the downsizing scenario @xcite .
@xcite analysed a sample of 609 sources in the cosmos field , with both optical spectra from the zcosmos - dr2 and a mips 24@xmath79 photometric measurement available . in their study
they compared different kinds of star formation rate indicators , _
i.e. _ uv light , line emissions and ir luminosity , thus allowing an investigation of the dust extinction affecting these diagnostics . as a main result , at least for what concerns our work here
, they find that the calzetti et al .
( 1994 ) attenuation law works well for de - reddening both the uv light and the h@xmath80 , h@xmath81 line fluxes .
although the median @xmath82 of our sample , as derived from the sed fitting , turns out to be somewhat lower than the one they find for the mips 24@xmath79 selected sample , we note that this might be partially due to a selection effect of the @xcite sample .
we postpone a more careful analysis of this issue to a future work , which will make use of the recently available spitzer space telescope mid
infrared observations @xcite . in figure [ ssfrl ]
it is also worth noting the evolution of the ssfr as a function of the environment . at a first glance
, there is not much difference between the different environments .
nonetheless , at all redshifts there is a tendency of the early type sample to present a plume toward the high ssfr zone , which is very pronounced at lower environmental densities .
in this work we have studied the evolution of the stellar mass content of disk - dominated and bulge - dominated galaxies with respect to the total mass budget up to @xmath83 , and its dependency from the local environment .
as expected , we find that early type objects always dominate the high mass end , while at progressively lower stellar masses late type objects increase their contribution to the total mass density . at each redshift
, a _ transition mass _ can be identified as the stellar mass where the transition from a bulge dominated to a disk dominated stellar mass budget takes place .
based on the good statistics available in this work , we are able to confirm that the morphological mix at the high mass end evolves with cosmic time , with the transition mass increasing with redshift . at @xmath47
we find a transition mass approximately consistent with the local value @xmath48m@xmath38 , while at @xmath49 the late - types and early - types contributions become comparable at @xmath50m@xmath38 .
the stellar mass function of disk - dominated galaxies is consistent with being constant over the redshift range [ 0.3 - 1.2 ] , with number densities declining by more than an order of magnitude in the mass range log m@xmath37 = [ 10.6 12 ] . on the other hand ,
the stellar mass function of bulge - dominated systems shows a pure normalization decline with redshift .
such different behavior of the stellar mass functions of late - type and early - type galaxies naturally sets the redshift evolution of the transition mass . the slope of total number density evolution with redshift is consistent , within the errors , with being the same in all mass ranges , at least up to the common redshift completeness .
this means that the galaxy stellar mass function , in the redshift range and stellar mass range explored , does not evolve in shape but only in normalization with redshift ( see also @xcite for similar conclusions at higher redshifts ) . according to our data we do not see any differential evolution between the high mass and the low mass end of the galaxy stellar mass function .
we have identified in the photometric redshift distribution a large scale structure underdensity at z@xmath5 0.5 .
it gets more and more significant with increasing stellar mass .
the morphological evolution in this underdense redshift bin is _ delayed _ at all stellar masses : the early - type fraction , both with respect to the total number density and to the late - type number density , is much lower than those in the contiguous redshift bins , in agreement with the expectations of the morphology - density relation .
we notice that the morphological mix in this underdense redshift bin is , at all masses probed , very close to the morphological mix at [email protected] , which has a comparable total comoving number density of objects .
we further explored the redshift evolution of the galaxy stellar mass function split by morphology and by environmental density . in overdense and medium - dense environments , the mass function shape is in good agreement with local determinations .
however , we find that the slope of the mass function is anti - correlated with the environmental density : the less dense is the environment the more steep is the mass function .
this trend has also been found locally by @xcite . in overdense regions , at all redshifts probed , a very massive early - type population seems to exist , which overcomes the local _ field _ mass function determination : we might be witnessing the appearance of giant elliptical galaxies in the highest density regions .
the morphologically - split mass content evolves with both redshift and local density , with a striking feature : at different redshifts , the morphological mix and the transition mass appear to mainly depend on the local number density .
we have presented both the fraction of early - type galaxies with respect to the total galaxy population and the ratio between early to late
type objects at different redshifts and for different environments , finding that a morphology density relation is already well in place at redshift 1 .
we have found that early and late type galaxies in the stellar mass vs. ssfr plane ( or stellar mass vs. rest - frame u - b color ) are well separated at all redshifts and in all environments .
the intermediate type galaxies fall in between , confirming their nature of transition objects . at increasing redshift , the peak of the early type galaxy distribution in such a plane is moving toward higher ssfr by a factor @xmath84 . at the same time , the peak of the late type galaxy distribution shifts by approximately a factor ten in the same redshift range .
the early and late type galaxy populations exhibit a significant segregation in mass : late - type galaxies dominate at low masses while early type and intermediate objects dominate the high mass tail .
therefore , early type and intermediate galaxies drive the evolution of star formation present in massive ( @xmath85 ) galaxies , while lower mass disk - dominated galaxies are mostly responsible for the global decline of the cosmic star formation from redshift one to the local universe .
while in general this picture seems to be quite similar in all environments , in low density regions there is a population of relatively massive , early - type galaxies , having high ssfr and blue colors .
we also explored , with the highly homogeneous dataset available , the age of the massive early type galaxy stellar populations as a function of environment .
the massive early type galaxies have similar characteristic ages , colors , and ssfrs , hence a similar formation redshift independently of the environment they belong to .
the age distributions in different environments show a difference in the mean ages ( neglecting the age - metallicity degeneracy ) which is in a good agreement with published results .
finally , in agreement with previuos work , we have found that the galaxies hosting the oldest stars in the universe preferentially belong to the highest density regions .
we thank the anonymous referee for constructive comments which helped us to improve significantly the presentation of our results .
mp and vs are grateful for support from the max - planck society and the alexander von humboldt foundation .
part of this work was supported by the german _ deutsche forschungsgemeinschaft , dfg _ sfb 375 .
mp thanks gabriella de lucia , daniele pierini and stefano zibetti for a careful reading of the manuscript and for many insightful comments and suggestions .
we gratefully thank the entire cosmos collaboration for the huge amount of work and high quality data released to the whole community .
the hst cosmos treasury program was supported through nasa grant hst - go-09822 . more information on the cosmos survey is available at http://www.astro.caltech.edu/cosmos .
the national radio astronomy observatory is a facility of the national science foundation operated under cooperative agreement by associated universities , inc .
this research has made use of the nasa/ ipac infrared science archive , which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
130 natexlab#1#1 , r. g. , nair , p. , mccarthy , p. j. , et al .
2007 , , 669 , 184 , r. g. , van den bergh , s. , & nair , p. 2003
, , 588 , 218 , r. g. , tanvir , n. r. , santiago , b. x. , et al . 1996 , , 279 , l47 , s. , walcher , c. j. , le fvre , o. , et al .
2007 , , 476 , 137 , i. k. , balogh , m. l. , bower , r. g. , et al . 2006 , , 373 , 469 , i. k. , glazebrook , k. , brinkmann , j. , et al .
2004 , , 600 , 681 , s. p. , nichol , r. c. , baldry , i. k. , et al .
2009 , , 393 , 1324 , m. , rix , h .- w . , somerville , r. s. , et al .
2005 , , 635 , 959 , a. e. , drory , n. , hill , g. j. , & feulner , g. 2005 , , 621 , l89 , e. f. , mcintosh , d. h. , katz , n. , & weinberg , m. d. 2003 , , 149 , 289 , r. , appenzeller , i. , bhm , a. , et al .
2001 , in eso / ecf / stsci workshop on deep fields , ed . s. christiani ( berlin : springer ) , 327 , r. , burstein , d. , & faber , s. m. 1992 , , 399 , 462 , n. , ford , h. , bouwens , r. , et al .
2004 , , 150 , 1 , e. & arnouts , s. 1996 , a&as , 117 , 393 , j. p. , holden , b. p. , franx , m. , et al .
2006 , , 644 , 30 , m. r. , hogg , d. w. , bahcall , n. a. , et al .
2003 , , 592 , 819 , a. , meisenheimer , k. , bell , e. f. , et al .
2006 , , 453 , 869 , r. j. , illingworth , g. d. , franx , m. , & ford , h. 2007 , , 670 , 928 , r. j. , illingworth , g. d. , thompson , r. i. , et al .
2004 , , 606 , l25 , r. g. , benson , a. j. , malbon , r. , et al .
2006 , , 370 , 645 , f. , lerchster , m. , seitz , s. , bender , r. , & snigula , j. 2008 , e - prints arxiv:0811.3211 , j. & ellis , r. s. 2000 , , 536 , l77 , g. & charlot , s. 2003 , , 344 , 1000 , k. , ellis , r. s. , conselice , c. j. , et al .
2006 , , 651 , 120 , k. , ellis , r. s. , & conselice , c. j. 2005 , , 625 , 621 , d. , kinney , a. l. , & storchi - bergmann , t. 1994 , , 429 , 582 , n. , capaccioli , m. , & donofrio , m. 1993 , , 265 , 1013 , p. , abraham , r. g. , ellis , r. s. , et al .
2007 , , 172 , 284 , p. , aussel , h. , ajiki , m. , et al .
2007 , , 172 , 99 , n. , brusa , m. , hasinger , g. , et al .
2009 , e - prints arxiv:0901.2347 , k. i. , lilly , s. j. , aussel , h. , et al .
2008 , , 680 , 939 , p. , guzzo , l. , franceschini , a. , et al .
2007 , , 172 , 270 , a. , dekel , a. , faber , s. m. , & guiderdoni , b. 2008 , , 389 , 567 , a. , cassata , p. , pozzetti , l. , et al .
2008 , , 482 , 21 , a. , daddi , e. , & renzini , a. 2006 , , 453 , l29 , s. , norberg , p. , baugh , c. m. , et al .
2001 , , 326 , 255 , s. , lacey , c. g. , baugh , c. m. , & frenk , c. s. 2000 , , 319 , 168 , c. j. 2003 , , 147 , 1 , m. c. , tremonti , c. a. , newman , j. a. , & zabludoff , a. i. 2008 , , 390 , 245 , l. l. & barger , a. j. 2008 , , 686 , 72 , l. l. , songaila , a. , hu , e. m. , & cohen , j. g. 1996 , , 112 , 839 , t. j. , younger , j. , hernquist , l. , & hopkins , p. f. 2008 , in iau symposium , vol .
245 , iau symposium , 6366 , d. j. , farrar , g. r. , norberg , p. , et al .
2005 , , 356 , 1155 , o. , iovino , a. , marinoni , c. , et al .
2006 , , 458 , 39 , m. , labb , i. , franx , m. , et al .
2009 , , 690 , 937 , g. , springel , v. , white , s. d. m. , croton , d. , & kauffmann , g. 2006 , , 366 , 499 , g. , de vaucouleurs , a. , corwin , h. g. , et al .
1991 , third reference catalogue of bright galaxies ( volume 1 - 3 , xii , 2069 pp . 7 figs ..
springer - verlag berlin heidelberg new york ) , g. 1948 , annales dastrophysique , 11 , 247 , a. , birnboim , y. , engel , g. , et al .
2009 , , 457 , 451 , m. , papovich , c. , ferguson , h. c. , & budav ' ari , t. 2003 , , 587 , 25 , s. & davis , m. 1987 , , 313 , 59 , a. , lynden - bell , d. , burstein , d. , et al .
1987 , , 313 , 42 , a. 1980 , , 236 , 351 , n. & alvarez , m. 2008 , , 680 , 41 , n. , bender , r. , feulner , g. , et al .
2004 , apj , 608 , 742 , n. , bender , r. , & hopp , u. 2004 , , 616 , l103 , l. , ivison , r. j. , maddox , s. , et al .
2009 , , 394 , 3 , g. , gabasch , a. , salvato , m. , et al .
2005 , apj , 633 , l9 , g. , goranova , y. , drory , n. , hopp , u. , & bender , r. 2005 , , 358 , l1 , a. , salimbeni , s. , grazian , a. , et al .
2006 , , 459 , 745 , a. , pozzetti , l. , donnarumma , i. , et al .
2004 , a&a , 424 , 23 , f. , de lucia , g. , monaco , p. , somerville , r. s. , & santini , p. 2009 , e - prints arxiv:0901.1130 , h. c. , clampin , m. , hartig , g. f. , et al .
2003 , in presented at the society of photo - optical instrumentation engineers ( spie ) conference , vol . 4854 , future euv / uv and visible space astrophysics missions and instrumentation , ed . j. c. blades & o. h. w. siegmund , 8194 , a. , rodighiero , g. , cassata , p. , et al .
2006 , , 453 , 397 , k. c. 1970 , , 160 , 811 , a. , goranova , y. , hopp , u. , noll , s. , & pannella , m. 2008 , , 383 , 1319 , a. , bender , r. , seitz , s. , et al .
2004 , , 421 , 41 , a. , salvato , m. , saglia , r. p. , et al .
2004 , apj , 616 , l83 , a. , charlot , s. , brinchmann , j. , white , s. d. m. , & tremonti , c. a. 2005 , , 362 , 41 , n. 1986 , , 303 , 336 , e. , salimbeni , s. , menci , n. , et al .
2005 , , 622 , 116 , m. , ferguson , h. c. , koekemoer , a. m. , et al .
2004 , , 600 , l93 , r. , rosati , p. , strazzullo , v. , et al .
2008 , , 488 , 853 , a. w. , trujillo , i. , & caon , n. 2001 , , 122 , 1707 , a. , seitz , s. , & pannella , m. 2006 , , 372 , 1425 , b. , mcintosh , d. h. , barden , m. , et al .
2007 , , 172 , 615 , j. , appenzeller , i. , gabasch , a. , et al .
2003 , , 398 , 49 , p. f. , cox , t. j. , kere , d. , & hernquist , l. 2008 , , 175 , 390 , j. , buat , v. , hernndez - fernndez , j. , et al .
2007 , , 670 , 279 , o. , capak , p. , salvato , m. , et al .
2009 , , 690 , 1236 , o. , salvato , m. , le floch , e. , et al .
2009 , e - prints arxiv:0903.0102 , m. g. & white , s. d. m. 2007 , , 376 , 2 , a. m. , aussel , h. , calzetti , d. , et al .
2007 , , 172 , 196 , o. , vettolani , g. , garilli , b. , et al .
2005 , , 439 , 845 , a. , massey , r. , kneib , j .- p .
, et al . 2007 , , 172 , 219 , j. 2006 , , 644 , l5 , s. j. , le fvre , o. , renzini , a. , et al . 2007 , , 172 , 70 , t. 2008 , , 179 , 319 , j. m. , primack , j. , & madau , p. 2004
, , 128 , 163 , p. , pozzetti , l. , & dickinson , m. 1998 , apj , 498 , 106 , d. , van dokkum , p. g. , forster schreiber , n. m. , et al .
2008 , e - prints arxiv:0811.1773 , d. h. , bell , e. f. , rix , h .- w .
, et al . 2005 , , 632 , 191 , b. , capak , p. , scoville , n. z. , et al . 2007 , , 172 , 117 , b. , dahlen , t. , hopkins , a. , et al . 2009 , , 690 , 1074 , m. ,
baldry , i. k. , & bamford , s. p. 2007
, , 382 , 801 , m. , carilli , c. l. , daddi , e. , et al .
2009 , e - prints arxiv:0905.1674 , m. , hopp , u. , saglia , r. p. , et al .
2006 , , 639 , l1 , c. y. , ho , l. c. , impey , c. d. , & rix , h .- w .
2002 , , 124 , 266 , p. g. , rieke , g. h. , villar , v. , et al .
2008 , , 675 , 234 , l. , bolzonella , m. , lamareille , f. , et al .
2007 , , 474 , 443 , s. , ferguson , h. c. , conselice , c. , et al .
2004 , , 604 , l9 , a. , rosati , p. , nonino , m. , et al .
2008 , e - prints arxiv:0806.4604 , h .- w . ,
barden , m. , beckwith , s. v. w. , et al .
2004 , apjs , 152 , 163 , g. , rix , h .- w . , franx , m. , et al .
2003 , , 599 , 847 , e. e. 1955 , , 121 , 161 , d. b. , salvato , m. , aussel , h. , et al .
2007 , , 172 , 86 , p. , fontana , a. , grazian , a. , et al .
2009 , e - prints arxiv:0905.0683 , m. t. , carollo , c. m. , lilly , s. j. , et al .
2007 , , 172 , 434 , m. & thompson , d. 2006 , , 648 , 299 , c. , carollo , c. m. , lilly , s. , et al .
2007 , , 172 , 406 , c. , carollo , c. m. , lilly , s. j. , et al .
2007 , , 172 , 494 , n. , abraham , r. g. , aussel , h. , et al .
2007 , , 172 , 38 , n. , aussel , h. , brusa , m. , et al .
2007 , , 172 , 1 , j. l. 1968 , atlas de galaxias australes ( cordoba , argentina : observatorio astronomico , 1968 ) , s. , mo , h. j. , white , s. d. m. , et al .
2003 , , 343 , 978 , l. , koo , d. c. , faber , s. m. , et al .
1999 , , 519 , 563 , g. p. , treu , t. , ellis , r. s. , moran , s. m. , & dressler , a. 2005 , , 620 , 78 , r. s. , lee , k. , ferguson , h. c. , et al .
2004 , , 600 , l171 , d. , maraston , c. , bender , r. , & mendes de oliveira , c. 2005 , , 621 , 673 , i. , feulner , g. , goranova , y. , et al .
2006 , , 373 , l36 , i. , frster schreiber , n. m. , rudnick , g. , et al .
2006 , , 650 , 18 , i. , rudnick , g. , rix , h .- w .
, et al . 2004 , , 604 , 521 , r. b. & fisher , j. r. 1977 , , 54 , 661 , a. 2008 , , 675 , l13 , p. g. , franx , m. , kriek , m. , et al .
2008 , , 677 , l5 , p. g. & van der marel , r. p. 2007
, , 655 , 30 , r. e. , blacker , b. , dickinson , m. , et al .
1996 , aj , 112 , 1335 , c. n. a. , faber , s. m. , koo , d. c. , et al .
2006 , , 647 , 853 , m. a. , schiminovich , d. , rich , r. m. , et al .
2007 , , 172 , 468 , x. z. , bell , e. f. , papovich , c. , et al .
2007 , , 661 , l41
in figure [ uno ] we show the comparison between the photometric redshifts used in this work and the recently released zcosmos dr2 spectroscopic redshifts catalog @xcite .
4506 objects are in common with our catalog .
the distribution of @xmath98 has a median value of 0.01 and an rms of 0.04 with only 2.5% of the distribution having a _ catastrophic _ photometric redshift , _
i.e. _ a @xmath98 greater than 0.2 .
this nicely confirms the results described in gabasch et al.(2008 ) , which were obtained by using a much smaller spectroscopic redshifts sample . in the left panel of figure [ due ]
we compare the photometric redshifts used in this work with the ones from the recently released ilbert et al .
( 2009 ) cosmos legacy catalog for a sample of 40717 objects common to both catalogs .
we note that not all the objects of our morphological catalog are matched because the ilbert et al .
( 2009 ) sample does not contain the photometric redshifts for all the x - ray sources detected in the xmm cosmos images @xcite .
the relative accuracy between the two photometric redshift realizations is about 5% , which is in a good agreement with the one expected from the absolute errors quoted in the two studies .
finally , only 4% of the common objects have photometric redshifts that differ by more than 20% . in the right panel of figure
[ due ] we show the redshift distribution of our sample obtained using the ilbert et al . ( 2009 ) catalog ( dotted line ) as compared to the one we are using in the work ( solid line , see figure 1 ) .
we note that the two distributions agree very well , including the detection of the large scale structure underdensity at [email protected] . | the cosmic evolution survey ( cosmos ) allows for the first time a highly significant census of environments and structures up to redshift one , as well as a full morphological description of the galaxy population . in this paper
we present a study aimed to constrain the evolution , in the redshift range @xmath0 , of the mass content of different morphological types and its dependence on the environmental density .
we use a deep multicolor catalog , covering an area of @xmath10.7 @xmath2 inside the cosmos field , with accurate photometric redshifts ( i @xmath3 26.5 and ) .
we estimate galaxy stellar masses by fitting the multi - color photometry to a grid of composite stellar population models .
we quantitatively describe the galaxy morphology by fitting psf convolved srsic profiles to the galaxy surface brightness distributions down to f814 = 24 mag for a sample of 41300 objects .
we confirm an evolution of the morphological mix with redshift : the higher the redshift the more disk - dominated galaxies become important .
we find that the morphological mix is a function of the local comoving density : the morphology density relation extends up to the highest redshift explored .
the stellar mass function of disk - dominated galaxies is consistent with being constant with redshift .
conversely , the stellar mass function of bulge - dominated systems shows a decline in normalization with redshift .
such different behaviors of late - types and early - types stellar mass functions naturally set the redshift evolution of the _ transition _ mass .
we find a population of relatively massive , early type galaxies , having high ssfr and blue colors which live preferentially in low density environments .
the bulk of massive ( @xmath4 ) early type galaxies have similar characteristic ages , colors , and ssfrs independently of the environment they belong to , with those hosting the oldest stars in the universe preferentially belonging to the highest density regions .
the whole catalog including morphological information and stellar mass estimates analysed in this work , is made publicly available . |
the discovery of a higgs boson @xcite is one of the major goals of the present searches in particle physics .
the higgs boson of the standard model ( sm ) is until now undiscovered but direct or indirect results give stringent lower bounds on its mass .
moreover , the problematic scalar sector of the sm can be enlarged and some simple extensions of the sm with several higgs bosons have been intensively studied for many years .
the simplest extension is the two higgs doublet model ( thdm ) @xcite , the two most commonly studied versions being classified as type i and ii .
they differ in how the higgs bosons are coupled to the fermions , although both versions possess identical particle spectra after electroweak symmetry breaking @xcite . from the 8 degrees of freedom
initially present in the 2 higgs doublets , 3 correspond to masses of the longitudinal gauge bosons , leaving 5 degrees of freedom which should be manifested as 5 physical higgs particles ( 2 charged higgs @xmath6 , 2 cp even @xmath7 , @xmath8 and one cp odd @xmath0 ) .
model type ii is the structure found in the minimal supersymmetric standard model ( mssm ) .
until now no higgs boson has been discovered , and from the null searches one can derive direct and indirect bounds on their masses .
the latest such limits are @xmath9 gev @xcite and @xmath10 gev @xcite gev @xmath11 gev . ] .
we note that in the general thdm which will be considered in this paper the existence of a very light @xmath8 or @xmath0 is not excluded by the current searches ( @xcite and refs therein ) .
higgs boson production in association with gauge bosons has been extensively studied in the literature : * higgsstrahlung ( @xmath12 ) both for the sm higgs boson @xcite or for higgs bosons originating from extended higgs sectors @xcite ( thdm and mssm ) . *
associated production with a photon ( @xmath13 ) @xcite , @xcite * associated production of charged higgs bosons with a w gauge boson both at @xmath14 colliders @xcite , @xcite , and at hadron colliders @xcite .
odd higgs bosons can be produced at @xmath14 colliders @xcite via @xmath15 and @xmath16 @xcite,@xcite .
pair production @xmath17 is small @xcite , as is the analogous process at @xmath18 colliders @xcite . in this paper
we consider the rare decay @xmath19 in the context of the thdm .
such a process is forbidden at tree level and proceeds via loops , with the rate at hadron colliders evaluated in ref .
calculations of the loop mediated decay @xmath20 have appeared in refs .
@xcite , @xcite , but not for the full process @xmath19 . although not expected to give a rate comparable to the mechanisms above , if observable this decay would be a test of the thdm at the one loop level , since this mechanism would be sensitive to new physics entering in the loop .
in addition it is important to have an accurate prediction for @xmath21 in order to distinguish between possible signals from a cp violating and cp conserving thdm . in a cp violating thdm tree
level mixing is allowed between the pure cp even states and pure cp
odd state , giving rise to 3 neutral higgs scalars ( @xmath22 ) with no definite cp quantum numbers . thus all three neutral scalars may be produced in the higgsstrahlung process @xmath23 , via a tree
level vertex @xmath24 . therefore it is important to know the possible magnitude of @xmath21 in the cp conserving thdm in order to be sure if a ( say ) detected boson in the higgsstrahlung channel with a low rate could be consistent with @xmath0 , or is actually evidence for a @xmath22 from a cp violating thdm with a small component of cp even scalar field .
such scalar pseudoscalar mixing may also arise in the mssm , being generated radiatively if one allows cp violating phases in the model @xcite . the paper is organized as follows . in section 2 ,
we review all the cp odd higgs boson interactions ( with gauge bosons and fermions ) , section 3 contains the notation , conventions and one loop calculations while in section 4 we present our numerical results . finally section 5 contains our conclusions .
we will use the following notation and conventions .
the momenta of the incoming electron and positron , outgoing gauge boson @xmath25 and outgoing cp odd higgs boson @xmath0 are denoted by @xmath26 , @xmath27 , @xmath28 and @xmath29 , respectively . neglecting the electron mass @xmath30 ,
the momenta in the center of mass of the @xmath14 system are given by : @xmath31 where @xmath32 denotes the center of mass energy , @xmath33 the scattering angle between @xmath34 and @xmath0 and @xmath35 @xmath36 and @xmath37 are the masses of the neutral gauge boson and of the cp
odd higgs boson .
+ the mandelstam variables are defined as follows : @xmath38 at the one loop order ( fig.1.ad ) , the differential cross
section reads : @xmath39 the one loop amplitude @xmath40 can be written as a sum of vertex and box contributions as follows : @xmath41 where @xmath42 ( @xmath43 ) denotes the vertex correction s channel photon exchange ( s channel z exchange ) , depicted in fig.1.a + fig.1.b , and @xmath44 denote the box contribution with @xmath8 and @xmath7 exchange ( fig.1.c ) .
all these contributions fully project onto six invariants as follows : @xmath45 where the invariants @xmath46 are given by : @xmath47 with @xmath48 the polarization of the vector boson @xmath25 . summing over the @xmath49 gauge boson polarizations
, the squared amplitude may be written as : + * case v = z * @xmath50 + ( m_z^2 - t ) ^2 [ \vert { { \cal m}_3 \vert } ^2 + \vert { { \cal m}_4 } \vert ^2 ] \nonumber \\ [ 0.2 cm ] & & + 4 ( m_z^2 - u ) re [ { { \cal m}_1 } { { \cal m}^*_5 } + { { \cal m}_2 } { { \cal m}^*_6 } ] + ( m_z^2 - u)^2 [ \vert { { \cal m}_5 \vert } ^2 + \vert { { \cal m}_6 \vert } ^2 ] \nonumber \\ [ 0.2 cm ] & & - 2 ( m_{a}^2 m_z^2 + m_z^2 s - t u ) re [ { { \cal m}_3 } { { \cal m}^*_5 } + { { \cal m}_4 } { { \cal m}^*_6 } ] \}\end{aligned}\ ] ] * case v=@xmath51 * @xmath52 ) + \nonumber \\ & & s u ( re [ { \cal m}_1^ { * } { \cal m}_5 + { \cal m}_2^ { * } { \cal m}_6 ] ) -s t u ( re [ { \cal m}_3^ { * } { \cal m}_5 + { \cal m}_4^ { * } { \cal m}_6 ] ) \end{aligned}\ ] ] in the two higgs doublet extensions of the standard model , there are different ways of coupling the higgs fields to matter .
the most popular are labelled as model type i and model type ii . in the former ,
the quarks and leptons couple only to the second higgs doublet @xmath53 , while in the latter , in order to avoid the problem of flavor changing neutral current ( fcnc ) @xcite , one assumes that @xmath54 couples only to down quarks ( and charged leptons ) and @xmath53 couples only to up quarks ( and neutral leptons ) .
the type ii model is the pattern found in the mssm . in general , the cp
odd higgs interaction with fermions is given by : @xmath55 here @xmath56(@xmath57 ) may refer to any generation of up quark ( down quark or charged leptons ) and the @xmath58 couplings are defined as follows : @xmath59 it is worth noting the models i and ii are not very different for the top bottom loop corrections at low @xmath60 because the term @xmath61 will dominate and it is common to both types . in the case of large @xmath60
the effects of down quarks and tau lepton are enhanced ( suppressed ) in model type ii ( model type i ) . in a general two higgs doublet model ,
the interactions between higgs bosons and gauge bosons are completely dictated by the local gauge invariances and so they are model independent regardless of whether the model is supersymmetric or not .
these interactions originate from the square of the covariant derivative in the higgs lagrangian which is given by : @xmath62^+(\partial_\mu + ig \vec{t_a } \vec{w_{\mu}^a } + ig'\frac{y_{\phi_i}}{2}b_\mu ) \phi_i \label{covder}\end{aligned}\ ] ] where : @xmath63 is the isospin operator , @xmath64 the hypercharge of the higgs fields , @xmath65 the @xmath66 gauge fields , @xmath67 the @xmath68 gauge field , and @xmath69 and @xmath70 are the associated coupling constants . + after expanding eq .
( [ covder ] ) and deriving all the interaction vertices , one finds , in accordance with the conservation of electromagnetic current , that the @xmath51@xmath0@xmath71 vertex vanishes at tree
level while the vertex @xmath71@xmath71@xmath0 vanishes by virtue of cp invariance @xcite .
+ therefore , at tree level , the contributions to @xmath72 and/or @xmath73 only come from the @xmath74channel via electron exchange and from the following @xmath75channels : @xmath76 ( @xmath77 ) where @xmath7 and @xmath8 are the two cp even neutral higgs fields .
all these contributions are proportional to the electron mass @xmath30 which we will neglect in our work . as a major consequence , in this limit
, the studied process will be only due to loop effects .
+ + from eq.([covder ] ) one can deduce the following vertices which are needed for our study : @xmath78 where @xmath79 and @xmath80 are the cosine and sine of the weinberg angle @xmath81 .
we need also : @xmath82 where @xmath83 may refer to @xmath56 or @xmath57 , and @xmath84 in a thdm , cp violation ( explicit or spontaneous ) is possible in the higgs sector if one allows the discrete symmetry to be broken softly by a term of dimension 2 ( @xmath85 ) . in such a case mixing
is permitted between the pure cp even states @xmath8 , @xmath7 and the pure cp
odd state @xmath0 , leading to three neutral higgs mass eigenstates which can not be assigned a definite cp quantum number .
in the notation of ref.@xcite these are referred to as @xmath86 , @xmath87 , @xmath88 , whose couplings obey various sum rules @xcite,@xcite which in the cp conserving case reduce to the familiar sum rules of the thdm / mssm e.g. one has ( where the couplings are normalized to sm higgs boson strength ) @xmath89 here @xmath90 run from 1 to 3 , @xmath91 is the coupling @xmath24 , and @xmath92 is the coupling @xmath93 .
@xcite explained how cp violation may be probed in the higgs sector by measuring non zero values for each of @xmath94 , @xmath95 and @xmath96 , or for each of @xmath93 , @xmath24,@xmath97 ( where @xmath98 ) .
a third way would be a positive signal in all three higgsstrahlung channels @xmath99 , and it is possible that the higgs masses are arranged such that the latter production mechanisms would be all open kinematically ( i.e. @xmath100 ) while pair production of higgs bosons are not .
therefore the cp violating thdm may provide measurable signals in the @xmath101 channel for all three neutral higgs bosons . hence in order to distinguish between a cp violating thdm and a cp conserving thdm in the above channels it is important to know the attainable value of @xmath21 in the cp conserving case . in refs .
@xcite,@xcite , using effective lagrangians , it was shown that a cp
odd higgs scalar with effective point like couplings @xmath102 and @xmath103 would give rise to angular distributions in the channel @xmath19 which would differ significantly from those in @xmath104 .
our analysis essentially determines the possible magnitude of these effective couplings @xmath102 and @xmath103 in the context of the thdm . in the notation of @xcite this would correspond to evaluating the arbitrary couplings @xmath105 . given the expected smallness of the rate for @xmath106 , any angular distribution analysis for this channel would most likely be hampered by small statistics .
if in the cp violating thdm all three @xmath24 couplings have a reasonable strength ( i.e. significant pseudoscalar scalar mixing ) , a detectable signal would be possible which could not be mimicked by the cp conserving thdm .
in addition there would be sufficient events to show that the angular distribution proceeded via the cp even scalar component .
the problematic case of interest to us is when a @xmath22 is dominantly pseudoscalar and thus has a smaller ( but still detectable ) rate in the @xmath107 channel . in this case
the signal from @xmath108 is background to any possible interpretation of a cp violating signal in this channel .
it is also this scenario when angular distributions would be affected by low statistics .
we have evaluated the pure radiative effects of the rare decay @xmath109 ( @xmath110 ) at the one loop level in the t hooft - feynman gauge .
the sum of all the one loop effects are ultra
violet ( uv ) convergent but since some feynman diagrams are uv divergent we will use the dimensional regularization scheme @xcite to deal with them .
the typical feynman diagrams for the virtual corrections of order @xmath111 are drawn in figure 1 . in the thdm ,
the @xmath112 and the @xmath113 vertices receive corrections only from fermion exchanges ( fig.1-a - b ) .
there are also box diagrams and their crossed analogies ( fig.1-c ) .
note that there is no mixing coming from the @xmath114 ( @xmath115 ) s channel self energy because the vertices @xmath116 ( @xmath117 ) and @xmath118 ( @xmath119 ) vanish , while the mixing @xmath114 has to be considered in the t channel ( fig.1.d ) .
owing to lorentz invariance the @xmath114 self energy is proportional to @xmath120 ; then , since the vector boson v=@xmath51 or z is on shell , the t channel amplitude will be proportional to @xmath30 and consequently vanishes .
note also that the tadpole diagrams have vanishing contributions . in the on shell scheme defined in @xcite
, it is found that there is no counter
term for the z
z@xmath0 , z@xmath51@xmath0 and @xmath51@xmath51@xmath0 vertices . consequently the one loop vertices z z@xmath0 , z@xmath51@xmath0 and @xmath51@xmath51@xmath0 have to be separately uv finite and this feature will provide us with a good check of our calculations .
all the feynman diagrams are generated and computed using feynarts @xcite and feyncalc @xcite packages supplied with some mathematica `` know - how '' .
we also use the fortran ff
package @xcite in the numerical analysis .
+ + * vertex correction : fig.1.a + fig.1.b : @xmath121 * + + in terms of the @xmath122 , @xmath123 and @xmath124 passarino veltman functions , @xmath42 and @xmath43 are given by has been evaluated in @xcite . ] : @xmath125\nonumber \\ & + & ( [ m_a^2 m_z^2 - m_z^4 + m_a^2 s + 2 m_z^2 s - s^2 ] ( { g_q^l}^2 + { g_{q}^r}^2 ) \nonumber \\ & + & 2 [ - m_a^4 + m_a^2 m_z^2 + m_a^2 s ] g_q^l g_q^r ) c_0(m_a^2 , m_z^2 , s , m_q^2 , m_q^2 , m_q^2))\end{aligned}\ ] ] here @xmath126 and @xmath127 denote the charge and the mass of the specific quark ; @xmath128 , @xmath129 , @xmath130 and @xmath130 are the couplings defined before . + as defined in the appendix a , the @xmath122 functions are uv divergent whereas the @xmath123 are uv convergent .
consequently it is easy to check that the one loop vertices defined below are uv convergent , as they should be .
+ + * vertex correction : fig.1.a + fig.1.b : @xmath131 * + + in a similar way we found : @xmath132 @xmath133 + * box corrections : fig.1.c * : + + box corrections are only present for @xmath106 , because the photon does not couple to neutral particles . @xmath134\end{aligned}\ ] ] all the @xmath135 have the same arguments : @xmath136 .
the amplitude of a crossed box can be obtained from the direct one by changing : @xmath137 , @xmath138 , @xmath139 and a global sign
. moreover @xmath140 can be obtained from @xmath141 just by changing @xmath142 in @xmath143 and a global sign .
( 100,100)(0,0 ) ( 4,0 ) ( 41,0 ) ( 88,0)(1,1)28 ( 88,0)(1,-1)28 ( 117,-27)(0,1)51 ( 116,25 ) ( 116,-25 ) ( 88,0 ) ( 92,16 ) ( 89,-19 ) ( 119,0 ) ( 116,25 ) ( 129,29 ) @xmath0 ( 115,-25 ) ( 99,11)(-1,-1)1 ( 117,0)(0,1)1 ( 101,-13)(1,-1)1 ( 55,-45 ) ( 246,0 ) ( 295,0)(1,1)28 ( 295,0)(1,-1)28 ( 324,-27)(0,1)51 ( 323,25 ) ( 323,-25 ) ( 295,0 ) ( 207,0 ) ( 299,16 ) ( 296,-19 ) ( 326,0 ) ( 323,25 ) ( 339,28 ) @xmath0 ( 324,-25 ) ( 306.2,11)(1,1)1 ( 324,0)(0,-1)1 ( 308,-13)(-1,1)1 ( 261,-45 ) @xmath133 ( 100,100)(0,0 ) ( -60,50 ) ( -20,50 ) ( -20,48 ) ( 29,48 ) ( -22,1 ) ( -26,22 ) @xmath144 ( -20,1 ) ( -19,0 ) ( -60,0 ) ( 29,2 ) ( 27,48 ) ( 55,53 ) @xmath0 ( 59,7 ) @xmath71 ( 25,0 ) ( 27,1 ) ( 87,20 ) ( 20,-24 ) ( 100,100)(0,0 ) ( -60,50 ) ( -20,50 ) ( 29,48 ) ( -20,50 ) ( 13,50 ) ( -22,1 ) ( -26,22 ) @xmath144 ( -20,1 ) ( -60,0 ) ( 42,48 ) ( 46,48 ) ( 55,53 ) @xmath0 ( 59,7 ) @xmath145 ( -19,1 ) ( 87,20 ) ( 20,-24 ) ( 20,-54 ) @xmath133 + +
in this section we focus on the numerical analysis .
we take the following experimental input for the physical parameters @xcite : * the fine structure constant : @xmath146 . * the gauge boson masses : @xmath147 and @xmath148 . in the on shell scheme we consider ,
@xmath149 is given by @xmath150 , which is not modified by loop corrections .
* the top bottom quark masses are taken to be : @xmath151 gev and @xmath152 gev .
the tau lepton mass is taken to be @xmath153 gev as pointed out in the introduction , in the general thdm the cp
odd higgs mass is not constrained by experiment , and only the sum @xmath154 is constrained to be heavier than 90 gev . therefore in our analysis
we are free to consider a very light cp odd higgs mass .
taking into account perturbative requirements on the yukawa couplings @xmath155 is constrained to be in the range : @xmath156 .
let us first discuss briefly the thdm radiative contributions to the cp
odd @xmath0 associated production with a photon which has been considered before in @xcite .
we stress here that we are in perfect agreement with @xcite both on analytical result and numerical result .
we would like to highlight in fig.2.a the enhancement of the cross
section for small @xmath60 .
one can reach a cross section of about 0.8 fb for @xmath157 ( 8 fb for @xmath158 ) .
fig.2.b shows the total cross section against @xmath60 for cp
odd @xmath0 associated production with a z gauge boson .
one can see also the enhancement of the cross section in the small @xmath60 regime .
this enhancement for small @xmath155 is common to both model i and model ii .
the most important contribution in the fermionic case comes from top quark loops . in the large @xmath60 regime in model
ii the corrections start to be sensitive to the bottom , light quarks and tau lepton loops . as mentioned in section 3 , in the case of
associated cp
odd higgs production with a z gauge boson , there are extra box contributions to the cross section . in all figures 2a , 2b ,
3a and 3b we have chosen @xmath159 gev , @xmath160 gev and @xmath161 .
it is found that box contributions to the cross section are rather small . for large @xmath60
the fermionic contributions also become small and of the same order of magnitude as the box contributions . in fig.3.a
we show the dependence of the cross
section on the cp odd higgs mass for @xmath162 gev in the case where @xmath163 and @xmath164 .
one can see that the cross
section is maximized for both low @xmath60 and small @xmath165 mass .
the kink in the figure corresponds to @xmath166 i.e. when the decay @xmath167 opens . at @xmath168
gev the cross
section for the sm higgs boson in the higgsstrahlung channel falls from @xmath169 fb to @xmath170 fb as the higgs mass increases from 200 gev to 300 gev .
fig 3a shows that for @xmath171 one finds @xmath172 falling from 0.2 fb to 0.1 fb in the same mass region .
if one wished to interpret a third ( by definition the smallest ) higgsstrahlung signal as evidence for a cp violating thdm , a cross section comfortably in excess of 1 fb would be required .
we recall from eq .
2.10 that @xmath173 ( normalized to sm strength ) , and so @xmath172 constitutes a non
negligible background . in fig.3.b
we show the total cross section as a function of the center of mass energy @xmath32 for @xmath174 gev and for @xmath175 .
one can find a total cross section of about 0.04 fb only for the cp
odd higgs mass in the intermediate range ( @xmath165=50150 gev ) and for @xmath176 gev . at high center of mass energy
the rate of production is rather small .
to conclude , we have computed the associated production of a cp odd higgs boson with a gauge boson @xmath177 in high energy @xmath14 collisions in the framework of the general two higgs doublet model .
the calculation is performed within the dimensional regularisation scheme .
we have shown that in the small @xmath2 regime the top quark loop contribution is enhanced leading to significant cross sections ( about a few fb ) , while in the large @xmath2 regime the cross section does not attain observable rates . the smallness of the cross section means that there is no chance of seeing such associated production at lep ii , while such production could be relevant for a next linear collider machine and would require a very high luminosity option .
* acknowledgment : * a. akeroyd was supported by the japan society for promotion of science ( jsps ) .
a. arhrib acknowledges the abdus salam international centre for theoretical physics for the kind hospitality during his visit where part of this work has been done .
higgs , phys .
b12 * ( 1964 ) 132 , phys .
* 13 * ( 1964 ) 508
gunion , h.e .
haber , g.l .
kane and s. dawson , _ the higgs hunter s guide _ ( addison wesley , reading , 1990 ) .
h. komatsu and j. kubo , phys .
* b157 * ( 1985 ) 90 ; nucl . phys .
* b263 * ( 1986 ) 265 .
cern preprint cern
ep/9960 , combined experimental limits .
opal collaboration , o. abbiendi _
et al _ , eur .
* c7 * ( 1999 ) 407 .
chankowski , m. krawczyk and j. zochowski , hep - ph/9905436 .
j. ellis , m. k. gaillard and d.v .
nanopoulos , nucl .
b106 * ( 1976 ) 292 ; b. l. ioffe and v. a. khoze , sov .
( 1978 ) 50 ; b. w. lee , c. quigg and h.b .
thacker , phys . rev .
* d16 * ( 1977 ) 1519 ; j. d. bjorken , proc .
summer institute on particle physics , slac report 198 ( 1976 ) .
j. f gunion , l. roszkowski , a. turski , h. e. haber , g. gamberini , b. kayser , s.f .
novaes , f. olness and j. wudka , phys . rev .
* d38 * ( 1988 ) 3444 ; proceeding of the workshop `` @xmath14 collisions at 500 gev : the physics potential '' , edited by zerwas .
a. djouadi , v. driesen , w. hollik and a. kraft , eur .
j. * c1 * ( 1998 ) 163 ; m. krawczyk , j. zochowski and p. mattig , eur .
j. * c8 * ( 1999 ) 495 .
a. djouadi , v. driesen , w. hollik and j. rosiek , nucl . phys .
* b491 * ( 1997 ) 68 .
shou hua zhu , hep
a. arhrib , m. capdequi peyranre , w. hollik and g. moultaka , in preparation .
d. a. dicus , j. l. hewett , c. kao and t. g. rizzo , phys . rev .
* d40 * ( 1989 ) 787 ; a. a. barrientos bendez and b. a. kniehl , phys . rev . *
d59 * 015009 ( 1999 ) ; s. moretti and k. odagiri , phys . rev .
* d59 * 055008 ( 1999 ) .
physics report * 299 * ( 1998 ) 1 .
a. djouadi , j , kalinowski and p. zerwas , z. phys .
* c54 * ( 1992 ) 255 .
s. dawson and l. reina , phys . rev .
* d60 * ( 1999 ) 015003 .
a. djouadi , v. driesen and c. junger , phys . rev .
* d54 * ( 1996 ) 759 .
shou - hua zhu , j. phys .
* g24 * ( 1998 ) 1703 . c. kao , phys .
* d46 * ( 1992 ) 4907 .
a. mendez and a. pomeral , phys .
b272 * ( 1991 ) 313 .
gunion , h.e . haber and c. kao , phys
* d46 * ( 1992 ) 2907 .
a. pilaftsis , phys . lett .
* b435 * ( 1998 ) 88 ; phys .
* d58 * ( 1998 ) 096010 ; a. pilaftsis and c.e.m .
wagner , hep
ph/9902371 ; d.a .
demir , hep ph/9901389 .
s.glashow and s.weinberg , phys.rev . * d15 * ( 1977 ) 1958 . a.a .
iogansen , n.g .
uraltsev and v.a .
khoze , sov .
* 36 * ( 1983 ) 717 .
g.keller and d. wyler , nucl .
b274 * ( 1986 ) 410 .
b. grzadkowski , j. gunion and j. kalinowski , hep - ph/9902308 .
b. grzadkowski , j. gunion , h.e . haber and j. kalinowski , phys .
* 79 * ( 1997 ) 982 . v. barger ,
k , cheung , a. djouadi , b.a .
kniehl and p.w .
zerwas , phys . rev .
* d49 * ( 1994 ) 79 .
k. hagiwara and m.l .
stong , z. phys . * c62 * ( 1994 ) 99 .
g. t hooft and m. veltman , nucl .
phys . * b44 * ( 1972 ) 189 ; + p. breitenlohner and d. maison , commun .
* 52 * ( 1977 ) 11 .
a. dabelstein and w. hollik mpi - ph-93 - 86 , contributed to workshop on e+ e- collisions at 500 gev : the physics potential . published in susy 94 workshop:109 - 118 a. dabelstein z.phys .
* c67*,(1995 ) , 495 and p. chankowski , s. pokorski and j. rosiek nucl . phys .
* b423 * ( 1994 ) 437 .
opal collaboration .
phys.lett . *
b370 * ( 1996 ) 174 and r. m. barnett et al .
rev . * d54 * ( 1996 ) 1 .
h.eck and j. kublbeck , guide to feynarts 1.0 , university of wurzburg , 1992 .
r.mertig , guide to feyncalc 1.0 , university of wurzburg , 1992 .
van oldenborgh , comput.phys.commun .
* 66 * ( 1991 ) 1 . c. caso
et al , eur .
j. * c3 * ( 1998 ) 1 .
g. passarino and m. veltman , nucl . phys .
* b160 * ( 1979 ) 151 ; + g. thooft and m. veltman , nucl .
b153 * ( 1979 ) 365 .
let us briefly recall the definitions of scalar and tensor integrals we use .
the inverses of the propagators are denoted by @xmath178 where the @xmath179 are the momenta of the external particles .
+ + * two point function : * @xmath180 @xmath181 is an arbitrary renormalization scale and @xmath182 is the space time dimension .
+ + * three point function : * @xmath183 where @xmath184 . +
* four point functions : * @xmath185 using lorentz invariance , one can write the vectorial function @xmath186 in terms of new scalar functions @xmath124 as follows : @xmath187 moreover one can always get all the `` new '' functions @xmath124 in terms of the fundamental ones , @xmath188 , @xmath122 , @xmath123 and @xmath189 . the analytical expressions of all the scalar functions can be found in ref .
@xcite 1 .
one loop diagram contributions to @xmath190 ( @xmath191 or @xmath71 ) .
fig.1.a + fig.1.b are fermionic vertex corrections .
fig.1.c are box diagrams and fig.1.d is the t channel self energy .
2 . * fig.2.a * : total cross section for @xmath192 ( vertex only ) as a function of @xmath60 for @xmath162 gev and several values of @xmath165 .
+ + * fig.2.b * : total cross section for @xmath193 ( vertex + boxes ) as a function of @xmath60 for @xmath162 gev and several values of @xmath165 .
fig.3.a * : total cross section for @xmath194 ( vertex + boxes )
asa function of @xmath165 for @xmath162 gev and several values of @xmath60
. + + * fig.3.b*:total cross section for @xmath194 ( vertex + boxes ) as a function of @xmath32 for @xmath175 and for several values of @xmath165 . | we study the associated production of a cp
odd higgs boson @xmath0 with a neutral gauge boson ( z or photon ) in high energy @xmath1 collisions at the one loop level in the framework of two higgs doublet models ( thdm ) .
we find that in the small @xmath2 regime the top quark loop contribution is enhanced leading to significant cross sections ( about a few fb ) , while in the large @xmath2 regime the cross section does not attain observable rates .
0.26 cm 0.26 cm 16.cm # 1 ( 100,200)(0,0 ) ( 0,0 ) ( 5,0)(8,-8)6(8,8)[bl ] ( 5,-8)(8,-8)6(8,8)[tr ] ( 49,-32)(0,0)#1 # 1#2 ( 100,200)(0,0 ) ( 25,0 ) ( 0,23)(8,-8)3(8,8)[tr ] ( 8,23)(8,-8)3(8,8)[bl ] ( 0,-23)(8,8)3(8,8)[br ] ( 8,-23)(8,8)3(8,8)[tl ] ( -10,26)(0,0)#1 ( -10,-26)(0,0)#2 # 1#2 ( 100,200)(0,0 ) ( 0,0 ) ( 5,0)(8,8)3(8,8)[tl ] ( 5,8)(8,8)3(8,8)[br ] ( 5,0)(8,-8)3(8,8)[bl ] ( 5,-8)(8,-8)3(8,8)[tr ] ( 34,30)(0,0)#1 ( 34,-30)(0,0)#2 # 1#2 ( 100,200)(0,0 ) ( 34,0 ) ( 0,35)(1,-1)35 ( -10,26)(0,0)#1 ( -10,-26)(0,0)#2 ( 0,-35)(1,1)35 # 1#2 ( 100,200)(0,0 ) ( 0,0 ) ( 0,0)(1,1)25 ( 34,26)(0,0)#1 ( 34,-26)(0,0)#2 ( 0,0)(1,-1)25 # 1#2 ( 100,200)(0,0 ) ( 0,0)(16,0)3(8,8)[t ] ( 8,0)(16,0)3(8,8)[b ] ( # 2,15)(0,0)#1 # 1#2 ( 100,200)(0,0 ) ( 0,0)(16,0)2(8,8)[t ] ( 8,0)(16,0)2(8,8)[b ] ( # 2,15)(0,0)#1 # 1#2 ( 100,200)(0,0 ) ( 0,0)(16,0)5(8,8)[t ] ( 8,0)(16,0)5(8,8)[b ] ( # 2,15)(0,0)#1 # 1#2 ( 100,200)(0,0 ) ( 0,0)(1,0)48 ( # 2,15)(0,0)#1 # 1#2 ( 100,200)(0,0 ) ( 0,0)(1,0)35 ( # 2,15)(0,0)#1 # 1#2 ( 100,200)(0,0 ) ( 0,0)(1,0)48 ( # 2,15)(0,0)#1 # 1 ( 100,200)(0,0 ) ( 0,0)(0,1)48 ( 10,24)(0,0)#1 # 1 ( 100,200)(0,0 ) ( 0,0)(0,1)48 ( 10,24)(0,0)#1 # 1 ( 100,200)(0,0 ) ( 0,0)(0,16)3(8,8)[tl ] ( 0,8)(0,16)3(8,8)[br ] ( 0,8)(0,16)3(8,8)[tr ] ( 0,0)(0,16)3(8,8)[bl ] ( 15,24)(0,0)#1 # 1#2 ( 100,200)(0,0 ) ( 0,0)(10.8,0)6(1,0)5.2 ( # 2,15)(0,0)#1 # 1#2 ( 100,200)(0,0 ) ( 0,0)(10.8,0)4(1,0)5.2 ( # 2,15)(0,0)#1 # 1 ( 100,200)(0,0 ) ( 0,0)(0,10.8)5(0,1)5.2 ( 10,24)(0,0)#1 # 1#2 ( 100,200)(0,0 ) ( 0,0 ) ( 0,0)(16,16)3(1,1)10 ( 0,0)(16,-16)3(1,-1)10 ( 27,36)(0,0)#1 ( 52,-36)(0,0)#2 # 1#2 ( 100,200)(0,0 ) ( 40,0 ) ( 0,-40)(16,16)3(1,1)5 ( 0,40)(16,-16)3(1,-1)5 ( -7,36)(0,0)#1 ( -7,-36)(0,0)#2 # 1 ( 100,200)(0,0 ) ( 0,0 ) ( 0,0)(16,16)3(1,1)10 ( 45,36)(0,0)#1 kek preprint 99/53 + lpm/99/31 + ufr - hep/99/17 + july 1999 + a.g .
akeroyd@xmath3 , a. arhrib@xmath4 , m. capdequi peyranre@xmath5 |
the planetary nebula ( pn ) swst 1 ( swings & struve 1940 ) is believed to be in the incipience of this stage of evolution , because of its small angular diameter ( kwok et al 1981 ; aaquist & kwok 1990 ) and high electron density , found by various studies to have a value in the range @xmath9 @xmath11 3@xmath1210@xmath13 to 10@xmath14 @xmath1 ( flower et al .
1984 ; de freitas pacheco & veliz 1987 , hereafter dfp87 ; de marco et al .
2001 , hereafter dm01 ) .
the low excitation of its nebular emission line spectrum and cool temperature of the central star , @xmath15 = 40,000 k ( dm01 ) are also suggestive of youth .
the central star of swst 1 , hd 167362 , has been classified as spectral type [ wc9 ] ( crowther et al .
1998 ) , a class of h - deficient , c - rich pn central stars that exhibit emission lines indicative of strong winds , although its emission lines are weaker than typical members of this class ( tylenda et al . 1993 ; grny & tylenda 2000 ) .
while the evolutionary origin of this class of stars is still unclear , it may be related to late thermal pulses ( e.g. herwig et al .
1999 ; blcker 2001 ) or binary interactions ( de marco & soker 2002 ) . in this paper
, we present measurements of nebular absorption lines against the continuum of the central star of swst 1 .
the _ far ultraviolet spectroscopic explorer _
( _ fuse _ ) spectrum probes material along a narrow `` pencil - beam '' toward the central star , and therefore samples a very small portion of the nebula compared to emission line studies .
the _ fuse _ waveband includes the resonance lyman and werner bands of molecular hydrogen , several lines of and ( which we use to derive the fe abundance in swst 1 ) , and @xmath161088.46 , which has been used previously to derive the ge abundance of five pne ( sterling et al .
2002 , hereafter sdb ; sterling & dinerstein 2003 ) .
this spectral region also contains several other lines of species in the and neutral atomic regions .
the structure of this paper is as follows : in 2 , we describe the observations and data reduction .
we discuss the method of analysis in 3 , and present the voigt component fits used to derive the column densities , central velocities , and doppler spread parameters of several nebular species . in 4 , we use these results to investigate properties of the ionized gas , such as the electron density and gas phase iron abundance . we also examine neutral and molecular gas in swst 1 ( 5 ) , deriving an upper limit to the column density of nebular molecular hydrogen in the line of sight , and investigating the excitation of the fine structure levels .
the implications of our study are discussed in 6 , and the results are summarized in 7 .
the _ fuse _ satellite provides access to the wavelength range 9051187 at a nominal resolution @xmath17 ( fwhm @xmath18 km s@xmath19 ) .
the design ( moos et al . 2000 ; sahnow et al . 2000 ) consists of four prime focus telescopes and rowland - circle spectrographs with microchannel plate detectors .
two of the channels are coated with al : lif , which provides optimal throughput for the spectral range 10001187 , while the other two have sic coatings for optimal reflectivity below 1000 .
each detector consists of two segments , so that eight channel segments provide coverage of the entire spectral range .
spectral redundancy ensures that nearly all wavelengths are covered by multiple segments .
swst 1 was observed in 2001 august with the large ( lwrs , @xmath20 ) aperture in time - tag mode for a total of 39,587 seconds .
these data are available at the multi - mission archive at the space telescope science institute ( mast ) under the data set identifications b0690101 and b0690102 .
we used calfuse ( v2.0.5 ) , the standard reduction pipeline available at johns hopkins university , to calibrate the raw data .
the exposures were then cross - correlated and co - added in order to increase the signal - to - noise ratio ( s / n ) .
we rebinned the data by three pixels , which improved the average s / n to @xmath1115 - 25 per rebinned sample , depending on the region of the spectrum .
this increased the effective pixel size from @xmath11 7 m to @xmath11 20 m , corresponding to a velocity sampling interval of @xmath11 6.0 km s@xmath19 near 1000 and @xmath11 5.2 km s@xmath19 near 1150 .
figure 1 shows an excerpt from the _ fuse _ lif 2a spectrum of swst 1 from 11201150 .
we checked for terrestrial airglow contamination of the spectrum ( which can be important for lines such as ) by comparing the exposures taken during orbital night to the full dataset .
the night spectrum is essentially indistinguishable from that including daytime data , and we therefore conclude that airglow emission has a negligible effect on our results .
we also employ archival _ hubble space telescope _ ( _ hst _ ) ultraviolet observations of swst 1 , acquired by the space telescope imaging spectrograph ( stis ; woodgate et al .
1998 ; kimble et al . 1998 ) .
this data was obtained in 2000 september , as part of guest observer program go8304 ( principal investigator s. kwok ) .
swst 1 was observed for a total of 2300 seconds with the @xmath21 aperture and e140 m echelle grating , which provides a spectral resolution of @xmath22 ( @xmath23 km s@xmath19 ) in the spectral region 11501730 .
the data was reduced with the calstis v2.11 processing pipeline ( lindler 1999 ) .
unfortunately , the e140 m spectrum has a low s / n , particularly below 1200 ( s / n @xmath24 ) , and therefore we make limited use of this data .
we analyze the _ fuse _ data in the segments with the best s / n ( i.e. largest effective area ) for a given wavelength region : lif 1a for 10001083 , and lif 2a for @xmath25 .
redundant coverage in the other spectral segments is used to confirm the lif 1a/2a results .
we normalize each absorption profile by fitting a low - order legendre polynomial to the continuum and dividing the spectrum by this function .
atomic wavelengths and data are taken from morton ( 2000 ; 2003 ) , while h@xmath5 wavelengths and oscillator strengths are from the compilation of lines by mccandliss ( 2003 ; see also http://www.pha.jhu.edu/`~`stephan/h2ools/h1h2data/ , based on abgrall et al .
1993a , b ) .
the line of sight toward the central star of swst 1 passes through two interstellar clouds , as well as the absorbing nebular material . in order to separate the different absorption components of lines ,
we perform voigt component fits to each line profile .
these fits yield detailed information about the nebular features , such as central velocity @xmath26 , doppler spread parameter @xmath27 , and column density @xmath6 .
the model profiles are convolved with an instrumental spread function , a gaussian of fwhm = 20 km s@xmath19 for _ fuse _ , and 6.5 km s@xmath19 for the stis e140 m data , to account for the systematic broadening effect from the instrumental resolution . in many cases , more than one line from a given species
can be measured .
when possible , we perform simultaneous fits to absorption lines originating in the same level , in order to better constrain the fits and reduce systematic errors . in a simultaneous fit
, the parameters @xmath26 , @xmath27 , and @xmath6 are forced to be the same for all lines from a given level ( although the central velocities may differ by a small offset , particularly for lines measured by different instruments ) .
simultaneous fitting is especially important for the medium - resolution _ fuse _ spectrum , where saturation may not be evident due to instrumental broadening .
we use this technique in several cases to test the importance of saturation effects , and to constrain the parameters of mildly saturated lines which , if modeled individually , would otherwise be highly uncertain .
this procedure is most useful when the fitted lines have significantly different @xmath28-values , and at least one is not saturated .
we use a different line fitting method in this paper than in previous work for the same dataset ( sdb ) .
while we measure several more nebular lines in swst 1 than sdb reported , our two studies have lines of , and in common . by comparing our table 1 with table 1 of sdb
, one may note that the column densities we measure here are systematically higher than those of sdb , by about 0.3 dex ( a factor of two ) .
this is not due to differences in the atomic data used ; indeed , the @xmath28-values for @xmath29 are _ larger _ in morton ( 2003 ) than morton ( 1991 ) .
the difference is likely due to the fact that we account for saturation effects in the present study .
we consider the results presented in this paper to be more robust than those of sdb , since we utilize simultaneous fitting , and more recent atomic data ( morton 2003 ) and solar abundances ( asplund et al . 2004 ) . given the low excitation of the swst 1 emission line spectrum ( dm01 ; dfp87 ) , it is not surprising that the only nebular species we detect in the _ fuse _ spectrum are neutrals and low ions : , , , , , and . of these ions
, s@xmath30 requires the highest energy to create ( 23.3 ev ) , which is quite modest by pn standards .
there is no evidence of absorption from more highly ionized species . in table 1
, we show the results of the line fits for all detected nebular species .
two lines are listed in column 1 when a simultaneous fit was performed , while one line is recorded otherwise .
the quoted 1-@xmath31 error estimates include continuum placement and statistical fit uncertainties .
we estimated the statistical errors with monte carlo simulations of the absorption line profiles 100 artificial spectra were generated for each profile , with the data points modified within the errors of the gaussian noise . in most cases , the systematic errors from continuum placement dominate the uncertainties . for lines arising from different levels of the same ion
, we further constrain their fits by forcing the lines to have the same @xmath26 and @xmath27 , although the column densities of the levels are allowed to differ .
this is a useful constraint for ions which exhibit absorption lines from several excited levels ( , , and ) , some of which may be saturated . since the excitation energies of the ground term levels for these three ions ( @xmath32 ev ) are small compared to the gas temperature ( 0.9 ev for @xmath33 k ; dm01 )
, the same value of @xmath27 should apply for lines from all the observed levels of these ions .
the results of table 1 are presented graphically in figures 2a and 2b , where the continuum - normalized profiles are displayed versus heliocentric velocity ( all velocities reported in this paper are in the heliocentric frame ) .
note that the line of sight toward swst 1 intersects two interstellar components , one near @xmath34 km s@xmath19 , and the other at + 20 km s@xmath19 ( these are seen most clearly in the profile fits to and ) .
these absorbing clouds are well - separated in velocity from nebular absorption . the profiles of @xmath351250.58 , 1253.81 and @xmath351304.86 , 1306.83 are from stis data , while all others are from the _ fuse _ spectrum .
the total model fit to each line profile ( convolved with the instrumental spread function ) is shown as a thick , solid line ; thin , dashed lines represent individual component fits _ before _
instrumental broadening .
the depth of the dashed lines therefore indicate the importance of saturation for each component . in figures 2a and 2b ,
the panels are grouped by species
double panels for simultaneous fits , and single ones for single line fits .
the absorbing ion is indicated above the panel is from the second excited level above the ground state in . ] , and the wavelength of the line is shown within the panel .
nearby absorption from other species is indicated in the figures .
the central velocities we measure agree well with the findings of dinerstein et al .
( 1995b ) , who used high resolution observations of the @xmath355889 , 5895 doublet to find a heliocentric velocity of @xmath36 km s@xmath19 for the circumstellar gas .
this agreement is somewhat surprising , since ionized species will not necessarily absorb at the same velocity as neutral sodium , which has an ionization potential of only 5.1 ev .
note that the species with the most aberrant nebular velocities , and , are blended with a heavily saturated interstellar component , and @xmath26 is highly uncertain .
sulfur absorption lines are useful for calibrating the abundances of other elements , as well as for determining the average electron density @xmath9 along the line of sight .
while the excited levels of the @xmath37 ground term are scantly occupied in the diffuse interstellar medium ( ism ) , their populations can be significant in the physical conditions of pne .
nebular resonance lines are often blended with saturated ism components , but lines from the excited levels do not suffer from this effect .
the column densities of the excited levels can thus be used to determine the column density of the ground state , under the usual assumption of statistical equilibrium .
since the nebular components of the resonance transitions of at @xmath351012.50 , 1190.20 are lost in the saturated ism components and blends with other species , we use absorption lines of @xmath38 and @xmath39 to extrapolate the ground state column density , and subsequently derive the total ionic column density .
the populations of these levels are also sensitive to @xmath9 , and we use the column density ratios to derive the line of sight electron density
. voigt component fits to and profiles are depicted in figure 2a , with the fit parameters detailed in table 1 .
we detected absorption from both excited levels of , at @xmath161015.78 and @xmath351021.11 , 1021.32 , in the _ fuse _ spectrum . since only one @xmath38 line provided useful information ( the other lines in the _ fuse _ and stis wavebands are too severely blended and/or saturated to significantly constrain the fit ) , we chose to simultaneously fit @xmath38 and @xmath39 , fixing @xmath26 and @xmath27 , but allowing the column densities of the two levels to differ . the ground term of consists of a single level , so its ionic column density is given by the simultaneous fit of @xmath351250.58 , 1253.81 .
@xmath161259.52 is more heavily saturated , and does not significantly constrain the ionic column density when simultaneously fit with the other lines of the multiplet .
the two lines we fit are mildly saturated , and therefore the derived column density is not well determined , but this ion does not contribute a significant amount of the total column of s. in order to extrapolate the ground state population of from those of the excited levels , we used the 5-level task `` nebular.ionic '' by shaw & dufour ( 1995 ) in iraf , assuming the electron temperature @xmath40 k and density @xmath41 @xmath1 derived from nebular emission lines ( dm01 ) .
however , at such a high density , the predicted ratio @xmath42 @xmath43@xmath44/@xmath43@xmath39 ) = 0.96 ( hereafter , @xmath45 refers to the population ratio of the @xmath46 to the @xmath47 levels above the ground state ) is nearly a factor of two smaller than the observed value of [email protected] .
note that saturation effects in @xmath38 @xmath161015.78 can not explain the discrepancy , for if this line were saturated , then @xmath49 would be even larger than the observed value .
furthermore , line - fit uncertainties are unable to bring the observed and predicted line ratios into accord forcing the column densities to have a ratio of 0.96 leads to an inferior fit ( higher @xmath50 ) for any value of @xmath27 .
we therefore explored the sensitivity of @xmath49 to the adopted atomic data and physical parameters .
using more recent transition probabilities for ( nahar 1993 ) than the default values for the nebular.ionic task ( mendoza & zeippen 1982 ) has almost no effect on the relative populations of the ground term levels .
however , it should be noted that both sets of values assume ls coupling , which predicts the same relative oscillator strengths of lines within a multiplet .
we also found that varying the electron temperature within a reasonable range ( 515@xmath51 k ) has little effect on the predicted level populations . on the other hand , @xmath52 is sensitive to the assumed electron density , and increases as @xmath9 decreases . in table 2 , we show the predicted values of @xmath53 and @xmath49 for @xmath40 k and a range of electron densities .
the model agrees best with the observed @xmath54 at @xmath55 @xmath1 , with the 1-@xmath31 errors allowing for the range @xmath5613600 @xmath1 .
it is clear that the observed column density ratio @xmath49 is inconsistent with an electron density as high as reported by dm01 , @xmath41 @xmath1 .
it is important to note that dm01 derived the electron density from the optical doublet ratio [ ] @xmath164072/@xmath166725 .
the @xmath57 lines have high critical densities , and are therefore biased toward dense regions of the nebula .
other density diagnostics , such as [ ] @xmath58 and [ ] @xmath59 , imply @xmath60 @xmath1 , but are unreliable since they are near their high density limits .
therefore , the [ ] @xmath164072/@xmath166725 density indicator is the most reliable one measured by dm01 . using the same diagnostic ratio ,
dfp87 also derived a high electron density , @xmath61 @xmath1 .
it is possible that the more highly ionized portion of the nebula has a lower density than that occupied by singly ionized species .
however , this argument is weakened by the fact that flower et al .
( 1984 ) derived @xmath61 @xmath1 from iue observations of ] @xmath62 .
is it possible that the nebular gas along the line of sight to the central star of swst 1 has a significantly lower density than other regions of the nebula ?
since the absorption occurs along a `` pencil beam '' sample of the nebular gas in front of the central star , and thus samples a much smaller portion of the nebula than emission line studies , the electron density could be this low if the nebula is inhomogeneous .
other results ( 4.2 , 5.1 ) indicate that this is indeed the case .
we adopt @xmath0 @xmath1 and @xmath40 k to derive @xmath63 ( see table 2 ) .
this yields @xmath43 @xmath161012.50 ) = 0.77@xmath43@xmath39 ) = @xmath64 @xmath8 , where the errors include the uncertainties in @xmath9 .
the adopted column densities for the ground term levels of are summarized in table 3 . since the _ fuse _
observations do not provide a means of directly determining the overall column density of ionized gas @xmath6(h@xmath65 ) , we derive the fe abundance in swst 1 by using sulfur as a reference element .
s is particularly useful for this purpose , since it is not significantly depleted into dust in the diffuse ism ( e.g. sembach & savage 1996 ; welty et al .
1999 ) , nor is it affected by nucleosynthesis in the progenitor star .
because we do not observe all ionization stages of fe , we use appropriate ionic ratios as surrogates for the elemental abundances . and have a similar ionization potential range to and ( 7.930.7 ev for fe@xmath65 and fe@xmath30 , 10.434.8 ev for s@xmath65 and s@xmath30 ) , and therefore we assume that @xmath43fe@xmath65 + fe@xmath30)/@xmath43s@xmath65 + s@xmath30 ) @xmath11 ( fe / s ) .
we use the recommended solar abundances of asplund et al .
( 2004 ) to derive the logarithmic abundance ratio relative to solar , [ fe / s ] = log(fe / s)@xmath66log(fe / s)@xmath67 . as is the case for , the ground terms of and have multiple fine structure levels that may be significantly populated under nebular conditions .
the column densities of some of these levels could not be measured directly , because absorption lines from the levels were undetected or saturated .
therefore , we computed the populations for these unobserved levels with models .
we describe the methods used in the following subsections , before presenting the derived fe abundance
. only resonance transitions of are detected , despite the fact that this ion has several low - energy levels which may be significantly populated .
the populations of the excited levels must therefore be predicted by models in order to correctly compute the total ionic column density .
iron is not included in the nebular.ionic task , so we use xstar ( kallman & bautista 2001 ) photoionization models to calculate fe ionic level populations .
xstar utilizes atomic data from the iron project ( hummer et al .
1993 ; see bautista & kallman 2001 and references therein ) , and computes populations of numerous levels for each ion .
we assume a uniform temperature and density nebula , with a blackbody central star of @xmath68 k , and adopt other model input parameters from table 12 of dm01 .
we tested the sensitivity of the level populations to the adopted parameters by varying elemental abundances , @xmath69 , radius of the nebula , and central star temperature and luminosity .
the ground term level populations of iron and sulfur ions are negligibly affected by these parameters , as is the derived abundance [ fe / s ] .
we computed models for several values of @xmath9 . from the fine structure levels ( 4.1.1 ) , we adopt @xmath55 @xmath1 , with an allowed range @xmath5613600 @xmath1 .
while fe@xmath65 may reside in the photodissociation region as well as the region , we find a relatively small total column density of fe@xmath65 compared to fe@xmath30 , so that uncertainties in the level populations have little effect on the derived fe abundance .
the resulting fe level population ratios are listed in table 3 , along with the predicted column densities for unobserved or saturated lines .
we find that a significant fraction of fe@xmath65 resides in excited levels for all of the considered @xmath9 ( 400010@xmath70 @xmath1 ) .
the predicted column densities of the excited levels of the ground term are consistent with the measured upper limits of lines in the _ fuse _ bandpass ( table 1 ) . to compute the total column density of fe@xmath65
, we sum the calculated populations of the thirteen levels of the @xmath71 , @xmath72 , and @xmath73 terms ( higher energy levels contribute less than 1% to the fe@xmath65 abundance ) .
the adopted column density for each level is given in table 3 .
excited levels in the @xmath74 ground term of can give rise to several detectable absorption lines in the _ fuse _ bandpass .
has no excited terms with energies below 2.4 ev , so it is sufficient to consider only the @xmath74 levels when computing the ionic column density .
we detect lines arising from all ground term levels ( table 1 ) , although the sole line of @xmath75 at @xmath161130.40 is quite weak .
the column densities derived from the lines of and @xmath38 are very uncertain due to saturation , particularly that of the resonance line @xmath161122.52 .
since a large fraction of ions may reside in the two lowest energy levels , we use xstar and the absorption lines from the three highest energy @xmath74 levels to estimate their column densities . the predicted level population ratios are given in table 3 ; in general , they correspond poorly to the observed ones . the discrepancies in @xmath53 and @xmath49 are due to saturation of the resonance and @xmath38 lines
the fitted column densities are lower limits .
therefore , we adopt the model column densities for and @xmath38 . while @xmath76 is in good agreement with the model predictions
, @xmath77 is too small by a factor of 3 - 5 .
although @xmath75 @xmath161130.40 is only marginally detected , it is very unlikely to be uncertain by such a large amount . since @xmath77 is not sensitive to the adopted physical parameters of the photoionization models , we examine the atomic data as a source of uncertainty .
it is possible that the oscillator strengths , or @xmath28-values , of the @xmath78 multiplet are in error . while the @xmath28-values of nahar & pradhan ( 1996 , hereafter np96 ) are 44% larger than those recommended by morton ( 2003 ; ekberg 1993 ) , the _ relative _ @xmath28-values within the multiplet are the same in the two studies , since both assume ls coupling and ignore relativistic corrections .
if either of these assumptions does not hold , the relative @xmath28-values of lines within the multiplet would change .
alternatively , the adopted transition probabilities ( np96 ) and collision strengths ( zhang 1996 ) used by xstar to derive the level populations could be uncertain . however , the ground term level population ratios derived with this data ( e.g. keenan et al . 2001 ; keenan 2004 , private communication ) agree to within 20% of those calculated with older atomic data ( keenan et al .
1992 ) , showing that our results are not strongly dependent upon the adopted atomic data set .
we are unable to explain the large discrepancy between the observed and predicted @xmath77 , although @xmath161130.40 may have an uncertain @xmath28-value . with this caveat in mind
, we use the observed @xmath76 and @xmath77 ( which are from unsaturated lines ) , and the calculated values for @xmath53 and @xmath49 .
the adopted column densities are listed in table 3 .
we compute the gaseous iron abundance using the sulfur and iron column densities summarized in table 3 , assuming that fe / s @xmath79(fe@xmath65 + fe@xmath30)/@xmath6(s@xmath65 + s@xmath30 ) .
the gaseous fe abundance is quite high , [ fe / s ] = @xmath800.12 , which implies that only about half of the nebular iron is depleted into dust along the line of sight to the central star .
if one uses the @xmath28-values calculated by np96 rather than those compiled by morton ( 2003 ) , [ fe / s ] decreases to @xmath810.12 .
the derived [ fe / s ] is shown in table 4 for the adopted @xmath55 @xmath1 and the 1-@xmath31 limits , 6400 and 13600 @xmath1 ; note that the uncertainties in table 4 are not the formal errors , since they do not include the uncertainty in @xmath9 .
for comparison , we also show the fe abundance derived with the @xmath9 value of dm01 ( @xmath41 @xmath1 ) , and of dfp87 and flower et al .
( 1984 ; @xmath61 @xmath1 ) .
the inferred value of [ fe / s ] decreases as @xmath9 increases , primarily because the ground state population relative to excited levels declines as gas density rises .
furthermore , if we have underestimated the saturation of the lines , the true value of [ fe / s ] would be lower ( because the actual @xmath43s@xmath65 ) would be larger than adopted here ) . even given these considerations , the gaseous iron abundance along the line of sight is clearly high compared to the [ fe / h ] observed in disk ( @xmath82 to @xmath83 dex ) and even halo ( @xmath84 dex ) diffuse interstellar clouds ( sembach & savage 1996 ; welty et al .
1999 ) .
we stress that the modest level of iron depletion derived from the _ fuse _ spectrum is _ not _ necessarily characteristic of the whole nebula . in the following subsection , we show that iron is more heavily depleted in other regions of swst 1 , by deriving the fe abundance from optical emission lines . from five optical [ ] lines , dfp87 computed [ fe / h ] = @xmath85 and [ fe / s ] = @xmath86 , using their value of the s abundance . however , this determination relies on dated atomic data , and therefore may not be reliable .
we re - derive the fe abundance from optical emission lines , using current atomic data and the optical echelle spectrum of dm01 , which was obtained with the anglo - australian telescope ( courtesy of o. de marco ; see also appendix b of dm01 ) .
we measured line fluxes in the optical spectrum with the `` splot '' routine in iraf , and de - reddened them using @xmath87 ( dm01 ) .
the resulting line intensities relative to h@xmath88 are reported in table 5 .
we identify several [ ] and [ ] lines , but none from higher ions . since most of the optical [ ] emission lines may be severely affected by fluorescence ( rodrguez 1999 )
, we derive the fe@xmath65 abundance using only [ ] @xmath89 , which is insensitive to fluorescent pumping ( lucy 1995 ; baldwin et al .
we identify 12 [ ] emission lines , which are of varying utility ( see below ) in determining the fe@xmath30 abundance .
several [ ] emission line ratios may be used as nebular diagnostics , as previously discussed by garstang et al .
( 1978 ) and keenan et al .
( 1993 ) . however , the density diagnostics are often degenerate , allowing for more than one value of @xmath9 ( bautista & pradhan 1998 ; zhang & liu 2002 ) . using the diagnostic diagrams displayed in figure 7 of bautista & pradhan ( 1998 ) , we find that the ratios @xmath90 and @xmath91 indicate either log(@xmath9/@xmath92 or 5.7 . on the other hand , @xmath93 agrees with the other two ratios only for the low log(@xmath9/@xmath1 ) value ; the high value significantly exceeds 6.0 . because of the agreement of all the [ ] density diagnostics at log(@xmath9/@xmath92 , and its close concordance to the value derived by dm01
, we adopt this value for the electron density . unfortunately
, none of the temperature - sensitive transitions discussed by zhang & liu ( 2002 ) were detected .
therefore we adopt @xmath94 k ( dm01 ) to derive the fe ionic abundances . in order to calculate the fe@xmath65 and fe@xmath30 abundances ( table 5 )
, we use xstar models ( see 4.2.1 ) to derive the fractional population in the upper level of each transition .
note that some of the [ ] lines give discrepant ionic abundances compared to other transitions ( table 5 ) . in particular ,
[ ] @xmath354607.0 , 4777.7 , 4881.0 , 4930.5 , and 4987.9 indicate significantly different ionic abundances than the other seven lines .
rodrguez ( 2002 ) noted that @xmath354607.0 and 4777.7 , which yield higher fe@xmath30 abundances , may be contaminated by blends with other species ; in addition , [ ] @xmath164930.5 may be contaminated by [ ] @xmath164931.2 .
however , blends can not explain the anomalously low fluxes of @xmath164881.0 and @xmath164987.9 .
measurement uncertainties are unlikely to be the cause for the inconsistency , particularly in the case of the relatively strong line @xmath164881.0 .
interestingly , both of these lines originate in the @xmath95 level . in table 5
, we compare the observed line intensities to those predicted by keenan et al .
( 2001 ; k01 ) and keenan et al .
( 1992 ; k92 , pre - iron project ) to examine the effects of different sets of atomic data . while the predicted line intensities are similar in most cases , the k92 values actually reproduce the @xmath96 intensities quite well . by comparison , k01 ( who use the same atomic data as we do here )
predict @xmath97 and @xmath98 to be too large by a factor of two or more .
it is possible that the atomic data for the @xmath96 multiplet is not reliable , but it should be noted that neither rodrguez ( 2002 ) nor perinotto et al .
( 1999 ) commented on inconsistencies from these two lines . in any case , we do not use @xmath354607.0 , 4777.7 , 4881.0 , 4930.5 , and 4987.9 in the derivation of the fe@xmath30 abundance .
we derive fe ionic abundances ( table 6 ) using [ ] @xmath168617.0 and the seven most reliable [ ] lines , assuming log@xmath99/@xmath92 and @xmath100 k. we vary log@xmath99/@xmath101 from 4.5 to 5.0 , in accord with previously derived electron densities ( dm01 ; dfp87 ; flower et al . 1984 ) , and assume an uncertainty of @xmath102 k for @xmath69 , as given by dm01 , to produce the error bars on the ionic abundances listed in table 6 . as discussed above ,
fe@xmath65 is a trace ion in the nebula , so the derived fe abundance is not strongly affected even if much of the fe@xmath65 resides in the photodissociation region , under significantly different physical conditions than we assume . in order to correct for unseen stages of ionization ( primarily fe@xmath103 ) , we adopt the ionization correction factor ( icf ) given by rodrguez ( 2002 ) : @xmath104 using data from tables 10 and 11 of dm01 ( we adopt the `` empirical abundance '' for o in their table 11 ) , we find icf(fe ) @xmath105 .
the relatively large error is due to the uncertainty in the total o abundance .
this results in [ fe / h]@xmath3 , or [ fe / s]@xmath4 , using the s abundance from dm01 . in other words , the fe abundance we derive from the optical data of dm01 agrees well with that of dfp87 .
iron is depleted by a factor of 10 or more according to the optical data , which starkly contrasts with the very modest depletion we find from the _ fuse _ data .
thus , it appears that the high gaseous fe abundance that characterizes the column toward the central star is not representative of the emitting ionized gas .
this result is strengthened by the fact that the [ fe / s ] derived from the _ fuse _ spectrum is not strongly dependent upon the assumed @xmath9 ( table 4 ) .
the slight nebular fe depletion along the line of sight implies that other refractory elements will also be relatively undepleted .
sdb derived the ge abundance in four pne ( including swst 1 ) from the resonance line @xmath161088.46 in the _ fuse _ band , and reported a range of possible ge abundances , depending on the unknown degree of depletion .
the range found for swst 1 spanned [ ge / s ] = ( 0.861.48)@xmath106 ( note the typographical error in table 2 of sdb ) , depending on whether ge is wholly gaseous , or is depleted to the same extent as in cold interstellar clouds ( @xmath107 dex ; savage & sembach 1996 ) .
the modest fe depletion indicates that [ ge / s ] is at the low end of the range found by sdb .
we have re - evaluated the ge abundance with our new method of analysis ( 3.1 ) and updated atomic data from morton ( 2003 ) .
using the ionization correction factor ge / s = @xmath6(ge@xmath30)/@xmath6(s@xmath65 + s@xmath30 ) , based on similarities of the ionization potential ranges , we find [ ge / s ] = @xmath1080.06 .
this agrees within the errors with the value [ ge / s ] = @xmath1090.15 calculated by sdb .
many theoretical and observational studies of the _
s_-process discuss _ n_-capture element yields relative to iron .
assuming that ge / fe = @xmath6(ge@xmath30)/@xmath6(fe@xmath30 ) , we derive [ ge / fe ] = @xmath110 using @xmath28-values from morton ( 2003 ) , or @xmath111 using values from np96 .
( note that [ ge / fe ] is not equivalent to [ ge / s ] [ fe / s ] ; the ionization potential range of fe@xmath65 does not coincide with that of ge@xmath30 , and is therefore not used to derive [ ge / fe ] ) . in table 4
, we display the ge abundance relative to s and fe for the range of densities implied by uv absorption lines , as well as the @xmath9 found from optical emission lines .
if iron is depleted , then the derived value of [ ge / fe ] is an upper limit .
however , [ ge / s ] is enhanced by at least a factor of 5 for the entire range of electron densities we consider .
the original motivation for _ fuse _ go program b069 , for which this observation of swst 1 was taken , was to study the molecular components of pne through examination of the h@xmath5 lyman and werner resonance lines seen in absorption against the continua of their central stars .
accordingly , targets were selected to have velocities well - separated from low - velocity foreground interstellar material , and to display near - infrared quadrupole emission lines from vibrationally - excited h@xmath5 or some other indicator of neutral or molecular material .
swst 1 was selected as a target for the h@xmath5 search because it meets both of these criteria .
swst 1 was found by dinerstein et al .
( 1995b ; hereafter dsu ) to possess nebular components in the `` d '' lines at 5890.0 , 5895.9 .
since neutral na has a very low ionization potential ( 5.1 ev ) and appears to correlate with other indicators of atomic and molecular material ( see table 5 of dsu ) , is presumed to be a tracer of neutral material in nebulae . in the case of pure absorption by foreground material ,
the lines can in principle be used to infer a column density of circumstellar material from the classical `` doublet ratio '' method ( e.g. dinerstein & sneden 1988 ) , which has been widely applied to diffuse interstellar clouds .
however , this simple approach is probably invalid for swst 1 , where the modest equivalent widths of the lines ( 55@xmath4815 m ) appear inconsistent with their ratio , which is close to the fully saturated value of unity .
the reason for this inconsistency is evident from figure 5 of dsu , where it can be seen that the nebular absorption in swst 1 is partially filled in by scattered photons ; the true absorption profiles are actually deeper than they appear
. the observed absorption minima fall at a heliocentric radial velocity of 47 km s@xmath19 , blueshifted from the systemic emission line value of 20 km s@xmath19 ; the blending of the emission and absorption components is not likely to affect the apparent velocity by more than a few km s@xmath19 .
circumnebular 21 cm absorption was subsequently reported by gussie & taylor ( 1995 ) , at a velocity of @xmath1138 km s@xmath19 . they did not , however , detect co @xmath112 = 21 millimeter emission , although all of the other four pne they detected in were also seen in co. huggins et al .
( 1996 ) also searched for co @xmath112 = 21 in swst 1 and did not detect it .
this suggests that swst 1 possesses only a modest amount of neutral / molecular material .
swst 1 is a source of near - infrared h@xmath5 emission , although its spectrum has not been studied as intensively as some other pne . in their appendix b , dm01 reported a detection of the 2.122 @xmath10 @xmath113 = 10 s(1 ) line at a strength about 5% that of br@xmath114 .
additional h@xmath5 lines , in particular @xmath113 = 10 s(0 ) 2.223 @xmath10 and several @xmath115-branch lines , are visible in the spectrum of dm01 , although the line fluxes are not cited in their table b2 .
likkel et al .
( 2004 ) also detect 2.122 and 2.223 @xmath10 , but no other h@xmath5 lines ; their data do not extend longward of 2.26 @xmath10 and therefore do not include the @xmath115-branch .
they measure a 2.223/2.122 @xmath10 flux ratio consistent with radiative excitation of moderately dense gas in a strong radiation field
. if radiative excitation is occurring , the lyman and werner lines in the _ fuse _ band , which initiate the fluorescent cascades , should be detectable as long as there is h@xmath5 along the absorption column .
we do not detect nebular h@xmath5 absorption in the _ fuse _ spectrum of swst 1 . in order to quantify this statement , we measure limits on the equivalent widths of nebular components of specific h@xmath5 transitions on the blue wings of the corresponding interstellar features . because the velocity shift is modest ( only 23 spectral resolution elements ) , this is not feasible for lines that are very saturated in the ism , namely those arising from the @xmath112 = 0 and 1 levels . instead , we examine lines from @xmath112 @xmath117 2 , which have less saturated and hence narrower interstellar components .
we select several lines , arising from each of the @xmath112 = 2 , 3 , 4 , and 5 levels , that are free from interference by other species and fall in regions with good s / n ratio in the continuum .
the selected transitions are listed in table 7 , and we display a representative sample of these lines in figure 3 . for comparison ,
we show @xmath118 and 1 lines in the upper panels , from which it can be seen that blending with the interstellar component ( thick solid line ) prevents meaningful measurements of limits on the nebular column density .
we use two methods to set upper limits on the equivalent widths of the nebular components . in some cases
, we were able to `` fit '' a nebular feature by interpolating a conservatively high continuum near the nebular velocity .
for an alternate estimate , we evaluated the rms noise level in the adjacent continuum , assumed a line width equal to the spectral resolution element , and took the corresponding 3-@xmath31 value as our upper limit .
the @xmath119 values from both methods , typically @xmath11
515 m , were converted to column densities @xmath120 by assuming that the lines are on the linear portion of the curve of growth .
the @xmath120 limits are given in columns 4 and 5 of table 7 , and depicted in figure 3 by dotted lines showing voigt profiles with the measured @xmath120 limit at the nebular velocity , assuming @xmath121 km s@xmath19 ( the value measured for nebular lines ) .
overall , our upper limits are approximately @xmath122 @xmath8 for @xmath112 = 2 and 3 , and @xmath123 @xmath8 for @xmath112 = 4 and 5 . in order to translate these level - specific column densities into a total h@xmath5 column , it is necessary to make assumptions about the excitation equilibrium . given @xmath124 , the fraction of h@xmath5 molecules in level @xmath112 , then it is possible to compute @xmath125 = @xmath120/@xmath124 for each @xmath112 .
for purposes of a rough estimate , we assume that the lowest few @xmath112 levels are dominated by collisional processes , so that their populations are predicted by boltzmann statistics . to normalize the level population @xmath120 = @xmath126 , we need the appropriate partition function @xmath127 = @xmath128 ( @xmath120 ) over all @xmath112 levels that contain significant populations . we compute partition functions for h@xmath5 for several representative values of the rotation temperature @xmath129 : 200 , 500 , 800 , and 1600 k. we include the 20 energy levels with the largest populations , which consist of the @xmath112 = 0 15 levels of the @xmath113 = 0 vibrational state , and @xmath112 = 0 5 for @xmath113 = 1 , which begins to interleave the energy level ladder of @xmath113 = 0 between @xmath112 = 8 and 9 .
for radiative excitation , a single value of @xmath129 does not strictly apply ; however , the effective values of @xmath129 for @xmath112 @xmath130 5 for moderately high radiation field and gas densities lie in the range @xmath11 200 500 k ( draine 2004 , private communication ) . furthermore , in bd+30@xmath1313639 , which has highly excited circumstellar h@xmath5 , @xmath129
@xmath11 800 k for @xmath113 = 0 , while the @xmath132 2 levels have substantially higher values of @xmath129 ( dinerstein & bowers 2004 ; dinerstein et al .
2005 , in preparation ) . in table 8
we convert the upper limits on the specific @xmath112 levels into the corresponding total column densities @xmath125(h@xmath5 ) . in general ,
the strongest constraints on the total amount of h@xmath5 come from the @xmath112 levels with the largest fractional populations , but the @xmath112 value with the largest fraction varies with temperature . for @xmath129 @xmath117 500 @xmath133
the tightest constraint comes from the @xmath112 = 3 lines : @xmath125(h@xmath5 ) @xmath11 2 @xmath12 10@xmath134 @xmath8 . for the lowest temperature
we calculate here ( more appropriate for lower - density gas and a weak radiation field ) , the best limit , @xmath135 7 @xmath12 10@xmath134 @xmath8 , is from @xmath112 = 2 .
it should be noted that the derived upper limit to @xmath125(h@xmath136 may be in error if the h@xmath5 is sufficiently cold that only the @xmath112 = 0 and 1 states are significantly populated , since we were not able to derive meaningful upper limits to the column densities of these levels . for example , if @xmath137 k , then 1.2% of the h@xmath5 would be in the @xmath138 state , so that our upper limit of @xmath125(h@xmath139 7 @xmath12 10@xmath134 @xmath8 ( derived for @xmath140 k ) would be too low by a factor of about 6 .
temperatures lower than 60 k would be necessary for our estimate to be too low by two orders of magnitude .
however , given that h@xmath5 in the diffuse ism , far from strong uv radiation fields , exhibits rotational temperatures of @xmath1418090 k ( shull et al .
2004 ; savage et al . 1977 ) , it is very unlikely that circumstellar material close to the luminous , hot central star of a pn would be this cold . the flux of h@xmath5 1 - 0 s(1 ) 2.122 @xmath10 was reported as @xmath142 10@xmath143 erg @xmath8 s@xmath19 by dm01 , and @xmath144 10@xmath143 erg @xmath8 s@xmath19 by likkel et al . ( 2004 ) .
while there are no direct determinations of the size of the h@xmath5-emitting region , the similarity of the fluxes measured in different aperture sizes implies that it must be relatively compact .
_ hst _ images of the ionized gas suggest a knotty ring of approximate dimensions 13 @xmath12 09 , or 4.3 @xmath12 10@xmath145 ster ( dm01 ) . since in most
pne the h@xmath5-emitting regions are slightly more extended than the ionized gas ( e.g. latter et al .
1995 ) , we assume that the h@xmath5 fills solid angle @xmath11 5 @xmath12 10@xmath145 ster .
this corresponds to a surface brightness for the 2.122 @xmath10 line of ( 2.6 @xmath48 0.4 ) @xmath12 10@xmath146 erg @xmath8 s@xmath19 ster@xmath19 .
even this large value , 20 times the surface brightness of the `` canonical '' fluorescent h@xmath5 pn hubble 12 ( dinerstein et al . 1988 ; ramsay et al .
1993 ) , is only a lower limit , since it is beam - averaged and the emission may not be uniform . for optically thin emission , this surface brightness implies a beam - averaged column density in the originating level of the transition ( @xmath113 = 1 , @xmath112 = 3 ) of 1.0 @xmath12 10@xmath147 @xmath8 . even in a highly excited cloud
, this level contains only a small fraction of the total h@xmath5 : for thermal excitation , @xmath148(800@xmath133 ) @xmath11 2@xmath1210@xmath149 and @xmath148(1600@xmath133 ) @xmath11 7@xmath1210@xmath146 .
therefore the total column density of h@xmath5 would have to be at least @xmath125(h@xmath5 ) @xmath150 10@xmath151 @xmath8 . comparing this estimate to our limits on the line of sight @xmath125(h@xmath5 ) from the uv data ,
we see that they differ by _ at least _ four orders of magnitude .
if the absorption line of sight contained the column of h@xmath5 indicated by the infrared emission lines , the nebular components of the uv lines would be deeply saturated , even if the h@xmath5 is very hot ( @xmath152 k ; dashed lines in figure 3 ) .
this disparity indicates an extreme asymmetry or clumpiness in the distribution of molecular material in swst 1 . despite the absence of nebular h@xmath5 absorption in swst 1 , some neutral species are clearly present in the _ fuse _ spectrum . some of the low - ionization species we detect , such as , , and , may exist in both the ionized nebula and in the transition region between ionized and neutral material ( the photodissociation region , or pdr )
however , we also detect absorption from , which has an ionization potential nearly identical to that of , and therefore is expected to reside primarily in the neutral zone . under the conditions of a pdr
, the excited fine structure levels of the @xmath153 ground term may be significantly populated , and therefore can be detected in absorption along with resonance transitions .
if the excitation of these levels is exclusively due to collisions , one can use their column density ratios as a measure of the kinetic gas temperature .
while the lines of the well - known resonance triplet at 1302 - 1306 in the stis waveband suffer from saturation and blending , the _ fuse _ band offers a variety of alternate multiplets , with a range of different @xmath28-values and hence sensitivities to saturation effects .
many of these lie below 1000 , where the spectrum is crowded and has low s / n , so we focus on the multiplet near 1040 .
although the ground state line ( @xmath161039.23 ) can not be used due to saturation and blending with the ism component , simultaneous fits show that the absorption lines from the excited levels in this multiplet are free from such effects ( see figure 2a ) . taking the ratio of the column densities derived from these fits ( table 1 ) , we find @xmath6(@xmath154)/@xmath6(@xmath1550.16 .
this value exceeds by about a factor of two the ratio of statistical weights ( 1/3 ) , which is the limiting value for lte in the infinite temperature case , @xmath156 @xmath157 @xmath158 ( where @xmath159 is the energy difference between the two levels ) .
this population inversion demonstrates incontrovertibly that collisions are not the primary excitation mechanism for the fine structure levels of in swst 1 .
several processes other than collisional excitation can populate the excited levels of .
these processes were discussed and evaluated in detail by grandi ( 1975a , b ) , in order to understand the strength of @xmath168446.5 and other permitted lines of in the orion nebula .
these excitation mechanisms include recombination into excited levels ; fluorescent excitation by the stellar continuum ; or fluorescent excitation by lyman lines , primarily ly @xmath88 but also to some extent ly @xmath160 and ly @xmath161 . in order to distinguish among these possibilities , we examine the intensities of various optical and near - infrared lines in swst 1 as reported in the literature .
we use the data of dm01 ( see 4.2.4 for more details ) to measure optical and near - infrared lines
. the resulting ratios of various line fluxes to @xmath168446.5 are summarized in table 9 .
we also list the corresponding ratios in orion ( esteban et al .
2004 ) , for which grandi ( 1975a , b ) concluded that fluorescent excitation by the stellar continuum was the only mechanism that could explain the observed pattern of lines .
recombination can be ruled out as the dominant excitation mechanism in swst 1 by virtue of the strength of @xmath162 relative to h@xmath163 : the line flux ratio is @xmath110.003 , over a factor of 30 larger than predicted for recombination ( rudy et al .
additional support comes from the weakness of @xmath167774.2 relative to @xmath168446.5 ; if recombination populated these levels , these two lines should have comparable strengths .
starlight excitation is usually cited as the mechanism which produces @xmath168446.5 in galactic nebulae ( grandi 1976 ) , but rudy et al .
( 1989 ) have argued , on the basis of measurements of near - infrared lines , that ly @xmath88 fluorescence is responsible for exciting in the compact pn ic 4997 .
however , lyman line fluorescence can not explain the strengths of lines from states other than @xmath164 , such as @xmath355554.9 , 6046.4 , and 13164 ( grandi 1975a ; rudy et al .
1989 ) , which can not be populated via line excitation .
the general consistency between the line intensities in swst 1 and orion seen in table 9 strongly supports the conclusion that the excitation of the higher levels of is dominated by stellar continuum fluorescence .
since the atoms will cascade to lower levels , eventually reaching the fine structure levels of the ground @xmath153 term , we suggest that the population inversion we find in the _ fuse _ spectrum is a result of this fluorescent cascade initiated by stellar continuum photons .
[ ] @xmath165
@xmath166 63 @xmath10 , the transition from the first excited fine structure level to the ground state , is one of the strongest cooling lines in pdrs . although [ ] 158 @xmath10 is the strongest line at low densities , the 63 @xmath10 line dominates the cooling of pdrs with intense incident uv fields and high densities , conditions that apply to the circumstellar pdrs of pne ( tielens & hollenbach 1985 ; hollenbach & tielens 1997 ) . the companion [ ] @xmath167
@xmath165 line at 145 @xmath10 is weaker than 63 @xmath10 due to the higher energy and lower statistical weight of its originating level . under the assumption that collisional excitation dominates the level populations , ratios among the three lines can be used as diagnostics of the local thermal parameters @xmath168 and @xmath156 ( e.g. watson 1984 ) .
however , the populations in the ground term can be redistributed by fluorescent excitation in a strong stellar radiation field .
as seen above , stellar continuum fluorescence can overpopulate the upper level of the [ ] 145 @xmath10 line relative to its value from purely collisional excitation . therefore ,
when starlight excitation affects the ground term levels of , the true pdr temperature will be lower than indicated by the line flux ratios of [ ] 63 , 145 , and [ ] 158 @xmath10 . in the gas column of swst 1 probed by _ fuse
_ , starlight fluorescent excitation is so dominant that there is no ambiguity about its importance .
however , in other pdrs , more modest contributions might be present but not easily recognized .
observations with nasa s _ kuiper airborne observatory _
( _ kao _ ) and the _ infrared space observatory ( iso ) _ have been used to determine the physical parameters of pdrs for several pne , although swst 1 itself has not been observed . in order to interpret the line ratios ,
a correction must be made for the elevated c / o abundances of these objects .
applying such corrections to measurements from the _ kao _ , dinerstein et al .
( 1995a , 1997 ) found log(@xmath168/@xmath169 , @xmath170 k for the pdrs associated with bd+30@xmath1313639 , ic 418 , ngc 6572 , and ngc 7027 .
( we cite only a lower limit for @xmath156 , since the ratios become insensitive to the precise value for @xmath171 k. ) it is interesting to note that optical and near - infrared line ratios indicate that stellar continuum fluorescence is important in _ all four _ of these pne ( bd+30@xmath1313639 aller & hyung 1995 , rudy et al .
1991b ; ic 418 sharpee et al .
2003 ; ngc 6572 hyung et al .
1994 , rudy et al . 1991a ; ngc 7027 baluteau et al .
1995 , rudy et al .
liu et al . ( 2001 ) observed the same four objects as well as 18 additional pne , with the long wavelength spectrometer on _
iso_. with the exception of ngc 7027 , the measured flux ratios of liu et al
. differ by up to a factor of two from those found by dinerstein et al .
, yielding lower temperatures ( @xmath172400 k ) and much higher densities ( log(@xmath168/@xmath173 ) for these pne .
the reason for this difference is not clear .
one possible source of the discrepancy is the correction for c / o , which is uncertain by as much as a factor of 23 in pne .
another factor is that the _ iso _ observations were taken with a larger entrance aperture than the _ kao _ data , and may include a greater contribution from more extended , cooler gas , where the stellar radiation field is weaker and collisional excitation dominates .
thus , the contribution of the fluorescently excited component may be diluted compared to that seen by _
kao_. this effect has probably been seen , albeit not recognized , in other sources such as the @xmath174 oph cloud ( liseau et al . 1999a ) and carina nebula ( mizutani et al . 2004 ) , where excess emission is seen in the 145 @xmath10 line compared to the predictions of pdr models and other diagnostics of gas temperature . indeed , liseau et al .
( 1999a ) specifically suggest that an unidentified nonthermal pumping mechanism may be responsible ( also see liseau et al .
our _ fuse _ observations of swst 1 provide an explanation for this widespread puzzle .
many of the physical properties that we derive from the _ fuse _ spectrum of swst 1 the electron density , the high gaseous iron abundance , and the absence of nebular h@xmath5 absorption differ dramatically from values derived previously from emission lines .
however , _ fuse _ probes only a `` pencil - beam '' of gas in the line of sight to the central star , a small portion of the nebula compared to emission line studies ; the differences therefore suggest that the absorption line of sight is not representative of conditions elsewhere in the nebula .
these results indicate a non - uniform distribution of ionized and molecular material in swst 1 , either due to a highly asymmetrical global structure , or extreme clumping .
it is possible that global structure can explain the observations .
the images of the optical emission in dm01 suggest a toroidal morphology , tipped with respect to our line of sight .
( assuming that the optical nebula is really a circular ring , the observed aspect ratio implies an inclination of @xmath1160 from pole - on . ) in this case , the _ fuse _ spectrum looks through a polar cone in which the gas has mostly been evacuated .
the line of sight to the central star , which traverses this cone , would then be a region deficient in both molecules and dust , where the gas density is low compared to the toroidal ring .
alternatively , small - scale clumping may be responsible for the nebular inhomogeneities .
in fact , such clumpiness may be quite common in pne .
for example , odell et al .
( 2002 ) find numerous knots in nearby pne , and gonalves et al .
( 2001 ) discuss 50 pne with detected low ionization structures .
small - scale structure may form quite early in the evolution of a pn , from instabilities created during the interaction of the fast and slow winds ( dwarkadas & balick 1998 ) or with the passage of the ionization front ( e.g. garca - segura & franco 1996 ) .
condensations in the progenitor s envelope during the agb phase could also contribute to knots in the ejecta ( dyson et al .
1989 ) , although huggins & mauron ( 2002 ) find a lack of evidence that this occurs .
recent studies have shown that pne with h - deficient central stars ( such as swst 1 ) may be particularly likely to form clumpy nebulae .
acker et al . ( 2002 ) and pea et al . ( 2003 ) find that pne with [ wc ] central stars tend to exhibit a higher degree of turbulence ( which implies the presence of small - scale instabilities ) than those with h - rich stars .
indeed , the high metal abundance and mass - loss rates of [ wc ] central stars likely prolong the momentum - conserving phase of the nebula ( acker et al .
2002 ) , which is prone to nonlinear thin shell instabilities ( dwarkadas & balick 1998 ) .
furthermore , clumpy winds from [ wc ] central stars , as detected by grosdidier et al .
( 2000 ) , can create or enhance inhomogeneities in the nebula . while grosdidier et al .
did not detect inhomogeneous outflows from the central star of swst 1 , they acknowledge that their observations were insufficient to detect wind variability on timescales longer than about an hour .
however , they detected clumpy winds from the central stars of ngc 40 and bd+30@xmath1313639 , which are similar to that of swst 1 .
since the angular scale of swst 1 is so small that even _ hst _
images can not fully resolve its intrinsic structure , it is not possible to assess whether the non - uniform distribution of nebular material is due to global asymmetries or clumping .
however , it is clear that the nebula is inhomogeneous , and the sight line to the central star does not intersect material with the properties exhibited by emission lines .
the spatial segregation of h@xmath5 into dense regions , implied by the absence of uv h@xmath5 absorption in an object with near - ir h@xmath5 emission , is not unexpected in view of the fragility of h@xmath5 molecules to photodissociation and the intensity of the stellar uv radiation field .
tielens ( 1993 ) estimated that the lifetime of h@xmath5 molecules outside of dense knots where self shielding and dust extinction protect them from photodissociation is on the order of a few years , and therefore predicted that h@xmath5 in pne must be confined to dense clumps .
recent high spatial resolution observations of h@xmath5 and co emission in the helix and ring nebulae , where small linear dimensions can be resolved , have confirmed this expectation ( huggins et al .
2002 ; speck et al .
even given the relative youth of swst 1 ( estimated as @xmath11 300 yr by dm01 ) , the timescales for the propagation of photodissociation fronts are sufficiently short ( e.g. hollenbach & natta 1995 ) that the h@xmath5 will be largely dissociated along paths which do not contain condensations with @xmath175 @xmath8 .
the origin of the inhomogeneities in the gas phase fe abundance and the corresponding dust - to - gas ratio in swst 1 is less clear .
our results show a difference of @xmath176 dex between the gas phase fe abundance in the uv absorption column and the emitting gas , where it is depleted by about a factor of @xmath177 , similar to values found in the warm ionized ism and regions ( sembach & savage 1996 ; welty et al .
1999 ; rodrguez 2002 ) .
the low optical fe@xmath30 abundance is corroborated by likkel et al .
( 2004 ) , who find similarly low fe@xmath30/h@xmath65 from near - infrared [ ] lines in swst 1 .
further evidence for the presence of dust in swst 1 comes from the detection of thermal dust emission in the infrared .
swst 1 exhibits the dual dust chemistry that is typical of pne with [ wc ] central stars ( cohen et al .
2002 ) , as indicated by the presence of features from both o - rich ( aitken et al . 1979 ; zhang & kwok 1990 ) and c - rich dust ( roche et al .
1996 ; szczerba et al .
the o - rich grains dominate by mass : silicates account for about 95% of the dust , with the other 5% giving rise to the aromatic emission bands usually attributed to pahs ( casassus et al .
2001 ) . on general considerations , fe should be strongly depleted in regions with abundant silicate grains ( draine 2003 ) , and has also been found to be underabundant in nebulae with c - rich dust ( e.g. shields 1978 ; perinotto et al .
1999 ) .
there are two possible scenarios that could produce the observed difference in fe depletion between the absorption column and the emitting regions .
if the ejected material initially had a uniform dust - to - gas ratio , then the dust must have been destroyed more efficiently in lower density regions of the nebula .
alternatively , variations in dust formation or growth efficiency in a clumpy outflow could have caused a non - uniform dust - to - gas ratio , and the initial inhomogeneity remains imprinted on the evolving nebula .
the main dust destruction agents in nebulae are shocks and energetic stellar photons .
although the central star of swst 1 drives a fast wind ( 900 km s@xmath19 ; dm01 ) , the wind - shocked gas should lie in the central cavity , and be highly ionized and heated to @xmath178 k ( e.g. chu et al .
shocks moving into the dense optically emitting material would be much slower ( tielens 1993 ) , and therefore less able to destroy the dust .
the fact that we measure radial velocities for the uv absorption lines that are consistent with the overall expansion velocity of the optical emitting gas ( dm01 ) argues that the absorbing gas was not accelerated by a shock .
in addition , the ionization balance seen from the uv absorption lines is consistent with that indicated by the emission lines .
therefore , based on kinematic and excitation arguments , shock processing of the gas and dust in swst 1 seems improbable . unlike h@xmath5 molecules , dust grains are relatively robust with respect to evaporation by stellar radiation , with the exception of small grains ( containing fewer than 2030 atoms ) that have small heat capacities relative to the energies of typical stellar uv photons .
these small grains can not re - radiate the absorbed photon energy via an equilibrium process , and can undergo large temperature fluctuations .
such grains , whether carbon or silicate - based , can be heated to temperatures exceeding the sublimation temperature of refractory elements after absorbing just one or two high - energy photons ( e.g. guhathakurta & draine 1989 ) .
observational studies have drawn mixed conclusions about radiative destruction of dust in pne . observed gas phase depletions of elements such as fe and mg
do not show any clear trends with nebular age ( middlemass 1988 ; perinotto et al .
similarly , it does not appear that dust - to - gas ratios derived from infrared emission are lower in older , more evolved pne ( hoare et al . 1992 ) .
such arguments depend on comparing different objects , which may not be appropriate if the initial dust formation efficiency differs among their progenitor stars .
another approach is to search for evidence of dust destruction within specific nebulae . in this regard , the fact that the dust - to - gas ratios derived for the ionized and neutral ( pdr ) regions in ngc 7027 agree to within a factor of two is taken as evidence against dust destruction in the ionized gas by hoare et al .
( 1992 ) . on the other hand
, there is evidence for a `` depletion gradient '' in some pne in the sense that some refractory elements ( e.g. al , ca , and mg ) are less depleted in regions of higher than those of lower ionization , suggesting that atoms have been released from grains into the gas phase by the stellar radiation field ( kingdon et al .
1995 ; kingdon & ferland 1997 ) . in swst 1
, we find very disparate gas phase fe depletions in different regions of the nebula _ with the same ionization states _ , which makes it seem unlikely that there has been a substantial difference in the amount of radiative dust destruction between these two regions .
this indicates that the disparate fe depletions are most likely caused by initial inhomogeneities in the dust formation or growth efficiency in the agb outflow .
such inhomogeneities could be caused by low temperature structures ( e.g. starspots ) in the atmosphere of the progenitor agb star , as suggested by frank ( 1995 ) and soker & clayton ( 1999 ) .
in addition , the grain growth efficiency is higher in denser regions of a clumpy outflow ( e.g. dorfi & hfner 1996 ) , so that the grain size distribution would be skewed toward larger radii in dense gas .
if the density inhomogeneities are preserved into the pn phase , dust in the more rarefied gas is more susceptible to photodestruction because of the smaller grain size distribution .
this implies that radiative dust destruction could have been more efficient in lower density regions of swst 1 , even though the incident radiation field is similar for the low and high density gas ( as indicated by the comparable ionization balance of the absorbing and emitting gas ) .
the dust deficit in the gas probed by the uv absorption lines may have a direct influence on the amount of h@xmath5 present , since the primary channel for h@xmath5 formation is through grain surface reactions .
the removal of this mechanism for reforming h@xmath5 is likely to lower the equilibrium h@xmath5 abundance that can be maintained in the presence of a dissociating and ionizing radiation field ( aleman & gruenwald 2004 and references therein ) .
therefore , although we attribute the extreme contrast in the h@xmath5 abundance primarily to photodestruction , it is possible that the dust inhomogeneities amplify and maintain the spatial variations in molecular material .
germanium ( @xmath179 ) isotopes can be produced by slow neutron(_n_)-capture nucleosynthesis ( the _ s_-process ) in pn progenitor stars during the agb phase of evolution .
nuclei created by the _ s_-process may be convectively transported to the stellar envelope along with carbon during third dredge - up ( tdu ; e.g. busso et al .
the fact that we find an enhanced ge abundance relative to solar in swst 1 implies that the _ s_-process and tdu were efficient in the progenitor star . we measure a ge enrichment [ ge / s ] = @xmath1080.06 , or [ ge / fe ] = @xmath110 , which can be compared with theoretical models of stellar _
s_-process enrichments , given the metallicity . the s abundance derived by dm01 and
dfp87 implies that the progenitor star of swst 1 may have had a metallicity as low as @xmath180 .
however , the o and ne abundances indicate that the metallicity is closer to solar .
the derived ge overabundance is slightly larger than predicted by recent theoretical models of the _ s_-process in this metallicity range ( busso et al .
2001 ; busso 2003 , private communication ; goriely & mowlavi 2000 ) . however , many uncertainties exist in the strength of tdu and in the effect of convective overshoot and rotation on the _ s_-process ( e.g. herwig et al . 2003 ; siess et al .
therefore , we consider the derived ge enrichment to be consistent with evolutionary models , within uncertainties in the metallicity and yields .
we now summarize the principal results of our study : \1 .
we have analyzed the _ fuse _ spectrum of nebular gas in front of the central star of the compact pn swst 1 .
the absorbing nebular gas exhibits a low degree of ionization , consistent with previous studies of this object .
we find that the absorbing species in swst 1 lie at a heliocentric velocity of @xmath181 to @xmath182 km s@xmath19 , in good agreement with na d measurements .
\2 . from the column density ratio of fine structure levels
, we derive an electron density @xmath0 @xmath1 , which is at least a factor of three lower than found in previous optical and uv emission line studies .
we derive the abundance of iron relative to sulfur , accounting for populations of all of the low - energy levels of the relevant ions .
we find [ fe / s ] = @xmath800.12 or @xmath183 , depending on the adopted @xmath28-values .
the gaseous iron abundance is insensitive to uncertainties in @xmath9 and @xmath69 , and is high in comparison to interstellar medium values , and to the fe abundance derived from optical [ ] emission lines in swst 1 ( [ fe / h]@xmath3 and [ fe / s]@xmath4 ) .
this indicates the presence of marked inhomogeneities in the dust - to - gas ratio in different regions of the nebula .
the high gaseous fe abundance indicates that other refractory elements should not be strongly depleted in the sight line .
thus , the elemental ge abundance lies at the low end of the range suggested by sdb .
we re - derive [ ge / s ] and find a value of [email protected] .
the large ge overabundance implies that the progenitor of swst 1 experienced _
s_-process nucleosynthesis and efficient third dredge - up while on the agb .
. we do not detect nebular h@xmath5 absorption in the _ fuse _ spectrum .
we estimate an upper limit to the column density of @xmath43h@xmath7 @xmath8 .
this limit is more than four orders of magnitude smaller than the beam - averaged column density derived from near - infrared vibrationally excited h@xmath5 lines .
apparently , little of this molecular material lies along the line of sight to the central star .
the low electron density , non - detection of nebular h@xmath5 , and high gaseous fe abundance derived from the _ fuse _ spectrum of swst 1 implies that the gas lying in the direction of the central star has markedly different properties from the regions of the nebula sampled by emission lines . since the _ fuse _ spectrum samples only the gas along the line of sight toward the central star , this indicates that both the ionized and molecular gas in swst 1 are non - uniformly distributed .
however , it is not possible to determine whether the nebular inhomogeneities arise from global asymmetries or small - scale clumping of nebular material , since swst 1 s angular size is small enough that its internal structure is not resolved in _ hst _
images .
swst 1 exhibits absorption features from excited fine structure levels of neutral oxygen at 1040.94 and 1041.69 .
these levels must have been populated by a non - collisional process , as indicated by the inverted population structure .
using optical line ratios , we find that fluorescent pumping by the stellar continuum is the most likely excitation mechanism .
the fine structure levels of the ground term give rise to lines at 63 and 145 @xmath184 m , whose diagnostic value is compromised if the level populations are affected by fluorescent processes , rather than being determined purely by collisional excitation . the _ hst _ data presented in this paper were obtained from the multimission archive at the space telescope science institute ( mast ) .
stsci is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas 5 - 26555 .
support for mast for non - hst data is provided by the nasa office of space science via grant nag 5 - 7584 and by other grants and contracts .
we are indebted to o. de marco for making the optical spectrum of swst 1 available to us , and s. mccandliss for providing h@xmath5 line data .
we also thank f. keenan , who supplied us with level population data , and m. barlow , b. draine , and s. federman for helpful suggestions and discussion .
t. kallman provided invaluable assistance with the operation of the xstar code .
this work was supported by nasa contracts nag 5 - 9239 , nag 5 - 11597 , and nsf grant ast 97 - 31156 .
dinerstein , h. l. , haas , m. r. , erickson , e. f. , & werner , m. w. 1995a , in airborne astronomy symposium on the galactic ecosystem , asp conf . ser .
73 , eds .
m. r. haas , j. a. davidson , & e. f. erickson ( san francisco : asp ) , 387 llccc & 1048.22 , 1066.66 & @xmath1851.6 & [email protected] & [email protected] ( 14 ) + & 1143.23 , 1144.94 & @xmath1861.3 & [email protected] & [email protected] ( 13 ) + @xmath187 & 1148.28 & & & @xmath188 1.51 ( 13 ) + @xmath39 & 1150.69 & & & @xmath188 1.55 ( 13 ) + @xmath189 & 1148.96 & & & @xmath188 1.17 ( 13 ) + @xmath75 & 1154.40 & & & @xmath188 1.00 ( 13 ) + & 1122.52 & @xmath1901.7 & 7.0@xmath191 & 6.76@xmath192 ( 14 ) + @xmath187 & 1124.87 , 1128.04 & @xmath1901.7 & 7.0@xmath191 & 2.09@xmath193 ( 14 ) + @xmath39 & 1128.72 , 1131.91 & @xmath1901.7 & 7.0@xmath191 & [email protected] ( 14 ) + @xmath189 & 1129.19 , 1131.19 & @xmath1901.7 & 7.0@xmath191 & [email protected] ( 14 ) + @xmath75 & 1130.40 & @xmath1901.7 & 7.0@xmath191 & [email protected] ( 13 ) + & 1088.46 & @xmath1940.5 & [email protected] & [email protected] ( 12 ) + & 1039.23 & @xmath1951.5 & [email protected] & [email protected] ( 15 ) + @xmath187 & 1040.94 , 1304.86 & @xmath1963.3 & 11.4@xmath197 & [email protected] ( 15 ) + @xmath39 & 1041.69 , 1306.03 & @xmath1963.3 & 11.4@xmath197 & [email protected] ( 14 ) + & 1250.58 , 1253.81 & @xmath1982.8 & [email protected] & [email protected] ( 14 ) + @xmath187 & 1015.78 & @xmath1990.7 & [email protected] & [email protected] ( 15 ) + @xmath39 & 1021.11 , 1021.32 & @xmath1990.7 & [email protected] & [email protected] ( 14 ) + lcc 4000 & 1.83 & 2.87 + 5000 & 1.42 & 2.49 + 6000 & 1.14 & 2.21 + 7000 & 0.96 & 2.01 + 8000 & 0.85 & 1.85 + 8800 & 0.77 & 1.74 + 9000 & 0.77 & 1.71 + 10000 & 0.70 & 1.63 + 15000 & 0.51 & 1.31 + 20000 & 0.41 & 1.14 + 25000 & 0.36 & 1.04 + 31600 & 0.32 & 0.96 + observed & & [email protected] + lcccccc & 0 & @xmath200 & 0.77 & 0.531.05 & & [email protected] ( 14 ) + & 1 & @xmath201 & 1.74 & 1.362.12 & [email protected] & [email protected] ( 15 ) + & 2 & @xmath202 & 1.00 & 1.001.00 & 1.00 & [email protected] ( 14 ) + & 0 & @xmath203 & 1.00 & 1.001.00 & 1.00 & [email protected] ( 13 ) + & 1 & @xmath204 & 0.24 & 0.160.32 & @xmath188 0.28 & [email protected] ( 13 ) + & 2 & @xmath205 & 0.14 & 0.100.19 & @xmath188 0.29 & [email protected] ( 12 ) + & 3 & @xmath206 & 8.8 ( -2 ) & 6.412.3 ( -2 ) & @xmath188 0.22 & [email protected] ( 12 ) + & 4 & @xmath207 & 4.7 ( -2 ) & 3.56.4 ( -2 ) & @xmath188 0.19 & [email protected] ( 12 ) + & 5 & @xmath208 & 1.02 & 0.861.08 & & [email protected] ( 13 ) + & 6 & @xmath209 & 0.15 & 0.100.22 & & [email protected] ( 12 ) + & 7 & @xmath210 & 7.4 ( -2 ) & 4.711.7 ( -2 ) & & [email protected] ( 12 ) + & 8 & @xmath211 & 4.9 ( -2 ) & 3.27.7 ( -2 ) & & [email protected] ( 12 ) + & 9 & @xmath212 & 2.7 ( -2 ) & 1.94.2 ( -2 ) & & [email protected] ( 12 ) + & 10 & @xmath213 & 9.2 ( -3 ) & 5.616.7 ( -3 ) & & [email protected] ( 11 ) + & 11 & @xmath214 & 4.8 ( -3 ) & 2.89.3 ( -3 ) & & [email protected] ( 11 ) + & 12 & @xmath215 & 2.4 ( -3 ) & 1.44.6 ( -3 ) & & [email protected] ( 11 ) + & 0 & @xmath216 & 14.1 & 9.918.7 & @xmath217 3.4 & [email protected] ( 15 ) + & 1 & @xmath218 & 1.64 & 1.611.64 & 1.05@xmath219 & [email protected] ( 14 ) + & 2 & @xmath220 & 1.00 & 1.001.00 & 1.00 & [email protected] ( 14 ) + & 3 & @xmath221 & 0.73 & 0.700.76 & [email protected] & [email protected] ( 14 ) + & 4 & @xmath222 & 0.38 & 0.320.45 & [email protected] ( -2 ) & [email protected] ( 13 ) + lccc 6400 & @xmath2230.02 & [email protected] & [email protected] + 8800 & @xmath800.02 & [email protected] & [email protected] + 13600 & @xmath2240.02 & [email protected] & [email protected] + 31600 & @xmath2250.02 & [email protected] & [email protected] + 10@xmath70 & @xmath2260.02 & [email protected] & [email protected] + lcccccc @xmath227@xmath228 $ ] & 8617.0 & 8615.9 &
[email protected] & & & 2.0 (-8 ) + @xmath227@xmath228 $ ] & 4607.0 & 4606.6 & [email protected] & 1.0 & 0.8 & 8.0 ( -7 ) + & 4658.1 & 4657.6 & [email protected] & 12.3 & 12.3 & 5.5 ( -7 ) + & 4701.5 & 4701.0 & [email protected] & 5.3 & 5.3 & 5.2 ( -7 ) + & 4733.9 & 4733.3 & [email protected] & 2.5 & 2.7 & 5.6 ( -7 ) + & 4754.7 & 4754.2 & [email protected] & 2.3 & 2.2 & 5.3 ( -7 ) + & 4769.4 & 4768.9 & [email protected] & 1.8 & 1.7 & 5.3 ( -7 ) + & 4777.7 & 4777.1 & [email protected] & 1.2 & 1.3 & 7.5 ( -7 ) + & 4881.0 & 4880.4 & [email protected] & 10.5 & 4.4 & 2.1 ( -7 ) + & 4930.5 & 4930.1 & [email protected] & 0.7 & & 8.5 ( -7 ) + & 4987.9 & 4986.6 & [email protected] & 2.0 & 0.8 & 2.5 ( -7 ) + & 5084.8 & 5084.2 & [email protected] & 0.4 & 0.3 & 5.1 ( -7 ) + & 5270.4 & 5269.9 & [email protected] & 6.5 & 4.0 & 5.5 ( -7 ) + lr fe@xmath65/h@xmath65 & [email protected] (-8 ) + fe@xmath30/h@xmath65 & [email protected] ( -7 ) + icf & [email protected] + fe / h & [email protected] ( -7 ) + @xmath227fe / h@xmath228 $ ] & @xmath2290.24 + @xmath227fe / s@xmath228 $ ] & @xmath2300.33 + lcccc 8 - 0 r(2 ) & 1003.98 & 1.662 ( -2 ) & & @xmath188 7.89 ( 13 ) + 8 - 0 p(2 ) & 1005.39 & 9.911 ( -3 ) & @xmath188 7.59 ( 13 ) & @xmath188 1.40 ( 14 ) + 0 - 0 q(2 ) & 1010.94 & 2.381 ( -2 ) & & @xmath188 4.99 ( 13 ) + 7 - 0 p(2 ) & 1016.46 & 1.023 ( -2 ) & & @xmath188 9.44 ( 13 ) + 4 - 0 p(2 ) & 1053.28 & 9.021 ( -3 ) & & @xmath188 1.05 ( 14 ) + 2 - 0 p(2 ) & 1081.27 & 4.699 ( -3 ) & & @xmath188 1.95 ( 14 ) + 1 - 0 p(2 ) & 1096.44 & 2.366 ( -3 ) & @xmath188 4.57 ( 14 ) & @xmath188 3.55 ( 14 ) + 7 - 0 r(3 ) & 1017.42 & 1.838 ( -2 ) & @xmath188 5.37 ( 13 ) & @xmath188 6.07 ( 13 ) + 4 - 0 r(3 ) & 1053.98 & 1.336 ( -2 ) & @xmath188 8.32 ( 13 ) & @xmath188 5.61 ( 13 ) + 4 - 0 p(3 ) & 1056.47 & 9.556 ( -3 ) & @xmath188 6.46 ( 13 ) & @xmath188 8.37 ( 13 ) + 3 - 0 p(3 ) & 1070.14 & 7.537 ( -3 ) & & @xmath188 1.24 ( 14 ) + 2 - 0 r(3 ) & 1081.71 & 6.363 ( -3 ) & @xmath188 2.82 ( 13 ) & @xmath188 1.43 ( 14 ) + 1 - 0 p(3 ) & 1099.79 & 2.526 ( -3 ) & & @xmath188 2.90 ( 14 ) + 5 - 0 r(4 ) & 1044.54 & 1.548 ( -2 ) & @xmath188 5.50 ( 13 ) & @xmath188 4.77 ( 13 ) + 4 - 0 r(4 ) & 1057.38 &
1.294 ( -2 ) & @xmath188 3.55 ( 13 ) & @xmath188 7.49 ( 13 ) + 3 - 0 p(4 ) & 1074.31 & 7.744 ( -3 ) & & @xmath188 9.73 ( 13 ) + 2 - 0 p(4 ) & 1088.80 & 5.151 ( -3 ) & @xmath188 1.23 ( 14 ) & @xmath188 1.98 ( 14 ) + 0 - 0 q(5 ) & 1017.83 & 2.389 ( -2 ) & & @xmath188 4.39 ( 13 ) + 3 - 0 r(5 ) & 1075.24 & 9.279 ( -3 ) & & @xmath188 9.39 ( 13 ) + cccccccccc 2 & @xmath231 ( 13 ) & 7.3 ( -2 ) & 1.4 ( -1 ) & 1.3 ( -1 ) & 9.1 ( -2 ) & 6.8 ( 14 ) & 3.5 ( 14 ) & 3.7 ( 14 ) & 5.5 ( 14 ) + 3 & @xmath231 ( 13 ) & 2.4 ( -2 ) & 2.2 ( -1 ) & 3.0 ( -1 ) & 2.8 ( -1 ) & 2.1 ( 15 ) & 2.3 ( 14 ) & 1.6 ( 14 ) & 1.8 ( 14 ) + 4 & @xmath232 ( 13 ) & 4.0 ( -4 ) & 2.5 ( -2 ) & 5.6 ( -2 ) & 7.8 ( -2 ) & 1.0 ( 17 ) & 1.6 ( 15 ) & 7.1 ( 14 ) & 5.1 ( 14 ) + 5 & @xmath232 ( 13 ) & 2.0 ( -5 ) & 1.8 ( -2 ) & 7.3 ( -2 ) & 1.7 ( -1 ) & 2.0 ( 18 ) & 2.2 ( 15 ) & 5.5 ( 14 ) & 2.3 ( 14 ) + lcc ] @xmath233 & [email protected] ( -1 ) & 8.3 ( -2 ) + ] @xmath234 & [email protected] ( -2 ) & 2.7 ( -2 ) + ] @xmath235 & [email protected] ( -2 ) & 2.8 ( -2 ) + ] @xmath236 & [email protected] ( -2 ) & 4.3 ( -2 ) + @xmath237 & [email protected] ( -1 ) & 1.0 ( -1 ) + @xmath238 & [email protected] ( -2 ) & 9.8 ( -2 ) + @xmath239 & [email protected] ( -2 ) & 1.2 ( -1 ) + @xmath240 & [email protected] ( -2 ) & + @xmath241 & [email protected] & + @xmath242 & 129 & 113 + @xmath243 & 102 & 326 + | we present _ far ultraviolet spectroscopic explorer _
( _ fuse _ ) observations of the young , compact planetary nebula ( pn ) swst 1 along the line of sight to its central star hd 167362 .
we detect circumstellar absorption lines from several species against the continuum of the central star .
the physical parameters of the nebula derived from the _ fuse _ data differ significantly from those found from emission lines .
we derive an electron density @xmath0 @xmath1 from the column density ratio of the excited fine structure levels , which is at least a factor of 3 lower than all prior estimates .
the gaseous iron abundance derived from the uv lines is quite high ( [ fe / s ] = @xmath2 ) , which implies that iron is not significantly depleted into dust . in contrast ,
optical and near - infrared emission lines indicate that fe is more strongly depleted : [ fe / h]@xmath3 and [ fe / s]@xmath4
. we do not detect nebular h@xmath5 absorption , to a limit @xmath6(h@xmath7 @xmath8 , at least four orders of magnitude lower than the column density estimated from infrared h@xmath5 emission lines .
taken together , the lack of h@xmath5 absorption , low @xmath9 , and high gaseous fe abundance derived from the _ fuse _ spectrum provide strong evidence that dense structures ( which can shield molecules and dust from the destructive effects of energetic stellar photons ) are not present along the line of sight to the central star . on the other hand ,
there is substantial evidence for dust , molecular material , and dense gas elsewhere in swst 1 .
therefore , we conclude that the nebula must have an inhomogeneous structure .
we detect nebular absorption at 1040.94 and 1041.69 from the two excited fine structure levels of neutral oxygen .
these levels give rise to far - infrared emission lines at 63 and 145 @xmath10 which are often used to infer gas properties , particularly temperature , under the assumption that they are collisionally excited .
we find that the fine structure levels in swst 1 have an inverted population ratio .
this requires a non - thermal excitation mechanism , which we identify as fluorescent excitation by the stellar continuum . to the extent that fluorescence affects the level populations , the far - infrared [ ] line strengths
can not be directly used as diagnostics of density and temperature . |
the problem of checking the _ validity _ of quasiequations in finitely generated ( i.e. , generated by a finite set of finite algebras ) quasivarieties or , similarly , checking consequences from finite sets of formulas in finite - valued logics , is decidable and has been investigated extensively in the literature .
in particular , uniform methods for generating proof systems to check validity such as tableaux , resolution , and multisequent calculi , have been developed , as have standard optimization techniques for these systems such as lemma generation and indexing ( see , e.g. , @xcite ) .
however , checking the _ admissibility _ of quasiequations in finitely generated quasivarieties , or similarly , checking the admissibility of rules in finite - valued logics , is not so well - understood .
the problem is decidable , but a naive approach leads to computationally unfeasible procedures even for small sets of small algebras .
the main goal of this paper is to define uniform methods that generate computationally acceptable proof systems for checking admissibility in an arbitrary finitely generated quasivariety .
intuitively , a rule is said to be admissible in a logical system if it can be added to the system without producing any new theorems .
more formally , a quasiequation is admissible in a class of algebras @xmath1 if every @xmath1-unifier of the premises is a @xmath1-unifier of the conclusion , where a @xmath1-unifier of an equation @xmath2 is a substitution @xmath3 such that @xmath4 is valid in @xmath1 .
admissibility plays a fundamental meta - level role in describing `` hidden properties '' of classes of algebras and logical systems .
for example , establishing the completeness of a logical system with respect to some restricted class of algebras ( perhaps just one standard algebra ) often involves showing that a certain rule or quasiequation is admissible ; see , e.g. , @xcite for applications of the admissibility of rules in the context of fuzzy logics .
also , the closely related problem of deciding unifiability of concepts can be a useful tool for database redundancy checking in description logics @xcite .
moreover , it may be possible to automatically obtain admissible rules for classes of algebras and logics that can then be used to simplify reasoning steps or to speed up derivations for checking validity .
admissibility ( in tandem with unification ) has been studied intensively in the context of intermediate and transitive modal logics and their algebras @xcite , leading in some cases to proof systems for checking admissibility @xcite .
axiomatizations and characterizations have also been obtained for certain families of finite algebras and many - valued logics , in particular ukasiewicz logics ( or classes of mv - algebras ) @xcite and other fuzzy logics @xcite , fragments of the substructural logic r - mingle @xcite , and classes of de morgan algebras @xcite . however , a general theory , covering arbitrary finite algebras and finite - valued logics , has so far been lacking .
the starting point for this work is the observation ( see lemma 4.1.10 of @xcite and corollary [ c : admfin ] below ) that for a finite set of finite algebras @xmath1 , admissibility in the quasivariety @xmath5 amounts to validity in the free algebra on @xmath6 generators @xmath7 , where @xmath6 is the maximum cardinality of the algebras in @xmath1 . since by birkhoff s theorem on the structure of free algebras @xcite , this algebra @xmath7 is finite , checking admissibility in @xmath5 is decidable . on the other hand , even for small @xmath6 and a small set of small algebras @xmath1
, the size of @xmath7 may be prohibitively large for checking validity .
this is striking since validity and admissibility in @xmath5 may coincide , @xmath5 is then called _ structurally complete _ , or at least coincide for quasiequations with @xmath5-unifiable premises , in which case , @xmath5 is called _ almost structurally complete_. in other cases , @xmath5-admissibility may correspond to validity in other , often quite small , algebras .
we provide general algorithms here that discover such algebras , or , more precisely , generate finite sets of finite algebras such that the @xmath5-admissibility of a quasiequation corresponds to validity in the quasivariety generated by these algebras .
it is shown , moreover , that these are the smallest sets of algebras with this property with respect to a standard well - ordering on the multiset of their cardinalities .
we proceed as follows .
first , in section [ s : prelim ] , we recall some basic notions from universal algebra
. then in section [ s : fingen ] , we introduce some key ideas and methods for finitely generated quasivarieties ; in particular , we apply a standard multiset well - ordering to the cardinalities of algebras in generating sets for quasivarieties , and provide an algorithm for finding the ( unique up to isomorphism ) minimal generating set of a finitely generated quasivariety .
section [ s : admfree ] provides characterizations of admissibility , unifiability , structural completeness , and almost structural completeness .
these characterizations are then exploited in section [ s : algorithms ] to define corresponding algorithms , and illustrated using a selection of well - known finite algebras , confirming some known results from the literature , and establishing new ones . in section [ s :
logics ] , the approach is extended to finite - valued logics , where the designated values as well as the underlying finite algebra play a significant role .
finally , in section [ s : concluding ] , we conclude with some remarks on future directions for this research .
let us first recall some basic ideas from universal algebra , referring to @xcite for further details . given an algebraic language @xmath8 ( i.e. , without relation symbols ) , an _
@xmath9 is an algebraic structure consisting of a set @xmath10 ( the universe ) and an @xmath6-ary function @xmath11 corresponding to each @xmath6-ary function symbol @xmath12 of @xmath8 ( as usual , calling nullary functions _
constants _ ) .
we call @xmath9 _ finite _ if @xmath10 is a finite set and @xmath8 consists of finitely many function symbols
congruence _ on an @xmath8-algebra @xmath9 is an equivalence relation @xmath13 on @xmath10 satisfying for each @xmath6-ary function symbol @xmath12 of @xmath8 : @xmath14 implies @xmath15 .
the congruences of @xmath9 form a complete lattice @xmath16 with bottom element @xmath17 and top element @xmath18 , where the meet of a set of congruences on @xmath9 is just the intersection of those congruences . given @xmath19 , the _ quotient algebra of @xmath9 by @xmath13 _ is the @xmath8-algebra @xmath20 with universe @xmath21 consisting of the equivalence classes @xmath22 for @xmath23 with functions defined for each @xmath6-ary function symbol @xmath12 of @xmath8 by @xmath24 .
term algebras @xmath25 are defined over a set of variables @xmath26 in the usual way , writing just @xmath27 when @xmath26 is a fixed countably infinite set , and letting @xmath28 stand for arbitrary members of the universe @xmath29 called _
@xmath8-terms_. an _
@xmath8-equation _ is an ordered pair of @xmath8-terms , written @xmath2 .
@xmath8-clause _ is defined as an ordered pair @xmath30 of finite sets of @xmath8-equations , written @xmath31 , and called an _ @xmath8-quasiequation _ if @xmath32 and an _ @xmath8-negative clause _ if @xmath33 . as usual ,
if the language is clear from the context , we may omit the prefix @xmath8 .
let us fix @xmath1 to be a class of @xmath8-algebras , noting that often in what follows @xmath1 will consist of a finite set of @xmath8-algebras @xmath34 , and in this case we typically omit brackets .
given a finite set of @xmath8-equations @xmath35 , we write @xmath36 and say that the @xmath8-clause @xmath31 is _ @xmath1-valid _ , if for every @xmath37 and homomorphism @xmath38 , @xmath39 recalling that @xmath40 .
we also say that @xmath41 is _
@xmath1-satisfiable _ if @xmath42 for some @xmath37 and homomorphism @xmath38 .
the class @xmath1 is said to be _
axiomatized _ by a set of @xmath8-clauses @xmath43 if @xmath1 is the class of @xmath8-algebras @xmath9 such that all @xmath8-clauses in @xmath43 are @xmath9-valid .
@xmath1 is called a _ variety _ , _ quasivariety _ , or _ antivariety _ if it is axiomatized by a set of @xmath8-equations , @xmath8-quasiequations , or @xmath8-negative clauses , respectively . the variety @xmath44 , quasivariety @xmath5 , and antivariety @xmath45 _ generated by _
@xmath1 are the smallest variety , quasivariety , and antivariety containing @xmath1 , respectively .
moreover , let @xmath46 , @xmath47 , @xmath48 , @xmath49 , @xmath50 , @xmath51 , and @xmath52 be the class operators of taking homomorphic images , isomorphic images , subalgebras , products , ultraproducts , non - empty ultraproducts , and homomorphic preimages , respectively .
then @xmath53 , @xmath54 , and @xmath55 , and if @xmath1 is a finite set of finite algebras , these last two equivalences refine to @xmath56 and @xmath57 ( see @xcite for details ) .
given a language @xmath8 and a set of variables @xmath26 such that either @xmath58 or @xmath8 contains at least one constant symbol , the term algebra @xmath25 exists and admits a congruence : @xmath59 following @xcite , we let @xmath60 and define the _ free algebra of @xmath1 over @xmath61 _ : @xmath62 then @xmath63 has the universal mapping property for @xmath1 over @xmath61 : namely , for each @xmath37 , any map from @xmath61 to @xmath10 extends to a homomorphism from @xmath63 to @xmath9 ( ( * ? ? ?
* theorem ii.10.10 ) ) .
if also @xmath64 , then @xmath65 ( ( * ? ?
* theorem ii.10.12 ) ) .
note that @xmath66 whenever @xmath67 ( also @xmath68 if @xmath1 contains at least one non - trivial algebra ) .
hence we may consider for each cardinal @xmath69 , the ( unique up to isomorphism ) _ free algebra of @xmath1 on @xmath69 generators _ @xmath70 , where @xmath71 is a subalgebra of @xmath72 for cardinals @xmath73 .
we note also that @xmath74 and that @xmath75 implies @xmath76 ( see ( * ? ? ?
* corollary ii.11.10 ; exercise ii.11.2 ) ) .
a quasivariety @xmath0 is said to be _ finitely generated _ if @xmath77 for some finite ( generating ) set @xmath1 of finite @xmath8-algebras .
our first goal in this section will be to define a reasonable measure for comparing these generating sets ; we will then describe a method for obtaining a `` smallest '' generating set according to this measure .
more precisely , we apply the standard multiset well - ordering defined by dershowitz and manna in @xcite to the multiset of the cardinalities of algebras in @xmath1 .
we show that by decomposing finite @xmath8-algebras into their @xmath0-subdirectly irreducible components and appropriately refining the set of algebras obtained , we arrive at a ( unique up to isomorphism ) minimal set of @xmath8-algebras that still generates @xmath0 . recall that a _ multiset _ over a set @xmath78 is an ordered pair @xmath79 where @xmath80 is a function @xmath81 , and is called _ finite _ if @xmath82 is finite . for a well - ordered set @xmath83 , the _ multiset ordering @xmath84 _ on the set @xmath85 of finite multisets over @xmath78 is defined by @xmath86 if @xmath87 implies that for some @xmath88 , @xmath89 and @xmath90 .
it then follows that @xmath84 is a well - ordering of @xmath85 ( see @xcite ) . as usual
, we write a finite multiset of elements from @xmath78 as @xmath91 $ ] where @xmath92 may include repetitions . a set of finite @xmath8-algebras @xmath93
will be called a _ minimal generating set _ for the quasivariety @xmath94 if for every set of finite @xmath8-algebras @xmath95 : @xmath96 \leq_m [ |b_1| , \dots , |b_k|].\ ] ] there are of course many measures that could be used to compare generating sets .
indeed , it may seem unreasonable to view one hundred algebras with three elements as an improvement on a single algebra with four elements .
however , the order is optimal in the following sense . in general , checking a quasiequation with @xmath97 variables in a finite algebra @xmath9 requires checking @xmath98 assignments of variables to elements of @xmath9 . but
then checking validity in @xmath93 will involve checking fewer assignments of variables than checking validity in @xmath95 if @xmath99 \leq_m [ |b_1| , \dots , |b_k|]$ ] for quasiequations with sufficiently many variables . to obtain minimal generating sets for @xmath5 where @xmath1 is a finite set of finite algebras , we consider representations of the algebras in @xmath1 using smaller algebras of @xmath5 .
an algebra @xmath9 is called a _
subdirect product _ of algebras @xmath34 if there exist surjective homomorphisms @xmath100 for @xmath101 such that the induced homomorphism @xmath102 is an embedding . in this case
, @xmath80 is called a _ subdirect representation of @xmath9 _ and @xmath34 are called _ subdirect components _ ( for this representation ) of @xmath9 .
if @xmath0 is a quasivariety and @xmath103 , then @xmath9 is called a _ @xmath0-subdirect product of @xmath104 _ and @xmath80 is called a _ @xmath0-subdirect embedding_. @xmath9 is called _
@xmath0-subdirectly irreducible _ if for every @xmath0-subdirect embedding @xmath105 , @xmath9 is isomorphic to @xmath106 for some @xmath107 .
we will make essential use of the following : [ lem : qdecomp ] let @xmath0 be a quasivariety and @xmath108 .
then @xmath9 is a @xmath0-subdirect product of @xmath0-subdirectly irreducible members of @xmath0 .
[ lem : irr_emb ] if @xmath77 for a finite set @xmath1 of finite algebras and @xmath9 is a @xmath0-subdirectly irreducible algebra , then @xmath109 .
observe now that if @xmath0 is a quasivariety and @xmath108 is a @xmath0-subdirect product of @xmath104 , then each @xmath106 is in @xmath0 by definition , and @xmath9 is isomorphic to a subalgebra of a product of members of @xmath0 . hence we obtain : [ lem : reducegens ] let @xmath1 be a class of @xmath8-algebras and suppose that @xmath110 is obtained from @xmath1 by either ( a ) replacing @xmath37 with @xmath34 where @xmath9 is a @xmath5-subdirect product of @xmath104 , or ( b ) replacing @xmath111 with @xmath112 where @xmath113
. then @xmath114 . in particular , replacing each algebra @xmath9 in a finite set @xmath1 of finite algebras with the @xmath5-subdirectly irreducible algebras in some @xmath5-subdirect representation of @xmath9 , then removing any algebra that embeds into another algebra in the set , produces a minimal generating set for @xmath5 that is unique up to isomorphism .
[ t : mingenset ] suppose that @xmath115 where @xmath106 is a finite @xmath0-subdirectly irreducible algebra for @xmath116 and @xmath117 for @xmath118 .
then @xmath93 is the unique minimal generating set for @xmath0 up to isomorphism .
let @xmath115 where @xmath106 is a finite @xmath0-subdirectly irreducible algebra for @xmath116 and @xmath117 for @xmath118 .
suppose for a contradiction that @xmath119 and @xmath120 < _ m [ |a_1| , \dots , |a_n|]$ ] . without loss of generality , we can suppose that @xmath121 is @xmath0-subdirectly irreducible for @xmath122 ; otherwise , by lemmas [ lem : qdecomp ] and [ lem : reducegens ] , @xmath121 can be replaced with the @xmath0-subdirectly irreducible components of a @xmath0-subdirect representation of @xmath121 and we obtain a smaller ( according to @xmath123 ) generating set of algebras for @xmath0 .
it follows that there exists a largest @xmath124 such that there are strictly more occurrences of @xmath97 in @xmath125 $ ] than in @xmath120 $ ] , and for each @xmath126 , the number of occurrences of @xmath127 in @xmath125 $ ] and @xmath120 $ ] are equal .
each @xmath106 is finite and @xmath0-subdirectly irreducible , and hence by lemma [ lem : irr_emb ] , embeds into some @xmath121 where @xmath128 . if every @xmath106 of size @xmath97 embeds into , and is hence isomorphic to , a @xmath121 of size @xmath97 , then ( by the pigeonhole principle ) there must be two isomorphic algebras in @xmath129 , a contradiction
hence , suppose without loss of generality that @xmath130 embeds into @xmath131 with @xmath132 and @xmath133 .
but notice now that @xmath131 is also @xmath0-subdirectly irreducible and hence embeds into some @xmath106 with @xmath134 .
so @xmath135 , a contradiction .
finally , consider any minimal generating set @xmath136 for @xmath0 , and suppose for a contradiction that @xmath137 for some @xmath138 .
then by lemma [ lem : irr_emb ] , @xmath139 properly embeds into @xmath140 for some @xmath141 .
but also by lemma [ lem : irr_emb ] , @xmath140 embeds into @xmath142 for some @xmath143 .
it follows that @xmath139 can be embedded into the strictly larger algebra @xmath142 .
but then @xmath136 is not a minimal generating set for @xmath0 , a contradiction . in the remainder of this section
, we develop some results for @xmath0-subdirect products and @xmath0-subdirectly irreducible algebras , and present an algorithm for obtaining minimal generating sets for finitely generated quasivarieties .
first , let us recall the following theorem of birkhoff , which establishes a useful relationship between subdirect representations of a given algebra @xmath9 and sets of congruences on @xmath9 .
[ lem : correspondence ] if @xmath9 is a subdirect product of the family @xmath144 , then there exist for @xmath145 , congruences @xmath146 such that @xmath147 and @xmath148 .
conversely , for any family of congruences @xmath149 on @xmath9 , the quotient @xmath150 is a subdirect product of the family @xmath151 . the set of _
@xmath0-congruences on @xmath9 _ is defined as @xmath152 .
clearly , the above lemma also holds for @xmath0-subdirect representations of an algebra @xmath9 with respect to @xmath0-congruences .
note , moreover , that the number of congruences needed to obtain a subdirect representation of a finite algebra @xmath9 is at most @xmath153 , the maximal number of coatoms of the congruence lattice @xmath16 .
[ cor : correspondence ] let @xmath0 be a quasivariety and @xmath108
. 1 . if @xmath9 is a @xmath0-subdirect product of the family @xmath144 , then there exist for @xmath145 , @xmath0-congruences @xmath154 such that @xmath147 and @xmath148 .
conversely , for any family of @xmath0-congruences @xmath149 on @xmath9 with @xmath148 , @xmath9 is a @xmath0-subdirect product of the family @xmath151 .
@xmath9 is @xmath0-subdirectly irreducible iff the bottom element @xmath155 of @xmath156 is meet - irreducible ( i.e. , if @xmath148 for @xmath157 , then @xmath158 for some @xmath145 ) .
the problem of finding the congruence closure for a given equivalence relation on a finite algebra , i.e. , the smallest congruence containing this equivalence , can be solved in polynomial time .
this result was used in @xcite to provide a polynomial time algorithm for calculating a subdirect representation of a finite algebra .
the problem of finding the @xmath0-congruence closure of an equivalence relation on a finite algebra with respect to a finitely generated quasivariety @xmath0 appears to be much harder , however .
instead , we use here the following characterization of @xmath0-subdirectly irreducible algebras as the basis for a procedure that constructs @xmath0-subdirectly irreducible components for a @xmath0-subdirect representation of a given finite algebra without needing to calculate the @xmath0-congruence lattice .
[ lem : charqsubdir ] for a finite set @xmath1 of finite algebras and @xmath159 , the following are equivalent : 1 .
@xmath9 is @xmath5-subdirectly irreducible .
2 . @xmath160 . for convenience ,
let @xmath161 ( 1)@xmath162(2 ) we proceed contrapositively . if @xmath163 , then by corollary [ cor : correspondence](a ) , @xmath9 is a @xmath5-subdirect product of algebras in @xmath164 . but also by corollary [ cor : correspondence](b ) , since @xmath165 , @xmath9 is not @xmath5-subdirectly irreducible .
( 2)@xmath162(1 ) again , we proceed contrapositively .
if @xmath9 is not @xmath5-subdirectly irreducible , then combining lemma [ lem : qdecomp ] and corollary [ cor : correspondence ] , there exist @xmath166 such that @xmath148 and @xmath9 is a @xmath5-subdirect product of @xmath5-subdirectly irreducible algebras @xmath167 ( @xmath145 ) .
but then also by lemma [ lem : irr_emb ] , we have @xmath168 for each @xmath145 .
so @xmath169 , and hence @xmath163 .
* declare * @xmath170 : set * declare * @xmath171 : list * declare * @xmath9 : algebra * declare * @xmath172 : integer @xmath173 @xmath174 @xmath175 $ ] @xmath176 [ alg : mingenset : con ] @xmath177 embeds into @xmath178 [ alg : mingenset : s1 ] @xmath179 embeds into some @xmath180 \neq { { \ensuremath{{\mathbf{a}}}}}\}$ ] remove @xmath9 from @xmath181 [ alg : mingenset : dela ] @xmath182 [ alg : mingenset : e1 ] remove @xmath9 from @xmath181 [ alg : mingenset : e2 ] we now have all the ingredients necessary to describe an algorithm mingenset ( see figure [ alg : mingenset ] ) that calculates the ( unique up to isomorphism ) minimal generating set for a quasivariety @xmath77 , where @xmath1 is a finite set of finite @xmath8-algebras . by theorem [ t : mingenset ] , it suffices to find a set of @xmath0-subdirectly irreducible algebras that generates @xmath0 , where no member of the set embeds into another member of the set .
the algorithm proceeds by considering each @xmath37 in turn .
first , the congruence lattice @xmath16 is generated ( line [ alg : mingenset : con ] ) by checking for all equivalence relations if they are congruences .
next , the congruences @xmath183 such that @xmath20 embeds into @xmath9 or some other member of @xmath1 are collected in sets @xmath184 and @xmath185 , respectively .
if @xmath186 , then @xmath9 is @xmath0-subdirectly irreducible by lemma [ lem : charqsubdir ] , so the algorithm proceeds to the next algebra in @xmath1 .
otherwise @xmath187 and by lemma [ lem : charqsubdir ] , @xmath9 is not @xmath0-subdirectly irreducible . in this case , for each @xmath188 , the algebra @xmath20 is added to @xmath1 ( line [ alg : mingenset : addquot ] ) and @xmath9 is removed from @xmath1 ( line [ alg : mingenset : dela ] ) .
note that since the cardinalities of the added algebras are strictly smaller than the cardinality of the removed algebra , the new set of algebras is smaller according to the multiset ordering defined in section [ s : fingen ] .
hence this procedure is terminating .
moreover , the resulting finite set of finite algebras must generate the quasivariety @xmath0 ( by lemma [ lem : reducegens ] ) , contain only @xmath0-subdirectly irreducible algebras , and not contain any algebra that embeds into another member of the set ( lines [ alg : mingenset : e1][alg : mingenset : e2 ] ) .
hence by theorem [ t : mingenset ] , we obtain : for a finite set @xmath1 of finite @xmath8-algebras , mingenset(@xmath1 ) returns the ( unique up to isomorphism ) minimal generating set for the quasivariety @xmath5 .
we remark that although the algorithm mingenset does not need to calculate the @xmath0-congruence lattice , already calculating the congruence lattice of a finite algebra can take exponential time .
moreover , we make frequent use in our algorithm here of checking embeddings , which is in general an np - complete problem ( see @xcite ) .
let us fix a class of @xmath8-algebras @xmath1 .
a homomorphism @xmath189 is called a _
@xmath1-unifier _ of a set @xmath41 of @xmath8-equations if @xmath190 for every @xmath191 ; in this case , @xmath41 is said to be _
@xmath1-unifiable_. an @xmath8-quasiequation @xmath192 is _
@xmath1-admissible _ if every @xmath1-unifier of @xmath41 is a @xmath1-unifier of @xmath2 . in this section ,
we present characterizations of @xmath1-unifiability , @xmath1-admissibility , and related properties , emphasizing their close relationship to free algebras .
the characterizations obtained will then be used in section [ s : algorithms ] to develop algorithms for checking these properties in the context of finitely generated quasivarieties .
let us first take a closer look at @xmath1-unifiability , noting that a finite set @xmath41 of @xmath8-equations is @xmath1-unifiable if and only if the @xmath8-negative clause @xmath193 is not @xmath1-admissible ( equivalently , when @xmath1 contains a non - trivial algebra , if and only if the @xmath8-quasiequation @xmath194 with @xmath195 not occurring in @xmath41 is not @xmath1-admissible ) .
we will see that for checking @xmath1-unifiability , the `` optimal '' solution is to check satisfiability in the smallest finite subalgebra ( if such an algebra exists ) of @xmath196 .
[ l : unif ] the following are equivalent for any class @xmath110 of @xmath8-algebras : 1 .
@xmath41 is @xmath1-unifiable iff @xmath41 is @xmath110-satisfiable .
2 . @xmath197 .
recall that @xmath197 is equivalent to the condition that an @xmath8-negative clause @xmath193 is @xmath110-valid iff it is @xmath196-valid .
however , @xmath193 is @xmath110-valid iff @xmath41 is not @xmath110-satisfiable and @xmath193 is @xmath196-valid iff @xmath41 is not @xmath196-satisfiable . for the equivalence of ( 1 ) and ( 2 )
, it suffices therefore to show that @xmath41 is @xmath196-satisfiable iff @xmath41 is @xmath1-unifiable .
suppose first that @xmath198 satisfies @xmath41 .
then any homomorphism @xmath199 defined such that @xmath200 for each variable @xmath201 is a @xmath1-unifier of @xmath41 .
conversely , if @xmath202 is a @xmath1-unifier of @xmath41 , then the homomorphism @xmath198 defined by @xmath203 for each variable @xmath201 satisfies @xmath41 .
[ p : unif ] let @xmath1 be a class of @xmath8-algebras and @xmath204 . 1 .
@xmath41 is @xmath1-unifiable iff @xmath41 is @xmath205-satisfiable .
2 . if @xmath205 is a smallest finite subalgebra of @xmath196 and @xmath110 is a class of @xmath8-algebras such that @xmath41 is @xmath1-unifiable iff @xmath41 is @xmath110-satisfiable , then @xmath206 for each @xmath207 .
\(a ) by assumption , @xmath208 , so @xmath209 . but also , since @xmath210 and @xmath211 has the universal mapping property for @xmath44 over countably infinitely many generators , we obtain a homomorphism @xmath212 defined by @xmath213 for every variable @xmath201 for some fixed @xmath214 . hence @xmath215 $ ]
is a subalgebra of @xmath205 and @xmath216 .
so @xmath217 and the result follows by lemma [ l : unif ] .
\(b ) let @xmath205 be a smallest finite subalgebra of @xmath196 and suppose that @xmath110 is a class of @xmath8-algebras such that @xmath41 is @xmath1-unifiable iff @xmath41 is @xmath110-satisfiable .
then by lemma [ l : unif ] and part ( a ) , @xmath218 . hence if @xmath219 , then clearly @xmath206 .
[ ex : demorganuni ] the variety @xmath220 of de morgan algebras is generated as a quasivariety by the @xmath221-element algebra @xmath222 consisting of a distributive bounded lattice with an involutive negation defined as follows : ( -4,-0.5)(4,3 ) ( 0,0)(1,1 ) ( 1,1)(0,2 ) ( 0,0)(-1,1 ) ( -1,1)(0,2 ) ( 0,0)(-1,1)(1,1)(0,2 ) ( 0,-0.3)@xmath223 ( -1.2,1)@xmath224 ( 1.2,1)@xmath225 ( 0,2.3)@xmath226 ( 1.3,1.2)(1.5,1.4)(1.7,1)(1.5,0.6)(1.3,0.8 ) ( -1.3,1.2)(-1.5,1.4)(-1.7,1)(-1.5,0.6)(-1.3,0.8 ) ( 0,0.2)(0,1.8 ) since there are constants in the language of @xmath227 , the smallest algebra for checking @xmath220-unifiability is the @xmath228-element ground algebra @xmath229 : i.e. , the @xmath228-element boolean algebra .
that is , checking unifiability amounts to checking classical satisfiability .
e.g. , @xmath230 is not @xmath220-unifiable , since in the @xmath228-element boolean algebra , @xmath231 and @xmath232 .
the case of the `` constant - free '' variety @xmath233 of de morgan lattices , generated as a quasivariety by @xmath234 , is not so immediate .
however , there is also a smallest @xmath228-element subalgebra of @xmath235 with elements corresponding to @xmath236 and @xmath237 .
so checking @xmath233-unifiability amounts again to checking classical satisfiability .
let us turn our attention now to admissible quasiequations , recalling the following useful characterizations of admissibility in terms of free algebras and generated subvarieties .
[ t : eqadm ] the following are equivalent : 1 .
@xmath192 is @xmath1-admissible .
@xmath192 is @xmath5-admissible .
3 . @xmath238 .
4 . @xmath239 .
[ ex : lattices ] the following quasiequations , expressing meet and join semi - distributivity in the language of lattices with @xmath240 and @xmath241 , are satisfied by all free lattices ( see @xcite ) , and are therefore admissible in the variety of lattices : @xmath242 \{x \lor y { \approx}x \lor z\ } & { \rightarrow } & x \lor y { \approx}x \lor ( y \land z ) .
\end{array}\ ] ] similarly , the quasiequations below expressing torsion - freeness in a language of groups with @xmath243 , @xmath244 , and @xmath245 are satisfied by all free groups , and are hence admissible in the variety of groups : @xmath246 given a class @xmath1 of @xmath8-algebras , we will be interested in determining when the @xmath1-admissibility of quasiequations coincides with their @xmath110-validity in another class of @xmath8-algebras @xmath110 . by theorem [ t : eqadm ] , this is the case exactly when @xmath247 . the next result provides a further useful characterization of this situation .
[ p : adm ] the following are equivalent : 1 .
@xmath192 is @xmath1-admissible iff @xmath248 .
2 . @xmath247 .
3 . @xmath249 and @xmath250 .
\(1 ) @xmath251 ( 2 ) follows directly from theorem [ t : eqadm ] .
\(2 ) @xmath162
( 3 ) suppose that @xmath247
. then @xmath249 .
moreover , @xmath252 , so @xmath250 .
\(3 ) @xmath162
( 2 ) suppose that @xmath249 and @xmath250
. then clearly @xmath253 .
but also @xmath254 . that is ,
hence @xmath256 and @xmath257 . for some well - behaved classes of algebras ,
admissibility and validity coincide .
more precisely , a class @xmath1 of @xmath8-algebras is said to be _ structurally complete _ if it satisfies the condition : @xmath258 the following characterization then follows almost immediately from theorem [ t : eqadm ] : [ p : bergman ] the following are equivalent : 1 .
@xmath1 is structurally complete .
2 . @xmath259 .
3 . @xmath260 and @xmath261 implies @xmath262 .
the variety @xmath263 of boolean algebras , generated as a quasivariety by the @xmath228-element algebra @xmath264 , is structurally complete .
it suffices to observe that @xmath265 embeds into @xmath266 via a mapping that sends @xmath267 and @xmath268 to their respective equivalence classes , and hence that @xmath269 .
[ ex : m5n5 ] a _ modular lattice _ @xmath270 may be characterized as a lattice satisfying the equation @xmath271 .
famously , a lattice @xmath270 is non - modular if and only if the lattice @xmath272 ( below ) embeds into @xmath270 ( see ( * ? ? ?
* theorem i.3.5 ) ) .
but since @xmath272 is non - modular , also @xmath273 ( which must satisfy the same equations ) is non - modular .
so @xmath272 embeds into @xmath273 , and @xmath274 is structurally complete .
similarly , it is well - known that a lattice @xmath275 is distributive if and only if neither @xmath272 nor @xmath276 ( below ) embeds into @xmath270 ( see ( * ? ? ?
* theorem i.3.6 ) ) .
since @xmath276 is non - distributive and modular , also @xmath277 is non - distributive and modular .
so @xmath276 embeds into @xmath277 , and @xmath278 is structurally complete .
( 0,-2)(5.94,1.5 ) ( 0.86,1.2 ) ( 0.06,0.4 ) ( 0.06,-0.4 ) ( 0.86,-1.2 ) ( 1.66,0.0 ) ( 4.66,0.0 ) ( 4.66,1.2 ) ( 4.66,-1.2 ) ( 5.86,0.0 ) ( 3.46,0.0 ) ( 0.06,0.4)(0.86,1.2 ) ( 1.66,0.0)(0.86,1.2 ) ( 0.06,0.4)(0.06,-0.4 ) ( 0.06,-0.4)(0.86,-1.2 ) ( 0.86,-1.2)(1.66,0.0 ) ( 3.46,0.0)(4.66,-1.2 ) ( 4.66,-1.2)(5.86,0.0 ) ( 5.86,0.0)(4.66,1.2 ) ( 4.66,1.2)(3.46,0.0 ) ( 4.66,-1.2)(4.66,0.0 ) ( 4.66,0.0)(4.66,1.2 ) ( 0.9,-1.6)@xmath272 ( 4.7,-1.6)@xmath276 for certain other classes , admissibility and validity coincide for quasiequations with unifiable premises .
more precisely , we call a class @xmath1 of @xmath8-algebras _ almost structurally complete _ if it satisfies the condition : @xmath279 the next result provides a useful characterization of this situation : [ thm : asc ] the following are equivalent for any @xmath280 : 1 .
@xmath1 is almost structurally complete .
2 . @xmath281 .
3 . @xmath282 .
\(1 ) @xmath162
( 2 ) suppose that @xmath1 is almost structurally complete . to establish @xmath281 , it suffices to show that a quasiequation @xmath192 is valid in all algebras @xmath283 for @xmath37 iff it is valid in @xmath196 .
suppose first that @xmath284 .
then by theorem [ t : eqadm ] , either @xmath41 is not @xmath1-unifiable or @xmath192 is @xmath1-valid . in the first case , by proposition [ p : unif ] ,
@xmath41 is not @xmath112-satisfiable , so @xmath192 is valid in @xmath283 for all @xmath37 . in the second case
, @xmath192 is valid in @xmath285 for all @xmath37 .
conversely , if @xmath192 is valid in @xmath283 for each @xmath37 , then either @xmath41 is not @xmath112-satisfiable or @xmath192 is valid in each @xmath9 in @xmath1 . in the first case , by proposition [ p : unif ] ,
@xmath41 is not @xmath1-unifiable , so @xmath192 is valid in @xmath196 . in the second case , @xmath192 is valid in @xmath5 and hence valid in @xmath196 .
\(2 ) @xmath162 ( 1 ) suppose that @xmath286 .
then whenever @xmath192 is @xmath1-admissible , it is @xmath196-valid and hence also valid in @xmath283 for all @xmath37 .
if @xmath41 is @xmath1-unifiable , then , by proposition [ p : unif ] , it is @xmath112-satisfiable .
i.e. , there exists a homomorphism @xmath287 with @xmath288 . for any @xmath37 and homomorphism @xmath289 with @xmath290 , define @xmath291 by @xmath292 .
then , since @xmath192 is valid in @xmath283 for all @xmath37 , @xmath293 , so @xmath294 and @xmath295
. i.e. , @xmath296 .
so we have shown that @xmath297 .
\(2 ) @xmath162 ( 3 ) immediate .
\(3 ) @xmath162
( 2 ) suppose that @xmath282 .
then also , since @xmath298 for each @xmath37 , we obtain @xmath299 . hence by proposition [ p :
adm ] , @xmath281 .
[ ex : luk ] consider the @xmath228-element and @xmath300-element wajsberg algebras @xmath301 and @xmath302 where @xmath303 the algebra @xmath304 embeds into @xmath305 , as illustrated in the diagram below by the terms associated to elements , and has @xmath306 as a homomorphic image , as indicated by the arrows . hence @xmath306 is almost structurally complete .
note that it is not structurally complete since , for example , @xmath307 is @xmath306-admissible , but not @xmath306-valid . on the other hand ,
its implicational reduct @xmath308 is structurally complete , since it embeds into the free algebra @xmath309 .
+ ( -5,-1)(10,2.5 ) ( 0,0)(3,1)(0,3)(3,4)(0,6)(3,7)(21,0.5)(21,3.5)(21,6.5 ) ( -1,0)@xmath310 ( -1,3)@xmath311 ( -1,6)@xmath312 ( 4,1)@xmath313 ( 4,4)@xmath314 ( 4,7)@xmath315 ( 22,0.5)@xmath316 $ ] ( 22,3.5)@xmath317 $ ] ( 22,6.5)@xmath318 $ ] ( 0,0)(3,1 ) ( 0,3)(3,4 ) ( 0,6)(3,7 ) ( 0,0)(0,6 ) ( 3,1)(3,7 ) ( 21,0.5)(21,6.5 ) ( 13,0.5)(18,0.5 ) ( 13,3.5)(18,3.5 ) ( 13,6.5)(18,6.5 ) ( -7,-1)(11,-1 ) ( -7,2)(11,2 ) ( -7,5)(11,5 ) ( -7,8)(11,8 ) ( -7,-1)(-7,8 ) ( 11,-1)(11,8 ) ( -7,-2.5)@xmath319
[ s : algorithms ] in this section , we present algorithms for checking admissibility of quasiequations and related properties in finitely generated quasivarieties .
we also present results on admissibility ( some new , some well - known ) for a range of examples taken from the universal algebra and non - classical logic literature . these results , obtained using a delphi implementation of the algorithms described here , are collected at the end of this section in table [ table ] .
we begin by recalling birkhoff s result that the finitely generated free algebras for a finite set @xmath1 of finite algebras are themselves finite ; more precisely : [ l : finite ] for any set of finite algebras @xmath320 and @xmath321 : @xmath322 hence for the relatively easy task of checking @xmath1-unifiability , it suffices to find ( e.g. , through exhaustive search ) the ( unique up to isomorphism ) smallest subalgebra @xmath205 of the finite free algebra @xmath323 , noting that this is @xmath324 if the language contains constants .
it then follows by proposition [ p : unif ] that a set of equations @xmath41 is @xmath1-unifiable if and only if @xmath41 is @xmath205-valid , and indeed that there is no smaller algebra with this property . for checking @xmath1-admissibility
, we make use of a known result for finitely generated quasivarieties ( see ( * ? ? ?
* lemma 4.1.10 ) ) , obtained here as a corollary of proposition [ p : adm ] : [ c : admfin ] given a finite set @xmath1 of finite @xmath8-algebras with @xmath325 : 1 .
@xmath192 is @xmath1-admissible iff @xmath327 .
observe first that each @xmath37 is a homomorphic image of @xmath7 .
that is , define any surjective map from the @xmath6 generators of @xmath7 to @xmath10 ; this extends to a homomorphism from @xmath7 onto @xmath9 since @xmath7 has the universal mapping property for @xmath1 over @xmath6 generators .
so @xmath328 and , since also @xmath329 , ( a ) and ( b ) follow by proposition [ p : adm ] .
hence checking @xmath1-admissibility of quasiequations is decidable . however , even when @xmath1 consists of a small number of small algebras , free algebras on a small number of generators can be quite large .
for example , the free algebra @xmath330 ( see example [ ex : demorganuni ] ) has @xmath331 elements .
we therefore seek smaller algebras or finite sets of smaller algebras that also generate @xmath332 as a quasivariety .
in fact , since @xmath333 is finitely generated , we may apply the multiset ordering defined in section [ s : fingen ] , and seek a minimal generating set of finite algebras for this quasivariety that is unique up to isomorphism .
one strategy would therefore be to apply the algorithm mingenset directly to @xmath334 . however , this method is not feasible for large free algebras , since it involves the computationally labour - intensive task of building the congruence lattice of @xmath334 .
instead , we make auxiliary use of the following immediate corollary of proposition [ p : adm ] : [ c : adm ] given a class @xmath1 of @xmath8-algebras and @xmath335 such that @xmath336 : 1 .
@xmath192 is @xmath1-admissible iff @xmath248 . *
declare * @xmath337 : set * declare * @xmath338 : algebra @xmath339 @xmath340 @xmath341 @xmath342 [ alg : subprehom ] @xmath343 @xmath342 add @xmath112 to @xmath344 we describe an algorithm admalgs ( see figure [ alg : admalgs ] ) which takes as input a finite set @xmath1 of finite @xmath8-algebras and outputs the ( unique up to isomorphism ) minimal generating set for @xmath332 .
first , the procedure mingenset is applied to @xmath1 ( which typically would be a small set of small algebras ) to obtain the set of @xmath8-algebras @xmath345 .
then for each @xmath346 , a procedure free(@xmath347 ) is invoked to produce the smallest free algebra @xmath348 having @xmath9 as a homomorphic image .
( the procedure begins by checking the smallest free algebra @xmath349 or @xmath350 , then increases the number of generators one at a time . )
the algorithm then searches for progressively smaller subalgebras of @xmath348 which have @xmath9 as a homomorphic image .
more precisely , the procedure @xmath351 searches for a proper subalgebra of @xmath112 that is a homomorphic image of @xmath9 , returning @xmath112 if no such algebra exists ( line [ alg : subprehom ] ) .
this process terminates with a ( hopefully reasonably small ) algebra which is added to a set @xmath352 .
finally , the procedure mingenset is applied to @xmath352 .
for a finite set @xmath1 of finite @xmath8-algebras , admalgs(@xmath1 ) returns the ( unique up to isomorphism ) minimal generating set for the quasivariety @xmath333 .
[ ex : no_embedding ] note that given even just one algebra @xmath9 , it might not be the case that the smallest subalgebra of @xmath353 having @xmath9 as a homomorphic image is the smallest algebra that generates @xmath354 . consider the @xmath221-element algebra @xmath355 where the unary function @xmath12 and the free algebras @xmath356 are described by the following diagrams : + ( -1.5,0)(3,3 ) ( -4,7)@xmath224 ( -7.2,7)@xmath357 ( -4.2,6.5)(-4.2,4.5 ) ( -3.8,6.5)(-3.8,4.5 ) ( -4,4)@xmath225 ( -6.5,1.5)(-4.5,3.5 ) ( -7,1)@xmath358 ( -1.5,1.5)(-3.5,3.5 ) ( -1,1)@xmath359 ( 1.5,7)@xmath356 ( 2,5)@xmath360 ( 3,5)(4.5,5 ) ( 6,5)@xmath361 ( 7.5,5.2)(9,5.2 ) ( 7.5,4.8)(9,4.8 ) ( 9.5,5)@xmath362 ( 6,2.5)(6,3)(6,3.5 ) ( 2,1)@xmath363 ( 3,1)(4.5,1 ) ( 6,1)@xmath364 ( 7.5,1.2)(9,1.2 ) ( 7.5,0.8)(9,0.8 ) ( 9.5,1)@xmath365 the smallest subalgebra of @xmath366 with @xmath357 as a homomorphic image is the @xmath367-element free algebra @xmath368 .
however , @xmath369 , performed at the beginning of the algorithm admalgs , discovers that @xmath357 is a @xmath370-subdirect product of ( two copies of ) the @xmath300-element free algebra @xmath371 .
hence @xmath372 and this quasivariety is structurally complete . in the previous example , the fact that the algebra @xmath357 is structurally complete is discovered by the algorithm admalgs .
however , structural completeness can also be checked more directly using the following result : [ prop : charsc ] the following are equivalent for any finite set @xmath1 of finite @xmath8-algebras : 1 .
@xmath1 is structurally complete .
2 . @xmath373 where @xmath374 .
\(1 ) @xmath162 ( 2 ) if @xmath1 is structurally complete , then , by proposition [ p : bergman ] and corollary [ c : admfin ] , @xmath375 where @xmath374 .
so @xmath376 . but
each @xmath377 is @xmath378-subdirectly irreducible , so by lemma [ lem : irr_emb ] , @xmath9 embeds into @xmath7
. i.e. , @xmath373 .
\(2 ) @xmath162 ( 1 ) if each @xmath377 embeds into @xmath7 , then @xmath379 . so @xmath1 is structurally complete by proposition [ p : bergman ] .
[ ex : pseudocomp ] proposition [ prop : charsc ] has been used to confirm known structural completeness results for the @xmath300-element positive gdel algebra @xmath380 where @xmath381 is @xmath382 if @xmath383 , otherwise @xmath268 , and the stone algebra @xmath384 where @xmath385 . moreover
, a new structural completeness result has been established for the pseudocomplemented distributive lattice @xmath386 , obtained by adding a top element to the @xmath221-element boolean lattice , and calculating the pseudocomplement .
similarly , we can check almost structural completeness using the following result : [ prop : almost ] the following are equivalent for any finite set @xmath1 of finite @xmath8-algebras and @xmath280 : 1 .
@xmath1 is almost structurally complete .
2 . @xmath387 where @xmath374 .
\(1 ) @xmath162 ( 2 ) if @xmath1 is almost structurally complete , then by theorem [ thm : asc ] and corollary [ c : admfin ] , @xmath388 where @xmath389 .
in particular , @xmath390 .
but each @xmath391 is @xmath378-subdirectly irreducible , so by lemma [ lem : irr_emb ] , @xmath205 embeds into @xmath7 .
i.e. , @xmath387 .
\(2 ) @xmath162 ( 1 ) if @xmath387 , then @xmath392 .
so by theorem [ thm : asc ] , @xmath1 is almost structurally complete .
[ ex : sugimono3 ] the sobociski algebra @xmath393 with operations @xmath394 is a weak characteristic matrix for the multiplicative fragment @xmath395 of the substructural logic r - mingle : @xmath314 is a theorem of @xmath395 if and only if @xmath396 . using proposition [ prop : almost ]
, we establish that @xmath397 is almost structurally complete and that @xmath398 where @xmath399 .
note , moreover , that the implicational reduct @xmath400 of @xmath397 is structurally complete , which can be confirmed directly using proposition [ prop : charsc ] or discovered automatically by admalgs .
[ ex : sugimono4 ] consider now the algebra @xmath401 ( the reduct of a @xmath221-element algebra for r - mingle ) with a constant @xmath268 and operations @xmath402 admalgs produces a @xmath367-element algebra ( the product of a @xmath228-element algebra and a @xmath300-element algebra ) .
remarkably , if we drop @xmath403 to obtain @xmath404 , admalgs produces a @xmath221-element algebra that is not isomorphic to @xmath405 .
[ ex : demorgankleene ] recall from example [ ex : demorganuni ] that the algebras @xmath406 and @xmath227 generate the varieties of de morgan latices and de morgan algebras , respectively , as quasivarieties .
both @xmath406 and @xmath227 are homomorphic images of the corresponding free algebras on two generators ( with @xmath407 and @xmath331 elements , respectively ) but not on one generator . in the de morgan lattice case , admalgs finds a smallest suitable subalgebra isomorphic to @xmath408 , while for de morgan algebras , the smallest suitable algebra is isomorphic to @xmath409 with additional top and bottom elements .
these results were established `` by hand '' in @xcite ; our procedure here confirms them automatically .
similar results were also obtained in @xcite for kleene lattices and kleene algebras ( subvarieties of de morgan lattices and de morgan algebras , respectively ) which are generated as quasivarieties by the @xmath300-element chains @xmath410 and @xmath411 where @xmath403 swaps @xmath223 and @xmath226 and leaves @xmath412 fixed . in both cases , the smallest algebra for @xmath413 and @xmath414 , found automatically by admalgs is a @xmath221-element chain .
[ e : twoalgebras ] consider the @xmath228-element and @xmath300-element chains , @xmath415 and @xmath416 where @xmath403 swaps @xmath223 and @xmath226 and leaves @xmath412 fixed .
individually , these algebras are structurally complete . however , applying admalgs to @xmath417 , we find that @xmath1 is not structurally complete : both @xmath418 and @xmath419 are homomorphic images of the @xmath420-element free algebra @xmath421 , and the minimal generating set for @xmath333 consists of a single @xmath221-element algebra .
we remark finally that except for example [ e : twoalgebras ] , all the case studies considered in this section ( see table [ table ] ) involve quasivarieties generated by a single algebra with five elements or fewer , and produce minimal generating sets for checking admissibility also containing just one algebra . indeed in many of the cases considered , the quasivariety is shown to be either structurally complete or almost structurally complete .
the main obstacle at present to extending our experimental results to ( larger sets of ) larger algebras is the size of the resulting free algebras ( in the worst case , size @xmath422 for @xmath6 generators and an algebra of size @xmath423 ) , both for computing these algebras and then for generating suitable subalgebras .
nevertheless , we hope in future work to obtain procedures capable of treating larger examples by using heuristics to constrain the search for appropriate generating algebras .
.algebras for checking admissibility [ cols="^,^,<,<,^",options="header " , ]
[ s : logics ] the preceding characterizations of admissibility can be adapted relatively straightforwardly to finite - valued logics . however , in this setting , the designated values of the logic the `` true truth values '' must also be considered , and dealt with appropriately . here
we describe a method that given a finite - valued logic @xmath275 , provides another ( hopefully small ) finite - valued logic @xmath424 such that validity in @xmath424 corresponds to admissibility in @xmath275 .
the more general case , where a smallest finite set of logics is sought such that validity in all members of the set corresponds to admissibility in a logic ( or logics ) , is left as an exercise for the interested reader .
recall that a _ finite - valued logic _ @xmath425 for a language @xmath8 consists of a finite @xmath8-algebra @xmath9 and a set of _ designated values _ @xmath426 . given @xmath427
, we let @xmath428 denote that for all homomorphisms @xmath38 , whenever @xmath429 \subseteq d$ ] , also @xmath430 .
a term @xmath314 is _
@xmath275-valid _ if @xmath431
. we call a logic @xmath432 for a language @xmath8 a _ sublogic _ of a logic @xmath433 if @xmath130 is a subalgebra of @xmath434 and also @xmath435 .
consider now a finite - valued logic @xmath425 for a language @xmath8 and a finite set of terms @xmath427 .
we say that @xmath436 is _
@xmath275-unifiable _ if there exists a homomorphism @xmath437 such that @xmath438 for all @xmath439 and call @xmath3 in this case an _ @xmath275-unifier _ of @xmath436 .
a rule @xmath440 is said to be _
@xmath275-admissible _ if every @xmath275-unifier of @xmath436 is an @xmath275-unifier of @xmath314 .
now if we define the finite - valued logic @xmath441 where @xmath442 \in { { \ensuremath{{\mathbf{f}}}}}_{{\ensuremath{{\mathbf{a}}}}}(|a| ) : \ { \vdash_{{\mathrm{l } } } } { \ensuremath{\varphi}}\}$ ] , then we easily obtain the following analogue of theorem [ t : eqadm ] .
[ l : finval ] let @xmath425 be a finite - valued logic for a language @xmath8
. then @xmath440 is @xmath275-admissible iff @xmath443 .
the next result may then be understood as an analogue of proposition [ p : adm ] .
let @xmath444 be a finite - valued logic for a language @xmath8 , and let @xmath445 be a sublogic of @xmath446 such that there exists a surjective homomorphism @xmath447 satisfying @xmath448 \subseteq d_a$ ] .
then @xmath440 is @xmath275-admissible iff @xmath449 .
if @xmath440 is @xmath275-admissible , then by lemma [ l : finval ] , @xmath443 . since @xmath424 is a sublogic of @xmath446 , also @xmath449 .
conversely , suppose that @xmath449 and that @xmath3 is an @xmath275-unifier of @xmath436 .
notice that if @xmath450 , then @xmath451 and @xmath452 .
so @xmath3 is also an @xmath446-unifier and @xmath424-unifier of @xmath436 . but
@xmath453 and therefore @xmath454 .
now consider any homomorphism @xmath455 . since @xmath456 is a surjective homomorphism from @xmath112 to @xmath9 ,
there exists a homomorphism @xmath457 such that @xmath458 is the identity map on @xmath9 .
but @xmath454 and hence @xmath459 .
\subseteq d_a$ ] .
so @xmath461 .
the @xmath300-valued ukasiewicz logic @xmath462 and jakowski logic @xmath463 may both be presented using the @xmath300-element wajsberg algebra @xmath306 ( example [ ex : luk ] ) but with @xmath268 as designated value for @xmath462 and @xmath464 and @xmath268 as designated values for @xmath463 .
that is , @xmath465 and @xmath466 . in this case
, there is a smallest subalgebra of @xmath467 isomorphic to @xmath304 with a homomorphism that maps @xmath304 onto @xmath306 and sends the inherited designated values @xmath468 to @xmath268 , and @xmath469 to @xmath464 .
we therefore obtain a logic @xmath470 corresponding to admissibility in @xmath462 , and another logic @xmath471 corresponding to admissibility in @xmath463 .
the algorithms described in this paper have been implemented in the tool @xmath472 , a delphi xe2 implementation compiled for windows operating systems available for download online at https://sites.google.com/site/admissibility .
this tool has been used in particular to obtain minimal generating sets and to check structural completeness and almost structural completeness for all @xmath300-element groupoids ( i.e. , @xmath300-element algebras with a single binary operation ) of which there are @xmath473 up to isomorphism .
the cardinalities of the minimal free algebras required to check admissibility range from @xmath300 to @xmath474 , while the minimal generating algebras have at most @xmath475 elements .
@xmath476 of the groupoids are structurally complete and a further @xmath477 are almost structurally complete .
for further details of @xmath472 and these experiments , consult the system description @xcite .
although the algorithms described here produce minimal sets of algebras for testing admissibility ( and of course have to be run only once for a given finite set of finite algebras ) , the steps involved generating a suitable free algebra and its subalgebras and checking homomorphic images may be computationally unfeasible .
for this reason , our case studies have consisted so far mostly of algebras with @xmath478 or fewer elements where the appropriate free algebra is of a reasonable size ( less than @xmath479 elements , say ) .
this bottleneck could perhaps be addressed by introducing heuristics such as constructing and checking small subalgebras of free algebras incrementally , rather than beginning with the whole free algebra , or calculating only some and not all of the congruences in the @xmath480 algorithm .
we might also hope to obtain faster algorithms for cases where the given algebras admit certain properties such as congruence distributivity or permutability .
we recall finally that one of the motivations for investigating admissibility in arbitrary finite algebras or , similarly , finite - valued logics , is to obtain admissible quasiequations or rules that can be used to improve proof systems ( shortening derivations , constraining proof search , simplifying rules , etc . ) for these algebras or logics .
the work reported here provides a significant step towards this goal .
the next , equally ambitious , step will be to determine which rules have the potential to be useful for a given finite algebra or finite - valued logic and to add these appropriately to corresponding proof systems . | checking the admissibility of quasiequations in a finitely generated ( i.e. , generated by a finite set of finite algebras ) quasivariety @xmath0 amounts to checking validity in a suitable finite free algebra of the quasivariety , and is therefore decidable .
however , since free algebras may be large even for small sets of small algebras and very few generators , this naive method for checking admissibility in @xmath0 is not computationally feasible . in this paper , algorithms
are introduced that generate a minimal ( with respect to a multiset well - ordering on their cardinalities ) finite set of algebras such that the validity of a quasiequation in this set corresponds to admissibility of the quasiequation in @xmath0 . in particular , structural completeness ( validity and admissibility coincide ) and almost structural completeness ( validity and admissibility coincide for quasiequations with unifiable premises ) can be checked .
the algorithms are illustrated with a selection of well - known finitely generated quasivarieties , and adapted to handle also admissibility of rules in finite - valued logics . |
in the rome meeting i presented a derivation of the dynamical behavior of a beamed gamma ray burst ( grb ) remnant and its consequences for the afterglow light curve .
rhoads 1999 [ paper i ] ) . here ,
i summarize these results and apply them to test the range of beaming angles permitted by the optical light curve of grb 970508 .
suppose that ejecta from a grb are emitted with initial lorentz factor @xmath3 into a cone of opening half - angle @xmath0 and expand into an ambient medium of uniform mass density @xmath4 with negligible radiative energy losses .
let the initial kinetic energy and rest mass of the ejecta be @xmath5 and @xmath6 , and the swept - up mass and internal energy of the expanding blast wave be @xmath7 and @xmath8 .
then energy conservation implies @xmath9 so long as @xmath10 .
the swept - up mass is determined by the working surface area : @xmath11 , where @xmath12 and @xmath13 are the sound speed and time since the burst in the frame of the blast wave @xmath14 accreted material .
once @xmath15 , @xmath16 and the dynamical evolution with radius @xmath17 changes from @xmath18 to @xmath19 ( rhoads 1998 , & paper i ) .
the relation between observer frame time @xmath20 and radius @xmath17 also changes , from @xmath21 to @xmath22)$ ] .
thus , at early times @xmath23 , while at late times @xmath24 .
the characteristic length scale is @xmath25 , and the characteristic observed transition time between the two regimes is @xmath26 \right)^{1/3 } \zm^{8/3}$ ] , where @xmath27 is the burst s redshift .
we assume that swept - up electrons are injected with a power law energy distribution @xmath28 for @xmath29 , with @xmath30 , and contain a fraction @xmath31 of @xmath8 .
this power law extends up to the cooling break , @xmath32 , at which energy the cooling time is comparable to the dynamical expansion time of the remnant . above @xmath32 ,
the balance between electron injection ( with @xmath33 ) and cooling gives @xmath34 .
we also assume a tangled magnetic field containing a fraction @xmath35 of @xmath8 .
the comoving volume @xmath36 and burster - frame volume @xmath37 are related by @xmath38 , so that @xmath39 and @xmath40 .
the resulting spectrum has peak flux density @xmath41 at an observed frequency @xmath42 .
additional spectral features occur at the frequencies of optically thick synchrotron self absorption ( which we shall neglect ) and the cooling frequency @xmath43 ( which is important for optical observations of grb 970508 ) .
the cooling break frequency follows from the relations @xmath44 ( sari , piran , & narayan 1998 ; wijers & galama 1998 ) and @xmath45 . in the power law regime , @xmath46 , @xmath47 , and @xmath48 ; while in the exponential regime , @xmath49 , @xmath50 , and @xmath51 .
the spectrum is approximated by a broken power law , @xmath52 , with @xmath53 for @xmath54 , @xmath55 for @xmath56 , and @xmath57 for @xmath58 .
the afterglow light curve follows from the spectral shape and the time behavior of the break frequencies .
asymptotic slopes are given in table 1 .
for the @xmath59 regime , we study the evolution of break frequencies numerically .
the results for @xmath60 and @xmath61 are given in paper i. for @xmath43 , a good approximation is @xmath62 \hz } \\
& \times & \left(1 \over 1+z \right ) \left(c_s \over c/\sqrt{3 }
\right)^{17/6 } \left ( \xib \over 0.1\right)^{-3/2 } \\ & \times & \left ( \rho \cdot \cm^{3 } \over 10^{-24 } \gram \right)^{-5/6 } \left ( e_0 / 10^{53 } \erg \over \zm^2 / 4 \right)^{-2/3 } \left ( \zm \over 0.1 \right)^{-4/3 } ~~.\end{aligned}\ ] ] we fixed values of @xmath65 and @xmath66 , and then executed a grid search on the break time @xmath67 and normalization of the model light curve .
results are summarized in table 2 and figure 1 .
the final @xmath68 per degree of freedom is @xmath69 .
these large @xmath68 values make meaningful error estimates on parameters difficult .
let us suppose @xmath68 is large because details omitted from the models ( clumps in the ambient medium or blast wave instabilities ) affect the light curve , and so attach an uncertainty of @xmath70 to each predicted flux . adding this in quadrature to observational uncertainties when computing @xmath68
, we obtain @xmath71 . error estimates based on changes in @xmath68 then rule out @xmath72 at about the 90% confidence level even for our `` maximum beaming '' case ( @xmath73 , @xmath74 ) . to convert a supposed break time @xmath67 into a beaming angle @xmath0 , we need estimates of the burst energy per steradian and the ambient density .
wijers & galama ( 1998 ) infer @xmath75 and @xmath76 .
combining these values with @xmath77 gives @xmath78 .
@xmath79 and @xmath4 are substantially uncertain , but because @xmath80 , the error budget for @xmath0 is dominated by uncertainties in @xmath66 rather than in @xmath5 or @xmath4 .
this beaming limit implies @xmath81 , which is @xmath82 of the sky .
grb 970508 was at @xmath83 ( metzger et al 1997 ) .
we then find gamma ray energy @xmath84 \right)$ ] . if the afterglow is primarily powered by different ejecta from the initial grb , as when a `` slow '' wind ( @xmath85 ) dominates the ejecta energy , then our beaming limit applies only to the afterglow emission .
the optical fluence implies @xmath86 \right)$ ] .
the irreducible minimum energy is thus @xmath87 , using the smallest possible redshift and beaming angle .
we have reduced the beaming uncertainty , from the factor @xmath88 allowed by @xmath89-ray observations alone to a factor @xmath90 , and thus obtain the most rigorous lower limit on grb energy requirements yet .
galama , t. j. , et al 1998 , 500 , l97 garcia , m. r. et al 1998 , 500 , l105 metzger , m. r. , et al 1997 , , 387 , 879 rhoads , j. e. 1999 , ( paper i ) , astro - ph/9903399 ( @xmath91 ) rhoads , j. e. 1998 , astro - ph/9712042 sari , r. , piran , t. , & narayan , r. 1998 , 497 , l17 wijers , r. a. m. j. , & galama , t. j. 1998 , astro - ph/9805341 zharikov , s. v. , sokolov , v. v. , & baryshev , yu . v. 1998 , 337 , 356 | the beaming angle @xmath0 is the main uncertainty in gamma ray burst energy requirements today .
we summarize predictions for the light curves of beamed bursts , and model the r band light curve of grb 970508 to derive @xmath1 .
this yields an irreducible minimum energy requirement of @xmath2 ergs to power the afterglow alone . |
evolution of a supernova remnant ( snr ) with no neutron star in its center is described essentially by the shock waves propagating outward ( blast wave ) into the ambient matter and inward ( reverse - shock wave ) into the supernova ejecta . in young snrs ,
the shocked matter is heated up to temperatures higher than @xmath0 k and forms an optically - thin hot plasma .
this plasma emits x - rays with many emission lines of highly - ionized heavy elements .
since the time - scale of ionization by electron impact is longer than that of shock - heating of electrons , x - ray spectra of young snrs exhibit ionization states lower than those expected from the electron temperatures , which is called under - ionized or ionizing . the ionization state is represented by ionization temperature @xmath1 , the plasma temperature in collisional ionization equilibrium having the relevant ionization state .
thus , many of young snrs show @xmath2 ( under - ionized / ionizing ) for the electron temperature @xmath3 . on the contrary to the above standard scenario , @xcite found an over - ionized plasma ( @xmath4 ) in the _ asca _ observation from a mixed - morphology snr , ic443 .
they proposed that thermal conduction from the hot interior of the remnant to the cold exterior can explain the over - ionized plasma .
@xcite further investigated other five mixed - morphology snrs , w49b , w28 , w44 , 3c 391 and kes 27 , and found that w49b shows @xmath4 as well .
_ suzaku _ detected clear spectral structures of radiative - recombination continua , direct evidence for recombining plasmas , in the x - ray spectra of ic443 @xcite and w49b @xcite .
radiative recombination continua , and hence an over - ionized plasma is also found in another mixed - morphology snr , g359.10.5 @xcite .
although the samples are still limited , the over - ionized plasmas are all found in the mixed - morphology snrs associated in or near the star - forming complex with hii regions and molecular clouds .
hence these snrs are likely due to core - collapse supernovae ; the over - ionized plasma is possibly related to the massive progenitor and/or its environments .
one possibility is that the ionization by photons of a few tens kev at the initial phase of explosion of massive stars , e.g. , possible x - ray flash or afterglow of @xmath5-ray bursts , is responsible for the over - ionized state .
another possibility is rarefaction in the adiabatic evolution phase of the snr ; when the shock wave breaks out of the dense circumstellar matter ( csm ) , like the progenitor s stellar wind , into the rarefied interstellar medium ( ism ) , the shock - heated electrons would rapidly cool due to adiabatic expansion , leaving highly ionized states @xcite . in the present paper ,
we investigate the latter possibility .
@xcite calculated snr evolution and radiation for a spherically symmetric structure .
we extend this work to non - symmetric structures as are found in mixed - morphology snrs .
unlike @xcite , we do not take ionization or radiation processes into account ; we assume the relevant snr evolution is fully adiabatic .
we describe the model in the following section , calculation and results in section 3 , and discuss the results in section 4 .
we give a summary in section 5 and notes about electron temperature in appendix .
in a core - collapse supernova , the massive progenitor ( red or blue supergiant or wolf - rayet star ) is expected to blow a strong stellar wind in its pre - supernova phase and forms dense csm around the progenitor . if the stellar wind is spherically symmetric with a mass - loss rate @xmath6 and wind velocity @xmath7 , the csm density @xmath8 at a distance @xmath9 from the progenitor is given by @xmath10 for a while before supernova explosion , the stellar wind ceases . then the stellar wind forms a thick shell .
the inner and outer radii of the shell are given by @xmath11 and @xmath12 , respectively ( see figure [ figure1 ] ) , where time @xmath13 and @xmath14 are the duration of the wind activity and the elapsed time after the cease of the wind , respectively . in realistic stellar winds , more matter around the equatorial plane than the polar direction may be accumulated , because of the rotation of the progenitor ; disk - like winds of be stars for an extreme example .
we thus modify eq .
( 1 ) as , @xmath15 @xmath16 with an anisotropy parameter @xmath17 , which is the ratio of the major to minor axis of an equi - density surface of the wind matter .
hereafter , @xmath18 and @xmath19 are referred to the radii on the equatorial plane .
this distribution gives @xmath20 times higher density on the equatorial plane than in the polar direction at the same distance .
for the progenitor of sn 1987a , @xcite suggested that the ratio of equatorial to polar mass - loss rate was at least 20 during its red supergiant stage .
this value corresponds to @xmath21 . in our calculations
we adopt the values from 1 to 3 .
we calculate the evolution for the following five models : * isotropic csm same as @xcite for a reference . * anisotropic csm concentrated around the equatorial plane . * same as a2 , but more mass around the equatorial plane . * same as a2 , but with a lower density ism .
* same as b1 , but with a higher wind velocity .
the numerical values for each model are shown in table [ table1 ] , where values of @xmath18 , @xmath19 , @xmath22 and @xmath23 are those on the equatorial plane .
the mass - loss rate is assumed to be @xmath24 , the same as @xcite , for all the models .
( 80mm,52mm)figure1.eps cccccccc & & ism + model&@xmath17&@xmath18 ( 10@xmath25cm)&@xmath19 ( 10@xmath25cm)&@xmath22 ( amu@xmath26)&@xmath23 ( amu@xmath26)&@xmath7 ( @xmath27)&@xmath28 ( amu@xmath26 ) + a1&1 & 2.0&92.6&3.4@xmath29 & 1.6@xmath30&10@xmath31 & 1.6@xmath32 + a2&2 & 2.0&92.6&6.9@xmath29 & 3.2@xmath30&10@xmath31 & 1.6@xmath32 + a3&3 & 2.0&92.6&1.0@xmath33 & 4.8@xmath30&10@xmath31 & 1.6@xmath32 + b1&2 & 2.0&92.6&6.9@xmath29 & 3.2@xmath30&10@xmath31 & 1.6@xmath34 + b2&2 & 20&926&6.9@xmath30 & 3.2@xmath32&10@xmath35&1.6@xmath34 + + [ table1 ] we focus on the effect of the csm , particularly for anisotropic csm , on the early phase evolution of snrs .
for the supernova ejecta , we simply assume spherical distribution with a core of radius @xmath36 and extended envelope of radius @xmath37 , as @xcite @xmath38 @xmath39 where @xmath40 is the expansion velocity at @xmath37 .
@xmath41 and @xmath42 are the ejecta mass and the explosion energy ( kinetic energy ) , respectively .
these are assumed to be @xmath43 erg and @xmath44 .
the other parameters @xmath45 and @xmath46 are taken to be 6 and 0.49 , respectively .
all these values are the same as @xcite , and hence @xmath47 @xmath27 .
we numerically solve the euler equations utilizing the athena3d code ( @xcite ) : @xmath48 where @xmath8 , @xmath49 , @xmath50 and @xmath51 are the mass density , the pressure , the fluid velocity and the internal energy per unit mass , respectively .
the equation of state is written as @xmath52 with @xmath53 , and the mean molecular weight is taken to be 0.5 . in the hydrodynamical evolution
, we obtained mean temperature @xmath54 of ions ( protons ) and electrons weighted by their number density .
if no plasma mode works directly on electrons at the shock front , the ion temperature @xmath55 rises faster and then the energy of ions is transferred to electrons .
we calculate @xmath55 and @xmath56 from @xmath54 , assuming the transport by coulomb collisions ( @xcite ; see appendix ) .
although we include neither ionization nor radiation and simply assume the adiabatic evolution , the evolution of model a1 is in fairly good agreement with @xcite . the blast shock and reverse shock heat the csm to @xmath57 k , and the ejecta to @xmath58 k , respectively .
the blast wave expands as @xmath59 , while @xmath60 in @xcite , where @xmath61 is the blast - wave radius .
the blast wave breaks out at 41 yr , 2 years earlier than in @xcite who take radiation loss into calculation . after the break - out , the blast wave rapidly expands adiabatically , and hence the temperature and density of the shocked matter decrease approximately as @xmath62 and @xmath63 for @xmath64 .
when the pressure of the shocked csm becomes below that of the shocked ism , the second reverse shock occurs to propagate inward and reheats the csm and ejecta .
the second reverse shock reaches the ejecta at 830 yr , 120 years earlier than in @xcite .
the evolution is basically the same as model a1 , but is dependent on direction : more matter on the equatorial plane than in the polar direction . with increasing the value of @xmath17 , the break - out occurs earlier in the polar and later in the equatorial direction than in model a1 .
the former is due to a shorter distance to the csm , and the latter is higher density of the csm than those in model a1 .
in addition , the second reverse shock reaches the ejecta earlier in the polar direction and later in the equatorial direction than in model a1 . in model a2
, the break - out occurs at @xmath65 yr in the polar direction , and at @xmath66 yr in the equatorial direction as shown in figure [ figure2 ] .
the mean temperature @xmath54 of the shocked ejecta turns to rise at 340 yr and 1100 yr , in the polar and equatorial directions , respectively .
( 80mm,78mm)figure2.eps the break - out occurs at the same time as model a2 since the csm distribution is the same as model a2 .
the difference of model b1 from model a2 is the lower density of the ism by an order of magnitude .
the velocity of the blast wave after the break - out is slightly higher than that in model a2 by a factor of @xmath67 in the equatorial direction . as a result ,
the pressure of the shocked ism is about one - fifth of that in model a2 , and the second reverse shock occurs later .
the temperature @xmath54 of the shocked ejecta drops at 23 yr and 68 yr by the break - out , but turns to rise at 810 yr and 1800 yr in the polar and equatorial directions , respectively .
the difference from model b1 is the lower density of the csm by three orders of magnitude due to higher wind velocity by an order of magnitude .
even at such low densities , the break - out occurs at 170 yr and 450 yr in the polar and equatorial directions , respectively .
similarly to the other models , the temperature @xmath54 of the shocked ejecta drops at 280 yr and 770 yr , but turns to rise by the second reverse shock at 920 yr and 2400 yr in the polar and equatorial directions , respectively .
ccccccc & & + model&@xmath68 ( yr)&@xmath54 ( @xmath0k)&@xmath69 ( pc)&@xmath70 ( yr)&@xmath54 ( @xmath0k)&@xmath69 ( pc ) + a1&41&1.3&0.30&180&0.20&1.7 + a2-e&46&1.2&0.30&210&0.18&1.8 + a2-p&17&1.6&0.15&80&0.16&0.99 + a3-e&47&1.4&0.30&220&0.21&1.8 + a3-p&11&0.8&0.10&60&0.11&0.81 + b1-e&46&1.2&0.30&190&0.23&2.0 + b1-p&17&1.7&0.15&80&0.10&1.1 + b2-e&450&1.2&3.0&70&1.7&0.64 + b2-p&170&1.1&1.5&36&1.2&0.38 + + [ table2 ] in table [ table2 ] we summarize the temperatures and blast - wave radii at characteristic epochs , @xmath68 and @xmath70 ( see below ) , for each model . before the break - out
, the evolution of snr can be described by a self - similar solution , and the blast wave expands as @xmath71 .
almost independently of anisotropy of the csm ( models a13 ) , the values of @xmath72 are close to each other both in the polar and equatorial directions , as seen in the _ upper _ panel of figure [ figure3 ] . after the break - out ,
@xmath72 values are slightly larger in larger @xmath17 models , but approach the value of model a1 ( @xmath73 ) with time .
this implies that the blast wave approaches spherical symmetry with time , and the ratio of the blast - wave radius of the polar to equatorial direction becomes less than @xmath74 in 1000 years in any model of ours here .
if the density difference across the interface between the csm and ism is large enough ( models a13 and b1 ) , the blast wave gets faster just after the break - out by a factor of @xmath75 , almost independent of the density difference . on the other hand , if the density difference is as small as model b2 , the velocity increases by a factor of @xmath76 .
the expansion velocity of a spherical fluid initially at rest asymptotically approaches its maximum value of @xmath77 , where @xmath78 is a sound speed of the fluid @xcite .
similarly , denoting the shock velocity by @xmath79 , we may have the maximum expansion velocity after the break - out , as @xmath80^{1/2}\simeq 2.0\,v_{\mathrm s}(t_{\mathrm b}),\ ] ] where @xmath81 is the shock velocity of the blast wave immediately before the break - out at @xmath82 . here
we take the rankine - hugoniot relation of strong shocks into account .
this estimate is consistent with the result of our hydrodynamical calculation
. the high shock velocity and rarefaction caused by the break - out may be in favor of particle acceleration , because the former boosts the maximum energy of particles , and the latter works for particles to become non - thermal .
( 80mm,65mm)figure3.eps ( 80mm,70mm)figure4a.eps ( 80mm,70mm)figure4b.eps as demonstrated by @xcite , rarefaction by the blast - wave break - out rapidly cools the shock - heated matter to be an over - ionized / recombining plasma of @xmath83 .
this recombining plasma state lasts until the second reverse shock reheats the matter to @xmath84 .
although ionization is not calculated simultaneously with hydrodynamics , we can analyze the ionization state , i.e. , ionizing , equilibrium or recombining , of the shocked matter from the electron temperature and density .
the ionization state thus evaluated is consistent with @xcite , considering the variation of epochs and duration that depend on the csm / ism models , as discussed in the previous sections .
the _ upper _ panels of figures [ figure4]a ( model b1 ) and [ figure4]b ( model b2 ) show the evolution of the ion and electron temperatures averaged over the shocked matter of the ejecta . before the break - out
, @xmath56 becomes nearly equal to @xmath55 due to high densities of the csm . also after the break - out and until arrival of the second reverse shock ,
@xmath56 is nearly equal to @xmath55 , because cooling is due to adiabatic expansion . by the second reverse shock
, @xmath55 rises faster than @xmath56 , and then @xmath56 rises through the energy transport from ions , as shown in the _ upper _ panels of figure [ figure4 ] . in the _ lower _ panels ,
we show the average density of the shocked ejecta by the solid line .
ionization by electron - impact becomes equilibrium with time - scale @xmath85 given by @xmath86 , almost independent of the temperature or ion species @xcite .
we also plot this relation as @xmath87 with the dashed line in the _ lower _ panels .
well before the break - out , since the density is high enough ( @xmath88 ) due to the presence of csm , ionization quickly reaches its equilibrium at @xmath56 , i.e. @xmath89 .
as the snr expands , the average density of the shocked matter decreases approximately as @xmath90 .
when rarefaction occurs by the break - out , @xmath8 decreases faster as @xmath91 ( see section 3.1 ) .
therefore , @xmath8 becomes below @xmath92 , as seen in the _ lower _ panels of figure [ figure4 ] , and @xmath93 is decoupled from @xmath56 .
in other words , recombination no longer follows the rapid decrease of the electron temperature , and the ionization state freezes roughly at @xmath94 , where @xmath70 is the epoch at which @xmath95 .
the decoupling epoch @xmath70 is also shown in the _ lower _ panels of figure [ figure4 ] with the thin - dotted vertical line . in model b1 , @xmath96 yr in the equatorial direction . note that @xmath97 and @xmath98 in model b1 , where @xmath99 is the electron temperature immediately before the break - out .
models a13 show the similar behavior to model b1 described here . in model b2 , unlike other models or @xcite , the csm is located away from the progenitor , and hence the density is low .
therefore , the break - out occurs later but @xmath8 becomes below @xmath100 earlier than in model b1 . as seen in figure [ figure4]b , @xmath101 in model b2 , and @xmath102 yr in the equatorial direction .
in fact , as is seen in the _ upper _ panel , @xmath56 before the break - out is nearly constant at @xmath103 in model b2 . as a result , the ionization temperature of the recombining plasma is higher in model b2 than that in model b1 .
when the second reverse shock reaches the ejecta , the temperature turns to rise , as seen in the _ upper _ panels of figure [ figure4 ] .
if the electron temperature exceeds or becomes comparable to @xmath104 , the ionization state turns to be ionizing or nearly equilibrium .
this is the case for models a13 with a higher ism density . in model b1 , though the ism density is low , @xmath56 exceeds @xmath93 in 3000 years owing to the dense csm and lower @xmath93 . on the other hand , in model b2 ,
the density is too low for the second reverse shock to raise @xmath105 in such short time , and the over - ionization state remains much longer .
if @xmath55 becomes higher than @xmath106 at the shock front , @xmath107 may be attained by the energy transfer from ions . since @xmath108 ( reverse shock ) decreases with the age , however , the recombining state would last yet for thousands years .
after decoupled from the electron temperature , the ionization temperature decreases by recombination with time - scale @xmath85 .
we note here that the recombination time - scale @xmath109 yr @xcite is longer than the age of over - ionized snrs : @xmath110 yr for ic443 @xcite and @xmath111 yr for w49b ( @xcite , @xcite , hwang et al .
2000 ) .
all the snrs with over - ionized / recombining plasma so far observed are mixed - morphology snrs , which exhibit center - filled x - ray emission and shell - like radio emission . hence , we investigate distribution of the x - ray emission measure for snrs with anisotropic stellar wind .
we integrate the density square along the line of sight , as @xmath112 , and show the map ( black lines ) for model b2 in figure [ figure5 ] . here
the shocked matter of @xmath113 k , which is responsible for x - ray emission , is taken into calculations . at ages 980 yr and 1800 yr
, we can see a bar - like structure ( horizontal ) with diffuse wings in the equatorial view , while a thin shell in the polar view .
the emission measure is dominated by the shock - heated ejecta , which is a recombining plasma as seen from figure [ figure4]b . the temperature dependence of the line emissivity is weaker in recombining than in equilibrium , and the emission - measure map reflects roughly the x - ray surface brightness .
the bar structure is clearer at younger ages after the break - out , and diffuses out gradually with time . at 1800 yr
, the second reverse - shock front appears to surround the bar / wing structure , and thereby the bar - end structure is being distorted .
the second reverse - shock propagates inward and eventually sweeps the bar / wing structure out .
hence , the late - phase snr approaches shell - like both in the equatorial and polar views , as seen at 10000 yr in figure [ figure5 ] , where a little elongated shape with a narrow middle part is still seen in the equatorial view .
w49b shows a bar - like structure as well as a recombining plasma state @xcite .
the bar and its east - end observed with _
@xcite look similar to our calculation shown in figure [ figure5 ] .
thus we suggest that w49b is the case of nearly equatorial view for the supernova remnant of a massive progenitor which exploded in its past stellar wind .
it should be noted that , in our calculation , the structure of the bar - end is formed by the reverse shock , not by the collision with a molecular cloud near the bar - end .
the grey lines in figure [ figure5 ] represent the low - level emission measure , much smaller than that represented by the black lines .
one can see the blast - wave front in the grey lines .
the shocked matter in the grey lines is hardly observed in x - rays , but the shock possibly accelerates electrons to be of order of gev , which can be responsible for ghz synchrotron radio ; the condition after the break - out is favorable for particle acceleration , as mentioned in section 4.1 .
( 84mm,46.2mm)figure5a.eps + ( 84mm,46.2mm)figure5b.eps + ( 84mm,46.2mm)figure5c.eps
we investigate the evolution of snrs that explode in the progenitors stellar wind matter , considering possible environments of mixed - morphology snrs with over - ionized plasmas .
we summarize the results ; * when the blast wave breaks out of the wind matter into the ambient interstellar medium , the shocked matter cools rapidly due to adiabatic expansion . just after the break - out , the expanding velocity becomes faster by a factor up to two , and then gradually decreases to that of the extrapolated from the velocity trend before the break - out .
* before the break - out , the shocked matter reaches ionization equilibrium and equipartition @xmath114 , but deviates from equilibrium by rarefaction after the break - out .
consequently , the shock - heated ejecta turns to be a recombining plasma , since cooling due to adiabatic expansion is much faster than recombination . * the recombining state of the shocked ejecta lasts until the second reverse shock , which occurs by the interaction with the interstellar medium , propagates inward and reheats the ejecta .
if the density of the ejecta is too low to establish ionization equilibrium , however , the recombining state lasts longer .
* after the break - out in the adiabatic phase , since the emission measure of the shocked ejecta is much larger than that of the shocked ambient matter , the snr in x - ray wavelengths appears much brighter in the reverse - shocked inner region than the blast - shocked outer shell .
* when the stellar wind matter is not isotropic but denser in the equatorial direction due to the progenitor s rotation , the snr in the recombining state looks bar - like with wings in the equatorial view and thin shell - like in the polar view .
so that , the snr would show center - filled various shapes in x - rays , depending on the viewing angle . on the other hand , the blast - shocked matter , which is very faint in x - rays but could be observed in radio , forms a fairly complete shell outside .
* as the snr age increases , however , the second reverse shock sweeps the bar / wing structure out and merges into the whole ejecta eventually .
hence , the bar / wing structure disappears and the late - phase snr would look shell - like almost independently of the viewing angle . the authors are grateful to hiroya yamaguchi and midori ozawa for meaningful discussion about _ suzaku _ observations of recombination radiation from snrs .
km and kk were respectively supported by the grant - in - aid for scientific research 22540253 and 2054019 , from japan society for the promotion of science ( jsps ) .
in the present work , we assume that shocks heat substantially ions at the shock front , and then the electron temperature rises through the energy transport by coulomb collisions and diffusion in the post shock region . in figure [ figure6 ]
we compare the electron temperature thus obtained @xcite by the solid lines and temperature in the equipartition with ions at the shock front by the broken lines , with the observed electron temperatures ( the circles ) .
one can see the transfer by coulomb collisions from ions to electrons is a reasonable assumption to account for the observed data .
aharonian , f. et al .
2008 , , 490 , 685 blondin , j. m. , & lundqvist , p. 1993
, , 405 , 337 enoguchi , h. , tsunemi , h. , miyata , e. , yoshida , k. 2002 , , 54 , 229 gotthelf , e. v. , vasisht , v. 1997 , , 486 , l133 harrus , i. , slane , p. o. , smith , r. k. , hughes , j. p. 2001
, , 552 , 614 hwang , u. , gotthelf , e. v. 1997 , , 475 , 665 hwang ,
u. , petre , r. , & hughes , j. p. 2000 , , 532 , 970 itoh , h. , & masai , k. 1989 , , 236 , 899 katsuda , s. , petre , r. , hwang , u. , yamaguchi , h. , mori , k. , tsunemi , h. 2009 , , 61 , s155 kawasaki , m. , ozaki , m. , nagase , f. , inoue , h. , & petre , r. 2005 , , 631 , 935 kawasaki , m. t. , ozaki , m. , nagase , f. , masai , k. , ishida , m. , & petre , r. 2002 , , 572 , 897 keohane , j. w. , reach , w. t. , rho , j. , & jarrett , t. h. 2007 , , 654 , 938 kinugasa , k. , tunemi , h. 1999 , , 51 , 239 koo , b .- c . , kim , k .-
t . , seward , f. d. 1995 , , 447 , 211 lazendic , j. s. , slane , p. o. , hughes , j. p. , chen , y. , dame , t. m. 2005 , , 618 , 733 masai , k. 1994 , , 437 , 770 miyata , e. , katsuda , s. , tsunemi , h. , hughes , j. p. , kokubun , m. , porter , f .
s. 2007 , 59 , s163 nakamura , r. , bamba , a. , ishida , m. , nakajima , h. , yamazaki , r. , terada , y. , phlhofer , g. , wagner , s. j. 2009 , , 61 , s197 ohnishi , t. , koyama , k.,tsuru , t. g. , masai , k. , yamaguchi , h. , & ozawa , m. 2011 , , 63 , 527 ozawa , m. , koyama , k. , yamaguchi , h. , masai , k. , & tamagawa , t. 2009 , , 706 , l71 park , s. , kargaltsev , o. , pavlov , g. g. , mori , k. slane , p. o. , hughes , j. p. , burrows , d. n. , garmire , g. p. 2009
, , 695 , 431 park , s. , slane , p. o. , hughes , j. p. , mori , k. , burrows , d. n. , garmire , g. p. 2007
, , 665 , 1173 pye , j. p. , thomas , n. , becker , r. h. , & seward , f. d. , 1984 , , 207 , 649 reynolds , s. p. , borkowski , k. j. , hwang , u. , harrus , i. , petre , r. , dubner , g. 2006 , , 652 rho , j. , dyer , k. , borkowski , k. j. , reynolds , s. p. 2002
, , 581 , 1116 rho , j. , petre , r. 1998 , , 503 , l167 saken , j. m. , long , k. s. , blair , w. p. , winkler , p. f. 1995 , , 443 , 231 sasaki , m. , plucinsky , p. p. , gaetz , t. j. , smith , r. k. , edger , r. j. , slane , p. o. 2004 , , 617 , 322 smith , a. , peacock , a. , jones , l. r. , & pye , j. p. 1985
, , 296,469 stone , j. m. , gardiner , t. a. , teuben , p. , hawley , j. f. , & simon , j. 2008 , , 178 , 137 sun , m. , wang , z .- r . ,
chen , y. 1999 , , 511 , 274 tamura , k. 1995 , ph.d . thesis , osaka univ .
troja , e. , bocchino , f. , miceli , m. , & reale , f. 2008 , , 485 , 777 truelove , j. k. , & mckee , c. k. 1999 , , 120 , 299 yamaguchi , h. , koyama , k. , katsuda , s. , nakajima , h. , hughes , j. p. , bamba , a. , hiraga , j. s. , mori , k. , ozaki , m. tsuru , t. g. 2008 , , 60 , s141 yamaguchi , h. , ozawa , m. , koyama , k. , masai , k. , hiraga , j. s. , ozaki , m. , & yonetoku , d. 2009 , , 705 , l6 zeldovich , y. b. , & raizer , y. p. 1966 , physics of shock waves and high - temperature hydrodynamic phenomena , ed .
w. d. hayes & r. f. probstein , volume i ( academic press new york and london ) , 103 | we carry out 3d - hydrodynamical calculations for the interaction of expanding supernova ejecta with the dense circumstellar matter ( csm ) and the rarefied interstellar medium ( ism ) outside .
the csm is composed of the stellar wind matter from the progenitor in its pre - supernova phase , and assumed to be axially symmetric : more matter around the equator than in the polar direction driven by rotation of the progenitor .
because of high density of the csm , the ionization state of the shock - heated ejecta quickly becomes equilibrium with the electron temperature .
when the blast wave breaks out of the csm into the rarefied ism , the shocked ejecta cools rapidly due to adiabatic expansion , and hence an over - ionized / recombining plasma would be left .
the ejecta is reheated by the second reverse shock due to the interaction with the ism .
we calculate the emission measure of the supernova remnant ( snr ) along the line of sight , and find that the over - ionized plasma appears to be bar - like with wings in the edge - on ( equatorial view ) , while shell - like in the face - on ( polar view ) geometry with respect to the rotation axis .
the hot gas heated by the blast wave exists in the outermost region of the snr with a nearly complete shell , but the x - rays therefrom are too faint to be observable .
thus , depending on the viewing angle , the snr of the over - ionized plasma would exhibit center - filled morphology in x - rays , like w49b , a mixed - morphology snr .
the bar - like structure is swept out by the second reverse shock and disappears eventually , and then the snr becomes shell - like in both the equatorial and polar views in the later phase of the evolution . |
it is customary in the study of generalised surfaces to examine the widths generated by kinetic roughening @xcite , and then establish properties related to _ dynamical scaling_.
however , the kinetic roughening of sandpile cellular automata has never been investigated ; to begin with , therefore we postulate a principle of _ dynamical scaling for sandpile cellular automata _ in terms of the surface width @xmath11 of the sandpile automaton : @xmath12 as in the case of interfacial widths , these equations signify the following sequence of roughening regimes : 1 . to start with , roughening occurs at the ca sandpile surface in a time - dependent way ; after an initial transient , the width scales asymptotically with time @xmath6 as @xmath13 , where @xmath14 is the _ temporal roughening _ exponent .
this regime is appropriate for all times less than the crossover time @xmath15 , where @xmath16 = @xmath17 is the dynamical exponent and @xmath3 the system size .
2 . after the surface has _ saturated _ , i.e. its width no longer grows with time , the _ spatial roughening _ characteristics of the mature interface can be measured in terms of @xmath18 , an exponent characterising the dependence of the width on @xmath3 .
we define the surface width @xmath19 for a sandpile automaton in terms of the mean - squared deviations from a suitably defined mean surface ; in analogy with the conventional counterpart for interface growth @xcite , we define the instantaneous mean surface of a sandpile automaton as the surface about which the sum of column _ height _ fluctuations vanishes .
clearly , in an evolving surface , this must be a function of time ; hence all quantities in the following analysis will be presumed to be instantaneous . the mean slope @xmath20 defines expected column heights , @xmath21 , according to @xmath22 where we have assumed that column @xmath5 is at the bottom of the pile .
column height deviations are defined by @xmath23 the mean slope must therefore satisfy @xmath24 = 0\ ] ] since the instantaneous deviations about it vanish ; thus @xmath25 / l(l+1)\ ] ] ( we note that this slope is distinct from the quantity @xmath26 that is obtained from the average of all the local slopes @xmath27 , about which _ slope _
fluctuations would vanish on average ) .
the instantaneous width of the surface of a sandpile automaton , @xmath19 , can be defined as : @xmath28 / l } \ ] ] which can in turn be averaged over several realizations to give , @xmath29 , the average surface width in the steady state .
we also compute here the height - height correlation function , @xmath30 , which is defined by @xmath31 where the mean values are evaluated over all pairs of surface sites separated by @xmath32 lattice spacings : @xmath33 for @xmath34 .
this function is symmetric and can be averaged over several realizations to give the average correlation function @xmath35 .
before moving on to the quantitative descriptors of sandpile avalanching and surface roughening , we present some results using more qualitative indicators .
recent experiments @xcite on sandpile avalanches have indicated that there are at least two broad categories ; uphill avalanches , which are typically large , and triangular avalanches which are generally smaller in size .
we have found evidence of this in a @xmath36 disordered model of sandpile avalanches , which will be presented elsewhere @xcite ; but in this work we discuss analogues in @xmath37 , which are respectively wedge - shaped and flat avalanches .
the following data indicate that it is the larger wedge - shaped avalanches which alter surface slope and width , while the flatter , smaller avalanches alter neither very much .
this is in accord with earlier work , where it was found that larger avalanches are the consequence of accumulated disorder , while the smaller ones can cause disordered regions to build up along the sandpile surface @xcite .
figure [ fig1](a ) shows a time series for the mass of a large ( @xmath38 ) evolving disordered sandpile automaton .
the series has a typical quasiperiodicity @xcite .
the vertical line denotes the position of a particular large event , while figure [ fig1](b ) shows the avalanche size distribution for the sandpile .
note the peak , corresponding to the preferred large avalanches , which was analysed extensively in earlier work @xcite .
our data shows that the avalanche highlighted in figure [ fig1](a ) drained off approximately @xmath39 per cent of the mass of the sandpile , placing it close to the second peak of figure [ fig1](b ) .
figure [ fig1](c ) shows the outline of the full avalanche before and after this event with its initiation site marked by an arrow ; we note that , as is often the case in one dimension , the avalanche is uphill. the inset shows the relative motion of the surface during this event ; we note that the signatures of smoothing by avalanches are already evident as the precursor state in the inset is much rougher than the final state .
finally we show in figure [ fig1](d ) the grain - by - grain picture of the aftermath pile superposed on the precursor pile , which is shown in shadow .
an examination of the aftermath pile and the precursor pile shows that the propagation of the avalanche across the upper half of the pile has left only a very few disordered sites in its wake ( i.e. the majority of the remaining sites are @xmath4 type ) whereas the lower half ( which was undisturbed by the avalanche ) still contains many disordered , i.e. @xmath5 type sites in the boundary layer .
this leads us to suggest that the larger avalanches rid the boundary layer of its disorder - induced roughness , a fact that is borne out by our more quantitative investigations .
in fact , our studies have revealed that the very largest avalanches , which are system - spanning , remove virtually all disordered sites from the surface layer ; one is then left with a normal ordered sandpile , where the avalanches have their usual scaling form for as long as it takes for a layer of disorder to build up .
when the disordered layer reaches a critical size , another large event is unleashed ; this is the underlying reason for the quasiperiodic form of the time series shown in figure [ fig1](a ) . before moving on to more quantitative features ,
we show for comparison the sequence of figure [ fig1 ] , for 1 .
an ordered pile - figure [ fig2](a - d ) 2 . a small disordered pile - figure [ fig3](a - d ) we note the following features : * the small disordered pile has a mass time series ( figure [ fig3](a ) ) that is midway between the scale - invariance of the ordered pile ( figure [ fig2](a ) ) and the quasiperiodicity of the large disordered pile ( figure [ fig1](a ) ) . * the avalanche size distribution of the small disordered pile ( figure [ fig3](b ) ) is likewise intermediate between that of the ordered pile ( which shows the scale invariance observed by kadanoff et al .
@xcite ) and the two - peaked distribution characteristic of the disordered pile @xcite . * in both small and large disordered piles , we see evidence of large uphill avalanches which shave off a thick boundary layer containing large numbers of disordered sites , and leave behind a largely ordered pile ( see figure [ fig1](c - d ) and figure [ fig3](c - d ) ) .
by contrast the ordered pile loses typically two commensurate layers even in the largest avalanche , with a correspondingly unexciting aftermath state left behind in its wake ( figure [ fig2](c - d ) ) .
we conclude from this that there is , even at a qualitative level , a post - avalanche smoothing of the sandpile surface , beyond a crossover length , as found in earlier work on continuum models @xcite ; importantly , our discrete model reveals that this is achieved by the removal of ( orientational ) disorder , the implications of which we will discuss in our concluding section .
the existence of the crossover length , in terms of the mass time series , has also been observed in experiment @xcite .
inspired by the picture of smoothing avalanches , we have investigated many of the material properties of the sandpile in the special pre- and post - avalanche configurations . from these
we have drawn the following conclusions : * the _ mean slope _ of the disordered sandpile peaks ( see table [ table1 ] ) before a large avalanche and drops immediately after ; this statement is true for events of any size and thus remains trivially true for the ordered sandpile . * the _ packing fraction _ @xmath40 of the disordered sandpile increases after a large event , i.e. effective consolidation occurs during avalanching ( see table [ table1 ] ) .
this consolidation via avalanching mirrors that which occurs when a sandpile is shaken with low - intensity vibrations @xcite .
* however , a far deeper statement can be made about the comparison of the surface width for pre- and post- large event sandpiles ; table [ table1 ] shows that the surface width goes down considerably during an event , once again suggesting that a rough precursor pile is smoothed by the propagation of a large avalanche .
we have also investigated the dependence of various material properties of a disordered sandpile on the aspect ratio of the grains @xcite .
table [ table2 ] shows our results , and figure [ fig4 ] illustrates the variation of the avalanche size distribution .
there is a transition as aspect ratios of @xmath41 are approached from above or below ; we have shown above that piles with these critical aspect ratios manifest strong disorder @xcite in the sense of : * a second peak in the avalanche size distribution denoting a preferred size of large avalanches * large surface widths denoting an increased surface roughness * a strong correlation between interfacial roughness and avalanche flow since the mean surface width varies dramatically in the pre- and post- large event piles .
clearly , sandpiles containing grains with aspect ratios close to unity act essentially as totally ordered piles @xcite ; there is however a significant symmetry in the shape of the avalanche size distribution curves above and below the transition region ( see figure [ fig4](a ) and ( d ) ) .
these size distributions are reminiscent of those obtained in earlier work @xcite for the case of uniform disorder ( which referred to piles that have disorder throughout their volume rather than , as is the present case , disorder concentrated in a boundary layer ) .
these observations lead us to speculate that there exist at least three types of avalanche spectra : 1 . the scale - invariant statistics characteristic of ordered sandpiles 2 .
the strongly disordered statistics characterized by a second peak in the distribution ; which we have obtained for specific values of the aspect ratio in the case where the disorder is concentrated in a boundary layer 3 .
the more weakly disordered region ( characterized by a flatter size distribution of avalanche sizes which is , nevertheless , _ not _ scale - invariant ) obtained in the intermediate regimes of aspect ratio ( as well as in the case of uniform disorder ) .
it is clear that _ the presence of inherent inhomogeneities in grain shape ( which we describe quantitatively by aspect ratio ) or bulk structure ( which we describe by the classifications of uniform or boundary disorder ) in a sandpile induces the presence of strong disorder in avalanche statistics_. additionally we present , in figure [ fig5 ] , the mass - mass correlation function of a particular disordered sandpile ; the curve has a peak , which indicates the average time between avalanches .
since the avalanche size distribution for this sandpile includes a preponderance of large events , we conclude that the peak in the correlation function corresponds approximately to the _ time between large avalanches_. we also expect this timescale to manifest itself in the power spectrum of the avalanches ; and we expect it to vary strongly with the level and nature of disorder in the sandpile .
this work is in progress , as are efforts to relate the timescale found above to a characteristic spatial signature for large events .
we present in figure [ fig6 ] the normalised _ equal - time _ height - height correlation function @xmath42 for a disordered sandpile .
this shows that the height deviations ( from the instantaneous expected column heights ) , in a disordered sandpile with @xmath38 , are positively correlated over about @xmath43 columns , but also have a range where they are negatively correlated . in the inset we plot the related function @xmath44 . for separations
@xmath45 much less than the correlation length of the system , we should have @xcite : @xmath46 and , therefore , we would expect the function in the inset of figure [ fig6 ] to manifest a similar @xmath45-dependence .
a linear fit to points with @xmath47 ( shown by the line in the inset of figure [ fig6 ] ) indicates a power - law dependence of the form @xmath48 for @xmath49 , implying that @xmath50 .
as we will see below , this corresponds to the spatial roughening exponent of an _ ordered _ sandpile .
an explanation of this behaviour is included in the next section .
the hypothesis of dynamical scaling for sandpiles assumes that the roughening process occurs in two stages .
first , the surface roughening is time - dependent , eq .
( [ betaeqn ] ) ; then once the roughness becomes temporally constant , the surface is said to saturate , and all further deposition results in surface fluctuations governed by eq .
( [ alphaeqn ] ) .
however , there is a subtlety concerning the first ( i.e. time - dependent ) stage ; `` early '' model sandpiles are wedge - shaped and the transition to saturation is accompanied by a gradual build - up to a pile that has a single , sloping surface with a suitable angle of repose .
we have taken this process into account to measure the dynamic exponent @xmath14 , eq .
( [ betaeqn ] ) ; in this case surface widths are evaluated from the sloping portion of the pile . for the roughening exponent @xmath18 , eq .
( [ alphaeqn ] ) , we have measured surface widths from mature piles that have only a sloping surface .
our results are : * for disordered sandpiles ( @xmath51 ) we find @xmath52 ; for ordered sandpiles ( @xmath51 ) @xmath53 . * for disordered sandpiles
above a crossover size of @xmath54 we find @xmath55 ; while for ordered piles we find @xmath56 . *
based on the above values we find the dynamical exponent @xmath16 , has values of @xmath57 and @xmath58 for the disordered and ordered sandpiles .
the variation of the surface width , @xmath11 , as a function of @xmath3 , is shown in a log - log plot in figure [ fig7 ] .
this figure shows clearly the crossover in @xmath18 as a function of system size , for disordered sandpiles ; the scaling behaviour of ordered sandpiles is shown for comparison .
disordered sandpiles with sizes below @xmath59 have @xmath60 ; this is in accord with earlier work , @xcite where the second peak in the avalanche spectrum appeared only for disordered piles above crossover .
the existence of this crossover length has been variously interpreted as a length related to reorganisation in the boundary layer of a sandpile @xcite or to variations in the angle of repose in a ( disordered ) sandpile @xcite .
disorder appears to be crucial for the existence of such experimentally observed crossovers , since for example ordered models @xcite , @xcite show no crossover in their measurements of @xmath18 . the crossover effect in a disordered sandpile is also indicated by the height - height correlation function ( figure [ fig6 ] ) . for separations @xmath61 in disordered sandpiles ( with length @xmath62 ) ,
the exponent @xmath18 obtained from the small @xmath45 behaviour of the correlation function is that of the ordered sandpile .
this suggests that even in disordered sandpiles grains which are within a crossover length , @xmath59 , of each other tend to order i.e. the examination of the height - height correlation function for separations @xmath63 ( figure [ fig6 ] ) or the direct measurement of @xmath18 for system sizes @xmath64 , as reported above , yields the exponent of the ordered sandpile , @xmath65 .
the above values indicate that while there is not a change of universality class as one goes from an ordered to a disordered sandpile ( @xmath16 stays the same , within the error bars ) , the disordered pile is clearly rougher with respect to both temporal and spatial fluctuations ( @xmath18 and @xmath14 higher ) .
it is important to note that our measurements of surface exponents are taken over many realisations of the surfaces concerned .
thus , even though , as demonstrated in earlier sections , the surface of a disordered sandpile is temporarily smoothed by the propagation of a large avalanche , it begins to roughen again as a result of deposition ; the values of @xmath18 and @xmath14 that we measure are averages over millions of such cycles and hence reflect the roughening of the interface , in an average sense .
by contrast , no abnormally large events occur for the ordered sandpiles and this is reflected by the lower values of fluctuations and exponents .
the most striking aspect of these exponents is that they indicate that our present cellular - automaton model is a discrete version of earlier continuum equations @xcite , which were formulated independently , to model the pouring of grains onto a sloping surface .
the exponents for our disordered pile are within error bars , exactly those that were measured for the height fluctuations of the surface in case 2 , in ref .
@xcite , while those for the ordered pile are exactly those that were measured for the fluctuations of the avalanches generated by the mobile grains , in the same case .
this is in accord with the notion that the avalanches which flow on an ordered pile generate only mobile grains on the otherwise ordered surface , while as we have demonstrated above , avalanches that flow on a disordered pile , also _ change the configuration of the surface _ by altering the distribution of height fluctuations ( measured by the surface widths ) .
we are exploring these analogies further , but note that this agreement is already a strong validation of both models .
we have presented a thorough investigation of the effects of avalanching on a sandpile surface , focusing on the interrelationship between the nature of the avalanches and the surfaces they leave behind .
we have also postulated a principle of dynamical scaling for sandpile surfaces , and measured the roughening exponents for a sample disordered sandpile .
finally , we have related the characteristics of avalanching in our model system to those obtained experimentally .
our current investigations concern several questions left unanswered above .
these include the dependence of the crossover length @xmath59 on the disorder in the pile ; as well as a fuller investigation of the effect of the nature of disorder ( i.e. whether boundary or uniform ) .
we would expect our correlation functions to depend strongly on the nature and magnitude of the disorder and we are undertaking a full quantitative study .
lastly , we hope that an extension of the present analysis to higher dimensions will yield more extensive comparisons with experiments than is presently available .
h. m. jaeger , c. liu and s. r. nagel , phys .
lett . * 62 * , 40 ( 1989 ) ; g. a. held , d. h. solina , d. t. keane , w. j. haag , p. m. horn , and g. grinstein , phys .
* 65 * , 1120 ( 1990 ) .
h. m. jaeger , s. r. nagel and r. p. behringer , rev .
68 * , 1259 ( 1996 ) .
adrien daerr and stphane douady , nature * 399 * , 241 ( 1999 ) .
r. e. snyder and r. c. ball , phys .
e * 49 * , 104 ( 1994 ) .
anita mehta , j. m. luck and r. j. needs , phys .
e * 53 * , 92 ( 1996 ) .
p. biswas , a. majumdar , anita mehta , and j. k. bhattacharjee , phys .
e * 58 * , 1266 ( 1998 ) .
anita mehta and g. c. barker , europhysics lett . *
27 * , 501 ( 1994 ) ; anita mehta , g. c. barker , j. m. luck and r. j. needs , physica a * 224 * , 48 ( 1996 ) .
l. p. kadanoff , s. r. nagel , l. wu and s .-
m zhou , phys .
a * 39 * , 6254 ( 1989 ) ; p. bak , c. tang and k. wiesenfeld , phys .
a * 38 * , 368 ( 1988 ) .
anita mehta , physica a * 186 * 121 , ( 1992 ) ; anita mehta , in _ granular matter : an interdisciplinary approach _ , 1 , ed .
anita mehta , ( springer - verlag , new york , 1994 ) . throughout this paper
we refer to disordered sandpiles described in reference @xcite with parameters @xmath68 and @xmath69 , unless otherwise stated .
anita mehta in _ structure and dynamics of materials in the mesoscopic domain _
, eds . m. lal et al .
, 340 , ( imperial college press and the royal society , london , 1999 )
. j. krug , adv .
phys . * 46 * , 1 ( 1997 ) .
g. c. barker and anita mehta ( unpublished ) .
g. c. barker and anita mehta phys .
e * 53 * , 5704 ( 1996 ) .
anita mehta and g. c. barker , phys .
* 67 * , 394 ( 1991 ) .
j. b. knight , c. g. fandrich , c. n. lau , h. m. jaeger and s. r. nagel , phys .
e * 51 * , 3957 ( 1995 ) ; e. r. nowak , j. b. knight , e. ben - naim , h. m. jaeger and s. r. nagel , phys .
e * 57 * , 1971 ( 1998 ) . v. frette , k. christensen , a. malthe - sorenssen , j. feder , t. jossang and p. meakin , nature * 379 * , 49 ( 1996 ) .
l. sander , in _ solids far from equilibrium _
, ed . c. godrche , ( cambridge university press , cambridge , 1991 ) .
s. r. nagel , rev .
phys . * 64 * , 321 ( 1992 ) .
m. paczuski and s. boettcher , phys .
lett . * 77 * , 111 , ( 1996 ) .
t. hwa and m. kardar , phys . rev .
a * 45 * , 7002 , ( 1992 ) . | we describe the surface properties of a simple lattice model of a sandpile that includes evolving structural disorder .
we present a dynamical scaling hypothesis for generic sandpile automata , and additionally explore the kinetic roughening of the sandpile surface , indicating its relationship with the sandpile evolution . finally , we comment on the surprisingly good agreement found between this model , and a previous continuum model of sandpile dynamics , from the viewpoint of critical phenomena .
avalanches are the signatures of instabilities on an evolving surface : regions on a sandpile , for example , which protrude excessively from the surface , get dislodged by such mechanisms .
this most intuitive picture of avalanching @xcite is the one we seek to model and study : although presented here in the context of sandpiles , a similar picture may be relevant to intermittent granular flows along an inclined plane @xcite , or to sediment consolidation @xcite . as deposition occurs on a sandpile surface , clusters of grains
grow unevenly at different positions and roughness builds up until further deposition renders some of the clusters unstable .
these then start toppling , so that grains from an already unstable cluster flow down the sandpile , knocking off grains from other similar clusters which they encounter .
the net effect of this is to wipe off protrusions ( where there is a surfeit of grains at a cluster ) and to fill in dips , where the oncoming avalanche can disburse some of its grains . in short , the surface
is smoothed by the passage of the avalanche so that there is a rough precursor surface , and a smoothed post - avalanche surface . in earlier work @xcite , @xcite
, we have explored this issue via an analytical model involving coupled continuum equations .
here we use a cellular - automaton model @xcite of an evolving sandpile to look in more depth at the mechanisms by which a large avalanche smooths the surface .
our sandpile model is a disordered and nonabelian version of the basic kadanoff cellular automaton @xcite ; a further degree of freedom , that involves granular reorganisation within columns , is added to the basic model which includes only granular flow between columns . in this sense , each column is regarded as a cluster of grains so that we represent intracluster as well as intercluster grain relaxations , in accord with a previous understanding of sandpile dynamics @xcite .
our disordered model sandpile @xcite is built from rectangular lattice grains , that have aspect ratio @xmath0 arranged in columns @xmath1 with @xmath2 , where @xmath3 is the system size .
each grain is labelled by its column index @xmath1 and by an orientational index @xmath4 or @xmath5 , corresponding respectively to whether the grain rests on its larger or smaller edge .
the dynamics of our model have been described at length elsewhere @xcite but we review the essentials here : * grains are deposited on the sandpile with fixed probabilities of landing in the @xmath4 or @xmath5 position . * the incoming grains , as well as all the grains in the same column , can then flip to the other orientation stochastically ( with probabilities which decrease with depth from the surface ) .
this flip , or change of orientation , is our simple representation of _ collective dynamics in granular clusters _ since typically clusters reorganise owing to the slight orientational movements of the grains within them @xcite .
* column heights are then computed as follows : the height of column @xmath1 at time @xmath6 , @xmath7 , can be expressed in terms of the instantaneous numbers of @xmath4 and @xmath5 grains , @xmath8 and @xmath9 respectively : @xmath10 * finally , grains fall to the next column down the sandpile ( maintaining their orientation as they do so ) if the height difference exceeds a specified threshold in the normal way @xcite ( the pile is local , limited and has a fall number of two ) . at this point , avalanching occurs . in earlier papers , @xcite
, it was shown that the presence of disorder led , for large enough system sizes , to a preferred size of large avalanches . in the absence of disorder ,
scale - invariant avalanche statistics were observed . in this work
, we focus on the state of the sandpile surface @xcite in a bid to correlate its evolution with the onset and propagation of avalanches . |
almost all the analysis of the decaying light - curves of gamma - ray burst ( grb ) afterglows is done within the framework of external shocks driven into the circumburst medium by ultrarelativistic ejecta ( mszros & rees 1997 ) whose kinetic energy is the same in all directions .
a non - isotropic distribution of the energy per solid angle within the outflow is a natural extension of the afterglow model .
fireballs whose kinetic energy and initial lorentz factor fall - off with angle are power - laws have been considered for the first time by mszros , rees & wijers ( 1998 ) , who studied the effect of such distributions on the afterglow light - curve decay .
the faster dimming of the afterglow emission that a structured outflow can produce has been used by dai & gou ( 2001 ) to explain the steep fall - off of the optical light - curve of the afterglow 991208 .
postnov , prokhorov and lipunov ( 2001 ) have suggested that grb outflows may have a universal angular structure , the observed distribution of isotropic @xmath2-ray outputs ( which has a width of 3 dex ) being due to the observer location .
rossi , lazzati & rees ( 2002 ) and zhang & mszros ( 2002 ) have proposed that the light - curve breaks seen in several grb afterglows and the narrow distributions of the grb energy release ( frail 2001 ) and jet kinetic energy ( panaitescu & kumar 2002 ) may be due to the angular structure of fireballs . in this work , we present an analytical treatment of the afterglow light - curves from structured fireballs ( [ analytical ] ) , focusing on axially symmetric fireballs endowed with a power - law distribution of the energy ( [ plaw ] ) . in
[ criteria ] we give criteria which can be used to assess from the afterglow properties when structure and collimation of grb outflows is required by observations , and apply these criteria to the afterglows whose optical light - curves exhibited a break .
section [ models ] presents the numerical modeling of two grb afterglows , 990510 and 000301c , in the framework of structured jets , leading to a few important conclusions about the role of structure in these two cases .
in this section we calculate the evolution of the afterglow light - curve index @xmath3 , defined as the logarithmic derivative with respect to the observer time @xmath4 of the received flux @xmath5 ( @xmath6 ) for a fireball endowed with structure .
our aim is to obtain the dependence of of @xmath3 and its asymptotic values at early and late times on the fireball s angular structure , the sharpness of the afterglow light - curve break that the structure can produce .
the calculation of the afterglow light - curve requires the following ingredients : dynamics of the fireball , spectrum of its emission , and integration over the equal photon - arrival - time surface . for analytical calculations
, we shall ignore the tangential motions and mixing in the fireball and consider a simplified scenario where a fluid patch travels as if it were part of a uniform fireball . for simplicity ,
in our analytical calculations of the light - curve index we consider only adiabatic grb remnants , a case encountered when electrons acquire a negligible fraction of the post - shock energy or if they cool on a timescale longer than the dynamical time .
radiative losses could be important during the early afterglow , in which case the results presented in this section should be re - derived taking into account their effect on the evolution of the fireball lorentz factor @xmath2 [ ( [ gm ] ) ] .
this effect is included in the numerical calculations presented in [ 0510 ] and [ 0301 ] . for an adiabatic fireball ,
energy conservation gives that the lorentz factor of a fluid patch moving in the direction ( @xmath7 ) ( @xmath8 and @xmath9 being the polar and azimuthal angles in a spherical polar coordinate system aligned to the direction toward the observer ) decreases with its radial location @xmath10 according to : @xmath11^{1/2}\ , r^{-(3-s)/2 } \ ; , \label{gm}\ ] ] where @xmath12 is a constant , independent of direction , @xmath13 is the kinetic energy per solid angle of the ejecta moving in the ( @xmath7 ) direction , and @xmath14 describes the type of external medium : @xmath15 , @xmath16 for a homogeneous medium and @xmath17 for a wind ejected at constant speed and mass - loss rate .
equation ( [ gm ] ) holds for @xmath10 larger than the radius at which the fireball deceleration sets in and until the fireball becomes semi - relativistic .
hereafter we will focus on this case , as it applies to observer times of at least 10 days , when the afterglow light - curve breaks are seen .
the photons emitted by a patch ( @xmath18 ) arrive at observer at a time @xmath4 given by integrating @xmath19 where @xmath20 is the velocity of the patch , and the speed of light is unity . in the limit of relativistic motion ( @xmath21 )
most radiation received by observer comes from the shocked medium moving at small angles ( @xmath22 ) relative to the line of sight . with these two assumptions , equation ( [ tobs ] )
leads to the following equation for the radius @xmath10 of the patch moving in direction ( @xmath23 ) as a function of observer time @xmath24 the peak synchrotron flux @xmath25 received from a @xmath26 patch is proportional to @xmath27 , where @xmath28 is the doppler boost factor , @xmath29 is the comoving frame peak intensity at frequency @xmath30 and @xmath31 is the area of the patch .
the @xmath29 is proportional to the product @xmath32 of the surface density of energized electrons and the magnetic field within the shocked external medium .
therefore , the flux received at some frequency @xmath33 above the self - absorption frequency is @xmath34 where the integral is over all directions of ejecta motion , @xmath35 @xmath36 assuming a magnetic energy density that is a constant fraction of the post - shock energy density , ( which is the product the typical random lorentz factor acquired by protons , @xmath2 , and the comoving proton density , @xmath37 ) , and the factor @xmath38 is the ratio of the flux at frequency @xmath33 to that at the peak @xmath39 .
the time dependence of the flux given in equation ( [ fnu1 ] ) is due to the variation of @xmath2 , @xmath40 , and @xmath41 with the location @xmath10 of the patch , which is related to the observer time through equation ( [ time ] ) . by substituting @xmath40 and @xmath41 in equation ( [ fnu1 ] ) ,
the flux becomes @xmath42 the spectral factor @xmath43 is determined by the distribution of the radiating electrons .
this distribution is shaped by the continuous injection in the downstream region of shock - accelerated electrons with a power - law distribution in energy , @xmath44 , above a minimum electron energy @xmath45 , and by radiative cooling .
the resulting electron distribution exhibits a cooling break at an energy @xmath46 for which the cooling timescale ( @xmath47 ) equals the age of the fireball ( @xmath48 ) , therefore @xmath49 . above @xmath46
the effective electron distribution is @xmath50 if @xmath51 ( kardashev 1962 ) or @xmath52 if @xmath53 .
consequently , the synchrotron spectrum is piecewise power - law , with breaks at the injection frequency @xmath54 and cooling frequency @xmath55 : @xmath56 for @xmath57 , while for @xmath53 @xmath58 if @xmath33 is above both spectral breaks , then @xmath59 for either ordering of @xmath60 and @xmath61 .
thus , the time dependence of the factor @xmath38 is determined by those of the spectral breaks . after using equations ( [ gm ] ) and ( [ bmag ] )
, one obtains : @xmath62 for a given fireball structure @xmath63 , one can calculate the afterglow light - curve @xmath64 by integrating numerically equation ( [ fnu2 ] ) with @xmath10 , @xmath2 , and @xmath38 corresponding to arrival time @xmath4 and direction @xmath65 as given by equations ( [ time ] ) , ( [ gm ] ) , ( [ spek1])([spek3 ] ) , and ( [ nu ] ) . for
now we simplify the integral in equation ( [ fnu2 ] ) by using an approximate solution to equation ( [ time ] ) . in that equation , the second term in the left hand side is dominant for small @xmath8 , while the first term dominates at large @xmath8 .
the two terms become equal for @xmath66 $ ] , which describes a curve @xmath67 on the equal photon arrival - time surface of time @xmath4 given by @xmath68 from equation ( [ gm ] ) , the above expression for @xmath10 shows that @xmath69 also satisfies @xmath70 . for @xmath71 , equations ( [ gm ] ) and ( [ time ] ) give
@xmath72 while for @xmath73 @xmath74 the integral in equation ( [ fnu2 ] ) can then be approximated as the sum of an inner " integral over the @xmath75 disk , where @xmath10 and @xmath2 are given by equation ( [ rg1 ] ) , and an
outer " integral over directions outside this disk , where @xmath10 and @xmath2 are given by equation ( [ rg2 ] ) .
it can be shown by numerical integration that these two integrals have the same time dependence , thus the calculation of the afterglow light - curve can be simplified by retaining only the inner integral .
the product @xmath76 increases with @xmath8 , thus its maximal value on the inner disk is @xmath77 , therefore the factor @xmath78 in equation ( [ fnu2 ] ) is approximately unity . with these simplifications ,
the afterglow flux becomes : @xmath79 where @xmath43 is given by equations ( [ spek1])([spek3 ] ) with @xmath80 therefore the afterglow light - curve has the form @xmath81^{g(p , s ) } \ ; , \label{fnu4}\ ] ] where the exponents @xmath82 and @xmath83 depend on the electron index @xmath84 and type of external medium .
hereafter we shall consider fireballs whose kinetic energy per solid angle @xmath13 is axially symmetric , uniform within a core of size @xmath85 , and a power - law outside it : @xmath86 where @xmath87 and the angle @xmath88 measured from the symmetry axis of the fireball is related to the direction ( @xmath23 ) through a rotation : @xmath89 @xmath90 being the angle between the observer s line toward the fireball center and the fireball symmetry axis .
as shown by granot ( 2002 ) , the emission from the uniform core depends very weakly on the observer location .
the emission received from outside the fireball core should depend even less on the observer s off - set @xmath90 , thus , to a good approximation , we can consider @xmath91
. then equation ( [ tildelta ] ) gives @xmath92 } & { \rm if } \;\ ; \theta_c < \tildelta \end{array } \right .
\;.\ ] ] at early times , when @xmath93 , the afterglow emission arises from within the uniform fireball core . in this case , the afterglow light - curve is @xmath94 , which leads to @xmath95 with @xmath96 these are the usual results for a uniform fireball . at late times ,
when @xmath97 , the contribution of the ejecta outside the core to the afterglow emission is proportional to @xmath98 if @xmath99 , then the first term in the right hand side dominates , while for @xmath100 , the last term is the largest and has the same time dependence as the emission from the fireball core . using the expression of @xmath101 for each case , the transition value of the structural parameter is @xmath102 for @xmath103 , the fall - off of @xmath104 is sufficiently shallow that the emission from the ejecta outside the core is dominant , the afterglow light - curve @xmath105 decaying as @xmath106 with @xmath107 for a homogeneous medium and @xmath108 for a wind - like medium . for @xmath109 ,
@xmath104 decreases sufficiently fast away from the fireball axis that the afterglow emission is dominated by the uniform core and/or by the ejecta just outside the core . in either case
, the afterglow light - curve @xmath110 decays as @xmath106 with @xmath111 taking into account that for most well - observed afterglows , the electron index is between 1.5 and 2.5 , it follows that @xmath112 .
therefore , for fireballs that do not undergo significant lateral spreading at the observing time , the asymptotic index @xmath113 of the afterglow light - curve is independent of the fireball power - law structure if @xmath114 .
this is true also for observers located outside the fireball core ( [ offcore ] ) , at times when the core is visible . and at an angle @xmath8
relative to the direction toward the observer , becomes visible when @xmath115 , when the cone of its relativistically beamed emission starts to contain the direction toward the observer ] equations ( [ a1 ] ) and ( [ a20])([a2 ] ) allow the calculation of the steepening of the light - curve index @xmath116 produced by the passage of the cooling frequency @xmath61 through the observing band or by the core edge becoming visible .
for the former , @xmath117 either if the entire visible region of the fireball is within its core or if the core edge is observable but @xmath109 . for @xmath103 , the passage of @xmath61 ( which decreases in time for a homogeneous medium , but increases if the medium is wind - like ) yields @xmath118 . because @xmath76 increases with @xmath8 and @xmath119 , when the core edge becomes visible ( @xmath120 ) the core edge is outside the inner disk ( @xmath121 ) .
therefore @xmath122 satisfies equation ( [ rg2 ] ) , which , together with @xmath123 , yields @xmath124^{1/(3-s ) } \;. \label{tc}\ ] ] for @xmath125 the light - curve break across @xmath126 is @xmath127 for a homogeneous medium and @xmath128 for a wind , while for @xmath103 the break is @xmath129 therefore the maximal light - curve break that a non - spreading , power - law fireball can yield to an observer located with its uniform core is @xmath130 for a homogeneous medium and @xmath131 for a wind ( panaitescu , mszros & rees 1998 ) . the evolution of the light - curve index @xmath3 , obtained by calculating numerically the integral of equation ( [ fnu2 ] ) ( without the simplifications made subsequently ) , for an on - axis observer is shown in figure 1 for some values of the structural parameter @xmath132 . before the edge core becomes visible , the index @xmath133 given by equation ( [ a1 ] ) .
when the core edge becomes visible , at @xmath134 for @xmath16 and @xmath135 for @xmath17 ( @xmath136 being the observer frame deceleration timescale for the fluid moving toward the observer : @xmath137 ) , the index @xmath138 increases toward the @xmath113 given in equations ( [ a20 ] ) and ( [ a22 ] ) for @xmath139 and that given in equation ( [ a2 ] ) for @xmath140 . as shown in figure 1 , 80% of the light - curve decay steepening across @xmath126 lasts a factor 46 in time for @xmath16 and a factor 710 for @xmath17 . the slower transition for a wind - like medium is caused by the slower decrease of the fireball lorentz factor with time , given in equations ( [ rg1 ] ) and ( [ rg2 ] ) : for @xmath16 , @xmath141 if @xmath142 and @xmath143 if @xmath144 , while for @xmath17 , @xmath145 and @xmath146 , respectively
. the fireball deceleration being slower for a wind - like medium , it takes a longer time for the core edge to become fully visible , thus the transition between the asymptotic light - curve indices is smoother .
figure 1 also shows the light - curve steepening for a fireball with @xmath147 , corresponding to a uniform , collimated outflow with a sharp edge , undergoing lateral spreading . due to the widening of the jet aperture ,
@xmath148 is not constant , but increases slowly in time .
furthermore , @xmath149 is larger than without lateral spreading , reaching @xmath150 ( rhoads 1999 ) . for a uniform ,
spreading jet , interacting with a homogeneous medium , the fastest 70% of the analytically expected steepening @xmath151 is acquired over a factor 20 in time if @xmath152 and 10 if @xmath153 , while for a wind - like medium the corresponding factors are 100 and 1,000 , respectively .
therefore the `` jet - break '' transition is significantly slower and smoother if the surrounding medium is the wind that a massive grb progenitor expelled before the burst ( kumar & panaitescu 2000 ) . for observers located outside the uniform fireball core
, the pre - break light - curve fall - off is mitigated as the more energetic ejecta located closer to the fireball axis become visible , therefore the early time light - curve index @xmath154 decreases in time ( see also the light - curves presented by granot & kumar 2003 and wei & jin 2003 ) .
the fireball axis is seen at a time satisfying @xmath155^{1/(3-s ) } \;,\ ] ] similar to the time @xmath126 when the core edge becomes visible to an on - axis observer ( ) but with the core size @xmath85 replaced by the angle @xmath90 between the fireball axis and the center
observer line ( rossi , lazzati & rees 2002 ) .
well after @xmath156 , when the beaming factor @xmath157 is almost the same across the fireball core , the light - curve index for an off - core observer should reach that given in equations ( [ a20])([a2 ] ) for an on - axis observer .
thus , the flattening of the light - curve fall - off at @xmath158 yields larger light - curve breaks @xmath159 than for an on - axis observer .
this is the most important feature arising from the structure of the outflow .
figure 2 shows the evolution of @xmath138 obtained from equation ( [ fnu2 ] ) for an off - core observer located at @xmath160 .
as can be noticed , a stronger fall - off of the ejecta energy away from the fireball axis ( a larger parameter @xmath132 ) , leads to a more prominent flattening of the afterglow light - curve at @xmath161 , where @xmath162 for @xmath16 and @xmath163 for @xmath17 .
also as expected , at @xmath164 the index @xmath138 asymptotically reaches the values for an on - axis observer . note that , for the same parameter @xmath132 , the light - curve break @xmath159 across @xmath156 is larger for a homogeneous medium , and that the transition between the lowest value of @xmath138 and the @xmath113 at late time takes about a decade in time for a homogeneous medium and about two decades for a wind - like medium .
the slow transition in the latter case ( see also dai & gou 2001 and granot & kumar 2003 ) suggests that , for a wind - like stratified medium , light - curve breaks arising from the structure of the outflow may be too shallow compared the light - curve steepenings observed in some afterglows , which last less than a decade in time . the minimal value @xmath165 reached by the light - curve index before the break depends not only on the structural parameter @xmath132 , as shown in figure 2 , but also on the location of the observer , through the ratio @xmath166 , and on the slope @xmath84 of the electron distribution . as illustrated in figure 3
, the observer location has a much stronger effect on the sharpness of the light - curve break produced by the fireball structure if the external medium is homogeneous . in this case , for observer directions further away from the fireball axis , @xmath165 decreases and the transition between @xmath165 and the @xmath113 at late times lasts shorter ( relative to the break - time @xmath156 ) , the light - curve break becoming sharper . consequently , for a homogeneous medium , higher observer offsets will accommodate easier some of the observed sharp breaks . however , given the expected correlation between the grb peak flux and the energy of ejecta moving toward the observer , large observer offsets render the burst less likely to be detected .
furthermore , for homogeneous media , as shown in figure 3 , large offsets also yield pre - break light - curve indices that are too small compared to those observed .
taking @xmath167 as a more likely case , so that the resulting grb is sufficiently bright and the afterglow break sufficiently sharp , we show in figure 4 the dependence of the pre - break minimum light - curve index @xmath165 on the structural parameter @xmath132 and on the electron index @xmath84 . because most of the effect of the latter is through the @xmath168 factor in equation ( [ a1 ] )
, the @xmath169 has been offset by its value @xmath170 for a uniform fireball , to illustrate the less trivial effect of @xmath84 on @xmath165 .
the minimum light - curve index @xmath165 shown in figure 2 should be close to the pre - break index usually inferred by fitting the afterglow light - curve with a smoothed broken power - law ( beuermann 1999 ) , while the post - break index inferred from observations is either the @xmath113 given in equations ( [ a20])([a2 ] ) for a structured fireball , or the electron index @xmath84 for a jet ( uniform or with structure ) whose edge is visible .
furthermore , one can determine from multiband optical observations the afterglow intrinsic spectral slope @xmath171 , defined as @xmath172 , which depends only on the electron index @xmath84 : @xmath173 if there is a significant dust reddening in the host galaxy , @xmath171 can be determined with the aid of the @xmath174-ray spectral slope , if available , or of fits to the optical spectrum , if it is sufficiently curved to constrain the host extinction for an assumed dust - reddening law ( jensen 2001 , fynbo 2001 ) .
thus , observations provide three characteristics of the afterglow emission : pre- and post- break light - curve indices @xmath165 and @xmath113 , and spectral slope @xmath171 , which constrain three major fireball parameters : the structural index @xmath132 , the observer location @xmath166 , and the electron index @xmath84 .
an analytical determination of @xmath132 from the measured asymptotic light - curve indices is hampered by the complicated dependence of the pre - break index @xmath165 on @xmath132 and observer location ( figures 2 and 3 ) and by that the post - break index @xmath113 could be affected by the lateral spreading of the outflow . the observer location @xmath175 considered in figures 2 and 4 can be taken as representative , but , for a more secure approach to determining the fireball structure , one should fit the observed break with a numerically calculated afterglow light - curve , allowing for a free observer location . before proceeding on this path
, we discuss below how the observed asymptotic light - curve indices @xmath154 and @xmath113 and the spectral slope @xmath171 can be used to assess if a structured fireball is required by the observations and if the lateral spreading of the outflow is significant .
we make here the _ assumption _ that the asymptotic pre- and post - break light - curve indices are determined only by the fireball / jet angular structure , radial distribution of the external medium density , and slope of the shock - accelerated electron distribution . in principle
there could be other factors that determine these indices , such as : a refreshed shock produced by slower ejecta catching - up with the grb remnant ( panaitescu , mszros & rees 1998 ) , which may mitigate the afterglow dimming rate if sufficient energy is injected , or a dependence on the ejecta lorentz factor of the electron distribution index or fractional energy acquired by electrons or magnetic field . from equations ( [ a1 ] ) and ( [ beta ] ) it follows that , for a uniform outflow , the pre - break light - curve index is @xmath176
thus , _ if the observed @xmath154 is smaller than given above , a structured outflow is required _
( @xmath177 criterion ) , as well as an off - core observer location . given that for a fireball with a decreasing @xmath178
, the post - break light - curve index @xmath113 can not exceed the values given in equation ( [ a2 ] ) , _ if the observed @xmath113 exceeds _ @xmath179 _ then a highly collimated outflow ( jet ) _ , whose boundary becomes visible when the light - curve break is observed , _ is required _ ( @xmath180 criterion ) . in this case
, the post - break light - curve index reaches its maximal value : @xmath181 . from equation ( [ beta ] )
, it follows that @xmath182 combining equations ( [ ab1 ] ) and ( [ alfamax ] ) , the maximum steepening @xmath183 that a uniform jet can produce is @xmath184 thus , _ if the observed @xmath159 exceeds the above value , a structured jet is required _ ( @xmath185 criterion ) .
however , _ if the observed @xmath113 exceeds that given in equation ( [ alfamax ] ) _ , then the light - curve break can not be due entirely to the collimation of the ejecta , and _ another mechanism for light - curve breaks _ , such as a spectral break passing through the observing band , _ is required _
( @xmath186 criterion ) .
this passage will also produce a color change ( @xmath171 increases ) and an achromatic light - curve break . applying the above criteria to the optical emission of afterglows with light - curve breaks ,
it is possible to identify the cases where structured fireballs or jets are required : + @xmath187 from the @xmath177 criterion , angular structured is required by the pre - break index @xmath154 of the afterglows + @xmath188 980519 , 990510 , and 991216 , if @xmath17 , + @xmath189 000301c and 010222 , if @xmath190 , + @xmath191 990123 and 000926 , if @xmath17 and @xmath190 , + @xmath192 from the @xmath180 criterion , a jet undergoing lateral spreading is required by the post - break index @xmath113 of the afterglows + @xmath188 990510 , + @xmath189 990123 , 000926 , and 010222 , if @xmath193 , + @xmath191 991208 if @xmath190 , + @xmath194 from the @xmath185 criterion , a structured jet is required by the steepening @xmath159 observed for the afterglow 990510 , if @xmath17 , + @xmath195 from the @xmath186 criterion , the passage of a spectral feature through the optical domain is required ( li & chevalier 2001 , panaitescu & kumar 2002 ) by the steep post - break index @xmath113 of the afterglows 991208 and 000301c , if @xmath193 and , perhaps , by 000301c if @xmath190 .
equations ( [ ab1])([da ] ) allow one to assess if a structured outflow is necessary or sufficient to explain the pre and post - break optical light - curve indices .
a more conclusive test should investigate if the light - curve steepening produced by a structured fireball is as sharp as observed , and if the fireball emission at different wavelengths is consistent with the observations . for uniform outflows , the latter
is usually done using a snapshot broadband afterglow spectrum , but , for structured outflows , one has to take in account that , at the same observer time , the ejecta moving at different angles have different lorentz factors , thus their synchrotron emissions have with different characteristic frequencies . to perform such a test , the data should be fit numerically , followed by a comparison of the best fit @xmath196 obtained for the various model features that are investigated .
the model used hereafter takes into account the spread in the photon arrival time and doppler boost due to the curvature of the jet surface , radiative losses , the effect of inverse compton scatterings on the cooling frequency , and the self - absorption and interstellar scintillation at radio frequencies .
the uniform jet expands sideways at a comoving frame tangential velocity equal to the sound speed @xmath197 .
thus , the jet opening @xmath198 is given by @xmath199 where @xmath200 is time measured in the comoving frame .
the uniformity of the jet is assumed to be maintained during the jet lateral spreading .
the variation of the energy per solid angle @xmath13 of an infinitesimal ring @xmath201 during @xmath202 is @xmath203 , where @xmath204 is the spreading of the ring during @xmath202 .
therefore , the assumption of jet uniformity at any time is equivalent to that @xmath205 is a @xmath88-independent quantity .
then , from equation ( [ theta ] ) , it follows that the spreading of any infinitesimal ring is assumed to be governed by @xmath206 the collimation of ejecta affects the afterglow emission in three ways .
first , the size of the visible part of the jet ( @xmath207 ) stops increasing when @xmath208 , when the edge is seen , yielding a light - curve steepening ( panaitescu , mszros & rees 1998 ) .
second , the sideways expansion of the jet increases its sweeping area , which alters the jet dynamics ( rhoads 1999 ) .
third , the lateral spreading decreases the surface density of the swept - up medium , moving a fraction of the radiating electrons outside the visible region when the jet edge is not yet seen .
the last two effects are important when @xmath209 ( @xmath210 being the initial jet opening ) , which , with the aid of equations ( [ gm ] ) and ( [ theta ] ) , can be shown to occur slightly _ after _ the radius where the jet edge becomes visible if the external medium is homogeneous , and slightly _ before _ if the medium is wind - like .
the jet lateral spreading contributes to the light - curve steepening caused by seeing the jet edge , but also delays the completion of the steepening . because of the slower deceleration caused by a wind - like medium , this completion lasts longer for @xmath17 than for @xmath16 .
therefore , the light - curve breaks for spreading jets should be smoother than for structured fireballs , the effect being stronger for a wind - like medium than for a homogeneous one ( figures 1 and 2 ) . for a structured fireball
, we divide its surface in @xmath211 rings and calculate the ring location ( @xmath212 ) and width ( @xmath213 ) using equation ( [ deltheta ] ) , with the local sound speed @xmath214 obtained from the shock jump conditions and dynamics ( @xmath215 ) of the ring , and with @xmath216 . for each step
@xmath217 , the deceleration of the ring is calculated as if it were part of a uniform fireball of kinetic energy per solid angle @xmath218 , which is adjusted at each step to account for the ring spreading @xmath219 and radiative losses .
therefore , we ignore any fluid mixing expected to arise from the tangential flow , a factor which granot & kumar ( 2003 ) have found to have a weak effect on the afterglow emission . equation ( [ deltheta ] ) provides a simple one - to - one mapping of the jet structure at radius @xmath10 into that at @xmath220 , and maintains computationally inexpensive the numerical algorithm for the dynamics .
the uniform jet model has six free parameters : the initial energy density per solid angle within the ejecta @xmath221 , the jet initial half - opening @xmath210 , the external particle density @xmath222 ( or the parameter @xmath223 introduced by chevalier & li 1999 for an @xmath0 wind profile ) , the exponent @xmath84 of the electron distribution , and the fractional energies in electrons and magnetic field in the post - shock gas . for observers located within the initial jet opening ( which is a condition for seeing a grb ) ,
this location is irrelevant ( granot 2002 ) .
the structured jet model with a power - law distribution of the energy per solid angle has three additional parameters : the core size @xmath85 , the power - law exponent @xmath132 , and the observer location @xmath90 .
we apply the structured jet model to the two afterglows with the largest observed steepening @xmath159 : 990510 and 000301c . for 990510
, a jet undergoing lateral spreading at the time of observations is required for either type of external medium .
the purpose of modeling this afterglow is twofold : for @xmath16 we will determine how much structure is allowed by the pre - break light - curve index , for @xmath17 we will assess if a structured outflow improves significantly the best fit by sharpening the light - curve steepening .
the latter is also the purpose of modeling the afterglow 000301c with a structured jet , as its sharp light - curve break can not reproduced with a uniform jet and standard electron distribution even if the external medium is homogeneous .
the optical emission of the 990510 afterglow exhibited a break from an index @xmath224 to @xmath225 ( harrison 1999 ) at @xmath226 days ( see figure 6 ) .
the optical and @xmath174-ray slopes , @xmath227 ( stanek 1999 ) and @xmath228 ( kuulkers 2000 ) , indicate that @xmath229 . in the framework of uniform jets ,
we have previously found ( panaitescu & kumar 2002 ) that the optical light - curve steepening can be well accommodated ( @xmath230 for 69 degrees of freedom ) with a jet of initial kinetic energy @xmath231 ergs , initial opening @xmath232 , interacting with a homogeneous medium of density @xmath233 .
we have also found that a wind - like medium can not reproduce the steepness of the optical light - curve break , yielding an unacceptable best fit ( @xmath234 for 69 df ) .
for a _ homogeneous _ external medium , equation ( [ ab1 ] ) shows that @xmath235 for a uniform outflow , which is consistent with the observed index , thus a large structural parameter @xmath132 is disfavored , otherwise the early decay would be too shallow .
further , equation ( [ ab2 ] ) shows that , for a fireball , the post - break index @xmath113 can not exceed @xmath236 , which is too small compared to that observed , thus a spreading jet is required . in this case ,
equation ( [ alfamax ] ) leads to a post - break index @xmath237 , which is consistent with the measured index .
the above conclusion that the pre - break light - curve index of the afterglow of grb 990510 does not allow much structure in a spreading jet is illustrated in figure 5 , showing the @xmath196 of the fits obtained for several combinations of the structural parameter @xmath132 and jet core size - to - opening ratio @xmath238 , and for two particular observer locations : the jet axis and the jet edge . by decreasing the @xmath238 ratio or increasing @xmath132 ,
a stronger jet structure is enforced .
as can be noticed , only jets with @xmath239 for @xmath240 and @xmath241 for @xmath242 provide acceptable fits ( defined by a probability larger than 10% ) , while fits as good as that provided by a uniform jet ( @xmath230 ) require that @xmath243 if @xmath240 and @xmath244 if @xmath245 .
thus , for @xmath16 , the afterglow of grb 990510 is best explained by a structured jet with a variation of the energy per solid angle less than a factor @xmath246 across the jet surface .
the radio , optical and @xmath174-ray light - curves for the best fit obtained with a structured jet are shown in the left panel of figure 6 .
we note that its parameters are similar to those for a uniform jet
. for a _ wind - like _ external medium , equation ( [ ab1 ] ) gives @xmath247 for a uniform outflow , thus structure is required in this case to explain the slower decay observed at early time .
equation ( [ ab2 ] ) shows that a structured fireball yields @xmath248 , slightly smaller than the observed index , suggesting that collimation may also be required to accommodate the post - break decay of the this afterglow .
the right panel of figure 6 shows the best fit obtained with a structured jet interacting with a wind medium .
the jet structure improves the fit by @xmath249 relative to the best fit obtained with a uniform jet , which is statistically significant .
however , the optical light - curves steepen too slowly , overestimating the afterglow flux before and after the break and yielding @xmath250 for 66 df .
we conclude that a wind - like medium is not compatible with the observations of the 990510 afterglow even if the jet is endowed with structure .
the @xmath1-band light - curve index of the 000301c afterglow increased from @xmath251 to @xmath252 ( bhargavi & cowsik 2000 ) , a sharp break occurring at @xmath253 days ( figure 7 ) .
jensen ( 2001 ) found that , at @xmath254 days , the optical spectral slope is @xmath255 after correcting for the host reddening , determined from the curvature of the spectrum and assuming an smc reddening law .
@xmath174-ray observations were not made , thus the location of the cooling frequency is not constrained . if @xmath193 then the pre - break light - curve index ( ) @xmath256 for a uniform outflow would be too small compared with that observed .
therefore @xmath190 seems more likely for the 000301c afterglow . given that structured jets interacting with a wind - like medium produce light - curves that are too smooth , we focus here only on a homogeneous circumburst medium . for @xmath190 and a uniform fireball , equation ( [ ab1 ] ) yields @xmath257 , which is slightly larger than observed ,
thus a structured outflow is only marginally required , while equation ( [ ab2 ] ) leads to @xmath258 , which is well below the observed value , thus a jet is required .
then , according to equation ( [ alfamax ] ) , @xmath259 , which is marginally compatible with the index measured by bhargavi & cowsik ( 2000 ) .
other post - break asymptotic indices reported in the literature are larger ( albeit more uncertain ) , which suggests that a jet model may have difficulties in explaining the steep post - break fall - off of the @xmath1-band light - curve of 000301c . in the framework of uniform jets ,
we have found ( panaitescu 2001 ) that this indeed the case : the best fit obtained for @xmath16 has @xmath260 for 98 df and @xmath261 ergs , @xmath232 , @xmath262 , @xmath263 , failing to produce the observed steepening @xmath264 when the jet edge becomes visible . for this reason
, we have investigated a jet model where the distribution of shock - accelerated electrons has a break , so that a sharp light - curve fall - off is seen after the passage through the optical of the synchrotron characteristic frequency corresponding to this break .
further indication that the distribution of injected electrons is not a pure power - law is provided by the discrepancy between the post - break light - curve indices at radio ( @xmath265 ) and optical frequencies , and also by the @xmath266 color change between 2 and 5 days after the burst ( rhoads & fruchter 2001 ) , which implies a softening of the optical spectral slope @xmath267 that is too fast to be attributed to the passage of the cooling break . the best fit obtained with
the broken power - law injected electron distribution has @xmath268 for 96 df , being marginally acceptable .
figure 7 assesses the ability of a structured jet to accommodate the sharp break of the 000301c afterglow without recourse to a non - standard injected electron distribution .
the new best fit has @xmath269 for 95 df , which is a substantial improvement in comparison with the uniform jet model and a pure power - law electron distribution .
nevertheless , the best fit obtained with a structured jet is not acceptable and , clearly , poorer than the uniform jet model with a broken power - law injected electron distribution .
it does not reproduced well the steep post - break decay of the @xmath1-band light - curve and underestimates the 250 ghz emission .
we note that , for the best fit shown in figure 7 , the cooling frequency falls within the optical domain at a few days , a feature which is required to explain the observed curvature of the optical spectrum ( jensen 2001 ) .
in the standard picture of uniform jets , the measured pre- and post - break light - curve indices and the spectral slope offer an overconstrained problem , as , for a given type of external medium , all these quantities depend only on the exponent of the power - law distribution of shock - accelerated electrons .
the angular structure of grb jets and the observer location relative to the axis of the structured jet affects the pre - break index and , possibly , the post - break index also ( [[a2 ] ] , figures 24 ) . for most well - observed afterglows ,
a uniform jet interacting with a homogeneous external medium provides good fits ( panaitescu & kumar 2002 ) , indicating that the freedom allowed by the outflow structure is usually not required by current afterglow observations .
nevertheless , a simple analysis ( [ criteria ] ) of the light - curve indices and spectral slopes , shows that structured outflows are required if the circumburst medium has a wind - like profile ( @xmath17 ) , as , for such media , the resulting steep pre - break light - curve decay must be mitigated by the outflow angular structure . due to the slower deceleration of jets by wind - like media and the stronger sideways expansion
when the jet edge becomes visible , the light - curve breaks for @xmath17 and uniform jets are too smooth compared with the observations ( kumar & panaitescu 2000 ) .
it is then worth investigating if an angular structure in the outflow , yielding a shallower pre - break light - curve decay , sharpens sufficiently the light - curve steepening to make it consistent with the observations .
figures 2 , 3 , and the right panel of figure 6 show that the steepening of light - curves from structured fireballs / jets interacting with wind - like media lasts more than a decade in time , suggesting that such media are not able to accommodate all the observed afterglow breaks . using numerical calculations of the dynamics and synchrotron emission of structured jets ,
we find ( [ 0510 ] ) that little angular structure is allowed for the afterglow 990510 if the external medium is homogeneous .
although acceptable fits to the emission of this afterglow can be obtained even for a tenfold variation of the kinetic energy per solid angle from its axis to edge , fits of quality comparable to that obtained with a uniform jet require an energy variation across the jet surface of less than a factor @xmath246 .
allowing for structure in the jet improves significantly the best fit to 990510 obtained with a wind - like medium , nevertheless the best fit is not acceptable .
the ability of structured jets to yield larger light - curve breaks was also tested against the strong steepening observed for the afterglow 000301c ( [ 0301 ] ) .
we find that the addition of structure greatly improves the fit obtained with a pure power - law electron distribution ; however it does not fare as well as a jet model with a non - standard , broken power - law electron distribution .
nevertheless , angular structure in a wide jet may be a viable explanation for the shallower light - curve breaks ( rossi , lazzati & rees 2002 ) seen in other grb afterglows .
in fact , as discussed in [ criteria ] , a structured fireball is required by some afterglows if the external medium is wind - like . evidently , the best cases for a structured outflow will be those where the light - curve indices and spectral slopes are not consistent with each other within the framework of uniform fireballs . in such cases , equations
( [ ab1])([alfamax ] ) and the criteria derived in [ criteria ] offer a simple way to assess the importance of structure and collimation in the outflow .
furthermore , equations ( [ a1 ] ) , ( [ a20])([a2 ] ) , and the results shown in figure 5 can be used to constrain the angular distribution of the ejecta kinetic energy .
beuermann , k. 1999 , a&a 352 , l26 bhargavi , s. & cowsik , r. 2000 , apj 545 , l77 chevalier , r. & li , z .- y .
1999 , apj 520 , l29 dai , z. & gou , l. 2001 , apj 552 , 72 frail , d. 2001 , apj 562 , l55 fynbo , j. 2001 , a&a 373 , 796 granot , j. 2002 , apj 570 , l61 granot , j. & kumar , p. 2003
, apj , submitted ( astro - ph/0212540 ) harrison , f. 1999 , apj 523 , l121 jensen , b. 2001 , a&a 370 , 909 kardashev , n. 1962 , aj 6 , no.3 , 317 kumar , p. & panaitescu , a. 2000 , apj 541 , l9 kuulkers , e. 2000 , apj 538 , 638 li , z .- y . & chevalier , r. 2001 , apj 551 , 940 mszros , p. , rees , m.j .
1997 , apj 476 , 232 mszros , p. , rees , m.j . & wijers , r. 1998 , apj 499 , 301 panaitescu , a. , mszros , p. & rees , m.j . 1998 , apj 503 , 314 panaitescu , a. 2001 , apj 556 , 1002 panaitescu , a. & kumar , p. 2002 , apj 571 , 779 postnov , k. , prokhorov , m. & lipunov , v. 2001 , astronomy report 45 , 236 ( astro - ph/9908136 ) rhoads , j. 1999 , apj 525 , 737 rhoads , j. & fruchter , a. 2001 , apj 546 , 117 rossi , e. , lazzati , d. & rees , m.j .
2002 , mnras 332 , 945 stanek , k. 1999 , apj 522 , l39 wei , d. & jin , z. 2003 , a&a , submitted ( astro - ph/0212514 ) zhang , b. & mszros , p. 2002 , apj 571 , 876 | we investigate analytically the effect of an anisotropic distribution of the kinetic energy within a relativistic fireball on the decay of the afterglow emission , focusing on axially symmetric fireballs with a uniform core and a power - law decrease with angle outside the core .
the afterglow fall - off steepens after the core becomes fully visible . for observer directions within the core ,
simple formulae are derived for the afterglow decay after the break , while for off - core observers results are shown graphically .
some criteria for assessing from observations the necessity of an angular structure and/or collimation are given .
applying them to several afterglows with light - curve breaks , we find that jets endowed with structure are required only if the circumburst medium has a wind - like ( @xmath0 ) stratification .
numerical fits to the multiwavelength emission of the afterglows 990510 and 000301c show that , for the former , the angular distribution of the ejecta kinetic energy is not far from uniformity , and that , with a standard power - law electron distribution , the sharp steepening of the @xmath1-band light - curve of the latter is better accommodated by a structured jet than an uniform outflow .
structured outflows may accommodate the shallower light - curve breaks observed in other afterglows . |
coherent transport phenomena in mesoscale conductors with various geometries have attracted much attention over recent years due to their potential in the investigation of various resonance or bound - state features,@xcite imaging coherent electron wave flow , @xcite and electrical switching effects .
@xcite the conductance @xmath0 is a fundamental property of quasi - one - dimensional systems close to twice the quantum unit of conductance @xmath1 , where @xmath2 denotes the charge of an electron , the factor of 2 accounts for spin degeneracy , and @xmath3 is planck s constant .
moreover , the conductance depends sensitively on the particular arrangement of scatterers as well as the applied external fields in the mesoscopic system .
conducting structures subject to magnetic fields or periodically varying voltages are essential fundamental entities in mesoscopic physics .
the momenta of electrons are allowed to undergo a robust change if a magnetic field is applied , which in turn dynamically modifies the transport properties of a quantum system .
it was indicated that the conductance involving aharonov - bohm ( ab ) interference@xcite as a function of magnetic field exhibits step - like structures.@xcite recently , differential conductance of an ab interferometer was measured as a function of the bias voltage.@xcite varying either the magnetic field or the electrostatic confining potentials allows the interference to be tuned .
a mesoscale system driven by an external time - dependent potential allows charge carriers to make coherent inelastic scattering@xcite involving inter - subband and inter - sideband transitions.@xcite a number of interesting time - dependent transport related issues have been investigated such as time - modulated quasibound - state ( qbs ) features , @xcite quantum pumping effects , @xcite and nanomechanical rectifiers .
@xcite if the driving frequency is comparable to the subband level spacing , the pumping becomes nonadiabatic and manifests reversely shifted partial gap in the transmission as a function of energy.@xcite the driving force behind nonadiabatic pumping is the coherent inelastic multiple backscattering involving either absorption or emission of a quantized photon energy .
it is noteworthy that these effects are applicable to design a tunable current source or reversely to create a quantum motor driven by the generated electric current,@xcite or fast manipulations for quantum information processing.@xcite of particular interest are investigations of the interplay between the various effects of electron transport in mesoscopic systems . in this
study the main stress falls on the investigation of tunable quantum magnetoconductance that could be manipulated by the ac - biased quantum point contacts ( qpcs ) in the presence of a magnetic field perpendicular to the two - dimensional electron gas ( 2deg ) plane .
this can be achievable by designing a specific size and geometry of the quantum constriction by controlling split - gate voltages for manipulating the coupling strength between the leads and the open cavity region confined by the qpcs . in sec .
ii , we specify the setup and the geometry of the two - dimensional quantum channel structure , and present the theoretical framework as well as computational approach . in sec .
iii , the main transport spectroscopy features are demonstrated and discussed along with the underlying dynamical mechanisms .
concluding remarks are presented in sec .
the system under investigation is composed of split - gates confined qpcs embedded in a quantum channel with parabolic confining potential @xmath4 , and hence the electrons are transported through the broad channel with characteristic energy scale @xmath5 in the transverse direction . the electrons incident from the reservoirs propagating in the @xmath6-direction impinge on the qpc system scattered by a local time - periodic potential @xmath7 under the influence of a perpendicular magnetic field .
the system is supposed to be fabricated from a modulation doped gaas / algaas heterostructure hosting a 2deg system .
the hamiltonian thus consists of @xmath8 where we choose a value of effective mass @xmath9=@xmath10 of the charge carriers to corresponding to a gaas - based 2deg and @xmath11=@xmath12 denotes the magnetic length of an electron .
the local time - dependent scattering potential @xmath13 contains a static part with spatial dependent strength @xmath14 and a time - dependent part with spatial dependent strength @xmath15 , driving frequency @xmath16 and phase @xmath17 . in the presence of a magnetic field @xmath18=@xmath19 ,
the time - dependent schrdinger equation @xmath20 is inseparable in the ( @xmath21)-coordinates , but is separable in the mixed momentum - coordinate representation , @xcite namely transforming the total wave function @xmath22 into the wave function @xmath23 and expanded in terms of the eigenfunctions @xmath24 of an ideal quantum channel @xmath25 due to the effects of the lorentz force , the eigenfunctions of the parabolic confinement @xmath24 are shifted by @xmath26 with @xmath27 being the reciprocal of the effective magnetic length of the system . here the effective confining strength @xmath28 under a magnetic field is related to the bulk cyclotron frequency @xmath29 and the characteristic frequency @xmath30 for the parabolic confinement .
the resulting equation after the expansion is a coupled nonlocal integral equation in the momentum space describing the electron propagation of an asymptotic state occupying subband @xmath31 along the @xmath6-direction that can be expressed as @xmath32 here the electron energy @xmath33 of the subband @xmath31 contains the subband threshold @xmath34 for conduction ( @xmath31 = @xmath35 , @xmath36 , @xmath37 ) that is determined by the lateral confinement and the effective kinetic energy @xmath38 in the integrand of , the overlap integral @xmath39 constructing the matrix elements of the scattering potential indicates the electrons in the subband @xmath31 making inter - subband transitions to the intermediate states @xmath40 . due to the periodicity in time of the driving field ,
the time - dependent wave function with incident energy @xmath41 and the driving potential can be transformed into the frequency domain , namely @xmath42 and @xmath43 where the quasi - energy @xmath44 with @xmath45 and @xmath46 indicating the indices of sidebands induced by the external driving field .
the magnitude of the wave vector @xmath47 along the @xmath6-direction in the @xmath48 intermediate state can be expressed as @xmath49 this is convenient for us to obtain the multiple scattering identity containing intricate inter - subband and inter - sideband transitions @xmath50 where we have defined @xmath51 for simplicity . from , we can define the green function of the @xmath48 state as @xmath52 and the corresponding incident wave obeys @xmath53 after some algebra , we can obtain the lippmann - schwinger equation for the fourier components of the wave function scattered into the @xmath48 state @xmath54 by taking all the intermediate @xmath55 states into account .
we note that the above equation is not suitable for numerical calculations because the incident wave @xmath56 is a delta function in the fourier space .
to achieve exact numerical calculation , one has to define the @xmath57 matrix @xmath58 that couples all the intermediate @xmath48 and @xmath59 states .
the scattering potential is expanded in the fourier series in and the matrix elements calculated according to .
this yields a connection between the sidebands in the system which can be seen clearly when the matrix elements for the potential are inserted into for the @xmath57 matrix , @xmath60 where @xmath61 from which we can see that adjacent sidebands are coupled .
this allows us to write the momentum space wave function in terms of the @xmath57 matrix @xmath62 the transmission coefficients or amplitudes can be found by constructing the full wave function by inserting eq . back into the expansions done previously , i.e. eqs . and .
after performing the inverse fourier transform into the coordinate representation and the application of residue integration to examine only the contribution of the wave traveling in the + @xmath6-direction , we have the transmission amplitudes @xmath63 the two - terminal conductance is simply obtained by the landauer - bttiker transmission function@xcite @xmath64 \ , , \ ] ] where @xmath65 is constructed from the transmission amplitudes ( [ tnmm ] ) and our notation indicates transmission matrix contributed from the sideband @xmath46 connecting the incident electron flux in the various subbands in the source region to the outgoing electron flux in the subbands in the drain .
in this section we investigate magnetotransport properties of split - gated systems depicted in .
( color online ) schematic diagram of gate - voltage controlled split - gate constriction with no magnetic field : ( a ) sketch of a single split - gated quantum constriction ; ( b ) sketch of a double split - gated quantum constriction.,title="fig:",scaledwidth=45.0% ] ( color online ) schematic diagram of gate - voltage controlled split - gate constriction with no magnetic field : ( a ) sketch of a single split - gated quantum constriction ; ( b ) sketch of a double split - gated quantum constriction.,title="fig:",scaledwidth=45.0% ] the considered single qpc ( sqpc ) system shown in ( a ) can be modeled by the gaussian - shaped potential @xmath66 where @xmath67 is the distances of the gaussian potential peak away from the center of the wire in the @xmath68-direction .
the parameters for the potential in are @xmath69 mev , @xmath70 , @xmath71 , and @xmath72 such that the width of the qpc is @xmath73 nm and the distance of the split - gates is approximately @xmath74 nm . the double qpc ( dqpc ) system shown in ( b ) is described using four gaussian - shaped potentials @xmath75 where @xmath76 are the center coordinates of the gaussian potentials .
the parameters are the same as the except for @xmath77 . to investigate the electronic transport properties under a perpendicular magnetic field , we select the confinement parameter @xmath78 mev .
we assume that the quantum constriction is fabricated in a high - mobility gaas - al@xmath79ga@xmath80as heterostructure such that the effective rydberg energy @xmath81 mev and the bohr radius @xmath82 nm .
length parameters are scaled using the the effective magnetic length at zero magnetic field , referred to as @xmath83 @xmath84 while energy is either fixed in @xmath85 or given in units of the effective confinement strength @xmath86 .
we start by considering a sqpc placed at @xmath87 between two electron reservoirs , as shown in ( a ) .
we assume that the qpc could be induced by metallic split - gates situated on the top of the heterostructure and can be treated as an open structure with distance @xmath88 nm .
( color online ) conductance through a single qpc as a function of energy under magnetic field with strength from 0.0 to 4.0 tesla .
the parameters of the constriction potentials are the same with ( a).,scaledwidth=45.0% ] in , we show the conductance as a function of incident energy under magnetic field . by increasing the magnetic field strength from @xmath89 to @xmath90 t
, the subband threshold is red - shifted around @xmath91 , and the pinch - off regime is also reduced .
moreover , for a given incident electron energy , increasing the magnetic field may enhance the conductance which tends to approach the ideal quantization .
this is because of the formation of one - dimensional edge states in the channel suppressing the backscattering .
since there is no significant interference , we see that the conductance plateaus are monotonically increased as a function of energy for arbitrary magnetic fields implying that no ab oscillations could be induced in such a simple geometry and small source - drain bias regime .
to enhance the interference effects , we consider a dqpc made by two pairs of split - gates located at @xmath92 with @xmath77 [ see ( b ) ] forming a cavity with characteristic length @xmath93 nm .
( color online ) conductance as a function of energy with no magnetic field : ( a ) the conductance for the case of dqpc ( red solid curve ) in comparison with the case of sqpc ( green dashed curve ) ; ( b ) the logarithm conductance for the case of dqpc .
the parameters for the potentials of the dqpc are the same as ( b ) .
the resonances ( a)-(c ) shown by the black arrows are at @xmath94 , @xmath95 , and @xmath96 . , scaledwidth=45.0% ] in ( a ) , we show the conductance as a function of incident energy in the dqpc system with no magnetic field ( red solid curve ) in comparison with the case of a sqpc ( green dashed curve ) . by adding the second qpc ,
the conductance is strongly suppressed in the non - resonant energies .
in addition , it is interesting to see that the conductance brings forth resonance peaks instead of dips.@xcite this implies that the qpc increases the subband threshold , so that electrons with energy in the pinch - off regime manifest resonant transmission feature induced by the cavity formed by the dqpc . to obtain a deeper understanding for the resonance features shown by the black arrows in ( a )
, we plot the logarithm conductance shown in ( b ) .
it is clearly seen that the resonance ( a ) manifests dip structure , while the resonances ( b ) and ( c ) exhibit the fano line - shapes @xcite indicating the interference between the localized and the extended states .
their corresponding probability densities are shown in ( a)-(c ) .
( color online ) probability density for resonances in the conductance marked by ( a)-(c ) in ( a ) with corresponding energies @xmath97 ( a ) @xmath98 ; ( b ) @xmath95 ; ( c ) @xmath96.,title="fig:",scaledwidth=23.8% ] ( color online ) probability density for resonances in the conductance marked by ( a)-(c ) in ( a ) with corresponding energies @xmath97 ( a ) @xmath98 ; ( b ) @xmath95 ; ( c ) @xmath96.,title="fig:",scaledwidth=23.8% ] ( color online ) probability density for resonances in the conductance marked by ( a)-(c ) in ( a ) with corresponding energies @xmath97 ( a ) @xmath98 ; ( b ) @xmath95 ; ( c ) @xmath96.,title="fig:",scaledwidth=23.8% ] the number of probability density peaks within the cavity region implies the order of the resonances formed in the cavity implying that the resonances ( a)-(c ) are the second to the fourth resonances in the cavity .
the ratio of the distance between the nearby peaks to the incident wave length is around @xmath99 , this indicates long - lived resonance modes fitting the cavity in the dqpc .
we now turn to study the magnetotransport properties in the split - gated systems .
( color online ) conductance as a function of energy with @xmath100 .
the other parameters are the same as previous figures.,scaledwidth=45.0% ] in , we present abundant resonance features in conductance of sqpc ( green dashed curve ) and dqpc ( red solid curve ) under magnetic field @xmath101 tesla . for the case of sqpc
, the conductance is monotonically increased in the pinch - off regime ( @xmath102 ) .
the conductance quantization at @xmath103 demonstrates that electrons can be transported coherently within the edge channel without significant backscattering . for the case of a dqpc ,
the conductance manifests resonant transmission peaks in the low kinetic energy regimes of the first and the second subbands , while the conductance exhibits resonant reflection features in the high kinetic energy regime .
the resonance structures in the conductance are more dense due to the magnetic field . to get a better understanding for the on- resonance peaks ( a ) and ( b ) as well as the off- resonance valley structure ( c ) marked by the black arrows in
, we plot their corresponding probability densities in ( a)-(c ) .
( color online ) probability density for the peaks marked in .
( a ) a ring structure inside the cavity due to the magnetic field .
@xmath104 , @xmath105 .
( b ) an edge state .
@xmath106 , @xmath107 .
( c ) a scattering state .
@xmath108 , @xmath105.,title="fig:",scaledwidth=23.8% ] ( color online ) probability density for the peaks marked in .
( a ) a ring structure inside the cavity due to the magnetic field .
@xmath104 , @xmath105 .
( b ) an edge state .
@xmath106 , @xmath107 .
( c ) a scattering state .
@xmath108 , @xmath105.,title="fig:",scaledwidth=23.8% ] ( color online ) probability density for the peaks marked in .
( a ) a ring structure inside the cavity due to the magnetic field .
@xmath104 , @xmath105 .
( b ) an edge state .
@xmath106 , @xmath107 .
( c ) a scattering state .
@xmath108 , @xmath105.,title="fig:",scaledwidth=23.8% ] first , when the electron is transported with very low kinetic energy such as the case of ( a ) in which the electron is occupying the second subband @xmath109 .
we see that the localized states in the cavity can be well established forming double ab - oscillation paths , where the inner path manifests an entangled feature .
in addition , the qbss can be formed at both ends of the open cavity . secondly
, if the electron carries sufficient high kinetic energy such as the case of ( b ) in which the electron is occupying the first subband @xmath110 .
the lorentz force plays a dominant role on the transport such that the electron wave is pushed to the upper confinement and forms an edge state facilitating the flow of electrons through the system by suppressing backscattering in the system . for comparison with the case ( a )
, we show the non - resonance probability feature shown in ( c ) in which the electron is also occupying the second subband @xmath109 .
it is important that , in the non - resonant condition , the lorentz force is able to push the electron a little bit to the upper confinement and qbs can be formed only at the left qpc thus manifesting reflection feature with minimal conductance .
( color online ) conductance @xmath0 as a function of magnetic field @xmath111 for electrons with incident energy @xmath112 through dqpc .
the inset shows that the conductance with small ab oscillations can be seen superimposed on large oscillations .
the other parameters are @xmath113 , @xmath70 , @xmath71 , @xmath114 , and @xmath77.,scaledwidth=40.0% ] in , we show that the conductance versus magnetic field exhibits periodic oscillations .
the period @xmath115 of ab oscillations is inversely proportional to the effective area @xmath116 enclosed by the electron path , given by @xmath117 with @xmath118 being the flux quantum .
@xcite the effective area can be slightly changed by tuning the strength @xmath14 of split gates .
the ab oscillations with large period @xmath119 tesla is associated with the interference between the directly reflected electrons by the left qpc and the electrons go through an enclosed path forming a small area in the left qpc ( area i in ) .
the small oscillations superimposed on the larger ones shown in the inset of are formed due to the interference between the electrons directly reflected by the right qpc and the electrons going through the open cavity forming a large area in the dqpc ( area ii in ) .
( color online ) illustration of the possible paths that the electrons can take and the paths that they enclose in the dqpc system confined by double split gates ( shaded regions).,scaledwidth=30.0% ] we note in passing that our results demonstrate that the ab oscillations do not require a ring geometry .
@xcite interference is the most important effect to generate ab oscillations .
moreover , the robust zero conductance feature at around @xmath120 tesla exhibits that the dqpc could be applicable as a mesoscopic switching device . to explore the time - dependent transport in a dqpc with time - harmonic modulation
, we construct the model by using four gaussian - shaped potentials that are expressed as @xmath121 where strengths of the left qpc @xmath122 and the right qpc @xmath123 contain the same driving frequency @xmath16 with a phase difference @xmath124 .
this driven dqpc system is similar to the one depicted in ( b ) except for the external driving terms with amplitude @xmath15 .
( color online ) conductance versus incident energy with no magnetic field ( @xmath125 ) for a time - harmonic dqpc ( @xmath126 mev and @xmath127 , red solid ) in comparison with that of a static dqpc ( @xmath128 mev , green dashed ) .
the other parameters for the system are @xmath129 , @xmath70 , @xmath71 , @xmath114 , @xmath77 , and driving frequency @xmath130.,scaledwidth=45.0% ] conductance @xmath0 versus phase difference @xmath124 between the left and the right qpcs for the electron with energies marked by ( a ) and ( b ) in .,scaledwidth=45.0% ] in , we show the conductance as a function of incident energy for the time - harmonic dqpc ( @xmath131 mev and @xmath127 , red solid curve ) in comparison with the static dqpc ( @xmath128 mev , green dashed curve ) . under the influence of the time - harmonic driving potential , we find a small side peak in @xmath0 marked by ( a ) indicating that the electron is allowed to emit a photon with energy @xmath132 and jump to a state beneath the resonance , i.e. , the main peak marked by ( b )
. moreover , the electron kinetic energy plays a role to suppress such inter - sideband transitions . as we can see that the small peak becomes a shoulder structure for an electron with incident energy at around @xmath133 . in
, we show the conductance as a function of phase difference between the two qpcs .
the energies are fixed at the main peak @xmath134 ( green dashed curve ) and at the side peak @xmath135 ( red solid curve ) marked , respectively , by ( a ) and ( b ) in .
both cases are not very sensitive to the phase difference @xmath124 , but we see that @xmath124 at around @xmath136 can enhance the inter - sideband transitions .
we have developed a lippmann - schwinger model that has allowed us to explore the magnetotransport and time - dependent transport spectroscopy of coherent elastic and inelastic multiple scattering features relevant to quantum constricted sqpc and dqpc systems under a magnetic field perpendicular to the 2deg . we have demonstrated and analyzed the mechanisms causing the slow and the fast conductance oscillations due to ab interference in the dqpc system .
we hope that our numerical demonstrations on magnetotransport and time - dependent transport could be useful for the utilization of intricate coupling between subbands and sidebands towards the realization of quantum pumping circuits and fast manipulation of quantum information processing in mesoscopic systems .
this work was financially supported by the research and instruments funds of the icelandic state , the research fund of the university of iceland , the icelandic science and technology research programme for postgenomic biomedicine , nanoscience and nanotechnology , and the national science council of the republic of china through contract no .
nsc97 - 2112-m-239 - 003-my3 . | the authors report on modeling of transport spectroscopy in split - gate controlled quantum constrictions .
a mixed momentum - coordinate representation is employed to solve a set of time - dependent lippmann - schwinger equations with intricate coupling between the subbands and the sidebands .
our numerical results show that the transport properties are tunable by adjusting the ac - biased split - gates and the applied perpendicular magnetic field .
we illustrate time - modulated quasibound - state features involving inter - sideband transitions and the aharonov - bohm oscillation characteristics in the split - gated systems . |
the rovibronic spectra of @xmath0c@xmath1 , involving singlet , triplet , and quintet electronic states , are rich and complex in features as well as in anomalies .
this is true despite the fact that @xmath0c@xmath1 , hereafter simply called c@xmath1 , is a homonuclear diatomic molecule containing only 8 valence electrons .
there are 62 electronic states corresponding to the six possible separated - atom limits formed by different pairs of the c(@xmath21p ) , c(@xmath22d ) , and c(@xmath22s ) atoms .
18 of these states , six singlet , six triplet , and six quintet ones , correspond to the c(@xmath21p ) + c(@xmath21p ) asymptote .
as summarized in table 1 and figure 1 , the measured rovibronic spectroscopy of c@xmath1 is characterized by 19 known band systems , and a further three intercombination transitions , which cover the spectral range @xmath23 @xmath4 and now all three spin states . due to the lack of detailed transition data , see section 2.4 for details , only 14 band systems are considered in the present study .
all the states considered , except the and the 4 @xmath24 states , share the same c(@xmath21p ) + c(@xmath21p ) separated - atom limit .
the present study was initiated as , rather unusually , at least six of these bands have been used for astronomical observations , with a further one having been proposed for such use .
there are numerous astronomical observations of c@xmath1 spectra with a multitude of important applications for astrophysics .
c@xmath1 spectra have been observed in comets @xcite , in high - temperature stars @xcite , translucent clouds @xcite , and in the low - temperature interstellar medium @xcite .
high - quality studies of rotational and spatial distributions of c@xmath1 in comets are available @xcite .
the astronomical observations and their modelling are supported by a number of laboratory studies @xcite . to make maximum use of the observations one needs accurate rovibronic energy level information , the principal topic of the present paper .
both the swan @xcite and the deslandres
@xcite bands have been observed in fluorescence from comets .
indeed , two singlet , and , and four triplet , , , , and d @xmath24 , electronic states are needed to model c@xmath1 emission observed in comets . two spin - forbidden transition systems , @xmath25 and @xmath25 , are needed to explain the observed intensities in the swan band @xcite .
the astronomical observations include solar spectra , where c@xmath1 forms an important component of the photospheric carbon abundance @xcite .
so , for example , c@xmath1 can be observed in the sun s photosphere at visible wavelengths using the swan band @xcite and in the infrared via the phillips and ballik ramsay bands @xcite .
transitions in the swan band have also been observed in peculiar white dwarfs @xcite and the coronae borealis star v coronae australis @xcite , while the phillips band is prominent , for example , in the carbon star hd19557 @xcite .
transitions in the ballik ramsay band have also been observed in carbon stars @xcite .
interstellar c@xmath1 has been observed via the infrared phillips band , for example in the perseus molecular complex @xcite and toward the cyg ob2 association @xcite .
emission features from the swan band have been observed in the red rectangle @xcite .
@xcite observed absorbotion in the phillips , mulliken and herzberg f bands in their study of translucent clouds by the simultaneous use of observations from both space and ground - based telescopes . the rovibrational manifolds of the and states strongly overlap as the corresponding electronic excitation energy is less than half of the vibrational spacing in either state .
@xcite suggested that long - wavelength transitions between levels in the
band should provide a good method for monitoring interstellar c@xmath1 .
@xcite attempted to observe such lines in the hale
bopp comet , without success .
the accurate rovibronic energy levels presented in this paper can be used to provide accurate transitions for such features . as for the other bands , at ultraviolet wavelengths the international ultraviolet explorer ( iue )
was used to observe c@xmath1 towards x persei via the herzberg f band @xcite .
the hubble space telescope also has been used to monitor c@xmath1 absorption in the ultraviolet using both the mulliken and the herzberg f bands @xcite .
finally , we note that all of the bands mentioned are important in laboratory plasmas @xcite and flames @xcite .
for example , there are particular vibrational bands of the swan system known as the high pressure ( hp ) bands @xcite which can be prominent in such environments at atmospheric pressure
. studies of the spectra , and thus the band systems , of c@xmath1 date all the way back to @xcite and the dawn of spectroscopy , when c@xmath1 emissions were first observed in flames .
detailed studies by @xcite of the most prominent band of c@xmath1 also predate the development of the quantum mechanical tools required to interpret the spectroscopic results obtained . over the following century and a half ,
a number of other bands have been detected and studied .
indeed , four band systems have been identified during the last decade @xcite .
table [ tab : bands ] lists the known band systems of c@xmath1 relevant for the present study .
a large number of spectroscopic measurements exist for the different band systems of c@xmath1 .
all the studies which contain primary measured transitions at a reasonable level of accuracy are considered during the present analysis .
the experimental papers found and analysed by us are listed here based on the band systems : phillips @xcite , mulliken @xcite , bernath b @xcite , bernath b@xmath26 @xcite , freymark @xcite , ballik
ramsay @xcite , swan @xcite , fox
herzberg @xcite , deslandres dazambuja @xcite , messerle
krauss @xcite , + goodwin
cool @xcite , duck @xcite , krechkivska - schmidt @xcite and radi - bornhauser @xcite , as well as various spin - changing intercombination bands @xcite .
it is interesting to note that for a long time a triplet state , now named , was believed to be the lowest electronic state of c@xmath1 and not a singlet state , the true ground electronic state , .
the source of the confusion , detailed in @xcite , is that the energy difference between the two electronic states is only about one half of the vibrational fundamental of either state so that the swan system can be seen in absorption in many sources .
the spectroscopy and the spectroscopic constants of c@xmath1 have been the subject of several reviews @xcite .
@xcite reported the analysis of seven singlet and seven triplet states . @xcite reviewed the then available experimental and theoretical results .
the spectroscopic and kinetic properties of c@xmath1 , including 23 electronic states studied prior to 1992 , were reviewed by @xcite .
@xcite reviewed the spectroscopy of small carbon clusters which , once again , covered the extensive spectroscopic literature available for c@xmath1 .
the present paper surveys all rovibronically - resolved measurements made up to the end of 2015 . a large number of _ ab initio _ computations have been performed on c@xmath1 @xcite .
notable among the sophisticated electronic structure computations are those of bacskay @xcite ; for example , they led to the observation of the duck band and enabled the identification of the krechkivska
schmidt band system .
@xcite pointed out that `` at the present state of the art , theoretical pecs that reproduce the rotational - vibrational levels to spectroscopic accuracy ( 1 @xmath4 ) or near spectroscopic accuracy ( 10 @xmath4 ) are considered highly accurate . ''
reproduction of measured electronic excitation energies of c@xmath1 appears to be even more problematic , an accuracy of a few hundred @xmath4 seems to be the norm .
thus , it is still of interest to perform accurate quantum chemical computations on c@xmath1 and its rovibronic states .
results of a preliminary , first - principles analysis are reported here ( _ vide infra _ ) , used in particular for checking the experimental transitions and energy levels .
the main body of the present work is a marvel ( measured active rotational - vibrational energy levels ) @xcite analysis of the measured rovibronic states of c@xmath1 .
our analysis has been in progress for four years but a real breakthrough came with the study of @xcite , who observed 16 forbidden transitions between singlet and triplet states .
this study coupled , for the first time , the singlet and triplet components of the observed spectroscopic network ( sn ) @xcite of c@xmath1 , allowing a much improved analysis of its rovibronic energy level structure .
we analyze all the known bands of c@xmath1 with the exception of the five vuv bands due to @xcite and @xcite .
these bands involve upper energy levels which arise from a single experiment and are too high for us to be able to independently validate them .
finally , when a complete set of accurate energy levels are available for a molecule , they can be used , via the direct summation technique , to compute accurate ideal - gas thermodynamic functions , most importantly the high - temperature internal partition function , @xmath27 .
we do this here for c@xmath1 , complementing and improving several previous efforts @xcite and arrive at very precise and accurate values for @xmath27 of c@xmath1 up to 4,000 k. .
the blue , red , and green dots correspond to the singlet , triplet , and quintet rovibronic states , respectively .
the dots represent energy levels and only some of the transitions are visible , especially those connecting the singlet and the triplet as well as the triplet and the quintet rovibrational states.,height=566 ]
the marvel @xcite procedure and code is used in this study to obtain rovibronic energies of c@xmath1 by inverting all assigned experimental rovibronic transitions available in the literature for this molecule .
marvel is based on the concept of spectroscopic networks ( sn ) @xcite : sns are large , finite , weighted , rooted graphs , where the vertices are discrete energy levels ( with associated uncertainties ) , the edges are transitions ( with measured uncertainties ) , and in a simple picture the weights are provided by the transition intensities .
no weights are considered during the present study . within the experimental sn of a molecule
there can be several rooted components and several floating ones .
for many molecules the rooted components belong to ortho and para nuclear spin isomers .
however , the nuclear spin of @xmath0c is zero ; thus , @xmath0c@xmath1 has only one nuclear spin isomer and one root for its lowest electronic state , .
transitions between electronic states of different spin multiplicity are spin forbidden ; thus , until 2011 no lines were measured and assigned experimentally between the singlet and triplet and the triplet and quintet manifold of states of c@xmath1 .
first , @xcite observed transitions linking triplet and quintet electronic states , leading to the first observation and characterization of a quintet band of c@xmath1 .
later , @xcite managed to identify 16 spin - forbidden transitions between singlet and triplet states .
these spin - forbidden transitions proved to be particularly important during the marvel analysis of the experimental spectra of c@xmath1 . using these forbidden transitions the experimental sn of c@xmath1 simplifies as
now it contains only one principal component ( pc ) .
the label of the root of the pc is \{0 + 0 @xmath3 @xmath28 } , for the notation employed for the label see subsection [ s : qn ] .
the measured sn of c@xmath1 is shown pictorially in figure 2 , the singlet , triplet , and quintet energy levels are indicated with different colors . since figure 2 is a particular representation of a network ( @xmath29 , a graph ) , the arrangement of the nodes ( energy levels ) and links ( transitions ) is arbitrary , but clearly displays several important characteristics of the sn of c@xmath1 .
figure 2 vividly shows , for example , how weakly the singlet rovibrational energy levels are connected to the triplet core . during a marvel analysis
we simultaneously process all the available assigned and labelled experimental transitions .
the energy levels are obtained from the set of transitions via a weighted linear least - squares inversion protocol . as the marvel technique has been employed to determine experimental - quality energy levels of nine isotopologues of water @xcite , of three isotopologues of h@xmath30 @xcite , as well as of ammonia @xcite and ketene @xcite , the interested reader is referred to these publications for details about the different stages of a marvel analysis .
since c@xmath1 has both regular and inverted triplet states ( see the next subsection for details ) , for example the and of c@xmath1 are inverted , during the marvel analysis it was checked whether the labels of all experimental transitions follow the same convention . at the beginning , the marvel analysis of the spectra of c@xmath1 was complicated by the fact that there have been no truly high - accuracy variational or effective hamiltonian rovibronic energy levels available for this molecule at higher energies .
this situation greatly improved by the first - principles duo @xcite analysis performed as part of this study ( _ vide infra _ ) , allowing at least a preliminary validation of the marvel levels up to 35,000 @xmath4 . for marvel to work
properly one needs appropriate and unique labels .
the label used in the present study for a rovibronic energy level is built up from information concerning the uncoupled electronic state , the vibrational and rotational quantum numbers , and the symmetry of the rovibronic state .
the label finally chosen for each energy level , @xmath31 , fully characterizes the rovibronic states of a homonuclear diatomic molecule , such as c@xmath1 , but contains redundant information @xcite . here
@xmath32 is the quantum number corresponding to the total angular momentum ( an integer value ) characterizing the state , @xmath33 labels the total state parity , @xmath34 , in terms of the laboratory - fixed inversion operator @xmath35 @xcite , ` state ' is the customary term symbol of the electronic state before spin - orbit coupling is taken into account ( _ e.g. _ , x @xmath36 and b @xmath37 ) , and @xmath38 is the vibrational quantum number .
the @xmath39 , @xmath40 , label denotes the spin multiplet components : for singlet states @xmath41 , _ i.e. _ , @xmath42 , for triplet states @xmath43 , which corresponds to the standard spectroscopic notation @xmath44 , and @xmath45 ( @xmath3 , @xmath46 , @xmath45 refer to levels with @xmath47 , where @xmath48 is the quantum number corresponding to the angular momentum exclusive of nuclear and electron spin and it is usually not a good quantum number ) , and for quintet states @xmath49 @xcite .
the convention @xcite is such that whether the state is regular ( @xmath50 ) or inverted ( @xmath51 ) , the energy order is always @xmath52 for the triplet states .
the @xmath53 and @xmath54 subscript in the ` state ' label indicate if the symmetry of the electronic state is ` gerade ' ( positive ) or ` ungerade ' ( negative ) in terms of the molecule - fixed inversion operator , @xmath55 @xcite . the @xmath33 superscripts within the @xmath56 states indicate the parity of the electronic component with respect to vertical reflection , @xmath57 , in the molecular frame @xcite .
the rotationless parity @xmath58 is a widely used alternative to the total parity @xmath33 , this redundant information is included in our marvel label to help experimentalists . for c@xmath1 the allowed combination of the parity @xmath33 and the label g / u is ( @xmath59,g ) and ( @xmath5,u ) .
the rotationless parity @xmath58 is then obtained as follows : ( a ) for the @xmath28 states , the parity ( @xmath60 or @xmath61 ) can be recovered as @xmath62 , _
i.e. _ , for the even values of @xmath63 the parity of the @xmath28 states is @xmath60 , while for odd @xmath63s it is @xmath61 ; and ( b ) for the @xmath64 states , the parity is recovered as @xmath65 , _
i.e. _ , for even @xmath63s the parity is @xmath61 , and for odd @xmath63s it is @xmath60 .
the rigorous electric dipole selection rules for the rovibronic transitions are @xmath66 as mentioned , the nuclear spin statistical weights @xmath67 of the nuclear spin - zero @xmath0c@xmath1 molecule for the @xmath68 and @xmath69 states is 1 , while the @xmath70 and @xmath71 states have @xmath72 .
therefore , the latter states do not appear in spectroscopic experiments on @xmath0c@xmath1 .
the marvel input set of measured transitions was checked for a corresponding labeling error .
l l c l c l [ tab : exp ] + 15chkabeta&@xcite & @xmath73 & fts & 319/318 & + 88daabph & @xcite & @xmath74 & d - fts & 191/191 & + 88donibea & @xcite & @xmath75 & fts & 241/238 & + 77chmama & @xcite & @xmath76 & fts & 774/770 & + 63barab & @xcite & @xmath77 & & 574/532 & + 06pesi & @xcite & @xmath78 & & 8/8 & ( 2a ) + 04chyewoli & @xcite & @xmath79 & & 293/293 & ( 2b ) + 13naen & @xcite & @xmath80 & lif & 77/73 & + + + 39landsver & @xcite&@xmath81 & cae & 171/167 & ( 2c ) + 97sobllixu & @xcite & @xmath82 & lif & 179/165 & ( 2c ) + 95blliso & @xcite & @xmath83 & dl & 9/8 & + + + 88donibeb & @xcite & @xmath84 & fts & 507/507 & + 16chkabeta & @xcite & @xmath85 & fts & 1001/1001 & + + + 88donibeb & @xcite & @xmath86 & fts & 237/237 & + + + 97sobllixu & @xcite & @xmath87 & lif & 66/65 & ( 2c ) + 51freymark & @xcite&@xmath88 & & 376/354 & ( 2d ) + + + 15chkabeta & @xcite & @xmath89 & fts & 3510/3507 & + 85yacumeca & @xcite & @xmath90 & & 356/352 & ( 2e ) + 85rowamive & @xcite & @xmath91 & & 1309/1298 & ( 2f ) + 79amchma & @xcite & @xmath92 & & 2168/2139 & + 88daabsa & @xcite & @xmath93 & d - fts & 382/365 & + 06pesi & @xcite & @xmath94 & & 80/80 & + 11bosyknge & @xcite & @xmath95 & & 8/8 & + + + 13naen & @xcite & @xmath96 & lif & 168/168 & + 14naen &
@xcite & @xmath97 & lif & 150/141 & + 13bosyknge & @xcite & @xmath98 & & 23/23 & ( 2 g ) + 13yechwa & @xcite&@xmath99 & hcd & 276/273 & ( 2h ) + 07tahiam & @xcite & @xmath100 & fts & 3853/3771 & + 48phillips & @xcite&@xmath101&discharge&1181/1128 & + 02taam & @xcite & @xmath102 & hcd & 356/352 & + 03kayaguyu & @xcite & @xmath103 & & 153/150 & ( 2i ) + 85cusa & @xcite & @xmath104 & doppler - free&217/217 & ( 2j ) + 94prbe & @xcite & @xmath105 & jet cooled&39/39 & ( 2j ) + 83amiot & @xcite & @xmath106 & fts & 347/346 & + 99llew & @xcite & @xmath107 & & 138/138 & + 85susahi & @xcite & @xmath108 & & 194/194 & ( 2j ) + 10boknge & @xcite & @xmath109 & & 23/23 & ( 2j ) + 11bosyknge & @xcite&@xmath110 & & 46/46 & + + + 86hawi & @xcite & @xmath111 & & 100/100 & ( 2k ) + 49phillips & @xcite&@xmath112&discharge&1833/1664 & + 98brhakocr & @xcite & @xmath113 & & 10/10 & ( 2l ) + + + 13chyewa & @xcite&@xmath114 & hcd & 221/210 & + 13naen & @xcite & @xmath115 & lif & 513/513 & + 14naen & @xcite & @xmath116 & lif & 205/196 & + 07jonare & @xcite&@xmath117 & lif & 235/234 & + + + 15krbatrna & @xcite & @xmath118 & rempi & 67/67 & + 16krbawena & @xcite & @xmath119 & & 1/1 & ( 2 m ) + + + 15chkabeta & @xcite&@xmath120 & fts & 32/32 & + + + 15chkabeta & @xcite & @xmath121 & fts & 1/1 & + + + 11bosyknge & @xcite&@xmath122 & fwm & 68/68 & + + + 15bomago & @xcite & @xmath123 & fwm & 57/57 & + l l c l c l [ tab : expnot ] + 82erlama & @xcite & & & no & ( 3a ) + + + 63cagi & @xcite&&&no & + 68meme & @xcite & & & no & ( 3b ) + 68phdaxx&@xcite & & & yes & ( 3c ) + 88urbaja&@xcite&21,280 - 25,930&&no + 90chma & @xcite & & & no + 97kahuew&@xcite & & & no & ( 3d ) + 94cado&@xcite & & & no & ( 3e ) + 99llew & @xcite & & & yes & ( 3f ) + 05smpasc & @xcite & & cl & no + + + 63baraa & @xcite & & & yes & ( 3 g ) + + @xmath5b@xmath26 @xmath124 + 91baurja&@xcite&28,030 - 28,555 & lif & no
+ + + 06koremona&@xcite&&&no + + + 09najopare&@xcite&&&no&(3h ) + + + 69helama&@xcite & 69,000 - 73,000 & fd & yes & ( 3i ) + + + 96blliso&@xcite&43,510 - 43,545&lif & no .
+ + + 85anbope&@xcite&23,800 - 31,250 & e & & ( 3j ) + 87vahe & @xcite & 21,000 - 24,500 & lif & no & ( 3k ) + 88urbaja&@xcite&21,280 - 25,930 & lif & no & ( 3l ) + + + 67mekr&@xcite&&&no&(3 m ) + + + 88gocoa & @xcite & & & no&(3n ) + 88gocob & @xcite & & & yes&(3o ) + 89goco & @xcite & & & yes & ( 3p ) + first , we make comments on those observed bands of c@xmath1 which were not used in the marvel analysis . the situation with the deslandres
dazambuja band is a remarkable one .
the band was originally observed more than a century ago @xcite and , as discussed in the introduction , has been observed in a number of astronomical objects .
however , while a high - resolution line list is available for @xmath125c@xmath1 @xcite and there are a number of papers reporting laboratory observation of the band for @xmath0c@xmath1 @xcite , there are no high resolution line data for the band .
a high resolution re - measurement of the band for @xmath0c@xmath1 would appear to be welcome .
observation of the messerle
krauss band is reported in a single , short paper @xcite , which provides no line data .
similarly , there are so far no published lines for the recently detected kable
schmidt band @xcite .
the situation with the three herzberg and two goodwin cool vuv bands is somewhat different .
there are papers reporting detailed spectra for each of these bands @xcite .
however , each of them comes from a single uncorroborated measurement . as these bands
probe upper states which are too high in energy for reliable , independent theoretical predictions , it was decided to leave their inclusion in a marvel analysis for future work .
we note that omitting the energy levels associated with these bands is not critical for the partition sums and thermodynamic data determined as part of this study .
we also note that papers by @xcite refer to the so - called leblanc band comprising weak @xmath5 b@xmath26 @xmath126 transitions ; however , we could find no line data on this band or any papers by leblanc reporting it .
second , the set of comments below refer to data sources used in our marvel analysis and listed in table 2 .
( 2a ) 06pesi @xcite uncertainty assumed to be 0.01 @xmath4 ( petrova , private communication , 2015 ) .
+ ( 2b ) 04chyewoli @xcite not the original data are used but those data extracted from 15chkabeta @xcite .
+ ( 2c ) 97sobllixu @xcite and 39landsver @xcite uncertainty of these two sources is assumed to be 1.0 and 0.2 @xmath4 , respectively .
+ ( 2d ) 51freymark @xcite the stated uncertainty was doubled to 0.02 @xmath4 , as this seems to be a more adequate guess of the accuracy of these transitions .
+ ( 2e ) 85yacumeca @xcite results recorded using magnetic rotation .
the data are provided by curl ( private communication , 2015 ) .
the uncertainty was increased to 0.005 @xmath4 as values below this did not give consistent results during the marvel analysis .
+ ( 2f ) 85rowamive @xcite this source presents an analysis of measurements by @xcite ; the data were extracted from 15chkabeta @xcite . + ( 2 g )
13bosyknge @xcite new assignments and reassignments of the measurements by 13yechwa @xcite .
+ ( 2h ) 13yechwa @xcite assignments corrected following 13bosyknge @xcite .
+ ( 2i ) 03kayaguyu @xcite 0.007 @xmath4 uncertainty ; includes 26 transitions with @xmath127 .
+ ( 2j ) 85cusa @xcite , 85susahi @xcite , 94prbe @xcite and 10boknge @xcite data extracted from 13brbescba @xcite .
+ ( 2k ) 86hawi @xcite .
data provided by hardwick ( private communication , 2015 ) .
+ ( 2l ) 98brhakocr @xcite uncertainty assumed to be 0.02 @xmath4 . +
( 2 m ) 16krbawena @xcite give a single line position as part of their measurement of their multiphoton spectroscopic measurement of the ionization energy of c@xmath1 .
+ 0.2 cm third , the next set of comments refer to sources listed in table 3 ; these were not used in our marvel analysis for reasons listed below .
( 3a ) 82erlama @xcite this source contains lifetime data but no transition frequencies .
+ ( 3b ) 68meme @xcite no actual line data given in the paper but 11bosyknge @xcite presents some lines reassigned from this work which are included in our dataset , see table [ tab : exp ] .
+ ( 3c ) 68phdaxx @xcite a book with an extensive list of lines including higher bands but only 3607 of the 10910 lines could be be validated ; @xcite also found that these assignments did not match those of other work .
it was therefore decided to omit these lines from the final compilation .
+ ( 3d ) 97kahuew @xcite a precursor to 99llew @xcite .
+ ( 3e ) 94cado @xcite so - called high pressure ( hp ) band . + (
3f ) 99llew @xcite none of these data were selected by 13brbescba @xcite , so they were not considered in this study either .
+ ( 3 g ) 63baraa @xcite original observation by ballik and ramsay of their eponymous band .
the work was re - assigned by @xcite , but his results are not available .
+ ( 3h ) 09najopare @xcite report of a new c@xmath1 band but no data are provided and no follow - up study exists .
+ ( 3i ) 69helama @xcite report of three new vuv bands with line data .
upper states lie at too high energy for data to be validated .
+ ( 3j ) 85anbope @xcite report of extensive data for @xmath125c@xmath1 but with no dataset for @xmath0c@xmath1 .
+ ( 3k ) 87vahe @xcite report of a new c@xmath1 band : 25 transitions from one ( unknown ) band were recorded with an uncertainty of 0.2 @xmath4 and assigned @xmath63 quantum numbers ; these were supplied by heaven ( private communication , 2015 ) .
the calculations of @xcite suggest that this band actually belongs to c@xmath128 rather than c@xmath1 .
the band remains unassigned and the data were not included in our current analysis .
+ ( 3l ) 88urbaja @xcite no transition data reported in the paper . + ( 3 m ) 67mekr @xcite discovery paper giving many spectroscopic parameters but no primary transition data .
+ ( 3n ) 88gocoa @xcite discovery paper : follow - up work with data in 88gocob @xcite and 89goco @xcite .
+ ( 3o ) 88gocob @xcite new band 1 @xmath129 a @xmath11 with line data .
upper states lie at too high energy for data to be validated .
+ ( 3p ) 89goco @xcite new band 1 @xmath129 b @xmath12 with line data .
upper states lie at too high energy for data to be validated . in order to decide on their correctness ,
we have compared the experimental marvel rovibronic energies with their theoretical counterparts .
the latter approximate but complete set of energy levels is based on empirical potential energy curves ( pec ) , spin - orbit curves ( soc ) , and electronic angular momentum curves ( eamc ) of c@xmath1 as given by @xcite . the theoretical rovibronic energies were computed using a new diatomic nuclear motion program called duo @xcite .
duo solves the fully - coupled rovibronic schrdinger equation variationally using a combination of discrete variable representation ( dvr ) and rigid - rotor basis sets to represent the vibrational and the spin - rotational degrees of freedom in the hund s case ( a ) representation , respectively .
the final pecs , socs , and eamcs were obtained by refining _ ab initio _
curves obtained at the ic - mrci / aug - cc - pvqz level of electronic structure theory for the nine lowest electronic states of c@xmath1 , , , , b@xmath26 @xmath124 , , , , , and @xmath20 by fitting to the marvel energies . a detailed account of these computations will be reported elsewhere .
duo uses the following quantum numbers motivated by the hund s case ( a ) choice of the basis set : @xmath130 where @xmath131 , @xmath132 , and @xmath133 are the signed quantum numbers corresponding to projections of the electronic , spin , and total angular momenta , respectively ( the projection of the rotational angular momentum @xmath134 on the molecular axis is zero ) .
the duo quantum numbers characterizing the computed rovibronic states are used to check and complete the marvel labels .
it should be noted that duo uses an approximate assignment scheme based on the largest contribution to the wavefunction expansion @xcite .
although this scheme is very robust , it can sometimes lead to ambiguous sets of quantum numbers , especially for states in strong resonance with other rovibronic states .
comparison of the duo and marvel results helped us to identify problems in the experimental data , such as misassigned lines , duplicate transitions , and outliers .
@xmath4 from the energy @xmath135 of the state.,height=472 ] lrrr|crrr [ tab : tes ] & calc.@xmath136 & empirical & duo@xmath137&empirical & marvel@xmath138 & @xmath63 & @xmath139 + & 0.0 & 0.0 & 0.0 & 0.0 & 0.0 & 0 & 1 + & 509 & 720.0083(21)@xmath140 & 722.58 & 603.828@xmath141 & 603.817(4 ) & 2 & 1 + & 6233 & 6439.08382(58)@xmath140 & 6437.00 & 6250.164@xmath141 & 6250.149(7 ) & 0 & 3 + & 8374 & 8391.4062(19)@xmath140 & 8393.63 & 8271.606@xmath141 & 8271.607(7 ) & 1 & 1 + & 9371 & 9124.2@xmath142 & 9172.30 & 9277@xmath143 & 9280.215(5 ) & 1 & 2 + & 11966 & 12082.34355(54)@xmath144 & 12092.45 & 11859@xmath143 & 11867.825(5 ) & 2 & 1 + b@xmath26
@xmath124&15261 & 15410.33(36)@xmath144 & 15401.52 & 15197@xmath143 & 15196.509(5 ) & 0 & 1 + d @xmath24 & 20092 & 20022.5@xmath142 & 20030.92 & 19992@xmath143 & 19983.953(8 ) & 2 & 1 + @xmath145 & & & & 29258.5922(48)@xmath146 & 29860.921(5 ) & 2 & 1 + & & 40796.7@xmath142 & & & 40422.691 ( 50 ) & 2 & 1 + & & 43239.8@xmath142 & & & 43230.499 ( 138 ) & 1 & 1 + @xmath147 & & & & 51 049.799@xmath148 & 51651.142(20 ) & 1 & 1 + @xmath149 & & & & & 52106.042(94 ) & 1 & 1 + e@xmath150 & & 55034.7@xmath142 & & & 54936.672(150 ) & 0 & 1 +
in its ground electronic state c@xmath1 has a fairly strong bond @xcite .
the equilibrium bond distance of the state at 1.2425 is short for a double bond ( considerably shorter than the c = c bond of ethylene at @xmath151(c = c ) = 1.331 ) , but long for a triple bond ( in c@xmath1h@xmath1 @xmath151(cc ) = 1.203 ) .
this also means that the dissociation energy of the state is large , in fact more than 50,000 @xmath4 .
compared to this , the energy differences among the several feasible asymptotes comprised by the low - lying @xmath21p , @xmath22d , and @xmath22s states of the c atom are relatively minor . furthermore , the structure of the molecular orbitals ( mo ) of c@xmath1 is such that a large number of low - energy singlet , triplet , and quintet valence states are feasible and many of them are part of experimentally measurable rovibronic transitions ( see table 1 ) . in the state
the leading valence electron configuration is ( core)@xmath152 . by promoting electrons from the weakly antibonding @xmath153 and the strongly bonding @xmath154 mos and populating the @xmath155 mo a large number of electronic states arise .
fortunately , for obtaining proper , temperature - dependent ideal - gas thermochemical quantities up to about 4,000 k , it is sufficient to consider 9 electronic states , four singlet , four triplet , and one quintet states ( see table 4 ) . as mentioned already
, all these lowest - energy electronic states correlate with the c(@xmath21p ) + c(@xmath21p ) separated - atom limit .
not too surprisingly for such a simple molecule , a large number of electronic structure computations are available for c@xmath1 in the literature @xcite .
there are several issues which make the electronic structure computations extremely challenging for c@xmath1 .
first , there is a quasi - degeneracy of the fully occupied @xmath156 and the empty @xmath157 mos , explaining some of the unusual characteristics of the excited electronic states of c@xmath1 .
second , the existence of several low - lying excited electronic states leads to the occurrence of a considerable number of avoided crossings among the pecs as the cc distance is varied .
third , as pointed out by @xcite based on full configuration interaction ( fci ) computations , at least in the cases of the x , b , and b@xmath26 states , methods based on an unrestricted hartree
fock ( uhf ) reference provide correct but methods based on a restricted hf ( rhf ) reference provide incorrect results .
fourth , one must account for the strong multireference character of the electronic states and the near degeneracies changing rapidly along the cc distance .
fifth , rather large atom - centered , fixed - exponent gaussian basis sets are required for the correct and converged description of the valence states .
these difficulties explain why this deceptively simple diatomic molecule is still one of the favorites of developers of modern wavefunction - based electronic structure techniques @xcite . using ci methods , 27 bound valence states of c@xmath1 were computed by @xcite .
@xcite obtained results for all 62 electronic states in the valence manifold , including weakly bound and repulsive ones .
@xcite obtained results for rydberg states , as well .
electronic states with @xmath158 values up to 75,000 @xmath4 ( this is in fact the f @xmath11 state ) have been studied but for the present investigation the energy cut - off value was chosen to be 35,000 @xmath4 .
this limits the number of singlet , triplet , and quintet states to 4 , 4 , and 1 , respectively , 9 states altogether .
none of the higher - lying states will be considered in what follows .
note that rkr potential curves are given for several singlet , triplet , and quintet states in @xcite .
in this study , the so - called @xmath159 values obtained for the electronic states define directly the lowest measurable term energies of the states ( thus , they may not necessarily correspond to @xmath160 ) .
it is not that simple to determine the @xmath158 values of the excited electronic states of c@xmath1 , as these are not measurable quantities .
this can only be achieved if the zero - point vibrational energy ( zpve ) of all the states is determined . however , since all states are coupled in the duo computations , these do not come directly from our joint marvel and duo analysis .
in particular , for the singlet @xmath161 and @xmath162 electronic states there are no transitions to @xmath160 upper rovibronic states .
llrrrclcl [ tab : vib_singletfirst ] x @xmath126 & 1 & 1829.15 & 1827.4849(2 ) & 1827.486(5 ) + & 2 & 3630.35 & 3626.6835(2 ) &
3626.681(10 ) + & 3 & 5402.78 & 5396.6892(4 ) & 5396.686(9 ) + & 4 & 7145.40 & 7136.3507(6 ) & 7136.350(6 ) + & 5 & 8856.84 & 8844.1241(11 ) & 8844.124(7 ) + & 6 & 10536.37 & 10517.9659(39 ) & 10517.950(7 ) + & 7 & 12178.69 & 12154.9615(29 ) & 12154.961(6 ) + & 8 & 13783.76 & 13751.3944(38 ) & 13751.393(3 ) + & 9 & 15346.69 & 15302.8952(46 ) & 15302.893(7 ) + b@xmath163 & 1 & & 1420.4850(4 ) & 1420.488(9 ) + & 2 & & 2840.0048(4 ) & + & 3 & & 4261.0686(4 ) & 4261.071(1 ) + & 4 & & 5681.5113(6 ) & + e @xmath126 & 1 & & & 1592.316(200 ) + lrrrllrrr [ tab : vib_singletsecond ] a @xmath164 & 1 & 1 & 1584.008(4 ) & & b @xmath165 & 1 & 2 & 1384.440(2 ) + & 2 & 1 & 3143.805(5 ) & & & 2 & 2 & 2746.010(9 ) + & 3 & 1 & 4679.323(6 ) & & & 3 & 2 & 2746.010(9 ) + & 4 & 1 & 6190.503(7 ) & & & 4 & 2 & 5400.804(8 ) + & 5 & 1 & 7677.273(4 ) & & & 5 & 2 & 6694.148(10 ) + & 6 & 1 & 9139.523(9 ) & & & 6 & 2 & 7964.836(3 ) + & 7 & 1 & 10577.184(13 ) & & & 7 & 2 & 9214.047(3 ) + & 8 & & & & & 8 & 3 & 10446.668(5 ) + & 9 & 1 & 13378.377(1 ) & & & + & 10 & 1 & 14741.688(1 ) & & & + & 11 & 1 & 16079.978(7 ) & & & + & 12 & 1 & 17393.116(8 ) & & & + & 13 & 1 & 18680.944(7 ) & & & + & 14 & 1 & 19943.297(7 ) & & & + & 15 & 1 & 21179.968(7 ) & & & + & 16 & 1 & 22390.754(7 ) & & & + lrrr [ tab : vib_triplet ] a @xmath166 & 1 & 1617.985(10 ) + & 2 & 3212.620(9 ) + & 3 & 4783.940(5 ) + & 4 & 6331.973(1 ) + & 5 & 7855.893(100 ) + & 7 & 10835.820(1 ) + & 8 & 12290.393(1 ) + & 9 & 13721.623(5 ) + & 10 & 15129.564(1 ) + & 11 & 16514.369(1 ) + b @xmath167 & 1 & 1448.103(8 ) + & 2 & 2874.028(2 ) + & 3 & 4277.927(1 ) + c @xmath168 & 1 & 2031.833(8 ) + & 2 & 4034.776(7 ) + & 3 & 6007.745(8 ) + & 5 & 9859.060(1 ) + & 6 & 11734.338(10 ) + & 7 & 13573.601(1 ) + d @xmath24 & 1 & 1753.500(7 ) + & 2 & 3469.636(10 ) + & 3 & 5145.247(6 ) + & 4 & 6776.153(1 ) + & 5 & 8356.139(7 ) + & 6 & 9880.362(7 ) + & 7 & 11337.658(1 ) + & 8 & 12722.024(1 ) + & 9 & 14025.567(1 ) + & 10 & 15245.388(2 ) + figure [ fig : marvelels ] shows all marvel term values below 35,000 @xmath4 for four singlet and four triplet states , where @xmath169 @xmath4 has been subtracted from the computed energies to make the figure clearer : this means that near - horizontal sequences of levels for a particular electronic state are all associated with something one could call a single vibrational level .
figure [ fig : marvelels ] shows that the largest total angular momentum quantum numbers , @xmath170 , are 74 , 75 , 86 , and 70 for the , , , and states , respectively . as expected , as the vibrational excitation increases , the @xmath171 value usually decreases . finally , note that the coverage of rovibronic levels up to 35,000 @xmath4 from experiment is not complete ; assuming rigid rotation , data up to about @xmath172 is needed to have full coverage of the energy levels required during the thermochemical analysis .
this coverage is provided in this study by duo energy levels ( _ vide infra _ ) . due to
the strength of the cc bond in all the electronic states studied , the vibrational fundamental is substantial in almost all the bound electronic states of c@xmath1 .
in fact , for the ground electronic state the harmonic wavenumber is close to 2,000 @xmath4 , a high value for a relatively heavy molecule .
thus , the number of vibrational states is not that high , despite the large dissociation energy . in particular , @xcite computed 57 , 54 , 49 , and 36 bound vibrational levels for the , , , and b@xmath26 @xmath124 states , respectively . the number of vibrational levels characterized by our marvel analysis is considerably smaller , only 9 , 16 , 8 , and 3 , respectively .
the vibrational energies presented for the triplet electronic states given in table 7 can not be compared easily with existing literature values , since the marvel values are for a specific spin component of a rovibronic energy level .
the largest `` vibrational fundamental '' ( @xmath160 ) corresponding to the electronic states studied here is that of the c @xmath168 state , at 2031.833 @xmath4 .
consequently , this state must have the strongest cc bond .
the rotational constant of c@xmath1 for the state is relatively small , about 1.81 @xmath4 .
this results in a large number of rotational states for each vibrational level . as part of this study ,
duo results were obtained up to @xmath172 .
they served to check whether a marvel energy level is viable or not as well as they were generated to help the thermochemical analysis of this study ( _ vide infra _ ) .
the lowest missing marvel energy level is at 7,800 @xmath4 ; up to this energy the coverage is complete . as figure [ fig : marvelrovib ] shows , as the energy increases there are more and more experimentally unknown energy levels .
the coverage drops below 10% at about 30,000 @xmath4 .
it is interesting to note how close some of the rovibrational energy differences are to each other .
take the state as an example .
the @xmath173 energy difference for @xmath160 is 1617.985 @xmath4 .
the highly similar energy differences for the different spin components ( @xmath174 ) , ( @xmath175 ) , ( @xmath176 ) , ( @xmath177 ) , and ( @xmath178 ) , are 1618.063 , 1617.941 , 1617.985 , 1617.902 , and 1617.831 , respectively .
the reason behind this observation is that each spin component of each vibrational level of the state has a slightly different effective value for their rotational constant due to both electronic and vibrational effects , and these energy differences include rotational energies .
the first - principles duo energies , where such interactions are taken explicitly into account , results in similar energy differences .
@xcite and @xcite seem to be the first to address the thermochemical properties of the c@xmath1 molecule .
they both based their analysis on spectroscopic constants available to them and included several , but not all necessary , electronic states in their study .
more recent thermochemical studies include those by @xcite , @xcite , and @xcite .
the most reliable results appear to have been given by @xcite .
the internal partition function , @xmath27 , of c@xmath1 is computed here via the direct summation recipe @xcite using a mixture of experimental ( marvel ) and theoretical ( duo ) energy levels .
duo provides the full set of energies for the nine electronic states considered during the determination of the ideal - gas thermochemistry of c@xmath1 .
while this is a complete set , the energy levels are of limited accuracy .
thus , whenever possible , the duo energy levels are replaced by the incomplete but accurate set of marvel rovibronic energies .
the final energies are used to compute the internal partition sum , its first ( @xmath179 ) and second ( @xmath180 ) moments , and the specific heat ( @xmath181 ) as a function of temperature .
inaccuracies in @xmath27 have three sources of origin .
the first is the intrinsic uncertainty of the energy levels .
the second is the lack of a complete set of bound rovibronic energy levels .
the third is associated with the treatment , including the possible neglect , of the unbound states .
the main source of uncertainty in @xmath27 can be estimated straightforwardly using the uncertainty of each energy level and an error propagation formula .
all experimental ( marvel ) energy levels have an associated uncertainty , while an uncertainty of 5.0 @xmath4 was assumed for the uncertainties of all the duo levels .
( note that this uncertainty estimate is rather pessimistic . )
figure [ fig : qerror ] ( solid line ) shows the impact of the uncertainties of the energy levels on the uncertainty of @xmath27 .
it can be seen that up to about 2,500 k this type of uncertainty is dominant , but its maximum value , occurring at the lowest temperatures , is still less than 0.01% .
estimation of the second type of error is hard since ( a ) the exact number of the energy levels is unknown , and ( b ) the value of the partition function grows monotonically as more and more energy levels are considered in the direct sum .
therefore , only an approximate convergence can be reached at higher temperatures during the direct summation . to check the convergence of the partition function
we need a larger set of energy levels ; therefore , we computed approximate rovibronic energy levels for all the electronic states considered , using the spectroscopic constants published by @xcite up to 70,000 @xmath4 .
using the spectroscopic constants of 27 electronic states we could determine 332 347 extra energy levels above 35,000 @xmath4 .
figure [ fig : qerror ] ( dashed line ) shows the difference ( in % ) of the two data sets , ( marvel + duo ) and ( marvel + duo + approximate energy levels ) . although up to 2,000 k the partition function is fully converged ( the difference is less than @xmath182 % ) , above 2,000 k the difference begins growing appreciably .
nevertheless , the maximum uncertainty is still less than 0.1% at 4,000 k , which is acceptable for probably all practical applications . in case of c@xmath1 ,
no consideration of unbound states @xcite is necessary , due to the large dissociation energy of c@xmath1 .
the @xmath183 , @xmath179 , @xmath180 , and @xmath181 results are given in table 10 in 100 k intervals .
the full set of results at 1 k increments is given in the machine readable version of the table in the electronic edition .
figure [ fig : qdiff ] shows the result of the comparison of our @xmath27 with those of @xcite and @xcite .
it can be seen that the earlier studies always yield smaller numbers for the partition function of c@xmath1 than the present study .
a possible explanation is that we use a larger ( more complete ) set of energy levels .
note that the temperature range of the two earlier works begins from 1,000 k. figure [ fig : cpdiff ] shows the difference between our @xmath181 results and the janaf @xcite , @xcite , esa @xcite , and @xcite and @xcite data .
@xcite reported @xmath181 values from 0 k to 6,000 k , but @xcite corrected his values above 2,000 k. while there is good agreement with the esa and altman & clementi data at low temperatures , the difference begins to increase with the temperature .
conversely , for gurvich and janaf , the differences , especially compared to those of gurvich & veyts , are surprisingly large at lower temperatures .
@xcite also found this discrepancy cpncerning the gurvich & veyts ( nasa ) data at lower temperatures ; nevertheless , they could not explain this strange behavior .
we believe that the problem originates from the incorrect usage of @xmath184 values in @xcite . the standard ( spectroscopic ) energy expansions use @xmath158 as a minimum - to - minimum excitation energy ; in this case the rovibronic energy levels of the upper electronic state will be shifted by the difference of the zero - point energies , by about 0.5 @xmath185 .
to get the correct energies , the @xmath159 values should be used instead of the @xmath158 values .
the first excited electronic state usually lies much above the ground state ; therefore , this relatively small shift does not cause a significant problem if left out of consideration .
however , in the case of c@xmath1 , where the relative energy of the state is smaller than the vibrational fundamental of either state , the incorrect use of @xmath158 leads to wrong @xmath181 values at lower temperatures .
table 8 gives coefficients of the least squares fit to our computed partition function using the traditional form of @xcite @xmath186 in order to get the best reproduction of the directly computed values , the fit had to be performed in two separate temperature ranges .
the first range is @xmath187 k , the other is @xmath188 k. these fits can reproduce the values of log@xmath183 reasonably accurately , within 0.1% in either region .
nevertheless , to take full advantage of the high accuracy of the present thermochemical reults for c@xmath1 , the numerical results of the supplementary information should be used .
this supplementary information also contains the transitions file which forms the input for marvel , and which can be augmented with any future spectroscopic data on c@xmath1 and rerun , and the associated energies file which is the output from the marvel run .
c@xmath1 up to 4000 k. the solid ( black ) line shows the uncertainty which arises from the uncertainties of the known energy levels .
the dashed ( red ) curve represents the convergence error , in % , due to energy levels not included in the analysis.,height=472 ] results with janaf @xcite ( solid ) , @xcite ( dashed ) , esa @xcite ( dotted ) , and @xcite & @xcite ( dashed - dotted , see text ) values .
, height=472 ] crr [ tab : qcoeff ] @xmath189 & 3.6577362306 & 186.8558092069 + @xmath190 & @xmath57.0625294443 & @xmath5139.9034556834 + @xmath191 & 5.4057108902 & 42.4445304374 + @xmath192 & @xmath52.0744319852 & @xmath56.6283715970 + @xmath193 & 0.4447612738 & 0.5612382751 + @xmath194 & @xmath50.0504276952 & @xmath50.0241572167 + @xmath195 & 0.0023610646 & 0.00040521942 + rrrrlcl[tab : q ] 100.0 & 19.3786(1 ) & 19.3839(8 ) & 40.242(8 ) & 29.7333(2 ) + 200.0 & 41.6629(1 ) & 55.1971(7 ) & 170.867(15 ) & 40.2915(2 ) + 300.0 & 78.0529(1 ) & 137.7708(7 ) & 453.195(28 ) & 43.1578(2 ) + 400.0 & 133.6964(1 ) & 260.3942(7 ) & 812.605(39 ) & 39.7816(2 ) + 500.0 & 207.5463(1 ) & 411.2436(7 ) & 1225.949(48 ) & 37.2546(2 ) + 600.0 & 297.5930(1 ) & 585.6218(7 ) & 1700.047(59 ) &
36.0862(2 ) + 700.0 & 402.3796(1 ) & 782.8055(7 ) & 2244.006(70 ) & 35.6865(2 ) + 800.0 & 521.0433(1 ) & 1003.3874(7 ) & 2863.969(87 ) & 35.6538(2 ) + 900.0 & 653.1204(1 ) & 1248.2596(7 ) & 3564.19(15 ) & 35.7888(2 ) + 1000.0 & 798.3910(1 ) & 1518.3519(7 ) & 4348.50(38 ) & 36.0006(2 ) + 1100.0 & 956.7878(1 ) & 1814.6138(7 ) & 5221.04(92 ) & 36.2499(2 ) + 1200.0 & 1128.3464(1 ) & 2138.0495(7 ) & 6186.5(19 ) & 36.5204(3 ) + 1300.0 & 1313.1770(1 ) & 2489.7472(7 ) & 7250.5(37 ) & 36.8053(5 ) + 1400.0 & 1511.4500(2 ) & 2870.8933(7 ) & 8418.9(65 ) & 37.1015(7 ) + 1500.0 & 1723.3857(2 ) & 3282.7725(13 ) & 9698(10 ) & 37.4068(10 ) + 1600.0 & 1949.2475(3 ) & 3726.7596(19 ) & 11095(15 ) & 37.7195(14 ) + 1700.0 & 2189.3369(4 ) & 4204.3059(28 ) & 12616(22 ) & 38.0375(17 ) + 1800.0 & 2443.9883(6 ) & 4716.9251(38 ) & 14269(30 ) & 38.3585(21 ) + 1900.0 & 2713.5652(8 ) & 5266.1783(51 ) & 16060(41 ) & 38.6803(25 ) + 2000.0 & 2998.4562(11 ) & 5853.6614(65 ) & 17996(53 ) & 39.0006(29 ) + 2100.0 & 3299.0714(15 ) & 6480.9933(83 ) & 20084(66 ) & 39.3173(33 ) + 2200.0 & 3615.8393(19 ) & 7149.806(11 ) & 22332(81 ) & 39.6285(36 ) + 2300.0 & 3949.2042(24 ) & 7861.740(17 ) & 24744(98 ) & 39.9326(40 ) + 2400.0 & 4299.6233(32 ) & 8618.434(34 ) & 27329(116 ) & 40.2281(43 ) + 2500.0 & 4667.5647(49 ) & 9421.521(74 ) & 30092(135 ) & 40.5139(50 ) + 2600.0 & 5053.5054(88 ) & 10272.63(15 ) & 33039(156 ) & 40.7893(63 ) + 2700.0 & 5457.929(17 ) & 11173.38(32 ) & 36178(178 ) & 41.0535(92 ) + 2800.0 & 5881.327(33 ) & 12125.37(63 ) & 39513(201 ) & 41.306(14 ) + 2900.0 & 6324.193(64 ) & 13130.2(12 ) & 43051(225 ) & 41.546(23 ) + 3000.0 & 6787.02(11 ) & 14189.4(21 ) & 46798(251 ) & 41.775(36 ) + 3100.0 & 7270.32(21 ) & 15304.7(36 ) & 50760(280 ) & 41.991(56 ) + 3200.0 & 7774.59(36 ) & 16477.4(60 ) & 54941(316 ) & 42.195(83 ) + 3300.0 & 8300.33(61 ) & 17709.2(98 ) & 59348(363 ) & 42.38(12 ) + 3400.0 & 8848.04(98 ) & 19001(15 ) & 63986(429 ) & 42.56(17 ) + 3500.0 & 9418.2(15 ) & 20355(23 ) & 68859(528 ) & 42.73(23 ) + 3600.0 & 10011.4(23 ) & 21773(35 ) & 73971(673 ) & 42.89(32 ) + 3700.0 & 10628.0(35 ) & 23256(52 ) & 79328(882 ) & 43.03(42 ) + 3800.0 & 11268.6(52 ) & 24805(74 ) & 84933(1172 ) & 43.16(56 ) + 3900.0 & 11933.7(75 ) & 26421(105 ) & 90789(1562 ) & 43.28(71 ) + 4000.0 & 12624(11 ) & 28106(146 ) & 96899(2076 ) & 43.39(91 ) +
this study utilizes the marvel technique to accurately determine close to 6,000 experimental rovibronic energies of @xmath0c@xmath1 for six singlet , six triplet , and two quintet electronic states , including the eight lowest valence states which gives coverage up to 35 000 @xmath4 .
we survey all available laboratory high - resolution spectroscopic studies to provide input data for this process , resulting in 23,343 transitions connecting the 14 electronic states . while there are many spectroscopic studies available , in fact 42 were analyzed to yield the transitions analyzed , and there has been significant recent activity including the identification of several new band systems , there are also surprising gaps . for example , there is a detailed , fully rovibronically - resolved study of the deslandres
dazambuja ( @xmath5 ) band system for @xmath125c@xmath1 @xcite , but even a century after the original observation of this band @xcite there is no available high - resolution study for @xmath0c@xmath1 .
the recent observation of singlet triplet intercombination bands by @xcite helped us to achieve to link all rovibronic levels of @xmath0c@xmath1 into a single huge component within its experimental spectroscopic network ; thus , individual intercombination lines can now be predicted accurately using the results of our study .
this is a significant step towrd the astronomical detection of these transitions @xcite . to further aid this work and other astronomical studies involving c@xmath1 ,
a full rovibronic line list for @xmath0c@xmath1 is currently being constructed using the variational code duo by @xcite , as part of the exomol project @xcite . as to now
, the full set of marvel results comprising a file of validated transition frequencies and a file containing the resulting rovibronic energy levels are given in the supplementary information to this paper .
the highly accurate but limited set of experimental ( marvel ) energy levels augmented with the much less accurate but much more complete set of duo energy levels has been used to compute ideal - gas thermochemical functions for @xmath0c@xmath1 up to 4,000 k. the accuracy of the partition function is better than 0.1% even at the highest temperatures , considerably exceeding the accuracy of all previous studies .
this assures that the accuracy of the present isobaric heat capacity of @xmath0c@xmath1 is significantly better than that of any previous study .
we thank robert curl , john hardwick , michael heaven , and jian tang for supplying the ( unpublished ) data from their spectroscopic experiments , and tatiana petrova and peter radi for comments on their data .
peter radi is also thanked for his comments on the manuscript .
this work has received support from the european research council under advanced investigator project 267219 , and the scientific research fund of hungary ( grant otka nk83583 ) .
collaboration of the ucl and elte groups has greatly benefited from the support of two cost actions , codecs ( cm1002 ) and molim ( cm1405 ) .
some funding was provided by the nasa laboratory astrophysics program .
capitelli , m. , colonna , g. , giordano , d. , et al .
2005 , tables of internal partition functions and thermodynamic properties of high - temperature mars - atmosphere species from 50 k to 50000 k , tech . rep .
, esa scientific technical review | the carbon dimer , the @xmath0c@xmath1 molecule , is ubiquitous in astronomical environments .
experimental - quality rovibronic energy levels are reported for @xmath0c@xmath1 , based on rovibronic transitions measured for and among its singlet , triplet , and quintet electronic states , reported in 42 publications .
the determination utilizes the measured active rotational - vibrational energy levels ( marvel ) technique .
the 23,343 transitions measured experimentally and validated within this study determine 5,699 rovibronic energy levels , 1,325 , 4,309 , and 65 levels for the singlet , triplet , and quintet states investigated , respectively .
the marvel analysis provides rovibronic energies for six singlet , six triplet , and two quintet electronic states .
for example , the lowest measurable energy level of the state , corresponding to the @xmath2 total angular momentum quantum number and the @xmath3 spin - multiplet component , is 603.817(5 ) @xmath4 .
this well - determined energy difference should facilitate observations of singlet triplet intercombination lines which are thought to occur in the interstellar medium and comets . the large number of highly accurate and clearly labeled transitions that can be derived by combining marvel energy levels with computed temperature - dependent intensities should help a number of astrophysical observations as well as corresponding laboratory measurements .
the experimental rovibronic energy levels , augmented , where needed , with _ ab initio _ variational ones based on empirically adjusted and spin - orbit coupled potential energy curves obtained using the duo code , are used to obtain a highly accurate partition function , and related thermodynamic data , for @xmath0c@xmath1 up to 4,000 k. llll singlet & phillips & @xmath5 & @xcite + & mulliken & @xmath5 & @xcite + & _ herzberg f _ & f @xmath6&@xcite + & bernath b & @xmath5 & @xcite + & bernath b@xmath7 & b@xmath7 @xmath8 & @xcite + & _ deslandres dazambuja _ & @xmath5 & @xcite + & _ messerle krauss _ & c@xmath7 @xmath9 & @xcite + & freymark & @xmath5 & @xcite + & _ goodwin cool a _ & 1 @xmath10 a @xmath11 & @xcite + & _ goodwin cool b _ & 1 @xmath10 b @xmath12 & @xcite + triplet & ballik ramsay & @xmath5 & @xcite + & swan & @xmath5 & @xcite + & fox herzberg & e @xmath13 & @xcite + & _ herzberg f _ & f @xmath14 & @xcite + & _ herzberg g _ & g @xmath15 & @xcite + & krechkivska schmidt & 4 @xmath16 a @xmath17 & @xcite + & duck & @xmath5 & @xcite + & _ kable schmidt _ & e @xmath18 & @xcite + quintet & radi bornhauser & @xmath19 & @xcite + intercombination & triplet - singlet & @xmath5 & @xcite + & quintet - triplet & @xmath20 & @xcite + & singlet - triplet & @xmath5 & @xcite + + [ tab : bands ] c@xmath1 showing the bands considered in this work .
the dashed lines represent observed but unnamed intercombination bands.,height=604 ] |
the main thrust of this work is to understand the undulation phenomena in smectic a liquid crystals caused by a strong magnetic field .
the mathematical framework for our study is the model introduced by de gennes @xcite .
the landau de gennes energy describes the state of the liquid crystal in terms of a director @xmath1 and a complex order parameter @xmath2 for the layered structure .
the molecular mass density is defined by @xmath3 where @xmath4 is a locally uniform mass density , @xmath5 is the mass density of the smectic layers , and @xmath6 parametrizes the layers so that @xmath7 is the direction of the layer normal . also , @xmath8 is the wave number and @xmath9 is the layer thickness .
the state @xmath10 corresponds to no layered structure ( the sample is in a nematic phase ) .
if @xmath11 is a nonzero constant , then the sample is in a smectic state throughout .
the contribution to the energy due to the external magnetic field is accounted for by the magnetic free energy density@xmath12 where @xmath13 is the magnetic anisotropy , @xmath14 is the magnitude of the magnetic field and @xmath15 is a unit vector representing the direction of the field .
we assume that @xmath16 . as a consequence
, the director has a preferred orientation perpendicular to the direction of the applied magnetic field .
is assumed to be positive ; in that case the preferred orientation for the director is parallel to the field which fully specifies the orientation in that @xmath17 setting . ] in the one constant approximation case for the oseen - frank nematic energy , the total free energy for smectic a is given by , @xmath18 where the material parameters @xmath19 and temperature dependent parameter @xmath20 are fixed positive constants and @xmath21 is a rectangular box .
this models a smectic a liquid crystal with smectic layers parallel to the top and bottom boundaries .
in our study we assume that the magnetic field is applied in the @xmath22-direction , perpendicular to the layers , that is @xmath23with the director constraint @xmath24 , we rewrite the magnetic energy density @xmath25 by omitting the constant term .
we define the dimensionless order parameter , @xmath26 , and do the change of variables @xmath27 with @xmath28 , and obtain the following nondimensionalized energy @xmath29 where @xmath30 and @xmath31 with @xmath32 and @xmath33 . the values @xmath34 and @xmath35
are employed in @xcite .
the dimensionless parameter @xmath36 is in fact the ratio of the layer thickness to the sample thickness .
the case @xmath37 corresponds to a lower dimensional approximation and this is the point of view in the works @xcite where the emergence of interesting structures is captured in minimizers via a @xmath38-convergence approach .
we point out that his kind of variational dimension reduction has already been considered in the analysis of the related ginzburg - landau model @xcite .
one strength of our approach is that we do not require @xmath36 to be small but just that it avoids values at which resonances occur .
the other advantage is that the branches we find do not only start near the lowest critical field but at higher bifurcation values as well , so that metastable bifurcations are also obtained .
we are interested in studying the stability of the uniformly layered state : @xmath39 to that end , we impose periodic boundary conditions for @xmath40 and @xmath41 in the @xmath42 and @xmath43 directions , while we assume dirichlet boundary conditions for @xmath40 and @xmath41 on the boundary plates : @xmath44 for all @xmath45\times\lbrack-\pi / b,\pi / b]$ ] . with this boundary condition
, we assume that @xmath46 , where @xmath47 is the two torus identified with @xmath48\times\lbrack-\pi / b,\pi / b]$ ] and @xmath49 $ ] .
it is known that the application of an electric or magnetic field of large enough strength , or bare mechanical tension @xcite , can trigger the emergence of undulations in the smectic layers : this is called the helfrich - hurault effect @xcite . in @xcite , the effect of a magnetic field on a @xmath17 smectic a liquid crystal was considered ; a threshold value of the intensity , above which periodic layer undulations are observed , was estimated and a nontrivial solution curve was found .
the proof relied on crandall - rabinowitz bifurcation theory for simple eigenvalues and thus the solution curve branches off of the lowest eigenvalue .
in our situation , the spectral analysis of the linearized operator about the uniformly layered state gives a lowest eigenvalue which is not simple and so we can not follow the same route to study the instability of the undeformed state at the critical magnetic field . instead , we work in a space of maximal symmetries where the resonances are avoided ; this is akin to the palais principle of symmetric criticality @xcite .
the previous analysis yields the existence of branches stemming from the critical magnetic field with the symmetries of these one dimensional fixed point subspaces of irreducible representations(see section [ bifurcation ] for more details ) .
a similar approach has been used to obtain multi - pole periodic solutions to the gross - pitaevskii equation in the plane @xcite ; these solutions arise as global bifurcations of a nonlinear schrdinger equation .
our main result is the following .
[ tmain]there exists a critical magnetic field @xmath50 such that the uniformly layered state has two bifurcating solutions starting from @xmath51 except for a finite union of hypersurfaces in the set of parameters @xmath52 one of these branches is stable . in section [ bifurcation ] we give more precise information about these branches , in particular we find their symmetries and in [ localestimates ] we provide their local asymptotic profile .
we are also able to retrieve other bifurcation branches not covered by theorem [ tmain ] , namely for the resonant case @xmath53 ( see proposition [ propab ] for more details ) .
next , we investigate what happens at each cross section of the domain under the assumption that the magnetic field is large enough to render a planar configuration .
we find this reduced model bears strong resemblance with the one for micromagnetics considered in @xcite and from this we derive important qualitative information about minimizers in the limit @xmath54 including a general compactness statement and a sharp lower bound that allows us to compare between natural candidates for the optimal pattern .
finally , we provide some numerical evidence in support of our findings .
we now give an outline of the paper . in the next section
we introduce notation and the functional setting where our bifurcation problem is posed .
we then proceed to study the spectral problem of the linearized operator at the undeformed state and derive the critical field in section [ spectrum ] . in section [ bifurcation ]
we study the irreducible representations associated to @xmath55 and the corresponding isotropy groups ; from these we obtain two fixed one dimensional subspaces were we find the aforementioned bifurcation branches .
we move to the @xmath17 dimensional problem in section [ planar ] and deduce the periodicity of minimizers . in the final section we present the numerical simulations depicting the @xmath17 planar configurations and layers at the onset of undulations , confirming that the estimate of the critical field from our theory is consistent with the numerical results .
we start by introducing some notation which is going to be essential for the bifurcation analysis .
let @xmath56 be a group .
we recall that a hilbert space @xmath57 is a @xmath56-*representation * , if there is a morphism of groups given by the map @xmath58 .
also , a set @xmath59 is said to be * invariant * if @xmath60 for all @xmath61 , @xmath62 .
an * irreducible representation * is an invariant subspace of @xmath57 without a proper invariant subspace . for a subgroup @xmath63
, we consider the * * fixed point space**@xmath64 the * orbit * and the * isotropy group * of a point @xmath65 are defined as@xmath66 it is said that a map @xmath67 is @xmath56-*equivariant * if @xmath68 finally , a map @xmath69 is * invariant * if @xmath70 .
the following properties are used in this article .
we refer the reader to @xcite for a proof .
( @xcite)[actionprop ] 1 . if @xmath67 is @xmath56-equivariant and @xmath71 , then the linear map @xmath72 is @xmath14-equivariant .
if @xmath73 is an isometry in @xmath57 for all @xmath61 , i.e. @xmath74 , then the gradient of an invariant map @xmath75 is equivariant , i.e. @xmath76 is @xmath56-equivariant .
the restriction of an @xmath56-equivariant map @xmath77 to @xmath78 is well defined , i.e. @xmath79 for all @xmath71 .
the previous definitions and statement are adapted for a manifold @xmath80 , if @xmath81 is invariant by the linear transformation @xmath73 . in this case
the action is @xmath82 the landau - de gennes energy @xmath75 introduced in is well defined in the manifold @xmath83 since @xmath81 is contained in the hilbert space @xmath84 , we define the action of the group@xmath85 in this hilbert space as follows .
.1 in * 1 . * the rotations .
.1 in for @xmath86 @xmath87 .1 in * 2 . * the reflections .
.1 in the action of the planar reflections @xmath88 is given by@xmath89 where @xmath90 and @xmath91 the above defines the action of the direct product of the two copies of the orthogonal group .
.1 in * 3 .
* the @xmath92 component .
.1 in we define the action of the reflection @xmath93 by @xmath94 we have : [ invoff ] the map @xmath95 is @xmath56-invariant .
since @xmath75 is invariant by translations and reflections in @xmath42 and @xmath43 , due to the boundary conditions , then it is invariant by the action of @xmath96 .
moreover , it is clear that all the terms in @xmath75 are invariant by the action of @xmath97 except maybe for @xmath98 . to see that this term is invariant too ,
let us denote @xmath99 , then @xmath100 and using that @xmath101 we have@xmath102 therefore , we conclude @xmath75 is invariant under the action of the group @xmath56 .
the uniformly layered state @xmath103 is a trivial critical point of @xmath75 in @xmath81 .
since all the elements of @xmath56 leave this critical point invariant , their orbits are trivial and their isotropy groups are all of @xmath56 , i.e. @xmath104 and @xmath105 for all @xmath106 in @xmath81 .
we may parameterize a submanifold of @xmath81 by writing@xmath107 where @xmath108 and@xmath109 , where hereafter the subindex @xmath110 means a zero boundary conditions on @xmath111 and @xmath112 .
the map @xmath113 defines a diffeomorphism from the space @xmath114 into the submanifold @xmath115 this is the set of functions in @xmath81 such that @xmath1 takes values in the upper half of the unit sphere and @xmath116 is purely imaginary .
let us define @xmath117 , writing the energy functional @xmath75 using this parametrization of @xmath118 we are led to the map @xmath119 defined by@xmath120 thus @xmath121corresponds to the uniformly layered state @xmath122 , see ( [ cif ] ) .
the action of @xmath56 in these coordinates @xmath123 is given by@xmath124 and @xmath125 the first variation @xmath126 is characterized by the relation @xmath127 we define the map @xmath128 by @xmath129 which is @xmath56-equivariant , because the action of @xmath56 is isometric , by property @xmath130 of lemma [ actionprop ] . in order to obtain the second variation of the energy ( [ energy ] ) around the flat layer state @xmath122
we follow the definition and derivation of the critical magnetic field in @xcite . in these coordinates , consider a perturbation @xmath131 , i.e. we have@xmath132 where @xmath133 , @xmath134 and @xmath135 , since @xmath136 and @xmath137 are orthogonal . then , the second variation around the flat layer state @xmath138 is@xmath139 now , since we have that @xmath140 , then @xmath141 from the associated euler - lagrange equations , we have that the second variation at the flat layered state applied to @xmath123 is@xmath142 then the operator @xmath143 is given by @xmath144 where @xmath145 . of course , the linearized operator @xmath146 depends on the parameter @xmath147 and in the next subsection we make this dependence explicit and write @xmath148 when we present the bifurcation theorem needed for our main result .
the solutions described in theorem [ tmain ] correspond to critical points of @xmath149 we obtain these solutions as zeros of the map@xmath150 moreover , since @xmath151 is @xmath56-equivariant , then the map @xmath152 is well defined by property 3 of lemma [ actionprop ] , where@xmath153 therefore , we may consider @xmath154 where @xmath155 is the restriction of @xmath146 to @xmath78 and @xmath156 is the restriction of @xmath157 to @xmath158 to obtain our bifurcation branches we apply theorem [ t5 ] below to @xmath159 for suitable choices of @xmath160 assume that for some linear operator @xmath161 and some @xmath162 @xmath163 is a self - adjoint fredholm operator , @xmath164 is non trivial and @xmath163 is invertible for @xmath147 close but different from @xmath165 then we can define @xmath166 as the morse index of @xmath167 restricted to @xmath168 for @xmath147 close to @xmath169 and @xmath170 as the net crossing number of eigenvalues of @xmath171 at @xmath169 , that is @xmath172 [ t5]let @xmath173 be a family of @xmath56-equivariant maps ( indexed by @xmath147 ) satisfying @xmath174 [ biftheo]if @xmath163 is a self - adjoint fredholm operator and @xmath170 is odd , then @xmath175 has a local bifurcation of zeros from @xmath176 in @xmath177 .
we may adapt the theorem 3.5.1 in @xcite that uses degree theory to prove that the bifurcation branch form a local continuum . in the case @xmath178 and @xmath179 , the theorem can be obtained with an aplication of the implicit function theorem .
this is the condition of having a simple bifurcation which allows us to give estimates in the local branches parametrized by the amplitude , using the bifurcation equation as in @xcite . actually , applying a result @xcite based on conley index
, we can prove that if @xmath170 is different from zero , there is local bifurcation from @xmath176 .
however , this local branch does not need to be a continuum .
for the bifurcation analysis we need to characterize the spectrum of the linear map @xmath180 as a starting point we already know that the space @xmath181 has a spectral representation in fourier series given by @xmath182 where @xmath183 with @xmath184 .
next we proceed to further decompose the @xmath22-dependent fourier coefficients @xmath185 [ prop : linear ] the space @xmath186 has a representation as @xmath187 where @xmath188 and @xmath189 is the eigenfunction of @xmath146 with eigenvalue @xmath190 .
moreover , the eigenfunction @xmath191 crosses zero only at the critical field@xmath192 where @xmath193 in addition , the eigenfunctions satisfy the relations @xmath194 from the associated euler - lagrange equations , we have that the second variation at the flat layered state applied to @xmath123 is@xmath142 then the operator @xmath146 is given by @xmath144 where @xmath145 .
then the eigenfunctions @xmath189 and eigenvalues @xmath190 are such that @xmath195 from here after we omit the index @xmath196 to simplify notation .
we set @xmath197,@xmath198 we have that the condition ( [ ei ] ) is equivalent to @xmath199 writing @xmath200 the system ( [ euler ] ) becomes @xmath201f-(\alpha^{2}+\beta^{2})h & = \lambda f,\label{equation - f}\\ -\varepsilon g^{^{\prime\prime}}+[\varepsilon(\alpha^{2}+\beta^{2})+(\varepsilon^{-1}-\tau)]g-(\alpha^{2}-\beta^{2})h & = \lambda g,\label{equation - g}\\ -\varepsilon h^{^{\prime\prime}}+\varepsilon(\alpha^{2}+\beta^{2})h - f & = \lambda\varepsilon^{2}h .
\label{equation - h}\ ] ] with boundary conditions @xmath202 .
we consider the linear operator @xmath203h,\ ] ] defined in @xmath204 .
let the inverse be @xmath205 , it is well known that @xmath206 is compact .
then , the system ( [ equation - f ] ) and ( [ equation - h ] ) can be written as @xmath207 and @xmath208 where we have defined the polynomial @xmath209 let @xmath210 be an eigenvalue of @xmath206 , i.e. , @xmath211 , then @xmath212 satisfies the equation @xmath213 . from the dirichlet boundary conditions we see that the solutions are @xmath214{c}\sin(lz)\text { for } l\in2\mathbb{z}^{+}\\
\cos(lz)\text { for } l\in2\mathbb{z}^{+}\mathbb{-}1
\end{array } \right .
\text{.}\ ] ] therefore , the eigenvalues @xmath210 need to satisfy @xmath215 since @xmath206 is compact , from ( [ t - theta ] ) and the fact that @xmath216 where @xmath217 is a spectrum of @xmath206 and @xmath218 is a polynomial , we have @xmath219 this can be written as @xmath220 therefore eigenvalues are defined by the relations ( [ r1 ] ) and ( [ r2 ] ) , using @xmath221 then@xmath222 ^{2}=\left ( \lambda(1-\varepsilon^{2})+\tau-\varepsilon^{-1}\right ) \left [ \varepsilon p^{2}-\varepsilon^{2}\lambda\right ] + ( \alpha^{2}+\beta^{2})\ ] ] the previous relation is equivalent to@xmath223 now @xmath224 is when @xmath225 deriving respect @xmath147 we have for @xmath226 that@xmath227 evaluating at the critical value @xmath169 we have @xmath224 , then@xmath228 as @xmath229 .
therefore , the eigenvalue @xmath230 crosses zero at @xmath169 .
the solution to ( [ equation - g ] ) is given by @xmath231 which leads to the eigenfunction @xmath232 moreover , we would like to choose the eigenfunctions such that the condition @xmath233 becomes @xmath234 . then , the eigenfunction @xmath189 such that @xmath235 are@xmath236 thus the eigenvalue problem ( [ euler ] ) has eigenvalues @xmath230 which is zero at the critical value @xmath147 and corresponding eigenfunctions @xmath189 . we have considered the domain with sides @xmath59 in order to simplify the spectrum , as it appears in terms of @xmath237 , @xmath238 and @xmath239.we conclude : for all @xmath181 we have @xmath240 and @xmath241 [ remark : critical - field]in order to find the critical value of @xmath147 , we minimize ( [ linearfunctional ] ) over functions satisfying @xmath242 . therefore , one can see that the first critical field is the minimum of the @xmath243@xmath244 following the proof of proposition [ prop : linear ] , one can see that @xmath243 can be achieved from the system ( [ equation - f])-([equation - h ] ) with @xmath224 .
noticing that ( [ equation - f ] ) and ( [ equation - h ] ) are decoupled from the system , we rewrite them , with @xmath245 and @xmath246 , @xmath247 & -\psi^{^{\prime\prime}}+\frac{\delta}{\varepsilon}\psi=\frac{\delta } { \varepsilon}f.\end{aligned}\ ] ] thus this system determines @xmath50 in ( [ threshold1 ] ) and it has been studied for layer undulations in two dimensions @xcite with the magnetic field applied in the @xmath42 direction . thus the first eigenvalue obtained in @xcite for two dimensions also provides the estimate of the critical field for our three dimensional case .
in particular , theorem 4 of @xcite proves that the first eigenvalue is estimated as @xmath248 when the interval for @xmath22 is @xmath249 .
since the scaling used in this paper gives the interval @xmath250 for @xmath22 , it is easy to see that the first eigenvalue is estimated @xmath251
as anticipated , the first step is to identify the irreducible representations associated to @xmath252 due to schur s lemma , all eigenvalues @xmath230 of @xmath146 corresponding to this representation are the same , and then , if one does not consider the symmetries the resonances make the proof of the bifurcation difficult or impossible . in the second step ,
the strategy is to identify the isotropy groups that have a fixed point space of dimension one in the irreducible representation , and then , using the implicit function theorem in the restricted point space of this isotropy groups , we can obtain simple bifurcation , that is without resonances .
however , we do not claim to find them all , although the two we find are enough to characterize the loss of stability of the uniform state to a preferred branch while at the same time shows the emergence of other interesting solutions satisfying different symmetries , as soon as one crosses the critical field @xmath253 the symmetries of the bifurcation solutions are given by the isotropy groups with these properties .
the symmetries of the solutions are presented in the third part of this section , and the bifurcation theorem in the fourth subsection .
now , we proceed to find the irreducible representations .
it suffices to characterize the action on the elements of the fourier basis . for fixed @xmath254 ,
the action on @xmath255 is given by @xmath256 the action of @xmath56 in each fourier component is given by@xmath257 from this we obtain @xmath258 in a similar way , we have that@xmath259 moreover , since @xmath260 , then@xmath261 and then the reflection @xmath262 acts according to @xmath263 in a similar way , using the fact that @xmath264 , one can prove that @xmath265 furthermore , since @xmath266 , then @xmath267@xmath268 we conclude that the irreducible representations correspond to @xmath269 with the previous action when @xmath270 . in the case
that @xmath271 or @xmath1 are zero , the irreducible representations are @xmath272 , however we do not analyze this case because this produces a bifurcation of stationary solutions in @xmath42 or @xmath43 , which may be analyzed directly @xcite . for each representation
there are at least two kind of isotropy groups that have fixed point spaces of dimension one .
bear in mind that , as mentioned earlier , in this section we do not claim to find all of them , but at least two .
hereafter , we denote @xmath273 the irreducible representation for a fixed @xmath196 .
the irreducible representation for @xmath274 even is fixed by the element @xmath97 . moreover
, the point @xmath275 for @xmath276 has an isotropy group given by@xmath277 where @xmath278 denotes , as usual , the subgroup generated by the elements @xmath279 of @xmath280 .
the other isotropy groups corresponds to @xmath281 and is given by@xmath282 for @xmath274 odd one isotropy group corresponds to @xmath283 and is given by@xmath284 the second isotropy group corresponds to @xmath285 is given by@xmath286 it is not hard to prove that the fixed point spaces of these groups have real dimension equal to one , where the fixed point spaces are generated by @xmath287 and @xmath288 respectively .
the previous isotropy groups are relevant because they give the symmetries of the bifurcating solutions . before proving the existence of these branches ,
we want to present the symmetries of the bifurcating solutions , i.e. the fixed point spaces of the previous groups .
since the elements @xmath289 , @xmath290 and @xmath97 generate @xmath291 , solutions with this isotropy group must satisfy @xmath292 where @xmath293 and @xmath294 . from the first symmetry
, we have that @xmath295 then @xmath296 unless @xmath297 or @xmath298 , then @xmath299 and @xmath300 for @xmath301 and @xmath302 with @xmath303 with @xmath304 since the elements @xmath289 , @xmath305 generate @xmath306 , they must satisfy @xmath307 where @xmath267 @xmath101 .
therefore , @xmath308 is as before but with @xmath303 such that @xmath309 the isotropy group @xmath310 has the generators @xmath262 , @xmath311 , @xmath97 and @xmath312 , then solutions with this isotropy group must satisfy @xmath313 where @xmath293 , @xmath294 . in this case@xmath314
then @xmath296 unless @xmath315 . moreover , we have that @xmath316 the group @xmath317 has generators @xmath318 , @xmath319 and @xmath312 .
then , solutions with this isotropy group @xmath317 must satisfy @xmath320 as before , we have that @xmath296 unless @xmath315 and@xmath321 for each fixed @xmath322 , the uniformly layered state @xmath138 has two bifurcations of critical points of starting from the critical field @xmath323 in @xmath324 except for a finite union of hypersurfaces in the set of parameters @xmath325 .
one of the bifurcations has the symmetries of @xmath326 and the other one of @xmath310 for @xmath327 even , and @xmath328 and @xmath317 for @xmath327 odd .
we appeal to theorem [ t5 ] .
since the eigenvalue @xmath190 is zero only at @xmath243 , and @xmath329 as @xmath330 , then @xmath331 is different from zero except for a finite number of @xmath332 .
moreover , the condition @xmath333 defines a hypersurface in the space of parameters @xmath52 on the other hand @xmath334 is identically zero only if @xmath335 and @xmath336 and @xmath337 therefore , there are no resonances with others eigenvalues @xmath338 except for the set of parameters @xmath325 in this finite number of hypersurfaces .
thus , we conclude that the only eigenvalue @xmath190 close to zero at @xmath323 is @xmath338 .
moreover , the kernel of @xmath339 is given by the space corresponding to the irreducible representation @xmath340 , i.e. the eigenvalue @xmath338 has real dimension equal to four .
we then take advantage of the actions of the isotropy groups that have a fixed point space of real dimension equal to one in @xmath341 ; in either one of the fixed point spaces ( @xmath326 and @xmath310 for @xmath327 even , and @xmath328 and @xmath317 for @xmath327 odd ) of these groups the kernel has real dimension equal to one .
therefore , the restricted map @xmath212 to the fixed point space of the isotropy group . for this restricted map only one eigenvalue crosses zero , which then yields the desired result as an application of theorem [ biftheo ] .
unfortunately , there are resonances in the hypersurface of parameters @xmath53 , i.e. we have that @xmath342 . in this case
, the previous theorem does not apply , but we can improve the results of the previous theorem for this particular case .
[ propab ] in the case that @xmath53 , there is always a bifurcation of solutions from @xmath243 for the groups @xmath291 and @xmath306 , as in the previous theorem , and for the groups @xmath310 and @xmath317 if in addition the only solutions of @xmath343 are @xmath344 is equal to @xmath345 or @xmath346 . in this case
, we can use the same arguments as in the previous theorem , we are only left to verify that there are no resonances , that is , we only need to consider the cases when @xmath347 is identically zero(viewing both expressions as functions of @xmath36 ) . in the fixed point space @xmath291 and @xmath306 other resonances can not appear , since the fixed point space has functions @xmath308 with components @xmath348 equal to zero unless @xmath349 for @xmath301 , and then , we have that @xmath350 for @xmath351 unless @xmath352 . in the case of the groups @xmath310 and @xmath317 , we have from the assumptions that @xmath353 only for @xmath344 equal to @xmath345 and @xmath346 , then we have the double eigenvalue @xmath342 .
actually , this double eigenvalue is a consequence of the invariance of @xmath75 by another action @xmath354 given by @xmath355 , due to @xmath53 . using this extra symmetry
, it is not hard to see that in the fixed point space of @xmath210 and the group @xmath310 or @xmath317 , no additional resonances exist by assumptions .
therefore , the desired conclusion follows again in this case thanks to theorem [ t5 ] .
the assumptions in the case @xmath53 are true for many @xmath345 , including the case @xmath356 . as
all the eigenvalues @xmath357 cross in the same direction , then @xmath323 is always a bifurcation point regardless of the multiplicity .
but the previous theorems allows us to obtain specific information about different branches , the symmetries of the branches , the local continuity of the branches , and local estimates given in the following section .
we complete the bifurcation analysis by providing the asymptotic portrait of the branches found in theorem [ tmain ] .
as mentioned in the introduction , these properties follow from standard bifurcation arguments @xcite .
since the bifurcation is simple , we can estimate the local bifurcating branches in the cases @xmath358 .
.1 in * 1 . the branches with @xmath291
symmetry * .1 in the bifurcation with group @xmath291 has leading term @xmath359 @xmath360 , from where we gather .1 in .1 in where @xmath361 is a function in the fixed point space of @xmath326 of order @xmath362 .
moreover , using ( [ u_v ] ) we conclude that.1 in .1 in the previous solutions satisfy the corresponding symmetry ( [ oa ] ) . also , notice that @xmath308 is the parameterization of @xmath363 given in ( [ cif ] )
, then one can approximately characterize these functions in the local branches.1 in .1 in .1 in * 1 .
the branches with @xmath310 symmetry * .1 in for the group @xmath310 we have that @xmath364 , thus.1 in .1 in and as before we have the local solutions that satisfy the symmetry ( [ da ] ) given by.1 in .1 in for the group @xmath306 we have that @xmath365 , and so .1 in .1 in using ( [ u_v ] ) we have that the solution that satisfy ( [ ob ] ) locally is .1 in .1 in for the group @xmath317 we have that @xmath366,.1 in .1 in where a similar expression with the symmetry ( [ db ] ) may be obtained .
[ section : planar ] in this section , we study the de gennes free energy in the cross section of the rectangular box while assuming the constant smectic order parameter and find the characteristics of the minimizers well above the critical field . we also assume a sufficiently strong magnetic field which forces a planar configuration . thus by taking @xmath367 and @xmath368 with @xmath369 , the energy ( [ energy ] ) becomes @xmath370 where @xmath371 , @xmath372 .
we assume that @xmath373 is a flat torus identified with @xmath374 .
if @xmath375 is a minimizer of @xmath376 , then @xmath377 satisfies @xmath378 letting @xmath379 be a solution to ( [ vp - n ] ) and defining @xmath380 we have @xmath381 we therefore study @xmath382 in ( [ energy - planar ] ) with @xmath379 satisfying ( [ vp - n ] ) to obtain the configuration of minimizer of @xmath376 .
this formulation was also used in @xcite where the weak critical field was achieved .
the functional @xmath382 in two dimensions is analogous to a free energy for micromagnetics studied in @xcite . to see this
, we let @xmath383 then by setting @xmath384 , @xmath382 becomes @xmath385 where the demagnetizing field @xmath386 is a solution to @xmath387 this energy is not the same as a micromagnetics energy ( [ energy - m ] ) studied in @xcite due to the presence of a nonlocal term in ( [ energy - m ] ) .
the following results , however , will follow directly from @xcite . while the limiting @xmath271 in @xcite is a divergence - free vector field tangent to @xmath388 ,
the limit @xmath1 in our case is a curl - free @xmath389-valued vector field in @xmath373 whose components satisfy zero mass constraint .
we restate the following theorem from @xcite in terms of @xmath1 with the addition of mass constraint .
let @xmath390 $ ] be a geometric half - angle between @xmath391 and @xmath392 where @xmath393 are traces on each side of the one - dimensional jump set of @xmath394 . in terms of @xmath1 , @xmath395 is simply the half the angle between @xmath396 and @xmath397 .
[ prop : lower - bound ] let the sequences @xmath398 , and @xmath399 be such that @xmath400 then there exist a subsequence @xmath401 , @xmath402 , and @xmath403 such that @xmath404 if , in addition , @xmath405 , then @xmath406 where @xmath407 , \label{small - angle}\\ & a(x ) = 4|\left ( x-\frac{\pi}{2}\right ) \cos x - \sin x + \sqrt{2}|\quad\mbox{for } \quad x \in\left [ \frac{\pi}{4 } , \frac{\pi}{2}\right ] .
\label{large - angle}\ ] ] the compactness result may follow directly from lemma 2.1 in @xcite and the mass constraints of @xmath408 and @xmath409 in ( [ limit ] ) result from @xmath410 and periodicity of @xmath411 .
the lower bound inequality ( [ lower - bound ] ) follows from theorem 1 of @xcite .
note that constant configurations are not allowed due to the zero mass constraint in ( [ limit ] ) and periodicity of @xmath1 .
therefore , the line singularity appears , which is also observed in @xcite . here , we consider horizontal or vertical lines for the one - dimensional jump set rather than diagonal lines , in order to minimize the arc length of the jump set .
one can see that @xmath1 has at least two internal jumps due to the periodic boundary condition on @xmath1 .
therefore we should consider two situations , squares with two jumps on both horizontal and vertical 1d tori ( figure [ fig:1d-2d ] ( a ) ) and only vertical strips ( or horizontal strips ) with two parallel 1d - tori ( figure [ fig:1d-2d ] ( b ) ) .
for the latter case , we must have @xmath412 wall to ensure the mass constraint ( [ limit ] ) . by applying proposition [ prop :
lower - bound ] , we would like to compare these two configurations , 2d square pattern ( figure [ fig:1d-2d ] ( a ) ) and 1d stripe configuration ( figure [ fig:1d-2d ] ( b ) ) .
for the 1d stripes , the formula ( [ large - angle ] ) with @xmath413 gives the lower bound @xmath414 on the other hand , for the 2d square pattern , the formula ( [ small - angle ] ) with @xmath415 gives the better lower bound @xmath416 it was proved in @xcite that the lower bound in ( [ small - angle ] ) can be achieved by one dimensional structure in the wall for jumps less than or equal to @xmath417 .
thus the lower bound in ( [ energy-2d ] ) is optimal and hence we can conclude that 2d square pattern gives the lower energy than 1d stripe structure .
we consider the gradient flow ( in @xmath418 of the energy ( [ energy ] ) and study the behavior of the solutions . the gradient flow equations are @xmath419 } + \tau\left [ ( n \cdot\mathbf{e}_{1 } ) \mathbf{e}_{1 } + ( n \cdot\mathbf{e}_{2 } ) \mathbf{e}_{2 } \right ] \right ) , \\
\frac{\partial\psi}{\partial t } & = c \delta\psi-2c i n \cdot\nabla\psi- i c ( \nabla\cdot n ) \psi- \frac{1}{\varepsilon } \psi+ \frac{g}{\varepsilon } ( 1-|\psi|^{2})\psi , \end{aligned}\ ] ] where we have defined , for a given vector @xmath420 , the orthogonal projection onto the plane orthogonal to the vector @xmath1 as @xmath421 this projection appears as a result of the constraint @xmath422 .
this method has been used for smectic a liquid crystals @xcite . as initial condition
, we consider a small perturbation from the undeformed state .
more precisely , for all @xmath423 , @xmath424 where a small number @xmath425 and @xmath426 and @xmath427 are arbitrarily chosen .
we impose strong anchoring condition for the director field and dirichlet boundary condition on @xmath40 at the top and the bottom plates ; @xmath428 periodic boundary conditions are imposed for both @xmath1 and @xmath40 in the @xmath42 and @xmath43 directions .
we use a fourier spectral discretization in the @xmath42 and @xmath43 directions , and second order finite differences in the @xmath22 direction .
the fast fourier transform is computed using the fftw libraries @xcite . for the temporal discretization
, we combine a projection method for the variable @xmath1 @xcite , with a semi - implicit scheme for @xmath40 : given @xmath429 , we solve @xmath430 } + \tau\left [ ( \mathbf{n } ^{k } \cdot\mathbf{e}_{1 } ) \mathbf{e}_{1 } + ( \mathbf{n } ^{k } \cdot\mathbf{e}_{2 } ) \mathbf{e}_{2 } \right ] , \label{scheme - n}\\ \mathbf{n } ^{k+1 } & = \frac{\mathbf{n } ^{*}}{|\mathbf{n } ^{*}|},\label{projection}\\ \frac{\psi^{k+1 } - \psi^{k}}{\delta t } & = c \delta\psi^{k+1 } - 2 c i \mathbf{n } ^{k+1 } \cdot\nabla\psi^{k } - i c ( \nabla\cdot\mathbf{n } ^{k+1 } ) \psi^{k } - \frac{1}{\varepsilon } \psi^{k+1}\nonumber\\ \label{scheme - vp}\\[-2.ex ] & + \frac{g}{\varepsilon } ( 1-|\psi^{k}|^{2 } ) \psi^{k}.\nonumber\end{aligned}\ ] ] the second step ( [ projection ] ) ensures that @xmath431 at each grid point .
note that @xmath432 in ( [ projection ] ) since we consider the case where there are no point defects in the liquid crystal .
the consistency and convergence of the projection method are given in @xcite .
the method is first order accurate in time and second order accurate in space due to the first order accuracy of the projection method ( [ scheme - n])-([projection ] ) . to solve the implicit system
, we perform a discrete fourier transform in the @xmath42 and @xmath43 direction .
the resulting tridiagonal systems in the @xmath22 direction are solved using gauss elimination .
we take the domain size @xmath433 , i.e. , in the case of @xmath53 , and @xmath434 . the number of grid points in the @xmath42 , @xmath43 and @xmath22 directions are all @xmath435 . here
we use parameters @xmath436 and then we obtain from ( [ parameters ] ) @xmath437 we have solved this system in @xcite to understand the minimizer at a various field strength in a two dimensional domain , @xmath438 and @xmath439 . here
we consider a three dimensional domain with @xmath440 . in figure
[ fig : layerz ] we show the layer structures in the middle of the sample ( @xmath441 ) in response to various magnetic field strengths @xmath147 . in figures ,
we show the contour maps of @xmath40 since the level sets of @xmath40 represent the smectic layers .
one can see that the undeformed state is stable for values of @xmath147 below the critical field @xmath442 as seen in the left column of figure [ fig : layerxy ] .
if @xmath147 increases and reaches @xmath50 , layer undulations of 2d square patterns occur ( the second column of figure [ fig : layerxy ] ) .
the estimates of the critical field , @xmath443 , is consistent with our analytical result ( [ first - eigenvalue ] ) .
the third column of figure [ fig : layerxy ] shows that the amplitude of undulations grows in the increase of the field strengths .
figure [ fig : layerz ] depicts the contour map of the phase of @xmath40 in two vertical sections , @xmath444 and @xmath445 . as in the classic helfrich - hurault theory and result from two dimension @xcite , the figures show that the maximum undulations occur in the middle of the domain and layer perturbations decreases as approaching the bounding plates . in this section , we present numerical simulations of liquid crystal equilibrium states which minimize ( [ energy - planar ] ) to support analytical results given in section [ section : planar ] .
we consider the gradient flow ( in @xmath418 of the energy ( [ energy - planar ] ) in @xmath446 with @xmath447 .
the gradient flow equations are @xmath448 where we have used the same notation as in the section [ sec : numerics - complex ] . in figure
[ fig : planar ] ( a ) and ( b ) we illustrate the director and layer configuration and confirm the analytical results in section [ section : planar ] . numerical simulations illustrate 2d square patterns as an equilibrium state .
( see figures [ fig:1d-2d ] ( a ) and [ fig : planar](a ) . )
furthermore , the contour maps of @xmath408 and @xmath409 in figure [ fig : planar ] ( c ) and ( d ) clearly demonstrate the zero mass constraint given in ( [ limit ] ) . the numerical values of energy ( [ energy - planar ] ) at the equilibrium state with various values of @xmath36 are calculated and written in the table [ table : energy ] .
these values agree with the lower bound given in ( [ energy-2d ] ) .
a. contreras , c. garcia - azpeitia , `` global bifurcation of vortices and dipoles in bose - einstein condensates '' , to appear in _ comptes rendus mathematiques .
preprint available at arxiv:1511.06843 .
a. contreras , p. sternberg , `` gamma - convergence and the emergence of vortices for ginzburg - landau on thin shells and manifolds '' .
partial differential equations _ 38 , no . 1 - 2 , 2010 , 243274 . l. nirenberg .
`` topics in nonlinear functional analysis''._courant lecture notes in mathematics , _ volume 6 .
new york university , courant institute of mathematical sciences , new york ; american mathematical society , providence , ri , 2001 .
chapter 6 by e. zehnder , notes by r. a. artino , revised reprint of the 1974 original . | we investigate the effect of a strong magnetic field on a three dimensional smectic a liquid crystal . we identify a critical field above which the uniform layered state loses stability ; this is associated to the onset of layer undulations . in a previous work @xcite , garca - cervera and joo considered the two dimensional case and analyzed the transition to the undulated state via a simple bifurcation . in dimension
@xmath0 the situation is more delicate because the first eigenvalue of the corresponding linearized problem is not simple .
we overcome the difficulties inherent to this higher dimensional setting by identifying the irreducible representations for natural actions on the functional that take into account the invariances of the problem thus allowing for reducing the bifurcation analysis to a subspace with symmetries .
we are able to describe at least two bifurcation branches , one of which is stable , highlighting the richer landscape of energy critical states in the three dimensional setting . finally , we analyze a reduced two dimensional problem , assuming the magnetic field is very strong , and are able to relate this to a model in micromagnetics studied in @xcite , from where we deduce the periodicity property of minimizers . |
the fractional quantum hall effect ( fqhe ) is a fascinating physical phenomenon observed @xcite in high quality two dimensional electron systems due to their highly degenerate quantized band structure in a very strong perpendicular magnetic field .
the fqhe occurs in a partially filled landau level ( ll ) where coulomb interaction takes the most important part in describing @xcite its physics .
the fqhe at most of the filling factors in the lowest ll , is very successfully described @xcite as an integer quantum hall effect ( iqhe ) @xcite of composite fermions ( cfs ) which are fermionic bound states of electrons and even ( @xmath8 ) number of quantized vortices , denoted by @xmath9cfs . due to the berry phase
@xcite produced by attached vortices , cfs experience an effective magnetic field @xmath10 , where @xmath11 is the magnetic flux quantum and @xmath12 is the electron density of the system subjected to an external magnetic field @xmath13 , and occupy some of the effective landau levels , namely lambda levels ( @xmath14ls ) , formed by @xmath15 .
a complete filling of @xmath16 such @xmath14ls by non - interacting cfs successfully describes the fqhe states observed at the filling factors @xmath17 as the iqhe of cfs @xcite .
possibilities of fqhe of cfs in a partially filled @xmath14l arise when the interaction between cfs is taken into account .
the observation @xcite of fqhe at certain filling factors in the range @xmath18 suggests the role of the interaction between @xmath19cfs in the partially filled second @xmath14l .
subsequently , the mechanisms @xcite responsible for occurring fqhe of these @xmath19cfs at certain filling factors and thereby producing incompressible states for the observed filling factors have been suggested .
any filling factor within this range may be described by the effective filling factor @xmath20 of @xmath19cfs , as @xmath21 , with @xmath22 ( @xmath23 ) representing partially filled second @xmath14l with filling factor @xmath4 on top of the completely filled lowest @xmath14l .
@xcite have shown that incompressible fqhe states with full and partial spin polarizations can arise at @xmath24 3/8 , when the @xmath19cfs , at @xmath2 , capture two more vortices to turn into @xmath25cfs and form a chiral p - wave anti - pfaffian pairing . recall that anti - pfaffian pairing correlation @xcite is degenerate with its particle - hole conjugate partner , namely pfaffian pairing , @xcite for any two - body interaction .
therefore , three or higher body interaction of @xmath19cfs in the half filled @xmath14l are responsible for breaking the degeneracy . whereas an unusual interaction between the @xmath19cfs , supporting wjs , yi , and quinn ( wyq)@xcite mechanism of an unconventional suppression of cf pairs with relative angular momentum three , at @xmath26 1/3 and 2/3 , creates @xcite fully polarized incompressible states at @xmath24 4/11 and 5/13 , respectively .
minima in the longitudinal resistance , _ albeit _ lack of the signature of hall plateaus , observed by pan .
_ et al._@xcite a decade ago , have also indicated the existence of very weak fqhe states at 3/10 , 4/13 and 5/17 filling factors in the lowest ll .
later , a resonant inelastic light scattering experiment@xcite has supported the emergence of higher order fqhe states in the corresponding filling range @xmath27 .
these filling factors are described in terms of partially filled @xmath14ls produced by negative effective magnetic field experienced by @xmath25cfs as @xmath28 with @xmath29 .
the fqhe at @xmath30 = 3/10 , 4/13 , and 5/17 may be described by the fqhe of @xmath25cfs with a negative effective magnetic field at @xmath4 = 1/2 , 1/3 , and 2/3 respectively , _ a la _ , fqhe of @xmath19cfs with positive effective magnetic field at @xmath4 = 1/2 , 1/3 , and 2/3 describes @xcite fqhe at @xmath30 = 3/8 , 4/11 , and 5/13 .
one may thus expect that the mechanisms responsible for fqhe in the range @xmath31 will be the same for fqhe in the range @xmath32 having same @xmath4 .
this expectation is however not yet theoretically met , primarily because of the technical difficulties regarding fqhe states with negative effective magnetic field for @xmath25cfs . in this paper , we have examined the possibility of forming incompressible fqhe states along with their mechanisms at these filling factors .
we have modified previous algorithm @xcite for determining the projected basis states to an extent that we remove the issue of precision relating to very small values of certain variables .
this has helped us to construct the fully spin polarized low energy cf basis states at the lowest ll filling factors @xmath30 = 3/10 , 4/13 , and 5/17 , where each of the electrons captures four quantized vortices to turn into a @xmath33cf and experiences a negative effective magnetic field which in turn produces @xmath34 and @xmath35 respectively .
then , by the method of diagonalization of full coulomb hamiltonian in these truncated low energy cf basis , known as composite fermion diagonalization ( cfd)@xcite , we show that the incompressible fqhe states should occur at all these three filling factors with appropriate correlations between cfs in their second @xmath14l .
we further estimate the interaction energy of the ground states in the thermodynamic limit , and minimum energy gaps for neutral excitations from the cfd spectra .
the rest of the paper is organized as follows . in sec .
ii , we show the general construction of the relationship between number of flux quanta and number of particles for fully spin polarized states of @xmath36cfs experiencing a negative @xmath15 in the range of filling factors @xmath37 . the cfd technique @xcite used for the calculation of low - lying spectra is reviewed in sec .
origin of the appropriate correlations and occurrence of incompressibility at @xmath0 are described in sec .
iv , while sec .
v contains the similar descriptions for @xmath38 and @xmath7 .
we then estimate the thermodynamic limit of the ground state energy and minimum energy gap for neutral excitation at these filling factors in sec .
vi and summarize our results in sec .
appendix a is devoted for reviewing jain - kamilla @xcite and davenport - simon @xcite algorithms and the description of a modified algorithm developed by us for studying fqhe states with negative effective magnetic fields . in appendix b
, we estimate thermodynamic limit of the interaction energy for the previously proposed parafermionic wave function @xcite at @xmath39 and find that this energy is reasonably higher than our calculated energy using cfd technique .
when @xmath40 , the effective magnetic field for the @xmath25cfs becomes negative , _
i.e. _ , @xmath41 . in that case
, the electronic filling factors are related with the cf filling factors @xmath20 as @xmath42 when @xmath22 with @xmath43 being the filling of the second @xmath14l , we get the desired filling factor in the range @xmath37 . a complete filling of the lowest @xmath14l along with a partial occupation of second @xmath14l at @xmath4 = 1/2 , 1/3 , and 2/3 form fully spin polarized states at @xmath44 = 3/10 , 4/13 , and 5/17 , respectively . in a spherical geometry@xcite , @xmath45 electrons moving on the surface of a sphere experience a radial magnetic field generated by a magnetic monopole of strength @xmath46 placed at the center of that sphere .
an integer number of flux quanta denoted by @xmath47 pass through the surface of the sphere and the @xmath33cfs experience an effective negative flux @xmath48 in units of magnetic flux quantum .
the total number of single particle states in the lowest and the second @xmath14l are @xmath49 and @xmath50 , respectively .
the fractional filling @xmath4 of the second @xmath14l with @xmath51 cfs is related to the effective flux @xmath52 as @xmath53 where the `` flux - shift '' parameter @xmath54 determines topology of the fractional quantum hall state of @xmath25cfs at the filling factor @xmath4 .
the total number of particles is then @xmath55 .
the flux - particle relation for the filling factor @xmath56 in the range @xmath57 becomes @xmath58 we will consider different sets of @xmath59 belonging to several finite size systems at certain filling factors , namely , 3/10 , 4/13 , and 5/17 with specific values of @xmath54 .
exact diagonalization of the coulomb hamiltonian is the most accurate way to study the finite size systems , but the exponential increase of hilbert space dimension restricts us to use this technique for higher number of particles .
we thus use the cfd technique which can deal with much larger systems as it works in a much smaller subspace .
this method has already been proven@xcite to be the next best candidate for calculating energy with considerable accuracy in the neighboring filling factor range @xmath60 . in the cfd@xcite
, the diagonalization is performed in a truncated low energy composite fermion basis which ignores the @xmath14l mixing by neglecting all the configurations involving promotion of cfs into higher @xmath14ls .
this restriction helps us to work within a reduced subspace of exponentially smaller dimension than that of the hilbert space of the lowest ll . as an illustration ,
we tabulate exact dimension of the hilbert space for some of the systems described by @xmath59 at the filling factors 3/10 , 4/13 , and 5/17 in the table [ exactdmnsn ] at various angular momenta , and the reduced dimensions of these systems in the table [ cfddmnsn ] .
we first perform exact diagonalization for a system of particles and fluxes @xmath61 in the lowest ll to know the combinations of the basis states , which provide definite angular momenta .
we then construct states at @xmath20 , by elevating these basis states into the second ll on top of the fully filled lowest ll .
the angular momenta of the latter type of states will be the same as that of the former type of states since the lowest ll is fully filled .
there will be several states denoted by @xmath62 ( @xmath63 s labeling different states with a particular angular momentum ) , corresponding to each angular momentum , @xmath64 , as shown in table [ cfddmnsn ] .
we then composite fermionize ( as @xmath25cfs ) these states with the jastrow factor @xmath65 where @xmath66 with @xmath67 and @xmath68 being two spinor variables for fermions with angular coordinates @xmath69 and @xmath70 .
finally , we construct the state @xmath71 as @xmath72 by projecting onto the lowest ll represented by the operator @xmath73 .
all the @xmath74 s in eq.([wfn_psi ] ) are antisymmetric many body states , which can be obtained from slater determinants consisting of monopole harmonics@xcite of the form @xcite @xmath75 where @xmath76 = 0 , 1 , ... stands for the @xmath14l index , @xmath77 labels the degenerate states of @xmath76th @xmath14l , and the normalization coefficient @xmath78 is given by @xmath79 we develop an efficient algorithm , described in the appendix , for numerical calculation of these many body states @xmath80 for relatively large finite size systems studied by cfd .
though the invariance of total orbital angular momentum gives us the low energy correlated many body basis states @xmath81 at different @xmath64 sectors independently , the basis looses its orthogonality .
we follow the gram - schmidt procedure to obtain an orthogonal basis and then generate the full coulomb hamiltonian matrix in this reduced basis by metropolis monte carlo technique . performing 10
monte carlo runs for each system with @xmath82 monte carlo iterations in each run , we reduce the statistical uncertainty to a desired level .
finally , we diagonalize this matrix in each @xmath64 sector separately to get the low - lying energy eigen spectrum as well as eigenfunctions at any given filling factor .
.full lowest ll hilbert space dimensions in various angular momenta , @xmath64 with its @xmath83-component @xmath84 , for various systems ( @xmath85 ) corresponding to fully spin polarized 3/10 , 4/13 , and 5/17 filling factors . [ cols="^,>,>,>,>,>,>,>,>,>,>,>",options="header " , ]
for estimating ground state energies per electron for the incompressible fqhe states at @xmath38 , 5/17 , and 3/10 in the thermodynamic limit , we first subtract the background energy @xmath86 from the corresponding ground state interaction energies for finite systems , followed by the multiplication with the factor @xmath87 as the density correction and then divide by @xmath45 before plotting ( fig .
( [ gsgap](a ) ) against @xmath88 .
the thermodynamic limit is obtained by extrapolating the fitted lines up to @xmath89 .
these energies are @xmath90 , @xmath91 , and @xmath92 @xcite in the unit of @xmath93 ( where @xmath94 is the background dielectric constant and @xmath95 is the magnetic length ) , at @xmath96 , 4/13 and 5/17 , respectively .
we also plot the lowest energy gap for neutral excitations at these filling factors in fig .
[ gsgap](b ) .
though the results show a consistent non - zero gap for all systems , the strong finite size effects restrict us to do a reliable extrapolation of gaps to the thermodynamic limit . considering the
largest studied systems , we predict that the energy scale for the gap is @xmath97 , in consistence with our recent study @xcite of neutral collective modes . this tiny gap
( about two orders of magnitude lower than the theoretically calculated gap @xcite at neighboring 1/3 state ) indicates that the interaction between the cfs in a fractionally filled @xmath14 level is much weaker than that of electrons in the fractionally filled lowest landau level .
, obtained by cfd at fully spin polarized @xmath98 3/10 ( with apf shift ) , 4/13 and 5/17 are plotted against @xmath88 and extrapolated to the thermodynamic limit .
( b ) the minimum energy gap , @xmath99 , of neutral excitation is plotted as function of @xmath88 .
we do not consider the ground state energy as well as the excitation gap of the smallest systems with n = 6 and 8 to neglect the large deviation in results due to strong finite size effect .
, width=283 ]
the experimental signatures @xcite , though feeble , of the presence of fqhe states at @xmath30= 3/10 , 4/13 , and 5/17 within the range @xmath57 indicate the fqhe of @xmath25cfs arising out of the interaction between them in a partially filled @xmath14l . considering a model of interacting @xmath25cfs in a negative effective magnetic field
, we have shown that an incompressible fqhe state is very likely to occur at fully spin polarized @xmath30 = 3/10 , 4/13 and 5/17 due to unconventional correlations between the composite fermions in partially filled second @xmath14 level .
the character of these three states respectively replicate previously studied fully polarized 3/8 , 4/11 and 5/13 fqhe states which are formed due to @xmath19cfs in a positive effective magnetic field .
the ground states arise at @xmath96 due to an anti - pfaffian type chiral p - wave pairing of cfs in their half - filled second @xmath14l after each of them captures two more vortices , and at @xmath100 and 5/17 due to wyq type unconventional correlations between cfs in the partially filled second @xmath14l .
the small values of minimum energy gap for neutral excitations suggest that the states are very fragile and thus very high quality samples are necessary to obtain prominent quantum hall plateaus .
in this appendix , we first review jain - kamilla @xcite and davenport - simon @xcite algorithms for determining the lowest landau level projection for the basis states of the composite fermions when the effective magnetic field for the cfs is negative , _
i.e. _ , opposite to the applied magnetic field .
we will then find that neither of these algorithms will be suitable for finding cfd spectra for the states with @xmath25cfs and negative effective field .
we thus next develop an algorithm which enables us to determine the cfd spectra for these states .
these single particle states may be obtained @xcite from the following equation : @xmath101 \label{appndx_mh}\end{aligned}\ ] ] where @xmath102 is the jastrow factor for @xmath103-th particle . because of the order of derivatives in eq.([appndx_mh ] ) grows rapidly with @xmath45 ,
the calculation is numerical expensive and it has many other technical problems such as apparent singularity and precision .
we discuss below the algorithms and their limitations for determining these basis states . in jain - kamilla s lll projection
algorithm,@xcite the jastrow factor is pulled through the derivatives and written as @xmath104 \nonumber\\ & & = { j_j}^p\left [ { \widehat{u}_j}^{(|q|+m+s)}{\widehat{v}_j}^{(|q|-m+l - s)}\right],\nonumber\end{aligned}\ ] ] where @xmath105 since the exponents of @xmath106 and @xmath107 increase very rapidly with @xmath45 , their numerical values become huge when two particles come very close during monte carlo steps , and thereby rejection of the subsequent moves occurs . in particular , we can not deal with larger number of particles ( e.g. @xmath108 for @xmath109 ) using this type of projection algorithm .
davenport and simon @xcite developed an alternate method which avoids the troublesome denominators with higher exponents of @xmath110 when two particles come close during monte carlo .
they begin with expanding @xmath111 as @xmath112,\end{aligned}\ ] ] where @xmath113 represents a symmetric polynomial with degree @xmath114 in the @xmath115 variables of @xmath116 for @xmath117
. therefore each element of a slater determinant will have the form @xmath118 now the main job is to find an efficient algorithm for determining the polynomials @xmath119 .
the elementary symmetric polynomials @xmath120 is defined as @xmath121 the explicit form of the polynomials naturally takes very long time to compute .
davenport and simon@xcite calculated them using recursive relations , @xmath122 where @xmath123 and the restricted polynomial @xmath124 .
although the inefficiency occurred for jain - kamilla algorithm @xcite has been removed in this algorithm , it suffers the numerical issue of precision , particularly for large @xmath45 , as the value of @xmath125 for large @xmath126 becomes extremely large when @xmath127 .
this has been sorted out by calculating the numerical variables in a higher precision .
this , however , extremely slows down the numerical calculation .
nonetheless , they were able to find energies for some of the @xmath19cf states with negative effective fields . in case of @xmath25cfs with negative effective field , the form of the projected matrix elements will be @xcite @xmath128 with the summation restricted by the relation @xmath129 the numerical evaluation of eq.([projection_4cf ] ) also suffers an issue of precision , even after calculating the variables in the higher precision , that tremendously slows down the algorithm .
since we are interested in cfd studies involving the states consisting of large number of slater determinants and requiring a huge number of monte carlo iterations , the algorithm is not sufficient for minimizing the time needed for the calculation .
we thus suitably modify this algorithm for our purpose .
to get rid of all the denominators in the polynomials @xmath119 , we redefine the polynomials as @xmath130 by absorbing @xmath131 displayed in eqs .
( [ projection_2cf ] ) and ( [ projection_4cf ] ) . before calculating the modified polynomial @xmath132 ,
we first construct a collection of @xmath115 number of sets , defined by @xmath133 ( @xmath134 ) for @xmath45-th particle , consisting of @xmath135 s @xmath136 only .
the explicit elements of these sets : @xmath137 are the all possible products of @xmath138 $ ] numbers of @xmath135 s arranged in increasing order of @xmath139 . as a set @xmath140 contains @xmath141 number of elements ,
the number of elements in @xmath140 and @xmath142 will be same .
we similarly construct the sets @xmath143 by replacing @xmath144 s in place of @xmath135 s in the sets @xmath140 . in order to obtain the modified polynomials
@xmath145 , we store all the elements of @xmath146 and @xmath147 for @xmath45-th particle only , along with the information about corresponding particle indices of @xmath135 s and @xmath144 s by associating distinct binary numbers , respectively , present in them .
we readily see from the expressions ( [ mod_poly ] and [ sequence ] ) that @xmath148 are pure functions of @xmath135 s and @xmath144 s , respectively .
the other modified polynomials of degree @xmath149 contain both @xmath135 s and @xmath144 s ( @xmath150 ) .
they are @xmath151 where @xmath152 represents the @xmath139-th element of the set @xmath153 , @xmath154 represents the @xmath155-th element of the set @xmath142 , and @xmath156 is the number of elements in either of the sets @xmath143 and @xmath142 .
since the term @xmath157 in eq.([poly_n ] ) contains either the spinor @xmath158 or @xmath159 for any other @xmath76-th particle , it is possible to separate out the terms containing @xmath158 or @xmath159 in @xmath160 : @xmath161 where @xmath162 and @xmath163 are the sum of the terms containing @xmath158 and @xmath164 respectively . for any specific particle index @xmath76 , @xmath165
is obtained by collecting terms containing @xmath159 ; the terms can easily be identified with the above mentioned binary numbers .
the remaining terms automatically constitute @xmath166 .
the functional form of @xmath167 and @xmath168 is different for different particles . using these @xmath162 and @xmath169
, we can easily calculate the polynomials for any particle as @xmath170 therefore , just by calculating the polynomials for @xmath45-th particle , we obtain the polynomials for all other particles in a very simple manner . as we have removed the denominators from the polynomial , this algorithm of calculating different polynomials in recursive method does not suffer any issue of precision in evaluating the wave functions . although this modified algorithm has been used only for @xmath25cfs with negative effective flux , it is useful for cfs with both negative and positive effective flux because eq .
( [ mod_poly ] ) is suitable for any kind of cfs .
, obtained from monte carlo calculation for the parafermionic state at @xmath24 5/17 for @xmath1715 , 10 , and 15 are plotted against @xmath88 and extrapolated to the thermodynamic limit .
since the energy for @xmath172 may have large finite size effect , we consider the thermodynamic limit of the energy by linear fitting with @xmath173 and 15 only.,width=283 ] in this appendix , we calculate energy for the previously proposed @xmath174 parafermion wave function @xcite at the filling factor @xmath175 .
the read - rezayi ( rr ) @xcite parafermion wave functions are proposed as a generalization of pfaffian wave function @xcite with clustering of composite bosons in @xmath176 clusters .
the corresponding wave function for composite bosons with positive effective flux as a trial wave function @xcite for the ground state of fqhe at @xmath177 with the number of clusters @xmath139 reflecting a state of @xmath178 parafermions , and the number of vortices attached to each composite boson @xmath179 , is given by @xmath180 here @xmath181 is an operator that symmetrizes over all possible clustering of @xmath45 particles into @xmath139 sets of equal size .
equation ( [ z3_sequence ] ) reproduces @xmath39 , 3/10 , and 4/13 for @xmath182 and @xmath183 5 , 6 , and 8 respectively .
in spherical geometry , these parafermionic states have a generalized flux - particle relation @xmath184 .
the general flux - shift @xmath185 for all these states differ from that ( [ flux - particle ] ) in our interacting @xmath25cf model . with increasing value of @xmath139 ,
the symmetrization part of the wave function @xmath186 becomes computationally expensive .
we thus choose @xmath39 only for calculating interaction energy using monte carlo method .
figure [ gap_parafermion ] shows the energy per electron along with the consideration of the background interaction energy and density correction factor @xmath187 , for @xmath172 , 10 , and 15 .
we find that the thermodynamic limit of the energy of the state @xmath188 at @xmath39 is @xmath189 which is reasonably higher than the energy obtained in our cfd calculation
. therefore the trial wave function @xmath190 ( [ psi_517 ] ) whose overlap with the cfd ground state is very high ( table-[413ovrlp ] ) , should be the representative wave function for the filling factor @xmath7 .
another wave function for the ground state at @xmath39 was proposed using parafermionic physics .
we estimate ( see appendix [ app_parafermion ] ) the thermodynamic limit of the interaction energy for this state and find that it is reasonably higher than the calculated cfd energy here . | by developing an algorithm for evaluating the basis states for the composite fermions with negative effective magnetic field , we perform the composite - fermion - diagonalization study for the fully spin - polarized fractional quantum hall states at the filling factors @xmath0 , 4/13 , and 5/17 in the range @xmath1 . these observed states correspond to partially filled second effective landau level , for the composite fermions carrying four vortices , with filling factor @xmath2 , 1/3 , and 2/3 respectively , analogous to the previously studied states of composite fermions with two attached vortices in the range @xmath3 .
we show that the character of these states in the range @xmath1 replicates the same in the range @xmath3 having identical @xmath4 : chiral p - wave pairing with anti - pfaffian correlation of composite fermions carrying six quantized vortices produces incompressible state at @xmath0 ; an unconventional interaction between composite fermions , resulting from the suppression of fermion pairs with relative angular momentum three and producing fractional quantum hall effect of composite fermions in the second effective landau level with @xmath5 and its particle - hole conjugate filling factor 2/3 , reproduces incompressible states at @xmath6 and @xmath7 filling factors .
we further estimate the thermodynamic limit of the ground state energies and calculate the lowest energy gap for neutral collective excitations of these states . |
in the late 1980s the idea of using _ argumentation _ to model nonmonotonic reasoning emerged ( see @xcite as well as @xcite for excellent overviews ) . nowadays argumentation theory is a vibrant subfield of artificial intelligence , covering aspects of knowledge representation , multi - agent systems , and also philosophical questions . among other approaches which have been proposed for capturing representative patterns of inference in argumentation theory @xcite , dung
s abstract argumentation frameworks ( afs ) @xcite play an important role within this research area . at the heart of dung s
approach lie the so - called _ argumentation semantics _
@xcite for an excellent overview ) . given an af @xmath0 , which is set - theoretically just a directed graph encoding arguments and attacks between them , a certain argumentation semantics @xmath1 returns acceptable sets of arguments @xmath2 , so - called @xmath1-_extensions_. each of these sets represents a reasonable position w.r.t .
@xmath0 and @xmath1 . over the last 20 years a series of abstract argumentation semantics were introduced .
the motivations of these semantics range from the desired treatment of specific examples to fulfilling a number of abstract principles .
the comparison via abstract criteria of the different semantics available is a topic which emerged quite recently in the community ( @xcite can be seen as the first paper in this line ) .
our work takes a further step towards a comprehensive understanding of argumentation semantics .
in particular , we study the following question : do we really need the entire af @xmath0 to compute a certain argumentation semantics @xmath1 ?
in other words , is it possible to unambiguously determine acceptable sets w.r.t .
@xmath1 , given only partial information of the underlying framework @xmath0 . in order to solve this problem let us start with the following reflections : 1 . as a matter of fact , one basic requirement of almost all existing semantics
is that of conflict - freeness , i.e. arguments within a reasonable position are not allowed to attack each other .
consequently , knowledge about conflict - free sets is an essential part for computing semantics .
2 . the second step is to ask the following : which information on top on conflict - free sets has to be added ?
imagine the set of conflict - free sets given by @xmath3 .
consequently , there has to be at least one attack between @xmath4 and @xmath5 .
unfortunately , this information is not sufficient to compute any standard semantics ( except naive extensions , which are defined as @xmath6-maximal conflict - free sets ) since we know nothing precise about the neighborhood of @xmath4 and @xmath5 .
the following three afs possess exactly the mentioned conflict - free sets , but differ with respect to other + ( a1 ) at ( -0.6,0 ) [ circle , thick , draw , label = left:@xmath7@xmath4 ; ( b1 ) at ( 0.4,0 ) [ circle , thick , draw]@xmath5 ; + ( a2 ) at ( 2,0 ) [ circle , thick , draw , label = left:@xmath8 @xmath4 ; ( b2 ) at ( 3.0,0 ) [ circle , thick , draw ] @xmath5 ; + ( a3 ) at ( 4.6,0 ) [ circle , thick , draw , label = left:@xmath9 @xmath4 ; ( b3 ) at ( 5.6,0 ) [ circle , thick , draw ] @xmath5 ; + ( a1 ) to [ thick , bend right ] ( b1 ) ; ( b2 ) to [ thick , bend right ] ( a2 ) ; ( a3 ) to [ thick , bend right ] ( b3 ) ; ( b3 ) to [ thick , bend right ] ( a3 ) ; 3 .
the final step is to try to minimize the added information .
that is , which kind of knowledge about the neighborhood is somehow dispensable in the light of computation ?
clearly , this will depend on the considered semantics .
for instance , in case of stage semantics @xcite , which requests conflict - free sets of maximal range , we do not need any information about incoming attacks .
this information can not be omitted in case of admissible - based semantics since incoming attacks require counterattacks .
the above considerations motivate the introduction of so - called verification classes specifying a certain amount of information . in a first step ,
we study the relation of these classes to each other . we therefore introduce the notion of being _ more informative _ capturing the intuition that a certain class can reproduce the information of an other .
we present a hierarchy w.r.t . this ordering .
the hierarchy contains 15 different verification classes only .
this is due to the fact that many syntactically different classes collapse to the same amount of information .
we then formally define the essential property of a semantics @xmath1 being _ verifiable _ w.r.t . a certain verification class .
we present a general theorem stating that any _ rational _ semantics is exactly verifiable w.r.t .
one of the @xmath10 different verification classes . roughly speaking ,
a semantics is rational if attacks inbetween two self - loops can be omitted without affecting the set of extensions .
an important aside hereby is that even the most informative class contains indeed less information than the entire framework by itself .
in this paper we consider a representative set of standard semantics .
all of them satisfy rationality and thus , are exactly verifiable w.r.t . a certain class .
since the theorem does not provide an answer to which verification class perfectly matches a certain rational semantics we study this problem one by one for any considered semantics . as a result , only @xmath11 different classes are essential to classify the considered standard semantics . in the last part of the paper we study an application of the concept of verifiability .
more precisely , we address the question of strong equivalence for semantics lying inbetween known semantics , so - called intermediate semantics .
strong equivalence is the natural counterpart to ordinary equivalence in monotonic theories ( see @xcite for abstract argumentation and @xcite for other nonmonotonic theories ) .
we provide characterization theorems relying on the notion of verifiability and thus , contributing to a more abstract understanding of the different features argumentation semantics offer . besides these main results
, we also give new characterizations for strong equivalence with respect to naive extensions and strong admissible sets .
an _ argumentation framework _ ( af ) @xmath12 is a directed graph whose nodes @xmath13 ( with @xmath14 being an infinite set of arguments , so - called _ universe _ ) are interpreted as _
arguments _ and whose edges @xmath15 represent _ conflicts _ between them .
we assume that all afs possess finitely many arguments only and denote the collection of all afs by @xmath16 .
if @xmath17 we say that @xmath4 _ attacks _ @xmath5 .
alternatively , we write @xmath18 as well as , for some @xmath19 , @xmath20 or @xmath21 if there is some @xmath22 attacked by @xmath4 or attacking @xmath5 , respectively .
an argument @xmath23 is _ defended _ by a set @xmath24 if for each @xmath25 with @xmath26 , @xmath27 .
we define the _ range _ of @xmath28 ( in @xmath0 ) as @xmath29 .
similarly , we use @xmath30 to denote the _ anti - range _ of @xmath28 ( in @xmath0 ) as @xmath31 .
furthermore , we say that a set @xmath28 is _ conflict - free _ ( in @xmath0 ) if there is no argument @xmath32 s.t .
the set of all conflict - free sets of an af @xmath0 is denoted by @xmath34 . for an af @xmath35
we use @xmath36 and @xmath37 to refer to @xmath38 and @xmath28 , respectively .
furthermore , we use @xmath39 for the set of all self - defeating arguments .
finally , we introduce the union of afs @xmath0 and @xmath40 as @xmath41 . a _ semantics _
@xmath1 assigns to each @xmath12 a set @xmath42 where the elements are called @xmath1-_extensions_. numerous semantics are available .
each of them captures different intuitions about how to reason about conflicting knowledge .
we consider @xmath43 for admissible , naive , stable , preferred , complete , grounded , semi - stable , stage , ideal , and eager semantics @xcite .
[ def : semantics ] given an af @xmath12 and let @xmath44 .
1 . @xmath45 iff @xmath46 and each @xmath47 is defended by @xmath48 , 2 .
@xmath49 iff @xmath46 and there is no @xmath50 s.t .
@xmath51 , 3 .
@xmath52 iff @xmath46 and @xmath53 , 4 .
@xmath54 iff @xmath45 and there is no @xmath55 s.t .
@xmath51 , 5 .
@xmath56 iff @xmath45 and for any @xmath23 defended by @xmath48 , @xmath47 , 6 .
@xmath57 iff @xmath56 and there is no @xmath58 s.t .
@xmath59 , 7 .
@xmath60 iff @xmath45 and there is no @xmath55 s.t .
@xmath61 , 8 .
@xmath62 iff @xmath46 and there is no @xmath50 s.t .
@xmath61 , 9 .
@xmath63 iff @xmath45 , @xmath64 and there is no @xmath55 satisfying @xmath65 s.t .
@xmath51 , 10 .
@xmath66 iff @xmath45 , @xmath67 and there is no @xmath55 satisfying @xmath68 s.t .
@xmath69 . for two semantics @xmath1 ,
@xmath70 we use @xmath71 to indicate that @xmath72 for each af @xmath73 .
if we have @xmath74 and @xmath71 for semantics @xmath75 , we say that @xmath1 is _ @xmath76-@xmath70-intermediate_. well - known relations between semantics are @xmath77 , meaning , for instance , that @xmath78 is @xmath79-@xmath80-intermediate .
[ def : semantics_conditions ] we call a semantics @xmath1 _ rational _ if self - loop - chains are irrelevant .
that is , for every af @xmath81 it holds that @xmath82 , where @xmath83 .
indeed , all semantics introduced in definition [ def : semantics ] are rational .
a prominent semantics that is based on conflict - free sets , but is not rational is the @xmath84-semantics @xcite , since here chains of self - loops can have an influence on the sccs of an af ( see also @xcite ) .
the following definition captures the two main notions of equivalence available for non - monotonic formalisms , namely _ ordinary _ ( or _ standard _ ) _ equivalence _ and _ strong _ ( or _ expansion _ ) _ equivalence_. a detailed overview of equivalence notion including their relations to each other can be found in @xcite .
given a semantics @xmath1
. two afs @xmath0 and @xmath40 are * _ standard equivalent _ w.r.t .
@xmath1 ( @xmath85 ) iff @xmath86 , * _ expansion equivalent _ w.r.t .
@xmath1 ( @xmath87 ) iff for all afs @xmath88 : @xmath89 expansion equivalence can be decided syntactically via so - called _ kernels _ @xcite .
a kernel is a function @xmath90 mapping each af @xmath0 to another af @xmath91 ( which we may also denote as @xmath92 ) . consider the following definitions .
[ def : kernel ] given an af @xmath12 and a semantics @xmath1 .
we define @xmath1-kernels @xmath93 whereby @xmath94 = @xmath95 , + @xmath96 + @xmath97 , + @xmath98 + @xmath99 , + @xmath100 .
we say that a relation @xmath101 is _ characterizable through kernels _ if there is a kernel @xmath102 , s.t .
@xmath103 iff @xmath104 .
moreover , we say that a semantics @xmath1 is _ compatible with a kernel _
@xmath105 if @xmath106 iff @xmath107 .
all semantics ( except naive semantics ) considered in this paper are compatible with one of the four kernels introduced above . in the next section
, we will complete these results taking naive semantics and strong admissible sets into account .
@xcite [ the : strong ] for any afs @xmath81 and @xmath108 , 1 .
@xmath109 with @xmath110 , 2 .
3 . @xmath111
in order to provide an exhaustive analysis of intermediate semantics ( confer penultimate section ) we provide missing kernels for naive semantics as well as strongly admissible sets .
we start with the so - called _ naive kernel _ characterizing expansion equivalence w.r.t .
naive semantics . as an aside
, the following kernel is the first one which adds attacks to the former attack relation .
[ def : naivekernel ] given an af @xmath12 .
we define the _ naive kernel _ @xmath112
whereby @xmath113 the following example illustrates the definition above .
consider the afs @xmath81 and @xmath108 .
note that @xmath114 .
consequently , @xmath115 .
\(a ) at ( -0.6,0 ) [ circle , thick , draw , label = left:@xmath7@xmath4 ; ( b ) at ( 0.6,0 ) [ circle , thick , draw]@xmath5 ; ( c ) at ( 1.8,0 ) [ circle , thick , draw ] @xmath116 ; ( d ) at ( 3.0,0 ) [ circle , thick , draw ] @xmath117 ; ( a ) at ( -0.6,-1.0 ) [ circle , thick , draw , label = left:@xmath8@xmath4 ; ( b ) at ( 0.6,-1.0 ) [ circle , thick , draw]@xmath5 ; ( c ) at ( 1.8,-1.0 ) [ circle , thick , draw ] @xmath116 ; ( d ) at ( 3.0,-1.0 ) [ circle , thick , draw ] @xmath117 ; \(b ) to [ thick , loop , distance=0.5 cm ] ( b ) ; ( d ) to [ thick , bend right ] ( c ) ; ( b ) to [ thick , loop , distance=0.5 cm ] ( b ) ; ( c ) to [ thick , bend right ] ( d ) ; ( a ) to [ thick , bend right ] ( b ) ; ( b ) to [ thick , bend right ] ( c ) ; in accordance with definition [ def : naivekernel ] we observe that both afs possess the same naive kernel @xmath118 .
\(a ) at ( -0.6,0 ) [ circle , thick , draw , label = left:@xmath9@xmath4 ; ( b ) at ( 0.6,0 ) [ circle , thick , draw]@xmath5 ; ( c ) at ( 1.8,0 ) [ circle , thick , draw ] @xmath116 ; ( d ) at ( 3.0,0 ) [ circle , thick , draw ] @xmath117 ; \(b ) to [ thick , loop , distance=0.5 cm ] ( b ) ; ( d ) to [ thick , bend right ] ( c ) ; ( c ) to [ thick , bend right ] ( d ) ; ( a ) to [ thick , bend right ] ( b ) ; ( b ) to [ thick , bend right ] ( a ) ; ( b ) to [ thick , bend right ] ( c ) ; ( c ) to [ thick , bend right ] ( b ) ; the following theorem proves that possessing the same kernels is necessary as well as sufficient for being strongly equivalent , i.e. @xmath119 . for all afs @xmath0,@xmath40 , @xmath120 in @xcite
it was already shown that @xmath121 iff jointly @xmath122 and @xmath123 .
consequently , it suffices to prove that @xmath124 implies @xmath122 as well as @xmath123 and vice versa .
( @xmath125 ) given @xmath124 . by definition [ def : naivekernel ]
we immediately have @xmath122 .
assume now that @xmath126 and without loss of generality let @xmath127 . obviously , for any af @xmath128 , @xmath129 .
hence , there is an @xmath130 , s.t .
@xmath131 . thus , there are @xmath132 , s.t .
@xmath133 . furthermore ,
@xmath134 and since for any af @xmath128 , @xmath135 we obtain @xmath136 . consequently , we have to consider @xmath137 .
since @xmath133 , we obtain @xmath138 .
since @xmath124 is assumed we derive @xmath139 . by definition
[ def : naivekernel ] we must have @xmath140 contradicting the conflict - freeness of @xmath130 in @xmath108 .
( @xmath141 ) we show the contrapositive , i.e. @xmath142 implies @xmath143 or @xmath144 .
observe that for any af @xmath128 , @xmath145 .
consequently , if @xmath146 , then @xmath147 .
assume now @xmath148 .
without loss of generality let @xmath149 . since for any af @xmath128 , @xmath135 we obtain @xmath137 .
furthermore , @xmath150 implies @xmath151 and consequently , for any @xmath152 , @xmath153 . since @xmath154 we deduce @xmath155 .
hence , @xmath156 and thus , there exists a set @xmath157 , s.t .
@xmath158 ( compare ( * ? ? ?
* lemma 3 ) ) witnessing @xmath144 .
we turn now to _ strongly admissible sets _
( for short , @xmath159 ) @xcite
. we will show that , beside grounded @xcite and resolution based grounded semantics @xcite , strongly admissible sets are characterizable through the grounded kernel .
consider the following self - referential definition taken from @xcite .
[ def : strad ] given an af @xmath12 .
a set @xmath24 is _ strongly admissible _ , i.e. @xmath160 iff any @xmath161
is defended by a strongly admissible set @xmath162 .
the following properties are needed to prove the characterization theorem .
the first two of them are already shown in @xcite .
the third statement is an immediate consequence of the former .
[ pro : strad : prop ] given two afs @xmath0 and @xmath108 , then 1 .
@xmath163 , 2 .
if @xmath164 we have : @xmath165 for all @xmath166 , and 3 .
@xmath167 implies @xmath168 .
the following definition provides us with an alternative criterion for being a strong admissible set .
in contrast to the former it allows one to construct strong admissible sets step by step .
thus , a construction method is given .
[ def : stradnew ] given an af @xmath12 .
a set @xmath24 is _ strongly admissible _
, i.e. @xmath160 iff there are finitely many and pairwise disjoint sets @xmath169 , s.t . @xmath170 and @xmath171 is the so - called _ characteristic function _
@xcite with @xmath172 .
the term @xmath173 can be equivalently replaced by @xmath174 . ] and furthermore , @xmath175 .
[ pro : stradnew ] definitions [ def : strad ] and [ def : stradnew ] are equivalent .
for the proof we use @xmath176 as a shorthand for @xmath160 in the sense of definition @xmath105 .
@xmath177 given @xmath178 . hence , there is a finite partition , s.t .
@xmath170 , @xmath171 and @xmath175 .
observe that @xmath179 for any @xmath180 .
let @xmath161 .
consequently , there is an index @xmath181 , s.t .
@xmath182 . furthermore , since @xmath183 by definition
, we deduce that @xmath184 defends @xmath4 .
we have to show now that ( the smaller set w.r.t .
@xmath6 ) @xmath185 .
note that @xmath186 .
since we are dealing with finite afs we may iterate our construction . hence ,
no matter which elements are chosen we end up with a @xmath6-chain , s.t . @xmath187 and @xmath188 defends @xmath189 for some index @xmath190 , set @xmath191 and element @xmath192 .
this means , the question whether @xmath193 can be decided positively by proving @xmath194 . since the empty set does not contain any elements we find @xmath194 concluding @xmath195 .
+ @xmath196 given @xmath193 , consider the following sets @xmath197 : @xmath198 , @xmath199 , @xmath200 ,
, @xmath201 .
since we are dealing with finite afs there has to be a natural @xmath202 , s.t .
@xmath203 . consider now the union of these sets , i.e. @xmath204 .
we show now that @xmath205 and @xmath206 . by construction
we have @xmath207 .
moreover , @xmath208 . this can be seen as follows . by definition @xmath209 .
this means , @xmath210 .
since @xmath211 contains all elements defended by @xmath212 we obtain @xmath205 .
obviously , @xmath213 .
in order to derive a contradiction we suppose @xmath214 .
this means there is a nonempty set @xmath215 , s.t .
let @xmath217 .
observe that no element @xmath218 is defended by @xmath204 ( * ) .
since @xmath193 we obtain a set @xmath219 , s.t . @xmath220 and @xmath221 defends @xmath222 .
we now iterate this procedure ending up with a set @xmath223 , s.t . @xmath224 and @xmath225 defends @xmath226 contradicting ( * ) and concluding the proof .
the following example shows how to use the new construction method .
consider the following af @xmath0 .
\(a ) at ( -0.6,-0.6 ) [ minimum size = 7 mm , circle , thick , draw , label = left:@xmath227@xmath4 ; ( b ) at ( 0.6,-0.3 ) [ minimum size = 7 mm , circle , thick , draw]@xmath5 ; ( c ) at ( 1.8,0 ) [ minimum size = 7 mm , circle , thick , draw ] @xmath116 ; ( d ) at ( 1.8,-1.2 ) [ minimum size = 7 mm , circle , thick , draw ] @xmath117 ; ( e ) at ( 3.0,-0.6 ) [ minimum size = 7 mm , circle , thick , draw ] @xmath228 ; ( f ) at ( 4.2,-0.6 ) [ minimum size = 7 mm , circle , thick , draw ] @xmath229 ; \(e ) to [ thick , loop , distance=0.5 cm ] ( e ) ; ( a ) to [ thick , bend right ] ( b ) ; ( b ) to [ thick , bend left ] ( c ) ; ( c ) to [ thick , bend right ] ( e ) ; ( e ) to [ thick , bend right ] ( f ) ; ( f ) to [ thick , bend right ] ( e ) ; ( d ) to [ thick , bend right ] ( e ) ; we have @xmath230 .
hence , for all @xmath231 , @xmath160 .
furthermore , @xmath232 , @xmath233 and @xmath234 .
this means , additionally @xmath235 . finally , @xmath236 justifying the last missing set @xmath237 .
the following corollary is an immediate consequence of definition [ def : stradnew ] .
it is essential to prove the characterization theorem for strongly admissible sets .
[ cor : strad ] given an af @xmath0 and two sets @xmath238 . if @xmath38 defends @xmath239 , then @xmath240 is strong admissible if @xmath38 is .
the following lemma shows that the grounded kernel is insensitive w.r.t
. strong admissible sets .
[ lem : strad ] for any af @xmath0 , @xmath241 .
the grounded kernel is node- and loop - preserving , i.e. @xmath242 and @xmath243 .
furthermore , @xmath244 and @xmath245 as shown in ( * ? ? ?
* lemma 6 ) .
+ ( @xmath6 ) given @xmath246 .
the proof is by induction on @xmath247 indicating the number of sets forming a suitable ( according to definition [ def : stradnew ] ) partition of @xmath28 .
let @xmath248 . in consideration of the grounded kernel we observe @xmath249 , i.e. the set of unattacked arguments does not change .
since @xmath250 is assumed we are done .
assume now that the assertion is proven for any @xmath105-partition .
let @xmath28 be a @xmath251-partition , i.e. @xmath252 . according to induction hypothesis as well as corollary [ cor : strad ] it suffices to prove @xmath253 defends @xmath254 in @xmath255 .
assume not , i.e. there are arguments @xmath256 , @xmath257 s.t .
@xmath258 and for all @xmath259 , @xmath260 ( * ) .
since @xmath253 defends @xmath254 in @xmath81 we deduce the existence of an argument @xmath261 s.t .
thus , @xmath263 is redundant w.r.t . the grounded kernel . according to definition [ def : kernel ] and due to the conflict - freeness of @xmath253 we have @xmath264 and @xmath265 .
consequently , @xmath266 . since @xmath253 is a strong admissible @xmath105-partition in @xmath81 we obtain by induction hypothesis that @xmath253 is strong admissible in @xmath255 and therefore , admissible in @xmath255 ( proposition [ pro : strad : prop ] ) .
hence there has to be an argument @xmath259 , s.t .
@xmath267 , contradicting ( * ) .
( @xmath268 ) assume @xmath269 .
we show @xmath246 by induction on @xmath247 indicating that @xmath28 is a @xmath247-partition in @xmath255 . due to
@xmath249 the base case is immediately clear . for the induction step let @xmath28 be a @xmath251-partition ,
i.e. @xmath252 . by induction hypothesis we may assume that @xmath253 is strongly admissible in @xmath81 .
using corollary [ cor : strad ] it suffices to prove @xmath253 defends @xmath254 in @xmath81 .
assume not , i.e. there are arguments @xmath256 , @xmath257 s.t .
@xmath270 and for all @xmath259 , @xmath271 .
we even have @xmath260 since @xmath272 . consequently
, @xmath273 has to be deleted in @xmath255 . definition [ def : kernel ] requires @xmath274 contradicting the conflict - freeness of @xmath28 in @xmath255 .
for any two afs @xmath81 and @xmath108 we have , @xmath275 ( @xmath141 ) we show the contrapositive , i.e. @xmath276 . assuming @xmath277 implies @xmath278 ( theorem [ the : strong ] ) .
this means , there is an af @xmath128 , s.t .
@xmath279 . due to statement 3 of proposition
[ pro : strad : prop ] , we deduce @xmath280 proving @xmath281 .
( @xmath125 ) given @xmath282 . since
expansion equivalence is a congruence w.r.t .
@xmath283 we obtain @xmath284 for any af @xmath128 .
consequently , @xmath285 . due to lemma
[ lem : strad ] we deduce @xmath286 , concluding the proof .
in this section we study the question whether we really need the entire af @xmath81 to compute the extensions of a given semantics .
let us consider naive semantics .
obviously , in order to determine naive extensions it suffices to know all conflict - free sets .
conversely , knowing @xmath287 only does not allow to reconstruct @xmath81 unambiguously .
this means , knowledge about @xmath287 is indeed less information than the entire af by itself .
in fact , most of the existing semantics do not need information of the entire framework .
we will categorize the amount of information by taking the conflict - free sets as a basis and distinguish between different amounts of knowledge about the neighborhood , that is range and anti - range , of these sets .
[ def : classes ] we call a function @xmath288 ( @xmath289 ) which is expressible via basic set operations only _ neighborhood function_. a neighborhood function @xmath290 induces the _ verification class _ mapping each af @xmath81 to @xmath291 we coined the term neighborhood function because the induced verification classes apply these functions to the neighborhoods , i.e. range and anti - range of conflict - free sets .
the notion of _ expressible via basic set operations _ simply means that ( in case of @xmath292 ) the expression @xmath293 is in the language generated by the following bnf : @xmath294 consequently , in case of @xmath248 , we may distinguish eight set theoretically different neighborhood functions , namely @xmath295 a verification class encapsulates a certain amount of information about an af , as the following example illustrates .
[ ex : verification_class ] consider the following af @xmath81 : \(a ) at ( -0.6,0 ) [ circle , thick , draw , label = left:@xmath7@xmath4 ; ( b ) at ( 1,0 ) [ circle , thick , draw]@xmath5 ; \(c ) at ( 2.6,0 ) [ circle , thick , draw ] @xmath116 ; \(b ) to [ thick , loop , distance=0.5 cm ] ( b ) ; \(b ) to [ thick , bend right ] ( a ) ; ( a ) to [ thick , bend right ] ( b ) ; ( c ) to [ thick , bend right ] ( b ) ; now take , for instance , the verification class induced by @xmath296 , that is @xmath297 , storing information about conflict - free sets together with their associated ranges w.r.t .
it contains the following tuples : @xmath298 , @xmath299 , @xmath300 , and @xmath301 .
the verification class induced by @xmath302 contains the same tuples but @xmath303 instead of @xmath299 .
intuitively , it should be clear that the set @xmath304 suffices to compute stage extensions ( i.e. , range - maximal conflict - free sets ) of @xmath81 .
this intuitive understanding of _ verifiability _ will be formally specified in definition [ def : verifiable ] .
note that a neighborhood function @xmath290 may return @xmath247-tuples .
consequently , in consideration of the eight listed basic function we obtain ( modulo reordering , duplicates , empty set ) @xmath305 syntactically different neighborhood functions and therefore the same number of verification classes . as usual
, we will denote the @xmath247-ary combination of basic functions @xmath306 as @xmath307 with @xmath308 . with the following definition we can put neighborhood functions into relation w.r.t . their information
this will help us to show that actually many of the induced classes collapse to the same amount of information .
[ def : informative ] given neighborhood functions @xmath290 and
@xmath309 returning @xmath247-tuples and @xmath310-tuples , respectively , we say that @xmath290 is _ more informative _ than @xmath309
, for short @xmath311 , iff there is a function @xmath312 such that for any two sets of arguments @xmath313 , we have @xmath314 .
we will denote the strict part of @xmath315 by @xmath316 , i.e. @xmath317 iff @xmath311 and @xmath318 .
moreover @xmath319 in case @xmath311 and @xmath320 , we say that @xmath290 _ represents _ @xmath309 and vice versa .
\(a ) [ matrix of math nodes , column sep=0.41 cm , row sep=0.5 cm ] & & & + - + + & + & & & & - & - + + & & & & & & - + & & & + ; i / in 2 - 1/1 - 4 , 2 - 2/1 - 4 , 2 - 3/1 - 4 , 2 - 5/1 - 4 , 2 - 6/1 - 4 , 2 - 7/1 - 4 , 3 - 1/2 - 1 , 3 - 1/2 - 2 , 3 - 2/2 - 1 , 3 - 2/2 - 3 , 3 - 2/2 - 6 , 3 - 3/2 - 1 , 3 - 3/2 - 5 , 3 - 3/2 - 7 , 3 - 4/2 - 3 , 3 - 4/2 - 5 , 3 - 5/2 - 2 , 3 - 5/2 - 5 , 3 - 5/2 - 6 , 3 - 6/2 - 2 , 3 - 6/2 - 3 , 3 - 6/2 - 7 , 3 - 7/2 - 6 , 3 - 7/2 - 7 , 4 - 4/3 - 1 , 4 - 4/3 - 2 , 4 - 4/3 - 3 , 4 - 4/3 - 4 , 4 - 4/3 - 5 , 4 - 4/3 - 6 , 4 - 4/3 - 7 ( a - i ) ( a- ) ; [ lemma : verification_classes ] all neighborhood functions are represented by the ones depicted in figure [ fig : verification_classes ] and the @xmath321-relation represented by arcs in figure [ fig : verification_classes ] holds .
we begin by showing that all neighborhood functions are represented in figure [ fig : verification_classes ] .
clearly , each neighborhood function @xmath290 represents itself , i.e. @xmath322 . all neighborhood functions for @xmath248 are are depicted in figure [ fig : verification_classes ] .
we turn to @xmath323 .
consider the neighborhood functions @xmath324 , @xmath325 , and @xmath326 , defined as @xmath327 , @xmath328 , and @xmath329 for @xmath313 .
observe that @xmath330 .
hence , we can easily define functions in the spirit of definition [ def : informative ] mapping the images of the function to one another : * @xmath331 ; * @xmath332 ; * @xmath333 .
therefore , @xmath334 .
in particular , they are all represented by @xmath335 .
we can apply the same reasoning to other combinations of neighborhood functions and get the following equivalences w.r.t .
information content : @xmath336 ; @xmath337 ; @xmath338 ; @xmath339 ; and @xmath340 , with the functions stated first acting as representatives in figure [ fig : verification_classes ] . for the remaining functions returning @xmath341-tuples
we get @xmath342 by * @xmath343 ; * @xmath344 ; * @xmath345 . finally , every neighborhood function @xmath346 with @xmath347 is represented by @xmath348 since we can compute all possible sets from @xmath28 and @xmath130 .
now consider two functions @xmath290 and @xmath309 such that there is an arrow from @xmath349 to @xmath350 in figure [ fig : verification_classes ] .
it is easy to see that @xmath320 since , for sets of arguments @xmath28 and @xmath130 , @xmath307 is either contained in @xmath351 or obtainable from @xmath351 by basic set operations .
the fact that @xmath352 , entailing @xmath353 , follows from the impossibility of finding a function @xmath354 such that @xmath355 .
if the information provided by a neighborhood function is sufficient to compute the extensions , we say the semantics is verifiable by the class induced by the neighborhood function .
[ def : verifiable ] a semantics @xmath1 is _ verifiable _ by the verification class induced by the neighborhood function @xmath290 returning @xmath247-tuples ( or simply , @xmath349__-verifiable _ _ ) iff there is a function ( also called _ criterion _ ) @xmath356 s.t .
for every af @xmath357 we have : @xmath358 moreover , @xmath1 is _ exactly @xmath349-verifiable _ iff @xmath1 is @xmath349-verifiable and there is no verification class induced by @xmath309 with @xmath359 such that @xmath1 is @xmath350-verifiable . observe that if a semantics @xmath1 is @xmath349-verifiable then for any two afs @xmath81 and @xmath108 with @xmath360 and @xmath361 it must hold that @xmath362 .
we proceed with a list of criteria showing that any semantics mentioned in definition [ def : semantics ] is verifiable by a verification class induced by a certain neighborhood function . in the following , we abbreviate the tuple @xmath363 by @xmath364 .
@xmath365 @xmath366 instead of a formal proof we give the following explanations .
first of all it is easy to see that the naive semantics is verifiable by the verification class induced by @xmath367 since the naive extensions can be determined by the conflict - free sets .
stable and stage semantics , on the other hand , utilize the range of each conflict - free set in addition .
hence they are verifiable by the verification class induced by @xmath296 .
now consider admissible sets . recall that a conflict - free @xmath28 set is admissible if and only if it attacks all attackers .
this is captured exactly by the condition @xmath368 , hence admissible sets are verifiable by the verification class induced by @xmath369 .
the same holds for preferred semantics , since we just have to determine the maximal conflict - free sets with @xmath368 .
semi - stable semantics , however , needs the range of each conflict - free set in addition , see @xmath370 , which makes it verifiable by the verification class induced by @xmath371 . finally consider the criterion @xmath372 .
the first two conditions for a set of arguments @xmath28 stand for conflict - freeness and admissibility , respectively . now assume the third condition does not hold , i.e. , there exists a tuple @xmath373 with @xmath374 and @xmath375 .
this means that every argument attacking @xmath376 is attacked by @xmath28 , i.e. , @xmath376 is defended by @xmath28 .
hence @xmath28 is not a complete extension , showing that @xmath377 for each @xmath378 .
one can verify that all criteria from the list are adequate in the sense that they describe the extensions of the corresponding semantics .
we show now that the formal concepts of verifiability and being more informative behave correctly in the sense that the use of more informative neighborhood functions do not lead to a loss of verification capacity .
if a semantics @xmath1 is @xmath349-verifiable , then @xmath1 is verifiable by all verification classes induced by some @xmath309 with @xmath320 . as @xmath1 is verifiable by the verification class induced by @xmath290 it holds that there is some @xmath379 such that for all @xmath357 , @xmath380
now let @xmath320 , meaning that there is some @xmath354 such that @xmath381 .
we define @xmath382 and observe that @xmath383 , hence @xmath384 for each @xmath378 . in order to prove unverifiability of a semantics @xmath1 w.r.t . a class induced by a certain @xmath290 it suffices to present two afs @xmath81 and @xmath108 such that @xmath385 but , @xmath360 and @xmath386 .
then the verification class induced by @xmath290 does not provide enough information to verify @xmath1 .
in the following we will use this strategy to show exact verifiability .
consider a semantics @xmath1 which is verifiable by a class induced by @xmath290 . if @xmath1 is unverifiable by all verifiability classes induced by @xmath309 with @xmath359 we have that @xmath1 is exactly verifiable by @xmath290 .
the following examples study this issue for the semantics under consideration .
[ ex : exact_verify ] the complete semantics is @xmath387-verifiable as seen before . the following afs show that it is even exactly verifiable by that class .
( a1 ) at ( -0.6,0 ) [ circle , thick , draw , label = left:@xmath388@xmath4 ; ( b1 ) at ( 0.8,0 ) [ circle , thick , draw]@xmath5 ; ( a1 ) at ( 3.4,0 ) [ circle , thick , draw , label = left:@xmath389 @xmath4 ; ( b1 ) at ( 4.8,0 ) [ circle , thick , draw ] @xmath5 ; ( b1 ) to [ thick , loop , distance=0.5 cm ] ( b1 ) ; ( b1 ) to [ thick , loop , distance=0.5 cm ] ( b1 ) ; ( b1 ) to [ thick , bend right ] ( a1 ) ; ( a2 ) at ( -0.6,-1 ) [ circle , thick , draw , label = left:@xmath390@xmath4 ; ( b2 ) at ( 0.8,-1 ) [ circle , thick , draw]@xmath5 ; ( c2 ) at ( 2,-1 ) [ circle , thick , draw]@xmath116 ; ( a2 ) at ( 3.4,-1 ) [ circle , thick , draw , label = left:@xmath391 @xmath4 ; ( b2 ) at ( 4.8,-1 ) [ circle , thick , draw ] @xmath5 ; ( c2 ) at ( 6.2,-1 ) [ circle , thick , draw ] @xmath116 ; ( b2 ) to [ thick , loop , distance=0.5 cm ] ( b2 ) ; ( b2 ) to [ thick , loop , distance=0.5 cm ] ( b2 ) ; ( b2 ) to [ thick , bend right ] ( c2 ) ; ( c2 ) to [ thick , bend right ] ( b2 ) ; ( a2 ) to [ thick , bend right ] ( b2 ) ; ( c2 ) to [ thick , bend right ] ( b2 ) ; ( b2 ) to [ thick , bend right ] ( c2 ) ; ( a3 ) at ( -0.6,-2 ) [ circle , thick , draw , label = left:@xmath392@xmath4 ; ( b3 ) at ( 0.8,-2 ) [ circle , thick , draw]@xmath5 ; ( a3 ) at ( 3.4,-2 ) [ circle , thick , draw , label = left:@xmath393 @xmath4 ; ( b3 ) at ( 4.8,-2 ) [ circle , thick , draw ] @xmath5 ; ( b3 ) to [ thick , loop , distance=0.5 cm ] ( b3 ) ; ( b3 ) to [ thick , loop , distance=0.5 cm ] ( b3 ) ; ( a3 ) to [ thick , bend right ] ( b3 ) ; ( b3 ) to [ thick , bend right ] ( a3 ) ; ( a4 ) at ( -0.6,-3 ) [ circle , thick , draw , label = left:@xmath394@xmath4 ; ( b4 ) at ( 0.8,-3 ) [ circle , thick , draw]@xmath5 ; ( a4 ) at ( 3.4,-3 ) [ circle , thick , draw , label = left:@xmath395 @xmath4 ; ( b4 ) at ( 4.8,-3 ) [ circle , thick , draw ] @xmath5 ; ( b4 ) to [ thick , loop , distance=0.5 cm ] ( b4 ) ; ( b4 ) to [ thick , loop , distance=0.5 cm ] ( b4 ) ; ( a4 ) to [ thick , bend right ] ( b4 ) ; ( b4 ) to [ thick , bend right ] ( a4 ) ; ( b4 ) to [ thick , bend right ] ( a4 ) ; ( a5 ) at ( -0.6,-4 ) [ circle , thick , draw , label = left:@xmath396@xmath4 ; ( b5 ) at ( 0.8,-4 ) [ circle , thick , draw]@xmath5 ; ( a5 ) at ( 3.4,-4 ) [ circle , thick , draw , label = left:@xmath397 @xmath4 ; ( b5 ) at ( 4.8,-4 ) [ circle , thick , draw ] @xmath5 ; ( b5 ) to [ thick , loop , distance=0.5 cm ] ( b5 ) ; ( b5 ) to [ thick , loop , distance=0.5 cm ] ( b5 ) ; ( a5 ) to [ thick , bend right ] ( b5 ) ; ( b5 ) to [ thick , bend right ] ( a5 ) ; ( a5 ) to [ thick , bend right ] ( b5 ) ; ( a6 ) at ( -0.6,-5 ) [ circle , thick , draw , label = left:@xmath398@xmath4 ; ( b6 ) at ( 0.8,-5 ) [ circle , thick , draw]@xmath5 ; ( a6 ) at ( 3.4,-5 ) [ circle , thick , draw , label = left:@xmath399 @xmath4 ; ( b6 ) at ( 4.8,-5 ) [ circle , thick , draw ] @xmath5 ; ( b6 ) to [ thick , loop , distance=0.5 cm ] ( b6 ) ; ( b6 ) to [ thick , loop , distance=0.5 cm ] ( b6 ) ; ( b6 ) to [ thick , bend right ] ( a6 ) ; ( a6 ) to [ thick , bend right ] ( b6 ) ; first consider the afs @xmath400 and @xmath401 , and observe that @xmath402 . on the other hand @xmath400 and @xmath401
differ in their complete extensions since @xmath403 but @xmath404 .
therefore complete semantics is unverifiable by the verification class induced by @xmath324 .
likewise , this can be shown for the classes induced by @xmath405 , @xmath406 , @xmath407 , @xmath371 , and @xmath408 , respectively : * @xmath409 @xmath410 @xmath411 @xmath412 , but @xmath413 .
* @xmath414 , but @xmath415 . * @xmath416 , but @xmath417 . *
@xmath418 , but @xmath419 . * @xmath420 , but @xmath421 .
hence the complete semantics is exactly verifiable by the verification class induced by @xmath348 .
[ ex : semi_eager_unverifiable ] consider the semi - stable and eager semantics and recall that they are @xmath422-verifiable in order to show exact verifiability it suffices to show unverifiability by the classes induced by @xmath296 , @xmath423 , and @xmath369 ( cf .
figure [ fig : verification_classes ] ) ; @xmath424 and @xmath425 are taken from example [ ex : exact_verify ] above .
* @xmath426 , but @xmath427 . * @xmath428 , but @xmath429 . *
@xmath430 , but @xmath431 and @xmath432 .
( a1 ) at ( -0.6,0 ) [ circle , thick , draw , label = left:@xmath433@xmath4 ; ( b1 ) at ( 0.8,0 ) [ circle , thick , draw]@xmath5 ; ( c1 ) at ( 2,0 ) [ circle , thick , draw]@xmath116 ; ( a1 ) at ( 3.4,0 ) [ circle , thick , draw , label = left:@xmath434 @xmath4 ; ( b1 ) at ( 4.8,0 ) [ circle , thick , draw ] @xmath5 ; ( c1 ) at ( 6,0 ) [ circle , thick , draw ] @xmath116 ; ( c1 ) to [ thick , loop , distance=0.5 cm ] ( c1 ) ; ( c1 ) to [ thick , loop , distance=0.5 cm ] ( c1 ) ; ( b1 ) to [ thick , bend right ] ( a1 ) ; ( a1 ) to [ thick , bend right ] ( b1 ) ; ( b1 ) to [ thick , bend right ] ( a1 ) ; ( a1 ) to [ thick , bend right ] ( b1 ) ; ( b1 ) to [ thick , bend right ] ( c1 ) ; hence , both the semi - stable and eager semantics are exactly verifiable by the verification class induced by @xmath371 .
[ ex : grd_sad_unverifiable ] now consider the grounded and strong admissible semantics and recall that they are @xmath435-verifiable in order to show exact verifiability we have to show unverifiability by the classes induced by @xmath302 , @xmath436 , and @xmath423 ( cf .
figure [ fig : verification_classes ] ) ; again , the afs from example [ ex : exact_verify ] can be reused .
* @xmath437 , but @xmath438 and @xmath439 . * @xmath440 , but @xmath441 and @xmath442 * @xmath428 , but @xmath443 and @xmath444 .
hence , both the grounded and strong admissible semantics are exactly verifiable by the verification class induced by @xmath371 .
[ ex : last_unverifiable ] finally consider stable , stage , admissible , preferred and ideal semantics .
they are either @xmath445-verifiable ( @xmath79 and @xmath446 ) or @xmath447-verifiable ( @xmath448 , @xmath80 , and @xmath449 ) . in order to show that these verification classes are exact we have to show unverifiability w.r.t . the verification class induced by @xmath367 .
consider , for instance , the afs @xmath450 and @xmath451 from example [ ex : exact_verify ] .
we have @xmath452 , but @xmath453 , @xmath454 , and @xmath455 for @xmath456 , showing exactness of the respective verification classes . the insights obtained through examples [ ex : exact_verify ] , [ ex : semi_eager_unverifiable ] , [ ex : grd_sad_unverifiable ] , and [ ex : last_unverifiable ] show that the verification classes obtained from the criteria given above are indeed exact .
figure [ fig : classes_of_semantics ] shows the relation between the semantics under consideration with respect to their exact verification classes .
node[snode ] ( na ) @xmath457 : @xmath458 + + ( -3,0.8 ) node[snode ] ( stb ) @xmath445 : @xmath79 , @xmath459 + + ( 3,0 ) node[snode ] ( pref ) @xmath447 : @xmath448 , @xmath80 , @xmath460 + + ( -1.5,0.8 ) node[snode ] ( semi ) @xmath461 : @xmath78 , @xmath462 + + ( 4.5,0 ) node[snode ] ( grd ) @xmath463 : @xmath464 , @xmath159 + + ( -3,0.8 ) node[snode ] ( co ) @xmath465 : @xmath466 ; ( na ) edge ( stb ) ( na ) edge ( pref ) ( na ) edge ( grd ) ( pref ) edge ( semi ) ( stb ) edge ( semi ) ( semi ) edge ( co ) ( grd ) edge ( co ) ; we turn now to the main theorem stating that any _ rational _ semantics ( recall that all semantics we consider in this paper are rational ) is exactly verifiable by one of the @xmath10 different verification classes .
every semantics which is rational is exactly verifiable by a verification class induced by one of the neighborhood functions presented in figure [ fig : verification_classes ] .
first of all note that by lemma [ lemma : verification_classes ] , @xmath467 is the least informative neighborhood function and for every other neighborhood function @xmath290 it holds that @xmath468 .
therefore , if a semantics is verifiable by the verification class induced by any @xmath290 then it is exactly verifiable by a verification class induced by some @xmath309 with @xmath469 .
moreover , if a semantics is exactly verifiable by a class , then it is by definition also verifiable by this class .
hence it remains to show that every semantics which is rational is verifiable by a verification class presented in figure [ fig : verification_classes ] .
we show the contrapositive , i.e. , if a semantics is not verifiable by a verification class induced by one of the neighborhood functions presented in figure [ fig : verification_classes ] then it is not rational .
assume a semantics @xmath1 is not verifiable by one of the verification classes .
this means @xmath1 is not verifiable by the verification class induced by @xmath348 .
hence there exist two afs @xmath81 and @xmath108 such that @xmath470 and @xmath361 , but @xmath471 . for every argument @xmath4 which is not self - attacking , a tuple @xmath472
is contained in @xmath473 ( and in @xmath474 ) .
hence @xmath81 and @xmath108 have the same not - self - attacking arguments and , moreover these arguments have the same ingoing and outgoing attacks in @xmath81 and @xmath108 .
this , together with @xmath475 implies that @xmath476 ( see definition [ def : semantics_conditions ] ) holds .
but since @xmath471 we get that @xmath1 is not rational , which was to show .
note that the criterion giving evidence for verifiability of a semantics by a certain class has access to the set of arguments of a given framework .
in fact , only the criterion for stable semantics makes use of that .
indeed , stable semantics needs this information since it is not verifiable by any class when using a weaker notion of verifiability , which rules out the usage of @xmath36 .
a type of semantics which has aroused quite some interest in the literature ( see e.g. @xcite and @xcite ) are intermediate semantics , i.e. semantics which yield results lying between two existing semantics .
the introduction of @xmath1-@xmath70-intermediate semantics can be motivated by deleting _ undesired _ ( or add _ desired _ ) @xmath70-extensions while guaranteeing all reasonable positions w.r.t . @xmath1 .
in other words , @xmath1-@xmath70-intermediate semantics can be seen as sceptical or credulous acceptance shifts within the range of @xmath1 and @xmath70 .
a natural question is whether we can make any statements about compatible kernels of intermediate semantics . in particular ,
if semantics @xmath1 and @xmath70 are compatible with some kernel @xmath102 , is then every @xmath1-@xmath70-intermediate semantics @xmath102-compatible .
the following example answers this question negatively .
[ ex : stagle ] recall from theorem [ the : strong ] that both stable and stage semantics are compatible with @xmath477 , i.e. @xmath478 .
now we define the following @xmath79-@xmath446-intermediate semantics , say _ stagle _ semantics : given an af @xmath479 , @xmath480 iff @xmath481 , @xmath482 and for every @xmath483 we have @xmath484 .
obviously , it holds that @xmath485 and @xmath486 as well as @xmath487 , as witnessed by the following af @xmath81 : \(a ) at ( 1.0,0.0 ) [ circle , thick , draw , label = left:@xmath7@xmath4 ; ( b ) at ( 2.5,0.0 ) [ circle , thick , draw]@xmath5 ; ( c ) at ( 4.0,0.0 ) [ circle , thick , draw]@xmath116 ; \(a ) to [ thick , loop , distance=0.5 cm ] ( a ) ; ( b ) to [ thick , bend right ] ( c ) ; ( c ) to [ thick , bend right ] ( b ) ; ( a ) to [ thick , bend right ] ( b ) ; it is easy to verify that @xmath488 .
we proceed by showing that stagle semantics is not compatible with @xmath477 . to this end consider @xmath489 , which is depicted below .
\(a ) at ( 1.0,0.0 ) [ circle , thick , draw , label = left:@xmath490@xmath4 ; ( b ) at ( 2.5,0.0 ) [ circle , thick , draw]@xmath5 ; ( c ) at ( 4.0,0.0 ) [ circle , thick , draw]@xmath116 ; \(a ) to [ thick , loop , distance=0.5 cm ] ( a ) ; ( b ) to [ thick , bend right ] ( c ) ; ( c ) to [ thick , bend right ] ( b ) ; now , @xmath491 witnesses @xmath492 and therefore , @xmath493 .
since @xmath494 we are done , i.e. stagle semantics is indeed not compatible with the stable kernel .
it is the main result of this section that compatibility of intermediate semantics w.r.t .
a certain kernel can be guaranteed if verifiability w.r.t .
a certain class is presumed .
the provided characterization theorems generalize former results presented in @xcite .
moreover , due to the abstract character of the theorems the results are applicable to semantics which may be defined in the future . before turning to the characterization theorems we state some implications of verifiability .
in particular , under the assumption that @xmath1 is verifiable by a certain class , equality of certain kernels implies expansion equivalence w.r.t .
@xmath1 .
[ pro : cf - range - verifiable ] for any @xmath445-verifiable semantics @xmath1 we have @xmath495 in @xcite it was shown that ( i ) .
consider now a @xmath445-verifiable semantics @xmath1 . in order to show @xmath496 ( ii ) we prove @xmath497 ( * ) first .
it is easy to see that @xmath498 iff @xmath499 .
furthermore , since @xmath477 deletes an attack @xmath263 only if @xmath4 is self - defeating we deduce that ranges does not change as long as conflict - free sets are considered .
thus , @xmath500 .
now assume that @xmath501 and let @xmath502 for some af @xmath128 .
we have to show that @xmath503 . applying ( ii )
we obtain @xmath504 .
furthermore , using ( i ) we deduce @xmath505 .
finally , @xmath506 by applying ( ii ) , which concludes the proof .
the following results can be shown in a similar manner .
[ pro : cf - range-?-verifiable ] for any @xmath461-verifiable semantics @xmath1 we have @xmath507 [ pro : cf - range - inrange - verifiable ] for any @xmath465-verifiable semantics @xmath1 we have @xmath508 [ pro : cf - inrange - pm - verifiable ] for any @xmath463-verifiable semantics @xmath1 we have @xmath509 [ pro : cf - verifiable ] for any @xmath457-verifiable semantics @xmath1 we have @xmath510 we proceed with general characterization theorems .
the first one states that @xmath79-@xmath446-intermediate semantics are compatible with stable kernel if @xmath445-verifiability is given .
consequently , stagle semantics as defined in example [ ex : stagle ] can not be @xmath445-verifiable .
[ the : stablefirst ] given a semantics @xmath1 which is @xmath445-verifiable and @xmath79-@xmath459-intermediate , it holds that @xmath511 ( @xmath141 ) follows directly from proposition [ pro : cf - range - verifiable ] .
( @xmath125 ) we show the contrapositive , i.e. @xmath512 . assuming @xmath513 implies @xmath514 , i.e. there exists an af @xmath128 such that @xmath515 and therefore , @xmath516 . let @xmath517 and @xmath518 .
it is easy to see that @xmath519 and @xmath520 . since now both @xmath521 and @xmath522 it holds that @xmath523 and @xmath524 .
hence @xmath525 , showing that @xmath526 .
the following theorems can be shown in a similar manner . [
the : admfirst ] given a semantics @xmath1 which is @xmath461-verifiable and @xmath76-@xmath448-intermediate with @xmath527 , it holds that @xmath528 remember that complete semantics is a @xmath78-@xmath448-intermediate semantics .
furthermore , it is not characterizable by the admissible kernel as already observed in @xcite .
consequently , complete semantics is not @xmath461-verifiable ( as we have shown in example [ ex : exact_verify ] with considerable effort ) .
[ the : grdfirst ] given a semantics @xmath1 which is @xmath463-verifiable and @xmath464-@xmath159-intermediate , it holds that @xmath529
in this work we have contributed to the analysis and comparison of abstract argumentation semantics .
the main idea of our approach is to provide a novel categorization in terms of the amount of information required for testing whether a set of arguments is an extension of a certain semantics .
the resulting notion of verifiability classes allows us to categorize any new semantics ( given it is `` rational '' ) with respect to the information needed and compare it to other semantics .
thus our work is in the tradition of the principle - based evaluation due to @xcite and paves the way for a more general view on argumentation semantics , their common features , and their inherent differences . using our notion of verifiability
, we were able to show kernel - compatibility for certain intermediate semantics .
concerning concrete semantics , our results yield the following observation : while preferred , semi - stable , ideal and eager semantics coincide w.r.t.strong equivalence , verifiability of these semantics differs .
in fact , preferred and ideal semantics manage to be verifiable with strictly less information . for future work
we envisage an extension of the notion of verifiability classes in order to categorize semantics not captured by the approach followed in this paper , such as @xmath84 @xcite .
arieli , o. 2012 .
conflict - tolerant semantics for argumentation frameworks . in _ logics in artificial intelligence - 13th european conference , proceedings _ , volume 7519 of _ lecture notes in computer science _
springer .
baroni , p. , and giacomin , m. 2007a . comparing argumentation semantics with respect to skepticism . in _ symbolic and quantitative approaches to reasoning with uncertainty , 9th european conference , proceedings _ ,
volume 4724 of _ lecture notes in computer science _ , 210221 .
springer .
baumann , r. , and brewka , g. 2013 . analyzing the equivalence zoo in abstract argumentation . in _
14th international workshop on computational logic in multi - agent systems , proceedings _ ,
volume 8143 of _ lecture notes in computer science _ , 1833 .
springer .
baumann , r. , and spanring , c. 2015 .
infinite argumentation frameworks on the existence and uniqueness of extensions . in _ advances in knowledge representation , logic programming , and abstract argumentation - essays dedicated to g. brewka on the occ . of his 60th birthday _ , volume 9060 of _ lecture notes in computer science _ , 281295 .
springer . baumann , r. 2016 .
characterizing equivalence notions for labelling - based semantics . in _ principles of knowledge representation and reasoning : proceedings of the 15th international conference_. to appear .
caminada , m. 2014 .
strong admissibility revisited . in _ computational models of argument - proceedings of comma 2014 _ , volume 266 of _ frontiers in artificial intelligence and applications _ , 197208 . press .
dvok , w. ; linsbichler , t. ; oikarinen , e. ; and woltran , s. 2014 .
resolution - based grounded semantics revisited . in _ computational models of argument - proceedings of comma 2014 _ , volume 266 of _ frontiers in artificial intelligence and applications _ , 269280 . press .
turner , h. 2004 .
strong equivalence for causal theories . in _
7th international conference on logic programming and nonmonotonic reasoning , proceedings _ ,
volume 2923 of _ lecture notes in computer science _
, 289301 . springer . | _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ dung s abstract argumentation theory is a widely used formalism to model conflicting information and to draw conclusions in such situations .
hereby , the knowledge is represented by so - called argumentation frameworks ( afs ) and the reasoning is done via semantics extracting acceptable sets .
all reasonable semantics are based on the notion of conflict - freeness which means that arguments are only jointly acceptable when they are not linked within the af . in this paper , we study the question which information on top of conflict - free sets is needed to compute extensions of a semantics at hand .
we introduce a hierarchy of so - called verification classes specifying the required amount of information .
we show that well - known standard semantics are exactly verifiable through a certain such class .
our framework also gives a means to study semantics lying inbetween known semantics , thus contributing to a more abstract understanding of the different features argumentation semantics offer .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ |
pb(mg@xmath0nb@xmath1)o@xmath2 ( pmn ) is one of the earliest members of the growing family of lead ferroelectric relaxors , which , due to its application potential , has been attracting attention of researchers for many years .
however , despite significant efforts , many properties of lead compounds are still unclear and the very nature of the relaxor behavior has yet to be understood .
the difficulties stem from the high complexity of these materials with a high degree of compositional , chemical and structural disorder .
pb(mg@xmath0nb@xmath1)o@xmath2 has a perovskite structure with , on average , a cubic pm3 m symmetry , with mg@xmath3 and nb@xmath4 ions interchanging on b sites .
measurements of the dielectric constant show that the crystal does not undergo a sharp transition to a ferroelectric phase .
instead , the dielectric constant exhibits a broad maximum at t@xmath5 k@xcite .
below t@xmath6 k , it deviates from curie - weiss law@xcite and , at t@xmath7350 k , exhibits strong dispersion@xcite .
the structure of pmn is not homogeneous .
if mg and nb occupation ratio were uniformly and ideally equal 1:2 throughout the crystal , it would have a rhombohedral symmetry r3 m .
but , a network of superstructure clusters with face centered cubic symmetry ( fm3 m ) , destroys the picture .
the size of these clusters is of the order of 2 - 3 nm ; the distance between the centers of the neighboring clusters is near 2.5 nmboulesteix , harmer .
the most widely accepted hypothesis , so - called space - charge model , associates these fm3 m clusters with antiferroelectrically ordered@xcite areas where mg : nb occupation ratio is equal to 1:1 , i.e. with local composition pb(mg@xmath8nb@xmath9)o@xmath2@xcite .
such 1:1 clusters carry an excessive negative charge , which is compensated by the positively charged host matrix .
network of these charges acts as a source of random fields and may prevent the occurrence of the normal phase transition .
alternatively , the fm3 m clusters might be explained by the local order in the form of pb[b@xmath10b@xmath11@xmath9[b@xmath4]@xmath9o@xmath2 @xcite .
this hypothesis avoids assumption of the considerable space charges and calls for other reasons to explain relaxor behavior , including the random fields due to the disorder - induced local anisotropy@xcite , effects related to different ionic radii@xcite or both@xcite .
the presence of areas with special order causes crystalline structure to change in an intricate way . at the high temperature limit
, the crystal has a uniform primitive cubic ( pm3 m ) structure .
lowering the temperature leads to distinguishability between mg@xmath3 and nb@xmath4 sites and to formation in the ordered regions of a face centered ( fm3 m ) superstructure .
the temperature of the fm3 m clusters formation has yet to be determined .
husson s data suggests they presence at t@xmath13850 k @xcite . our data ( see below )
raise the upper limit to a temperature higher than 1000 k. lllllll research & 45 @xmath14 & 130 @xmath14 & 260 @xmath14 & 420 @xmath14 & 500 - 600 & 780 @xmath14 + group & & & & & @xmath14 & + husson@xcite & & o - b - o & mg - o - mg & nb - o ... nb & nb - o ... mg + ( 1990 ) & & & bending & stretching & stretching & stretching + dimza@xcite & & & & & & coexistence of tetragonal + ( 1992 ) & & & & & & and rhombohedral phases + krainik@xcite ( 1993 ) & to@xmath15 & & & & & + idink@xcite & + ( 1994 ) & + marssi@xcite & to@xmath15 & to@xmath16 & to@xmath2 & to@xmath17 & & to@xmath16+to@xmath17 or lo@xmath17 ( 1996 ) + ( 1996 - 98 ) & & & & & & nb - o - mg or lo@xmath17 ( 1998 ) + jiang@xcite ( 1999 ) & & + + & vv - e@xmath18 ? ? & & & & & + lushnikov@xcite & vh-2t@xmath19 & & & & & a@xmath20 + siny@xcite & or & & breathing mode + ( 1999 ) & vv and one vh & & of oxigen + & have fractal & & & & & + & character & & & & & + & another vh - t@xmath21 & & & & & + dujovne@xcite ( 2002 ) & ta+disorder & & & & & + cooling the sample leads to the nucleation of polar nanoregions ( pnr s ) characterized by the local distortions .
these distortions are responsible for the deviation of the dielectric constant from curie - weiss law at t@xmath22@xcite .
their presence also causes a non - linearity in the temperature dependence of the optical refraction index@xcite .
size of the pnr s is equal to 30 - 80 depending on temperature@xcite .
the appearance of polar distortions is a consequence of the ion off - centering that is a common property of many perovskites . by structural similarity
, one could expect that the mechanism of the polar regions formation in pmn is analogous to the one in kta@xmath23nb@xmath24o@xmath2 ( ktn)@xcite .
however , unlike ktn where nb@xmath4 is the only off - centered positive ion , pmn represents a more difficult case . here , not only _ both ions of b - type _ ,
mg@xmath3 and nb@xmath4 , are shifted by @xmath130.1 - 0.2 , but also pb@xmath3 , _ ion of a - type _ , is off - centered by @xmath130.25 - 0.35 @xcite .
the b - ions tend to shift along a ( 111 ) direction , thus having eight equivalent positions .
the shifts of the pb@xmath3-ions exhibit a spherically symmetric distributionvakhrushev .
it is often assumed@xcite that nb@xmath4 , due to its position in the cell and small radius , acts as the main ferroelectric agent . at high temperature ,
the off - centered ions are free to reorient among all allowed off - centered positions . cooling down and the growing role of the electric interactions causes the appearance of correlated regions consisting of several neighboring cells .
each such region is characterized by a giant electric dipole moment and a local distortion from the cubic symmetry .
the pnr s are capable of reorientational motion as whole units . in ktn ,
the development of these regions leads , at some critical temperature , to a phase transition .
however , in pmn , the polar regions develop under constraints imposed by the local anisotropy . as a result , no ferroelectric phase transition , but a glasslike freezing of the polar regions is observed@xcite . a bias electric field of magnitude @xmath25 kv / cm , applied in the ( 111 ) direction , is sufficient to compensate for the influence of the frustrating effects and to induce , at t@xmath26 k , a transition to the rhombohedral r3 m ferroelectric phase@xcite .
raman spectra of pmn have been measured and reported by several research groups@xcite .
but , the high complexity of the processes caused difficulties in their interpretation , especially due to possible coexistence of clusters with pm3 m , fm3 m and r3 m symmetry .
it has been shown that the major spectral lines are of first - order character@xcite . in cubic crystals first - order raman scattering
is prohibited by symmetry .
explanation of the appearance of first - order scattering _ in principle _ is the one side of the problem .
another side is the assignment of _ particular _ lines .
the question of the soft ferroelectric mode ( to@xmath15 ) and the strong raman line at @xmath1345 @xmath14 , which is close to where the to@xmath15 is expected to be found , has been like a stumbling block for investigators .
interpretations of the pmn raman spectrum offered by various research groups are summarized in table [ ramanconc ] .
this table reflects the variety of mutually excluding opinions ranging from those that explain the light scattering by _ disorder in general _ ( groups of hussonhusson , krainik@xcite , marssi@xcite , dujovne@xcite ) to those that connect it to the presence of some _ particular type of disorder _ , the one that can lead to the appearance of clusters with symmetry allowing first - order scattering , like face centered cubic ( jiang@xcite , lushnikov@xcite , sinysinymrs , sinyjrs , sinyf - a , sinyf - b , sinyf - rev ) , rhombohedral ( idinkidink ) or tetragonal ( dimza@xcite ) .
the assignement of particular lines is even more contradictory .
recently made first - principle calculations @xcite can confirm only few of them .
we believe , the arguments might be resolved by taking into account the possibility of coexistence of different mechanisms , each playing a role in the scattering processes .
recently published low branches of phonon dispersion curvesgehwak , wakstock , gehvak , nabervak , measured by neutron scattering , must be helpful for the phonon assignment of some of the raman lines . according to these data
, the strong 45 @xmath14 line has an energy close to the energy of the zone center ( zc ) to@xmath15 or the zone boundary ( zb ) ta phonon .
lowering the temperature , to@xmath15 zone center phonon exhibits softening , at t@xmath27t@xmath22 k , due to the drastic increase of damping , it disappears .
the overdamping continues down to t@xmath28 k@xcite .
it also causes significant , almost sixfold , broadening of the ta phonon branch@xcite .
it seems reasonable to expect that the raman line at 45 @xmath29 is also affected by the described phenomenon .
however , none of the known raman reports show any significant effects that could be attributed to this overdamping .
the formation of the lower - symmetry clusters in the host crystal is accompanied by relaxational and reorientational dynamics .
this dynamics should be reflected in the light scattering spectrum either directly or through interaction with the phonon modes .
theoretical considerations@xcite show that internal relaxational motion leads to the appearance of the central peak ( cp ) in the raman spectrum .
interactions of this motion with phonons can cause softening of them@xcite .
these predictions have been tested on several types of crystals with internal degrees of freedom , like alkali - halide - cyanide ( k(cn)@xmath24-kbr@xmath23 , k(cn)@xmath24-kcl@xmath23)rowea , roweb , lutya , lutyb compounds .
recently , we have analyzed@xcite the reorientational motion in ferroelectric kta@xmath23nb@xmath24o@xmath30 crystals . in this paper
we discuss the role of the relaxational and reorientational dynamics in the scattering processes from pmn crystal . in the paper
presented we offer a complex analysis of temperature dependencies of polarized ( vv ) and depolarized ( vh ) raman spectra from a single crystalline pmn sample .
this analysis is based on multiple peak decomposition .
an attempt to apply similar kind of analysis has earlier been made by siny et al.@xcite .
but , it was restricted to the low frequency vv spectra in a limited temperature range ( with maximum temperature @xmath13620 k ) , concentrating on the central peak ( cp ) only . to obtain the complete picture
, we consider the temperature evolution of the whole raman scattering spectrum ( up to 1000 @xmath14 ) in the 100 - 1000 k temperature range .
we interpret our results in connection with recently published neutron scattering data and with the concept of interaction of phonons with internal relaxational motion . as a result of this work
, we present our view on the picture of structural transformations in pmn crystal and on the origin of its raman spectrum .
we have investigated the raman scattering from @xmath31 cut pmn single crystal .
the crystal was grown by the high temperature solution technique using 30 wt% pbo as flux .
the growth conditions were optimized based on the pseudo - binary phase diagram established for pmn and pbo@xcite . the as - grown crystal exhibits a pseudo - cubic morphology with @xmath32 cub growth steps .
a @xmath33 cub - oriented crystal of a volume of 53 mm@xmath34 was cut from a large as - grown crystal and polished with fine diamond paste ( down to 0.25 mm ) .
it showed very high optical quality and satisfied the requirements for light scattering studies .
the scattering was excited by propagating in @xmath31 direction 514.5 nm light from a 200 mw ar@xmath35-ion laser , focused to a 0.1 mm spot .
the scattered light was collected at an angle of 90@xmath36 with respect to the incident beam ( i.e. , in @xmath37 direction ) by a double - grating isa jobin yvon spectrometer equipped with a hamamatsu photomultiplier r-649 . for most of the measurements ,
the slits were opened to 1.7 @xmath14 . however , in order to acquire more precise data in the central peak region , at the temperatures close to the maximum of the dielectric constant ( @xmath38 k ) , the slits were narrowed to 0.5 @xmath14 .
each polarization of the scattered light , @xmath39 ( vv ) and @xmath40 ( vh ) , was measured separately . in order to exclude differences in sensitivity of the monochromator to different polarizations of the light ,
a circular polarizer was used in front of the entrance slit . for control purposes
, we also took measurements without polarization analysis .
finally , to protect the photomultiplier from the strong raleigh scattering , the spectral region from -4 to + 4 @xmath14 was excluded from the scans .
the data were collected in the temperature range from 1000 to 100 k. the cooling rate was 0.5 - 1 k / min . every 50 - 20 k the temperature was stabilized and the spectrum recorded .
we agree with ref.@xcite that the reproducibility of the spectra in the temperature range approximately between 200 and 350 k was not very good . in this range ,
the spectra exhibited a strong dependence on the cooling rate and other experimental conditions .
fig.[cascade ] presents examples of light scattering spectra measured at different temperatures without polarization analysis . to facilitate comparison
, we corrected the spectra by the bose population factor : @xmath41 where @xmath42 figure [ vv - vh ] demonstrates scattering spectra measured in vv ( a ) and vh ( b ) geometries at different temperatures .
these spectra we present in uncorrected , `` as - measured '' , form . as can be seen from figs .
[ cascade ] and [ vv - vh ] , our spectra are consistent with those from refs.@xcite and @xcite . in the high temperature region ,
a typical spectrum consists of two strong lines centered approximately at 45 @xmath14 and 780 @xmath14 ( labelled a and b ) and of three broad bands ( c , d , e ) .
line a exhibits fine structure , which is explicitly seen on the spectra measured with polarization analysis ( see figs .
[ vv - vh ] and [ fitexamples ] ) .
lowering temperature leads to the splitting of the broad bands c , d and e into a number of narrower lines ( labelled by indices ) and to appearance of new line f. as seen from fig .
[ cascade ] , starting from 1000 k , the reduced intensity of the scattering , first grows until the temperature of @xmath43 k then sharply decreases to a minimum located close to the burns temperature t@xmath44 k , after which it increases again . at @xmath45
k its strength gets restored .
below this temperature , the intensities of the major lines ( especially a and b ) are determined primarely by the bose factor .
i.e. being corrected by it , they do not show significant temperature changes , demonstrating characteristic first - order behavior .
the shape similarity of the low- and high- temperature spectra shows that even at 1000 k the scattering has first - order character . in the perfecly cubic crystal ,
first - order scattering is prohibited .
therefore , one should assume the presence of distortions in the form of lower symmetry clusters , like fm3 m or r3 m .
at high temperature , these clusters might be present in the dynamic form , having short lifetime .
lowering the temperature , their lifetime increases and they become progressively more static .
the observed around t@xmath22 decrease of raman intensity is also clearly seen in fig .
[ vv - vh ] .
however , there are no reports on this effect in the literature ; it might have been overlooked .
this intensity drop indicates worsening of the optical quality of the crystal . in our opinion
, it is consistent with the formation of fm3 m clusters in their evolution from dynamic to static form .
the splitting of peaks is associated with the development of r3 m clusters .
the first signs of the splitting appear at t@xmath22 .
strarting from t@xmath46 k ( which is also close to where the dispersion of the dielectric constant appears ) , this effect becomes even more definite and the intensities of the components begin to grow .
below t@xmath28 k , which is the temperature of electric field - induced ferroelectric phase transition@xcite , the shape of the spectra is relatively stable .
a comprehensive analysis of our data and their comparison with results of other experiments allows to understand the meaning of the mentioned above temperatures t@xmath44 k , t@xmath47 k and t@xmath48 k in the temperature evolution of the raman spectra and the lattice structure .
to be able to analyze the data , we decomposed the measured spectra using multiple peak fitting procedure .
it was found that satisfactory fit could be achieved with the assumption that the central peak has a lorentzian shape and that each of the other peaks is described by the spectral response function , i.e. damped harmonic oscillator , modified by the population factor ( [ boze ] ) : @xmath49 where @xmath50 and @xmath51 are the damping constant and the mode frequency .
as all of the peaks are much better resolved at low temperatures , we started our fitting procedure at the low - temperature end of the data set ( at 110 k ) and , then , observed the evolution of the peaks with increasing temperature ( i.e. , in order , opposite to the order of measuring ) . at the same time , we tried to minimize the number of peaks necessary to achieve a reasonably good fit .
the control data set ( measured without polarization analysis ) has been used to calibrate the positions and widths of the weak and poorely resolved peaks from the vv and vh data sets . since a large number of parameters
is involved , results of a particular fit may depend on their initial values .
to stabilize the results , the best - fit values of parameters obtained at the previous temperature were used as initial values for the next one . in this manner ,
several sets of fits were obtained and analyzed .
it is remarkable , that in all of them the major parameters showed the same trends of behavior , confirming the importance of the mentioned above temperatures : t@xmath52 k , t@xmath53 k and t@xmath54 k. these trends are summarized in table [ peakproperties ] .
one of the sets have been selected to report the most interesting results in detail .
examples of the fits at the temperatures of 1000 and 110 k in vv and vh geometries are shown in fig .
[ fitexamples ] . for clarity
, we describe the observed phenomena from high to low temperatures , following the same order as in measurements ( unless the opposite stated ) .
[ cols="^,^ , < , < , < , < " , ]
the temperature dependences of the fitting parameters for the central peak ( cp ) are presented in fig .
[ central ] .
circles correspond to the vv and triangles to the vh component of the peak .
the existence of the cp is a direct consequence of the lattice relaxations , which are very sensitive to the restrictions imposed by the low - symmetry clusters .
if relaxations are fast@xcite , the cp is low - intense and broad , whereas their slowing causes growth and narrowing of the peak .
we would like to point out a striking similarity of the temperature behavior of the cp in pbmg@xmath0nb@xmath1o@xmath2 ( fig .
[ central ] ) and in kta@xmath55nb@xmath56o@xmath2 ( fig.3 in ref .
@xcite ) crystals .
we have shown@xcite that the temperature behavior of the cp in ktn can be explained by the model involving the relaxational motion of off - centered nb ions and its progressive restriction with temperature decrease . in the cubic phase ,
nb ions are allowed to reorient amongst eight equivalent @xmath57111@xmath58 directions .
the appearance of the pnr s followed by a sequence of phase transitions down to a rhombohedral r3 m phase , limits the ion motion to four , two and , finally , locks it in only one site .
this model is in agreement with the diffuse neutron scattering studies of similar system@xcite .
the similarity of the cp behavior in these two very different materials , suggests that the temperature evolution of the polar clusters in pbmg@xmath0nb@xmath1o@xmath2 passes through the similar stages as in kta@xmath55nb@xmath56o@xmath2 ( ktn ) , however , it is not accompanied by appearance of the long - range order . starting from the high - temperature end ( fig . [ central ] ) , the first important feature is the strong and narrow scattering in the vv geometry accompanied by a relatively weak scattering in the vh geometry .
this indicates the presence of a symmetric slow relaxational motion , involving 180@xmath59 reorientations of ions .
lowering the temperature , starting from @xmath60 k , a cessation of this motion causes an intensity decrease of the cp .
slow at first , this decrease becomes sharper and reaches minimum near t@xmath44 k. the prohibition of 180@xmath61 reorientations introduces anisotropy into the lattice causing the distinguishability between mg and nb occupied cites and to formation in the 1:1 ordered areas of the superstructure with , on average , fm3 m symmetry .
it also imposes first restrictions on the reorientational motion of the dynamical r3 m polar nanoregions .
now , they can reorient only amongst four neighboring @xmath57111@xmath58 directions , forming , on average , tetragonal - like distortions .
these processes are accompanied by the appearance of large ( of the order of wavelength of light ) dynamic fluctuations causing worsening of the optical quality of the sample . with further decrease of the temperature , the optical quality of the crystal
improves . from @xmath45
k , the four - cite reorientational motion of the pnr s starts to slow down , which is marked by the narrowing of the vv component of the cp and increase of its vv and vh intensities .
the simultaneous broadening of the vh component up to the maximum at @xmath62 k , indicates rearrangements in the crystalline structure leading to the appearance of the new type of restrictions on the ion motion .
analogy with ktn suggests that below @xmath62 k the motion of r3 m clusters becomes limited to two neigboring @xmath57111@xmath58 orientations , averaging in monoclinic - like distortions .
such a rearrangement causes some decrease in intensity of the vv component ( with minimum at @xmath63 k ) , while the vh intensity keeps growing . below @xmath63
k , the slowing down of the two - site relaxational motion causes narrowing the cp and increase of intensities of both components . from t@xmath47 k ,
these effects become especially dramatic .
further temperature decrease , starting from @xmath64 k , leads to the complete prohibition of intersite reorientational motion of r3 m clusters , i.e. to appearance of static r3 m clusters .
this is marked by a sharp decrease in intensity of both components of the cp .
at the temperature t@xmath48 k , the process of conversion of pnr s from dynamic to static form ( freezing ) is primarily finished .
below this temperature , the central peak is narrow and its intensity is small in both scattering geometries .
we should mention that the result of our analysis of the vv component of the cp exhibits similar tendencies to those in ref.@xcite , except at the low temperature end .
the discrepancy occurs due to the difference in experimental technique . in the central peak region
, we did the measurements with much higher resolution ( 0.5 @xmath14 , as compared to 2 @xmath14 from ref.@xcite ) .
consequently , we were able to provide a more correct separation of the slow relaxations from the elastic scattering , which was especially important at low temperatures . ) .
phonon frequencies , measured by neutron spectroscopy@xcite , are shown for comparizon . ]
figure [ line45 ] presents the temperature evolution of the fitting parameters of the peak a , showing its position ( a ) , reduced intensity ( b ) and damping constant ( c ) .
this peak has a triplet structure , containing one component in vv ( circles ) and two components in vh ( up and down triangles ) geometries .
the fitting parameters for this peak exhibit changes at mentioned above temperatures t@xmath22 , t@xmath65 and t@xmath66 , confirming their importance for the structural evolution of the crystal .
however , the origin of this peak ( see table . [ ramanconc ] ) requires qlarification . from a comparison with the frequencies of the phonon modes determined from neutron scattering ,
it is clear that this line can not be due to the zone center soft to@xmath15 mode ( black stars in fig .
[ line45 ] ) . on the other hand , the lower frequency vh component , and possibly the vv component
could be due to the disorder - induced zone boundary scattering on ta phonon ( white stars in fig .
[ line45 ] ) .
however , the higher frequency vh component would still not be accounted for . in an attempt to account for both vh components simultaneously ,
we have tried to make use of a coupled oscillator model . in the approximation of linear coupling between two harmonic oscillatorscurrat
this approach gains : @xmath67\left ( \frac{cy - bz}{b^{2}+c^{2}}\right ) , \ ] ] where @xmath68 @xmath69,$ ] @xmath70 @xmath71 in these equations @xmath72 and @xmath73 are the resonant frequencies , @xmath74 and @xmath75 are the damping constants , @xmath76 and @xmath77 are the structure factors for two oscillators , @xmath78 and @xmath79 are the real and imaginary parts of the coupling constant . )
using the model of coupled harmonic oscillators .
from top to bottom : resonant frequencies ( a ) , damping constants ( b ) and coupling constant ( c ) .
phonon frequencies , measured by neutron spectroscopy@xcite , are shown for comparizon . ]
it was shown@xcite , that a unitary transformation can be found , which would set either real or imaginary part of the coupling to be equal to zero .
we have chosen to use imaginary coupling and set @xmath80 .
we also set the structure factors @xmath76 and @xmath77 to be the same at all temperatures .
the rest of the parameters were treated as being temperature dependent . in the result of fitting
, we have found that the values of structure factors can be approximated as @xmath81 @xmath14 and @xmath82 @xmath14 .
the other parameters as a function of temperature are shown in fig.[coupled ] : the best - fit frequencies ( a ) , the damping constants ( b ) and the coupling coefficient ( c ) .
the examples of the fits ( inserts in fig.[fitexamples ] ) demonstrate that this model gives a good approximation to the shape of the peak a. however , a comparizon of the best - fit frequencies ( fig .
[ coupled]a ) with the neutron scattering data@xcite rules out the possibility to explain the higher - frequency vh component by the interaction between zc to@xmath15 ( black stars ) and zb ta ( white stars ) modes .
the lower frequency oscillator can still be associated with disorder - induced scattering on zb ta mode .
the possible explanation of the presence of two vh components in the peak a can be in the correspondence to ta modes with different polarizations in different crystalline planes of a distorted lattice .
( a similar effect we have seen in monodomain kta@xmath23li@xmath24o@xmath2 , where one or the other ta polarization was observed , depending on the orientation of the electric field . ) these differently polarized modes are influenced by the polar nanoregions , which explains the increase of damping ( fig .
[ coupled]b ) and coupling ( fig .
[ coupled]c ) coefficients with lowering temperature .
it is also necessary to point out an observation concerning the temperature dependences of the frequencies of the components of the triplet a in approximation of independent oscillators ( fig .
[ line45]a ) .
these dependences seem to show anomalies at intersections with zc to@xmath15 modes ( black stars ) .
for example , the vv component exhibits deeps at @xmath43 and 350 k and vh - i component has a knee at @xmath63 k. among the changes occuring to the raman spectrum with cooling the sample , the splitting of the phonon lines desrves special attention , since such a splitting is a direct indicator of the modifications of the local structure . as we already mentioned , the splitting occurs to the lines c , d , and e , which is especially well seen in the vv geometry ( fig .
[ vv - vh]a ) . at high temperatures ,
their widths are very big , so the lines almost merge in a single broad shoulder .
lowering the temperature , they become more discernible and the fine structure gradually develops .
the first signs of it appear at t@xmath44 k. below t@xmath83 k , the fine structure becomes more obvious .
these observations support the expressed above idea that the scattering originates from the distorted regions .
highly dynamic and disordered at high temperature , they become progressively more static with cooling down , and their motion becomes more correlated .
this process is reflected by the temperature evolution of the lines c , d , and e. below the temperature of t@xmath83 k , the splitting value of the line e exhibits especially interesting trend . as seen from fig .
[ split ] , with temperature decrease below t@xmath84 the distance between the e - line components increases .
moreover , this increase exhibits the characteristic order parameter behavior , following the ( t@xmath85t)@xmath86 dependence , which is demonstrated in the insert to fig .
[ split ] .
it also exhibits similar temperature tendency to that described in ref .
detailed understanding of this phenomenon requires further work , and it is a subject of separate paper . 780 @xmath14 . ) ] finally , we would like to stop on the line b located at the upper end of the spectrum .
[ line780 ] demonstrates the fitting paramenters for this line : position ( a ) , reduced intensity ( b ) , and damping coefficient ( c ) .
this line is very strong in the vv geometry but it is rather weak in the vh one ( i.e. characterized by a@xmath20 symmetry ) .
it seems to be the least influenced by the ordering processes in the crystal .
however , the temperature evolution of its parameters reflects development of the low - temperature phase , confirming the expressed above ideas about formation of dynamic polar clusters at t@xmath44 k , their slowing down , and appearance of the static order at t@xmath83 k. as it is seen from the table [ ramanconc ] , the appearance of this line in the raman spectrum has several contradictory explanations .
however , according to the recent first - principle calculations@xcite , this mode most likely is a consequence of oscillations of oxygen ions .
this paper offers a comprehensive analysis of the raman scattering spectra of pbmg@xmath0nb@xmath1o@xmath2 in the temperature range between 100 and 1000 k. from our analysis , the structural evolution of the lattice occurs in several stages .
even at 1000 k , the pmn crystal is characterized by the presence of dynamic distortions of the lattice from cubic symmetry , which are responsible for existence of first - order raman scattering at such a high temperature . at the burns temperature
t@xmath44 k , the first restrictions on the reorientational motion of the off - centered ions appear .
these restrictions lead to slowing of the motion and growth of correlations between neighboring regions with r3 m symmetry .
they also introduce distinguishability between sites occupied by mg and nb , which causes onset of fm3 m symmetry in the 1:1 ordered areas .
further cooling leads to progressive growth of restrictions on ionic motion and , starting from t@xmath83 k , to the appearance of the static polar nanoregions . at t@xmath48 k ,
the formation of the low - temperature phase is primarily finished .
the shape of the phonon lines is determined by multiphonon processes involving phonons with different polarizations propagating in different directions , which interact with dynamic and static disorder .
authors are very grateful to d.la-orauttapong and g.yong for useful advice and helpful discussions .
this work was partially supported by the onr grant # n00014 - 99 - 1 - 0738 .
t.egami , w.dmowski , m.akbas , p.k.davies , first - principles calculations for ferroelectrics , ed .
by r.e.cohen , fifth williamsburg workshop , williamsburg , va february 1998 , aip conf .
proc . , * 436 * , p.1 ( 1998 ) . | pb(mg@xmath0nb@xmath1)o@xmath2 is one of the simplest representatives of the class of lead relaxors and often serves as a model system for more complicated compounds . in this paper
we analyse both polarized and depolarized raman scattering spectra , measured in the temperature range between 1000 and 100 k , by multiple peak decomposition . basing on this analysis
, we present the picture of transformations in the crystal and the nature of the spectral lines . according to our model ,
the formation of the fm3 m symmetry in the chemically ordered regions as well as the appearance and freezing of the polar nanoregions are the consequences of the same phenomenon : ion off - centered displacements and their reorientational thermal motion .
the lowering temperature progressively limits this motion from eight to four , two and only one site .
raman scattering exists due to the presence of the fm3 m and polar clusters .
it is a product of the multiphonon interaction process , mediated by the presence of the dynamic and static disorder . |
intensity variations have long been recognized as a defining characteristic of active galactic nuclei ( agn ) .
variability is a powerful tool for probing agn geometry and physical properties such as the black hole mass , and the sizes and bulk motions of the outflows in their innermost regions that are well beyond the current imaging capabilities of telescopes in any part of the electromagnetic spectrum ( e.g. , wagner & witzel 1995 ; urry & padovani 1995 ; xie et al .
the variations can be particularly violent for those agns whose flux is dominated by relativistic jets of nonthermal radiation broadly pointing in our direction ( e.g. , begelman , blandford & rees 1984 ) .
intensities of such agns , called ` blazars ' , are known to vary across the entire electromagnetic spectrum and time scales from minutes to years have been observed .
for instance , in the soft x - ray band , several tev emitting blazars have been found to vary on a characteristic time scale of @xmath21 day , with the flares having sub - structures on shorter time scales of @xmath6s ( e.g. , tanihata et al .
2001 ; kataoka et al .
2001 ) . in the optical regime
, there have been many detections of intra - night optical variability ( inov ) , or optical microvariability , following the pioneering work of carini , miller & goodrich ( 1990 ) who first used ccd detectors as multi - object photometers for this purpose .
the shortest time scale found for essentially all inov events is around 1 hour , with an amplitude of a few percent ( e.g. , xie et al .
2001 ; romero et al . 2002 ; stalin et al .
. such rapid continuum variability of blazars is usually explained by invoking relativistic jets ( e.g. , marscher 1996 ; schlickeiser 1996 ; wiita 2006 ) . a study of 23 agn of the quasar / seyfert1 type ( i.e. , non - blazar ) yielded a @xmath7 upper limit of 0.03 mag for variability on hour - like timescales ( webb & malkan 2000 ) .
the literature does contain reports of a few inov detections on time scales much shorter than 1 hour .
examples include the papers by kidger & dediego ( 1990 ) , sagar , gopal - krishna & wiita ( 1996 ) , xie et al .
( 2001 ) , and dai et al .
( 2001 ) ; however , see romero et al .
( 2002 ) for a convincing critique of the latter two claims . to our knowledge
such assertions are lacking in the literature over the past 5 years or so , aside from a very recent paper presenting evidence for a quasi - periodic oscillation of @xmath8 15 min spanning most of a night of optical monitoring of the tev blazar s5 0716@xmath5714 ( rani et al .
2010 ) , which is also a member of our present sample . very recently ,
impiombato et al . (
2011 ) have reported a single event of duration @xmath225 minutes in @xmath9-band while searching for long and short term optical and infrared variability for blazar pks0537@xmath10441 in the data spanning for @xmath26 years ( 2004 - 2009 ) .
there is thus a need for renewed efforts capable of making robust detections of events of ultra - rapid inov events on sub - hour time scales . since @xmath11ray emission is known to be correlated with relatively large bulk doppler factors of the radiating plasma in the blazar jets for both the gev band probed by egret ( e.g. , kellermann et al .
2004 ; lister & homan 2005 ) and the higher energies probed by fermi / lat ( kovalev et al . 2009 ; savolainen et al .
2010 ) , tev blazars appear to be particularly promising candidates for detecting the most rapid inov . in tev blazars ,
the relativistic plasma of the inner jets almost certainly must move with a bulk lorentz factor @xmath12 4550 in order to escape absorption of the tev photons due to the pair production process in the radiation field present near the origin ( e.g. , krawczynski , coppi & aharonian 2002 ; begelman , fabian & rees 2008 ; for large lorentz factors in blazar jets , also see kundt & gopal - krishna 1980 , 2004 ) . according to one model
, such ultra - fast moving emission features may arise in situ within the jet , e.g. , from reconnection events in a poynting flux dominated jet ( giannios , uzdensky & begelman 2009 ) .
thus , it seems likely that in tev blazars the bulk lorentz factors of the optically radiating inner segments of the jets are also comparatively larger than those occuring in other blazars .
here we note that the marked paucity of apparent superluminal motion in the jets of tev blazars , highlighted by piner & edwards ( 2004 ) , can be reconciled with the above requirement of extreme bulk lorentz factors in a number of ways , e.g. , by taking into account a modest opening angle for the jet ( gopal - krishna , dhurde & wiita 2004 ; gopal - krishna , wiita & dhurde 2006 ; gopal - krishna et al . 2007 ; petrucci , boutelier & henri 2010 ) , or by postulating a spine - sheath geometry for the sub - parsec scale jets ( e.g. , attridge , roberts & wardle 1999 ; ghisellini , haardt & matt 2004 ) or , alternatively , if the bulk of the @xmath11rays come from an ultra - relativistically approaching volume element in the jet ( e.g. , giannios et al .
2009 ; also , gopal - krishna , singal & krishnamohan 1984 ) . in sect . 2 we describe the selection of our sample of tev blazars for intranight optical monitoring .
our observations are described in sect . 3 and the results summarized in sect . 4 , followed by a discussion in sect . 5
our sample of tev blazars consists of two sets .
set 1 is derived from the list of tev detected extragalactic agn , published by weekes ( 2008 ; table 2 of his paper ) .
application of the criteria , @xmath13 , @xmath14 and @xmath15 to that list leaves us with 9 tev blazars .
the redshift limit is designed to exclude the nearest blazars so that the optical image should have a point - like appearence .
this is to ensure that the host galaxy makes a negligible contribution to the image , a pre - requisite for high precision photometry .
the declination and the apparent magnitude limits are meant to ensure a sufficiently dense sampling and reasonably long duration ( @xmath16 hours ) for the continuum light curves obtained with the 1 @xmath10 2 metre class telescopes employed in this programme .
the 9 tev blazars constituting set 1 are : tev1219@xmath5283 , tev1429@xmath5427 , tev0809@xmath5524 , tev1218@xmath5304 , tev1011@xmath5496 , tev0716@xmath5714 , tev1553@xmath511 , tev0219@xmath5425 and tev1256@xmath10058 , listed in increasing order of distance from us .
the intra - night and long - term optical lightcurves of these 9 blazars are derived in this study and presented here .
our set 2 of tev blazars was derived from table 1 of abdo et al .
( 2010 ) , consisting of 709 tev detected agns , which is based on 11 months of monitoring with @xmath17 @xmath18 .
this set consists of 13 blazars , including 10 high polarization core - dominated quasars ( hpcdqs ) and 3 bl lac objects .
the hpcdqs were selected employing the following criteria : all cdqs labeled `` hp '' ( @xmath19 per cent ) in the compendium of vron & vron ( 2006 ) were subjected to the aformentioned selection criteria , namely , @xmath13 , @xmath14 and @xmath20 . to ensure the availability of inov data additional selection filters used are :
\(a ) we selected all the 7 hpcdqs from j. noble s phd thesis ( noble 1995 , table 3.1 ) .
these are j0239@xmath51637 , j1159@xmath52914 , j1256@xmath100547 , j1310@xmath53220 , j1643@xmath53953 , j2225@xmath100457 , j2253@xmath51608 .
of these j1256@xmath100547 ( 3c 279 ) is already a member of set 1 and j1643@xmath53953 ( 3c 345 ) is undetected by @xmath17 @xmath21 .
this gave us the first 5 tev blazars for set 2 .
\(b ) another 3 hpcdqs were taken from the polarimetric survey of wills et al .
( 1992 ) , limiting ourselves to the right ascension range @xmath22 @xmath23 and declination range @xmath10@xmath24 to + 40@xmath25 .
note that although these criteria yielded 5 hpcdqs ( j0423@xmath100120 , j0739@xmath50136 , j1058@xmath50133 , j1159@xmath52914 and j1256@xmath100547 ) , the last two of these are already a part of the noble ( 1995 ) sample ( a ) .
this left us with 3 additional blazars .
\(c ) one hpcdq , j1218@xmath100119 , was taken from the first phase of our inov program ( sagar et al .
2004 , stalin et al .
inov data have been taken from those papers for this source as well as for another two hpcdqs ( j0239@xmath51637 and j1310@xmath53220 ) that are in the part ( a ) of set 2 derived from noble ( 1995 ) .
note that these were the only 3 hpcdqs monitored in the first phase of our inov program .
\(d ) lastly , one hpcdq was added from the sample of romero , cellone & combi ( 1999 ) .
they have reported v - band intranight monitoring of a sample of southern agn that contains 4 hpcdqs , according to the vron & vron ( 2006 ) classification .
these are j0538@xmath104405 , j1147@xmath103812 , j1246@xmath102547 , and j1512@xmath100906 .
only the last of these hpcdqs was included in our sample ; because we could ensure a minimum 3 nights monitoring data ; the others are situated far to the south .
\(e ) we have also included the 3 bl lacs pks 0735@xmath5178 , oj 287 and b2 1215@xmath530 for which intra - night monitoring data of duration
@xmath26 4 hours are available in the literature .
this completes our set 2 that contains 13 tev detected blazars .
salient properties of the two blazar sets are listed in table 1 .
all these sources have flat radio spectra ( @xmath27 , where @xmath28 ) as well as high optical linear polarization , with @xmath29 falling in the range 3.5 to 44 per cent , except for j1428@xmath54240 ( which has a slightly steep integrated radio spectrum with @xmath30 , and a comparatively low optical polarization , @xmath29 = 2.5% ) .
the values of radio core luminosity ( @xmath31 ) , extended radio luminosity ( @xmath32 ) , and the radio core - dominance parameter ( @xmath33 ; ratio of core to extended radio luminosities ) at 5 ghz , were determined using the available very long baseline interferometry ( vlbi ) measurements at milli - arcsecond resolution and the integrated nrao vla sky survey ( nvss ) flux values at 1.4 ghz , taking a radio spectral index of zero for the core ( @xmath34 ) and @xmath35 for the extended emission .
the concordance cosmological model was assumed , with a hubble constant @xmath36 km sec@xmath37 mpc@xmath37 , @xmath38 and @xmath39 ( spergel et al . 2007 ) .
values of the radio loudness parameter ( @xmath40 ) were determined following stocke et al .
the absolute magnitudes , @xmath41 , were calculated taking the total galactic extinction from schlegel , finkbeiner & davis ( 1998 ) and assuming an optical spectral index @xmath42 of @xmath43 .
the observations were mainly carried out using the 104-cm sampurnanand telescope ( st ) located at the aryabhatta research institute of observational sciences ( aries ) , in naini tal , india .
it has ritchey - chrtien ( rc ) optics with a f@xmath4413 beam ( sagar 1999 ) .
the detector was a cryogenically cooled 2048 @xmath45 2048 chip mounted at the cassegrain focus .
this chip has a readout noise of 5.3 e@xmath46/pixel and a gain of 10 e@xmath46@xmath44analog to digital unit ( adu ) in the usually employed slow readout mode .
each pixel has a dimension of 24 @xmath47m@xmath48 which corresponds to 0.37 arcsec@xmath48 on the sky , covering a total field of 13@xmath49 @xmath45 13@xmath49 .
observations were carried out in 2 @xmath45 2 binned mode to improve the s@xmath44n ratio .
all the observations with the st were carried out using an _ r _ filter for which the ccd sensitivity is maximum .
the seeing ranged mostly between @xmath50 and @xmath23@xmath51 , as determined using 3 moderately bright stars within the ccd frame . for each night , the plot of seeing is provided in figure 1 in the corresponding bottom panel . the other telescope used for our monitoring of tev blazars is the 201-cm himalayan chandra telescope ( hct ) of the indian astronomical observatory ( iao ) located at hanle , india , which is also of the rc design with a f@xmath449 beam at the cassegrain focusiao ] .
the detector was a cryogenically cooled 2048 @xmath45 4096 chip , of which the central 2048 @xmath45 2048 pixels were used .
as the pixel size is 15 @xmath47m@xmath48 the image scale of 0.29 arcsec@xmath44pixel covers an area of 10@xmath49 @xmath45 10@xmath52 on the sky .
the readout noise of the ccd is 4.87 e@xmath46/pixel and the gain is 1.22 e@xmath46@xmath44adu .
this ccd was used in an unbinned mode .
the seeing ranged mostly between @xmath53 to @xmath54 .
lastly , one night of blazar monitoring data were obtained using the 200-cm iucaa girawali observatory ( igo ) telescope located near pune , india .
it has an rc design with a f@xmath4410 beam at the cassegrain focustele@[email protected] ] .
the detector was a cryogenically cooled 2110@xmath452048 chip mounted at the cassegrain focus .
the pixel size is 15 @xmath47m@xmath48 so that the image scale of 0.27 arcsec@xmath44pixel covers an area of 10@xmath49 @xmath45 10@xmath52 on the sky .
the readout noise of this ccd is 4.0 e@xmath46/pixel and the gain is 1.5 e@xmath46@xmath44adu . the ccd was used in an unbinned mode .
the seeing ranged between @[email protected] and @[email protected] on that particular night .
the observations were made using _
_ filters .
the exposure times were typically 10 to 12 minutes for the aries and igo observations and ranged between 3 to 6 minutes for the observations from iao ( depending on the brightness of the source , the lunar phase and the sky transparency for the night ) .
the field positioning was adjusted so as to always have within the ccd frame 23 comparison stars within about a magnitude of the blazar , in order to minimize the possibility of getting spurious variability detection ( e.g. , cellone , romero & araudo 2007 ) .
for all three telescopes , bias frames were taken intermittently and twilight sky flats were obtained on each night .
each blazar in set 1 was monitored for a minimum of 3 nights .
likewise , for the blazars in set 2 , this requirement of minimum 3 nights is met , except for the case of j2253 + 1608 ( 3c 454.3 ) for which only 2 nights of monitoring data are available . the preprocessing of the images ( bias subtraction , flat - fielding and cosmic - ray removal ) was done by applying the standard procedures in iraf and midas softwares .
the instrumental magnitudes of the blazar and the comparison stars in the image frames were determined by aperture photometry using daophot ( stetson 1987 ) .
the magnitude of the blazar was measured relative to the steady comparison stars present on the same ccd frame ( table 2 ) . in this manner ,
differential light curves ( dlcs ) of each blazar were produced relative to 3 comparison stars ( fig .
1 ) . for each night
, the selection of optimum aperture size for the photometry was done by examining the observed dispersions in the star - star dlcs for different aperture radii starting from the median seeing ( fwhm ) value on that night to 4 times that value .
we selected the aperture that showed minimum scatter for the steadiest dlc found for the various pairs of the comparison stars ( e.g. , stalin et al .
the inov dlcs are shown in figure 1 for all the 9 blazars of set 1 . from set 2
we present here the dlcs for just j0854@xmath52006 ( oj 287 ) since it was also monitored in the present work ( fig .
the dlcs for another few members of the set 2 , which too were monitored in the present work , can be found in goyal et al .
( 2011 ) as they form part of the samples discussed in that paper .
figure 2 displays the long - term optical variability ( ltov ) dlcs for sources that could be observed in the same colour filter on a minimum of 3 nights .
table 3 summarizes the observations and the derived results for our entire sample of 22 tev blazars , 9 of which were monitored in the present study ( set 1 ) , while for the remainder ( set 2 ) the inov data were largely taken from the literature ( sect . 2 ; table 3 ) . the 6th , 7th and 8th columns give , respectively , the monitoring duration on the respective night , the number of data points ( n@xmath56 ) in the dlc , and the rms of the steadiest star
star dlc obtained using two of the comparison stars .
the next columns in table 3 contain the measures of the source variability ( inov ) on each night .
the fractional amplitude of inov , @xmath4 , is given in col .
( 9 ) , while @xmath57 , a widely used indicator of variability status , is in col .
the classification ` variable ' ( v ) or ` non - variable ' ( n ) as decided using the @xmath3-test , basically defined following the criteria of jang & miller ( 1997 ) , is in col . (
@xmath57 for a night is derived by combining the @xmath3-values estimated for individual dlcs of the blazar on that night . for a given dlc , @xmath3
is defined as the ratio of its standard deviation , @xmath58@xmath59 and @xmath60 , where @xmath61 is the average of the rms errors of its individual data points . as the photometric errors given by the @xmath62 package
is known to be underestimated , a compasatory factor @xmath63 is determined that would make the rms of the dlc consistent with the rms of its individual data points ( see edelson et al .
2002 , also , stalin et al .
in this way , the computed value of @xmath63 is found to be @xmath21.5 ( stalin et al . 2004a , 2004b , 2005 ; gopal - krishna et al .
2003 ; sagar et al .
2004 ) . however , our analysis for the present dataset yields @xmath64 and so we have adopted this value here .
we computed @xmath57 from the @xmath3 values ( as defined above ) derived for the individual dlcs of a given blazar relative to 3 different comparison stars which were monitored simultaneously with the blazar on the same ccd chip .
this gave us 3 values of @xmath3 for a given blazar .
these values were converted into probabilities and multiplied to obtain @xmath57 for the blazar ( sagar et al .
2004 for details ) .
this has the advantage of using the available multiple dlcs of an agn , relative to different comparison stars .
the agn is termed ` v ' for @xmath65 , corresponding to a nominal confidence level above 0.99 .
the probable ` variable ' ( pv ) classification applies when , @xmath66 , corresponding to a nominal confidence level between 0.95 to 0.99 .
it has been recently argued by de diego ( 2010 ) , however , that @xmath3 is not a proper statistic as it is based on ratios of standard deviations and not on ratios of variances ; only the latter are distributed in such a way that @xmath67 tests can be used to assign proper confidence levels .
he shows that the standard @xmath68-statistic , where @xmath69 is a more appropriate choice for characterizing agn light curves .
it also has the advantage that the somewhat uncertain parameter , @xmath63 , cancels out .
12 gives the @xmath68-value for each night along with the number of degrees of freedom shown in parentheses .
since we can compute the @xmath68-statistic only for our data , for the bulk of blazars in set 2 ( not observed by us ) only the @xmath3-values are available . in computing the @xmath68-values we examined the various star star light curves to decide if any of the comparison stars might have even slightly varied on that night .
we considered the steadiest two of the three comparison stars and selected the one closer to the blazar in apparent magnitude . the two dlcs ( namely , blazar@xmath10star and star@xmath10star ) involving the selected comparison star
were then used in the @xmath68-test ( eq .
1 ) . for a fair comparison to the @xmath3-test , we take for the @xmath68-test a significance of @xmath70 0.99 to correspond to a definitely variable source ( v ) and a significance between 0.95 and 0.99 to correspond to the pv classification .
these indicators of variability status as computed using @xmath68-test are tabulated in the 13th column .
the last column gives the reference(s ) for the inov data used here .
the peak - to - peak inov amplitude is calculated using the definition ( romero et al . , 1999 )
@xmath71 with @xmath72 = maximum in the agn s differential light curve , @xmath73 = minimum in the agn s differential light curve , and @xmath74= @xmath75@xmath76 . the inov duty cycle ( dc ) for our entire sample of 22 tev blazars ( table 3 )
was then computed following the definition of romero et al .
( 1999 ) ( see also , stalin et al .
2004a ) : @xmath77 where @xmath78 is the duration of the blazar monitoring session on the @xmath79 night , corrected for the blazar s cosmological redshift , @xmath80 .
note that since the monitoring durations for any given source on different nights were not equal , the computation of dc has been weighted by the actual monitoring duration @xmath81 on the @xmath79 night .
the parameter @xmath82 was set equal to 1 if inov was detected ; otherwise @xmath82 = 0 .
we realize that 2 of our 9 blazars in set 1 lie at rather small redshifts ( table 1 ; j1221 + 2813 at @xmath80 = 0.102 and j1428@xmath54240 at @xmath83 ) , raising the possibility of a significant contribution from the host galaxy to the flux falling within the circular photometry aperture centered at the blazar .
as stressed by , e.g. , cellone , romero & combi ( 2000 ) , the host s varying contribution to the flux within the aperture changes as the seeing varies and thus the atmospheric seeing changes can yield spurious inov detection .
the paper also presents simulated dlcs for agn , relative to a suitable comparison star , for cases where the emission from the agn host galaxy ( elliptical or spiral ) is comparable to that from the agn itself and the atmospheric seeing undergoes a large intranight variation . for a wide range in the host galaxy size ,
they find that even if the host s flux is comparable to that of the agn ( an extreme situation from the perspective of the present sample ) , the dlcs show a negligible variation ( @xmath84 ) as long as the aperture radius exceeds @xmath85 and the seeing remains within this limit ( as also applicable to the present study ) . therefore
, since our choice of aperture size always meets this condition , we do not expect any of our dlcs classified as ` v ' to be spurious , i.e. , being an artefact of variable atmospheric seeing during the night .
note also that in a very small number of cases the inov results taken from the literature were for the v band ; however , for the present purpose we do not distinguish them from our data which were essentially always taken in the r band .
based on the @xmath3-test the computed inov dc for our sample of 22 tev blazars is found to be @xmath86 per cent ( 116 nights ; table 3 ) , which increases to @xmath87 per cent if five ` pv ' cases of ` probable ' inov are also included . however , for the nights showing larger inov amplitudes , @xmath88 , the inov dc is @xmath89 per cent . since the redshift
the blazar j1555 + 1111 is known to be controversial , with values ranging from 0.25 to 0.50 ( see , treves , falomo & uslenghi , 2007 and magic collaboration : albert et al . 2008 ) we have computed the @xmath90 vlaues for our entire sample of blazars ( 116 nights ) , taking the lower and upper @xmath80 values for j1555 + 1111 .
the computed values are @xmath90 = 58.6% and @xmath90 = 58.9% , respectively .
thus , the uncertainty in @xmath80 of j1555 + 1111 does not significantly affect the estimated @xmath90 for the sample ( i.e. , dc @xmath259% ) .
using the @xmath68-test , the corresponding @xmath90 values for the 9 blazars monitored presently , is @xmath91 per cent ( @xmath92 per cent , if the two ` pv ' cases are included ) but only @xmath93 per cent for the cases having @xmath88 ( 55 nights ; table 3 ) .
the basic parameters for all these 9 blazars monitored by us have been taken from the compilations by abdo et al .
( 2010a ) and weekes ( 2008 ) and their inov and ltov lightcurves determined in the present work are shown in fig . 1 & fig . 2 , respectively . as seen from table 1 ,
the set 1 of tev blazars consists of five hbls , two ibls , one lbl and one fsrq .
* j0222@xmath54302 ( 3c 66a ; @xmath94 ) + this blazar ( bramel et al . 2005 ) has been classified as an intermediate - peaked bl lac object ( ibl ) whose nonthermal emission peaks in the range @xmath95 hz ( perri et al . 2003 ) . during 19982000
it was monitored in our programme on 7 nights , for durations ranging between 5 and 10 hours per night .
inov was detected on 6 of the nights ( sagar et al .
the only sub - hour feature seen in those dlcs is a 1.5 per cent ` glitch ' detected on the night of 13 nov .
1999 at 19.6 u.t . in table 3
we have combined those published data with the two dlcs obtained in the present work . of these ,
the dlc on 28 sept .
2009 showed confirmed inov using both @xmath57 and @xmath68 statistics . also , a clear event of _ internight _ variability was observed when the source faded by @xmath20.04 mag between 27 and 28 sept .
2009 ( fig .
* j0721@xmath57120 ( s5 0716 + 714 ; @xmath96 ) + this blazar with a ` lbl ' classification is the archetypal intra - day variable ( e.g. , wagner et al .
1996 ) with a history of frequent high - amplitude flux variations ( e.g. , gupta et al .
2008 and references therein ) . from their @xmath97-band light curves
wagner et al .
( 1996 ) had reported a significant `` flickering '' on time scale as short as @xmath98 minutes .
recently , its high temporal resolution study revealed quasi - periodic oscillations of about 15 minutes at @xmath99 significance ( rani et al .
the presence of such short time scales in an optical light curve can provide important clues to relativistic beaming at these wavelengths ( e.g. , fabian & rees 1979 ; guilbert , fabian & rees 1983 ) . during february and march 1994
this blazar was the target of a 4-week long inov monitoring campaign ( bvri ) under our programme , using two indian telescopes : the 1-m st and the 2-m vainu bappu telescope ( sagar et al . 1999 ) .
ghisellini et al .
( 1997 ) have reported a multi - colour optical monitoring campaign around that time , revealing a moderately active state of this blazar . in the indian campaign , a monitoring duration of 23 hours
was typically achieved on each night .
no evidence was found for inov on time scale shorter than 1 hr , but 3 prominent events on @xmath100 hour time scales were detected . also , _
variability with @xmath101 per cent was frequently detected during the 4-week long indian campaign .
results from other recent monitoring campaigns on this source performed at different sites should be available soon ( j. webb ; a. gupta , private communications ) .
+ the present one night s monitoring of this blazar detected a clear flare lasting @xmath102 hour ( fig . 1 ; table 3 ) .
however , since the 1.5 hour duration of our monitoring is much less than our selection criterion of @xmath264 hour , we have not included this observation in the computation of dc for this blazar ( table 3 ) .
note that neither of the aforementioned published observational campaigns revealed inov time scale @xmath103 hour , even for this highly variable blazar , except for results reported by wagner et al .
( 1996 ) and some recent observations reported by rani et al .
* j0809@xmath55218 ( 1es 0806@xmath5524 ; @xmath104 ) + to our knowledge , the present work is the first intranight optical monitoring of this hbl ( 4 nights , table 3 ) .
a highly unusual aspect of this source , as revealed by the 5 ghz vlbi observations , is a two - sided jet structure on parsec scale straddling the core ( chen et al .
2006 ) , which itself has a brightness temperature of @xmath105 k , markedly lower than the inverse - compton limit ( kellermann & paulini - toth 1969 ) .
these unusual characteristics are , however , consistent with the absence of strong relativistic boosting in this hbl , as also reflected by its quite stable gev flux ( chen et al .
2006 ) . on none of the 4 nights
did we detect inov in this blazar ( table 3 ) , although on longer term ( @xmath106 month timescale ) , a clear fading by @xmath20.45 mag was detected .
thereafter , a distinct brightening by @xmath2 0.04 mag occured between the observations taken 4 days apart ( fig . 3 ) .
* j1015@xmath54926 ( 1es 1011@xmath5496 ; @xmath107 ) + out of the 3 nights observed by us this hbl showed a clear variability on one night , as confirmed by both @xmath57 and @xmath68-statistics ( table 3 ; 7 mar .
2010 ) . on the night of 19 feb 2010
it was found to be a probable variable using @xmath57 but classified as `` v '' on application of @xmath68-statistics . on a longer time scale , a steady fading by @xmath20.11 mag was seen over the 1@xmath10month span covered by our monitoring ( fig .
* j1221@xmath53010 ( 1es 1218@xmath5304 ; @xmath108 ) + out of 3 nights observed by us this blazar showed a hint of variablity on one night on applying the @xmath68-statistics , but remained non - variable on all the three nights , according to the @xmath57 criterion ( table 3 ) .
however , on a longer term , we detected a clear fading by @xmath20.1 mag between the first two epochs , which were separated by 10 days .
this was followed by a phase of 0.4 mag brightening over @xmath2 2 months , when it was last monitored by us ( fig .
* j1221@xmath52813 ( w comae / on231 , @xmath109 ) + this is the nearest source in our sample and also the first intermediate type bl lac ( ibl ) to be detected at tev energies .
gupta et al . (
2008 ) have reported r - band monitoring of this blazar on 11 jan .
2007 for 3.24 hours , but no inov was detected .
we monitored this blazar on 4 epochs .
it showed a confirmed variablity on 3 epochs using @xmath57 and on all 4 epochs when the @xmath68-test was applied ( table 3 ) .
in addition , a clear event of _ internight _ variation was seen when the source faded by @xmath20.1mag between 19 and 20 mar 2004 .
* j1256@xmath100547 ( 3c 279 , @xmath110 ) + this fsrq of the ovv type was the first blazar to be detected as an egret/@xmath111-ray source ( hartman et al .
1992 ) and the first fsrq to be detected at very high energies ( vhe , i.e. @xmath70 100 gev ) ( magic collaboration : albert et al .
this source is also particularly interesting , because with a redshift of 0.538 , it is the most distant vhe blazar yet found ( abdo et al .
2009 ) ; a strong absorption of its vhe emission due to extragalactic background light ( ebl ) is expected , yet not seen ( e.g. , costamante et al .
gupta et al . (
2008 ) have reported @xmath0 band intra - night monitoring with the 1-m yunnan observatory telescope on two nights in early 2007 ( for durations of 4 hour and 2.3 hours ) , but inov was not detected .
a negative result was also reported by romero et al .
( 2002 ) from their v - band intranight monitoring lasting 3.8 hours on 06 aug .
+ application of either of the two statistics shows it to be a confirmed variable on all the 3 nights it was monitored by us , attaining peak - to - peak inov amplitudes of 4 , 10 and 22 per cent , respectively ( table 3 ) .
further , the blazar brightened by @xmath20.9 mag between the first two epochs of monitoring that were separated by about 2 months , and was about 2.0 mag fainter when last observed a little more than 3 years later ( fig . 3 ) .
* j1428@xmath54240 ( h1426@xmath5428 , @xmath112 ) + this weak tev source has its synchrotron emission peak at the highest frequency known for any blazar ; hence termed as `` extreme '' hbl ( aharonian et al .
2002 ; also , costamante et al .
its radio structure consists of a core surrounded by a faint halo ( giroletti et al .
the light curves presented here are its first intranight optical monitoring observations .
the @xmath57 values indicated it was non - variable on all 3 nights we monitored it ; however , on application of the @xmath68-test it was found to be variable on 1 of the 3 nights ( table 3 ) . *
j1555 + 1111 ( pg 1553@xmath5113 , @xmath113 ) + this most distant hbl with a firm tev detection ( abdo et al .
2009 ) is known to show a highly time dependent variability behaviour at different frequencies .
its radio structure is marked by an extremely large bend ( by @xmath114 ) of the parsec - scale jet towards the outer lobe ( rector , gabuzda & stocke 2003 ) .
it showed inov on one of the two nights it was monitored during mid-1999 as part of our inov programme ( stalin et al .
those data have been included in our computation of the inov duty cycle ( table 3 ) . in our monitoring , confirmed variability was detected on all 3 nights both using @xmath57 and @xmath68 statistics . in the longer term , a clear brightening by @xmath2 0.25 mag
was observed between the first two epochs of our monitoring , which were separated by almost a year . a clear case of _ internight _ variability also occured when the source brightened by another @xmath20.04 mag between 15 and 16 may 2010 ( fig .
3 ) . set 2 consists of 1 hbl , 3 lbls , 8 fsrqs and 1 bl lac object of unspecified hbl / lbl classification .
their inov and ltov data are available in the literature cited in table 3
. only for oj 287 do we present here new monitoring data taken by us on 2 nights ( table 3 ; fig .
* j0238@xmath51637 ( ao 0235 + 164 ; @xmath115 ) + this lbl had been monitored by noble ( 1995 ) on 5 nights and later on 3 nights in the first part of our programme ( sagar et al .
on all these 8 nights it was found to be a confirmed variable , with @xmath4 ranging between 520 per cent ( table 3 ) .
romero et al .
( 2002 ) monitored it on another 6 nights and found @xmath4 to range between 744 per cent .
in addition , goyal et al .
( 2011 ) have monitored this blazar on one night and it showed a gradual decline by @xmath2 7 per cent during the first 6 hours on that night .
thus , the inov duty cycle of this bl lac is essentially @xmath116 per cent ! * j0423@xmath100120 ( pks 0420@xmath1001 ; @xmath117 ) + this blazar has been newly monitored by us on 3 nights , covering a 7 years time span ( goyal et al .
2011 ) it showed a confirmed variability on 19 nov .
2003 , with an inov amplitude of @xmath4 @xmath118 per cent and was a ` probable variable ' on 25 oct .
2009 ( @xmath3-test ) .
the @xmath68-test shows it to be a confirmed variable on all the 3 nights . as for ltov , it faded by @xmath21.9-mag between the first 2 epochs of monitoring roughly a year apart and then brightened by @xmath20.8-mag when we last monitored it on 25 oct . 2009 . * j0738@xmath51742 ( pks 0735 + 178 ; @xmath119 ) + this blazar is a rather unique case of relative intra - night quiescence that has persisted over the past two decades ( goyal et al . 2009 ; britzen et al .
2010 ) , although mild inov ( @xmath4 @xmath1203 per cent ) was detected on 5 out of the total 17 nights . on month - like timescale
it has exhibited @xmath20.5 mag variations ( sagar et al .
2004 ; ciprini et al . 2007 ;
goyal et al .
2009 ) . * j0739@xmath50137 ( pks 0736@xmath501 ; @xmath121 ) + not only did this highly polarized cdq remain a confirmed variable ( by both @xmath3 & @xmath68 statistics ) on 2 of the 3 nights of our monitoring ( goyal et al .
2011 ) , it also showed a clear _ internight _ variability , fading by @xmath2 4 per cent between 5 and 6 dec .
* j0854@xmath52006 ( oj 287 ; @xmath122 ) + this lbl was monitored by sagar et al .
( 2004 ) on 4 nights and it showed confirmed inov each time .
it also showed large ltov , first fading by 0.6 mag over @xmath2 1 year and then brightening by 1.63 mag on a 2-year time span . in fig . 2 we present the light curves derived from our newly acquired data on 2 nights .
the object showed a clear brightening by @xmath123 per cent in 4 hours on 12 apr 2005 .
however , on 5 feb 2005 , it remained a non - variable ( by the @xmath3-test ) , although a @xmath124test showed it to be variable .
* j1058@xmath50133 ( pks 1055@xmath501 ; @xmath125 ) + this fsrq with a variable polarization was a confirmed variable on 2 of the 3 nights we monitored it ( goyal et al .
it also showed a gradual brightening by @xmath126 magnitude between 25 mar .
2007 and 23 apr . 2007 . *
j1159@xmath52914 ( 4c 29.45 ; @xmath127 ) + confirmed inov was detected on all the 6 nights this fsrq was monitored by noble ( 1995 ) , with @xmath4 ranging between 4 and 12 per cent ( table 3 ) .
it has also shown large variations on _ internight _ time scales .
firstly , it brightened by 6 per cent between 19 and 21 jan .
1994 and then faded by @xmath20.5 mag between 21 and 22 jan.1994 .
it again brightened by 7 per cent between 22 and 23 jan .
1994 and was last found having faded by @xmath20.3 mag on the following night .
* j1217@xmath53007 ( b2 1215@xmath530 ; @xmath128 ) + this blazar was monitored on 4 epochs by sagar et al .
( 2004 ) , showing confirmed inov on 2 epochs ( table 3 ) .
variability was also detected on a longer time scale .
* j1218@xmath100119 ( pks 1216@xmath1001 ; @xmath129 ) + inov was detected on all the 4 nights this blazar was monitored by sagar et al .
( 2004 ) . on longer time scales
, it showed a 2 per cent fading between the first 2 epochs of monitoring , followed by @xmath130 per cent fading over the next 2 days and , finally , 14 per cent brightening within the next 24 hours ( table 3 ) .
* j1310@xmath53220 ( b2 1308@xmath532 ; z = 0.997 ) + this fsrq showed a confirmed inov ( @xmath4 up to 3.4 per cent ) on one out of the 4 nights it was monitored by sagar et al .
( 2004 ) . in the longer term
, it showed a strong variability ( stalin et al .
2004a ) . *
j1510@xmath100906 ( pks 1510 - 08 ; @xmath131 ) + this fsrq showed confirmed inov on 2 of the 3 nights it was monitored by us over the span of 5 years ( goyal et al .
, in prep . ) . in the long term , it brightened by @xmath132 mag over the course of 4 years and then faded by 0.25 mag over the next 19 days when we last monitored it on 20 may 2009 . in earlier inov campaigns by romero et al .
( 2002 ) and stalin et al .
( 2005 ) , this blazar was found to be non - variable . *
j2225@xmath100457 ( 3c 446 ; @xmath133 ) +
this blazar showed inov on one of the 2 nights it was monitored by noble ( 1995 ) ( table 3 ) . on 8 oct 2010 when we monitored it ( goyal et al .
2011 ) , it was found to be non - variable using the @xmath134test , but variable by the @xmath68-test ( table 3 ) .
* j2253@xmath51608 ( 3c 454.3 ; @xmath135 ) + inov was detected on both the nights it was monitored by noble ( 1995 ) ( table 3 ) .
the fairly large size of the tev blazar sample covered in the present study and the fact that all the key blazar subclasses , namely hbl , ibl , lbl and fsrq ( sect . 2 ;
table 1 ) are included in the sample , reassures us that the inov results reported here should be representative of tev blazars .
an explicit goal of the present work , the first systematic study to define the inov characteristics of tev blazars , was to search for variations on sub - hour time scales . as pointed out in sect .
1 , tev blazars are particularly promising targets to look for such ultra - rapid optical continuum variability , since their innermost jets are thought to have extremely large bulk lorentz factors , @xmath136 . despite examining the high - quality intranight monitoring observations of all 22 tev blazars in our sample , taken on 116 nights over a total duration of 677 hours , no clear event on sub - hour time scales
was found , even though the data sampling rate ( typically , once every 10 minutes or so ) was adequately dense to have revealed any such cases .
this demonstrates that the occurrence of optical continuum variability on sub - hour like time scales must be exceedingly rare , at least for amplitudes above @xmath2 1 per cent , the detection threshold typically reached in the deepest inov searches made so far .
it may nonetheless be noted that our intranight dlcs ( fig .
1 ) do exhibit a few clear cases of rather sharp `` bumps '' , notably in the cases of j0721@xmath57120 ( 1 feb .
2005 ) , j1221@xmath52813 ( 20 mar .
2004 ) and j1256@xmath100547 ( 26 and 28 jan .
2006 ) . as mentioned in sect . 1 ,
parsec scale radio jets in @xmath11ray ( egret ) blazars are known to show faster apparent ( superluminal ) motion as well as relatively higher brightness temperatures in their cores , compared to non - egret blazars ( e.g. , kellermann et al . 2004 ; jorstad et al .
2001 ; taylor et al .
2007 ; lister et al .
indeed , in an earlier part of our inov programme , evidence was reported for egret detected blazars to exhibit a stronger inov , again suggesting a link between inov and the jet speed ( stalin et al .
an independent evidence for such a physical link has emerged in the present study , from a comparison of the inov duty cycles determined for the two major subclasses of bl lacs , namely ` low - peaked - bl lacs ( lbl ) ' and ` high - peaked bl lacs ( hbls ) ' .
the synchrotron emission from lbls peaks in the near - infrared / optical domain while for hbls the synchrotron peak falls in the uv / soft x - ray regime ( urry & padovani 1995 and references therein ) . of the two classes ,
lbls are known to display greater activity and this is generally attributed to their emission being dominated by faster jets which are probably also better aligned to our direction ( e.g. , ghisellini & maraschi 1989 ; sambruna , maraschi & urry 1996 ; kharb , gabuzda & shastri 2008 ) .
consistent with this picture , it has been found that lbls display stronger inov than do hbls ; the duty cycle has been estimated to be @xmath270 per cent for lbls and @xmath23050 per cent for hbls ( e.g. , romero et al . , 1999 , 2002
; see also , heidt & wagner 1998 ) .
the present study allows us to check , for the first time , whether the difference persists even when the two blazar subclasses are subjected to the condition of tev detection . from table 1 ,
our sample consists of 7 hbls ( with the lower hump of the sed peaking above @xmath137 hz ) and 15 other blazars which can be together termed lbls ( see , abdo et al .
2010b ; li et al .
2010 ) including j1218@xmath100119 which is a flat - spectrum core dominated quasar .
the estimated inov duty cycles are 38 per cent for hbls ( 26 nights data ) and 63 per cent for lbls ( 90 nights data)(table 3 ) . if only the cases of inov amplitudes @xmath138 per cent are considered , the duty cycle is 22 per cent for hbls and 50 per cent for lbls .
both the estimates for the tev detected lbls are in close agreement with those reported in gopal - krishna et al .
( 2003 ) , for radio selected bl lacs ( believed to be predominently lbls ) .
figure 4 shows the histograms of inov dcs for hbls and lbls in our sample .
the k - s test rejects at the @xmath139 per cent confidence level the hypothesis that the two distributions are drawn from the same parent population .
thus , it is evident that the strong tendency for hbls to show a milder inov , vis - - vis lbls , persists even if the comaprison is restricted to their tev detected subsets .
this result must be accounted for in physical models of inov and tev emission from blazars .
a striking outcome of our search for ultra - rapid inov is the detection of a curious , sharp feature in the light curve of the hbl j1555@xmath51111 ( on 24 june 2009 ; fig . 1 ) , the most optically luminous member of set 1 ( table 1 ) .
intriguingly , instead of being a flare , as is more typical , this essentially symmetric feature is a clear `` dip '' , with an amplitude of @xmath2 4.2 per cent .
its initial , falling side extends 0.5 hour and the rising side is of a slightly longer duration .
we treat this detection as robust , since ( i ) the profile of the `` dip '' is well resolved ( 8 data points ) ; ( ii ) its peak amplitude is @xmath70 20 times the rms noise of individual data points ; ( iii ) the variation is visible with equal clarity and amplitude on the dlcs of the blazar against all the 3 comparison stars ; ( iv ) the 3 comparison stars are all within 0.8 magnitude of the blazar ; ( v ) all 3 comparison stars remained rock - steady through the monitoring duration ; and ( vi ) the atmospheric seeing too remained steady throughout the monitoring session ( fig . 1 ) .
all these points , along with the fact that in the dlcs , the brightness levels seen immediately prior to and just after the dip are fairly steady and well matched , _ make this feature by far the most credible , if not the only , case of an emission dip ( on an hour - like time scale ) detected in the optical continuum light received from a blazar_. in addition , the blazar was substantially brighter on both other nights it was monitored by us ( table 3 ) , making it quite unlikely that the elevated flux levels seen just before and after the dip in the light curves of 24 june 2009 are a result of multiple flares occuring at those times .
the robust detection of the dip , combined with its temporal sharpness , calls for an explanation . a plausible scenario for
this rare feature would be that the optical continuum from the jet was temporarily absorbed by a foreground dusty cloud in the nuclear region .
an emerging picture of agn broad emission line clouds suggests they form at the distances from the central continuum source where the temperature becomes low enough for dust formation ( i.e. , @xmath140 k ) and these dusty clouds then form a dust - driven wind ( e.g. , czerny & hryniewicz 2010 ; also , krongold , binnete & hernandez - ibarra 2010 ) . recalling that the low - level detection ( or non - detection ) of emission lines in the case of bl lacs could well be due to a paucity of energetic photons to ionise the gas clouds existing around the central engine ( e.g. , ghisellini et al .
2009 ) , it is plausible that a dusty cloud , or a stream of such clouds happens to be on our line of sight to a superluminally moving optical synchrotron ` knot ' in the jet . to see
if this could explain the `` dip '' , we need an estimate of the size of an optically emitting knot in blazar jets . here
a reasonable expectation would be that optical knots already exist in the jet before the point where the knot becomes visible in the radio band ( e.g. marscher et al .
2008 ) ; that point typically occurs at a distance , @xmath141 @xmath142 times the gravitational radius of the central supermassive black hole ( lobanov 2010 ) . for a typical supermassive black hole of mass @xmath143
, the radio visibility point would thus be about @xmath144 cm away from the nucleus , from which a size of around @xmath137 cm can be reasonably inferred for the optical knot in the jet .
the observed time scale of the optical continuum dip ( @xmath140 sec ) , if interpreted in terms of an occultation of the superluminal knot by a dusty gaseous cloud ( see above ) , would then require a relative transverse speed of order of @xmath145 ( see gopal - krishna & subramanian 1991 for such a scenario ) .
the corresponding bulk lorentz factor of the inner jet , @xmath146 for viewing angle @xmath147 2@xmath25 , is not unreasonable for a tev blazar ( sect .
one prediction of this scenario is that any such rapid intensity dips should appear more pronounced in b - band light curves , because of their greater sensitivity to dust obscuration , as compared to the r - band light curves .
finally , searches for minute - like time scales in the optical light curves of blazars , a key motivation for the present work , have acquired added significance in the present fermi / lat era ( atwood et al .
a few recent studies of blazars have revealed a tight physical relationship between @xmath11ray and optical flaring events , strongly suggesting a spatial coincidence between their origins , which could well be in the jet s knot crossing a standing shock , located up to several parsecs from the central engine ( see , e.g. , agudo et al .
2010 , and references therein ) .
more specifically , a recent study of blazars ( ammando et al . 2010 ) has revealed that polarized optical emission from their jets is spatially coincident with the site of the tev flares observed on minute - like time scales .
this emerging scenario provides further motivation for more concerted efforts to search for optical variability of tev blazars on minute - like time scales .
the authors are thankful to the anonymous referee for helpful suggestions .
this research has made use of nasa / ipac extragalactic database ( ned ) , which is operated by the jet propulsion laboratory , california institute of technology , under contract with national aeronautics and space administration .
we thank n. k. chakradhari for carrying out observations for us .
the authors wish to acknowledge the support received from the staff of the iao and crest of iia and the iucaa - girawali observatory ( igo ) of iucaa .
abdo a. a. , et al . , 2009 ,
apj , 700 , 597 abdo a. a. , et al . , 2010a , apj , 716 , 30 abdo a. a. , et al . ,
2010b , apjs , 188 , 405 aharonian f. , et al .
, 2002 , a&a , 384 , 23 agudo i. , et al . , 2010 ,
preprint ( astroph/1011.6454v1 ) ammando f. d. , 2010 , in scineghe 2010 workshop , september 8 - 10 , 2010 , trieste , italy . to appear in il nuovo cimento c - colloquia on physics .
preprint ( astroph/1012.1120 ) andruchow i. , romero g. e. , cellone s. , 2005 , a&a , 442 , 97 attridge , j. m. , roberts , d. h. , wardle , j. f. c. , 1999 , apj , 518 , l87 atwood w. b. , 2009 , apj , 697 , 1071 begelman m. c. , blandford r. d. , rees m. j. , 1984 , rvmp , 56 , 255 begelman m. c. , fabian a. c. , rees m. j. , 2008 , mnras , 384 , l19 bramel d. a. , et al . , 2005 , apj , 629 , 108 britzen s. , et al . , 2008 , a&a , 484 , 119 britzen s. , et al . , 2010 , a&a , 515 , 105 carini m. t. , miller h. r. , goodrich b. d. , 1990 , aj , 100 , 347 cellone s. a. , romero g. e. , combi j. a. , 2000 , aj , 119 , 1534 cellone s. a. , romero g. e. , araudo a. t. , 2007 , mnras , 374 , 357 ciprini s. , et al , 2007 , a&a , 467 , 465 chen y. , gu m. , shen z .- q . , fan z. , 2006 , mnras , 370 , 1885 costamante l. , et al . , 2001 , a&a , 371 , 512 costamante l. , aharonian f. , buehler r. , khangulyan d. , reimer a. , reimer o. , 2009 , preprint ( astroph/0907.3966 ) czerny b. , hryniewicz k. , preprint ( astroph/1010.6201 ) dai b. z. , xie g. z. , li k. h. , zhou s. b. , liu w. w. , jiang z. j. , 2001 , aj , 122 , 2901 de diego j. a. , 2010 , aj , 139 , 1269 edelson r. , turner t. j. , pounds k. , vaughan s. , markowitz a. , marshall h. , dobbie p. , warwick r. , 2002 , apj , 568 , 610 fabian a. c. , rees m. j. , 1979 , mnras , 187 fan j. h. , cheng k. s. , zhang l. , liu c. h. , 1997 , a&a , 327 , 947 giannios d. , uzdensky d. a. , begelman m. c. , 2009 , mnras , 395 , 29 ghisellini g. , haardt f. , matt g , 2004 , a&a , 413 , 523 ghisellini g. , maraschi , l. , 1989 , apj , 340 , 181 ghisellini g. , et al . , 1997 , a&a , 327 , 61 ghisellini g. , maraschi , l. , tavecchio , f. 2009 , mnras , 396 , l105 giroletti m. , giovannini g. , taylor g. b. , falomo r. , 2004 , apj , 613 , 752 giroletti m. , giovannini g. , taylor g. b. , falomo r. , 2006 , apj , 646 , 801 gopal - krishna , singal , a.k . , krishnamohan , s. , 1984 , a&a , 140 , l19 gopal - krishna , subramanian k. , 1991 , nature , 349 , 766 gopal - krishna , stalin c. s. , sagar r. , wiita p. j. , 2003 , apj , 586 , l25 gopal - krishna , dhurde , s. , wiita p. j. , 2004 , apjl , 615 , l81 gopal - krishna , wiita p. j. , dhurde s. , 2006 , mnras , 369 , 1281 gopal - krishna , dhurde s. , sirkar p. , wiita p. j. , 2007 , mnras , 377 , 446 goyal a. , et al . , 2009 , mnras , 399 , 1622 goyal a. , gopal - krishna , wiita p. j. w. , anupama g. c. , sahu d. k. s. , sagar r. , 2011 , submitted guilbert p. w. , fabian a. c. , rees m. j. , 1983 , mnras , 205 , 593 gupta a. c. , fan j. h. , bai j. m. , wagner s. j. , 2008 , aj , 135 , 1384 hartman r. c. , et al .
, 1992 , apjl , 385 , 1 heidt j. , wagner s. j. , 1998 , a&a , 329 , 853 helmboldt j. f. , et al . , 2007 , apj , 658 , 203 impiombato d. , et al . , 2011 , apjs , 192 , 12 jannuzi b. t. , smith p. s. , elston r. , 1993 , apjs , 85 , 265 jorstad s. g. , marscher a. p. , mattox j. r. , wehrle a. e. , bloom s. d. , yurchenko a. v. , 2001 , apjs , 134 , 181 kataoka j. , et al .
, 2001 , apj , 560 , 659 kellermann k. i. , pauliny - toth i. i. k. , 1969 , apj , 155 , 71 kellermann k. i. , et al .
, 2004 , apj , 609 , 539 kharb p. , gabuzda d. , shastri p. , 2008
, mnras , 384 , 230 kidger m. r. , dediego , j. a. , 1990 , a&a , 227 , l25 kovalev y. y. , nizhelsky n. a. , kovalev y. a. , berlin a. b. , zhekanis g. v. , mingaliev m. g. , bogdantsov a. v. , 1999 , a&as , 139 , 545 kovalev y. y. , et al . , 2005 aj , 130 , 2473 kovalev y. y , et al . , 2009 apjl , 696 , 17 krawczynski h. , coppi p. s. , aharonian f. , 2002 , mnras , 336 , 721 krongold y. , binette l. , hernandez - ibarra f. , 2010 , apj , 724 l203 kundt w. , gopal - krishna , 1980 , nature , 288 , 149 kundt w. , gopal - krishna , 2004 , japa , 25 , 115 li h. z. , xie g. z. , yi t. f. , chen l. e. , dai h. , 2010 , apj , 709 , 1407 lister m. l. , homan d. c. , 2005 , apjl , 130 , 1389 lister m. l. , homan d. c. , kedler m. , kellermann k. i. , kovalev y. y. , ros e. , savolainen t. , zensus a. , 2009a , apjl , 696 , 22 lister m. l. , et al . , 2009b ,
aj , 138 , 1874 lobanov a.p .
, in `` steady jets and transient jets '' , bonn , 7 - 8 april 2010
. published in memorie della societa astronomica italiana , v.81 , n.4 ( 2010 ) magic collaboration ; albert j. , et al . , 2008 , science , 320 , 1752 marscher a. p. , 1996
, in blazar continuum variability , eds .
miller , j.r .
webb , and j.c .
noble , asp conf
. ser .
110 , ( san francisco : asp ) , p. 248 marscher a. p. , jorstad s. g. , mattox j. r. , wehrle a. e. , 2002 , apj , 577 , 85 marscher , a.p .
2008 , nature , 452 , 966 mihov b. , bachev r. , slavcheva - mihova l. , strigachev a. , semkov e. , petrov g. , 2008 , an , 329 , 77 monet d. g. , et al .
, 2003 , aj , 125 , 984 noble j. c. , 1995 , phd thesis , georgia state university perri m. , et al .
, 2003 , 407 , 453 petrucci p. o. , boutelier t. , henri g. , 2010 , preprint ( astroph/1010.5895 ) piner b. g. , edwards p. g. , 2004 , apj , 600 , 115 piner b. g. , pant n. , edwards p. g. , 2008 , apj , 678 , 64 pollock j. t. , webb j. r. , azarnia g. , 2007 , apj , 133 , 487 rani , b. , gupta , a.c . , joshi , u.c . , ganesh , s. , wiita , p.j . , 2010 , apj , 719 , l153 rector t. a. , gabuzda d. c. , stocke j. t. , 2003 , aj , 125 , 1060 romero g. e. , cellone s. a. , combi j. a. , 1999 , a&as , 135 , 477 romero g. e. , cellone , s. a. , combi j. a. , andruchow i. , 2002 , a&a , 390 , 431 sagar r. , 1999 , curr .
sci , 77 , 643 sagar r. , gopal - krishna , mohan v. , pandey a. k. , bhatt b. c. , wagner s.j . , 1999 ,
a&as , 134 , 453 sagar r. , stalin c. s. , gopal - krishna , wiita p. j. , 2004 , mnras , 348 , 176 sambruna r. m. , maraschi l. , urry c. m. 1996 , apj , 463 , 444 savolainen t. , homan d. c. , hovatta t. , kadler m. , kovalev y. y. , lister m. l. , ros e. , zensus j. a. , 2010 , a&a , 512 , 24 schlegel d. j. , finkbeiner d. p. , davis m. , 1998 , apj , 500 , 525 schlickeiser r. , 1996 , a&as , 120 , 481 spergel d. n. , 2007 , apjs , 170 , 377 stalin c. s. , gopal - krishna , sagar r. , wiita p. j. , 2004a , japa , 25 , 1 stalin , c. s. , gopal - krishna , sagar , r. , wiita , p. j. , 2004b , mnras , 350 , 175 stalin c. s. , gupta a. c. , gopal - krishna , wiita p. j. , sagar r. , 2005 , mnras , 356 , 607 stetson p. b. , 1987 , pasp , 99 , 191 stocke j. t. , foltz c. b. , weymann r. j. , christiansen w. a. , 1984 , apj , 280 , 476 stockman h. s. , moore r. l. , angel j. r. p. , 1984
, apj , 279 , 485 takalo l. o. , sillanp a. , nilsson k. , 1994 , a&as , 107 , 497 stocke j. t. , morris s. l. , weymann r. j. , foltz c. b. , 1992 , apj , 396 , 487 tanihata c. , urry c. m. , takahashi t. , kataoka j. , wagner s. j. , madejski g. m. , tashiro m. , kouda m. , 2001 , apj , 563 , 569 taylor g. b. , et al . , 2007 apj , 671 , 1355 treves a. , falomo r. , uslenghi m. , 2007 , a&a , 473 , 17 urry c. m. , padovani p. , 1995 ,
pasp , 107 , 803 vron - cetty m .- p . ; vron p. , 2006
, a&a , 455 , 773 visvanathan n. , wills b. j. , 1998 , aj , 116 , 2119 wagner s. j. , witzel a. , 1995 , ara&a , 33 , 163 wagner s. j. , et al . , 1996 , aj , 111 , 2187 webb w. , malkan m. , 2000 , apj , 540 , 652 weekes t. c. , 2008 , in high energy gamma - ray astronomy , ed . f. a. aharonian , w. hofmann & f. rieger , aip conf .
, 1085 , 3 wehrle a. e. , morabito d. d. , preston r. a. , 1984 , apj , 89 , 336 wiita p. j. , 2006 , in blazar variability workshop ii : entering the glast era asp conference series , vol .
350 , eds .
r. miller , k. marshall , j. r. webb , and m. f. aller .
( san francisco : asp ) , p. 183
wills b. j. , wills d. , breger m. , antonucci r. r. j. , barvianis r. , 1992 , apj , 398 , 454 xie g. z. , li k. h. , bai j. m. , dai b. z. , liu w. w. , zhang x. , xing s. y. , 2001 , apj , 548 , 200 ccccccccccccccc + iau name & other name & type & r.a.(j2000 ) & dec(j2000)&_b _ & @xmath148 & z & @xmath149&@xmath150&@xmath151 & @xmath152 & @xmath153 & @xmath154 & @xmath155 + & & & ( h m s ) & ( @xmath156 @xmath157 @xmath158 ) & & & & & & ( w / hz ) & ( w / hz ) & & & ( max ) + ( 1 ) & ( 2 ) & ( 3 ) & ( 4 ) & ( 5 ) & ( 6 ) & ( 7 ) & ( 8) & ( 9 ) & ( 10 ) & ( 11 ) & ( 12 ) & ( 13 ) & ( 14 ) & ( 15 ) + + j0222@xmath54302&3c 66a & ibl & 02 22 39.6 & @xmath543 02 08 & 15.71 & @xmath1024.96 & 0.444 & 16.8@xmath159 & @xmath100.20@xmath160 & @xmath161@xmath162 & @xmath163&-0.22 & 2.98 & 5.8@xmath164 + j0721@xmath57120&s5 0716 + 71 & ibl & 07 21 53.3 & @xmath571 20 36 & 15.50 & @xmath1024.73 & 0.310 & 29.0@xmath165 & @xmath50.36 & @xmath166@xmath167 & @xmath168 & 0.36 & 2.43 & 10.07@xmath169 + j0809@xmath55218&1es 0806 + 524 & hbl & 08 09 49.2 & @xmath552 18 58 & 15.30 & @xmath1023.25 & 0.138 & & @xmath100.25 & @xmath170@xmath171 & @xmath172 & 0.30 & 1.79 & + j1015@xmath54926&gb 1011 + 496 & hbl & 10 15 04.2 & @xmath549 26 01 & 16.56 & @xmath1022.88 & 0.200 & & @xmath100.26 & @xmath173@xmath174 & @xmath175 & 0.03 & 2.59 & + j1221@xmath53010&pg 1218 + 304 & hbl & 12 21 21.9 & @xmath530 10 37 & 16.50 & @xmath1022.72 & 0.182 & 6.6@xmath176 & @xmath100.08 & @xmath177@xmath171 & @xmath178 & 0.58 & 1.85 & + j1221@xmath52813&on 231 & hbl & 12 21 31.7 & @xmath528 13 58 & 16.81 & @xmath1021.21 & 0.102 & 4.3@xmath176 & @xmath100.09@xmath160 & @xmath179@xmath167 & @xmath180 & 0.70 & 3.01 & 3.2@xmath181 + j1256@xmath100547&3c 279 & fsrq & 12 56 11.1 & @xmath1005 47 21 & 18.01 & @xmath1023.24 & 0.538 & 44.0@xmath165 & @xmath50.48@xmath160 & @xmath182@xmath167 & @xmath183 & 0.50 & 4.51 & 20.58@xmath169 + j1428@xmath54240&h 1426 + 428 & hbl & 14 28 32.7 & @xmath542 40 21 & 16.95 & @xmath1021.60 & 0.129 & 2.5@xmath184 & @xmath100.58 & @xmath185@xmath171 & @xmath186 & 0.34 & 1.96 & 2.09@xmath187 + j1555@xmath51111&pg 1553 + 11 & hbl & 15 55 43.1 & @xmath511 11 24 & 15.00 & @xmath1025.42 & 0.360@xmath188&3.5@xmath189 & @xmath50.54 & @xmath190@xmath191 & @xmath192 & 0.88 & 1.85 & + + j0238@xmath51637&ao 0235 + 164 & lbl & 02 38 38.9 & @xmath516 37 00 & 16.46 & @xmath1025.47 & 0.940 & 43.9@xmath165 & @xmath50.71@xmath160 & @xmath193@xmath167 & @xmath194 & 0.80 & 3.12 & + j0423@xmath100120&pks 0420 - 01 & fsrq & 04 23 15.8 & @xmath1001 20 33 & 17.50 & @xmath1024.17 & 0.915 & 20.0@xmath165 & @xmath50.19@xmath160 & @xmath195@xmath167 & @xmath196 & 0.27 & 3.69 & 7.35@xmath169 + j0738@xmath51742&pks 0735 + 17 & lbl & 07 38 07.4 & @xmath517 42 19 & 16.76 & @xmath1024.04 & @xmath700.424 & 36.0@xmath165 & @xmath100.28@xmath160 & @xmath197@xmath167 & @xmath198 & 0.61 & 3.40 & 11.8@xmath199 + j0739@xmath50137&pks 0736 + 01 & fsrq & 07 39 18.0 & @xmath501 37 04 & 16.90 & @xmath1021.96 & 0.191 & 5.6@xmath200 & @xmath100.10@xmath160 & @xmath201@xmath167 & @xmath202 & 1.20 & 3.45 & 14.44@xmath169 + j0854@xmath52006&oj 287 & lbl & 08 54 48.8 & @xmath520 06 30 & 15.91 & @xmath1024.30 & 0.306 & 37.2@xmath165 & @xmath50.20@xmath160 & @xmath203@xmath167 & @xmath204 & 0.41 & 2.91 & 11.70@xmath169 + j1058@xmath50133&pks 1055 + 01 & fsrq & 10 58 29.6 & @xmath501 33 58 & 18.74 & @xmath1023.34 & 0.888 & 5.0@xmath184 & @xmath50.06@xmath160 & @xmath205@xmath167 & @xmath206 & 0.54 & 4.26 & 11.0@xmath169 + j1159@xmath52914&4c 29.45 & fsrq & 11 59 31.9 & @xmath529 14 45 & 14.80 & @xmath1027.00 & 0.729 & 28.0@xmath165 & @xmath100.34@xmath160 & @xmath193@xmath167 & @xmath207 & 0.57 & 2.60 & 24.8@xmath169 + j1217@xmath53007&b2 1215 + 30 & hbl & 12 17 52.0 & @xmath530 07 01 & 16.07 & @xmath1022.45 & 0.130 & 8.0@xmath176 & @xmath100.12@xmath160 & @xmath208@xmath171 & @xmath209 & 0.17 & 2.60 & + j1218@xmath100119&pks 1216 - 010 & bl@xmath210 & 12 18 34.9 & @xmath1001 19 54.0 & 16.17 & @xmath1025.14 & 0.554 & 6.9@xmath211 & @xmath50.01@xmath160 & @xmath212@xmath213 & @xmath214 & 0.22 & 3.01 & + j1310@xmath53220&b2 1308 + 32 & fsrq & 13 10 28.7 & @xmath532 20 44 & 15.61 & @xmath1026.69 & 0.997 & 28.0@xmath165 & @xmath100.09@xmath160 & @xmath215@xmath167 & @xmath216 & 0.49 & 2.71 & 27.2@xmath169 + j1512@xmath100906&pks 1510 - 08&fsrq & 15 12 50.5 & @xmath1009 06 00 & 16.74 & @xmath1023.49 & 0.360 & 7.8@xmath165 & @xmath100.10@xmath160 & @xmath217@xmath167 & @xmath218 & 1.35 & 3.48 & 20.2@xmath169 + j2225@xmath100457&3c 446 & fsrq & 22 25 47.2 & @xmath1004 57 01 & 18.83 & @xmath1023.66 & 1.404 & 17.3@xmath165 & @xmath100.13@xmath160 & @xmath219@xmath167 & @xmath220 & 0.34 & 4.58 & 17.34@xmath169 + j2253@xmath51608&3c 454.3 & fsrq & 22 53 57.7 & @xmath516 08 53 & 16.57 & @xmath1025.11 & 0.859 & 16.0@xmath165 & @xmath100.28@xmath160 & @xmath221@xmath167 & @xmath222 & 0.74 & 3.99 & 14.19@xmath169 + columns : ( 1 ) source name ; ( 2 ) most popular name as given in vron & vron ( 2006 ) ; ( 3 ) classification into low- , intermediate- , or high - frequency peaked blazars or flat spectrum radio quasar ( fsrq ) ; ( 4 ) right ascension ; ( 5 ) declination ; ( 6 ) apparent b - magnitude ; ( 7 ) absolute b - magnitude ; ( 8) redshift ; ( 9 ) the measured optical polarization ; ( 10 ) radio spectral index ; ( 11 ) radio core luminosity ; ( 12 ) extended emission radio luminosity ; ( 13 ) radio core dominance parameter @xmath33 ( see text ) ; ( 14 ) radio loudness @xmath40 ; ( 15 ) the fastest apparent speed observed in the parsec - scale jet ( in units of the speed of light ) . +
footnotes : + column 3 : abdo et al .
( 2010a ) , @xmath210 sed classification is not available .
+ column 8 : @xmath188 reference for redshift : rector et al .
( 2003 ) , see also , treves , falomo & uslenghi ( 2007 ) + column 9 : ( a ) takalo , silanp & nilsson ( 1994 ) ; ( b ) fan et al .
( 1997 ) ; ( c ) stockman , moore & angel ( 1984 ) ; ( d ) jannuzi , smith & elston ( 1993 ) ; ( e ) wills et al .
( 1992 ) ; ( f ) visvanathan & wills ( 1998 ) ; ( g ) andruchow , romero & cellone ( 2005 ) .
+ column 10 : @xmath160 is the radio spectral index derived using simultaneous flux measurements from kovalev et al .
( 1999 ) by means of least square method ( @xmath223 ) .
the remaining values of @xmath224 have been calculated using 6 and 20 cm .
flux densities from vron & vron ( 2006 ) .
+ column 11 : reference for vlbi flux : ( h ) marscher et al .
( 2002 ) ; ( i ) kovalev et al .
( 2005 ) ; ( j ) giroletti et al . ( 2006 ) ; ( k ) helmboldt et al . ( 2007 ) ; ( l ) wehrle , morabito & preston ( 1984 ) ; ( m ) rector et al . ( 2003 ) . + column 14 : @xmath40 is the _
k_-corrected rest frame ratio of the 5 ghz to 2500 flux densities , following stocke et al .
( 1992 ) ; reference for the radio flux is vron & vron ( 2006 ) .
+ column 15 : reference for @xmath155 : ( n ) britzen et al .
( 2008 ) ; ( o ) lister et al .
( 2009b ) ; ( p ) britzen et al .
( 2010 ) ; ( q ) kellermann et al .
( 2004 ) ; ( r ) piner et al .
+ j0222@xmath54302 & 1 & 14.11.98 & st & r & 6.5 & 118 & & 5.4 & 6.0 & v & & & ( a ) + & & 13.11.99 & st & r & 5.7 & 123 & & 5.5 & @xmath706.6 & v & & & ( a ) + & & 24.10.00 & st & r & 9.1 & 73 & & 4.3 & 5.8 & v & & & ( a ) + & & 26.10.00 & st & r & 10.1 & 82 & & 3.2 & 3.5 & v & & & ( a ) + & & 01.11.00 & st & r & 9.0 & 103 & & 2.2 & 2.9 & v & & & ( a ) + & & 24.11.00 & st & r & 5.1 & 71 & & & & n & & & ( a ) + & & 01.12.00 & st & r & 5.1 & 59 & & 8.0 & @xmath706.6 & v & & & ( a ) + & & 27.09.09 & st & r & 4.1 & 34 & 0.28 & 1.0 & 1.6 & n & 1.15(33 ) & n & ( b ) + & & 28.09.09 & st & r & 6.2 & 38 & 0.19 & 1.1 & 2.6 & v & 3.75(37 ) & v & ( b ) + & & & & & & & & & & & & & + j0238@xmath51637 & 2 & 27.10.90 & & v & 4.5 & & 1.00 & 5.0 & & v & & & ( c ) + & & 29.10.90 & & v & 8.3 & & 1.20 & 18.0 & & v & & & ( c ) + & & 05.11.91 & & r & 6.3 & & 1.20 & 20.0 & & v & & & ( c ) + & & 07.11.91 & & r & 8.0 & & 1.20 & 5.0 & & v & & & ( c ) + & & 08.11.91 & & r & 9.4 & & 1.30 & 16.0 & & v & & & ( c ) + & & 03.11.99 & & v & 6.7 & & 1.40 & 27.3 & 10.0 & v & & & ( d ) + & & 04.11.99 & & v & 6.6 & & 1.20 & 24.5 & 6.1 & v & & & ( d ) + & & 05.11.99 & & v & 7.0 & & 1.20 & 34.5 & 8.9 & v & & & ( d ) + & & 06.11.99 & & v & 6.7 & & 0.70 & 11.0 & 4.4 & v & & & ( d ) + & & 07.11.99 & & v & 6.6 & & 0.90 & 44.3 & 14.4 & v & & & ( d ) + & & 12.11.99 & st & r & 6.6 & 39 & & 12.8 & 6.6 & v & & & ( a ) + & & 14.11.99 & st & r & 6.2 & 34 & & 10.2 & 3.2 & v & & & ( a ) + & & 22.10.00 & st & r & 7.9 & 39 & & 7.6 & 2.6 & v & & & ( a ) + & & 22.12.00 & & v & 7.2 & & 0.70 & 7.0 & 3.3 & v & & & ( d ) + & & 18.11.03 & hct & r & 7.4 & 39 & 0.40 & 6.8 & @xmath705.54 & v & 58.71(38 ) & v & ( e ) + & & & & & & & & & & & & & + j0423@xmath100120 & 2 & 19.11.03 & st & r & 6.3 & 36 & 0.25 & 1.6 & 3.60 & v & 7.81(35 ) & v & ( e ) + & & 08.12.04 & st & r & 6.0 & 11 & 0.26 & 1.8 & 0.96 & n & 10.51(10 ) & v & ( e ) + & & 25.10.09 & st & r & 4.0 & 18 & 0.38 & 2.9 & 2.00 & pv & 6.45(17 ) & v & ( e ) + & & & & & & & & & & & & & + j0721@xmath57120 & 1 & 23.03.04 & & r & 7.0 & & & 6.0 & & v & & & ( f ) + & & 01.02.05 & st & r & 1.5 & 24 & 0.25 & 3.1 & 5.1 & v & 22.15(23 ) & v & ( b ) + & & 12.01.07 & & r & 3.6 & 185 & 0.70 & 6.3 & 2.8 & v & & & ( g ) + & & 20.03.07 & & r & 4.2 & 195 & 0.50 & & 1.6 & n & & & ( g ) + & & & & & & & & & & & & & + j0738@xmath51742 & 2 & 26.12.98 & st & r & 7.8 & 49 & & 1.8 & 1.13 & n & & & ( a)&(h ) + & & 30.12.99 & st & r & 7.4 & 64 & & 1.0 & 0.61 & n & & & ( a)&(h ) + & & 25.12.00 & st & r & 6.0 & 42 & & 1.6 & 1.02 & n & & & ( a)&(h ) + & & 25.12.01 & st & r & 7.3 & 43 & & 1.0 & 2.8 & v & & & ( a)&(h ) + & & 20.12.03 & hct & r & 5.8 & 36 & 0.23 & 1.0 & 1.78 & n & 1.86(35 ) & n & ( h ) + & & 10.12.04 & st & r & 5.8 & 28 & 0.16 & 1.3 & 3.00 & v & 8.40(27 ) & v & ( h ) + & & 23.12.04 & st & r & 5.0 & 11 & 0.10 & 1.2 & 3.10 & v & 17.83(10 ) & v & ( h ) + & & 02.01.05 & st & r & 4.9 & 20 & 0.22 & 0.8 & 0.97 & n & 1.42(19 ) & n & ( h ) + & & 05.01.05 & st & r & 5.8 & 23 & 0.15 & 1.0 & 2.25 & pv & 7.00(22 ) & v & ( h ) + & & 09.01.05 & st & r & 6.7 & 28 & 0.19 & 1.3 & 3.20 & v & 9.78(27 ) & v & ( h ) + & & 09.11.05 & st & r & 3.8 & 17 & 0.17 & 0.7 & 2.00 & pv & 3.35(16 ) & pv & ( h ) + & & 16.11.06 & st & r & 4.5 & 19 & 0.29 & 1.1 & 0.95 & n & 4.35(18 ) & v & ( h ) + & & 29.11.06 & st & r & 5.8 & 26 & 0.19 & 1.0 & 0.83 & n & 3.42(25 ) & v & ( h ) + & & 17.12.06 & st & r & 5.6 & 24 & 0.19 & 0.9 & 1.06 & n & 5.72(23 ) & v & ( h ) + & & 15.12.07 & st & r & 6.6 & 28 & 0.29 & 1.9 & 3.53 & v & 10.37(27 ) & v & ( h ) + & & 16.12.07 & st & r & 6.6 & 27 & 0.19 & 1.0 & 1.45 & n & 3.87(26 ) & v & ( h ) + & & 22.11.08 & st & r & 5.6 & 27 & 0.19 & 0.8 & 0.33 & n & 1.84(26 ) & n & ( h ) + & & & & & & & & & & & & & + j0739@xmath50137 & 2 & 05.12.05 & hct & r & 5.3 & 10 & 0.38 & 3.4 & 2.93 & v & 36.84(9 ) & v & ( e ) + & & 06.12.05 & hct & r & 6.0 & 09 & 0.54 & 3.1 & 3.50 & v & 6.23(8 ) & v & ( e ) + & & 09.12.05 & hct & r & 5.5 & 14 & 0.54 & 1.3 & 0.38 & n & 1.45(13 ) & n & ( e ) + & & & & & & & & & & & & & + j0809@xmath55214 & 1 & 04.02.05 & hct & r & 6.8 & 27 & 0.20 & 1.2 & 2.31 & pv & 1.86(26 ) & n & ( b ) + & & 05.12.05 & hct & r & 4.3 & 08 & 0.24 & 0.8 & 0.59 & n & 3.40(7 ) & n & ( b ) + & & 08.12.05 & hct & r & 4.9 & 14 & 0.11 & 0.4 & 0.15 & n & 1.45(13 ) & n & ( b ) + & & 09.12.05 & hct & r & 4.8 & 12 & 0.24 & 0.5 & 0.32 & n & 1.27(11 ) & n & ( b ) + & & & & & & & & & & & & & + j0854@xmath52006 & 2 & 29.12.98 & st & r & 6.8 & 19 & & 2.3 & 2.80 & v & & & ( a ) + & & 31.12.99 & st & r & 5.6 & 29 & & 3.8 & 6.50 & v & & & ( a ) + & & 28.03.00 & st & r & 4.2 & 22 & & 5.0 & 5.80 & v & & & ( a ) + & & 17.02.01 & st & r & 6.9 & 48 & & 2.8 & 2.70 & v & & & ( a ) + & & 05.02.05 & hct & r & 7.7 & 40 & 0.23 & 0.8 & 1.69 & n & 2.57(39 ) & v & ( b ) + & & 12.04.05 & st & r & 4.1 & 54 & 0.32 & 7.5 & 5.20 & v & 71.39(54 ) & v & ( b ) + & & & & & & & & & & & & & + j1015@xmath54926 & 1 & 06.02.10 & st & r & 5.5 & 24 & 0.25 & 1.0 & 0.98 & n & 2.08(23 ) & n & ( b ) + & & 19.02.10 & st & r & 5.6 & 41 & 0.30 & 1.8 & 2.17 & pv & 5.06(40 ) & v & ( b ) + & & 07.03.10 & st & r & 5.2 & 34 & 0.36 & 3.4 & 4.30 & v & 11.99(33 ) & v & ( b ) + & & & & & & & & & & & & & + j1058@xmath50133 & 2 & 25.03.07 & st & r & 5.8 & 11 & 0.12 & 2.0 & 3.20 & v & 58.56(10 ) & v & ( e ) + & & 16.04.07 & st & r & 3.8 & 15 & 0.17 & 0.8 & 0.53 & n &
1.37(14 ) & n & ( e ) + & & 23.04.07 & st & r & 4.4 & 10 & 0.23 & 1.8 & 2.57 & v & 12.51(9 ) & v & ( e ) + & & & & & & & & & & & & & + j1159@xmath52914 & 2 & 19.01.94 & & r & 4.0 & & 0.90 & 12.0 & & v & & & ( c ) + & & 20.01.94 & & r & 5.5 & & 0.80 & 3.0 & & v & & & ( c ) + & & 21.01.94 & & r & 5.1 & & 0.90 & 10.0 & & v & & & ( c ) + & & 22.01.94 & & r & 5.4 & & 1.10 & 11.0 & & v & & & ( c ) + & & 23.01.94 & & r & 5.1 & & 1.00 & 4.0 & & v & & & ( c ) + & & 24.01.94 & & r & 4.0 & & 1.30 & 7.0 & & v & & & ( c ) + & & & & & & & & & & & & & + j1217@xmath53007 & 2&20.03.99 & st & r & 7.0 & 21 & & 3.50 & 5.50 & v & & & ( a ) + & & 25.02.00 & st & r & 5.9 & 28 & & & & n & & & ( a ) + & & 31.03.00 & st & r & 5.0 & 27 & & & & n & & & ( a ) + & & 19.04.02 & st & r & 6.8 & 23 & & 1.80 & 4.90 & v & & & ( a ) + & & & & & & & & & & & & & + j1218@xmath100119 & 2 & 11.03.02 & st & r & 8.0 & 22 & & 7.3 & 3.20 & v & & & ( a ) + & & 13.03.02 & st & r & 8.5 & 24 & & 3.8 & 2.60 & v & & & ( a ) + & & 15.03.02 & st & r & 3.9 & 11 & & 5.5 & 3.50 & v & & & ( a ) + & & 16.03.02 & st & r & 8.2 & 22 & & 14.1 & @xmath705.54 & v & & & ( a ) + & & & & & & & & & & & & & + j1221@xmath53010 & 1 & 08.03.10 & igo & r & 6.2 & 15 & 0.24 & 0.9 & 1.18 & n & 1.96(14 ) & n & ( b ) + & & 18.03.10 & st & r & 4.7 & 25 & 0.49 & 1.0 & 0.26 & n & 2.12(24 ) & n & ( b ) + & & 22.05.10 & st & r & 3.9 & 19 & 0.88 & 3.3 & 0.41 & n & 2.23(18 ) & pv & ( b ) + & & & & & & & & & & & & & + j1221@xmath52813 & 1 & 19.03.04 & hct & v & 5.4 & 74 & 0.32 & 5.2 & @xmath705.54 & v & 16.65(72 ) & v & ( b ) + & & 20.03.04 & hct & v & 6.6 & 97 & 0.49 & 8.2 & @xmath705.54 & v & 62.42(96 ) & v & ( b ) + & & 18.03.05 & st & r & 4.0 & 26 & 0.33 & 2.0 & 1.74 & n & 3.21(25 ) & v & ( b ) + & & 05.04.05 & st & r & 6.9 & 38 & 0.20 & 3.2 & 4.50 & v & 27.71(37 ) & v & ( b ) + & & & & & & & & & & & & & + j1256@xmath100547 & 1 & 26.01.06 & st & r & 4.2 & 19 & 0.17 & 2.5 & @xmath705.54 & v & 13.59(18 ) & v & ( b ) + & & 28.02.06 & st & r & 6.1 & 40 & 0.25 & 10.2 & @xmath705.54 & v & 403.89(39 ) & v & ( b ) + & & 20.04.09 & st & r & 4.9 & 20 & 0.44 & 22.0 & @xmath705.54 & v & 1069.51(19 ) & v & ( b ) + & & & & & & & & & & & & & + j1310@xmath53220 & 2 & 26.04.00 & st & r & 5.6 & 16 & & & & n & & & ( a ) + & & 17.03.02 &
st & r & 7.7 & 19 & & 3.4 & 3.1 & v & & & ( a ) + & & 24.04.02 & st & r & 5.8 & 14 & & & & n & & & ( a ) + & & 02.05.02 & st & r & 5.1 & 15 & & & & n & & & ( a ) + & & & & & & & & & & & & & + j1428@xmath54240 & 1 & 21.04.04 & hct & v & 5.8 & 32 & 0.46 & 2.8 & 1.82 & n & 2.90(31 ) & v & ( b ) + & & 22.04.09 & st & r & 4.0 & 16 & 0.37 & 1.1 & 0.28 & n & 1.99(15 ) & n & ( b ) + & & 29.04.09 & st & r & 6.4 & 27 & 0.73 & 2.1 & 0.65 & n & 1.42(26 ) & n & ( b ) + & & & & & & & & & & & & & + j1512@xmath100906 & 2 & 28.04.98 & & v & 3.8 & & 0.30 & & & n & & & ( i ) + & & 29.04.98 & & v & 4.0 & & 0.40 & & & n & & & ( i ) + & & 30.04.98 & & v & 4.0 & & 0.80 & 4.2 & 1.45 & n & & & ( d ) + & & 06.06.99 & & v & 7.2 & & 0.90 & 3.4 & 1.05 & n & & & ( d ) + & & 07.06.99 & & v & 7.3 & & 0.90 & 4.2 & 1.27 & n & & & ( d ) + & & 14.06.05 & st & r & 4.0 & 09 & 0.12 & 1.6 & 1.53 & n & 12.02(8 ) & v & ( e ) + & & 01.05.09 & st & r & 5.6 & 22 & 0.31 & 4.7 & 2.63 & v & 29.85(21 ) & v & ( e ) + & & 20.05.09 & st & r & 4.8 & 23 & 0.43 & 3.1 & 0.98 & n & 4.20(22 ) & v & ( e ) + & & & & & & & & & & & & & + j1555@xmath51111 & 1 & 05.05.99 & st & r & 3.6 & 20 & & 2.3 & @xmath706.60 & v & & & ( i ) + & & 06.06.99 & st & r & 7.1 & 40 & & & & n & & & ( i ) + & & 24.06.09 & st & r & 3.8 & 23 & 0.17 & 4.2 & @xmath705.54 & v & 50.81(22 ) & v & ( b ) + & & 15.05.10 & st & r & 6.1 & 25 & 0.29 & 2.8 & @xmath705.54 & v & 12.99(24 ) & v & ( b ) + & & 16.05.10 & st & r & 6.1 & 31 & 0.30 & 2.3 & 4.63 & v & 6.33(30 ) & v & ( b ) + & & & & & & & & & & & & & + j2225@xmath100457 & 2 & 29.09.88 & & r & 4.8 & & 0.80 & 9.0 & & v & & & ( c ) + & & 01.10.88 & & r & 6.7 & & 0.90 & & & n & & & ( c ) + & & 08.10.10 & & r & 5.1 & 16 & 0.64 & 6.8 & 1.12 & n & 14.35(15 ) & v & ( e ) + & & & & & & & & & & & & & + j2253@xmath51608 & 2 & 12.09.90 & & r & 6.9 & & 1.00 & 6.0 & & v & & & ( c ) + & & 13.09.90 & & r & 7.5 & & 1.10 & 4.0 & & v & & & ( c ) + @xmath226
v= variable , n = non - variable , pv = probable variable + @xmath188 dof = degrees of freedom + @xmath210reference for inov data : ( a ) sagar et al .
( 2004 ) ; ( b ) present work ; ( c ) noble ( 1995 ) ; ( d ) romero et al .
( 2002 ) ; ( e ) goyal et al .
( 2011 ) ( f ) pollock , webb & azarnia ( 2007 ) ; ( g ) gupta et al .
( 2008 ) ; ( h ) goyal et al .
( 2009 ) ; ( i ) stalin et al . | in this first systematic attempt to characterise the intranight optical variability ( inov ) of tev detected blazars , we have monitored a well defined set of 9 tev blazars on total 26 nights during 2004 - 2010 . in this @xmath0 ( or @xmath1)-band monitoring programme only one blazar was monitored per night and the minimum duration was close to 4 hours , the average being 5.3 hours per night . using the ccd for strictly simultaneous photometry of the blazar and nearby reference stars ( n - star photometry ) , an inov detection threshold of @xmath2
12 per cent was achieved in the densely sampled differential light curves derived from our data .
we have further expanded the sample by including another 13 tev blazars , taking advantage of the availability in the literature of inov data , including those published earlier in our programme .
the selection criteria for this set of 13 blazars conform to the basic criteria we had adopted for the first set of 9 blazars we have monitored presently . this enlarged , well defined representative sample of 22 tev blazars , monitored on a total of 116 nights ( including 55 nights newly reported here ) , has enabled us to arrive at the first estimate of the inov duty cycle ( dc ) of tev detected blazars . applying the conservative , but commonly employed , @xmath3-test ,
the inov dc is found to be 59 per cent , which decreases to 47 per cent if only inov fractional amplitudes ( @xmath4 ) above 3 per cent are considered .
these observations also permit , for the first time , a comparison of the inov characteristics of the two major subclasses of tev detected bl lacs , namely lbls and hbls , for which we find the inov duty cycles to be @xmath2 63 per cent and @xmath2 38 per cent , respectively .
this demonstrates that the previously recognized inov differential between lbls and hbls persists even when only their tev detected subsets are considered . despite dense sampling
, the intranight light curves of the 22 tev blazars have not revealed even a single feature on time scale substantially shorter than 1 hour , even though the inner jets of tev blazars are believed to have exceptionally large bulk lorentz factors ( and correspondingly stronger time compression ) .
an intriguing feature , clearly detected in the light curve of the hbl j1555@xmath51111 , is a 4 per cent ` dip ' on a 1 hour timescale .
this unique feature could have arisen from absorption in a dusty gas cloud , occulting a superluminally moving optical knot in the parsec scale jet of this relatively luminous bl lacs object .
[ firstpage ] galaxies : active galaxies : jets bl lacertae objects : general galaxies : photometry |
the most widely studied bimodality in the galaxy properties is that observed for their colours , that is the clear separation of galaxies into the red sequence and blue cloud populations ( e.g. * ? ? ?
this has the advantage that it is easy to measure , particularly at high - redshifts , allowing studies to follow the evolution of the bimodality to @xcite , and demonstrating its existence at even @xcite .
the main drawback of using optical colours is that they do not necessarily fully relate to the underlying star - formation history , in particular galaxies can appear red not only because they are passive , but also through high levels of dust extinction produced by star - bursts .
a number of studies have followed separately the evolution of the red and blue galaxy luminosity functions to , in particular with regard to characterising the build - up of the red sequence population .
there is little evolution in the number density of massive red sequence galaxies with , with only a factor two increase in the stellar mass density from to the present day @xcite , or a factor six since @xcite . at lower stellar masses
however a much more rapid evolution is observed . by the stellar mass density of red galaxies with
has dropped by an order of magnitude , while by this cut - off mass has risen to @xcite .
the red sequence thus appears to be built up from the top down , with the most massive galaxies in place first at , and the faint end incrementally filled in at lower redshifts .
the epoch and rapidity over which the red sequence is assembled is found to be a function of environment .
@xcite follow the build - up of the red sequence as a function of environment at three redshifts , ( based on sdss data ) , and . in cluster environments the data are consistent with there being no significant growth of the red sequence since , except for a hint of evolution in the lowest mass bin between and .
similarly , @xcite find evidence for a deficit of faint red sequence galaxies in 18 ediscs clusters at with respect to present day clusters .
they find that the observed decrease in the numbers of faint red sequence galaxies are consistent with a simple model whereby these galaxies are formed as the result of blue galaxies in clusters at having their star - formation suppressed by the hostile cluster environment . in field environments the fraction of galaxies on the red sequence is lower than in clusters at all luminosities and all redshifts to @xcite , showing that the colour - density relation is already in place at .
there is also a deficit of faint red sequence galaxies in field regions , both at and at the present day , indicating that the assembly of the red sequence is still incomplete in low density environments @xcite . in (
* ; * ? ? ?
* papers i and ii ) we presented an analysis of star - formation in galaxies as a function of both luminosity and environment in order to constrain the physical mechanisms that drive the star - formation histories of galaxies of different masses . in paper
i we used the fourth data release of the sloan digital sky survey ( sdss dr4 ; * ? ? ?
* ) to investigate the possible mass dependency of the sf - density and age - density relations in the vicinity of the supercluster centred on the rich cluster abell 2199 . for giant galaxies we found gradual age - density and sf - density trends extending to the lowest densities studied , with the clusters dominated by old , passively - evolving galaxies while in field regions we found equal fractions of old , passively - evolving and young , star - forming galaxy populations which were completely interspersed .
in contrast for the dwarf galaxy population we found a sharp transition from the virialized regions of clusters and groups which were still dominated by old , passively - evolving galaxies , to outside where virtually all dwarf galaxies were young with ongoing star - formation .
the few old , passively - evolving dwarf galaxies outside of the clusters were always found to reside in poor groups or as a satellite to a massive galaxy . in paper ii we extended the analysis to the entire sdss dr4 , using a sample of 27753 galaxies in the redshift range that is % complete to . in high - density regions we found 70% of galaxies to be passively - evolving independent of luminosity . in the rarefied field
however , the fraction of passively - evolving galaxies was found to be strongly luminosity - dependent , dropping from 50% for to zero by .
indeed for the lowest luminosity bin studied none of the galaxies in the lowest density quartile were found to be passive . throughout the sdss dr4 dataset we found no passively - evolving dwarf galaxy more than two virial radii from a massive halo , whether that be a cluster , group or massive galaxy .
these results imply fundamental differences in the formation and evolution of giant and dwarf galaxies .
recently a theoretical framework has been proposed whereby the differences are related to : ( i ) the increasing star - formation efficiencies and decreasing gas consumption time - scales with galaxy mass resultant from the kennicutt - schmidt law @xcite ; ( ii ) the way that gas from the galaxy halo cools and flows onto the galaxy @xcite and which affects its ability to maintain star - formation over many gyr ; as well as ( iii ) agn feedback which can effectively permanently shut down star - formation in massive galaxies @xcite .
the lack of passively - evolving dwarf galaxies in isolated field regions implies that _ internal _ processes , such as agn feedback , merging or gas consumption through star - formation , are not able to completely shut down star - formation in these galaxies
. instead star - formation in dwarf galaxies is only terminated once they become satellites in massive halos , probably through the combined effects of tidal forces and ram - pressure stripping . in this article
we examine what consequences these differences have on the form of the red sequence in the local universe , and its variation from within the high - density environments of clusters and groups , to the lowest density regions of the rarefied field . in @xmath1 [ data ] we describe the dataset used , that is the same spectroscopic sample as in paper ii , which allows us to classify the galaxies as passive , star - forming or agn , as well as fully define the local environment of each galaxy . in @xmath1 [ redseq ]
we examine the extent and make - up of the red sequence as a function of environment , based on the colour - magnitude diagram .
in particular we examine the issue of the contamination of the red sequence by dusty star - forming galaxies as found by @xcite and @xcite , and which are likely to represent a major problem for high - redshift samples . to resolve these problems in @xmath1 [ galex ] we take advantage of the significant coverage of the sdss dr4 footprint by near - ultraviolet ( nuv ) imaging from the _ galex _ public archive , to obtain colour - magnitude diagrams in both cluster and field environments .
we show that this provides a much cleaner separation of passive and star - forming galaxies , and find that in isolated field regions the red sequence appears truncated at , confirming the lack of passively - evolving dwarf galaxies in these rarefied environments .
we also examine the issue of aperture bias on our classification of galaxies , as our spectroscopic sample covers the redshift range for which aperture biases are known to be important , at least for massive galaxies @xcite . finally in @xmath1 [ discussion ] we present our discussion and conclusions . throughout
we assume a concordance @xmath2cdm cosmology with , and .
the galaxy sample is the same as that used and described in paper ii , consisting of 27753 galaxies in the redshift range from the low - redshift subsample @xcite from the new york university - value added catalogue ( nyu - vagc ; * ? ? ?
this is a subsample of galaxies from the sdss dr4 spectroscopic dataset for which additional checks have been made for badly deblended objects and absolute magnitudes have been calculated using version 3.2 of the software k - correct @xcite .
we consider only those galaxies belonging to the two large contiguous regions in the north galactic cap , excluding the three narrow stripes with .
we use the stellar indices of the mpa / jhu sdss dr4 catalogues @xcite , in which the emission - line fluxes are corrected for stellar absorption using a continuum fitting code based on the @xcite population synthesis models @xcite . in paper ii
we showed that the distribution of the h@xmath3 equivalent widths of galaxies is strongly bimodal , allowing them to be robustly separated into passively - evolving and star - forming populations about a value .
additionally we identify agn using the versus diagnostic diagrams of @xcite as those galaxies lying above the @xmath4 lower limit of the models defined by @xcite .
when either the [ oiii ] or h@xmath5 lines are unavailable ( @xmath6 ) , the two - line method of @xcite is used , with agn identified as having . the local environment of each galaxy is quantified using a variant of the adaptive kernel estimator @xcite , whereby each galaxy is represented by a gaussian kernel in redshift - space , @xmath7 , of width 500kms@xmath8 in the radial direction , and whose transverse width is defined by the distance to its third nearest neighbour within 500kms@xmath8 .
the choice both of the method and of the kernel dimensions is designed to resolve the galaxy s environment on the scale of its host dark matter halo , as it is the mass of a galaxy s host halo and whether the galaxy is the central or a satellite galaxy , that is believed to be the dominant factor in defining its global properties such as its star - formation history or morphology ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . in the case of galaxies within groups or clusters ,
the local environment is measured on the scale of their host halo , while for galaxies in field regions the local density is estimated by smoothing over its 510 nearest neighbours or scales of 15mpc . in paper
ii we tested the efficiency of this density estimator by applying it to the public galaxy catalogues from the millennium simulations @xcite , finding it very sensitive to the presence of even very poor groups comparable to the local group containing as few as four galaxies , which represent the preferred mass - scale of major mergers for galaxies of stellar mass @xcite . by selecting galaxies with a pure field sample
is produced with no contamination from group members .
in contrast 60% of and 90% of galaxies lie within the virial radius of a galaxy group or cluster .
the high quality photometric and spectroscopic data of the sdss have allowed thorough analyses of the bivariate colour - magnitude distribution of galaxies to be determined , demonstrating its bimodal nature , particularly in terms of galaxy colours ( e.g. * ? ? ?
* ; * ? ? ?
@xcite show that the colour distribution of galaxies in any given magnitude bin is well described by double gaussians for the entire magnitude range covered , allowing galaxies to be split into red and blue populations .
the relative contribution of the red sequence population is found to increase with luminosity / mass , as well as increasing local density @xcite .
the relative simplicity of observing this bimodal colour - magnitude distribution of galaxies has allowed it to be followed out to @xcite . to relate these observational results to theoretical predictions for the growth and evolution of galaxies
it is broadly assumed that red sequence galaxies are passively - evolving galaxies dominated by old stellar populations formed at , while blue galaxies are star - forming galaxies whose optical emission is dominated by young stellar populations .
the main uncertainty in such a model is that the optical colours of galaxies do not necessarily fully relate to the underlying star - formation history as described above . in particular galaxies
can appear red both because they are passive , but also through high levels of dust extinction produced by star - bursts .
a much more reliable measure of the recent star - formation history of galaxies than their optical colour can be made from their spectra , in particular the level of h@xmath3 emission . in figure [ u_r_makeup ]
we look at how the colour - magnitude diagram is broken up into ( a ) passively - evolving , ( b ) star - forming and ( c ) agn components as determined from their spectra . in each panel
the black contours indicate the _ global _ bivariate density distribution of galaxies in colour - magnitude space .
the black contours are spaced logarithmically , such that the galaxy density doubles every two contours . over this redshift range the sdss spectroscopic sample
is expected to be complete to , the main source of incompleteness being due to fibre collisions , but at fainter magnitudes galaxies will only be observable within a certain redshift range , which in some cases may be much less than overall redshift range of the sample . to correct for this issue , we weight each galaxy by v@xmath9 where v@xmath10 is the maximum volume over which the galaxy could be observed within the survey volume , v@xmath11 . for each bin in colour and m@xmath12 magnitude
we calculate the relative contribution of passively - evolving galaxies ( ; not agn ) , star - forming galaxies ( ; not agn ) and agn , in panels ( a - c ) respectively . for each bin in the colour - magnitude diagram containing at least one galaxy , the _ global _ fraction of passively - evolving galaxies
is indicated both by the numeric value , and the colour shading of the box , with increasingly intense red colours indicating higher passive galaxy fractions in that bin .
panels ( b ) and ( c ) show respectively the contributions of star - forming galaxies ( blue - shaded boxes ) and agn ( green - shaded boxes ) .
the bimodality of galaxy properties in colour - magnitude space is clear , with a population of red galaxies that extends over the full magnitude range covered forming the red sequence , and a second population of blue galaxies that show a greater dispersion in colour , but tend to be less luminous in general than their red counterparts .
panel ( a ) shows that passively - evolving galaxies are well confined to the red sequence , with virtually no passive galaxies showing blue optical colours with .
the c - m relation of red sequence galaxies shows the well known slope due to metallicity effects .
we apply the method of @xcite to estimate the slope and width of the c - m relation , using the biweight algorithm @xcite to estimate the dispersion about the relation , and varying the slope to minimize the biweight scale of the deviations about the median ( see e.g. * ? ? ?
* ) . considering just the passively - evolving galaxies we obtain a best - fitting relation of : @xmath13 ( shown by the red solid lines in panels a - c ) , and a width @xmath14 .
the best fit closely follows the main axis of the maximum in the galaxy density contours due to the red sequence population , demonstrating the validity in using only the passive galaxies to obtain the fit .
we identify red sequence galaxies as those which have colours redder than the black dashed line , which corresponds to a relation bluer than the red sequence , and where equal numbers of passive and star - forming galaxies are observed when averaged over all environments . in panel
( b ) we see that virtually all blue galaxies are identified as star - forming from their spectra , and hence to a first approximation the correlation between galaxy colour and the current star - formation rate appears good .
however it is also notable that a significant fraction of galaxies with red optical colours are in fact star - forming , particularly at the faint end , where 53% of red sequence galaxies ( i.e. above the dashed line ) are spectroscopically classed as star - forming .
finally in panel ( c ) we see that agn are generally red and confined to the most luminous galaxies in the sample where they make up of the galaxy population , this fraction falling to close to zero by as shown in paper ii .
hence , while passive galaxies are red , _ red galaxies are not necessarily passive .
_ in an analysis of the sdss main sample galaxies covered by infrared imaging from the swire survey , @xcite find that 17% of red sequence galaxies are dusty , star - forming galaxies ( identified by their high 24@xmath15 m to 3.6@xmath15 m flux ratios and h@xmath3 emission ) , while @xcite find that dusty , star - forming galaxies make up more than one - third of the red sequence population in the the a901/2 supercluster system at @xmath16 , finding them preferentially in the medium - density outskirts of the clusters .
given that the relative fractions of passively - evolving / red and star - forming / blue galaxies of a given luminosity / stellar mass are strongly dependent on local environment , particularly at the faint end @xcite , the contamination of the red sequence by star - forming galaxies is also likely to be strongly dependent on environment . in fig .
[ makeup_rho ] we show how the optical colour - magnitude diagram and the relative contribution from star - forming galaxies changes from field (; left panel ) to cluster (; right panel ) environments .
the contours in each panel indicate the bivariate volume - corrected number density in colour - magnitude space as in fig .
[ u_r_makeup ] , this time for field and cluster galaxies respectively .
the bivariate distribution of field galaxies is similar to that for the global population , mainly as the majority ( 64% ) of galaxies are found in field regions .
the main difference is that in field regions the red sequence is less well populated , particularly at faint magnitudes . in cluster regions , the bivariate colour - magnitude distribution of galaxies
is instead dominated by the red sequence , which can be clearly seen to extend to the completeness limit of the sample at , the `` blue cloud '' appearing now merely as a tail of galaxies extending bluewards from the red sequence .
looking now at the fraction of red sequence galaxies which are in fact spectroscopically classed as star - forming , it is clear that the contamination of dusty star - forming galaxies is much greater in field regions than in clusters , and in all environments the contamination is greater at faint magnitudes .
it is also notable that most galaxies that are above the red sequence are in fact star - forming , their red colours simply due to dust extinction .
we summarize these trends in table [ fracs ] which shows the fraction of red sequence galaxies ( i.e. above the red dashed lines in figures [ u_r_makeup ] and [ makeup_rho ] ) which are spectroscopically classed as star - forming as a function of both environment and luminosity .
globally , we find that 31% of red sequence galaxies ( with ) are star - forming , similar to that found by @xcite , but larger than the 17% obtained by @xcite .
this difference is most likely due to their sample being biased towards more luminous galaxies than ours , and in fact we find just 18% of galaxies to be star - forming .
.fraction of red sequence galaxies classified as star - forming as a function of both environment and luminosity [ cols="^,^,^,^ " , ] in paper ii we found that the fraction of passively - evolving galaxies is a strong function of both luminosity / stellar mass and local environment .
in particular we found that in low - density regions corresponding to field environments well outside the influence of clusters and groups , the fraction of passively - evolving galaxies is a strong function of luminosity , dropping from for galaxies to zero by .
moreover , those few passively - evolving galaxies outside groups and clusters were invariably found to be satellites to massive galaxies .
this was put forward as strong evidence that dwarf galaxies do not become passive through internal mechanisms , but only through environment - related processes after they become satellites within massive halos . in figure [ galexcm ]
we compare the colour - magnitude diagrams of galaxies in field (;
left panel ) and cluster (; right panel ) environments . in paper
ii we found that the few passively - evolving dwarf galaxies in field regions were almost always satellites to massive galaxies , and that throughout the sdss dr4 dataset there were no passively - evolving dwarf galaxies more than two virial radii from a massive halo , whether that be a cluster , group or massive galaxy . hence to confirm this result we remove those field galaxies within a projected distance of 400kpc and a radial velocity within 300kms@xmath8 of a galaxy .
these comprise around 10% of the total field galaxy population .
our resultant field galaxy sample thus only contains galaxies that are unlikely to have had any encounter with a massive halo in their past .
the nuv - optical red sequence should then represent only those galaxies which have become passive through internal processes such as merging , agn feedback and gas exhaustion through star - formation .
we separate galaxies into star - forming ( shown as blue circles ) and passively - evolving ( red squares ) galaxies , to measure the correspondence between the current star - formation rate and the colour of galaxies . galaxies with optical agn signatures
are indicated by green triangles ( for those with or magenta triangles . in cluster regions (; right panel )
there is a clear nuv - optical red sequence dominated by passively - evolving galaxies that extends to at least the completeness limit of the survey .
there appears to be a break in the red sequence at , with massive red sequence galaxies having similar colours while fainter galaxies become increasingly blue .
similar trends have been seen by @xcite who find the dichotomy between giant and dwarf red sequence galaxies to be even stronger when using the fuv - optical or fuv - nir colours .
@xcite in contrast find no evidence of a break in the nuv - optical red sequence , fitting it instead by a straight line for .
fitting separate c - m relations to bright and faint passively - evolving galaxies we obtain slopes of @xmath17 and @xmath18 respectively , and hence the break appears significant .
considering now a c - m relation with break at the overall dispersion is reduced slightly to , with no significant difference between bright and faint magnitudes . at all luminosities
we see an extended tail of galaxies with bluer colours , being mostly agn at the bright - end and star - forming galaxies at fainter magnitudes .
although many of these galaxies have corresponding to the peak of the blue cloud distribution seen in fig .
[ nuv_r_makeup ] , a significant fraction are found with transitional colours being in the `` green valley '' with .
we see no obvious separation between the red sequence and the blue cloud populations in cluster environments . in isolated field regions
however , the dominant feature of the colour - magnitude diagram becomes the blue cloud population at , made up almost entirely of star - forming galaxies , except at the very bright tip where agn take over .
the red sequence is now sparsely populated at all magnitudes and appears truncated below . the galaxy luminosity function , which describes the number of galaxies per unit volume as a function of luminosity , is a powerful tool to constrain galaxy evolution and transformations , since it is directly related to the galaxy mass function . in particular , by separating galaxies by their colour into the red sequence and blue cloud populations , and obtaining the type - specific luminosity functions , the growth of the red sequence and the transformation of blue to red galaxies can be followed .
in figure [ lfs ] we plot the red and blue volume - corrected galaxy luminosity functions ( as red and blue symbols respectively ) in both field and cluster environments , based on the galaxies shown in fig .
[ galexcm ] , where the dashed line is used to separate the red and blue galaxy populations .
we correct for incompleteness at faint magnitudes as before by weighting each galaxy by v@xmath19 . for each luminosity function
we determine the best - fitting single @xcite function based on a maximum - likelihood analysis , shown by the black curves .
the resultant best - fit parameters are presented in table [ lumparams ] , where the reported errors represent the 68.3% confidence limits in @xmath3 and m@xmath20 .
the uncertainties in @xmath3 and m@xmath20 are expected to be correlated , and in figure [ lfparams ] we present the 1 , 2 and 3@xmath21 confidence limits in m@xmath22 and @xmath3 defined as the regions containing 68.3 , 95.4 and 99.7% of the probability of finding the parameters within the contours .
the filled and empty contours correspond respectively to the galaxy luminosity functions in field and cluster environments .
we see little sign of variations in the luminosity function of blue galaxies from field to cluster environments , with marginal evidence for an increase in the luminosity of m@xmath20 from field to cluster environments , but no difference in the faint - end slope .
in contrast the luminosity function of the red sequence populations in field and cluster environments are inconsistent at level . in particular
, the red sequence luminosity function in field regions peaks at and drops rapidly at magnitudes fainter than , with no red galaxies at .
this decline results in a very shallow faint - end slope , and we are also able to fit the red sequence field galaxy population equally well by a single gaussian function with and as shown by the dashed curve in fig [ lfs ] .
although we are limited by the small number of red galaxies in our sdss - galex field sample , it is possible that the red sequence is completely truncated below in these rarefied environments .
ccccc lf params & @xmath3 & m@xmath23 & @xmath24 + + red & @xmath25 & @xmath26 & 49.3 + blue & @xmath27 & @xmath28 & 163.2 + + red & @xmath29 & @xmath30 & 135.9 + blue & @xmath31 & @xmath32 & 110.3 + .
the red contours indicate the 1 , 2 and confidence limits for the parameters @xmath3 and m@xmath12 for the @xmath0-selected red sequence lf , while the blue contours indicate the corresponding confidence limits for the blue cloud population .
the filled and empty contours correspond respectively to the galaxy luminosity functions in field and cluster environments .
, width=302 ] one of the major concerns of using the fibre - obtained sdss spectra for nearby galaxy samples , such as those used in papers
i and ii , is the effect of aperture biases , due to the spectra being obtained through 3 arcsec diameter apertures rather than over the full extent of the galaxy .
significant radial star - formation gradients are possible within galaxies , particularly those undergoing nuclear star - bursts or spiral galaxies with prominent passively - evolving bulges , that can result in the `` global '' star - formation rate extrapolated from spectra containing flux from only the galaxy nucleus being significantly over or underestimated .
@xcite indicate that star - formation rates based on spectra obtained through apertures containing less than of the integrated galaxy flux can be over or underestimated by a factor , and to ensure the sdss fibres sample more than 20% require galaxies to be at . clearly in order to use the sdss dataset to study star - formation in dwarf galaxies this is not possible , as at @xmath33 they are already too faint to be included in the sdss spectroscopic sample .
-band ( a ) and nuv - band ( b ) images of a bright face - on early type spiral from our sample , highlighting the effects of the fibre apertures on classifying galaxies as passive from their sdss spectra .
the red circles indicate the size of the 3@xmath34 diameter apertures of the sdss spectroscopic fibres .
( c ) the fraction of galaxies as a function of absolute magnitude that are classified as passive from their sdss spectra and also have blue uv - optical integrated colours ( ) indicative of recent star - formation.,width=302 ] -band ( a ) and nuv - band ( b ) images of a bright face - on early type spiral from our sample , highlighting the effects of the fibre apertures on classifying galaxies as passive from their sdss spectra .
the red circles indicate the size of the 3@xmath34 diameter apertures of the sdss spectroscopic fibres .
( c ) the fraction of galaxies as a function of absolute magnitude that are classified as passive from their sdss spectra and also have blue uv - optical integrated colours ( ) indicative of recent star - formation.,width=302 ] in papers i and ii we are primarily interested in the simple classification of galaxies into passive and star - forming , and so the main issue is the number of early - type spiral galaxies which may appear passive from spectra that sample only their bulge , but have also normal star - forming disks .
as demonstrated earlier , the integrated colour gives a robust separation of passive and star - forming galaxies .
hence we can quantify the level of aperture bias as the fraction of galaxies that are classified as `` passive '' from their spectra by having , yet have global uv - optical colours indicative of star - forming galaxies , defined as those with . in figure [ galaxy ]
we show the @xmath35band ( panel a ) and nuv - band ( panel b ) images of a typical misclassified galaxy from our sample ( @xmath36 , ) , with the apertures used to obtain the sdss spectra indicated by the red circles .
the galaxy is a clearly face - on early - type spiral , whose @xmath37-band flux is dominated by a bulge , but also with clear spiral arms . in the nuv - band image however , this bulge disappears almost completely , while the extended uv - flux from star - formation in the disk and spiral arms is now dominant .
the reason for the misclassification is clear , the sdss aperture covers only the central bulge which is passively - evolving as apparent from the `` hole '' in the uv - emission in the nuclear regions , yet misses entirely the extended star - forming regions from the outer disk and spiral arms . in panel
( c ) we show the fraction of galaxies misclassified due to aperture effects as a function of absolute magnitude .
we find that the level of misclassification due to aperture effects is strongly luminosity dependent , dropping from eight per cent at ( 20 out of 246 ) to zero for galaxies .
we indicate that aperture biases are significant for galaxies , where seven per cent are spectroscopically classified as passive , yet whose blue integrated colours are indicative of recent star - formation .
as expected many of these galaxies appear as face - on spiral galaxies with prominent bulges , whose predominately old and passive stellar populations dominate the flux within the sdss spectral apertures . in contrast
we find that none of the 1375 galaxies in the sdss sample covered by galex nuv photometry are misclassified , indicating that the classification of such low - luminosity galaxies based on their h@xmath3 emission as measured through the sdss fibres is robust against aperture biases
. we can understand this luminosity dependence for the level of aperture bias as the combination of two effects : ( i ) more luminous galaxies at the same distance will have larger apparent sizes , and so the fraction of flux covered by the sdss fibres will be reduced ; and ( ii ) low - luminosity galaxies tend to be either late - type spirals or dwarf ellipticals and hence do not have such significant radial gradients in their star - formation rates .
the luminosity function of early - type spirals ( sa+b ) for which aperture biases are by far the most important has a gaussian distribution centred at and width @xcite , and hence are rare at .
we have combined _ galex _ nuv photometry from the 3rd data release with a volume - limited sample of local sdss dr4 galaxies to examine the make - up and the environmental dependencies of the optical and uv - optical colour - magnitude ( c - m ) diagrams .
we consider the same sdss dr4 spectroscopic sample of 27753 galaxies in the redshift range as used in paper ii , that is complete to . from this sample ,
4065 galaxies are also covered by nuv photometry from the 490.1deg@xmath38 area covered by both the sdss dr4 spectroscopic survey and _ galex _ gr3 .
we find that of red sequence galaxies in the optical c - m diagram show signs of ongoing star - formation from their spectra having .
this contamination is greatest at faint magnitudes and in field regions where as many as three - quarters of red sequence galaxies are star - forming .
this has important consequences for understanding the build - up of the red and blue sequences being interpreted as the hierarchical assembly of star - forming and passively - evolving galaxies , as the colour of the galaxy can not be always reliably related to its star - formation history .
instead a significant fraction ( and in some cases the majority ) of galaxies on the red sequence have star - formation histories more in common with the blue cloud population and appear red simply due to dust , the presence of which is directly related to them having active ongoing star - formation .
the effect of the contamination of the faint end of the red sequence by dusty star - forming galaxies is likely to become increasingly important at higher redshifts , where the global star - formation rates @xcite , and hence the effects of dust extinction are that much higher .
this produces a significant overestimation of the amount of stellar mass already in the red sequence at a given epoch , particularly at the low mass end , and an overestimation of the look - back time by which the stellar mass is assembled in passively - evolving galaxies .
similarly , the amount of stellar mass in the blue sequence or star - forming galaxies will be underestimated , as will the global amount of star - formation in the blue sequence .
these effects could bias the interpretation of studies looking at the evolution of the optically - defined red and blue sequences with redshift in terms of the hierarchical assembly and conversion of accreted gas into stars in passive and star - forming galaxies ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , the result of which could be the over - estimation of the importance of dry mergers to create the present day population of passively - evolving galaxies .
we find that instead the colour allows a much more robust separation of passively - evolving and star - forming galaxies , with a clear red sequence of passively - evolving galaxies at and a well separated blue sequence of star - forming galaxies at .
we find that globally only 8% of uv - selected red sequence galaxies are star - forming , i.e. one quarter of the contamination seen in the optical red sequence .
this robust separation allows the build - up of the uv - selected red sequence with redshift and environment to be directly interpreted in terms of the assembly of stellar mass in passively - evolving galaxies ( see also * ? ? ?
* ; * ? ? ?
* ) . in cluster and group environments
we find the uv - selected red sequence to be fully in place to at least , being well described by a single schechter function with a faint - end slope @xmath39 . in sharp contrast
we find that in isolated field regions the number density of uv - optical red sequence galaxies declines rapidly at magnitudes fainter than and is completely truncated at .
this confirms the findings of paper ii that no passively - evolving dwarf galaxies are found more than two virial radii from a massive halo , whether that be a group , cluster of massive galaxy .
moreover we show that the classification of dwarf galaxies into passively - evolving and star - forming from their sdss fibre - obtained spectra as used in paper ii is robust to aperture effects , finding that none of the 1375 galaxies with integrated colours indicative of recent star - formation were also spectroscopically classified as passive with . hence in isolated field regions ( where the majority of galaxies are found )
we find that the build - up of the red sequence of passively - evolving galaxies is incomplete , forming from the top down and being truncated at . in these regions only _ internal _ processes such as merging , supernovae and agn feedback mechanisms , and gas exhaustion due to star - formation can be responsible for completely stopping star - formation in galaxies , and hence can not be effective in low - mass galaxies ( for a discussion see * ? ? ?
* ) as otherwise passively - evolving dwarf galaxies would be ubiquitous . instead the passively - evolving dwarf galaxies that dominate ( in terms of numbers ) groups and clusters and make up the faint - end of the cluster red sequence must have had their star - formation quenched through processes directly related to their environment such as ram - pressure stripping or tidal shocks .
@xcite have followed the build - up with redshift of the passive and star - forming galaxy populations as selected from their rest - frame colour and/or their spectral energy distributions obtained by fitting their @xmath40 optical+ir photometry .
they are able to obtain a similarly robust separation of rest - frame uv - selected red and blue sequence galaxies to at least and probably as early as @xmath41 . for each redshift bin
( using either spectroscopic redshifts from the vimos vlt deep survey or photometric redshifts ) they obtain separate stellar mass functions for the quiescent / red and active / blue galaxy populations .
they find that the stellar mass density of red sequence galaxies has increased by a factor two since , and that this increase can be accounted for entirely by the shutdown of star - formation in active galaxies without requiring additional growth through dry mergers , in agreement with the results of @xcite who derive the global star - formation contribution of blue and red galaxy populations from ultraviolet and _ spitzer _
24@xmath15 m luminosities .
these results support the downsizing paradigm whereby the red sequence is built - up from the top down , being already largely in place at the bright end by @xcite , and the faint end filled in at later epochs in clusters and groups through environment - related processes such as ram - pressure stripping or galaxy harassment .
this filling in of the faint end appears to occur mainly at , and occurs earlier in the richest clusters .
@xcite find few faint red galaxies in four clumps belonging to a large - scale structure at suggesting that the red sequence is truncated within groups at , whereas by combining several rich clusters at @xcite find the red sequence to be fitted by a schechter function with @xmath42 to at least .
@xcite find an increase of a factor two in the dwarf - to - giant ratio of red sequence galaxies from to , while @xcite find evidence for a further doubling of the dwarf - to - giant ratio in the cluster red sequence since .
@xcite find that this incremental filling in of the faint cluster red sequence population since is consistent with a simple model whereby blue galaxies in clusters at have their star - formation suppressed by the hostile cluster environment . at all epochs to @xcite find a deficit of faint red sequence galaxies in field regions , with respect to cluster and group environments , although their deficits are not as dramatic as presented here due to the use of photometric redshifts and statistical subtraction methods to define the field sample , as well as the use of rest - frame @xmath43 colour - magnitude diagrams to define the red sequence population . to obtain clearer results for the field dwarf galaxy population at these redshifts would require a large - scale spectroscopic survey of galaxies to in order to define both the redshifts and environments of each galaxy , something that is within reach of present surveys ( e.g. deep2 , vvds ) , at least to .
cph acknowledges the financial supports provided through the european community s human potential program , under contract hprn - ct-2002 - 0031 sisco .
the authors thank gianni busarello , amata mercurio and francesco la barbera for stimulating discussions during this project .
funding for the sdss and sdss - ii has been provided by the alfred p. sloan foundation , the participating institutions , the national science foundation , the u.s .
department of energy , the national aeronautics and space administration , the japanese monbukagakusho , the max planck society , and the higher education funding council for england .
the sdss web site is http://www.sdss.org/. the sdss is managed by the astrophysical research consortium for the participating institutions .
the participating institutions are the american museum of natural history , astrophysical institute potsdam , university of basel , university of cambridge , case western reserve university , university of chicago , drexel university , fermilab , the institute for advanced study , the japan participation group , johns hopkins university , the joint institute for nuclear astrophysics , the kavli institute for particle astrophysics and cosmology , the korean scientist group , the chinese academy of sciences ( lamost ) , los alamos national laboratory , the max - planck - institute for astronomy ( mpia ) , the max - planck - institute for astrophysics ( mpa ) , new mexico state university , ohio state university , university of pittsburgh , university of portsmouth , princeton university , the united states naval observatory , and the university of washington .
this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
galex ( _ galaxy evolution explorer _ ) is a nasa small explorer , launched in april 2003 .
we gratefully acknowledge nasa s support for the construction , operation , and science analysis for the galex mission , developed in cooperation with the centre national detudes spatiales of france and the korean ministry of science and technology .
99 adelman - mccarthy , j. k. , et al .
2006 , apjs , 163 , 38 arnouts s. , et al .
2007 , preprint ( astro - ph/0705.2438 ) baldry i. k. , et al .
2004 , apj , 600 , 681 baldry i. k. , et al .
2006 , mnras , 373 , 469 baldwin j. a. , phillips m. m. , terlevich r. , 1981 , pasp , 93 , 5 balogh m. l. , baldry i. k. , nichol r. , miller c. , bower r. , glazebrook k. , 2004 , apjl , 615 , 101 beers t.c .
, flynn k. , gebhardt k. , 1990 , aj , 100 , 32 bell e. f. , et al .
2004 , apj , 608 , 752 bell e. f. , et al .
2007 , preprint ( astro - ph/0704.3077 ) blanton m. r. , berlind a. a. , hogg d. w. , 2006 , preprint ( astro - ph/0608353 ) blanton m. r. , et al . 2003 , aj , 125 , 2348 blanton m. r. , et al . 2005a , apj , 631 , 208 blanton m. r. , et al .
2005b , aj , 129 , 2562 boselli a. , et al .
2005 , apjl , 629 , 29 brinchmann j. , et al .
2004 , mnras , 351 , 1151 bruzual a. g. , charlot s. , 2003 , mnras , 344 , 1000 bundy k. , et al .
2006 , apj , 651 , 120 chiosi c. , cararro g. , 2002 , mnras , 335 , 335 cirasuolo m. , et al .
2006 , preprint ( astro - ph/0609287 ) croton d. j. , et al .
2006 , mnras , 368 , 11 davoodi p. , et al .
2006 , mnras , 371 , 1113 dekel a. , birnboim y. , 2006 , mnras , 368 , 2 de lapparent v. , 2003 , a&a , 408 , 845 de lucia g. , et al .
2007 , mnras , 374 , 809 giallongo e. , et al .
2005 , apj , 622 , 116 glazebrook k. , et al . 2004 , nature , 430 , 181 haines c. p. , merluzzi p. , mercurio a. , gargiulo a. , krusanova n. , busarello g. , la barbera f. , capaccioli m. 2004 , mnras , 371 , 55 haines c. p. , la barbera f. , mercurio a. , merluzzi p. , busarello g. , 2006 , apjl , 647 , 21 ( paper i ) haines c. p. , gargiulo a. , la barbera f. , mercurio a. , merluzzi p. , busarello g. , 2007 , mnras in press , preprint ( astro - ph/0707.1262 ) ( paper ii ) hopkins p. f. , hernquist l. , cox t. j. , kere d. , 2007 , preprint ( astro - ph/0706.1243 ) kauffmann g. , et al .
2003 , mnras , 341 , 33 kauffmann g. , et al .
2004 , mnras , 353 , 713 kauffmann g. , et al .
2006 , preprint ( astro - ph/0609436 ) kaviraj s. , et al .
2006 , preprint ( astro - ph/0601029 ) kennicutt r. c. , jr .
1998 , ara&a , 36 , 189 kere d. , katz n. , weinberg d.h .
, dav r. , 2005 , mnras , 363 , 2 kewley l. j. , dopita m. a. , sutherland r. s. , heisler c. a. , trevena j. , 2001 , apj , 556 , 121 kewley l. j. , jansen r. a. , geller m. j. , 2005 , pasp , 117 , 227 lemson g. , kauffmann g. , 1999 , 302 , 111 lpez - cruz o. , barkhouse w. , yee h. k. c. , 2004 , apj , 614 , 679 , martin d. c. , et al . 2005 , apjl , 619 , 1 martin d. c. , et al .
2007 , preprint ( astro - ph/0703281 ) miller c. j. , et al .
2003 , apj , 597 , 142 morrissey p. , et al .
2005 , apjl , 619 , 7 morrissey p. , et al .
2007 , preprint ( astro - ph/0706.0755 ) noeske k. g. , et al .
2007 , apjl , 660 , 43 pisani a. , 1996 , mnras , 278 , 697 rich r. , et al .
2005 , apjl , 619 , 107 salim s. , et al .
2005 , apjl , 619 , 39 salim s. , et al .
2007 , preprint ( astro - ph/0704.3611 ) schlegel d. j. , finkbeiner d. p. , davis m. , 1998 , apj , 500 , 525 schawinski k. , et al .
2006a , preprint ( astro - ph/0601036 ) schawinski k. , et al . 2006b , nature , 442 , 888 schechter p. , 1976 ,
apj , 203 , 297 springel v. , di matteo t. , hernquist l. , 2005a , mnras , 361 , 776 springel v. , et al .
2005b , nat , 435 , 629 stoughton c. , et al .
2002 , aj , 123 , 485 stott j. p. , smail i. , edge a. c. , ebeling h. , smith g. p. , kneib j .-
, pimbblet k. a. , 2007 , apj , 661 , 95 strateva i. , et al .
2001 , aj , 122 , 1861 strateva i. , et al .
2006 , a&a , 450 , 909 tanaka m. , et al .
2005 , mnras , 362 , 268 tanaka m. , et al .
2007 , mnras , 377 , 1206 tremonti c. a. , et al .
2004 , apj , 613 , 898 willmer c. n. a. , et al .
2006 , 647 , 853 wolf c. , gray m. e. , meisenheimer k. 2005 , a&a , 443 , 435 wyder t. , et al .
2007 , prepint ( astro - ph/0706.3938 ) yang x. , mo h. j. , van den bosch f. c. , jing y. p. 2005 , mnras , 356 , 1293 yi s. k. , et al .
2005 , apjl , 619 , 111 zheng x. z. , et al .
2007 , preprint ( astro - ph/0702208 ) | we combine _ galex _ near - uv photometry with a volume - limited sample of local sdss dr4 galaxies to examine the composition and the environmental dependencies of the optical and uv - optical colour - magnitude ( c - m ) diagrams .
we find that of red sequence galaxies in the optical c - m diagram show signs of ongoing star - formation from their spectra having .
this contamination is greatest at faint magnitudes and in field regions where as many as three - quarters of red sequence galaxies are star - forming , and as such has important consequences for following the build - up of the red sequence .
we find that the @xmath0 colour instead allows a much more robust separation of passively - evolving and star - forming galaxies , which allows the build - up of the uv - selected red sequence with redshift and environment to be directly interpreted in terms of the assembly of stellar mass in passively - evolving galaxies .
we find that in isolated field regions the number density of uv - optical red sequence galaxies declines rapidly at magnitudes fainter than and appears completely truncated at .
this confirms the findings of haines et al .
( 2007 ) that no passively - evolving dwarf galaxies are found more than two virial radii from a massive halo , whether that be a group , cluster or massive galaxy .
these results support the downsizing paradigm whereby the red sequence is assembled from the top down , being already largely in place at the bright end by , and the faint end filled in at later epochs in clusters and groups through environment - related processes such as ram - pressure stripping or galaxy harassment .
[ firstpage ] galaxies : clusters : general galaxies : evolution galaxies : luminosity function galaxies : stellar content |
the total hamiltonian ( @xmath86 ) describing electron - phonon interactions consists of the _ system _ hamiltonian @xmath8 which includes the electronic degrees of freedom , the free hamiltonian @xmath87 of the phonon bath describing fluctuations in the protein structure , considered as a set of @xmath88 independent bosonic baths @xmath89 coupled to each of the sites ( chromophores ) , and the linear coupling of fluctuations to the excitations @xmath90 : where @xmath92 are the usual creation and annihilation operators , respectively , satisfying @xmath93 = \delta_{ik}$ ] . for simplicity
, we will assume an identical coupling to the environment for each site @xmath94 with @xmath95 being the vibrational spectral density .
in the exciton basis , @xmath8 is diagonal , @xmath12 , and we will write @xmath96 . in that basis the hamiltonian can be expressed as @xmath97 and @xmath98 { \ensuremath{\left|{e_m}\right\rangle}}{\ensuremath{\left\langle{e_n}\right|}}$ ] , where we have neglected the lamb and stark shifts corresponding to @xmath99 in the last term .
a state @xmath100 , or sink , has been introduce to model the transfer to the rc , where the excitation will end up eventually as a relaxation process coming from site 3 , so the behaviour of the system will account for an additional channel @xmath101 .
although not fully equivalent , when using the master equation in the exciton picture we will model this irreversible transfer to the sink as happening at rate @xmath102 . in all cases
we study the time evolution of the sink s population following the selective excitation of site @xmath103 , @xmath104 . in the considered 3 sites model system ,
@xmath105 , while in the 8 site model of fmo , it is @xmath106 . in this section
we analyse the 3-site case beyond the concerns exposed in the markovian treatment , with the aim to show how manifestations of the phonon antenna also occur in a paradigmatic example of non - markovianity .
for that matter we will assume that all levels in the network are strongly coupled to identical independent harmonic oscillators , all tuned to the same frequency @xmath52 , representing therefore the coupling to a unique discrete mode .
the coupling between sites would be given by the same matrix @xmath8 .
this model presents the advantage of enabling a direct analysis in the local site basis , which lets us _ avoid _ making the approximation @xmath102 for the transfer rate to the sink node .
the local hamiltonian described in eq .
( [ mainequation ] ) fits this model if we assume that the sum of @xmath107 s restricts to just one mode .
the dissipation of each mode will be modelled using a master equation formalism , where the harmonic oscillators relax at the same rate ( @xmath108 ) to a ( markovian ) thermal bath at temperature @xmath109 , according to the lindblad superoperator : @xmath110 [ a_i \rho a_i^\dagger - \frac{1}{2}a_i^\dagger a_i \rho- \frac{1}{2 } \rho a_i^\dagger a_i ] \\
+ \sum_i \gamma [ n(\omega_h ) ] [ a_i^\dagger \rho a_i - \frac{1}{2 } a_i a_i^\dagger \rho- \frac{1}{2 } \rho a_ia_i^\dagger ] \end{gathered}\ ] ] the coupling constant @xmath111 is chosen to match the quasi lorentzian profile previously presented in @xcite . there
, it is shown that a two level system coupled to a bath with that profile behaves equivalently as if it were coupled to a damped oscillator with coupling constant @xmath112 within this approach , the oscillator is damped within an environment of ohmic nature @xmath113 .
although that is not exactly our case , it can be used as a first approximation to study the impact of non - markovian features .
the inset of fig .
[ fig : nonmarkov3sitescouplings ] shows some changes in the phenomenon .
it takes place close to the expected value , although the shift is prominent .
the explanation relies on the direct coupling to the harmonic oscillators , which is so strong that leads to a modification in the excitonic levels of the whole system , the final configuration being way more complicated than the one explained in section ii .
the concept remains valid , but of course the intuition about the configuration of the antenna is not so precise .
[ fig : nonmarkov3sitescouplings ] shows a plot with the mode fixed at 200 @xmath77 .
as we see , the values of the hamiltonian are close to a maximum in transfer yield showing once more the forming of an antenna . for the sake of modelling appropriately the spectral function used for the fmo case , we have relied on the function proposed by adolphs and renger @xcite . in that reference a combination of super - ohmic densities
is used and then a single effective high - energy vibrational mode @xmath115 is added with huang rhys factor @xmath116 and @xmath117 . in our case , instead of using a delta function , we introduce a pure lorentzian with the same characteristics but a linewidth of @xmath118 . with those prescriptions the spectral function can be written as we have envisioned a figure of merit to check how well can the concept of _ phonon antenna _ alone predict the main features of the landscape as shown in fig .
[ fig : nonmarkov8sitescouplings]a . for the same variations of @xmath34 and @xmath59
we have studied the level structure and computed the simple value : @xmath123 the higher the value of @xmath124 , the more assisted by the environment the transition to the sink would be , as inferred from the scheme described .
[ fig : figureofmerit ] shows the map of @xmath124 in the same range of parameter values as fig .
[ fig : nonmarkov8sitescouplings ] . | a non - trivial interplay between quantum coherence and dissipative environment - driven dynamics is becoming increasingly recognised as key for efficient energy transport in photosynthetic pigment - protein complexes , and converting these biologically - inspired insights into a set of design principles that can be implemented in artificial light - harvesting systems has become an active research field . here
we identify a specific design principle - _ the phonon antenna _ - that demonstrates how inter - pigment coherence is able to modify and optimize the way that excitations spectrally sample their local environmental fluctuations .
we place this principle into a broader context and furthermore we provide evidence that the fenna - matthews - olson complex of green sulphur bacteria has an excitonic structure that is close to such an optimal operating point , and suggest that this general design principle might well be exploited in other biomolecular systems . _
introduction. _ experimental techniques such as optical 2d fourier transform spectroscopy have recently begun to probe the ultrafast photophysics of energy transport in a range of pigment - protein complexes ( ppcs ) taken from photosynthetic green sulphur bacteria , marine algae and higher plants @xcite .
all of the key photosynthetic light reactions ( photon capture , energy transport and charge generation ) are carried out in ppc structures @xcite , and the often @xmath0 quantum efficiency with which they carry out these functions has created considerable interest in understanding and exploiting their design features for artifical solar energy applications @xcite .
surprisingly , recent experiments have found direct evidence for long - lasting quantum coherence amongst the excitons which transport energy in these complexes . in the case of the fenna - matthews - olson ( fmo )
complex , these coherences can persist on picosecond timescales in cryogenic conditions and are still observable at room temperatures @xcite .
several phenomenological theories have subsequently shown that there is an optimal mixture of coherent inter - pigment energy transport and stochastic environmental noise that may lead to faster and higher - yield energy delivery in ppc architectures , suggesting that quantum effects may underpin their efficient function @xcite .
microscopic investigations have also recently shown the key role of both discrete and continuous environmental fluctuation spectra in stabilising the long - lasting coherences observed in spectroscopy as well as facilitating efficient transport @xcite . in this letter
we build upon these microscopic studies to describe an explicit design principle through which coherent excitonic coupling is _ exploited _ to optimize energy transport in the typical environmental conditions found in ppcs .
this is based on the idea that coherent couplings can allow a pigment network to be spectrally ` tuned ' into configurations where the emerging relaxation ( transport ) pathways may extract maximum noise strength from the environment and thus proceed faster . in analogy to the well - established idea that excitonic interactions in light - harvesting antenna complexes allow pigments to absorb a broader range of the ambient light spectrum @xcite , we refer to this principle of advantageous , coherently - modified sampling of the protein noise spectrum as _ the phonon antenna mechanism _ @xcite . _ the phonon antenna. _
subject to dephasing noise as described in the text .
an excitation on site @xmath1 will be transferred to the sink ( site @xmath2 ) via sites @xmath3 and @xmath4 .
the population transfer from @xmath4 to the sink is incoherent ( dotted arrow ) .
( b ) diagonalising the network hamiltonian results in new eigenstates @xmath5 .
for the parameters in ( b ) , the energy differences between eigenstates correspond to maxima of a hypothetical environmental spectral function shown in ( c ) .
environment - induced relaxation between these states ( single - headed arrows ) is thus optimised , as transition rates are proportional to the spectral density at the exciton energy differences .
, width=302 ] to illustrate the concept , we begin by setting out the ( canonical ) model of exciton transport in a ppc network whose function is to deliver excitonic energy to a reaction centre ( where it is used to generate electrons for subsequent dark stages of the photosynthesis ) @xcite .
the network consists of a set of chromophores ( sites ) which are held in space by a protein matrix .
each chromophore possesses an optically excited state @xmath6 , whose energy @xmath7 depends on the local environment .
the electronic hamiltonian @xmath8 is then given by @xmath9 , the sum of single exciton contributions on each site @xmath10 with local energy energy @xmath7 and coherent coupling terms describing the inter - site exchange of excitations with dipolar coupling strength @xmath11 .
the states obeying @xmath12 are in general delocalized excitonic states , which can be also be expressed in the site basis as @xmath13 . the ultimate transfer of the exciton to a reaction center
is modelled by including an isolated site ( labeled @xmath2 and referred to as the ` sink ' ) which is populated irreversibly from a particular local site @xcite .
non - electronic degrees of freedom , due to the surrounding protein environment and/or internal vibrational modes of the chromophores , also couple to the excitons , inducing local fluctuations of the site energies @xmath7 .
these ` environmental ' fluctuations ( phonons ) are modelled as originating from a continuous _ bath _ of harmonic oscillators coupled to the exciton and constitute the primary source of energy relaxation and dephasing for the excitonic system @xcite - see appendix a. coherent inter - site couplings in @xmath8 can be stronger or of the same order than the characteristic strength of the environmental interactions @xcite .
in such cases delocalised exciton eigenstates , for which @xmath8 is diagonal , therefore provide the _ preferred _ basis to analyze the system s transport dynamics . within this framework
, we can employ secular redfield theory to obtain the following markovian equation of motion for the population @xmath14 of the exciton states @xmath15 @xcite : @xmath16 where the transition rates between eigenstates are given by : @xmath17 , & \mbox{$\omega_{mn } < 0$}\end{array}\end{aligned}\ ] ] with @xmath18 here the irreversible process that locally populates the sink from site @xmath19 is assumed to have an ( environment - independent ) rate @xmath20 @xcite , which is typical for transfer to reaction centers . in the exciton picture
this leads to population transfer to the sink at a rate @xmath21 .
our figure of merit for assessing the transport efficiency of a given network is the population in the sink @xmath22 at some time @xmath23 following the local excitation of a selected site .
the population dynamics is entirely driven by the environment and , as a result of energy conservation , the relaxation rate of an exciton transition @xmath24 ( which requires the emission of a phonon ) is thus dependent on both the phonon density of states and the coupling strength to phonons at frequencies equal to the energy difference @xmath25 .
both these properties are described by the environmental spectral function @xmath26 and the transition rate @xmath27 is proportional to @xmath28 . as typical ppc spectral functions
often have ( at least ) one maximum at a finite frequency @xcite and may exhibit sharp features due to long - lived vibrational modes in the environment , it becomes possible to _ tune _ the hamiltonian parameters to create excitonic states with energy differences @xmath25 at local maxima of the spectral function and thus optimise @xmath27 .
_ this is the basis of the phonon antenna effect_. figure [ antennapic ] illustrates the phonon antenna mechanism in a sequential transport scenario .
the simplest model consists of a network made up of three sites , @xmath29 , with local energies @xmath30 , @xmath31 , as shown in figure [ antennapic]a .
sites @xmath1 and @xmath3 are assumed to be nearly degenerate and strongly coupled , with a smaller coupling between sites 2 and 3 , and almost no coupling between sites 1 and 3 ; that is , @xmath32 .
initially , only site @xmath33 is excited .
for these parameters , the coherent coupling @xmath34 splits the degeneracy of states @xmath33 and @xmath35 and the exciton eigenstate of @xmath8 are approximately @xmath36 , @xmath37 , @xmath38 , with energies @xmath39 and @xmath40 . the initial condition can be re - expressed as @xmath41 , and an energy diagram for these new states and initial condition is shown in figure [ antennapic]b . the efficiency of transport to site @xmath4 ( and thence to the sink ) now depends on the dissipative rates @xmath42 and @xmath43 of the environment - induced transitions @xmath44 and @xmath45 , respectively .
let us consider now an spectral function possessing a maximum at an energy @xmath46 , as depicted in fig .
[ antennapic]c ) .
when @xmath47 , we find that @xmath48 and a ladder of equally spaced states appears with a splitting @xmath46 .
thus a clever choice of @xmath34 simultaneously optimizes _ both _ transition rates @xmath42 and @xmath43 and generates efficient energy transport to the sink .
this simple system illustrates the essential idea behind implementing an efficient phonon antenna : _ electronic interactions are tuned ( e.g. by controlled or evolutionary adaption of distances and relative orientation of chromophores ) to achieve resonance with maxima of the environmental vibrational spectra_. this already suggests how in more general networks , coherent couplings and site energies could be tuned to enhance ( or inhibit ) a particular set of energetic transitions ( which in a multi - chromophore network could also correspond to a particular spatial pathway ) . the more structure in the spectral function , the greater the possibilities for phonon antenna effects to be harnessed , as shown in fig.[antennapic]c where a shoulder feature in @xmath26 also enhances the direct @xmath49 transition . in anticipation of our analysis of the fmo complex
, we now perform a quantitative analysis of this three - site model system . motivated by actual values in typical ppcs ,
the considered hamiltonian parameters are , in units of @xmath50 : @xmath51 the spectral density function is taken to be of a quasi - lorentzian shape centered in @xmath52 @xcite : @xmath53 where @xmath54 is fixed in order to have a reorganization energy @xmath55 , and @xmath56 .
we assume that initially site 1 is selectively excited . within this parameter regime , the inset in figure [
fig : markov3sitescouplings ] shows the population in the sink after @xmath3 ps as @xmath46 is varied .
as expected , a clear peak in the transport efficiency emerges when the spectral function is peaked slightly above the value @xmath57 which would correspond to perfect matching of the inter - exciton energy differences to the maximum of the spectral density .
figure [ fig : markov3sitescouplings ] shows the sink populations after @xmath3 ps for fixed @xmath46 and @xmath58 , while varying the coupling @xmath34 and the site energy @xmath59 .
we observe a symmetric structure in the transport efficiency with respect to @xmath60 , but find that the degenerate configuration @xmath61 is not a global maximum of the efficiency .
the highest efficiency occurs for @xmath62 and @xmath63 .
this larger value of the efficiency is the result of a slightly enhanced coupling ( overlap ) between sites @xmath3 and @xmath4 that results from shifting site @xmath3 to a lower ( local ) energy .
however , the differences in exciton energies @xmath64 and @xmath65 remain close to the maximum of the spectral density .
indeed we notice that the reduction in the optimal values of @xmath34 as @xmath59 decreases maintains energy differences close to the spectral density maximum .
the shift in the global maximum illustrates the role of the @xmath66 factor in determining the energy and coupling structures which optimally exploit the functional form of the environment s spectral function .
for a temperature @xmath67 .
actual values of the hamiltonian values , lie as expected ( blue dot ) , at the top of a maximum in transfer yield . in the inset ,
population in the sink after 2 ps for a spectral density peaked at @xmath52 , given a temperature @xmath67 . as an example
were the matching takes place , the phonon antenna is formed and adjusted to a peak @xmath68 .
the other peak around @xmath69 but with half the height corresponds to an antenna effect directly connecting sites @xmath70 and @xmath71 .
small deviations come from @xmath72 being not zero .
, scaledwidth=45.0% ] _ the phonon antenna in the fmo complex : modified redfield simulation. _ the fmo complex is a ppc present in green sulfur bacteria which acts as a`molecular ' wire by mediating exciton transfer from the light - harvesting chlorosomes ( antenna complex ) to the bacterial reaction center @xcite .
it has a trimeric structure with each of the three monomers containing seven chromophores bound to a common protein structure . under low light conditions
, it can be assumed that there is only one excitation at a time , and that all monomers operate independently .
the excitation is thought to originate at an eighth site ( common to several monomers ) which injects an exciton locally at sites @xmath1 or @xmath73 of a given monomer @xcite .
the exciton then relaxes through the manifold of exciton states , and is directed towards site @xmath4 ( which is the primary component of the lowest energy exciton state ) . due to its proximity to the reaction centre ( rc ) ,
the exciton on site @xmath4 is then rapidly transferred out of the fmo complex @xcite .
experimental and theoretical studies have described two main relaxation pathways in the fmo complex @xcite . here
we shall focus on the dominant pathway that proceeds through site @xmath1 to site @xmath4 , although our numerical analysis is based on simulations using the modified redfield approach as described in @xcite of the entire eight - site fmo structure with an initial excitation on site @xmath74 .
the sink is connected to site @xmath4 .
unless stated otherwise , the hamiltonian parameters are taken from @xcite , and the spectral density of the environment is based on adolphs and renger fit to fluorescence narrowing experimental data @xcite . for different values of the coupling between sites 1 and 2 ( @xmath34 ) and the energy of site 2 ( @xmath75 ) .
we clearly see that the actual values for @xmath34 and @xmath75 lie at the top of a maximum .
the inset shows @xmath76 for the actual couplings as observed for the fmo , assuming an spectral density function such as that of @xcite but peaked at @xmath52 .
b ) local site and exciton energies for the @xmath8 values corresponding to the fmo , as taken from @xcite .
the arrows show how the splitting might act as an antenna for the energy peak close to 180 @xmath77,scaledwidth=45.0% ] fig .
[ fig : nonmarkov8sitescouplings]a shows the final population transferred to the sink for a simulation of a total time @xmath78 at @xmath79 as a function of @xmath34 and @xmath75 .
the transport efficiency shows a complex landscape of maxima and minima for the parameter region explored , but we find that the hamiltonian parameters of ref .
@xcite ( marked as a blue cross ) are located at an optimal point .
figure [ fig : nonmarkov8sitescouplings]b shows the energy level structure in both the local and exciton bases for the physical hamiltonian parameters of @xcite .
we see in the exciton basis that the energy differences corresponding to the relaxation transitions connecting sites @xmath80 and @xmath4 are well matched to the maximum of the experimentally - determined spectral function , and therefore this relaxation channel forms a striking example of an operational phonon antenna in a biological transport system . the inset of fig .
[ fig : nonmarkov8sitescouplings ] shows @xmath81 for fixed hamiltonian parameters and a varying peak position @xmath46 in the spectral density ( as described in appendix d ) .
again , we find that @xmath82 is a multi - peaked function , but the most prominent maximum appears at @xmath83 , which is very close to the experimentally - fitted peak in the spectral function of adolphs and renger ( @xmath84 ) @xcite . _
the phonon antennae in a broader context _ the phonon antennae principle which states that _ electronic interactions are tuned to achieve resonance with maxima of the environmental vibrational spectra and thus strongest response _ may be viewed in a broader context in which the tuning of electronic degrees of freedom to match the frequency of external signals is exploited to achieve enhanced system response to facilitate sensing and transport .
in fact , the phonon antennae principle is closely related to the hartmann - hahn condition @xcite which is of considerable importance in sensing devices . here
the challenge is that the sensor , e.g. an electron spin , may not be resonant with an external system generating a signal ( e.g. a different electron or nuclear spin ) .
this problem may be overcome by continuous driving of the electron spin in the sensor at a strength that leads to a splitting of dressed states that matches the externally generated signal and thus permits a strong response of the sensor ( transitions between upper and lower dressed state ) @xcite .
a further example in which tuning is not achieved by driving is the recently proposed theory that olfaction ( our sense of smell ) is explained , at least partially , as originating from phonon assisted electron transport where electron source and drain in the receptor are separated by an energy splitting that is matched to a specific vibrational frequency of an odorant molecule .
electron tunneling , hence strong response , will occur only if the molecular vibrations matches this energy gap @xcite . _
discussion. _ taken together , these results suggest that the physically important relaxation pathway between sites @xmath1 and @xmath4 is mediated by pigments which are spectrally and spatially positioned by the protein to efficiently sample the spectral function of the protein s fluctuations . whether this optimality is a determinant in the emergence of this structure in nature is beyond the scope of this study , but it is striking how well the phonon antenna concept can be used to rationalise the site energies and couplings of the pigments participating in this pathway .
other considerations , such as structural stability and steric constraints imposed by the need to integrate the structure into the cellular environment are likely to drive the supramolecular structure of ppcs , however the multiplicity of local maxima we find _ does _ provide the opportunity to find an optimal working point for transport in the _ vicinity _ of a structure arising from these other factors .
indeed , we find that the fmo complex parameters correspond to a local maximum of the @xmath85 pathways efficiency .
interestingly , changes in temperature will also alter the inter - exciton energy that maximises the relaxation rates in eq .
( [ rate ] ) , and an evolving or adaptive system might be able to change its coupling structure to maintain the inter - exciton energy differences at the maximum of the thermal noise strength .
we also point out that while the redfield formalism allows us to give a physically clear description of the essential phonon antenna mechanism , the use of markovian redfield theory for ppc problems neglects a number of important dynamical effects ( environmental memory , stokes shift , discrete modes in the spectral function ) which are crucial for understanding the _ persistence _ of inter - eigenstate coherence that is generated in experiments @xcite .
these effects have been neglected here for simplicity , but exploring these effects in the context of the phonon antenna mechanism could lead to an even more sophisticated principle , involving transient , broad - band sampling of the spectral function and evolving exciton spectra and localisation lengths which could , potentially , activate or deactivate relaxation pathways dynamically .
finally , the concept of using coherent interactions to alter the sampling of a ( fixed ) environmental spectral function could be readily achieved in a variety of atomic and condensed matter systems @xcite , and could also be used as a sensitive probe for measuring spectral functions by measuring transport rates as functions of a controllable coupling parameter . _
acknowledgements. _ this work was supported by the alexander von humboldt - foundation , the eu strep project picc and the eu integrated project q - essence .
awc acknowledges support from the winton programme for the physics of sustainability .
m. del rey was supported by the spanish micinn project fis2011 - 29287 , cam research consortium quitemad s2009-esp-1594 , csic jae - predoc2010 grant and the fundacin botn .
the authors would also like to thank rienk van grondelle for fruitful discussions .
30 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 , , , , , , , , * * , ( ) . , , , , , , , ,
* * , ( ) . , , , , , , , * * , ( ) .
, , , , , , * * , ( ) . ,
_ _ ( , ) . , , , _ _ ( , ) . , , , , * * , ( ) . , , , , , , , * * , ( ) . , , , ,
* * , ( ) . , * * ( ) . , , , , , * * , ( ) . , , , , * * , ( ) . , , , ,
* * , ( ) . , , , , , * * ( ) . , , , , , * * , ( ) . , , , , , ( ) , . , , , , * * , ( ) . , * * , ( ) . , , , * * , ( ) . , , , , ( ) . , , , , , , , * * , ( ) . , * * , ( ) . , , , , , , * * , ( ) . , * * ( ) . , _ _ ( , ) , isbn .
, , , * * , ( ) . , , , ,
* * , ( ) . ,
* * , ( ) . , , , , * * , ( ) . , * * , ( ) . , * * , ( ) .
p. london , j. scheuer , j.m .
cai , i. schwartz , a. retzker , m.b .
plenio , j. isoya , r. fischer , l. mcguiness , b. naydenov and f. jelezko , , , , , ( ) .
j.m .
cai , f. jelezko , m.b .
plenio and a. retzker .
l. turin , * * , ( ) .
, , , * * , ( ) . |
i would like to thank benot and paolo for their very competent supervision of this master thesis .
river landscapes exhibit many different forms in all climatic regions of the world . nevertheless , one is also able to observe common features which often can be found to be the product of specific water and sediment interaction (
see for example @xcite and references therein ) .
the science studying the coupled water and sediment dynamics to explain formation and alteration of river courses is called river morphodynamics and includes research on mainly longitudinal structures like long sediment waves and more complex 2-dimensional structures , among them are alternate and multiple bars ( see figure [ fig : bars ] ) . in the past , these research areas have been investigated using experimental setups ( see @xcite for bar experiments for example ) , numerical simulations ( see @xcite for example ) and theoretical approaches based on linear stability analysis ( @xcite , @xcite and @xcite ) .
linear stability analysis is a concept that allows to study the asymptotic fate ( @xmath0 ) of a linear or linearized system which is slightly perturbated away from a spatially homogeneous solution .
the method was originally developed by @xcite and has been applied frequently to hydrodynamic topics since .
for example , @xcite used linear stability analysis to show that the capacity of a river to develop alternate or multiple bars is closely linked to its aspect ratio ( river width divided by depth ) .
this benchmark result is depicted in figure [ fig : cst ] and we will reproduce it later in the present work . + it is well - known though that in addition to water and sediment transport , riparian vegetation can play a crucial role in river pattern development ( see @xcite ) . in particular , it is recognized that riparian vegetation affects river morphology through modification of the flow field , bank strength and erosion / sedimentation processes in the riverbed / floodplain ( @xcite ) . however , due to the very complex nature of the dynamic interactions between vegetation and sediment transport and flow , riparian vegetation evolution was often not taken into account explicitly .
instead , it was added as a correcting factor for bed roughness and bank stability ( @xcite ) . while the treatment of vegetation as a correction factor may be justified when
looking at short timescales where riparian vegetation density does not change much , this is not the case for river pattern formation which occurs over much longer timescales and where vegetation takes an active role in the process .
for instance , dynamic interaction between riparian vegetation and flow and sediment is thought to be crucial in the formation of anabranching river patterns on vegetated bars and in ephemeral rivers in dry regions ( figure [ fig : anabranch ] a and b ) .
additionally , we can find similar patterns on the inside of a meandering bend in large streams ( scroll bars , figure [ fig : anabranch ] c ) .
+ recently , researchers have added riparian vegetation dynamics to numerical morphodynamic models and included some of the feedback mechanisms that are thought to occur in nature .
namely , @xcite took into account vegetation induced impedance to sediment transport and increase in bank stability and @xcite modeled the interaction of bank stabilizing vegetation and a meandering riverbed .
furthermore , @xcite proposed an analytical morphodynamic model coupled with an equation for riverbed vegetation dynamics .
but , until today vegetation dynamics was never included in a stability analysis of morphodynamic equations .
in fact , several difficulties arise when trying to formulate a physical vegetation model suitable for stability analysis .
for example , in modeling the sediment stabilizing effect of plant root systems is often taken into account as a threshold in the sediment transport function below which no erosion occurs .
however , such a threshold possesses mathematical properties that are not suitable for a stability analysis .
+ is the river s aspect ratio and @xmath1 is the dimensionless longitudinal wavenumber which characterizes the spatial periodicity of the bars , width=340 ] in the present work , we propose a minimal model for the evolution of riverbed vegetation density which takes into account only very basic mechanisms and is thus suitable for a stability analysis . using this vegetation model and a standard morphodynamic framework ,
we would like to explore the possibility of such a coupled morphodynamic - vegetation system ( ecomorphodynamic equations ) to explain the formation of anabranching patterns .
we would like to know which are the determining variables and to what extent the ecomorphodynamic analysis differs from the state of the art river morphodynamics .
hence , we perform an analytical linear stability analysis on the linearized set of ecomorphodynamic equations which describe a model river whose riverbed is colonized by plants .
this river is assumed to be of constant width with inerodible banks , the riverbed consists of cohesionless , erodible material ( sand / gravel ) of uniform size and the river s sediment transport capacity is thought to always exceed the threshold above which sediment transport occurs .
additionally we assume sediment transport to be mainly bedload .
+ we begin by formulating an equation which describes the evolution of riverbed vegetation density ( section [ sec : veg ] ) and discuss the different terms and its validity .
important mechanisms to be considered are vegetation growth , distribution by means of seeding and resprouting , and death through flow impact induced uprooting .
this equation is then coupled with a standard 1-dimensional and 2-dimensional river morphodynamic framework ( sections [ sec:1d_gov ] and [ sec:2d_gov ] respectively ) which consists of depth - averaged fluid and sediment continuity as well as a formulation for momentum balance in the fluid .
these systems are subsequently linearized and perturbated around a spatially homogeneous solution and the conditions for which the wavelike perturbations amplify are investigated using linear stability analysis .
this is done for a 1d - framework to study instability towards long sediment waves in section [ sec:1d ] and for a 2d - framework to study the formation of alternate and multiple bars .
the main focus in this work is on highlighting the fundamental role that vegetation dynamics can have in this process together with known mechanisms of sediment dynamics .
stability analysis of morphodynamic equations generally does not include the active role of vegetation explicitly due to the complex nature of the interaction mechanisms .
hereafter , we develop an analytic model for riverbed vegetation dynamics and discuss its validity for different conditions . for simplicity ,
we model vegetation as rigid , non - submerged cylinders with constant radius and submerged height equal to water depth .
we then call @xmath2 the vegetation density defined as number of plants per unit area of riverbed and we model its growth by the logistic term @xmath3 with carrying capacity @xmath4 and specific vegetation growth rate @xmath5 .
furthermore , we assume that vegetation growth is stimulated by nearby existing vegetation by means of seeding and resprouting ( i.e. positive local feedback ) and we model it using the diffusion term @xmath6 with @xmath7 the streamwise vegetation diffusion constant and @xmath8 the streamwise coordinate .
we finally want to quantify vegetation death caused by flow drag for which we only consider the direct uprooting effect of flow drag on non - submerged and rigid vegetation ( type i mechanism after @xcite ) . in this case , a fluid parcel which impacts on the vegetation is decelerated from mean stream velocity to zero .
furthermore , the rate of fluid that impacts on the vegetation is also proportional to stream velocity while the vegetation cross - section per cubic meter of river is proportional to water depth and vegetation density .
we therefore propose the vegetation uprooting term ( see also @xcite ) @xmath9 where @xmath10 is a proportionality constant , @xmath11 the water depth and @xmath12 the streamwise velocity .
putting together equations ( [ eq : veg1 ] ) to ( [ eq : veg3 ] ) we get the rate of change of vegetation density as @xmath13 however , in real rivers flow is not constant throughout the year .
typically , large parts of a river s cross - section are only flooded during a limited amount of time per year which allows vegetation to colonize these surfaces during non - flooded periods
. therefore , equation ( [ eq : veg_const ] ) , except for certain special cases ( see figure [ fig : veg_paolo ] , where vegetation seems to grow while being completely submerged most of the time ) , is not really applicable for vegetation growth in natural streams since it considers all processes to happen simultaneously . in reality
however , vegetation grows and seeds during the vegetation period ( which is part of the non - flooded period ) and is uprooted during the flooding period . to simplify our analysis , we assume constant and continuous flow and thus we have to integrate growth and seeding into the flooding period . in the following , we call @xmath14 the drought period without vegetation growth , @xmath15 the vegetation period and @xmath16 the duration of the flooding ( see figure [ fig : timescales ] for illustration where
@xmath15 and @xmath14 have been separated for simplicity ) .
the duration of a complete cycle ( for example a year or half a year depending on the specific conditions ) is then given by @xmath17 .
if we assume that vegetation density does not vary much during a complete cycle i.e. @xmath18 where @xmath19 is the value of @xmath2 at the end of cycle i , then we can approximate the difference @xmath20 by the continuous time derivative @xmath21 .
we write the change of @xmath2 after one cycle as @xmath22\frac{\tilde{t}_v}{\tilde{t}_d+\tilde{t}_v+\tilde{t}_f}-\tilde{\alpha}_d\tilde{y}\tilde{u}^2\tilde{\phi}_i\frac{\tilde{t}_f}{\tilde{t}_d+\tilde{t}_v+\tilde{t}_f}.\ ] ] by approximating the finite differences by derivatives we get @xmath23\frac{\tilde{t}_v}{\tilde{t}_d+\tilde{t}_v+\tilde{t}_f}-\tilde{\alpha}_d\tilde{y}\tilde{u}^2\tilde{\phi } \frac{\tilde{t}_f}{\tilde{t}_d+\tilde{t}_v+\tilde{t}_f}\ ] ] and since we assumed @xmath14,@xmath15 and @xmath16 to be constant , we can integrate them into the proportionality constants to end up with @xmath24 where @xmath25 , @xmath26 and @xmath27 .
we can see that merging together the different mechanisms results in a relative increase of the growth and diffusion constant with respect to the uprooting constant if @xmath28 .
so even if in general the vegetation uprooting coefficient is much higher than the growth coefficient , this can be compensated by the small timescale ratio @xmath29 to get a regime where mutual feedback is possible .
this is the case for example in the marshall river ( see hydrograph in figure [ fig : hydrograph ] ) and also for bar flooding in the thur river ( see for example @xcite ) .
thus , the differential equation ( [ eq : veg_fin ] ) may also be valid in the case of non - constant flow if the modeling assumptions are met .
+ we quickly want to discuss two of the most important modeling assumptions adopted above , namely : * vegetation density change during a cycle is small compared to its actual value * the only uprooting effect is due to direct flow drag on non - submerged rigid vegetation the first assumption can be assumed to be valid if one considers the case of well developed vegetation . the vegetation coverage is dense enough to not allow much more biomass to be produced and at the same time a large part of the vegetation is robust enough to outlive the flooding period .
the second point refers to the fact that we only consider direct flow drag ( thus neglecting erosion which exposes the root system ) .
additionally , we need rigid vegetation like small trees or bushes with mean vegetation height @xmath30 greater than water depth @xmath11 in order for our assumption to be valid . for non - rigid vegetation , the exponent of @xmath12 in the uprooting term ( equation [ eq : veg3 ] ) should be somewhere between 1 and 2 while in the case of completely submerged vegetation the surface impacted by flow drag would be reduced by a factor of @xmath31 and thus , @xmath11 would be replaced by @xmath30 in equation ( [ eq : veg3 ] ) .
in this section we perform a linear stability analysis of the 1-dimensional ecomorphodynamic equations . the 1d - framework is valid in case flow , bed and vegetation can be assumed to be homogeneous in the direction transverse to the flow .
after the derivation of the dimensionless governing equations , linear stability is assessed .
we first reproduce some well - known results ( @xcite and @xcite ) and then we go on to evaluate the effect that riverbed vegetation dynamics has on these results .
figure [ fig : saintvenant ] depicts the model scheme adopted with the streamwise coordinate @xmath8 , the lateral ( normal ) coordinate @xmath32 ( not used in the 1d - analysis ) and the vertical coordinate @xmath33 .
the riverbed of constant width is assumed to consist of sandy , non - cohesive material which causes friction and may be transported by the fully turbulent flow .
furthermore , we consider the case of a straight channel ( see for example @xcite for curved channels ) with non - erodible banks .
additionally , vegetation as described in section [ sec : veg ] is able to colonize the whole riverbed .
then , assuming hydrostatic pressure distribution and the river width to be considerably larger than the flow depth , flow velocity may be depth - averaged and thus we get the well - known 1-dimensional de saint - venant momentum conservation law @xmath34+\frac{\tilde{\tau}}{\tilde{y}}=0.\ ] ] where the first and second term are the local and convective acceleration respectively , the third term represents hydrostatic pressure distribution , term 4 is the streamwise slope of the river and term 5 is the bed friction term . recall that @xmath12 is the streamwise velocity and @xmath11 is the water depth , while @xmath35 is the bed elevation .
we have to keep in mind that we are talking about long waves throughout our analysis ( pattern wavelength is larger than channel width ) in order for the depth - average as well as the 1d - formulation to make sense . as a closure relationship for bed friction @xmath36
, we choose the simple chezy equation and write @xmath37 where @xmath38 is the overall chezy coefficient .
the overall chezy coefficient depends on both , the bed roughness and the roughness induced by vegetation . according to @xcite ,
it can be expressed for non - submerged and rigid vegetation as @xmath39 where @xmath40 is the bed roughness which can be calculated by fixing manning coefficient @xmath41 , @xmath42 is the stokes drag coefficient and @xmath43 is the vegetation diameter .
+ subsequently , flow continuity is formulated as @xmath44 thus neglecting flow diversion by vegetation and assuming that sediment density in the water is low , therefore omitting the sediment term .
note that the flow diversion effect could easily be added but is left out here to keep the analysis simple . in order to account for sediment continuity
, we then write the well - known 1d - exner equation , valid for non - cohesive sediment with uniform grain size as @xmath45 where @xmath46 is bed porosity and @xmath47 is sediment flux per unit width .
we assume well - developed sediment transport ( always above the critical threshold ) , mainly in the form of bed load transport and therefore , as was done by @xcite , we adopt @xmath48 where @xmath49 is a parameter .
this is an approximation of the original meyer - peter / m@xmath50ller formula which states @xmath51 with @xmath52 the dimensionless shear stress and @xmath53 the critical dimensionless shear stress . omitting the threshold @xmath53 ( assuming sediment transport to be always above the threshold ) and knowing that @xmath52 is proportional to @xmath54 we get back our simplified power law . +
finally , we model riverbed vegetation dynamics using @xmath55 as explained in section [ sec : veg ] . equations ( [ hans11 ] ) and equation ( [ hans13 ] )
are conventionally called the de saint - venant s ( sv ) equations . if sediment dynamics is added , we speak of de saint - venant - exner equations ( sve ) or morphodynamic equations . since we added vegetation dynamics to sve , we name it the de saint - venant - exner - vegetation equations ( svev ) or the ecomorphodynamic equations . to perform a linear stability analysis , it is convenient to work with dimensionless quantities
. therefore , to write equations ( [ hans11 ] ) and ( [ hans13 ] ) to ( [ hans15 ] ) in dimensionless form , we introduce the change of variables ( motivated by the approach of @xcite ) [ eq:1d_cha ] @xmath56 where @xmath57 is the normal water depth and @xmath58 is the velocity at normal water depth . using change of variables ( [ eq:1d_cha ] ) , we obtain ( arranged in a way to have the time derivative on the left - hand side ) [ eq:1d_dim ] @xmath59-c_b\frac{u^2}{y}-c_v\phi u^2\\ \frac{\partial{y}}{\partial{t}}&=-y\frac{\partial{u}}{\partial{s}}-u\frac{\partial{y}}{\partial{s}}\\ \frac{\partial{\eta}}{\partial{t}}&=-\gamma u^2\frac{\partial{u}}{\partial{s}}\\ \frac{\partial{\phi}}{\partial{t}}&=\nu_g \phi(1-\phi)+\nu_d\frac{\partial{^2\phi}}{\partial{s^2}}-\nu_d\phi yu^2,\end{aligned}\ ] ] where @xmath60 , @xmath61 , @xmath62 , @xmath63 , @xmath64 , @xmath65 and @xmath66 . a linear stability analysis consists of studying the behavior of a linearized system when slightly perturbated away from a spatially homogeneous solution ( see @xcite ) . in the case of river morphology ,
a common choice for a homogeneous solution consists of a river with flat bed and constant slope under constant , uniform flow conditions ( see section [ sec : veg ] for the generalization to non - constant flow ) .
then , the reaction of the linearized system to small perturbations on every state variable is investigated whose physical meaning may be a variation in sediment supply or channel width for example ( @xcite . regardless of its shape , such a local perturbation can readily be interpreted as a wave packet and thus a velocity perturbation wave packet @xmath67 can be written as a fourier series with continuous wavenumber k @xmath68 where @xmath69 is the perturbation amplitude and @xmath70 the velocity perturbation . in a linear system , each sinusoidal component of the perturbation wave packet can then be treated separately to evaluate if there is growth towards periodic spatial patterns of the riverbed . in the following ,
we first derive the homogeneous solution of ( [ eq:1d_dim ] ) and then linearize and perturbate the equations around the homogeneous solution .
we begin with looking for spatially homogeneous solutions @xmath71 using normal flow conditions .
so , @xmath72 , @xmath73 and @xmath74 ( where @xmath75 is the slope at normal flow conditions ) . using the dimensionless governing equations
, we can find @xmath75 and @xmath76 as [ eq:1d_hom ] @xmath77\\ \phi_0=&\frac{\nu_g-\nu_d}{\nu_g}.\end{aligned}\ ] ] note that the equations also allow a trivial solution with @xmath78 which corresponds to a riverbed without vegetation .
this solution becomes the only physically relevant solution in case @xmath79 .
since the aim of this work is to evaluate the influence of vegetation on river patterns , the solution with @xmath78 is not interesting .
the non - zero dimensionless homogeneous solution can finally be summarized as @xmath80 with @xmath75 and @xmath76 as defined in ( [ eq:1d_hom ] ) .
the linearization is done by introducing into the dimensionless equations ( [ eq:1d_dim ] ) the perturbated homogeneous solution @xmath81 with @xmath69 the perturbation parameter and @xmath82 the perturbation ansatz . as we want to look for regular spatial patterns , we choose the perturbation ansatz as @xmath83 with @xmath84 the real dimensionless perturbation wavenumber and @xmath85 the perturbation vector .
while we are used to deal with sinusoidal patterns of velocity , water depth and bed elevation this is less common for vegetation density .
figure [ fig : phi0 ] depicts sinusoidal vegetation density patterns around a mean vegetation density of @xmath76 .
we can see that this formulation only is valid if @xmath76 is larger than the vegetation perturbation amplitude .
in fact , if this is not the case we get locally negative values for vegetation density which does not make sense physically .
so we have to bear in mind that vegetation needs to be well - developed in order for our analysis to be valid .
+ the cosine of equation [ eq:1d_ansatz ] can then be written as @xmath86 it can easily be seen that we get a complex conjugated system of equations when inserting the perturbation ansatz into ( [ eq:1d_dim ] ) .
thus , one can write the perturbated homogeneous solution as @xmath87 where c.c . denotes the complex conjugate .
note that the perturbation term of ( [ eq:1d_perhom ] ) for a given wavenumber @xmath84 is nothing else than one component of the wave packet introduced in ( [ eq : wavepacket ] ) .
then by only keeping the @xmath88 terms we get [ eq:1d_perteq ] @xmath89 the system of equations ( [ eq:1d_perteq ] ) can then be written as @xmath90 where a is the following 4 x 4 matrix : @xmath91 equations ( [ eq:1d_system ] ) and operator ( [ eq:1d_matrix ] ) define a system of ordinary , homogeneous differential equations with constant coefficients which describes the initial ( linear ) temporal evolution of the initially perturbated system . to find general solutions of this system , we have to introduce the concept of a normal operator : an operator is normal if @xmath92 , where @xmath93 is the complex conjugate transpose of @xmath94 .
if @xmath94 was a normal operator , the matrix eigenfunctions would form an orthogonal basis and we could write the general solution as @xmath95 where i is the rank of the matrix ( 4 in this case ) , @xmath96 are coefficients and @xmath97 are the complex eigenvalues of @xmath94 . in the limit of large t , this solution is dominated by the exponential with the largest temporal growth rate ( maximum of the real parts of @xmath97 ) and thus the solution decays to zero if the maximum growth rate is below zero and it diverges for a positive maximum growth rate .
however , in the context of river morphology a is not a normal operator and therefore its eigenfunctions do not form an orthogonal basis .
that is , transient growth occurs ( @xcite ) and ( [ eq:1d_solution ] ) is not generally valid anymore .
however , asymptotically the exponential with the largest real part of the eigenvalues is still going to dominate and thus describes the behavior of the system . as in this work
we are only interested in the long - term behavior of perturbations , we thus can still state that the initially small perturbations will be amplified in the long - term linear regime if the real part of any @xmath97 is positive . and if the largest growth rate occurs for a finite wavenumber @xmath84 , this mode would be amplified stronger than all other modes contained in the wave packet and thus would dominate after some time due to the exponential character of the growth rate .
thus , we can retain the following important points : * the system is stable ( perturbation is not amplified ) with respect to a perturbation mode with wavenumber @xmath84 if @xmath98 * the system is unstable ( perturbation is amplified ) with respect to a perturbation mode with wavenumber @xmath84 if @xmath99 * the system is unstable towards regular spatial patterns if the highest growth rate @xmath100 occurs at finite wavenumber @xmath84 additionally , the phase velocity of a perturbation can be computed using the imaginary part of the eigenvalues as @xmath101 which gives information about the propagation of the perturbation : if @xmath102 then the perturbation propagates downstream and conversely if @xmath103 the perturbation propagates upstream . in this section ,
the results of the stability analysis of matrix a , which was derived in section [ sec:1d_lin ] , are presented and interpreted .
the eigenvalues are calculated numerically and plotted by mathematica for different parameter values while the pattern images are computed using matlab .
additionally , in the simplest case of no vegetation and no sediment transport , the instability condition can be calculated analytically .
the aim of the analysis is to find parameter regions where the fastest growing initial perturbation has a finite wavenumber and thus the system can evolve to a regular pattern upon perturbation . in the following two subsections ,
we first repeat the calculations done by lanzoni et al .
, 2006 for the cases of hydrodynamic equations ( sv ) and hydrodynamic equations coupled with sediment dynamics ( sve ) .
then in [ sec:1d_sta_svv ] and [ sec:1d_sta_svev ] , we analyze the hydrodynamic equations coupled with vegetation dynamics ( svv ) and finally the hydrodynamic equations coupled with sediment and vegetation dynamics ( svev ) .
+ .5 and phase velocity @xmath104 of sv equations as a function of wavenumber @xmath84 for @xmath105 ( orange ) , @xmath106 ( blue ) , @xmath107 ( green ) , @xmath108 ( black ) ; parameter values are @xmath109 and @xmath110,title="fig:",width=283 ] + .5 and phase velocity @xmath104 of sv equations as a function of wavenumber @xmath84 for @xmath105 ( orange ) , @xmath106 ( blue ) , @xmath107 ( green ) , @xmath108 ( black ) ; parameter values are @xmath109 and @xmath110,title="fig:",width=283 ] + the stability analysis of the de saint - venant equations only consists of analyzing a 2 x 2 matrix ( taking the upper left part of matrix a with @xmath111 ) which gives the following characteristic equation for the eigenvalues @xmath112 : @xmath113 solving this equation for @xmath114 , we get the condition @xmath107 , independently from @xmath115 and @xmath84 .
figure [ fig_sv1 ] shows the temporal growth rate for different values of @xmath116 .
for @xmath107 the growth rate is zero . if @xmath108 however , the growth rate increases asymptotically with increasing @xmath84 while for @xmath117 it is always negative.this means that perturbations are amplified only if @xmath118 and that the most unstable mode is the one where the wavenumber @xmath84 tends to infinity and so the wavelength is equal to zero . according to @xcite this instability is linked to the formation of roll waves .
but , in the linear regime no instability towards regular patterns with finite wavelength is possible in a river with fixed bed and without riverbed vegetation .
+ additionally , figure [ fig_sv2 ] shows that perturbations propagate downstream only if the flow is subcritical and in both directions if flow is supercritical .
if sediment dynamics is added to the de saint - venant equations ( which means that the bed material may be transported by the flow ) , the eigenvalues of a 3 x 3 matrix have to be computed ( upper left part of a ) .
it turns out that if the morphodynamic timescale is small compared to the hydrodynamic timescale which is normally the case ( @xmath119 , see @xcite for some values ) , the first two ( hydrodynamic ) modes are essentially the same as in the previous paragraph .
there is however a third mode ( called morphodynamic mode ) that appears because of sediment dynamics .
the temporal growth rate and the phase velocity are depicted in figure [ fig_sve ] for the morphodynamic mode only ( scaled by @xmath120 ) .
+ the growth rate of the morphodynamic mode is below zero ( equal to zero at k=0 ) for all values of @xmath116 and @xmath84 which means that the morphodynamic mode is ( as were the hydrodynamic modes ) not able to produce instability towards finite patterns . finally , as we can see from figure [ fig_sve4 ]
the migration of the perturbations is downstream if the flow is subcritical and upstream if the flow is supercritical .
all results of paragraphs [ sec:1d_sta_sv ] and [ sec:1d_sta_sve ] are in agreement with the findings of @xcite which confirms the correctness of the stability analysis performed .
.5 and phase velocity @xmath104 of the morphodynamic mode of sve equations as a function of wavenumber @xmath84 for @xmath105 ( orange ) , @xmath106 ( blue ) , @xmath107 ( green ) , @xmath108 ( black ) ; parameter values are @xmath109 , @xmath110 and @xmath121,title="fig:",width=283 ] .5 and phase velocity @xmath104 of the morphodynamic mode of sve equations as a function of wavenumber @xmath84 for @xmath105 ( orange ) , @xmath106 ( blue ) , @xmath107 ( green ) , @xmath108 ( black ) ; parameter values are @xmath109 , @xmath110 and @xmath121,title="fig:",width=283 ] the effect of taking sediment dynamics into account was shown in the previous section .
now , we analyze the de saint - venant equations coupled with vegetation dynamics ( but with a fixed bed geometry ) .
again , no analytical solution is available and thus we analyze @xmath122 numerically for different parameters .
+ .values for constant parameters of the analysis [ cols="^,^,^,^ " , ] [ table:2d_sve ] and aspect ratio @xmath123 : no instability ( red ) , alternate bars ( light blue ) , multiple bars ( darker blues for increasing bar order @xmath124 ) ; for parameter values see table [ table:2d_sve],title="fig:",width=491 ] + it was also observed in nature and found using linear stability theory ( @xcite ) that the higher the aspect ratio of a river the higher the bar order ( number of bars in the transverse direction ) is .
figure [ fig:2d_sve3 ] depicts parameter domains where alternate bar and multiple bar regimes respectively dominate ( based on the highestl growth rate ) .
the aspect ratio @xmath123 seems to be the decisive parameter in a reasonable range of froude numbers between 1 and 2 .
however , the froude number for flooding in the marshall river is between 0.3 and 0.4 according to @xcite and thus @xmath125 is also important to determine the bar regime of a river .
and , once the froude number exceeds a certain maximum value , instability towards bar formation does no longer exist .
overall , we can say that results of earlier stability analyses of bar instability could be reproduced in this work .
bar instability triggered by 2-dimensional sediment dynamics is sensitive to a river s aspect ratio ( width to depth ratio ) and also to froude number for low values of @xmath125 .
finally , it is worth noting that the term accounting for gravitational effects of a weak lateral slope ( second term of ( [ eq : weak_slope ] ) ) is crucial in order to reproduce this well - known result .
@xcite was the first to propose this relation which was later confirmed experimentally by @xcite with both suggesting the parameter @xmath126 to be between 0.5 and 0.6 . for different froude numbers , the black dots mark the maximum of each curve ; parameters used : @xmath127 , @xmath128 , @xmath129 , @xmath130 , @xmath131 , @xmath132 , @xmath133 and table [ table_param ] + , title="fig:",width=491 ] + as it was done in the analysis of the 1-dimensional equations , we want to analyze separately the effect of vegetation dynamics in order to better understand its potential contribution to pattern formation .
thus , a fixed bed is presumed and the eigenvalues of equation ( [ eq:2d_matrix ] ) removing line 4 and column 4 are analyzed in the following .
+ first , we want to look at instability towards alternate bars ( m=1 ) and we can see in figure [ fig:2d_svv_f ] that , similarly to the 1d analysis , we can find the maximum growth rate to be at finite longitudinal wavenumber @xmath134 for a certain range of froude numbers , which means that instability towards finite patterns exists . yet , as explained before , in the 2d - case we are also interested in patterns with @xmath135 due to their similarity with channels in the marshall river .
such longitudinally homogeneous patterns seem to occur at slightly higher froude numbers than patterns with finite @xmath134 , but not too high in order to still allow interaction between vegetation growth and mortality .
note that for the moment longitudinal as well as lateral vegetation diffusion ( seeding and resprouting ) are put to zero .
+ and aspect ratio @xmath123 : the color code indicates relative growth rate and a value of @xmath136 ( red ) means that no patterns exist ; the black line shows the maximum growth rate for given @xmath123 and thus indicates the value of the dominating longitudinal wavenumber ; parameter values are @xmath127 , @xmath137 , @xmath129 , @xmath138 , @xmath131 , @xmath132 , @xmath133 and values indicated in table [ table_param],title="fig:",width=491 ] + and aspect ratio @xmath123 : the color code indicates relative growth rate and a value of @xmath136 ( red ) means that no patterns exist ; the black line shows the maximum growth rate for given @xmath123 and thus indicates the value of the dominating longitudinal wavenumber ; parameter values are @xmath127 , @xmath137 , @xmath129 , @xmath130 , @xmath131 , @xmath139 , @xmath140 and values indicated in table [ table_param],title="fig:",width=491 ] + so at first glance , vegetation dynamics behaves quite alike in a 2d model than in the 1d one . to compare the instability created by vegetation dynamics to the one induced by sediment dynamics
, we can have a look at figure [ fig:2d_svv1 ] .
surprisingly , the pattern domain features some characteristics that are quite similar to the one depicted in figure [ fig:2d_sve1 ] : a minimum value is required for the aspect ratio @xmath123 and the dominating longitudinal wavenumber increases with increasing aspect ratio .
however , as visible in both figures [ fig:2d_svv_f ] and [ fig:2d_svv1 ] the maximum growth rate tends to occur at considerably higher @xmath134 for vegetated rivers which could be thought to be physically unrealistic ( too short pattern wavelength could undermine the hypothesis of shallow water equations ) .
one has to bear in mind though that in the 2d model length scales are normalized with respect to half - river - width @xmath141 and not normal water depth @xmath57 .
thus , if we take @xmath142 , @xmath128 and @xmath109 for example we get @xmath143 which is much smaller than river width @xmath144 but still in a reasonable order of magnitude . by tuning the parameters
, we can quite easily get wavelengths that make more sense .
for instance if we set the the vegetation diffusion rate @xmath145 we get dimensionless longitudinal wavenumbers on the order of 3 to 4 for @xmath128 .
this leads to a physical longitudinal wavelength between 30 and 40 meters which is close to the actual river width and thus much more realistic ( compare figure [ fig:2d_svv1_d ] to figure [ fig:2d_svv1 ] to see the influence of vegetation diffusion on the dominating wavenumber ) .
we conclude that the inclusion of vegetation diffusion contributes to a more physically realistic result and we thus keep this value constant in the following analyses , keeping in mind though that it is not an absolutely indispensable part of the pattern producing mechanism .
+ and aspect ratio @xmath123 : no instability ( red ) , alternate bars ( light blue ) , multiple bars ( darker blues ) do not occur ; parameter values are @xmath129 , @xmath130 , @xmath146 , @xmath139 , @xmath140 and values indicated in table [ table_param ] + , title="fig:",width=491 ] + and aspect ratio @xmath123 : the color code indicates the value of the most unstable longitudinal wavenumber @xmath134 , negative numbers ( red ) mean no instability ; parameter values are @xmath127 , @xmath129 , @xmath130 , @xmath131 , @xmath139 , @xmath140 and values indicated in table [ table_param ] + , title="fig:",width=491 ] + and aspect ratio @xmath123 : the color code indicates the value of the most unstable longitudinal wavenumber @xmath134 , negative numbers ( red ) mean no instability ; parameter values are @xmath147 , @xmath129 , @xmath130 , @xmath131 , @xmath139 , @xmath140 and values indicated in table [ table_param ] + , title="fig:",width=491 ] + once we have seen the similarities of alternate bars inducing mechanisms of sediment and vegetation dynamics , we would like to know if that is still true for the formation of multiple bars .
thus , we repeat the multiple bar analysis of figure [ fig:2d_sve3 ] including vegetation dynamics instead of sediment dynamics . the result can be seen in figure [ fig:2d_svv3 ] : as expected , only a small range of froude numbers allows pattern formation which is due to vegetation growth balance ( as explained in section [ sec:1d_sta_svv ] ) .
surprisingly though , as opposed to the results of sediment dynamics , alternate bars grow always faster than multiple bars in the linear regime .
this means that a fixed riverbed that is under the influence of vegetation dynamics only does not tolerate instability towards multiple bars .
however , a number of vegetated patterns that occur in nature exhibit multiple bars ( up to 10 for the marshall river , even more in the case of rills on fluvial bars , see figure [ fig : anabranch ] ) .
this apparent contradiction of the model and reality could still be due to the fact that sediment transport , whose influence will be analyzed in the next section , was not considered until here .
+ figure [ fig:2d_svv3 ] only shows which kind of bars are occurring for a certain parameter configuration , but it does not give any information about the dominant longitudinal wavenumber .
we could also wonder if in reality there is no instability towards multiple bars at all since we only can see the alternate bar domain in the former figure . to answer these questions , we can have a look at figures [ fig:2d_svv3_1 ] and [ fig:2d_svv3_4 ] which show the instability domain and most unstable longitudinal wavenumber for the case of alternate bars ( m=1 ) and multiple bars ( m=4 ) respectively .
we can see that instability towards multiple bars does indeed exist but its growth rate being always smaller than the growth rate of alternate bars we can not perceive it in figure [ fig:2d_svv3 ] .
an interesting feature that is visible on both figures is that about half the domain seems to have the dominating longitudinal wavenumber equal to 0 which means that riverbed vegetation ( and also flow and depth ) are longitudinally homogeneous in this domain .
+ we then want to have a closer look at the vegetation growth balance which at first seems to be quite similar than in the 1-dimensional model .
indeed , the @xmath116 versus @xmath4 plot in figure [ fig:2d_svv_phi_old ] strongly resembles what we have seen before for vegetation dynamics in a 1d river . to the right ( in orange ) , there is a domain where no physically possible solutions can exist while in the middle there is a domain ( blue ) where vegetation growth and death through uprooting are balanced to allow formation of vegetation patterns at finite longitudinal wavenumber .
the only major difference lies in the fact that the domain with dominating longitudinal wavenumber @xmath134 equal to zero ( green ) may be interpreted as a pattern forming domain due to the lateral wavenumber being finite ( as explained before ) .
this domain with instability towards longitudinally homogeneous patterns ( @xmath135 ) is slightly wider than the corresponding one we saw in the 1d - analysis ( i.e. small red band in figure [ fig_pattern_old ] ) which is due to the diffusion coefficients @xmath148 and @xmath149 being put to a non - zero value in the 2d - analysis .
the fact that in the 2d - analysis the pattern domain directly borders on the domain with non - physical solution means that this domain boundary can be given analytically using the condition @xmath150 , which yields : @xmath151 and vegetation carrying capacity @xmath4 : the color code indicates the value of the most unstable longitudinal wavenumber @xmath134 , negative numbers ( red ) mean no instability ; parameter values are @xmath127 , @xmath128 , @xmath130 , @xmath131 , @xmath139 , @xmath140 and values indicated in table [ table_param],title="fig:",width=491 ] + we can also understand now why in the 1d analysis increasing the vegetation diffusion coefficient lead to a decreasing pattern domain ( see figure [ fig_pattern_d ] ) .
in fact , increasing the diffusion coefficient increases the part of the domain with the dominant longitudinal wavenumber @xmath134 equal to zero .
but , in the 1d - context , a longitudinal wavenumber equal to zero means that no patterns exist since no lateral variability is possible . in contrast to figure [ fig_pattern_d ] , if we plotted @xmath116 versus the diffusion coefficient for the 2d - analysis , we would just get two straight boundaries at constant froude number .
+ equation ( [ eq : froude_boundary ] ) also allows us to calculate the boundary at the higher froude number of figures [ fig:2d_svv3_1 ] and [ fig:2d_svv3_4 ] ( boundary is independent of bar order ) : @xmath152 furthermore , figure [ fig:2d_svv_phi_old ] shows that longitudinally homogeneous patterns ( vegetated , longitudinal channels ) only occur at very low relative vegetation density .
we write relative density because @xmath76 is normalized using @xmath4 which means that in case @xmath4 is large the real vegetation density does not necessarily have to be small .
figure [ fig:2d_svv_phi_c ] shows the dominating longitudinal wavenumber in the @xmath153 space and we can see that the same conclusions regarding vegetation balance are true than for the 1d - analysis : the largest longitudinal wavenumber @xmath134 occurs if growth and mortality through uprooting are well balanced , larger @xmath134 occur for larger values of froude number and vegetation carrying capacity . concluding the analysis of the vegetation growth balance
, we can say that , as in the 1d model , the pattern domain seems to be simply connected ( one domain without holes ) and it continues to open up as @xmath4 goes to infinity .
instability towards alternate bars could be detected , but not towards multiple bars .
+ and aspect ratio @xmath123 : the color code indicates relative growth rate and a value of @xmath136 ( red ) means that no patterns exist ; the black line shows the maximum growth rate for given @xmath123 and thus indicates the value of the dominating longitudinal wavenumber ; parameter values are @xmath127 , @xmath137 , @xmath121 , @xmath129 , @xmath130 , @xmath131 , @xmath139 , @xmath140 and values indicated in tables [ table_param ] and [ table:2d_sve],title="fig:",width=491 ] + and aspect ratio @xmath123 : the color code indicates relative growth rate and a value of @xmath136 ( red ) means that no patterns exist ; the black line shows the maximum growth rate for given @xmath123 and thus indicates the value of the dominating longitudinal wavenumber ; parameter values are @xmath147 , @xmath137 , @xmath121 , @xmath129 , @xmath130 , @xmath131 , @xmath139 , @xmath140 and values indicated in tables [ table_param ] and [ table:2d_sve],title="fig:",width=491 ] + and aspect ratio @xmath123 : the color code indicates relative growth rate and a value of @xmath136 ( red ) means that no patterns exist ; the black line shows the maximum growth rate for given @xmath123 and thus indicates the value of the dominating longitudinal wavenumber ; parameter values are @xmath127 , @xmath154 , @xmath121 , @xmath155 , @xmath130 , @xmath131 , @xmath139 , @xmath140 and values indicated in tables [ table_param ] and [ table:2d_sve],title="fig:",width=491 ] + after having studied separately the effects on 2-dimensional river pattern formation of sediment dynamics and vegetation dynamics , we finally move on to the instability analysis of the complete 2d - model which includes sediment and vegetation dynamics and thus represents a river with movable bed and vegetation coverage .
we saw before that sediment dynamics as well as vegetation dynamics are able to induce pattern formation in certain domains of the parameter space .
we now want to know which of these effects remain or whether they even combine to form something not seen in the analysis of either sediment or vegetation dynamics alone .
+ we start with looking at figure [ fig:2d_svev1 ] ( @xmath137 ) which indeed shows positive growth rates for a range of aspect ratio @xmath123 .
clearly , this pattern domain seems to be a superposition of figures [ fig:2d_sve1 ] ( note that this figure is with @xmath105 instead of @xmath137 ) and [ fig:2d_svv1 ] with sediment influenced positive growth rates to the left and vegetation influenced ones to the right .
both parts of the pattern domain have a lower boundary ( minimum aspect ratio @xmath123 ) , but the domain allegedly created by sediment dynamics has positive growth rates at lower longitudinal wavenumbers ( higher wavelenghts ) than vegetation dynamics .
figure [ fig:2d_svev1_4 ] shows the same phenomena for multiple bars ( m=4 ) . in both figures
, we can see that for these parameters ( see captions ) sediment dynamics does determine the dominating longitudinal wavenumber @xmath134 ( black line ) for lower values of aspect ratio @xmath123 while vegetation dynamics is dominant at higher aspect ratios ( around 10 for alternate bars and around 40 for multiple bars in figure [ fig:2d_svev1_4 ] ) . and aspect ratio @xmath123 : the color code indicates the value of the most unstable longitudinal wavenumber @xmath134 , negative numbers ( red ) mean no instability ; parameter values are @xmath127 , @xmath121 , @xmath129 , @xmath130 , @xmath131 , @xmath139 , @xmath140 and values indicated in tables [ table_param ] and [ table:2d_sve ] + , title="fig:",width=491 ] + and aspect ratio @xmath123 : the color code indicates the value of the most unstable longitudinal wavenumber @xmath134 , negative numbers ( red ) mean no instability ; parameter values are @xmath127 , @xmath121 , @xmath155 , @xmath130 , @xmath131 , @xmath139 , @xmath140 and values indicated in tables [ table_param ] and [ table:2d_sve ] + , title="fig:",width=491 ] + then , if we take a much lower value for @xmath116 ( along with a lower vegetation carrying capacity of @xmath156 ) we get a completely different picture which is shown in figure [ fig:2d_svev1_f01 ] : all of a sudden , the pattern domain completely changes and we have only one domain that occurs with a lower and upper limit for the aspect ratio as well as a limited domain for longitudinal wavenumber that .
figure [ fig:2d_svev3 ] depicts the same situation from another angle , namely in the @xmath116-@xmath123 space .
we can identify one single parameter domain leading to instability including a lower limit for a certain value for @xmath123 .
however , if we reduce vegetation carrying capacity ( thus decreasing the influence of vegetation ) , we can see that two different instability domains appear in figure [ fig:2d_svev3_phi ] which was not the case in the 1d analysis .
not surprisingly though , the larger domain to the right resembles strongly the vegetation induced domain already seen in section [ sec:2d_sta_svv ] which also means that the domain to the left should probably be due to sediment dynamics . comparing figures [ fig:2d_svev3 ] and [ fig:2d_svev3_phi ]
, we can also see that the domain at larger froude numbers decreases and is slightly more to the left in the latter figure .
this is another hint that the right domain comes from vegetation dynamics : when decreasing vegetation carrying capacity , the froude number has to decrease as well otherwise the river s uprooting capacity would overwhelm vegetation growth .
+ and aspect ratio @xmath123 : the color code indicates the value of the most unstable longitudinal wavenumber @xmath134 , negative numbers ( red ) mean no instability ; parameter values are @xmath147 , @xmath121 , @xmath129 , @xmath130 , @xmath131 , @xmath139 , @xmath140 and values indicated in tables [ table_param ] and [ table:2d_sve ] + , title="fig:",width=491 ] + next , we would like to know what happens to instability towards multiple bars . in the previous sections , we could show that instability towards multiple bars exists for flow dynamics coupled with sediment dynamics but not for flow dynamics coupled with vegetation dynamics .
figure [ fig:2d_svev3_phi_4 ] shows the domains and dominating longitudinal wavenumber for multiple bars ( m=4 ) .
we can see that both domains are slightly shifted upwards ( towards positive @xmath123 ) which was already seen before in the analysis with sediment dynamics .
moreover , higher order multiple bars develop higher longitudinal wavenumbers and thus shorter longitudinal wavelengths than alternate bars .
+ one important question remains though : which bar order will dominate in parameter domains where alternate bars and several orders of multiple bars can potentially exist ?
figure [ fig:2d_svev3_mult ] answers this question partially by showing that in the case of a movable bed with vegetation the domain to the right is always unstable towards alternate bars .
sediment dynamics induced patterns are not visible due to vegetation processes completely dominating river bed dynamics .
figure [ fig:2d_svev3_mult_phi10 ] shows what happens in not very highly vegetated riverbeds ( @xmath156 ) . as in the case of alternate bars ( figure [ fig:2d_svev3_phi ] ) ,
there is sediment induced instability to the left , but the domain to the right seems to contain both sediment incued instability ( towards higher order of multiple bars with increasing @xmath123 ) and vegetation induced instability to the very right .
this means that instability towards multiple bars in vegetated rivers is indeed possible , but only at rather low froude numbers ( either at about f=0.2 - 0.3 or at f=0.6 - 0.7 in this case ) . yet ,
as it can be seen in figures [ fig:2d_svev3 ] to [ fig:2d_svev3_phi_4 ] , instability towards patterns with very low longitudinal wavenumbers occurs only in the domain at higher froude number . at very low froude numbers ,
only rather high wavenumbers are reached in the asymptotic limit and thus , longitudinal channels ( which are characterized by @xmath134 close to zero ) do not occur . since the froude number for flooding events in the marshall river is rather low ( 0.3 - 0.4 , @xcite )
, our results do not predict channels for this river which is contrary to reality .
however , for not very highly vegetated riverbeds ( @xmath155 in figure [ fig:2d_svev3_mult_phi10 ] ) multiple bars may occur at froude numbers up to about 0.7 .
in such a parameter configuration , we can thus have instability towards multiple bars for the marshall river , but at rather high longitudinal wavenumbers , thus resembling more a braiding pattern than a channeled riverbed .
+ and aspect ratio @xmath123 : no instability ( red ) , alternate bars ( light blue ) , multiple bars ( darker blues for increasing bar order @xmath124 ) ; parameter values @xmath121 , @xmath129 , @xmath130 , @xmath131 , @xmath139 , @xmath140 and values indicated in tables [ table_param ] and [ table:2d_sve ] + , title="fig:",width=491 ] + and aspect ratio @xmath123 : no instability ( red ) , alternate bars ( light blue ) , multiple bars ( darker blues for increasing bar order @xmath124 ) ; parameter values @xmath121 , @xmath155 , @xmath130 , @xmath131 , @xmath139 , @xmath140 and values indicated in tables [ table_param ] and [ table:2d_sve ] + , title="fig:",width=491 ] + interestingly , figure [ fig:2d_svev3_phi ] indicates that with decreasing vegetation carrying capacity @xmath4 the two parts of the domain move closer together which implies that they eventually could merge once @xmath4 falls below a certain threshold .
we want to have a closer look at this by plotting @xmath4 against froude number in figure [ fig:2d_svev_phi_c ] .
although it is difficult to separate sediment induced instability from vegetation induced , this can be done when the current figure is compared to figure [ fig:2d_svv_phi_c ] .
the sediment domain consists of the thickening of the domain to the very left as well as the thin greenish domain that joins the vegetation induced domain at the left .
then , it can be seen that the domains actually merge when @xmath4 is low enough .
this means that although the pattern domain as a whole is not simply connected anymore ( there are holes in the domain ) it is still is connected .
+ to conclude , we again plot vegetation carrying capacity @xmath157 against @xmath116 in figure [ fig:2d_svev_phi_old ] and include contour lines for dimensionless homogeneous vegetation density @xmath76 ( in black ) as well as physical homogeneous vegetation density @xmath157 ( plants per @xmath158 , in yellow ) . as explained earlier , we need well - developed vegetation ( meaning @xmath76 well above zero ) in order to not have negative vegetation density because of the sinusoidal oscillations .
this means that the vegetation density wave amplitude always has to be smaller than the actual vegetation density .
we do nt have a way to know the wave amplitude , but still at least we have to wonder whether the model is valid for the part of the pattern domain close to the line @xmath78 .
this would mean that the part of the domain where the dominant @xmath135 could be actually non - physical due to the assumptions not met .
anyway , such patterns with dominant longitudinal wavenumber equal to zero would be alternate bars ( remember that alternate bars always grow faster than multiple bars in this region of pattern domain ) .
this would result in an asymmetric channel where either the left or the right side would be filled with sediment while water is flowing on the other side .
thus , what makes actually physically sense in figure [ fig:2d_svev_phi_c ] is instability towards alternate bars with finite @xmath134 at higher froude numbers and instability towards multiple bars with rather high @xmath134 at lower froude numbers .
+ we now quickly want to give an estimate for a longitudinal wavenumber of multiple bars . looking at figure [ fig:2d_svev3_phi_4 ] for example , we can see that a typical longitudinal wavenumber is @xmath159 for m=4 in the region where these multiple bars dominate .
this yields for @xmath160 @xmath161 which is not completely unreasonable as an order of magnitude , but too low in comparison of a river width of @xmath162 .
thus , the parameter values of the model would need to be checked and investigated order to get a more reasonable result .
+ and vegetation carrying capacity @xmath4 : the color code indicates the value of the most unstable longitudinal wavenumber @xmath134 , negative numbers ( red ) mean no instability ; parameter values are @xmath127 , @xmath163 , @xmath121 , @xmath130 , @xmath131 , @xmath139 , @xmath140 and values indicated in tables [ table_param ] and [ table:2d_sve],title="fig:",width=491 ] + finally , we want to find out whether there is evidence that the pattern domain at low froude numbers might be linked to actual vegetation density in the river .
we can see on figure [ fig:2d_svev_phi_old ] that the domain to the left can best be characterized by a very high @xmath76 ( black lines , see figure [ fig:2d_svv_phi_old ] for values ) .
this is astonishing to some extent since this domain is thought to be governed by sediment dynamics .
but then again , a very high relative ( dimensionless ) vegetation density does not necessarily mean a lot of variation .
so , one could describe this domain to be representative for rivers with stable vegetation density close to carrying capacity .
the vegetation would not be influenced much by flow due to low uprooting capacity ( low froude number ) and sediment dynamics would thus be governing the river s instability mechanisms in this region of the parameter space as we suspected in the beginning .
we reproduced known results of river instability towards alternate and multiple bars using a 2-dimensional de saint - venant - exner framework .
as expected , the higher a river s width - to - depth ratio ( aspect ratio @xmath123 ) the more bars a river tends to develop laterally , leading to the formation of multiple bars .
the froude number does has a crucial role at low f and no patterns exist above roughly @xmath107 . when analyzing the 2d - de saint - venant equation combined with vegetation dynamics , we find , similarly to the 1-dimensional model , that there exists a domain where vegetation growth and mortality by means of uprooting compete and thus instability towards finite patterns prevails .
however , a minimum value for the aspect ratio is required to induce such instability and only instability towards alternate bars occurs ( the exponential growth rate of alternate bars in the linear regime always exceeds the growth rate of multiple bars of any order ) .
the froude number , which is directly proportional to stream velocity , is very important to balance vegetation dynamics since a higher velocity increases the river s uprooting capacity ( see equation ( [ eq : veg3 ] ) ) .
+ when analyzing the full model including sediment as well as vegetation dynamics , a pattern domain of essentially two parts is detected .
one part occurs at low froude numbers and high dimensionless vegetation density ( independently of actual vegetation carrying capacity ) and mainly possesses features of sediment transport induced instability : independence of froude number , multiple bar order increases with increasing aspect ratio and higher dimensionless wavenumber for higher aspect ratios .
the other part stems from vegetation growth balance and inherits equally its attributes : domain highly dependent on froude number , instability towards alternate bars with rather low longitudinal wavenumbers at low vegetation density , but no instability towards multiple bars .
the two parts of the pattern domain are separated for higher values of vegetation carrying capacity @xmath4 and linked when @xmath4 falls below a certain threshold ( which depends on the other vegetation parameters @xmath164 ) .
the goal of this work was to shed light on the influence of riparian vegetation on formation of morphological river patterns .
we thus performed a linear stability analysis on the 1d and 2d ecomorphodynamic equations which include analytical models for flow , sediment as well as vegetation dynamics .
the vegetation model was kept very simple in order to be suitable for a stability analysis and it included terms for vegetation growth , diffusion through seeding / resprouting and mortality by means of uprooting caused by flow shear . at first , this equation was developed for rivers with constant flow , but was shown to also apply ( under certain conditions ) to variable flow and even ephemeral rivers .
+ our analysis of the 1d model showed that vegetated rivers indeed exhibit instability towards longitudinal sediment waves due to the competitive interaction between vegetation growth and mortality .
then , instability towards river patterns with lateral structure ( bars ) was assessed using the 2d ecomorphodynamic equations and it was discovered that two different kinds of instabilities occur .
instability at lower froude numbers is mainly driven by sediment dynamics and leads to formation of alternate and multiple bars with the bar order increasing with increasing river width - to - depth ratio ( aspect ratio ) . at higher froude numbers ,
only instability towards alternate bars was detected , independently of the aspect ratio .
+ we also looked at the value of the most unstable longitudinal pattern wavenumber .
generally , it was found that the values identified were reasonable ( corresponding wavelength has the same order of magnitude than river width ) although sometimes at the higher limit of what is allowed by the model assumptions .
however , we were not able to identify instability towards longitudinally infinite multiple channels as it occurs in some reaches of the marshall river ( figure [ fig : anabranch ] b ) .
actually , the longitudinal wavenumbers found for multiple bar instability at low froude numbers were too high to form such channels .
other parameter configurations or more detailed modeling would probably be required to match reality in this case .
+ in general , experimental verification of the equation adopted and research on the parameter values would be needed to adjust our purely theoretical model to reality .
in addition , flume experiments could also help to quantify other aspects of vegetation and sediment transport interaction that we did not take into account .
for instance , we only model the effect of vegetation on sediment transport indirectly by increasing the river s bed roughness . yet , other processes like scouring that increases sediment ablation around plants ( see @xcite for a scouring model around bridge piers ) and riverbed stabilization by plant s root systems are probably important as well but too difficult to model analytically at this stage .
further improvement of the vegetation model could include finding expressions for vegetation uprooting by gradual exposure of plants root system ( type ii mechanism of @xcite ) and flow diversion produced mainly by rigid vegetation .
+ finally , since we were dealing with a non - normal operator in our stability analysis , transient growth of the system can occur which was not considered in the present work .
therefore , as was done by @xcite for the morphodynamic equations , a non - modal analysis of our linear ecomorphodynamic operator is conceivable in the future to evaluate the importance of such transient growths .
the reason is that sometimes this transient behavior can actually be more relevant in reality than the asymptotic fate depending on the timescale of interest .
moreover , in addition to performing a stability analysis which only takes into account growth or decay at the linear level , we could extend our research by adding a non - linear numerical simulation of the initial perturbations .
in fact , as the perturbations amplify non - linearities of the system may become important and eventually dominate , thus determining the asymptotic fate of the system . in any case , all these possibilities of improvement of current modeling of interaction of riparian vegetation and river morphology show that we are still barely scratching the surface of this complex subject .
baptist m. , babovic v. , rodriguez uthuruburu j. , keijer m. , uittenbogaard r. e. , mynett a. , verwey a. 2007 . on inducing equations for vegetation resistance .
_ journal hydr .
45(4 ) : 435 - 450 .
blondeaux p. , seminara g. 1985 .
a unified bar - bend theory of river meanders .
_ j. fluid mech .
449 - 470 .
callander r.a .
instability and river channels .
_ j. fluid .
36 , part 3 , pp .
465 - 480 .
camporeale c. , perucca e. , ridolfi l. , gurnell a. m. 2013 .
modeling the interactions between river morphodynamics and riparian vegetation .
_ reviews of geophysics _ 51:1 - 36 .
camporeale c. , ridolfi l. 2009 .
nonnormality and transient behavior of the de saint - venant - exner equations .
_ water resources research _ vol.45 , w08418 .
chiodi f. , andreotti b. , claudin p. 2012 .
the bar instability revisited .
_ j. fluid .
_ , under consideration colombini m. , seminara g. , tubino m. 1987 .
finite - amplitude alternate bars , _ j. fluid mech .
213 - 232 .
edmaier k. , burlando p. , perona p. 2011 .
mechanisms of vegetation uprooting by flow in alluvial non - cohesive sediment .
earth syst .
_ 15:1615 - 1627 .
engelund f. 1981 .
the motion of sediment particles on an inclined bed . _ tech
denmark isva prog .
15 - 20 . engelund f. , skovgaard o. 1973 . on the origin of meandering and braiding in alluvial streams .
_ j. fluid mech .
57 , part 2 , pp .
289 - 302 . federici b. and paola chris .
dynamics of channel bifurcations in noncohesive sediments .
_ water resources research _ vol .
39 , no . 6 , 1162 .
federici b. and seminara g. 2003 . on the convective nature of bar instability .
_ j. fluid mech .
_ vol.487 , pp .
125 - 145 .
gurnell a. , petts g. e. 2006 .
trees as riparian engineers : the tagliamento river , italy . _ earth surface processes and landforms _ 31:1558 - 1574 .
jansen j. , nanson g. c. 2010 .
functional relationship between vegetation , channel morphology and flow efficiency in an alluvial ( anabranching ) river .
_ j. geoph .
res . _ 115,f07441 .
lanzoni s. , siviglia a. , frascati a. , seminara g. 2006 .
long waves in erodible channels and morphodynamic influence .
_ water resources research _ vol.42 , w06d17 . melville b. w. , sutherland a. j. 1988 .
design method for local scour at bridge piers
. _ j. hydraul .
_ 114:1210 - 1226 .
murray a. b. , paola c. 2003 .
modelling the effect of vegetation on channel pattern in bedload rivers . _ earth surface processes and landforms _ 28 , 131 - 143 .
parker g. 1976 .
on the cause and characteristic scales of meandering and braiding in rivers .
_ j. fluid mech .
76 , part 3 , pp .
457 - 480 .
pasquale n. , perona p. , schneider p. , shrestha j. , wombacher a. , burlando p. 2011 .
modern comprehensive approach to monitor the morphodynamic evolution of a restored river corridor .
earth syst .
_ 15 , 1197 - 1212 .
perona p. , crouzy b. , mclelland s. , molnar p. , camporeale c. 2014 .
ecomorphodynamics of rivers with converging boundaries . _ earth surface processes and landforms _ , in press .
perucca e. , camporeale c. , ridolfi l. 2007 .
significance of the riparian vegetation dynamics on meandering river morphodynamics .
_ water resources research _ vol .
43 , w03430 .
seminara g. 2010 .
fluvial sedimentary patterns .
fluid mech .
_ 42,43 - 66 .
talmon a.m. , struiksma n. , van mierlo m.c.l.m .
laboratory measurements of the direction of sediment transport on transverse alluvial - bed slopes .
_ journal of hydraulic research _ vol .
4 . tooth s. , nanson g. c. 2000 . forms and processes of two highly contrasting rivers in arid central australia , and the implications for channel pattern discrimination and prediction .
_ geological society of america bulletin _ 116(7 - 8 ) , 802 - 816 .
turing a. m. 1952 the chemical basis of morphogenesis . _
philosophical transactions of the royal society of london _ series b , biological sciences 237(641 ) , 37 - 72 .
wu w. , shields jr .
f. d. , bennett s. j. , wang s. s. y. 2005 . a depth - averaged two - dimensional model for flow , bed transport and sediment topography in curved channels with riparian vegetation . _ water resources research _ vol 41 , w03015 . | although riparian vegetation is present in or along many water courses of the world , its active role resulting from the interaction with flow and sediment processes has only recently become an active field of research .
especially , the role of vegetation in the process of river pattern formation has been explored and demonstrated mostly experimentally and numerically until now . in the present work
, we shed light on this subject by performing a linear stability analysis on a simple model for riverbed vegetation dynamics coupled with the set of classical river morphodynamic equations .
the vegetation model only accounts for logistic growth , local positive feedback through seeding and resprouting , and mortality by means of uprooting through flow shear stress . due to the simplicity of the model
, we can transform the set of equations into an eigenvalue problem and assess the stability of the linearized equations when slightly perturbated away from a spatially homogeneous solution .
if we couple vegetation dynamics with a 1d morphodynamic framework , we observe that instability towards long sediment waves is possible due to competitive interaction between vegetation growth and mortality .
moreover , the domain in the parameter space where perturbations are amplified was found to be simply connected .
subsequently , we proceed to the analysis of vegetation dynamics coupled with a 2d morphodynamic framework , which can be used to evaluate instability towards alternate and multiple bars .
it is found that two kinds of instabilities , which are discriminated mainly by the froude number , occur in a connected domain in the parameter space . at lower froude number ,
instability is mainly governed by sediment dynamics and leads to the formation of alternate and multiple bars while at higher froude number instability is driven by vegetation dynamics , which only allows for alternate bars . |
we obtain a nearly model - independent description of the isgur - wise ( iw ) function @xcite , @xmath5 , in terms of a single measurable parameter the slope at zero - recoil , @xmath1 .
only modest assumptions about the shape of the heavy - light wavefuction are required , and these are consistent with an established lattice result .
we obtain a simple functional form for the iw function in terms of @xmath1 : @xmath6 ^ 2 } \,.\ ] ] we also demonstrate that the effect of radiative qcd corrections and @xmath7 corrections can be described as the product of a term linear in @xmath8 and the iw function .
the resulting function is also approximately an iw function of shifted slope .
thus , fitting the above functional form to the measured form factors provides an accurate description of the extrapolation to zero - recoil . over the range of expected slopes ,
our result is in good agreement with the work of @xcite which is based on dispersion relations .
recent analyses @xcite of @xmath9 and @xmath10 decays extract two physical form factors and two ratios of physical form factors .
we show that this information can be directly related to qcd predictions .
in section [ intro ] we provide some background formalism behind form factors for semileptonic @xmath3-decay and the isgur - wise function .
we present our description of the iw function in terms of the slope parameter , @xmath1 , in section [ iw ] . in section [ ff ]
we discuss the effects of corrections to the heavy - quark limit and their equivalence to slope - shifted iw functions .
the relationship of hqet and qcd lattice simulation to semileptonic data is explored in section [ qcd ] .
as the observed sample of semileptonic @xmath3 decays accumulates , the need for a rigorous method of analysis becomes more pressing . even in the heavy quark limit
, a universal form factor is required for each light degree of freedom state of the meson . to compute the isgur - wise ( iw ) functions @xmath5 , where @xmath11 , requires considerable knowledge of the heavy - light meson dynamics .
for this reason , the iw function is usually thought of as `` model - dependent '' and not susceptible to reliable calculation . in the first portion of this paper
we seek a balance between rigorous constraints and phenomenological usefulness .
we will demonstrate , from heavy quark symmetry and some qualitative lattice results , that the iw function is accurately determined by specifying only the slope parameter , @xmath1 .
experimentalists commonly adopt the straightforward procedure of expanding form factors about the meson zero - recoil point ( @xmath12 ) , @xmath13 burdman @xcite has pointed out that the effect of the curvature term @xmath14 is significant but that , for statistical reasons , much predictive power is lost when it is a free parameter .
a `` model - independent '' relation between @xmath14 and the slope parameter @xmath1 has been proposed by boyd , grinstein and lebed @xcite .
this method has been modified by caprini and neubert @xcite and by caprini , lellouch , and neubert in expanded form @xcite .
this result has been criticized however by boyd , grinstein , and lebed @xcite .
these latter authors propose a similar but weaker relation between @xmath14 and @xmath1 . in section [ iw ]
, we propose a rigorous one - parameter expression for the iw function . the second observation we make here concerns the relation of the iw function to the actual ( physical ) form factors with @xmath2 and qcd radiative corrections .
we show in section [ ff ] that all of these corrections can be distilled into a new effective iw function with a shift in slope .
the analysis of experimental data is thereby greatly simplified .
the slopes appearing in @xmath15 and @xmath16 will be different and this difference can be compared to the theoretical predictions .
the decay rate for @xmath17 can be written @xcite as , @xmath18 where for @xmath19 @xmath20 @xmath21 and for @xmath22 @xmath23 \;,\ ] ] @xmath24 the two form factors @xmath25 and @xmath26 can be expressed @xcite in terms of the fundamental form factors @xmath27 , @xmath28 , @xmath29 , @xmath30 , @xmath31 , and @xmath32 .
these form factors can in turn be related to the isgur - wise function through @xmath33\xi(w ) \;.\ ] ] where @xmath34 and @xmath35 .
the @xmath36 describe perturbative radiative corrections and are in principle predictable within the heavy quark effective theory .
the @xmath37 represent the power ( @xmath2 ) deviations from heavy quark symmetry and require further theoretical assumptions @xcite . in section [ ff ]
we will point out that , for those form factors which do not vanish in the heavy quark limit , the pre - factors in ( [ 1 - 4 ] ) can nevertheless be approximately absorbed into an effective iw function which has a flavor and spin - dependent slope @xmath38
the iw function appears to require detailed information on the nature of heavy - light dynamics . in this section ,
we propose a method to distill most of this knowledge into a single parameter the slope @xmath1 .
almost all model dependence will be seen to vanish once the slope is specified . in the heavy - light limit
, heavy quark symmetry provides a unique prescription @xcite relating the iw function to the wavefunction and light degrees of freedom energy e , @xmath39 the expectation value may involve a multi - component wavefunction such as from the dirac equation . for simplicity of notation , we assume a single component wavefunction @xmath40 .
the expectation value is consequently defined by @xmath41 using @xmath42 , we may expand ( [ iw1 ] ) about the zero recoil point @xmath12 to yield : @xmath43 [ < \ ! 1 \ !
> - \frac{4e^2}{6 } ( \frac{w-1}{w+1})<\!r^2\!>+\ldots ] \\
\xi(w ) & \approx & 1 - \rho^2(w-1)+\ldots\end{aligned}\ ] ] where @xmath44 we now use ( [ iw4 ] ) to eliminate the energy @xmath45 in the general expression ( [ iw1 ] ) to obtain : @xmath46 where @xmath47 . in the above ( [ iw5 ] )
, if we know the slope ( @xmath1 ) and the radial ground state heavy quark wavefunction , we can compute the iw function for all @xmath8 .
we can make the wavefunction dependence more explicit by introducing the dimensionless quantities @xmath48 where @xmath49 is a hadronic scale factor .
we define the @xmath50 moment of @xmath51 as @xmath52 so that wave function normalization and @xmath53 become : @xmath54 the iw function ( [ iw5 ] ) is then @xmath55 it should be noted that both the normalization constant @xmath56 and the hadronic scale parameter @xmath49 do not appear in the above expression for the iw function .
the iw function thus depends only on the dimensionless slope parameter @xmath1 and one dimensionless function @xmath57 which is essentially the heavy - light wavefunction .
a few years ago , the heavy - light wavefunction was evaluated in quenched lattice simulation @xcite .
the result for the ground state is shown in fig .
one may observe that the lattice wavefunction closely resembles a simple exponential .
we use this lattice result as a guide and parametrize the wavefunction as @xcite : @xmath58 when @xmath59 the wavefunction is a simple exponential and the lattice simulation is closely reproduced . in fig .
[ wfcn ] , we also show that by choosing @xmath60 and @xmath61 we very conservatively bracket the observed wavefunction . assuming the heavy - light wavefunction is given by ( [ 2 - 11 ] ) with @xmath62 the iw function is determined for all @xmath8 once the slope is specified . in fig .
[ iwk ] we show the predicted iw function with @xmath63 .
the central curve corresponds to @xmath59 with a corridor implied by the limits of ( [ 2 - 12 ] ) .
we note that at @xmath64 , which is approximately the largest allowed for @xmath0 decay , the corridor width is only about 0.02 , or 3% .
this small uncertainty shows that most of the iw shape is well - determined by the slope alone .
a particularly important special case is when @xmath59 .
straightforward evaluation of ( [ bigiw ] ) with @xmath59 in ( [ 2 - 11 ] ) yields the iw function @xmath65 ^ 2 } \;.
\label{our - iw}\ ] ] the above isgur - wise function will be accurate within the corridor shown in fig .
an important observation from specific theoretical models @xcite is that the subleading isgur - wise form factors which characterize the @xmath2 corrections are to a good approximation linear in @xmath8 .
in addition , the radiative qcd corrections are monotonic and vary slowly with @xmath8 , so that they too can be approximated linearly . hence each of the @xmath66 form factors ( [ 1 - 4 ] ) can be written as @xmath67\xi(w ) \,,\ ] ] where both @xmath68 and @xmath69 are small , dimensionless constants .
the advantage of this observation is that an analysis of the experimental form factor can now be carried out without a priori knowledge of the @xmath70 and @xmath71 coefficients . to take advantage of the above , we start with the expansion ( [ iw3 ] ) and note that for small @xmath72 , iw functions of slopes @xmath73 and @xmath1 are related by @xmath74 over a wider range of @xmath8 a more accurate expansion is @xmath75 \,.\ ] ] to illustrate the above approximation
, we show in fig .
[ iwcomp ] a simple physical form factor ( with @xmath76 ) , @xmath77 \;.\ ] ] also on fig .
[ iwcomp ] is a shifted iw function @xmath78 with slope chosen by the above prescription .
we note that the two form factors differ by less than 2% at @xmath64 .
we now apply this result to the form factors @xmath66 . comparing ( [ ff4 ] ) to ( [ ff1 ] ) in the case where @xmath79 we see that @xmath80 ; . \label{ff5a}\end{aligned}\ ] ] we drop the small product @xmath81 to obtain @xmath82 with @xmath83 we may therefore conclude that the physical form factors are nearly equivalent to iw form factors with shifted slopes and normalizations . as pointed out by neubert @xcite ,
the sub - leading from factor @xmath84 contributes identically to all of the @xmath85 which remain in the heavy quark limit .
we therefore simplify the analysis by absorbing this @xmath84 contribution into a new iw type function which maintains spin - symmetry , but now has flavor dependence .
it is this slope @xmath1 which we refer to as the iw slope in the following .
the absorption of @xmath84 into the iw function is allowed as long as @xmath84 remains small .
the evidence for this is ambiguous .
a qcd sum rule evaluation @xcite gives a wide range of possible values for @xmath84 , some of which are large .
evidence that @xmath84 is small comes , indirectly , from the dispersive analysis of caprini , lellouch and neubert @xcite where the @xmath86 is found for a given slope . in fig .
[ ffs ] we show the cln prediction for @xmath87 compared to our @xmath59 prediction ( [ 2 - 15 ] ) for the same slope .
the two curves are nearly identical , which is expected if @xmath84 and other corrections are small and can be absorbed into an effective iw function .
the semileptonic @xmath3 decays yield definite but limited information about qcd corrections .
it is important to understand exactly what in a theoretical framework is being tested by experiment . in this section ,
we examine each observable quantity to see which aspect of qcd is being tested .
we consider the semileptonic decays separately . for @xmath19 ,
the form factor is @xcite @xmath88\end{aligned}\ ] ] where @xmath89 and hence @xmath90 the only measurable parameter here is the physical form factor slope , @xmath91 , which has been investigated at cleo @xcite .
if the data improves sufficiently , the difference between @xmath91 and the slope of a different physical form factor ( such as @xmath92 ) could provide information about qcd . to gain a rough idea of the sizes of @xmath93 and @xmath94
we compute the wilson coefficients and substitute the qcd sum rule approximations for the actual values of the subleading iw form factors .
the results are included in the table ii of appendix a. there are some directly applicable tests of hqet provided by qcd lattice simulations .
accurate lattice simulations are currently possible at the zero recoil point ( @xmath12 ) .
a recent calculation @xcite obtains : @xmath95 in hqet , these coefficients are due to the radiative corrections and @xmath96 corrections @xcite .
these are enumerated in appendix a. we observe that @xmath97 has no @xmath96 corrections as required by luke s theorem @xcite .
the result from the wilson coefficients alone of @xmath98 is consistent with the lattice result ( [ h+ ] ) .
the comparison with @xmath99 of ( [ h- ] ) is more interesting .
the @xmath96 correction here contains the subleading form factor @xmath100 .
this form factor has been estimated by qcd sum rules @xcite , by isgw2 @xcite , and by a salpeter equation method @xcite . while these various calculations yield reasonably consistent results for the subleading form factor @xmath101 , the expectations for @xmath102 vary widely . from appendix
a we see that @xmath103 using the calculated wilson coefficients and the parameters @xmath104 gev , @xmath105 gev , and @xmath106 gev we can evaluate the previous expression to yield @xmath107 the lattice result in ( [ h- ] ) suggests that the second term above is negative , i.e. @xmath108 .
this prediction is not completely rigorous as the size of the effect is of order a few percent and second - order @xmath96 corrections could be of that order .
in addition , the lattice calculation for @xmath109 has currently been calculated only to tree level , but a 1-loop calculation is underway @xcite . however , the preliminary indication is that @xmath100 is less than the central value of 0.62 which comes from qcd sum rules . at the same time
however , the much smaller values of @xmath100 which come from isgw2 and the salpeter model are also disfavored because they produce a result for @xmath109 which is too negative . in this case , there is additional information provided by the @xmath111 decay distribution @xcite . as suggested by cln @xcite and followed by cleo @xcite this decay can be analyzed in terms of the @xmath30 form factor and two ratios of form factors .
the cleo experiment , which to date has the best measurements of the @xmath16 form factor paramaters , has adopted the convention of fitting the slope of @xmath30 , @xmath112 and two form factor ratios : @xmath113 where @xmath114 .
these ratios are expected to be nearly independent of @xmath8 so they are treated as constants in the fit .
the form factor @xmath92 can be expressed in terms of these parameters by combining equation ( [ a-3 ] ) from appendix a and the definitions ( [ rs ] ) .
we would also like to write @xmath26 as a shifted iw function , @xmath115 \;,\ ] ] and hence @xmath116 where @xmath117 .
since @xmath118 is the sum of @xmath85 contributions and kinematical factors , the @xmath119 and @xmath120 are the result of an expansion about @xmath12 .
the result is @xmath121 \ , .
\label{muds}\end{aligned}\ ] ] as in the @xmath122 case , we can obtain rough approximations of @xmath120 , @xmath119 , @xmath123 and @xmath124 .
the results are included in table ii of appendix a. the above expression ( [ muds ] ) has a remarkable consequence .
the slope difference @xmath125 can be related to the zero - recoil quantities , @xmath68 above .
it is therefore possible in the near future to test ( for the first time ) the `` intercept '' quantities @xmath68 by direct comparison to experiment . at this point only @xmath124 has been computed in lattice simulation @xcite , but the corresponding values of @xmath126 , @xmath127 , and @xmath128 could be evaluated .
at present the experimental situation @xcite is also not precise enough : @xmath129 but this result can also be considerably sharpened .
we emphasize that the above prediction ( [ muds ] ) is an important check of the values of the intercepts , @xmath68 , which are in turn of critical importance in the extraction of the ckm parameter @xmath130 . in the heavy quark limit , @xmath131 .
these ratios deviation from unity , experimentally and theoretically , indicates deviation from heavy quark symmetry and , consequently , can test theoretical predictions for the symmetry - breaking corrections .
the measurements of @xmath132 and @xmath133 ( currently with large error bars ) agree with theoretical predictions , and thus roughly confirm the predictions of hqet . here
we investigate the exact nature of the predictions and their value as tests of hqet . treating the form factor ratios as constants , while reasonable theoretically and necessary for the fit , makes the tests of symmetry - breaking corrections less informative .
@xmath134 and @xmath135 are essentially reduced to @xmath136 and @xmath137 , and because many of the corrections vanish at @xmath12 , they can not be tested .
for example , the subleading isgur - wise function @xmath138 which has only been calculated up to the errors inherent to qcd sum rules is completely lost in this procedure .
if we apply the expressions given in appendix a to the various form factors at @xmath12 , eliminate those corrections which vanish at zero - recoil , and express the results in terms of the wilson coefficients and subleading iw form factors , we obtain : @xmath139 + \frac{4\chi_2(1)}{2m_c}(r^*-1 ) + \bar{\lambda}(\frac{1}{2m_b}-\frac{r^*}{2m_c } ) } { c_1 ^ 5(1)}\end{aligned}\ ] ] if we use the theoretical estimates as rough guides to the size of each term , we can determine what predictions are actually being tested .
the subleading combination @xmath140 is expected to be small on the basis of the @xmath99 considerations and is further reduced by @xmath141 .
consequently , after evaluating the wilson coefficients @xmath142 ( see table i of appendix a ) , @xmath132 must be greater than unity and provides a fairly direct probe of the hqet parameter @xmath143 , which is the light degrees of freedom energy to this order . in the expression for @xmath133 ,
@xmath144 is generally agreed to be quite small and @xmath145 .
thus after calculating the wilson coefficients , a measurement of @xmath133 provides an additional probe of the value of the subleading form factor @xmath100 . using the wilson coefficients in appendix a , as well as @xmath146 gev , @xmath105 gev , @xmath147 and @xmath106 gev ,
the expression for @xmath133 becomes @xmath148 the term containing @xmath100 is the dominant correction as @xmath144 is small , so a precise measurement of @xmath133 effectively measures @xmath100 .
we have addressed the question of how to analyze semileptonic @xmath3 decay while minimizing both the amount of theoretical assumption and the number of parameters .
we first establish that specifying the slope parameter @xmath1 accurately determines the entire isgur - wise function .
to do so , we need only an approximation to the light degrees of freedom wavefunction provided by a qcd simulation .
our result ( [ fcnform ] ) is @xmath149 ^ 2 } \;.\ ] ] the form factors @xmath66 either vanish or approach the iw function in the heavy quark limit .
these amplitudes can be accurately written as @xmath150\xi(w,\rho^2 ) \,,\ ] ] where @xmath151 is 0 or 1 and @xmath68 and @xmath69 are small dimensionless constants .
we observe that by altering the iw slope @xmath1 to @xmath152 the physical form factors can be expressed as @xmath153 where @xmath152 .
that this works well is shown by fig .
[ iwcomp ] . from the above we
conclude that semileptonic data are parametrized by an intercept value @xmath154 and an effective isgur - wise slope parameter @xmath155 . to find the ckm element @xmath130 one must use a theoretical estimate of @xmath68 . for @xmath10 decay
the parameters are @xmath156 , @xmath157 , @xmath132 , and @xmath133 . in this case one
has a consistency condition ( [ muds ] ) relating the @xmath68 and the difference between actual @xmath158 slope and @xmath112 .
we further point out that @xmath132 is nearly model - independent while @xmath133 depends sensitively on various estimates of the subleading from factor @xmath159 .
also , the value of the form factor @xmath160 at zero - recoil , @xmath99 , appears to offer an additional probe of the value @xmath159 .
finally , one might ask , what is the essential advantage of the slope shift scheme described here ?
the answer is , economy of parameters and a decoupling from theoretical assumptions . as seen in fig .
[ iwcomp ] , one of the effective iw function slopes is equivalent to an ( unknown ) iw slope with subleading corrections .
if these corrections were securely known our scheme offers no advantage . on the other hand ,
if one tried to fit both the iw slope and hqet @xmath161 parameters , a hopeless parameter correlation would arise .
we believe that it is best to rely as little as possible on theory and to use a direct phenomenological approach . as pointed out previously@xcite , one must have some theoretical constraint on the curvature term in ( [ 1 - 1 ] ) .
we show here that the shape of the form factor is specified once the slope parameter is given . later ,
after the decay distributions have been parametrized , the fitted parameters can be compared to predictions .
this research was supported in part by the u.s .
department of energy under grant no .
de - fg02 - 95er40896 and in part by the university of wisconsin research committee with funds granted by the wisconsin alumni research foundation . the matrix elements for @xmath19 in ( [ 1 - 3 ] ) can be expressed in terms of \{@xmath85 } as @xmath163 and the squared matrix element for @xmath164 in ( [ 1 - 3 ] ) as @xmath165 \nonumber \\ & & + ( w+1)^2[(w - r^*)h_{a_1}-(w-1)(r^*h_{a_2}+h_{a_3})]^2\,.\label{a-3}\end{aligned}\ ] ] the @xmath85 can , as indicated in ( [ 1 - 4 ] ) , be expressed by the short - range corrections @xmath166 and the @xmath2 corrections @xmath167 together as @xmath168 \nonumber \\
h_{a_3}(w ) & = & \xi(w)[c_{1}^{5 } + c_{3}^{5 } + \varepsilon_c(l_2-l_3-l_5+l_6 ) + \varepsilon_b(l_1-l_4)]\nonumber \\
h_{a_2}(w ) & = & \xi(w)[c_{2}^{5 } + \varepsilon_c(l_3+l_6 ) ] \nonumber \\
h_{a_1}(w ) & = & \xi(w)[c_{1}^{5 } +
\varepsilon_c(l_2-\frac{w-1}{w+1}l_5 ) + \varepsilon_b(l_1-\frac{w-1}{w+1}l_4 ) ] \nonumber \\
h_+(w ) & = & \xi(w)[c_1 + \frac{w+1}{2}(c_2+c_3 ) + ( \varepsilon_c+\varepsilon_b)l_1 ] \nonumber \\
h_-(w ) & = & \xi(w)[\frac{w+1}{2}(c_2-c_3 ) + ( \varepsilon_c+\varepsilon_b)l_4 ] \label{a-4}\end{aligned}\ ] ] where @xmath169 , the @xmath142 are the wilson coefficients which are discussed extensively in @xcite and the @xmath170 are combinations of the sub - leading isgur wise form factors and have been approximated by qcd sum rules @xcite : @xmath171 where @xmath172 is the `` binding energy '' of a heavy meson . using the detailed expressions given in @xcite and @xcite and the expressions above
, we can extract @xmath68 and @xmath69 for each form factor @xmath85 which appears in ( [ ff1 ] ) .
they appear in the following table .
we use the values @xmath173 and @xmath174 and calculate the wilson coefficients at the scale @xmath175 .
99 n. isgur and m.b .
wise , phys .
b232 , 113 ( 1989 ) ; b237 , 527 ( 1990 ) .
boyd , b. grinstein , and r.f .
lebed , phys .
* b353 * , 306 ( 1995 ) . i. caprini and m. neubert , phys .
b380 , 396 ( 1996 ) .
i. caprini , l. lellouch , and m. neubert , nucl .
b530 , 151 ( 1998 ) .
boyd , b. grinstein , and r.f .
lebed , phys .
d56 , 6895 ( 1997 ) .
duboscq et al . , phys .
76 , 3898 ( 1996 ) j. bartelt et al , phys .
82 , 3746 ( 1999 ) .
g. burdman , phys .
b284 , 133 ( 1992 ) .
e.g. , s. veseli and m.g .
olsson , phys .
d54 , 886 ( 1996 ) .
a. falk and m. neubert , phys .
d47 , 2965 ( 1993 ) .
m. sadzikowski and k. zalewski , z. phys .
c59 , 677 ( 1993 ) ; h. hgassen and m. sadzikowski , z. phys .
c64 , 427 ( 1994 ) ; m.g . olsson and s. veseli ,
d51 , 2224 ( 1995 ) ; m.r .
ahmady , r.r .
mendel and j.d .
talman , phys .
d52 , 254 ( 1995 ) . s. veseli and m.g .
olsson , phys .
b367 , 302 ( 1996 ) .
a. duncan , e. eichten , and h. thacker , phys . lett .
b303 , 109 ( 1993 ) ; phys .
d51 , 5101 ( 1995 ) .
olsson and s. veseli , phys .
b397 , 263 ( 1997 ) .
m. neubert , phys . rep .
245 , 259 ( 1994 ) .
s. hashimoto et al .
d61 , 014502 ( 2000 ) m. luke , phys .
b252 , 447 ( 1990 ) .
z. ligeti , y. nir , and m. neubert , phys .
d49 , 1302 ( 1994 ) .
d. scora and n. isgur , phys .
d52 , 2783 ( 1995 ) .
a. abd el - hady et al . , phys .
d51 , 5245 ( 1995 ) .
simone et al . , nucl.phys.proc.suppl .
73 , 399 ( 1999 ) . | we demonstrate that the @xmath0 form factors can be accurately predicted given the slope parameter @xmath1 of the isgur - wise function .
only weak assumptions , consistent with lattice results , on the wavefunction for the light degrees of freedom are required to establish this result .
we observe that the qcd and @xmath2 corrections can be systematically represented by an effective isgur - wise function of shifted slope .
this greatly simplifies the analysis of semileptonic @xmath3 decay .
we also investigate what the available semileptonic data can tell us about lattice qcd and heavy quark effective theory .
a rigorous identity relating the form factor slope difference @xmath4 to a combination of form factor intercepts is found .
the identity provides a means of checking theoretically evaluated intercepts with experiment . |
there have been many detections of ultraviolet resonance line absorption by highly ionized , presumably hot , gas in the galactic halo ( e.g. , sembach & savage 1992 , hurwitz & bowyer 1996 ) , but only three claimed detections of ultraviolet resonance line emission from this gas .
first , martin & bowyer ( 1990 ) reported detections of the ( unresolved ) ( @xmath0 1548/1550 ) emission in 4 out of their 8 lines of sight with a maximum strength of @xmath4 ergs @xmath1 sr@xmath2 s@xmath2 .
they also detectected o iii ] ( @xmath5 1663 ) emission at about half the intensity and much lower significance .
second , out of the ten hopkins ultraviolet telescope ( hut ) targets which could be profitably used for studies of the galactic halo , @xcite detected emission in four directions at levels on the order of @xmath6 ergs @xmath1 sr@xmath2 s@xmath2 ( @xmath7 photons @xmath1 sr@xmath2 s@xmath2 ) , with 2 @xmath8 upper limits of roughly the same level on the other six targets .
finally , there have been three recent observations of emission using the _ fuse _ satellite all at a level of about only 5000 photons @xmath1 sr@xmath2 s@xmath2 : @xcite at ( l , b ) = ( 315.00 , -41.33 ) ; and @xcite at ( l , b ) = ( 284.2 , 74.5 ) and ( l , b ) = ( 57.6 , 88.0 ) . the most constraining upper limit is the _ minisat-01 _
90% confidence upper limit of @xmath9 ergs @xmath1 sr@xmath2 s@xmath2 ( 1200 photons @xmath1 sr@xmath2 s@xmath2 ) from @xcite with earlier limits of about @xmath10 ergs @xmath1 sr@xmath2 s@xmath2 ( 7600 photons @xmath1
sr@xmath2 s@xmath2 ) from @xcite and @xcite .
note that all values cited for the emission are for the integrated emission over both lines of the doublet .
@xcite have reprocessed 17 years ( 1977 1994 ) of data from the _ voyager _ 1 and 2 archives with a focus on the continuum emission due to dust scattering . in the present paper
, we will discuss limits , from the same data set , on ( 1032/1038 ) line emission from the ism .
although new instruments are now providing important results , the _ voyager _ data are still the only source of information on the emission over many different lines of sight .
the two _ voyager _ spacecraft were launched in 1977 and have taken fuv ( 500 - 1700 ) spectra of astronomical objects ever since .
each spacecraft includes a wadsworth - mounted objective grating spectrometer ( uvs ) with a field of view of 01 x 087 and a spectral resolution of 38 for aperture filling diffuse sources .
a full description of the uvs instruments and the _ voyager _ mission is given by @xcite .
the data processing is described in @xcite . because we were only interested in the diffuse background , all other observations ( planets , stars and nebulae ) were discarded .
the remaining data consist of 426 observations of diffuse background .
the doublet ( @xmath0 1032/1038 ) is clearly visible in the _ voyager _ spectra of bright sources such as supernovae remnants @xcite and the eridanus superbubble @xcite , where the doublet is much brighter than the heliospheric hydrogen ly @xmath11(@xmath12 1026 ) emission .
however , the emission from the diffuse halo gas is much less than the ly @xmath11 emission on whose wings it lies .
fortunately , because the lyman lines are optically thick , the ly @xmath11/ly @xmath13 ratio is constant throughout the heliosphere and we can use the ly @xmath13line to scale the ly @xmath11 line .
we determined the ratio between the two lines using uvs observations in which only the heliospheric lines were present and then used this empirical ratio to scale the ly @xmath11 line in each observation ( see @xcite for a full description of this procedure ) .
we subtracted this scaled ly @xmath11 intensity from the observed spectrum and determined the upper limit from the remainder . because the ly @xmath11 line is at almost the same position as the line , there is a tradeoff between their respective derived intensities .
note , however , that the difference in the central wavelengths of the ly @xmath11 and emission is large enough ( 1 resolution element ) that can not fully , or even largely , replace the ly @xmath11 contribution . in our procedure , we have restricted the heliospheric ly @xmath11/ly @xmath13 ratio to fall between empirically determined limits ; if , on the other hand , we allow the ly @xmath11/ly @xmath13 ratio to vary freely , our limits on the emission will be correspondingly poorer .
we have carried out this exercise for each of our targets but , because of the varying amount of heliospheric emission , can not compare points on an individual basis . in general , our limits
go up ( become less constraining ) by a factor of 2 3 .
thus our best upper limit rises from 2600 photons @xmath1 sr@xmath2 s@xmath2 to 8500 photons @xmath1 sr@xmath2 s@xmath2 and the number of pointings with limits under 50,000 photons @xmath1 sr@xmath2 s@xmath2 drops to 127 from 244 .
( details of the individual targets are available from the authors on request . )
@xcite have claimed that @xcite and @xcite , have significantly underestimated the errors in the _ voyager _ data . as the data analysis in this work rests heavily on the earlier papers , particularly that of @xcite , we are compelled to address these criticisms . a careful reading of the edelstein et al .
paper shows that there are virtually no differences between their results and ours , despite their claims . from their table 2 edelstein et al .
obtain a 1 @xmath8 uncertainty of 125 photons @xmath1 sr@xmath2 s@xmath2 @xmath2 , identical to the limit claimed by @xcite .
however , they also obtain a residual of 320 photons @xmath1 sr@xmath2 s@xmath2 @xmath2 whereas @xcite found a null signal .
this difference can be traced to edelstein et al .
estimating two large numbers from a figure in @xcite , subtracting the two and claiming that as the residual .
if they had actually used the original data in digital format , the correct procedure , their results would have agreed exactly with @xcite .
importantly for this work , and contrary to their claims , @xcite have shown that the counting errors in our analysis procedure are reasonable . because the signal in the _ voyager _ spectra is dominated by the rtg particle background ( due to radioactivity in the radioisotope thermoelectric generator ) and the scattered ly @xmath13 ,
we have explored the possibility that systematic errors in the subtraction of the two components are affecting our limits . should there be a feature in either of these components coincidentally at the position of
, we would expect the limits to be correlated with that component , because the strength of that feature would be necessarily correlated with the level of the continuum ( the sum of the rtg and ly @xmath13 contributions ) . over the 17 years of _ voyager _
observations , both the rtg level and the ly @xmath13 emission declined : the former because of the decline in the radioactivity of the plutonium power source and the latter because of the increasing distance of the spacecraft from the sun .
thus , if there were any significant systematic errors associated with the background subtraction , our limits would be strongly correlated with the year of observation .
no such effect is detectable in our data ( shown in the case of the ly @xmath13 emission in figure 1 ) , implying that systematic effects due to the subtraction of the rtg and ly @xmath13 backgrounds are unimportant .
we can demonstrate empirically that our quoted error bars are reasonable through a listing of each of the errors in one of our targets ( table 1 ) .
note that we have arbitrarily chosen the location with our most constraining limit .
we have listed in table 1 the integrated counts under the line in the total spectrum and in each of the modeled components of the raw _ voyager _ data : the rtg spectrum ; the ly @xmath13 template ; ly @xmath11 emission ; and the diffuse continuum ( due to dust - scattered starlight ) .
the poisson errors are also listed , with the rtg and total errors reflecting the fact that each rtg event generates 3 counts @xcite . from these errors
we then calculate a total uncertainty , assuming uncorrelated errors . for this target
we obtain a 1 @xmath8 uncertainty of 308 counts corresponding to a signal of 1400 photons @xmath1 sr@xmath2 s@xmath2 .
this is entirely consistent with the 90% confidence level of 2600 photons @xmath1 sr@xmath2 s@xmath2 that we derived using our modeling procedure ( the modeling is described in detail in @xcite ) .
our quoted uncertainties have taken into account all the statistical errors and come from a @xmath14 minimization according to the procedure described by @xcite .
essentially , we changed the value of the emission , while allowing the other parameters to vary freely through the allowed parameter space , until the @xmath15 emission rose to unacceptable levels .
we detect no emission in any of 426 uvs observations of the diffuse radiation field but do set upper limits on such emission in each direction .
the best of these limits is 2600 photons @xmath1 sr@xmath2 s@xmath2(@xmath16 ergs @xmath1
sr@xmath2 s@xmath2 ) in the resonance line doublet in the direction ( l , b ) = ( 117.3 , 50.6 ) .
this direction is quite close to hd 121800 ( l , b = 113.0 , 49.8 , spectral type b1.5 v , distance = 2.2 kpc ) towards which @xcite obtained a column density of @xmath17 @xmath1 using orfeus . using these values and equation 5 of @xcite , and confining the temperature range to that
for which the fraction of oxygen atoms in the state is within 10% of its maximum value in collisional ionization equilibrium plasma ( t = @xmath18 k @xcite ) , we find an upper limit on the electron density of less than 0.010 @xmath3 .
assuming that the emitting gas has a solar abundance of helium atoms and that the hydrogen and helium are fully ionized , there will be 1.9 particles per electron and thus the thermal pressure will be less than 12,000 k @xmath3 , close to the thermal pressure of 15,000 k @xmath3 in the local bubble derived by @xcite from observations of the 1/4 kev soft x - ray flux seen by rosat .
the 94 locations in which we set 90% confidence upper limits of better than @xmath19 ergs @xmath1 sr@xmath2 s@xmath2 ( 25,000 photons @xmath1 sr@xmath2 s@xmath2 ) are plotted in figure 2 and those in which we set limits of better than @xmath20 ergs @xmath1 sr@xmath2 s@xmath2 ( @xmath21 10,000 photons @xmath1 sr@xmath2 s@xmath2 ) are listed in table 2 .
several of our observations are near the locations observed by @xcite using hut and we both set similar upper limits in those ( with our _ voyager _ limits in general being more constraining ) . only in their target 3
( ugc 5675 ; l = 218.2 , b = 56.4 ) do we obtain inconsistent results , with @xcite quoting a flux of 23,000 @xmath22 6000 photons @xmath1 sr@xmath2 s@xmath2while we place a 90% upper limit of 10@xmath23 photons @xmath1 sr@xmath2 s@xmath2at ( l , b ) = ( 216.8 , 55.3 ) about 1@xmath24 away . of course , it is entirely possible that there are truly spatial variations of this scale in the ism .
we also have several observations near the four high latitude locations where martin & bowyer ( 1990 ) detected emission but in none can we do more than say that the /c iv ratio is not inconsistent with the theoretical ratios reported from a variety of physical conditions ( e.g. cooling flows : @xcite ; shock heated gas : @xcite ; fountains : @xcite ; halo supernova remnants : @xcite ) .
very recent results concerning galactic diffuse emission include the _ fuse _ detections by @xcite and @xcite at a level of 5000 photons @xmath1 sr@xmath2 s@xmath2and the _
minisat-01 _ all - sky upper limit of 1200 photons @xmath1 sr@xmath2 s@xmath2by @xcite . combined with the present _
voyager _ upper limits , it appears that much of the sky has an emission of significantly less than 10,000 photons @xmath1 sr@xmath2 s@xmath2 .
only the 4 hut detections of @xcite show higher fluxes . a mission dedicated to the observation and mapping of faint line emission from the galactic halo would surely yield bountiful results .
much of this work was done while jm was at the johns hopkins university and was supported by nasa grants nag5 - 2398 and nag5 - 2299 .
an anonymous referee contributed greatly to a discussion of the error analysis in this work .
we thank doyle hall and matthew earl for their contributions to the data analysis .
blair , w. p. , vancura , o. , & knox , k. s. 1995 , aj , 110 , 312 benjamin , r. a. & shapiro , p. r. 1993 , american astronomical society meeting , 183 , 111404 dixon , w. v. , davidsen , a. f. , & ferguson , h. c. 1996 , apj , 465 , 288 dixon , w. v. , sallmen , s. , hurwitz , m. , & lieu , r. 2001 , apj , 552 , l69 edelstein , j. , bowyer , c. s. , korpela , e. , lampton , m. , trapero , j. , gomez , j. f. , morales , c. , & orozco , v. 1999 , american astronomical society meeting , 195 , 5302 edelstein , j. , bowyer , s. , & lampton , m. 2000 , apj , 539 , 187 edgar , r. j. & chevalier , r. a. 1986 , apj , 310 , 27 hartigan , p. , raymond , j. , & hartmann , l. 1987 , apj , 316 , 323 holberg , j. b. & watkins , r. 1992 , _ voyager _ ultraviolet spectrometer guest observer and data analysis handbook , version 1.1 holberg , j. b. 1986 , apj , 311 , 969 hurwitz , m. & bowyer , s. 1996 , apj , 465 , 296 korpela , e. j. , bowyer , s. , & edelstein , j. 1998 , apj , 495 , 317 lampton , m. , margon , b. & bowyer , s. 1976 , apj , 208 , 177 martin , c. & bowyer , s. 1990 , apj , 350 , 242 murthy , j. , i m , m. , henry , r. c. , & holberg , j. b. 1993 , apj , 419 , 739 murthy , j. , hall , d. t. , earl , m. , henry , r. c. , & holberg , j. b. 1999 , apj , 522 , 904 sembach , k. r. , & savage , b.d .
1992 , apjs , 83 , 147 shapiro , p. r. & moore , r. t. 1977 , apj , 217 , 621 shelton , r. l. 1998 , apj , 504 , 785 shelton , r. l. et al .
2000 , head , 32.4413s .
shelton , r. l. et al .
2001 , apj ( in press ) shull , j. m. & slavin , j. 1994 , apj , 427 , 784 snowden , s. l. , egger , r. , finkbeiner , d. p. , freyberg , m. j. , & plucinsky , p.
p. 1998 , apj , 493 , 715 ccc 117.3 & 50.6 & 2,600 + 272.5 & -67.4 & 4,100 + 67.8 & 5.2 & 5,700 + 60.3 & -22.5 & 6,500 + 117.3 & 50.8 & 6,700 + 200.7 & 9.6 & 7,000 + 71.6 & -59.6 & 7,400 + 189.6 & 32.3 & 8,600 + 91.1 & 61.4 & 8,700 + 115.7 & 72.6 & 9,000 + 32 & 70.5 & 9,100 + 331.7 & 60.5 & 9,200 + 99.3 & 80.3 & 9,500 + 225.7 & 68.3 & 9,900 + 190 & 33.3 & 10,000 + 216.8 & 55.3 & 10,000 + 346.6 & -52.3 & 11,000 + | we have examined 426 _ voyager _ fields distributed across the sky for ( @xmath0 1032/1038 ) emission from the galactic diffuse interstellar medium .
no such emission was detected in any of our observed fields .
our most constraining limit was a 90% confidence upper limit of 2600 photons @xmath1 sr@xmath2 s@xmath2 on the doublet emission in the direction ( l , b ) = ( 117.3 , 50.6 ) .
combining this with an absorption line measurement in nearly the same direction allows us to place an upper limit of 0.01 @xmath3 on the electron density of the hot gas in this direction .
we have placed 90% confidence upper limits of less than or equal to 10,000 photons @xmath1 sr@xmath2 s@xmath2on the emission in 16 of our 426 observations . |
the transport properties of a two - dimensional electron gas ( 2deg ) subjected to a nonhomogeneous perpendicular magnetic field ( periodically modulated or not ) have been the focus of a great deal of research in recent years .
@xcite current fabrication technologies permit to create nonhomogeneous magnetic fields on a nanometer scale by deliberately shaping or curving the 2deg,@xcite or by integration of superconducting@xcite or ferromagnetic materials@xcite on top of the 2deg .
this will add a new functional dimension to the present semiconductor technology and will open avenues for new physics and possible applications.@xcite theoretically , the effect of nonhomogeneous magnetic fields on a 2deg have been studied both in the ballistic and the diffusive regime .
the resulting perpendicular magnetic field can act as a scattering centre,@xcite but can also bind electrons,@xcite and so influence the transport properties of the 2deg . in transport calculations one needs the electron states , which are obtained by solving the schrdinger equation .
mller@xcite studied theoretically the single particle electron states of a 2deg in a wide quantum waveguide under the application of a nonuniform magnetic field and showed that in the case of a magnetic field modulation in one direction , transport properties also become one dimensional and electron states propagate perpendicularly to the field gradient.@xmath0 making use of this decoupling , the electron states for different nonhomogeneous magnetic field profiles along one dimension were investigated , i.e. for a periodically modulated magnetic field @xcite , for magnetic quantum steps , barriers and wells in an infinite 2deg @xcite and in a narrow waveguide.@xcite in this paper we consider an infinite 2deg subjected to a step - like magnetic field , i.e abruptly changing in magnitude or polarity at @xmath1 , in one dimension ( taken to be the @xmath2-direction ) .
preliminary results were presented in ref . .
first the situation for two opposite homogeneous magnetic fields with the same strength will be considered.@xmath0the classical trajectories correspond to _ snake orbits _ and were already used in the seventies@xcite to describe electron propagation parallel to the boundary between two magnetic domains.@xmath0back then , one was interested in understanding the electron transport through multi - domain ferromagnets and it turned out to be more convenient to work with the classical trajectrories than with the corresponding electron states , which allows one to use a semi - classical theory which reduces the complexity of the theory considerably .
we are interested in transport through a 2deg situated in a semiconductor in which the fermi energy is orders of magnitude smaller than in the metallic systems of ref . .
we will study thoroughly the quantum mechanics of such electron states in a 2deg subjected to this step magnetic field profile , and we will compare them with their classical counterpart . we will discuss the energy spectrum and the corresponding electron states , and derive several properties.@xmath0we will show the existence of states wich have a velocity opposite to the expected classical orbits .
additionally , we will show that adding a background magnetic field modifies the spectrum and the states considerably.@xmath0 the paper is organized as follows.@xmath0 in sec .
ii we present our theoretical approach.@xmath0 in sec .
iii we calculate the energy spectrum , the wavefunctions and their corresponding group velocity , and compare this with their quantum mechanical counterpart.@xmath0 in sec .
iv we study the influence of a background magnetic field on the quantum mechanical and classical behaviour . in sec .
v we focus on the negative velocity state , and finally , in sec .
vi , we construct time dependent states , and interpret them classically for several magnetic field profiles.@xmath0
we consider a system of noninteracting electrons moving in the @xmath3-plane in the absence of any electric potentials.@xmath0the electrons are subjected to a magnetic field profile @xmath4.@xmath0first , we will study the electronic states near the edge of two magnetic fields with opposite strength @xmath5 , \ ] ] which is independent of the @xmath6-coordinate .
next , we will consider the influence of a background magnetic field @xmath7 on these states , which results in the magnetic field profile @xmath5 + b.\ ] ] in the following we will use @xmath8 and @xmath9 , to denote respectively the magnetic field on the left and the right hand side of the magnetic edge .
the one - particle states in such a 2deg are described by the hamiltonian @xmath10 ^{2}.\ ] ] taking the vector potential in the landau gauge , @xmath11 we arrive at the following 2d schrdinger equation @xmath12 ^{2}+2e\right\ } \psi ( x , y)=0,\ ] ] where the magnetic field is expressed in @xmath13 , all lengths are measured in the magnetic length @xmath14 , energy is measured in units of @xmath15 , with @xmath16 the cyclotron frequency and the velocity is expressed in units of @xmath17.@xmath18 and @xmath19 commute due to the special form of the gauge , and consequently the wavefuntion becomes @xmath20 which reduces the problem to the solution of the 1d schrdinger equation @xmath21 \phi _ { n , k}(x)=e_{n , k}\phi _ { n , k}(x ) , \label{eq:1d_schrod}\ ] ] where it is the @xmath22-dependent effective potential @xmath23 ^{2 } , \label{eq : potential}\ ] ] which contains the two dimensionality of the problem.@xmath0we will solve eq .
( [ eq:1d_schrod ] ) numerically by use of a discretization procedure.@xmath0 in some limiting cases analytical results can be obtained .
let us first consider the case when no background magnetic field is present.@xmath24the situation is then symmetric , and more easily to solve.@xmath0the effective potential for this case is shown in fig .
[ fig : trajectories](a ) for @xmath25 ( dotted curve ) and @xmath26 ( solid curve).@xmath0we notice from eq .
( [ eq : potential ] ) that this potential is built from two parabolas , with minima situated at @xmath27 , and @xmath28 , thus respectively on the left and right hand side of the magnetic edge .
the total potential has for @xmath29 two local minima respectively at @xmath30 and @xmath31 , while for @xmath32 it has only one minimum at @xmath1 .
before we describe the energy spectrum of the snake orbits and their corresponding properties , we first discuss the limiting behaviour . for @xmath34 , the minima of the parabolas are situated far from each other.@xmath0the electrons are in the landau states of two opposite magnetic fields , one on the left , the other on the right , and they are not interacting with each other.@xmath0the electron wavefunctions are given by @xmath35 , and @xmath36 , respectively , where @xmath37 is the hermite polynomial.@xmath0for decreasing @xmath22 the parabolas shift towards each other , and the electrons will start to `` feel '' each other.@xmath0 in terms of wavefunctions , this results in a parabolic cylinder function @xmath38 $ ] matched at @xmath1 , with the condition that @xmath39 or @xmath40 , for the symmetric and the antisymmetric wavefunction , respectively .
this leads to a change in energy of the electron states , which can be understood as a lifting of the degeneracy of the two original electron wavefunctions .
the energy can then be written as @xmath41 with the corresponding wavefunctions @xmath42.@xmath0one can see that the presence of the second parabola results in two effects : ( 1 ) a decrease of@xmath43 due to the finite presence of the wave function in the other parabola , and ( 2 ) a splitting of the energy level due to the overlap , i.e. one level ( @xmath44 ) shifts upwards , while the other ( @xmath45 ) shifts down.@xmath46for @xmath34 , this results in the following first - order approximation to @xmath47 and @xmath48 : @xmath49 in the other limit @xmath50 , the effective potential can be approximated by a triangular well @xmath51.@xmath0solutions for this potential consist of airy functions , again matched at @xmath52 with the condition that @xmath53 or @xmath54 which results , respectively into the anti - symmetric wavefunction @xmath55 $ ] and a symmetric one @xmath56 $ ] , respectively with energy @xmath57 , \label{e2n(-infinity ) } \\ e_{2m+1}\left ( k\rightarrow -\infty \right ) & = & \frac{1}{2}\left [ k^{2}-z_{ai , m+1}\left ( 2\left| k\right| \right ) ^{2/3}\right ] , \label{e2n+1(-infinity)}\end{aligned}\ ] ] where @xmath58 ( @xmath59 , @xmath60 , @xmath61 , @xmath62 , @xmath63 ^{2/3}$ ] ) and @xmath64 ( @xmath65 @xmath66 @xmath67 @xmath68 @xmath69 ^{2/3}$ ] ) , denote respectively the @xmath70 @xmath71 zero of the airy function and of its derivative.@xmath0one can see that for increasing negative @xmath22 , the difference between the two energy branches increases , which is to first order linear in @xmath72 .
namely the more negative @xmath22 , the narrower the well , thus the more the energy levels are shifted from each other.@xmath0 solving eq .
( [ eq:1d_schrod ] ) numerically gives rise to the energy spectrum shown ( solid curves ) in fig .
[ fig : spectrum](a).@xmath0for @xmath73 , we obtain the earlier mentioned landau levels , which are labelled with the quantum number @xmath74.@xmath0each level is twofold degenerate.@xmath46for decreasing @xmath22 , the degeneracy is lifted and they separate into two different branches with eigenstates @xmath75 and @xmath76 and eigenvalues @xmath77 and @xmath78 , and corresponding quantum numbers @xmath79 and @xmath80 .
this quantum number @xmath81 does not only result from arranging the levels according to their lowest energy , starting with @xmath82 , but it also reflects the number of nodes of the corresponding wavefunction.@xmath0notice that the levels have now a non - zero derivative , i.e. electrons propage in the @xmath6-direction , and their group velocity is given by @xmath83.@xmath0(the minus sign appears here because in eq .
( [ eq : wavefunction ] ) we took @xmath84 ) .
this group velocity is plotted ( solid curves ) in fig .
[ fig : velocity](a ) for the 6 lowest levels . for @xmath85 electrons
are in a landau level , and consequently there is no net current in the @xmath6-direction.@xmath0decreasing @xmath22 , results in a net current in the @xmath6-direction , which is positive for the upper branches @xmath86 , but is initially negative for the branches @xmath87.@xmath0for more negative values of @xmath22 it increases almost linearly with increasing @xmath88 , which becomes the first order analytical result @xmath89 , obtained by differentiating ( [ e2n(-infinity ) ] ) and ( [ e2n+1(-infinity)]).@xmath0 the center of the classical orbit corresponds to a zero in the effective potential.@xmath0the energy spectrum can be divided up into three regions which can be classicaly understood by the electron orbits drawn in fig .
[ fig : trajectories](a).@xmath0 in region ( a ) the electrons move in closed orbits either in the magnetic field on the left or on the right hand side . since its cyclotron radius is smaller than the distance to the magnetic field discontinuity , they feel a homogeneous magnetic field .
there is no net velocity . in region
( b ) the cyclotron radius intersects the magnetic field discontinuity slightly , i.e. in such a way that the moving electron and the center of its orbit are on the same side .
the electron is nevertheless able to penetrate in the opposite magnetic field region , which results in a ( rather small ) propagation in the @xmath6 direction @xmath90 .
for @xmath91 the center of the orbit is exactly on the edge between the two opposite magnetic fields.@xmath0 in region ( c ) the center is located in the opposite magnetic field region of which the electron is moving in , resulting in a faster propagation of the electron in the @xmath6-direction .
these different regions are also indicated in fig .
[ fig : spectrum](a ) .
we can also make a quantitative classical study of the velocity , starting from the quantum mechanical energy spectrum.@xmath0since in classical mechanics there is no quantization , we make use of the obtained quantum energy spectrum in order to find the classical energy and thus the radius of the cyclotron orbit.@xmath0classically , the energy is contained in the circular velocity @xmath92 through @xmath93 .
for any given quantum mechanical @xmath94-value we obtain classically the circular velocity @xmath95 . now
if we consider @xmath96 to be the center of the electron orbit , we can calculate for every @xmath22-value the classical velocity @xmath97 , since we also know @xmath98 and the cyclotron radius @xmath99.@xmath24using geometric considerations , we obtain the following relation @xmath100^{2}}/\arccos[k / v_{\varphi } ( k ) ] , \label{eq : clasvel1}\ ] ] which is shown in fig .
[ fig : velocity](a ) by the dotted curves . comparing this with its quantum mechanical counterpart ,
we notice that for @xmath101 values good agreement is found , but for @xmath102 there is a large discrepancy.@xmath0moreover , one can see that negative velocities can not exist classically.@xmath0 the critical @xmath22-value , @xmath103 , for which no classical propagating states can exist , i.e. the electron describes just a circular orbit in a homogeneous magnetic field , has to be equal to the cyclotron radius @xmath104 , which leads to the boundary drawn in fig . [
fig : spectrum](a ) ( dotted parabola ) .
with a background magnetic field three different configurations : a ) @xmath105 , b ) @xmath106 , and c ) @xmath107 must be considered.@xmath0 in the following we will study the snake orbits in these configurations.@xmath0 applying a background magnetic field @xmath108 , results in a situation which is very similar to the previous one.@xmath0again the two magnetic fields have opposite sign , but in this case they also have a different strength , i.e. @xmath109.@xmath0again we can calculate analytically the correction to the energy in the limit @xmath34.@xmath0for an electron on the right hand side in the @xmath110landau state of a magnetic field with strength @xmath111 , the deviation from the landau energy due to the presence of the other parabola in the effective potential is given by the following matrix element , which to second - order reads @xmath112 -\frac{2^{m-2}}{m!\sqrt{\pi } } \left ( 1+\frac{1% } { p}\right ) k^{2m-1}e^{-k^{2}}.
\label{rhr}\end{aligned}\ ] ] for an electron on the left hand side , i.e. in the smaller magnetic field @xmath113 region , in the @xmath114 landau level , this results in @xmath115 -\frac{2^{m-2}}{m!\sqrt{\pi } } % \left ( 1+\frac{1}{p}\right ) k^{2m-1}e^{-k^{2}}. \label{lhl}\end{aligned}\ ] ] also in this case the energy is smaller then the corresponding landau energy.@xmath0the downward energy shift decreases for increasing @xmath116.@xmath0 if @xmath116 is an integer , landau states on the left and right hand side , respectively with quantum number @xmath117 and @xmath74 , coincide for @xmath118.@xmath0as a consequence these states have an overlap , which reads to first order @xmath119 one can see that for decreasing magnetic field , i.e. increasing @xmath116 , this function decreases because of the exponential factor.@xmath0the electron wavefunction in the lower magnetic field region is extended over a larger region , and further away from the other ( center at @xmath120).@xmath0the overlap therefore decreases with increasing @xmath116.@xmath0as a result of this , the energy for @xmath34 and @xmath121 is given by @xmath122 and @xmath123 . for @xmath124
, we obtain the previous result , but for increasing @xmath116 , the second order term in eqs .
( [ rhr ] ) and ( [ lhl ] ) becomes more important than eq .
( [ eq : matrix_lhr ] ) , because of the exponential factor.@xmath0the splitting is lifted , and the main contribution to the negative velocity for @xmath118 arises from eq .
( [ eq : matrix_lhr ] ) due to the finite extend of the wavefunction in the other parabola . as an example we studied numerically the case
when a background magnetic field @xmath125 is applied , i.e. @xmath126 and @xmath127.@xmath0as one can see in fig .
[ fig : trajectories](b ) , this results in two parabolas with different minima and confinement strength.@xmath0the resulting spectrum ( see fig .
[ fig : spectrum](b ) ) is very similar to the one of the previous case , but unlike the previous symmetrical case , not all states are twofold degenerate for @xmath34.@xmath0we now obtain two different sets of landau states , corresponding to electrons moving in different magnetic field regions with different strength.@xmath0 in this case the second landau level on the left coincides with the first on the right.@xmath46the classical picture for the three different regions corresponds to the one drawn in fig .
[ fig : trajectories](b ) , and is also similar to the previous case , except for the different cyclotron radii.@xmath0with this picture in mind , one can again calculate the classical velocity , which turns out to be identical to eq .
( [ eq : clasvel1 ] ) . from fig .
[ fig : velocity](b ) we notice that again we obtain good agreement for @xmath102 , but for @xmath32 , there is a large discrepancy.@xmath0the negative velocity can also in this case not be explained classically .
the critical @xmath22-value @xmath128 for which snake orbits are classicaly possible are indicated by the parabola in fig .
[ fig : spectrum](b).@xmath0 when a background magnetic field @xmath106 is applied , we obtain the magnetic barrier studied in ref . , where the magnetic field is different from zero only in the region @xmath129 , i.e. @xmath130 and @xmath131.@xmath0from fig . [
fig : trajectories](c ) one can see that in this case the potential is made up of only one parabola and on the left side it is a constant @xmath132.@xmath0the energy spectrum and corresponding velocities for this particular case are shown in fig .
[ fig : spectrum](c ) and [ fig : velocity](c ) , respectively.@xmath0we notice that for @xmath118 we again obtain landau states , which correspond to bound states on the right hand side of the magnetic edge.@xmath0consistent , as being a limiting case of the former magnetic field states , i.e. @xmath133 , the energy decreases with decreasing @xmath22 and there is no splitting of the energy levels.@xmath0thus now we only have states which propagate with negative velocity to which we can not assign a classical interpretation.@xmath0 also in this case we can divide up the spectrum into three regions : ( a ) the electrons move in closed orbits in the magnetic field region on the right hand side , ( b ) electrons are free , propagate forward and are reflected on the barrier and ( c ) electrons are free , propagate backward and are reflected on the magnetic edge.@xmath0notice that for a free electron , the energy is larger than @xmath132 , since now the electron also propagates in the @xmath2-direction and consequently has an additional kinetic energy @xmath134 .
classically , propagating states in the magnetic field region do not exist , only landau states do.@xmath0the boundary where these classical trajectories are possible is again given by @xmath128 . by applying a background magnetic field with strength larger than @xmath136
, we arrive at the situation where @xmath137 .
the magnetic fields on the left and the right hand side have the same sign , but a different strength , i.e. @xmath138 to obtain the energy in the limits @xmath33 , we again can approximate the wavefunction as being in a landau state in the corresponding magnetic field.@xmath0we found @xmath139 \nonumber \\ & & -\frac{2^{m-2}}{m!\sqrt{\pi } } \left ( 1+\frac{1}{p}\right ) k^{2m-1}e^{-k^{2}},\end{aligned}\ ] ] for an electron on the right hand side in the @xmath110landau state of a magnetic field with strength @xmath111.@xmath0for an electron on the left hand side , in the smaller magnetic field @xmath140 in the @xmath114 landau level , we have @xmath141 + \frac{2^{m-2}}{m!\sqrt{\pi } } % \left ( 1+\frac{1}{p}\right ) k^{2m-1}e^{-k^{2}},\end{aligned}\ ] ] which results in a negative velocity.@xmath0 the energy spectrum and the velocity of these eigenstates for the case when @xmath142 , i.e. @xmath143 , @xmath144 , are plotted respectively in fig .
[ fig : spectrum](d ) and [ fig : velocity](d).@xmath0the center of the orbit is situated on the right side for @xmath102 , for @xmath32 it is on the left side . for @xmath33 ,
the electrons move in a homogeneous magnetic field ( on the left ( @xmath50 ) or right ( @xmath145 ) hand side of @xmath1 ) , and thus @xmath146 . from fig .
[ fig : trajectories](d ) one notices that there is only one minimum in the effective potential because the minima of both parabolas are now situated on the same side.@xmath0the trajectories corresponding with regions ( a ) , ( b ) , and ( c ) are depicted in fig .
[ fig : trajectories](d).@xmath0the trajectories in region ( a ) are similar to those in ( a ) but now for a magnetic field on the left hand side , i.e. with smaller strength .
geometrical considerations yield the following classical velocity @xmath147 ^{2 } } \nonumber \\ & & \times \{b^{l}\arccos \left [ -k / v_{\varphi } ( k)\right ] \nonumber \\ & & \hspace{0.5cm}+b^{r}\arccos \left [ k / v_{\varphi } ( k)\right ] \}^{-1 } , \label{eq : clasvel2}\end{aligned}\ ] ] which is plotted in fig .
[ fig : velocity](d ) as dotted curves together with the quantum mechanical group velocity.@xmath0one can see that , in contrast to the previous cases , the negative velocity can be understood as classical snake orbits , but these snake orbits all run in the same @xmath6-direction and now there are no states with @xmath148 notice that the quantum mechanical velocity exhibits a small oscillatory behaviour on top of a uniform profile .
these whiggles can be understood from the electron distribution over the two parabolas ( see fig .
[ fig : whiggles ] ) . with increasing @xmath22 , the electron distribution is shifted from the left parabola to the right one . due to the wavelike character of this distribution , the probability for an electron to be in the right parabola ( integrated solid region in the inset of fig .
[ fig : whiggles ] ) exhibits whiggles as function of @xmath22 , with @xmath81 maxima as shown in fig .
[ fig : whiggles ] .
energetically it is favourable for an electron state to have as much as possible electron probability in the lower potential region .
consequently , when the electron probability in the lower potential region attains a maximum , a maximum downward energy shift will be introduced on top of the overall energy change , and this will result in a maximum in the group velocity .
formally , the existence of the quantum mechanical negative velocity state can be attributed to the fact that shifting two one dimensional potential wells towards each other results in a significant rearrangement of the energy levels in the composite potential well.@xmath0because the composed well is broader , some states , p.e .
the ground state , have an energy which is lower than in each of the individual narrower wells.@xmath0 in this particular case , the wave vector @xmath22 measures the distance of the two wells to each other , and consequently this energy decrease results in a negative group velocity @xmath149 . in this section
we focus on these negative group velocity states.@xmath0 since the negative velocity states are present for any background magnetic field @xmath7 , but can only be understood classically in the situation @xmath136 , we will investigate the group velocity for a fixed @xmath22 value with varying background magnetic field.@xmath0 in fig .
[ fig : spectrumb ] the spectrum is plotted as function of the applied background field @xmath7 .
we have chosen @xmath150 because in this case a large negative velocity is obtained for the lowest level when @xmath151.@xmath0notice that for @xmath152 : ( 1 ) almost all levels decrease in energy with increasing background field ; ( 2 ) there is an anti - crossing for @xmath153 ( dotted line ) .
this anti - crossing occurs when @xmath154 , with @xmath81 the landau level index . for this condition
some of the landau levels are degenerate in the limit @xmath34 ( see fig . [
fig : spectrum](b ) for the case @xmath155 ) ; and ( 3 ) for @xmath156 the separation between the levels decreases to zero and a continuous spectrum is obtained with a separate discrete level at the anti - crossing line . the continuous spectrum for @xmath157 results from the scattered states in the potential of fig . [ fig : spectrum](c ) , while the discrete state is the bound state in this potential . for the considered @xmath22-value , i.e. @xmath158
, only one bound state is found for @xmath106 .
the corresponding group velocity @xmath159 is shown in fig .
[ fig : velocityb ] .
notice that the maximum negative velocity is obtained near the anti - crossings in the energy spectrum ( fig .
[ fig : spectrumb ] ) . near @xmath160
the splitting in the energy spectrum ( see fig . [
fig : spectrum ] ) is largest and as a consequence one of the levels is pushed strongly down in energy and consequently @xmath161 becomes strongly negative .
notice that : ( 1 ) every level has some @xmath162 region at which @xmath163 , and ( 2 ) for @xmath164 the velocity @xmath165 , while ( 3 ) the enveloppe of @xmath166 in fig .
[ fig : velocityb ] reaches for @xmath162 the @xmath163 value of the @xmath106 state . for @xmath167
we have @xmath163 for all states . using expressions ( [ eq : clasvel1 ] ) and ( [ eq : clasvel2 ] )
, we can also calculate the classical velocity corresponding to the energy spectrum in fig .
[ fig : spectrumb].@xmath0this is shown in fig .
[ fig : clasvelb].@xmath0we notice that for @xmath168 , the classical velocity has a similar behaviour as the quantum mechanical one , except for the anti - crossings and the lack of negative velocities , which do not have a classical analogon.@xmath0these negative velocities appear suddenly for @xmath169 and exhibit more or less the same behaviour.@xmath0 as was already apparent from the above study a necessary condition for the existence of the non - classical edge states is the presence of two local minima in the effective potential . in the limiting case
@xmath106 the second minima is the limiting case of a flat region in @xmath170 for @xmath171 .
but not all these states have a negative velocity .
how can we classify them ? from fig .
[ fig : spectrum](a , b ) one notices that initially ( for rather small @xmath7 values ) the parabola @xmath172 separates the region where only states with positive group velocity exist , from the region where also negative velocity states are present .
this is due to the fact that the value of this parabola equals the barrier height between the two parabolic wells for the corresponding @xmath22-value .
when the energy exceeds this barrier , the shape of the wavefunction is not determined anymore by the separate parabolas , but by the overall composite well width . for decreasing @xmath22 the well is squeezed , and thus all the energy levels are pushed upwards , resulting in a positive group velocity .
although this is not an exact rule which can not be extended rigorously throughout the @xmath152 regime , it nevertheless provides insight into the @xmath22-values ( or @xmath7 values ) for which these negative velocities states arise .
inspection of the wavefunctions shows that there is a feature which marks the negative velocity states , and which relates indirectly to the presence of the different potential wells .
it turns out that if the wavefunction or its first derivative exhibits a dip at some @xmath2 which satisfies @xmath173 and @xmath174 , or @xmath175 and @xmath176 and the condition that @xmath177 , then the state has a non - classical negative velocity .
this is true for every @xmath22 value , as long as @xmath152 .
this is illustrated in fig .
[ fig : waves_min ] where we plot the wavefunction for @xmath158 , @xmath178 , with background magnetic field @xmath179 and @xmath180 .
the above dip in the wavefunction or its derivative ( indicated by the dashed circle in fig .
[ fig : waves_min ] ) is a result of the different potential wells , which have their separate influence on the shape of the wavefunction , and therefore hamper the matching .
the difference in @xmath181 being zero ( or not ) , can be interpreted as a generalization of matching the individual states in an asymmetric ( symmetric way ) when the landau states are degenerate at @xmath34 .
one can make different attempts to link a classical picture to quantum mechanics.@xmath0often the comparison starts with a schematic classical picture which is then supported by comparing the quantum mechanical probability density with the classical one , obtained through calculation of the classical electron trajectory solving newtons equation.@xmath0for a 1d problem one can also verify the classical motion by inspection of the velocity parallel to the edge.@xcite for a cilindrical symmetric problem , the classical electron motion can be inferred from the magnetic moment or the circular current distribution of the electron state.@xcite in this paper , a quantitative comparison was made by use of a quantum mechanical velocity parallel to the 1d magnetic field discontinuity . in the following , we will try a different approach where we will construct time dependent states , and in doing so we will introduce another feature , i.e. the oscillation frequency perpendicular to the magnetic edge .
we already mentioned before that the solutions for this kind of problem are the parabolic cylinder functions @xmath182 $ ] , matched in such a way that we have symmetric and anti - symmetric wavefunctions as is shown for @xmath26 in fig .
[ fig : trajectories](a ) . at @xmath85
, these symmetric and antisymmetric states are twofold degenerate ( see the two wavefunctions corresponding to the solid square in fig . [
fig : trajectories](a ) ) . due to this degeneracy
any linear combination of these states is also an eigenstate . if we take the following linear combination @xmath183 we arrive at the well known landau states , i.e. wavefunctions of electrons located in two different homogeneous magnetic field profiles .
one electron is moving clockwise , while the other is moving counterclockwise.@xmath0for decreasing @xmath22 this degeneracy is lifted.@xmath0although taking linear combinations of states with a different energy yields a time dependent solution , we will extrapolate this picture towards all the other states .
we choose a new orthonormal but time dependent basis : @xmath184 with @xmath185 . the resulting energy spectrum is shown in fig .
[ fig : spectrum](a ) by the dashed curves.@xmath0the corresponding velocities are plotted in fig .
[ fig : vel_oscil_0](a ) . for every branch
there are two states @xmath186 and @xmath187 . with these new quantum states
much better agreement is obtained with the corresponding classical results ( dotted curves in fig . [
fig : vel_oscil_0](a ) ) .
because of the addition of the two eigenstates the negative velocity almost disappeared.@xmath0only the lowering of the energy , as was mentioned in the limiting case ( i.e. @xmath34 ) , results in a small negative velocity , which ca nt be understood even in this picture.@xmath0also the boundary which indicates when classical states propagate is in much better agreement now.@xmath0 since we now have time dependent states , we can calculate a new feature : the oscillation frequency @xmath188 in the @xmath2-direction.@xmath0the time dependent probability densities of the @xmath186 and @xmath187 states have the following form : @xmath189\left\langle 2m|x\right\rangle \left\langle 2m+1|x\right\rangle ) , \nonumber \\ & = & |\left\langle m_{-}|x\right\rangle ( t+\pi /\omega _ { x})|^{2 } , \label{eq : prob_time}\end{aligned}\ ] ] where @xmath190 is the quantum mechanical oscillator frequency in the @xmath2-direction.@xmath0 classically we can calculate this frequency @xmath188 again , using simple geometrical considerations , which results in @xmath191 both results are plotted in fig .
[ fig : vel_oscil_0](b ) , and we obtain reasonably good agreement between the quantum ( solid curves ) and classical ( dotted curves ) results.@xmath0notice that for @xmath192,@xmath0classically @xmath193 , which means that the electron does not oscillate between the two different magnetic field regions ( i.e. it is not a snake orbit state ) , but it oscillates in a homogeneous magnetic field and consequently we obtain the time independent eigenstates corresponding to the landau levels . of course this approach is only useful if proper linear combinations are possible.@xmath0unfortunately this is not the case when a background magnetic field @xmath194 is applied.@xmath0 the above approach is also fruitful in the case when @xmath137 .
we can again add adjacent levels , two by two , similar as described before.@xmath0we can repeat exactly as was done before , and we arrive again at the time dependent states of eq .
( [ eq : time]).@xmath0the energy spectrum of these states when @xmath195 is shown in fig .
[ fig : spectrum](d ) , by dashed curves.@xmath0from fig .
[ fig : vel_oscil_15](a ) , we notice that the classical velocity is in better agreement then before , since the amplitude of the whiggles is lowered , due to the summation.@xmath0the quantum mechanical oscillation frequency in the @xmath2-direction is again given by @xmath196 and plotted in fig .
[ fig : vel_oscil_15](b).@xmath0we notice that since there are no degenerate states , we always have oscillating electrons.@xmath0for @xmath34 the electron oscillates with frequency @xmath197 , while for @xmath198 the electron oscillates with frequency @xmath199 , i.e. the electrons circle around in their seperate homogeneous magnetic fields.@xmath46this can also be seen from the classical oscillation frequency in the @xmath2-direction , which is given by @xmath200 ^{-1}.\ ] ] notice that also here whiggles in @xmath161 are present ( see fig .
[ fig : vel_oscil_15](b ) ) which are not present in the classical results.@xmath0it is clear that proper linear combinations can always be made , as long as@xmath201 .
we studied the electron states near discontinuities in the magnetic field .
different 1d magnetic field profiles , i.e. steps , were considered .
the quantum mechanical energy spectrum was obtained and the group velocity of the states was calculated .
their corresponding classical orbits were found and the propagating states which are located at the magnetic field discontinuity correspond to snake orbits .
quantum mechanical magnetic edge states were found which move along the magnetic field step in opposite direction to the classical snake orbits and which can not be understood classically .
we were able to construct non stationary quantum mechanical states which closely approximate the classical solution for the symmetrical case @xmath202 and for the more general case @xmath203 .
m. johnson , b. r. bennett , m. j. yang , m. m. miller , and b. v. shanabrook , appl .
lett . * 71 * , 974 ( 1997 ) ; f. g. monzon , m. johnson , and m. l. roukes , appl .
. lett . * 71 * , 3087 ( 1997 ) ; j. reijniers and f. m. peeters , appl .
* 73 * , 357 ( 1998 ) .
y. i. mankov , soviet physics - solid state * 14 * , 62 ( 1972 ) ; y. i. balkarei and l. n. bulaevskii , soviet physics - solid state * 14 * , 2018 ( 1973 ) ; a.s . rozhavsky and r. i. shekhter , solid state commun .
* 12 * , 603 ( 1973 ) ; g. g. cabrera and l. m. falicov , phys .
( b ) * 61 * , 539 ( 1974 ) ; _ ibid .
( b ) * 62 * , 217 ( 1974 ) ; l. berger , j. appl .
phys . * 49 * , 2156 ( 1978 ) ; y. v. zakharov and y. i. mankov , phys .
( b ) * 125 * , 197 ( 1984 ) . | we study the electron motion near magnetic field steps at which the strength and/or sign of the magnetic field changes .
the energy spectrum for such systems is found and the electron states ( bound and scattered ) are compared with their corresponding classical paths .
several classical properties as the velocity parallel to the edge , the oscillation frequency perpendicular to the edge and the extent of the states are compared with their quantum mechanical counterpart .
a class of magnetic edge states is found which do not have a classical counterpart . |
it was recognized many years ago that the cores of clusters of galaxies have sufficient x - ray luminosity to cool 10 to 1000 solar masses of x - ray emitting plasma every year ( e.g. @xcite , @xcite , @xcite ) .
the details of the cooling process are still debated , however , in most models , parcels of cooling plasma collect at the center of the cluster ( @xcite ) , forming what is referred to as a cooling - flow .
this basic picture has remained controversial , however , and a number of observational discrepancies and alternate theoretical interpretations have been raised in the literature . while it is well - established that cores of clusters have short ( @xmath2 @xmath5 yr ) cooling
times ( e.g. @xcite , @xcite , @xcite ) and cluster cores have been shown conclusively to contain much lower x - ray temperature plasma than the ambient hot outer regions ( @xcite , @xcite , @xcite ) , there is no consensus about how much mass has cooled from x - ray temperatures .
some evidence exists for copious amounts of cooler gas emitting in the uv ( @xcite ) , and in h@xmath6 emission ( @xcite , @xcite ) .
in addition , molecular hydrogen has been detected ( @xcite , @xcite ) , and the existence of significant quantities of dust has been inferred from infrared emission ( @xcite , @xcite , @xcite ) . however , hi absorption measurements have found no evidence for cold condensed clouds ( e.g. @xcite ) , and while co emission has been detected ( @xcite ) the amount of cold molecular material is still a factor of 10 below what is predicted by x - ray cooling estimates .
therefore , the connection between these observations and the x - ray data is still unclear , and the exact quantity and location of cooling - flow byproducts is unresolved .
we refer to this as the _ classic cooling - flow problem_. the bulk of the thermal energy of the cooling intracluster medium is thought to radiate at x - ray wavelengths , and thus studying the x - ray spectrum is critical to understanding cooling - flows and testing cooling - flow models .
if the plasma cools homogeneously , the density profile of the cores of clusters should be much steeper than observed ( @xcite ) , which has led to the conclusion that the cooling plasma must condense locally into smaller clouds distributed over a large volume ( tens of kpc ) , i.e. in a multi - phase medium .
however , even ignoring the details of the resulting spatial distribution , simple thermodynamic arguments show that the integrated x - ray spectrum of such a cooling flow can be robustly predicted .
if the blobs of plasma cool in thermal isolation at constant pressure , and the dominant energy loss mechanism is via x - radiation , then the differential luminosity distribution , i.e. the luminosity radiated per unit temperature interval , must be proportional to the mass deposition rate , @xmath7 : @xmath8 where @xmath9 is boltzmann s constant , and @xmath10 is the mean mass per particle .
the only free parameter in this expression is @xmath7 , which can be estimated from the density distribution inferred from the x - ray image of the cluster core .
the resulting spectrum can be calculated using a collisional equilibrium spectral synthesis model with an assumed set of elemental abundances , and normalizing the contribution in each temperature interval as given in equation ( 1 ) . to a good approximation
, a cluster spectrum should consist of two components : 1 ) the cooling - flow spectrum , as described above , and 2 ) an isothermal spectrum evaluated at the temperature of the background cluster gas .
data acquired by the reflection grating spectrometer ( rgs ) on xmm - newton enable this robust spectral prediction to be quantitatively tested for the first time .
surprisingly , the observed spectra reveal a remarkable systematic deficit of emission at the lowest temperatures , as compared to the multi - phase model ( @xcite , @xcite , @xcite , @xcite , @xcite , @xcite ) .
this result has been confirmed through medium resolution spatially - resolved spectroscopy using the xmm - newton european photon imaging cameras ( epic ) and chandra acis observations , where spectral fits have yielded significantly smaller @xmath7 s than expected ( @xcite , @xcite , @xcite , @xcite , @xcite , @xcite ) .
we refer to the observed deficit of the predicted soft x - ray emission as the _ soft x - ray cooling - flow problem_. as we discuss in this paper , the soft x - ray cooling - flow problem may or may not be related to the classic cooling - flow problem . here
we present a systematic study of 14 clusters with the rgs to quantify in detail the soft x - ray cooling - flow problem .
we present four basic results : 1 ) we find no evidence for x - ray absorption by cold material , which had been inferred from lower spectral and angular resolution data @xcite .
2 ) we find a severe lack of emission from lower temperature ions expected in the standard cooling - flow model .
3 ) we demonstrate the ubiquity of significant plasma just below the ambient gas temperature , @xmath0 , down to @xmath1 in roughly the predicted amount .
this last result is the most perplexing and is difficult to reconcile with proposed explanations for the cooling - flow problems .
4 ) we offer an empirical parameterization of the temperature distribution which is consistent with the entire sample .
the paper is organized as follows . in
2 we discuss the expected spectrum of the isobaric radiative multi - phase model and various associated temperature diagnostics . in
3 we discuss the spectral capabilities of the rgs for sources of moderate extent . in
4 we describe our analysis methods and present the high resolution spectra . in 5
we discuss several qualitative aspects of the spectra . in 6 we describe the model used to set cooling luminosity limits . in 7
we present the results . in 8
we discuss the results with respect to potential modifications of the cooling - flow scenario .
it has been argued that the intracluster medium ( icm ) should remain in collisional ionization equilibrium , even as it cools .
both the recombination time scale and electron equilibriation time scale through coulomb collisions , are much less than the cooling time scale , evaluated at characteristic x - ray temperatures and densities in the icm ( @xcite , @xcite ) . at high temperatures , the x - ray spectrum is produced through bremsstrahlung and line emission from abundant elements .
the spectra are dominated by lines from hydrogen - like and helium - like ions , and iron l shell emission . in principle , the temperature distribution can be determined by three different methods : 1 ) through the shape of the exponential cut - off of the bremsstrahlung spectrum , 2 ) the ratios of hydrogen - like to helium - like lines , and 3 ) through the distribution of line emission from iron l - shell ions .
however , the bremsstrahlung spectrum constrains only the highest temperatures present in the spectrum , and the ratio of hydrogen - like to helium - like lines is relatively unconstraining since both charge states are present over a very broad range of temperatures .
therefore , fe l emission is the most useful diagnostic , and provides the primary method for determining the temperature distribution presented in this paper .
the isobaric multi - phase model yields a unique spectral signature given an isothermal outer cluster temperature and a set of assumed abundances .
this is shown in figure 1 for a maximum temperature of 8 kev and for 1/3 solar abundances . in figure 2
, we divide the temperature distribution below 6 kev into four bands .
the curves are the spectra produced by emission from the multi - phase model in bins of 3 to 6 , 1.5 to 3 , 0.75 to 1.5 , and 0.375 to 0.75 kev .
important line blends are compiled in table 1 .
several important diagnostics and aspects of the model can be noted .
the first is that line emission from ions between fe xxiv and fe xvii provide tight constraints on the temperature distribution .
there is roughly comparable emission in the emission line blends from each fe l ion .
also note that fe xvii line emission is predicted to be three times as strong as fe xviii .
o viii and other hydrogen - like ions are produced at a range of temperatures , and thus do not provide strong constraints on the temperature distribution .
the overall normalization is proportional to the mass deposition rate , and the spectrum should be a strict superposition of all the individual temperatures up to the ambient temperature .
in order to test aspects of this model in detail , the fe l spectrum needs to be resolved for a spatially - extended object .
this could not be accomplished with previous instrumentation ; however , the launch of the xmm - newton rgs experiment now makes it possible . for a full description of the rgs experment see @xcite .
the two rgs spectrometers have 160 @xmath11 of combined collecting area and with a spectral resolution given roughly by : @xmath12 for an extended source larger than 10 arcseconds .
the wavelength band extends from 5 to 38 @xmath13 , which samples si k shell transitions to c k transitions .
the field of view is effectively 5 arcminutes by 1 degree , which is large enough to capture the entire cooling - flow region .
the former is set by the width of the ccd array in the cross - dispersion direction , and the latter is set by self - vignetting of the telescope shells . for the cooling - flow clusters , data selection cuts are tailored to include mostly emission from the cores of the clusters . the background is produced primarily by a variable flux of soft protons spread nearly uniformly across the focal plane .
there is no source - free region for extended sources , so it has to be modeled .
the cross - dispersion direction provides a one - dimensional image ( 11 arcseconds fwhm ) of the source , which we use to construct a model of the surface brightness of the source .
our sample of clusters ( listed in table 2 ) is a biased selection of 14 compact clusters and groups of galaxies chosen to exploit the spectral sensitivity of the rgs .
the sample includes clusters and groups at a range of temperature from 1 to 10 kev .
all of these clusters and groups were expected to host cooling - flows ranging from 1 to 1000 solar masses per year ( see table 3 ) . to produce effective - area and exposure - corrected spectra of these objects ,
we adopt the following procedure : we process the events by a development version of the science analysis system ( sas ) software ( version 5.3 ) , which accomplishes event reconstruction , aspect correction , ccd - pulseheight corrections , and focal plane reconstruction .
we only use events in time intervals where the background was less than 6 counts per second in the 0.35 to 1.9 kev energy band .
we then select photon events satisfying a 2 arcminute wide cross - dispersion cut .
events must also satisfy a first or second order joint dispersion - coordinate / ccd - pulseheight cut based on the characteristic resolution of the ccd , and additional broadening due to the 2 arcminute source extent . in this way
, we select photons roughly produced in a 2 arcminute square centered on each cluster .
there is also a significant contribution of photons from outside this square , which is self - consistently in our modeling .
finally , a wavelength is assigned to each photon based on the nominal center of the cluster and the instrument bore - sight . in order to account for the instrument response and efficiency of these selection cuts
, we use a monte carlo method , as discussed in detail by @xcite .
the monte carlo method is used to account for off - axis behavior of the response and the effect of arbitrary selection cuts and transformations made on the data .
additionally , it is required in the astrophysical modeling as we discuss in 6 .
the monte carlo calculation contains all known effects of the mirror shells , grating arrays , and ccd response .
it has calibration limitations similar to the response in sas version 5.3 .
the monte carlo method generates focal plane coordinates and ccd - pulseheights based on a model for the x - ray spatial and spectral distribution .
we generate photons having a flat wavelength spectrum and surface brightness profile given by a modified @xmath14 model ( described in 6 ) .
the parameters are listed in table 3 .
the monte carlo method uses focal plane maps of the exposure time produced by the sas to select detected events .
the same selection cuts that are applied to the data , are applied to the simulated photons as well to account properly for the efficiency of the selection regions . extracting the simulated photons as we did for the source photons
produces a wavelength histogram in units of @xmath15 .
we then take the histogram of the source photons and subtract a set of background simulated events .
the background model is described in detail in 6 .
the final histogram is divided by the exposure - area histogram and produces the fluxed spectra shown in figures 3a , 3b , and 3c . the spectra are not corrected for absorption in the interstellar medium .
the spectra shown in the three panels of figure 3 are sorted inversely by total luminosity , and therefore also roughly inversely by temperature and @xmath7 .
below , we discuss the spectra grouped by their outer temperature . * 7 - 10 kev clusters * : the rgs spectrum of abell 1835 has already been presented in @xcite .
the spectrum exhibits intermediate temperature ( @xmath16 @xmath4 3 kev ) plasma in addition to the 8 kev background plasma as shown by the detection of fe xxiv and xxiii .
o viii ly@xmath6 is clearly detected .
@xcite identified the gross inconsistencies between the measured spectrum of this source and the spectrum predicted by the standard cooling - flow model .
these constraints are derived from the lack of fe xvii - xxii .
abell 665 has a similar spectrum , but the observation of that source was plagued by a very high background rate .
* 4 - 7 kev clusters * : abell 1795 , hydra a , abell 496 , and abell 4059 are 4 to 7 kev clusters .
fe xxiv - xxii , o viii , mg xii , ne x , and si xiv emission lines are detected in all cases .
the strong fe xxiv - xxii complex indicates significant cooling down to @xmath17 @xmath4 2 to 3 kev .
there are no detections of fe xvii - xxi , however , in clear contradiction to the predictions of the cooling - flow model for these systems .
* 2 - 4 kev clusters * : the lowest temperature clusters , 2a0335 + 096 , srsic 159 - 03 , abell 262 , abell 1837 , mkw3s , abell 2052 , and m87 have the bulk of their emission from the fe l temperature range .
these spectra show clear detections fe xxiv - xix , o viii , mg xii , ne x , si xiv . various different ions produce stronger emission lines depending on the cluster temperature .
weak emission from the helium - like charge states are observed for silicon and magnesium , and in the best resolved spectra hydrogen - like c and n are detected .
these spectra also generally have weak emission from fe xvii and xviii indicating some plasma below 1 kev , but less than would be predicted by the standard cooling - flow model . *
1 kev groups * : ngc 533 is the lowest temperature system and has emission from the same ions as in the 2 - 4 kev clusters . in this case , fe xvii and xviii are strong due to the lower temperature of the system . however
, this spectrum is also inconsistent with the standard cooling - flow predictions , since fe xvii is not much stronger than fe xviii and ovii is not detected . *
composite spectrum * : as further confirmation of the reality of some of the detections discussed above , we generate a composite spectrum by combining all of the individual clusters .
this composite is shown at the bottom of figure 3 .
it was constructed by first shifting the wavelengths of photons by the cosmological redshift and then co - adding the counts - weighted spectra .
the individual area - exposures were also added using the same technique .
the counts - weighted spectrum was then divided by the sum of the area - exposures .
we excluded m87 from this analysis since it would otherwise dominate the result . in the composite spectrum , lines from si xiv & xiii , al xiii & xii , mg xii & xi , all fe l ions , ne x , o viii , c vi , and n vii are all clearly visible .
no neutral o k edge is observed at the redshift of the clusters .
this would be the most prominent absorption feature if a large amount of cold material with a large covering fraction was absorbing the flow .
a feature which appears to be a blue - shifted o k absorption edge is due to galactic absorption , and its depth is consistent with the mean interstellar column density .
several qualitative conclusions can be drawn from the observations : first , in all cases , there is a deficit of emission expected from lower temperatures .
second , in most cases , a continuous distribution of temperatures is nevertheless required . finally , absorption by intervening cool gas at the redshift of the cluster can not explain the dearth of soft x - ray emission . in 6 below , we perform quantitive spectral fits to further explore the departures from the cooling - flow model .
in this section , we set quantitative limits on emission from various parts of the temperature distribution . to derive these limits
, we use the monte carlo methods presented by @xcite , and implicitly used by @xcite and @xcite .
the basic procedure is based on an astrophysical model for the spatial and spectral dependence of the emission .
the monte carlo approach is required because a different spectrum at each projected spatial position is required to test the cooling - flow model in detail .
we randomly generate photons with an associated dispersion angle , cross - dispersion angle , and ccd pulseheight value .
the resulting count distributions are then compared with the raw data , after the various data selection cuts and transformations . in fitting for global parameters , such as the background normalization
, we find the best fitting solution by using multivariate methods , as described in @xcite .
for the abundances and temperatures , we use a @xmath18 statistic of the combined extracted first and second order spectra for both instruments and we iteratively adjust the surface brightness distribution to match the cross - dispersion profile . for the cluster emission
we adopt a relatively simple model , so as to reduce sensitivity to fitting biases . for the surface brightness
, we use a spherical @xmath14 profile where the core radius is left free and the @xmath14 parameter is fixed to the value determined from epic spectral fits ( @xcite ) .
additionally , we allow the normalization of the emission inside of a three - dimensional radius , @xmath19 , to be larger than the normalization outside of the radius . in this way
, the spatial profile can be much more peaked than the standard @xmath14 profile .
this three dimensional distribution is then projected on the sky .
the precise shape of the spatial distribution is not critically important for determining the cooling luminosity limits as long as it reproduces the rough behavior of the emission profile . outside of @xmath19 ,
the emission is set to an isothermal temperature . inside that radius ,
we fit for the normalization of the differential emission measure distribution below the upper temperature , @xmath0 .
the emission measure is divided into several temperature bins between @xmath20 and @xmath0 , @xmath21 and @xmath20 , @xmath22 and @xmath21 , and @xmath23 and @xmath22 .
this choice is arbitrary , but it provides robust fitting solutions since the fractional ionization curves are roughly equally spaced in the logarithm of the temperature .
preliminary fits indicate that only negligible hot ambient plasma coexists inside the cooling radius , so additional isothermal emission inside this radius is ignored . within each temperature bin
, we use the isobaric radiative cooling - flow model to predict the shape of the emission measure distribution , but this is not critically important since we do not have the spectral sensitivity to sample the emission measure distribution in very fine intervals .
this approach clearly forces the coolest emission to conform to a specified spatial distribution .
further detailed analyses are required to pinpoint its exact spatial location .
our spatial model , however , seems to be compatible qualitatively with the relatively small differences in the observed emission lines spatial profiles of m87 ( @xcite ) .
we assume uniform spatial abundances throughout the cluster emission .
this considerably simplifies the fitting procedure , and makes the limits on the coolest emission conservative .
it is known that abundance gradients exist in clusters , but they are generally flat within the cooling - flow region and weaken whenever more freedom is given to the temperature distribution @xcite .
the iron , neon , oxygen , magnesium , and silicon abundances are left as free parameters .
all other elements are tied to iron since they contribute few counts . in abell 1835 and
abell 665 , the magnesium , neon , and silicon abundances are tied to iron since the emission is from very high temperatures .
the absorption column density is left as a free parameter to account for variations along the line of sight to the cluster . in ngc 533
the absorption is set to the galactic value since it has little low energy continuum emission .
we also ignore the effects of resonant scattering ( cf .
@xcite ) , which redistribes emission line photons within our aperture , but otherwise results in the same detected emission line flux . for the background ,
we use a semi - empirical model calibrated on blank sky lockman hole observations ( xmm revolutions 0070/0073 ) .
the model includes a spatial model for soft protons , low energy detector readout noise , and characterizations of the in - flight al k and f k calibration sources .
all parameters are frozen in this model except the relative normalization of the particle component , and the overall normalization which can vary by factors of 10 from observation to observation .
the background is relatively flat in wavelength .
the model has the following free parameters : local column density , normalization of each part of the cooling flow emission measure distribution , abundances of magnesium , neon , silicon , oxygen , and iron , background temperature ( @xmath0 ) , particle background normalization , position of the source in the cross - dispersion direction , core radius , cooling radius , and overall normalization . for each spectrum
, we apply selection cuts discussed in 4 .
however , the background model parameters and position of the source are determined before applying any data selection cuts .
in addition to the model described above , we also set limits on a model wherein an intrinsic absorber embedded in the cooling - flow volume is invoked to suppress the expected soft emission . if the absorber is evenly embedded , the transmission function is given by , @xmath24 , where @xmath25 is the photoelectric optical depth as a function of energy , as opposed to the usual exponential form .
for this case , we otherwise take the isobaric cooling - flow emission measure distribution , without allowing any variation from one temperature bin to the next .
the errors are quoted at the 90@xmath26 statistical confidence level for setting limits on the parameters . the uncertainty in the wavelength scale of 8 @xmath27 and the characteristic line spread function uncertainty of 5 @xmath27 make negligible contributions to these errors for extended sources like clusters .
the effective area uncertainty is of order 10@xmath26 .
additionally , there is known to be non - statistical noise at the 5@xmath26 level of the source flux in the rgs spectra , caused by systematic dark current variations which can sometimes produce false features .
either of these effects can produce correlated errors in adjacent wavelength channels .
the global temperature , absorption column density , and overall normalization depend on many wavelength channels and are thus dominated by systematic uncertainties of this kind .
these parameters are therefore never quoted with uncertainties below 200 ev , @xmath28 , or 10@xmath26 in the normalization , respectively , which would all modify the spectrum characteristically by @xmath29 10@xmath26 .
we adopt the mekal plasma model @xcite as implemented in xspec v11 ( @xcite ) as the collisional ionization equilibrium model used to predict the spectrum .
the absorption cross - sections are taken from @xcite .
abundances are quoted relative to @xcite .
galactic absorption column densities are compared to @xcite .
we assume @xmath30 km / s / mpc , @xmath31 , @xmath32 throughout our analysis .
finally , all measured spectral mass deposition rates are compared to those derived in the corresponding epic data set .
that analysis takes into account temperature and density gradients by solving the hydrodynamic equations , assuming only radiative cooling and mass drop out according to standard techniques ( e.g. @xcite ) .
the mass deposition rates are interpolated at the radius for our @xmath19 parameter and are within uncertainties ( typically 50@xmath26 ) of other published values .
a more extensive discussion of the epic analysis will be presented elsewhere ( @xcite ) .
our derived spectral fits are shown in figure 4a , 4b , and 4c and the cross - dispersion profiles are shown in figure 5 .
the best fit parameters are given in tables 3 , 4 and 5 .
in each plot the red line shows the fitted empirical model .
the green line shows the standard isobaric cooling - flow model without altering the temperature distribution and maintaining the same normalization . in almost all cases , the spectral features and the intrinsic line profiles
are well - fit for the empirical model . due to the low signal to noise of the observation of abell 1837 and abell 665
, we could not place strong constraints on the cooling - flow model for those clusters .
the other clusters , however , show clear deviations from the cooling - flow model .
there are two significant residuals that should be noted .
the long wavelength region of the spectrum is not perfectly fit .
we believe this is due to errors in the effective area calibration , imperfect background modeling , and spatially broadened ( @xmath29 3 ) emission lines of o viii , o vii , n vii , and c vi emission lines from the diffuse soft x - ray background ( see @xcite ) .
these discrepancies have a minimal effect on our conclusions about the temperature distribution , however , since those rely primarily on the fe l region .
another problem is evident in the detailed line fitting of the spectrum of ngc 533 .
this is probably due to an underprediction of the 2p-3s lines of fe xvii and fe xviii ( at 16 and 17 , respectively )
. there is a known problem with the excitation rates for these transitions @xcite , as has been discussed for the case of ngc 4636 by @xcite and capella by @xcite .
the differential luminosity distribution for all clusters is plotted in figure 6 .
the luminosity is normalized with respect to the prediction from the isobaric radiative multi - phase model using equation ( 1 ) and @xmath7 from table 4 .
the presence of points above the dotted line is not unexpected , since these can result from the background uncooled hot plasma .
this can also result from the ambiguity in defining a distinct cooling radius .
however , the data should closely track the line @xmath33 at lower temperatures , if the cooling - flow model is correct .
many upper limits , however , are well below that line at temperatures near a third of the background temperatures .
as can be seen , the well - measured clusters show clear deviations from isothermality with significant plasma existing down to one - quarter of the background temperature . in figure 7
, we plot the same results with the horizontal axis now scaled to the background temperature . here
the discrepancies from the cooling - flow prediction look more systematic .
in particular , the differential luminosity distribution appears to be roughly consistent with the expression , @xmath34 where @xmath35 to @xmath36 instead of 0 , as expected for the isobaric radiative cooling - flow model .
there is still significant scatter in the normalization and slope of this relation , which could be due to various systematic effects , or to real differences between the cooling - flows .
further detailed analyses that start with the assumption of equation ( 3 ) might yield further insight into the scatter .
however , it is clear that no cluster has a temperature distribution consistent with the @xmath37 case , and that in each case the failure of the cooling - flow model occurs at a fraction of the virial temperature rather than at a fixed value .
the derived abundances as a function of ambient temperature are plotted in figure 8 .
the abundance of iron declines slightly with more massive clusters as indicated in earlier asca observations ( @xcite , @xcite ) , although the abundance measured here is somewhat higher .
the emission weighted iron abundance ranges from one third to 70@xmath26 with higher values in lower mass clusters .
this presumably reflects the increase in stellar mass fraction for lower mass systems .
there are no obvious trends in the other abundance ratios and there is considerable scatter in the points .
this is possibly caused by subtle fitting biases .
in particular , the abundances reflect the emission - weighted abundance average in an aperature that varies from cluster to cluster and there is often an anti - correlation between the absolute abundances and the width of the differential emission measure distribution .
additionally , abundances of elements other than iron are measured off the peak of their fractional ionic abundance , and depend sensitively on the temperature distribution .
nevertheless , all of the abundances seem to cluster around particular ratios .
the oxygen to iron ratio is 70 @xmath38 20 @xmath26 ( 1 @xmath39 ) of the solar value .
the magnesium to iron ratio is 100 @xmath38 40 @xmath26 , the neon to iron ratio is 110 @xmath38 40 @xmath26 , silicon to iron is 230 @xmath38 80@xmath26 .
abundance patterns for at least these four elements resemble those found in @xcite suggesting a common enrichment history between ellipticals and clusters .
these particular ratios do not appear to fit into a simple superposition of type ia yields plus type ii supernovae integrated over a single initial mass function , as in @xcite .
they more closely resemble low - metallicity high - mass supernovae yields , but further analyses with more attention to additional elements are needed to establish the pattern completely and determine the spatial distribution .
additionally , carbon and nitrogen abundances could be roughly estimated in m87 and abell 496 to be @xmath40 @xmath29 2 and @xmath41 @xmath29 2.5 with large uncertainties .
this implies , however , a non - negligible fraction of chemical enrichment has been processed through stellar winds .
further detailed analyses of the measured abundances and their spatial distributions will be presented elsewhere ( @xcite ) .
the derived intervening column densities are generally consistent with the expected galactic values .
the only large deviation is with 2a0335 + 096 , which has a high galactic column density and the measurements could be more affected by systematic uncertainties .
in addition , allowing an intrinsic absorber with the standard cooling - flow model produced worse fits than the standard cooling - flow model alone .
although the absorber can remove significant cold emission with column densities in excess of @xmath42 , it also removes the low energy tail of the bremsstrahlung due to the core , which provides a significant fraction of the continuum .
limits on intrisic absorbers are given in table 4 and are generally less than @xmath43 .
in 7 , we demonstrated the presence of significant quantities of plasma just below the ambient temperature , @xmath0 , down to @xmath1 and the clear failure of the cooling - flow model at lower temperatures .
we also showed the consistency of the sample with the empirical differential luminosity distribution given by equation ( 3 ) with the parameter , @xmath6 , equal to @xmath4 1 to 2 . here , we consider several ideas which have been suggested to explain these results .
we first discuss mechanisms which modify the isobaric cooling - flow model , but involve no real modification to the general paradigm of radiatively - driven flows in gravitationally - relaxed cluster cores .
then , we discuss substantial modifications to the general cooling - flow process by inclusion of additional heating mechanisms or non x - ray cooling channels . we have compared the data to the standard isobaric cooling - flow model . a modification to this model
is expected if a weak magnetic field is amplified as cooling plasma compresses in order to maintain pressure equilibrium . at the lowest temperatures then
, these blobs cool isochorically .
this would only modify the expression in equation ( 1 ) by changing the @xmath44 to @xmath45 , which is still inconsistent with the data displayed in figures 6 and 7 .
the compression of the gas due to the dark matter - dominated gravitational potential is included in our calculation of @xmath7 by measuring directly both the temperature and density spatial gradients , and has generally been included in previous work ( e.g. @xcite , @xcite ) .
this is important if much of the gas is actually flowing , and the effect on the temperature distribution is given in @xcite .
it has a relatively small effect on the spectra apart from the normalization , however , and should not affect the luminosity at the lowest temperatures since most of the plasma is thought to drop out of the ambient medium before it flows ( @xcite ) .
even if there is no mass drop out , however , the @xmath44 in equation ( 1 ) , would only be replaced by @xmath46 where @xmath47 is the radial logarithmic temperature gradient and @xmath48 is the radial logarithmic density gradient .
@xmath49 is always observed to be negative in the cores of clusters .
two recent explanations to the soft x - ray cooling - flow problem have been proposed that involve no real modification to the multi - phase model in equation ( 1 ) .
first , @xcite suggested that differential absorption by colder material could absorb more of the emission from cold ions .
this is somewhat implausible , since one would expect hot and cold parcels of plasma to have roughly equal probability of being obscured .
nevertheless , the result could be interpreted in the context of this model if the effective absorption column density scaled as a strong function of the occupied volume .
this would require the absorber to have an almost identical spatial distribution to the currently cooling parcels of gas from the cooling - flow and a small filling factor .
it is impossible to rule out this model in its most extreme form , since an infinite column density selectively applied to emission at a given temperature is , of course , identical to no emission at that temperature .
the most model - independent limits against this explanation comes from @xcite looking at absorption in spectra of agn in clusters .
an explanation offered by @xcite and developed in @xcite involved a locally non - uniform distribution of metals in the icm , which modifies the spectrum significantly .
this requires that the metal abundance variations occur on scales smaller than cooling parcels of plasma , and that the variations sample a particular bimodal distribution . in its simplest form
, this model would predict that the cooling - flow predictions would exceed the measured spectra near roughly a common temperature near 1 kev , where line cooling dominates over bremsstrahlung for super - solar abundances .
that is not what is observed , however , and this model is therefore not compatible with our results for the low temperature groups . a variety of additional physical processes
have been proposed to operate in the cores of clusters , which could , in principle , account for the soft x - ray cooling - flow problem ( @xcite , @xcite , @xcite , @xcite , and @xcite and references therein ) .
these include time - dependent heating by agn outflows ( e.g. @xcite , @xcite , @xcite ) , electron thermal conduction from the outer regions of clusters ( e.g. @xcite , @xcite , @xcite , @xcite , @xcite , @xcite ) , continual sub - cluster mergers ( @xcite ) , and interactions of the icm with dark matter ( @xcite ) .
others invoke rapid cooling induced by mixing at interfaces with cold clouds ( @xcite , @xcite ) , dust mixing ( @xcite ) , or acceleration of relativistic particles .
most calculations have succeeded in establishing that the above mechanisms are energetically important , i.e these mechanisms have total heating powers that can roughly cancel radiative cooling energy losses , or that alternative cooling channels have sufficient luminosity to cool the plasma through other means .
the total energy balance ( i.e. the sum of cooling luminosity minus heating power ) determines the total mass deposited from the cooling - flow , and thus directly addresses the classic cooling - flow problem .
the total energy balance , however , only affects the soft x - ray cooling - flow problem by reducing the total normalization of equation ( 3 ) by a fixed value . it does not change the predicted differential luminosity distribution . the critical difference , however , between the soft x - ray cooling - flow problem and the classic cooling - flow problem is that the latter requires a clear explanation for why x - ray cooling does not appear to be carried to completion .
it is difficult to find a relevant heating , mixing , or cooling time scale that would be comparable to the x - ray cooling time . if the time scale for a given process is too short it will overwhelm the cooling - flow ( @xmath50 ) , and if the time - scale is too long , it is dynamically unimportant ( @xmath37 ) .
below , we discuss the proposed physical processes and whether they can account , both energetically and dynamically , for the missing soft x - ray luminosity .
there are three requirements for additional heating mechanisms to be compatible with the observations .
first , the total time averaged heating power , @xmath2p@xmath51 , has to roughly cancel the radiative losses so that the expression , is near unity .
the denominator in the expression varies among the clusters in our sample by four orders of magnitude , so the proposed heating process must operate on a variety of scales .
second , the heating has to be distributed spatially throughout the cooling - flow volume to cancel cooling everywhere .
third , the process has to be self - regulating , so that the time scale for heating remains comparable to the cooling time for all clusters . *
agn outflows : * time - dependent agn outflow heating models have been considered by a variety of authors ( e.g. @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite ) .
buoyant bubbles carrying relativistic plasma appear to be a common phenomena in clusters with central agns .
the thermal energy seems to be enough to heat cooling - flows through cosmic ray interactions and mechanical heating , but it is unclear whether this energy gets properly channeled into the cooling volume ( e.g. @xcite , @xcite ) . note that it is essential that the heat be distributed evenly throughout the region which is thermally unstable .
in addition these models must be made self - regulating to counteract cooling at a rate proportional to the mass deposition rate , and with periods of heating roughly as long as periods of cooling .
clearly , it requires a significant degree of fine - tuning . * electron thermal conduction to the background plasma : * there is considerable thermal energy in the outer regions of clusters that can destroy any existing cooling - flow through electron thermal conduction ( e.g. @xcite , @xcite , @xcite ) .
the size of cooling - flows are only a few electron mean free paths in the absence of magnetic fields .
the critical question is to what level is conduction suppressed by tangled magnetic fields , an issue which continues to be debated theoretically ( @xcite , @xcite ) .
observationally , conduction is suppressed by factors near 100 in identified cold fronts ( @xcite , @xcite , @xcite ) .
@xcite , @xcite , @xcite have demonstrated that the heat flow from the outer regions of clusters with a small suppression ( @xmath51 @xmath53 ) in the spitzer conductivity would appear to cancel radiative losses in many clusters .
the spatial distribution of the heating and overall energetic requirements appear to be satisfied by conduction models , but there is no explanation for why the cluster would cool to their current temperature distribution . since conduction suppresses temperature gradients by definition , this mechanism alone does not solve the dynamical problem presented here .
* sub - cluster mergers : * @xcite has shown that clusters previously thought to be fully relaxed exhibit temperature fronts consistent with plasma exhibiting large bulk motions in the gravitational potential .
this provides another source of energy that has not been dissipated and therefore leads to an increase of the cooling time above previous estimates ( @xcite ) , so that the time scales for radiative cooling and dynamical relaxation may be similar .
there is no reason for this process to be self - regulating , however , and we would expect a much larger difference in the temperature distributions , depending on each cluster s merger history .
this explanation also requires a conspiracy of factors to both cancel radiative cooling in global energetics , and ensure that the mergers occur with a frequency that allows some cooling .
future numerical simulations may test this further , and presumably observations of cooling - flows at a different epoch would not show the same effects .
the radiative isobaric cooling - flow model assumes that all of the thermal energy is released in x - rays at high temperatures .
there may , however , be additional contributions from other cooling processes .
there are three main requirements for additional cooling channels to explain the observations , which are similar to , but slightly different from the heating requirements .
the first is that the total power in the coolant be comparable to the missing soft x - ray luminosity .
any coolant with power , @xmath54 , reduces the total x - ray emission by a factor , the second requirement is that the ratio @xmath56 needs to have a temperature dependence consistent with equation ( 3 ) or that the cooling channel is self - regulating in the same sense as the discussed heating models .
the third requirement is that the energy should be released with a similar spatial distribution to the lowest temperature x - rays .
* uv cloud interfaces : * @xcite and @xcite discussed the possibility that hot electrons are cooled conductively by interfaces with cold clouds .
this leads to emission in the uv where the cooling function is the highest , and is consistent energetically with the large observed h@xmath6 luminosities ( @xcite , @xcite ) of @xmath57 to @xmath58 ergs / s in cooling - flows .
however , by itself , this model does not explain the observed temperature distribution in the soft x - ray band , since high temperature electrons are cooled by this process as well .
in addition , in such a picture highly charged ions should also impact the cloud interfaces , resulting in charge exchange , which produces copious soft x - ray line emission .
the spatial distribution of h@xmath6 is remarkably similar to the coolest x - rays ( @xcite ) and the total luminosity is marginally sufficient to account for the missing soft x - ray luminosity ( @xcite ) , but again the dynamical problem of partial cooling is not easily solved by this mechanism alone .
* dust mixing : * some clusters have been established as strong ir sources , typically of order @xmath59 to @xmath60 ergs / s @xcite .
@xcite demonstrated that in all systems studied , the missing thermal energy in soft x - rays is more than accounted for by the ir luminosity .
hot electrons impinging on dust grains could collisionally heat the grains and this in turn cools the hot electrons .
the cooling function for this process has been discussed in @xcite and @xcite and is 100 times larger than the x - ray cooling function for a typical ism composition of dust grains .
as such , this explanation appears promising , but there is no strong temperature dependence to either the x - ray or dust cooling function that would systematically reduce the overall level of soft x - ray emission . presumably , the success of this process depends on magnetic field topologies and a steady - state process of destruction and creation of dust grains .
suppression of conduction by magnetic structures in some regions could play a critical role in regulating both uv cooling and dust mixing , and fast magnetic reconnection could in turn realign those structures . the dynamical problem remains unsolved here too , and if dust cooling is important for cluster x - ray plasmas , it creates an even larger cooling - flow problem .
* relativistic particles and alfvn waves : * the increase in magnetic field density as cooling blobs compress may allow for considerable energy to be released via alfvn waves or mildly relativistic particles .
in fact , the steeper synchrotron indices and short particle lifetimes in cluster radio mini - haloes have been invoked to argue for reacceleration of particles in cooling - flows ( e.g. @xcite , @xcite ) .
this process should , however , only liberate a small fraction of the thermal energy and it provides no natural explanation for the observed temperature distribution .
implicit in our discussion above is that only one mechanism explains the observations . given that many of the above ideas are energetically viable , that may be unlikely .
it may be that the soft x - ray cooling - flow problem and the classic cooling - flow problem are solved by different solutions .
it may also be possible that a combination of processes could have different effects .
for example , heating cooling - flows both from the inside with an agn and from the outside through conduction , results in a more stable heating method than each mechanism does individually ( @xcite ) .
we have demonstrated that the soft x - ray cooling - flow problem is not an isolated phenomenon and appears on a range of scales .
the empirical characterization offered here may provide clues to the ultimate solution . clearly , the model for the multiphase distribution is inadequate .
more detailed comparisons between observations at x - ray and other wavelengths may more carefully address which of the physical mechanisms are important energetically .
much more theoretical work is needed in addressing why any of the proposed mechanisms would dynamically conspire to produce the observed temperature distribution .
these descriptions , however , are just beginning to be developed and it remains a theoretical challenge either to cool kev plasma with little resulting soft x - ray radiation , or to balance radiative cooling both energetically and dynamically .
this work is based on observations obtained with xmm - newton , an esa science mission with instruments and contributions directly funded by esa member states and the us ( nasa ) .
work on the rgs at columbia university and u. c. berkeley is supported by nasa .
the laboratory for space research , utrecht is supported by nwo , the netherlands organization for scientific research .
jrp acknowledges many helpful conversations with other members of the columbia team .
lllllll & & kev + fe xxiv & 10.6 , 11.2 & 0.9 @xmath61 4.0 & & o viii & 19.0 , 16.0 & @xmath510.2 + fe xxiii & 11.0 , 11.4 & 0.8 @xmath61 2.0 & & o vii & 21.6 , 22.0 & 0.1 @xmath61 0.2 + & 12.2 & & & si xiv & 6.2 & @xmath511.0 + fe xxii & 11.8 , 12.2 & 0.6 @xmath61 1.5 & & si xiii & 6.6 , 6.7 & 0.2 @xmath61 1.0 + fe xxi & 12.2 , 12.8 & 0.5 @xmath61 1.0 & & al xiii & 7.2 & @xmath511.2 + fe xx & 12.8 , 13.5 & 0.4 @xmath61 1.0 & & al xii & 7.8 , 7.9 & 0.3 @xmath61 1.2 + fe xix & 13.5 , 12.8 & 0.3 @xmath61 0.9 & & mg xii & 8.4 & @xmath510.7 + fe xviii & 14.2 , 16.0 & 0.3 @xmath61 0.8 & & mg xi & 9.2 , 9.3 & 0.1 @xmath61 0.6 + fe xvii & 15.0 , 17.1 & 0.2 @xmath61 0.6 & & ne x & 12.2 & @xmath510.4 + & 15.3 , 16.8 & & & ne ix & 13.5 , 13.7 & 0.1 @xmath61 0.3 + n vii & 24.8 & @xmath510.1 + c vi & 33.7 & @xmath510.1 + lrrrrr & & ks & & mpc & kpc / arcsec + a 1835 & 0101 & 36 & 0.2528 & 1598 & 6.2 + a 665 & 0242 & 20 & 0.1818 & 1042 & 4.3 + a 1795 & 0100 & 40 & 0.0622 & 296 & 1.4 + hydra a & 0183 & 38 & 0.0538 & 253 & 1.2 + ser 159 - 03 & 0077 & 38 & 0.0580 & 274 & 1.3 + 2a0335 + 096 & 0215 & 26 & 0.0347 & 158 & 0.7 + a 4059 & 0176 & 54 & 0.0460 & 213 & 1.0 + a 496 & 0211 & 29 & 0.0328 & 149 & 0.7 + mkw 3s & 0129 & 39 & 0.0442 & 204 & 1.0 + a 2052 & 0128 & 33 & 0.0353 & 161 & 0.8 + a 262 & 0203 & 36 & 0.0163 & 71 & 0.3 + a 1837 & 0200 & 50 & 0.0372 & 170 & 0.8 + m87 & 0097 & 42 & 0.0043 & 19 & 0.1 + ngc 533 & 0195 & 48 & 0.0185 & 82 & 0.4 + lrrr [ tab : spatial ] & & arcsec & arcsec + a 1835 & 0.86 & 79 & 25 + a 665 & 0.67 & 53 & 55 + a 1795 & 0.74 & 147 & 55 + hydra a & 0.73 & 138 & 44 + ser 159 - 03 & 0.79 & 72 & 29 + 2a0335 + 096 & 0.79 & 156 & 63 + a 4059 & 0.78 & 72 & 156 + a 496 & 0.64 & 380 & 72 + mkw 3s & 0.72 & 111 & 25 + a 2052 & 0.67 & 258 & 51 + a 262 & 0.51 & 163 & 31 + a 1837 & 0.67 & 134 & 12 + m87 & 0.51 & 178 & 35 + ngc 533 & 0.58 & 119 & 18 + lrrrrrrrr & @xmath62 & @xmath62 & @xmath62 & & + & & & + a 1835 & [email protected] & 2.32 & @xmath26 & [email protected] & & & & 0.27 + a 665 & [email protected] & 4.24 & & & & & & 0.33 + a 1795 & [email protected] & 1.19 & @xmath22 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + hydra a & [email protected] & 4.94 & @xmath22 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + ser 159 - 03 & [email protected] & 1.79 & @xmath22 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + 2a0335 + 096 & [email protected] & 17.6 & @xmath225 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + a 4059 & [email protected] & 1.10 & @xmath24 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + a 496 & [email protected] & 4.58 & @xmath25 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + mkw 3s & [email protected] & 3.03 & @xmath22 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + a 2052 & [email protected] & 2.73 & @xmath23 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + a 262 & [email protected] & 5.37 & @xmath212 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + a 1837 & [email protected] & 4.30 & & & & & &
+ m87 & [email protected] & 2.54 & @xmath22 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + ngc 533 & 3.10 & 3.10 & @xmath218 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + lrrrrrrrrr [ tab : dem ] & @xmath63 & @xmath63 & @xmath63 & @xmath63 & @xmath64 & kev & @xmath63 + a 1835 & @xmath2200 & @xmath2300 & 800@xmath38200 & 5800@xmath38800 & 6@xmath65 & [email protected] & 1000 + a 665 & & & & & 2@xmath66 & [email protected] & + a 1795 & @xmath230 & @xmath230 & @xmath280 & 380@xmath3840 & 2@xmath65 & [email protected] & 300 + hydra a & @xmath2100 & 35@xmath3820 & 120@xmath3860 & 180@xmath3850 & 7@xmath67 & [email protected] & 180 + ser 159 - 03 & @xmath230 & @xmath230 & @xmath260 & 210@xmath3830 & 3@xmath67 & [email protected] & 79 + 2a0335 + 096 & @xmath284 & 20@xmath3810 & 40@xmath3820 & 420@xmath3850 & 4@xmath67 & [email protected] & 170 + a 4059 & @xmath210 & 10@xmath385 & 40@xmath3820 & 100@xmath3810 & 3@xmath67 & [email protected] & 60 + a 496 & @xmath210 & @xmath215 & 25@xmath3810 & 120@xmath3820 & 8@xmath67 & [email protected] & 72 + mkw 3s & @xmath264 & @xmath210 & @xmath220 & 45@xmath3810 & 3@xmath67 & [email protected] & 13 + a 2052 & @xmath210 & @xmath210 & 15@xmath385 & 100@xmath3820 & 4@xmath67 & [email protected] & 50 + a 262 & @xmath210 & @xmath22 & [email protected] & 10@xmath381 & 9@xmath68 & [email protected] & 2.0 + a 1837 & & & & & 8@xmath68 & [email protected] & + m87 & @xmath212 & @xmath20.6 & [email protected] & [email protected] & 4@xmath68 & [email protected] & 2.4 + ngc 533 & @xmath25 & @xmath21.6 & [email protected] & [email protected] & 6@xmath69 & [email protected] & 1.5 & + error bars , the empirical best fit model ( red ) , and the standard cooling - flow model ( green ) . both the model and the data are fluxed and background subtracted .
the cooling - flow model is not a best fit , but calculated by merely taking the soft x - ray flux in the empirical model and using the standard isobaric temperature distribution .
evident in all spectra is the severe overprediction of emission lines from the lowest temperatures.,width=672 ] by the estimate derived from the imaging data .
each color represents a different cluster and lines connect the points , which are not upper limits .
points above the line are consistent with ambient plasma which has not cooled , but the isobaric multiphase model would predict that emission should closely track the dashed line at @xmath33 or the line @xmath70 for isochoric cooling .
a large number of upper limits and points lie below those lines , however .
the spectra generally become inconsistent with the model near @xmath71 of the ambient temperature.,width=648 ] | we present high resolution x - ray spectra of 14 putative cooling - flow clusters of galaxies obtained with the reflection grating spectrometer on xmm - newton .
the clusters in the sample span a large range of temperatures and mass deposition rates .
various of these spectra exhibit line emission from o viii , ne x , mg xii & xi , al xiii & xii , si xiv & xiii , n vii , and c vi as well as all fe l ions .
the spectra exhibit strong emission from cool plasma at just below the ambient temperature , @xmath0 , down to @xmath1 , but also exhibit a severe deficit of emission , relative to the predictions of the isobaric cooling - flow model at lower temperatures ( @xmath2 @xmath3 ) .
in addition , the best - resolved spectra show emission throughout the entire x - ray temperature range , but increasingly less emission at lower temperatures than the cooling - flow model would predict .
these results are difficult to reconcile with simple prescriptions for distorting the emission measure distribution , e.g. by including additional heating or rapid cooling terms .
we enumerate some theoretical difficulties in understanding the soft x - ray spectra of cooling - flows independent of the classic problem of the failure to detect the cooling - flow sink . empirically , the differential luminosity distribution is consistent with being proportional to the temperature to the power of @xmath4 1 to 2 , instead of being independent of the temperature , as expected in the standard multi - phase model .
the primary differences in the observed low temperature spectra are ascribed to differences in the ambient temperature . |
early experimental studies of magnetotransport in two - dimensional ( 2d ) electron systems@xcite indicated that the @xmath0 factor of electrons in these systems may differ significantly from its bulk value .
it was established@xcite that the magnitude of the @xmath0 factor for electrons confined to @xmath3 si surfaces exceeds @xmath4 and increases from @xmath5 to @xmath6 with decreasing the concentration of electrons from @xmath7@xmath8 to @xmath9@xmath8 . shortly after the publication of experimental results@xcite it was suggested by janak@xcite that the enhancement of the @xmath0 factor can be accounted for by the electron - electron interactions .
the argument of janak represents a 2d version of the fermi liquid theory@xcite and goes as follows . in applied weak magnetic field ,
@xmath10 , the quasiparticle energies for the two spin projections can be written as @xmath11 where @xmath12 is the momentum , @xmath13 is the spectrum of a free electron , @xmath14 is the bare zeeman splitting , and @xmath15 is the self - energy @xmath16 where @xmath17 is the fourier component of the effective interaction between the electrons , and @xmath18 is the fermi function . solving the system eq .
( [ janak ] ) , eq .
( [ sigma ] ) in the zero - temperature limit , the effective @xmath0-factor can be presented as @xmath19 where @xmath20 in eq .
( [ effective ] ) @xmath21 is the effective mass @xmath22 and @xmath23 is the fermi momentum .
the enhancement factor @xmath24 is given by @xmath25 in the random phase approximation ( rpa ) one has@xcite @xmath26 for @xmath27 , where @xmath28 is the dielectric constant of the material , and @xmath29 is the interaction parameter of the 2d gas . with @xmath17 in the form
( [ diel ] ) , @xmath21 and @xmath24 can be evaluated analytically yielding@xcite @xmath30 @xmath31 where the function @xmath32 is defined as @xmath33 in the high - density limit ( @xmath34 ) the enhancement factor ( [ renorm ] ) takes the form ( see also ref . ) @xmath35 note that the theory@xcite neglects the frequency dependence of @xmath36 . as a result ,
( [ renorm ] ) predicts that interactions reduce the effective mass .
in fact , taking the frequency dependence into account@xcite leads to @xmath37 already within the rpa ( see , however , the recent numerical simulations@xcite ) .
later magnetotransport experiments@xcite on quantum well structures in narrow band semiconductors provided an evidence for a splitting of the conduction band in a zero magnetic field .
the analysis of the beating patterns in electron shubnikov - de haas oscillations led the authors@xcite to the conclusion that such a splitting can be accounted for by adding the spin - orbit ( so ) term @xmath38 to the hamiltonian of a free electron . here
@xmath39 is the so coupling constant , @xmath12 is the wave vector , @xmath40 is the unit vector normal to the plane of the quantum well , @xmath41 are the pauli matrices
. the term eq .
( [ hso ] ) was first introduced by bychkov and rashba@xcite to explain the experimental results on electron spin resonance@xcite and a cyclotron resonance of holes@xcite in gaas / algaas heterostructures . in order to obtain more detailed information about the so - induced splitting of the conduction band , the evolution of the shubnikov - de haas oscillations with a tilting of magnetic field was traced@xcite .
subsequently , the energy spectrum of a 2d electron in a tilted magnetic field in the presence of the so coupling was studied theoretically@xcite . recently a zero - field splitting in different 2d systems was inferred experimentally either from the shubnikov - de haas@xcite or from the commensurability oscillations@xcite ( in a spatially modulated sample ) patterns . in the domain of weak magnetic fields
the shubnikov - de haas oscillations are smeared out .
however , it was demonstrated both experimentally@xcite and theoretically@xcite that the so coupling still manifests itself in this domain through the weak localization corrections to the conductance .
early works@xcite in this direction were succeeded by detailed studies@xcite . in the present paper
we investigate theoretically the interplay between the so coupling and the electron - electron interactions .
namely , we address the question whether the interactions cause the renormalization of the coupling constant @xmath39 in eq .
( [ hso ] ) as it is the case for the @xmath0-factor .
we assume that the bare so coupling is weak enough @xmath42 . with the so term
( [ hso ] ) the hamiltonian of non - interacting electrons can be presented in the form @xmath43 where the projection operators @xmath44 and @xmath45 are defined as @xmath46 so that @xmath47 . in eq .
( [ p+- ] ) @xmath48 is the azimuthal angle of the wave vector @xmath12 .
the energy spectrum consists of two branches @xmath49 following the fermi liquid theory , the self - energy in the presence of the so coupling becomes an operator @xmath50.\ ] ] our main observation is that in the presence of the interaction , the operator @xmath51 still retains the structure of eq .
( [ h0 ] ) @xmath52 where @xmath53 are the _ scalar _ functions of the absolute value of the wave vector @xmath12 @xmath54\\ \nonumber \pm\frac{1}{2}\int\frac{d^2{\bf k}^{'}}{(2\pi)^2}\cos(\phi_{\bf k}-\phi_{{\bf k}^{'}})v_{eff}(|{\bf k}-{\bf k}^{'}|)\biggl[f_0(e_f - e_{+}(k^{'}))-f_0(e_f - e_{-}(k^{'}))\biggr].\end{aligned}\ ] ] for renormalized energy spectrum , we have @xmath55 by solving the system eq .
( [ sigmapmk ] ) and eq .
( [ ek ] ) , we get the following result for the renormalized so splitting @xmath56 where @xmath57 is the bare so splitting and the renormalization factor is determined as @xmath58 if @xmath59 does not depend on @xmath60 ( when interactions are short - ranged due _
e.g. _ to the presence of a gate electrode close to the 2d plane ) , then we have @xmath61 .
however , in general , the integral is positive .
thus , we conclude that the exchange interaction leads to the _ enhancement _ of the so coupling . within the random phase approximation
when @xmath36 has the form eq .
( [ diel ] ) , the integral ( [ lambdaso ] ) can be calculated analytically and expressed through the function @xmath32 as follows @xmath62 comparison of the last expression with eq .
( [ renorm ] ) indicates that @xmath63 .
in fact , this relation holds not only for @xmath17 in the form ( [ diel ] ) , but for arbitrary effective interaction and represents a 2d version of the corresponing relation in the fermi liquid theory@xcite . to verify this relation ,
it is convenient to perform transformation to the real space where the interaction has the form @xmath64 . then from eqs .
( [ mass ] ) and ( [ lambdaso ] ) it is easy to check that @xmath65 where @xmath66 is the bessel function . combining ( [ deltaso ] ) with ( [ real ] ) , we get @xmath67 in fig .
1 we plot both @xmath24 and @xmath68 as a function of interaction parameter @xmath1 .
it is seen that @xmath24 is much bigger than @xmath68 , which has a maximum at @xmath69 and does not exceed @xmath70 percent . on the other hand
, it is known that at large @xmath1 the random phase approximation overestimates the screening effect which , in turn , suppresses @xmath68 . to extend the fermi liquid description to higher @xmath1 , it is customary @xcite to modify the random phase dielectric function @xcite as follows @xmath71 where @xmath72=@xmath73 is the fourier component of the coulomb interaction and @xmath74=@xmath75 is the lindhard susceptibility of the free electron gas .
the factor @xmath76 ( local field correction ) describes the reduction of the screening at large @xmath77 ( small distances ) .
for @xmath78 we recover eq .
( [ diel ] ) for the effective interaction @xmath79 . in later works@xcite a different approximation for @xmath17 was put forward @xmath80 ^ 2\chi(q),\ ] ] where @xmath81 is defined as @xmath82 for the local field correction @xmath76 the authors@xcite adopted the following form @xmath83 where @xmath84 , and @xmath85 , @xmath86 are the numerical factors .
( [ effpoten ] ) is written neglecting the spin - fluctuation - induced vertex corrections .
combining eqs .
( [ effpoten ] ) , ( [ localfield1 ] ) and ( [ lambdaso ] ) , we get the following expression for the enhancement factor of so coupling @xmath87-g_{\infty}^2r_s\sin\phi/2}{(r_s+\sqrt{2}\sin\phi/2)(a^2+\sin^2\phi/2)-g_{\infty}r_s\sin\phi/2\sqrt{a^2+\sin^2\phi/2}}.\ ] ] in fig .
2 we present the dependence @xmath88 calculated numerically within the region up to @xmath89 .
following@xcite , we took @xmath85 from numerical calculations ( at discrete values of @xmath1 ) of the pair correlation function@xcite and following ref
. assumed @xmath90 within the entire domain .
it is seen that instead of falling down ( as in fig .
1 ) @xmath88 increases with @xmath1 , when the local factor is included . note , however , that approximately constant value for @xmath86 was established only within a limited interval @xmath91 in refs .. we used the same value for calculation at higher @xmath1 in order to illustrate that the enhancement factor can take appreciable values in this domain .
the alternative approach to the effective interaction in 2d electron gas with @xmath92 is described in ref . .
in this paper the local field factor in the conventional form ( [ dielectric ] ) of @xmath93 was fitted in such a way that the static characteristics of the system , calculated with this @xmath93 , are consistent with the monte carlo results of tanatar and ceperley@xcite . according to ref .
the local field correction has the form @xmath94,\ ] ] where the parameters @xmath95 , @xmath96 are listed in ref . at discrete values of @xmath1 up to @xmath97 .
the numerical results for @xmath68 calculated for these values by substituting @xmath76 in the form ( [ iwamoto ] ) into eq .
( [ dielectric ] ) are shown in fig .
they indicate that for @xmath2 the enhancement is quite pronounced .
however , the reliability of both approaches can still be questioned .
now let us address the situation when the zeeman splitting and so coupling are present simultaneously .
we will study the interplay between the interaction - induced enhancement of the @xmath0 factor and of the so coupling .
first assume that zeeman splitting is caused by a perpendicular magnetic field .
the bare hamiltonian in this case takes the form @xmath98 where the modified projection operators @xmath99 are introduced . in eq .
( [ modprojection ] ) @xmath100 is defined as @xmath101 the bare energy spectrum is given by @xmath102 so that the splitting of the spectrum at @xmath103 equals @xmath104 the general expression eq .
( [ sigmak ] ) for the self - energy retains its form in the present case after changing @xmath105 by @xmath106 , where the renormalized projection operators have the form of eq .
( [ modprojection ] ) @xmath107 with renormalized parameter @xmath108 , which should be determined self - consistently together with renormalized spectrum @xmath109 .
since in the present case the operators @xmath106 differ from @xmath110 , the consequence ( [ form1 ] ) of eq .
( [ sigmak ] ) is not valid anymore .
instead we get the following system of equations @xmath111,\end{aligned}\ ] ] @xmath112,\end{aligned}\ ] ] @xmath113=\frac{\gamma_0(k)}{1+\gamma_0 ^ 2(k)}\biggl[e_+^{(0)}(k)-e_-^{(0)}(k)\biggr]+\\ \nonumber + \int\frac{d^2{\bf k}^{'}}{(2\pi)^2}\cos(\phi_{\bf k}-\phi_{{\bf k}^{'}})v_{eff}(|{\bf k}-{\bf k}^{'}|)\biggl[f_0(e_f - e_{+}(k^{'}))-f_0(e_f - e_{-}(k^{'}))\biggr].\end{aligned}\ ] ] subtracting eq .
( [ eq2 ] ) from eq .
( [ eq1 ] ) , we get @xmath114=\frac{1-\gamma_0 ^ 2(k)}{1+\gamma_0 ^
2(k)}\biggl[e_+^{(0)}(k)-e_-^{(0)}(k)\biggr]+\\ \nonumber + \int\frac{d^2{\bf k}^{'}}{(2\pi)^2}v_{eff}(|{\bf k}-{\bf k}^{'}|)\frac{1-\gamma^2(k^{'})}{1+\gamma^2(k^{'})}\biggl[f_0(e_f - e_{+}(k^{'}))-f_0(e_f - e_{-}(k^{'}))\biggr].\end{aligned}\ ] ] now we can apply to eq .
( [ eq3 ] ) and eq .
( [ eq4 ] ) the same argument that led to renormalization of @xmath115 and @xmath116 respectively . in the zero - temperature limit
this results in the following system of equations @xmath117\biggl[1-\frac{m^*}{m}\lambda_{so}\biggr]=\frac{\gamma_0(k)}{1+\gamma_0 ^ 2(k)}\biggl[e_+^{(0)}(k)-e_-^{(0)}(k)\biggr],\ ] ] @xmath118\biggl[1-\frac{m^*}{m}\lambda_{z}\biggr]=\frac{1-\gamma_0 ^ 2(k)}{1+\gamma_0 ^ 2(k)}\biggl[e_+^{(0)}(k)-e_-^{(0)}(k)\biggr].\ ] ] dividing eq .
( [ quadra2 ] ) by eq .
( [ quadra1 ] ) we get a closed quadratic equation for @xmath108 @xmath119 substituting the solution of this equation back into eq .
( [ eq1 ] ) , we get the renormalized splitting of the spectrum @xmath120 @xmath121 using the definition ( [ gammak ] ) of @xmath122 , we can rewrite the last result in the following concise form @xmath123 finally , with the use of eqs .
( [ effective ] ) and ( [ deltaso ] ) , we arrive to the conclusion that renormalized splitting @xmath124 is related to renormalized values @xmath125 and @xmath126 in the same way as the bare values ( eq .
( 32 ) ) @xmath127 consider now the case when the zeeman splitting is caused by a parallel magnetic field applied along the @xmath128-direction .
then the hamiltonian can be written as @xmath129 where the energy spectrum @xmath130 depends both on the amplitude and orientation of @xmath12 with respect to the magnetic field . in eq .
( [ parallelfield ] ) , the projection operators @xmath131 are defined as @xmath132 with the angle @xmath133 related to the azimuthal angle of vector @xmath12 as follows @xmath134 the bare splitting of the energy spectrum at @xmath135 is equal to @xmath136 performing calculations similar to those for perpendicular field , it is easy to check that in the present case the relation between the renormalized splitting @xmath124 and @xmath137 , @xmath115 preserves the form ( [ deltaparallel ] ) @xmath138
the above calcualations were based on the concept of effective interaction between electrons , @xmath17 . generally speaking
, fermi liquid theory relates the observable values to the bare parameters of electron gas by means of interaction function @xcite having the form @xmath139,\ ] ] where @xmath140 and @xmath141 are spin matrices , @xmath142 and @xmath143 are the symmetric and antisymmetric parts of the interaction function , respectively . in eq .
( [ intfun ] ) @xmath144 are dimensionless quantities .
the concept of effective interaction used above is equivalent to the assumption @xmath145 . the way to extend our theory in order to take into account
the difference between @xmath146 and @xmath147 is to modify the self - energy eq .
( [ sigmak ] ) as follows @xmath148 \\ \nonumber + \int\frac{d^2{\bf k}^{'}}{(2\pi)^2}\hat{p}^{-}({\bf k}^{'})\biggl[v_{eff}(|{\bf k}-{\bf k}^{'}|)f_0(e_f - e_{-}(k^ { ' } ) ) \\
\nonumber + w_{eff}(|{\bf k}-{\bf k}^{'}|)f_0(e_f - e_{+}(k^{'}))\biggr].\end{aligned}\ ] ] here @xmath17 corresponds to the effective interaction in eq .
( [ sigmak ] ) , whereas @xmath149 accounts for the difference between @xmath146 and @xmath150 .
it is straightforward to check that with self - energy operator eq .
( [ modsigma ] ) , the projection operators @xmath44 and @xmath45 which , in principle , should be determined self - consistently , still retain the form eq .
( [ p+- ] ) .
thus we can repeat the derivation for the enhancement of @xmath115 in a similar way as in sec .
ii . the difference is , however , that the relation eq .
( [ real ] ) does not hold anymore .
indeed , the renormalization of the effective mass@xcite is now determined by the effective interaction @xmath151 through @xmath152 , while @xmath68 is determined by the first fourier harmonics of @xmath153 ; accordingly , @xmath154 is determined by the zero s fourier component of @xmath153 .
consequently , in terms of the dimensionless landau parameters @xcite , we get the following generalization of eq .
( [ simple ] ) @xmath155 note that , while the dependence of @xmath156 on @xmath1 in two dimensions was a subject of monte carlo studies @xcite , the dependence @xmath157 in 2d can not be extracted form the current literature .
the main goal of this paper is to demonstrate that alongside with fundamental characteristics , @xmath158 , of interacting electron gas , which describes the enhacement of the @xmath0-factor , and was studied in many works , there exists another fundamental characteristics @xmath88 which describes the interaction - induced enhancement of the so coupling .
we calculated this function analytically in the limit of high concentrations and estimated numerically at low concentrations .
note that throughout the paper we assumed the bare so coupling to be small : @xmath159 .
however , in the limit of high concentrations ( @xmath34 ) the corresponding condition is more strict : @xmath160 , which is equivalent to @xmath161 . in the intermediate region @xmath162 the enhancement factor is given by @xmath163 since experimentally the concentration of carriers is varied by changing the gate voltage@xcite , there exists another simple reason for the dependence of the so coupling on the concentration .
indeed , the change of the gate voltage causes the redistribution of the confining potential , which , in turn @xcite , affects the parameter @xmath39 .
this mechanism should be dominant at high concentrations when @xmath68 is small .
note in conclusion , that if the bare so splitting is caused by the dresselhaus mechanism@xcite , which originates from the absence of the inversion symmetry in the bulk , the renormalization of the corresponding splitting of the spectrum , @xmath164 , has the same form as eq .
( [ simple ] ) : @xmath165 . as a result ,
when both @xmath164 and @xmath115 are present , the splitting of the energy spectrum is given by the same formula as for noninteracting electrons@xcite @xmath166 finally , let us point out that in conventional magnetotransport oscillations experiments performed up to now@xcite the typical concentrations of electrons was quite high @xmath167@xmath8 .
as a result , the typical values of the interaction parameter @xmath1 were rather low ( @xmath168 ) . only in 2d hole
gas@xcite the condition ( @xmath169 ) was fulfilled . for low values of @xmath1
our theory predicts that the renormalization of the so coupling is weak .
however , in recent experiments on the electron gases in silicon metal - oxide - semiconductor field - effect transistors@xcite and alas quantum wells@xcite , as well as in hole gases in sige quantum wells@xcite , gaas inverted semiconductor - insulator - semiconductor structures@xcite , and gaas - algaas heterostructures@xcite the values of @xmath1 ranged from@xcite @xmath170 to@xcite @xmath171 .
for such large @xmath1 we predict a strong renormalization of the so coupling , which might be of relevance for metal - insulator transition observed in these systems .
for example , the strong renormalization of so coupling at large @xmath1 might cause an instability of electronic spectrum in a clean system , so that the system in zero magnetic field would undergo a transition into an exotic `` chiral phase '' at some critical density . to be more specific , in the latest publication@xcite the critical density for metal - insulator transition in n - type gaas was reported to be @xmath172 @xmath8 . to accomodate all these electrons within the lower branch of the spectrum , corresponding to chirality `` @xmath173 '' ( see eq .
( 15 ) ) , the effective coupling constant @xmath39 should exceed @xmath174ev@xmath175 m .
on the other hand , the constant @xmath39 for relatively high - density gaas / algaas structure with @xmath176@xmath8 can be extracted from ref . to be @xmath177ev@xmath175 m .
thus , a two times interaction - induced enhancement of so coupling ( although fig.2 for @xmath178 shows a lesser value ) would make a chiral phase feasible
. we are grateful to r. r. du , d. mattis , and y. s. wu for useful discussions .
we also thank l. e. zhukov for assistance in preparation of the manuscript .
m. e. r. ackowledges the discussions with h. u. baranger and y. imry and the hospitality of the aspen center for physics .
a. bychkov and e. i. rashba , pisma zh .
fiz . * 39 * , 64 ( 1984 ) [ jetp lett . *
39 * , 78 ( 1984 ) ] .
a. bychkov and e. i. rashba , j. phys .
c * 17 * , 6039 ( 1984 ) .
d. stein , k. v. klitzing , and g. weinmann , phys .
lett . * 51 * , 130 ( 1983 ) .
h. l. stormer , z. schlesinger , a. chang , d. c. tsui , a. c. gossard , and w. wiegmann , phys .
lett . * 51 * , 126 ( 1983 ) .
see the most recent publication y. lyanda - geller , phys .
lett . * 80 * , 4273 ( 1998 ) , and references therein .
the effect of the so term on magnetotransport in the insulating regime was studied theoretically in t. v. shahbazyan and m. e. raikh , phys .
lett . * 73 * , 1408 ( 1994 ) .
s. yarlagadda and g. f. giuliani , phys .
b * 38 * , 10966 ( 1988 ) .
g. e. santoro and g. f. giuliani , solid .
state commun .
* 67 * , 681 ( 1988 ) .
s. yarlagadda and g. f. giuliani , solid .
state commun .
* 69 * , 677 ( 1989 ) .
m. jonson , j. phys .
c * 9 * , 3055 ( 1976 ) .
m. y. simons , a. r. hamilton , m. pepper , e. h. linfield , p. d. rose , d. a. ritchie , a. k. savchenko , and t. g. griffiths , phys .
lett . * 80 * , 1292 ( 1998 ) .
y. hanien , d. shahar , j. yoon , c. c. li , d. c. tsui , and h. shtrikman , phys .
b. * 58 * , r13338 ( 1998 ) . | = 0.90 cm we study theoretically the renormalization of the spin - orbit coupling constant of two - dimensional electrons by electron - electron interactions .
we demonstrate that , similarly to the @xmath0 factor , the renormalization corresponds to the enhancement , although the magnitude of the enhancement is weaker than that for the @xmath0 factor . for high electron concentrations ( small interaction parameter @xmath1 )
the enhancement factor is evaluated analytically within the static random phase approximation . for large @xmath2
we use an approximate expression for effective electron - electron interaction , which takes into account the local field factor , and calculate the enhancement numerically .
we also study the interplay between the interaction - enhanced zeeman splitting and interaction - enhanced spin - orbit coupling . |
recent analyses of extracellular recordings performed in two motor areas of behaving monkeys have tried to clarify how information about movements is trasmitted and received from higher to lower stages of processing , and to identify distinct roles of the two areas in the planning and execution of movements @xcite .
although this study failed to produce clearcut results , it remains interesting to try and understand , from a more theoretical point of view , how information about multi - dimensional correlates of neural activity may be transmitted from the input to the output of a simple network .
in fact , a theoretical study is still lacking , which explores how the coding of stimuli with continuous as well as discrete dimensions is transferred across a network . in a previous report @xcite the mutual information between the activity ( ` firing rates ' ) of a finite population of @xmath0 units ( ` neurons ' ) and a set of correlates , which have both a discrete and a continuous angular dimension , has been evaluated analytically in the limit of large noise .
this parametrization of the correlates can be applied to movements performed in a given direction and classified according to different `` types '' ; yet it is equally applicable to other correlates , like visual stimuli characterized by an orientation and a discrete feature ( colour , shape , etc .. ) , or in general to any correlate which can be identified by an angle and a `` type '' . in this study , we extend the analysis performed for one population , to consider two interconnected areas , and we evaluate the mutual information between the firing rates of a finite population of @xmath1 output neurons and a set of continuous+discrete stimuli , given that the rate distribution in input is known . in input , a threshold nonlinearity has been shown to lower the information about the stimuli in a simple manner , which can be expressed as a renormalization of the noise @xcite .
how does the information in the output depend on the same nonlinearity ?
how does it depend on the noise in the output units ? is the power to discriminate among discrete stimuli more robust to transmission down one set of random synapses , than the information about a continuously varying parameter ?
we address these issues by calculating the mutual information , using the replica trick and under the assumption of replica symmetry ( see for example @xcite ) .
saddle point equations are solved numerically .
we analyze how the information trasmission depends on the parameters of the model , i.e. the level of output and input noise , on the ratio between the two population sizes , as well as on the tuning curve with respect to the continuous correlate , and on number of discrete correlates .
the input - output transfer function is a crucial element in the model .
the binary and the sigmoidal functions used in many earlier theoretical and simulation studies @xcite fail to describe accurately current - to - frequency transduction in real neurons .
such trasduction is well captured instead , away from saturation , by a threshold - linear function @xcite .
such a function combines the threshold of real neurons , the linear behaviour typical of pyramidal neurons above threshold , and the accessibility to a full analytical treatment @xcite , as demonstrated here , too . for the sake of analytical feasibility ,
however , we take the input units to be purely linear . therefore it should be kept in mind , in considering the final results that the threshold nonlinearity is only applied to the output units .
in analogy to the model studied in @xcite we consider a set of @xmath0 input units which fire to an external continuous+discrete stimulus , parametrized by an angle @xmath3 and a discrete variable @xmath4 , with a gaussian distribution : @xmath5 ; \label{dist}\ ] ] @xmath6 is the firing rate in one trial of the @xmath7 input neuron , while the mean of the distribution , @xmath8 is written : @xmath9 @xmath10 where @xmath11 is a quenched random variable distributed between @xmath12 and @xmath13 , @xmath14 is the preferred direction for neuron @xmath15 . according to eq.([tuning_tot ] )
neurons fire at an average firing rate which modulates with @xmath3 with amplitude @xmath16 , or takes a fixed value @xmath17 , independently of @xmath3 , with amplitude @xmath18 .
we assume that quenched variables are uncorrelated and identically distributed across units and across the @xmath19 discrete correlates : @xmath20^{nk } \label{ro_eps}\ ] ] @xmath21^n=\frac{1}{(2\pi)^n}.\ ] ] in @xcite it has been shown that a cosinusoidal shaped function as in eq.([tuning_tot ] ) is able to capture the main features of directional tuning of real neurons in motor cortex .
moreover it has been shown that the presence of negative firing rates in the distribution ( [ dist ] ) , which is not biologically plausible , does not alter information values , with respect to a more realistic choice for the firing distribution , in that it leads to the same curves except for a renormalization of the noise .
output neurons are activated by input neurons via uncorrelated gaussian connection weights @xmath22 .
each output neuron performs a linear summation of the inputs ; the outcome is distorted by a gaussian distributed noise @xmath23 and then thresholded , as in the following : @xmath24^+;\,\,\,\,i=1 .. m , j=1 ..
n \label{output}\ ] ] in eq.([output ] ) @xmath25 is a threshold term , @xmath26 is a ( 0,1 ) binary variable , with mean @xmath27 , which expresses the sparsity or dilution of the connectivity matrix , and @xmath28 @xmath29 @xmath30 @xmath31^+=x\theta(x).\ ] ]
we aim at estimating the mutual information between the output patterns of activity and the continuous+discrete stimuli : @xmath32 @xmath33 where the distribution @xmath34 is determined by the threshold linear relationship ( [ output ] ) , @xmath35 is given in eq.([dist ] ) and @xmath36 is a short notation for the average across the quenched variables @xmath37,@xmath38,@xmath39,@xmath40 and on the noise @xmath41 .
we assume that the stimuli are equally likely : @xmath42 .
eq.([info ] ) can be written as : @xmath43 with : @xmath44 @xmath45 \right\rangle_{\varepsilon,\vartheta^0,c , j,\delta}. \label{outent}\end{aligned}\ ] ] the analytical evaluation of the _ equivocation _ @xmath46 can be performed inserting eq.([xi_eta ] ) in the expression ( [ equiv ] ) , and using the replica trick to get rid of the logarithm : @xmath47 to take into account the threshold - linear relation ( [ output ] ) we consider the following equalities : @xmath48 inserting eq.([anna ] ) in eq.([replica ] ) one obtains : @xmath49 -1\right ) .
\label{replica2}\end{aligned}\ ] ] the average across the quenched disorder @xmath27,@xmath50,@xmath51 in eq.([replica2 ] ) can be performed in a very similar way as shown in @xcite : using the integral representation for each @xmath51 function , gaussian integration across @xmath50,@xmath51 is standard ; the average on @xmath27 can be performed assuming large the number @xmath0 of input neurons .
the final outcome for the _ equivocation _ reads : @xmath52^m -1\right ) , \label{replica3 } \end{aligned}\ ] ] where we have put @xmath53 .
integration on @xmath54 is straightforward .
integration on @xmath55 can be performed introducing @xmath56 auxiliary variables @xmath57 via @xmath51 functions expressed in their integral representation . considering the expression ( [ dist ] ) for the input distribution and with some rearrangement of the terms
the final result can be expressed as : @xmath58^m-1\right ) , \nonumber\end{aligned}\ ] ] where : @xmath59 the evaluation of the entropy of the responses @xmath60 , eq.([outent ] ) , can be carried out in a very similar way , introducing replicas in the continuous+discrete stimulus space .
the final result reads : @xmath61^{n+1 } \left\langle e^{-\sum_{\alpha,\beta}(\delta_{\alpha\beta}-\sigma^{-1}_{\alpha\beta } ) \tilde{\eta}(\vartheta_\alpha , s_\alpha)\tilde{\eta}(\vartheta_\beta , s_\beta)/2\sigma^2 } \right\rangle_{\varepsilon,\vartheta^0}^n\right.\label{outent2}\\ & & \left.e^{-\frac{m}{2}tr \ln g } \left[\int_{-\infty}^0 \prod_\alpha \frac{d\xi^\alpha}{\sqrt{2\pi } } e^{-\sum_{\alpha,\beta}(\xi^\alpha-\xi_0 ) ( g^{-1}_{\alpha\beta}/2)(\xi^\beta-\xi_0)}+\int^{\infty}_0 \frac{d\xi}{(2\pi)^{\frac{n+1}{2 } } } e^{-\sum_{\alpha,\beta } ( g^{-1}_{\alpha\beta}/2)(\xi^-\xi_0)^2}\right]^m-1\right ) .
\nonumber\end{aligned}\ ] ]
the integrals in eq.([equiv2]),([outent2 ] ) can not be solved without resorting to an approximation . in analogy to what is used in @xcite , we use a saddle - point approximation ( which in general would be valid in the limit @xmath62 ) and we assume replica symmetry @xcite in the parameters @xmath63 , @xmath64 .
this allows to explicitely invert and diagonalize the matrices @xmath65,@xmath66 : @xmath67 the assumption of replica symmetry seems to have more subtle implications in the present situation .
these will be discussed below . in replica symmetry
the mutual information can be expressed as follows : @xmath68}\right.\nonumber\\ & & \left .- e^{n\left[(n+1)z_0^b\tilde{z}_0^b - n(n+1)z_1^b\tilde{z}_1^b- \frac{r}{2}\left(tr\ln g(z_0^b , z_1^b)+f(z_0^b , z_1^b)\right)-\frac{1}{2}tr\ln\sigma(\tilde{z}_0^b,\tilde{z}_1^b)- h^b(\tilde{z}_0^b,\tilde{z}_1^b)\right]}\right\ } , \label{info_lim}\end{aligned}\ ] ] with @xmath69 ; \label{f}\ ] ] @xmath70 ; \label{ha}\ ] ] @xmath71^{n+1 } \left\langle e^{-\sum_{\alpha,\beta}\left(\delta_{\alpha\beta}-\sigma^{-1}_{\alpha\beta}\right ) \tilde{\eta}(\vartheta_\alpha , s_\alpha)\tilde{\eta}(\vartheta_\beta , s_\beta)/2\sigma^2 } \right\rangle_{\varepsilon,\vartheta^0}^n\right ] .
\label{hb}\ ] ] we have set @xmath72 and @xmath73,@xmath74,@xmath75,@xmath76 are the solutions of the saddle point equations : @xmath77;\nonumber\\ z_1^{a , b}&=&-\frac{1}{n}\frac{\partial}{\partial\tilde{z}_1}\left[\frac{1}{2 } tr\ln\sigma(\tilde{z}_0,\tilde{z}_1)+h^{a , b}(\tilde{z}_0,\tilde{z}_1)\right];\nonumber\\ \tilde{z}_0^{a , b}&=&\frac{\partial}{\partial z_0}\frac{r}{2}\left [ tr\ln g(z_0,z_1)+f(z_0,z_1)\right];\nonumber\\ \tilde{z}_1^{a , b}&=&-\frac{1}{n}\frac{\partial}{\partial z_1}\frac{r}{2}\left [ tr\ln g(z_0,z_1)+f(z_0,z_1)\right].\end{aligned}\ ] ] all the equations must be evaluated in the limit @xmath78 .
it is easy to check that all terms in the exponent in eq.([info_lim ] ) are order @xmath79 .
in fact , since when @xmath78 only one replica remains , one has : @xmath80_{|_{n=0}}.\end{aligned}\ ] ] therefore , from the saddle point equations , @xmath74 are order @xmath79 and @xmath81 is also order @xmath79 : @xmath82 since @xmath83 , it is easy to check by explicit evaluation that , when @xmath78 , all the @xmath84 diagonal terms among the matrix elements @xmath85 are order @xmath79 and all the @xmath86 out - of - diagonal terms are order @xmath13
. then all terms in the exponent of eqs.([ha]),([hb ] ) are order @xmath79 , and we can expand the exponentials , which allows us to perform the quenched averages across @xmath87 . considering the expression of @xmath88 , eq.([tuning_tot ] ) , one obtains : @xmath89\right ) ; \label{ha_hb}\end{aligned}\ ] ] @xmath90 ^ 2\rangle_{\varepsilon,\vartheta^0}\nonumber\\ & = & ( \eta^0)^2\left[(a_2+\alpha^2 - 2\alpha a_1)\langle\varepsilon^2\rangle_{\varepsilon}+\alpha^2 + 2\alpha(a_1-\alpha)\langle\varepsilon\rangle_{\varepsilon}\right ] ; \label{lambda1}\end{aligned}\ ] ] @xmath91 ^ 2 \langle\tilde{\eta}(\vartheta_1,s_1)\tilde{\eta}(\vartheta_2,s_2 ) \rangle_{\varepsilon,\vartheta^0}\nonumber\\ & = & ( \eta^0)^2\left[(a_1-\alpha)^2\left(\frac{k-1}{k } \langle\varepsilon\rangle_{\varepsilon}^2+\frac{1}{k}\langle\varepsilon^2 \rangle_{\varepsilon}\right)+\alpha^2 + 2\alpha(a_1-\alpha ) \langle\varepsilon\rangle_{\varepsilon}\right ] ; \label{lambda2}\end{aligned}\ ] ] @xmath92 a similar expansion in @xmath79 for @xmath93 and for @xmath94 allows to derive explicitely the saddle point equations : @xmath95 \lambda_\eta^1+\frac{1}{\left(1 + 2\sigma^2\tilde{z}_1^b\right)^2}\lambda_\eta^2;\nonumber\\ \tilde{z}_1^{a , b}&=&-c\sigma^2_j\frac{r}{2}\left\{\sigma\left(\frac{\xi^0}{\sqrt{p+q}}\right)\frac{\xi^0}{\left(p+q\right)^{\frac{3}{2}}}-\frac{1}{p}{{\rm erf}}\left(\frac{\xi^0}{\sqrt{p+q}}\right)\right.\nonumber\\ & & \left.+\int_{-\infty}^{\infty } dt \left[1+\ln\left({{\rm erf}}\left(-\frac{\xi^0-t\sqrt{q}}{\sqrt{p}}\right)\right)\right]\sigma\left(\frac{\xi^0-t\sqrt{q}}{\sqrt{p}}\right)\frac{1}{p^\frac{3}{2}}\left[\xi^0-t\frac{q+p}{\sqrt{q}}\right]\right \};\nonumber\\ \label{saddlepoint}\end{aligned}\ ] ] where : @xmath96 @xmath97 from the expression of @xmath73 in eq.([saddlepoint ] ) , it is easy to verify that the dependence on @xmath98 in eq.([ha_hb ] ) , which might affect the information in eq.([info_lim ] ) , cancels out with the products @xmath99,@xmath100 which should contribute to the information in the limit @xmath78 ( see eq.([info_lim ] ) ) . therefore , since @xmath73 is known and @xmath75 depends only on @xmath76 , the mutual information can be expressed as a function of @xmath75,@xmath76 , which in turn are to be determined self - consistently by the saddle point equations .
the average information per input cell can be written , finally : @xmath101 + \gamma_2^b(\tilde{z}_1^b)-\gamma_2^a(\tilde{z}_1^a)\right\ } , \label{final_info}\ ] ] with @xmath102;\end{aligned}\ ] ] @xmath103 @xmath104.\ ] ] the expression for the mutual information only contains terms linear in either @xmath0 or @xmath1 .
since the last of the saddle - point equations , ( [ saddlepoint ] ) , contains @xmath105 , if one fixes @xmath0 and increases @xmath1 the information grows non - linearly , because the position of the saddle point varies .
it turns out that , as shown below , the growth is only very weakly sublinear , at least when @xmath106 .
analogously , fixing @xmath1 and varying @xmath0 we would find a non - linearity due to the @xmath105-dependence of the saddle point .
if @xmath105 is fixed and @xmath0 and @xmath1 grow together , the information rises purely linearly . what our analytical treatment misses out , however , is the nonlinearity required to appear as the mutual information approaches its ceiling , the entropy of the stimulus set .
the approach to this saturating value was described at the input stage @xcite , where also the initial linear rise ( in @xmath0 ) was obtained in the large noise limit @xcite .
therefore , our saddle point method is in same sense similar to taking a large ( input ) noise limit , @xmath107 , to its leading ( order @xmath108 ) term .
it is possible that the saddle point method could be extended , to account also for successive terms in a large noise expansion .
this would probably require integrating out the fluctuations around the saddle point , but by carefully analysing the relation of different replicas to different values of the quenched variables .
we leave this possible extension to future work . the present calculation , therefore ,
although employing a saddle point method which is usually applicable for large @xmath0 and @xmath1 , should be considered effectively as yielding the initial linear rise in the mutual information , the one observed with @xmath1 small .
eq.([saddlepoint ] ) for @xmath76 has been solved numerically using a matlab code .
convergence to self - consistency has been found already after @xmath109 iterations with an error lower than @xmath110 .
fig.[fig1 ] shows the mutual information as a function of the output population size , for an input population size equal to @xmath111 cells .
this is contrasted with the information in the input units , about exactly the same set of correlates , calculated as in @xcite , by keeping only the leading ( linear ) term in @xmath0 .
in fact , in @xcite the mutual information carried by a finite population of neurons firing according to eq.([dist ] ) had been evaluated analytically , in the limit of large noise , by means of an expansion in @xmath112 . to linear order in @xmath0
the analytical expression for the information carried by @xmath0 input neurons reads : @xmath113 where @xmath114 , @xmath115 are defined , again , as in eqs.([lambda1 ] ) , ( [ lambda2 ] ) . in analogy to what had been done in @xcite
we have set @xmath116 . as evident from the graph ,
also the output information is essentially linear up to a value of @xmath117 , and quasi - linear even for @xmath118 .
it should be remined , again , that our saddle point method only takes into account the term linear in @xmath0 in the information _ input _ units carry about the stimulus .
it is not possible , therefore , for eq.([final_info ] ) to reproduce the saturation in the mutual information as it approaches the entropy of the stimulus set ( which is finite , if one considers only discrete stimuli ) .
the nearly linear behaviour in @xmath1 thus reflects the linear behaviour in @xmath0 induced , in the intermediate quantity ( the information available at the input stage ) , by our saddle point approximate evaluation . as it is clear from the comparison in fig.[fig1 ] ,
when the two populations of units are affected by the same noise the input information is considerably higher than the output one .
this is expected , since output and input noise sum up while influencing the firing of output neurons , but also because the input distribution is taken to be a pure gaussian , while the output rates are affected by a threshold .
if the input - output tranformation were linear and the output noise much smaller than the input one , one would expect that output and input units would carry the same amount of information .
briefly , in a linear network with zero output noise one has : @xmath119 considering eqs.([xi_eta]),([dist ] ) , an _ effective _ expression for the distribution @xmath120 can be obtained by direct integration of the @xmath51 functions @xmath121 via their integral representation , on @xmath122 : @xmath123 @xmath124 @xmath125 this distribution is then used to evaluate both the equivocation , eq.([equiv ] ) , and the entropy of the responses , eq.([outent ] ) .
we do not report the calculation , that is straightforward and analogous to the one reported in @xcite .
the final result , which is valid for a finite population size @xmath1 , and up to the linear approximation in @xmath126 , is analogous to eq.([input ] ) : @xmath127 thus , we expect that taking the limits @xmath128 and @xmath129 simultaneously in eq.([final_info ] ) , we should get to the same result : the output information should equal the input one when @xmath130 grows large . from eq.([final_info ] ) it is easy to show that : @xmath131;\ ] ] when @xmath132 one obtains exactly the linear limit , eq.([asymptot ] ) .
we have verified this analytical limit by studying numerically the approach to the asymptotic value of the mutual information .
fig.[fig3 ] shows the dependence of output information on the output noise @xmath133 , for 4 different choices of the ( reciprocal of the ) threshold , @xmath134 .
a large value , @xmath135 , implies linear output units . as expected , the output information , which always grows for decreasing values of the output noise , for @xmath135 approaches asymptotically the input information .
for increasing values of the output noise , the information vanishes with a typical sigmoid curve , with its point of inflection when the output matches the input noise .
we have then examined how the information in output ( compared to the input ) depends on the number @xmath19 of discrete correlates and on the width of the tuning function ( [ tuning2 ] ) , parametrized by @xmath136 , with respect to the continuous correlate .
fig.[fig4 ] shows a comparison between input and output information for a sample of 10 cells , as a function of @xmath19 .
both curves quickly reach an asymptotic value , obtained by setting @xmath137 in eq.([lambda2 ] ) for @xmath138 .
the relative information loss in output is roughly constant with @xmath19 .
a comparison is shown with the case where correlates are purely discrete , which is obtained by setting @xmath139 in eq.([tuning2 ] ) .
the curves exhibit a similar behaviour , even if the rise with @xmath19 is steeper , and the asymptotic values are higher .
this may be surprising , but it is in fact a consequence of the specific model we have considered , eq.([tuning_tot ] ) , where a unit has the same tuning curve to each of the discrete correlates , only varying its amplitude with respect to a value constant in the angle . as @xmath137 ,
most of the mutual information is about the discrete correlates , and the tuning to the continuous dimension , present for @xmath140 , effectively adds noise to the discrimination among discrete cases , noise which is not present for @xmath139 . with respect to the continuous dimension
, the selectivity of the input units can be increased by varying the power @xmath136 of the cosine from 0 ( no selectivity ) through 1 ( very distributed encoding , as for the discrete correlates ) to higher values ( progressively narrower tuning functions ) .
fig.[fig5 ] reports the resulting behaviour of the information in input and in output , for the case @xmath141 ( only a continuous correlate ) and @xmath142 ( continuous+discrete correlates ) .
increasing selectivity implies a `` sparser '' @xcite representation of the angle , the continuous variable , and hence less information , on average .
however if the correlate is purely continuous there is an initial increase , before reaching the optimal sparseness . it should be kept in mind , again , that the asymptotic equality of the @xmath141 and @xmath142 cases is a consequence of the specific model , eq.([tuning_tot ] ) , which assigns the same preferred angle to each discrete correlate .
the resolution with which the continuous dimension can be discriminated does not , within this model , improve with larger @xmath19 , while the added contribution , of being able to discriminate among discrete correlates , decreases in relative importance as the tuning becomes sharper .
figures [ fig4 ] and [ fig5 ] show that , as long as the output noise is non zero and the threshold is finite , information is lost going from input to output , but the information loss does not appear to depend on the structure and on the dimensionality of the correlate . note that , while the purely continuous case has been easily obtained by setting @xmath141 in the expression of @xmath143 , eq.([lambda2 ] ) , for the purely discrete case it is enough to set @xmath139 .
we have attempted to clarify how information about multi - dimensional stimuli , with both a continuous and a discrete dimension , is transmitted from a population of units with a known coding scheme , down to the next stage of processing .
previous studies had focused on the mutual information between input and output units in a two - layer threshold - linear network either with learning @xcite or with simple random connection weights @xcite .
more recent investigations have tried to quantify the efficiency of a population of units in coding a set of discrete @xcite or continuous @xcite correlates .
the analysis in @xcite has been then generalized to the more realistic case of multi - dimensional continuous+discrete correlates @xcite .
this work correlates with both research streams , in an effort to define a unique conceptual framework for population coding .
the main difference with the second group of studies is obviously the presence of the network linking input to output units .
the main difference with the first two papers , instead , is the analysis of a distinct mutual information quantity : not between input and output units , but between correlates ( `` stimuli '' ) and output units . in @xcite
it had been argued , for a number @xmath19 of purely discrete correlates , that the information _ about _ the stimuli reduces to the information about the `` reference '' neural activity when @xmath137 .
the reference activity is simply the mean response to a given stimulus when the information is measured from the variable , noisy responses around that means ; or it can be taken to be the stored pattern of activity , when the retrieval of such patterns is considered , as in @xcite .
true , the information about the stimuli saturates at the entropy of the stimulus set , but for @xmath137 this entropy diverges , only the linear term in @xmath0 is relevant @xcite , and the two quantities , information about the stimuli and information about the reference activity , coincide .
our present saddle point calculation is only able to capture , effectively , the mutual information which is linear in the number of input units , as mentioned above .
it fails to describe the approach to the saturating value , the entropy of the set of correlates , be this finite or infinite .
therefore , ours is close to a calculation of the information about a reference activity - in our case , the activity of the input units .
the remaining difference is that we can take into account , albeit solely in the linear term , the dependence on @xmath19 ( through the equation for @xmath138 , eq.([lambda2 ] ) ) , without having to take the further limit @xmath137 . due to the presence of a threshold and of a non zero output noise
the information in output is lower than that in input , and we have shown analytically that in the limit of a noiseless , linear input - output transfer function the ouptput information tends asymptotically to the input one .
we have not , however , introduced a threshold in the input units , which would be necessary for a fair comparison . in an independent line of research , recent work @xcite
has also quantified the contribution to the mutual information , in a different model , of cubic and higher order non - linearities in the transfer function , by means of a diagrammatic expansion in a noise parameter . in @xcite
it has been shown that the effect of a threshold in the input units on the input information results merely in a renormalization of the noise .
the resulting effect on the output information remains to be explored , possibly with similar methods .
considering mixed continuous and discrete dimensions in our stimulus set , we had been wondering whether the information loss in output depended on the presence or absence of discrete or continuous dimensions in the stimulus structure .
we have shown that for a fixed , finite level of noise this loss dose not depend significantly on the structure of the stimulus , but solely on the relative magnitude of input and output noise , and on the position of the output threshold .
a recent work has shown that the interplay between short and long range connectivities in the hopfield model leads to a deformation of the phase diagram with the appearence of novel phases @xcite .
it would be interesting to introduce short and long range connections in our model , and to examine how the coding efficiency of output neurons depends on the interaction between short and long range connections
. this will be the object of future investigations . | in a previous report we have evaluated analytically the mutual information between the firing rates of @xmath0 independent units and a set of multi - dimensional continuous+discrete stimuli , for a finite population size and in the limit of large noise . here , we extend the analysis to the case of two interconnected populations , where input units activate output ones via gaussian weights and a threshold linear transfer function .
we evaluate the information carried by a population of @xmath1 output units , again about continuous+discrete correlates .
the mutual information is evaluated solving saddle point equations under the assumption of replica symmetry , a method which , by taking into account only the term linear in @xmath0 of the input information , is equivalent to assuming the noise to be large . within this limitation , we analyze the dependence of the information on the ratio @xmath2 , on the selectivity of the input units and on the level of the output noise .
we show analytically , and confirm numerically , that in the limit of a linear transfer function and of a small ratio between output and input noise , the output information approaches asymptotically the information carried in input .
finally , we show that the information loss in output does not depend much on the structure of the stimulus , whether purely continuous , purely discrete or mixed , but only on the position of the threshold nonlinearity , and on the ratio between input and output noise . |
this edition of the _ quark matter _ series of conferences has witnessed an impressive number of new theoretical and experimental results , which is a testament to the activity in this vibrant field .
we discuss in turn some of the new results that were presented at the meeting .
starting with calculations of the electromagnetic emissivity of the quark - gluon plasma , photon emission rates have been known at leading order ( lo ) in @xmath0 , the strong fine structure constant , for little over a decade @xcite .
given that rhic phenomenology seems to favor values @xmath1 , one may wonder about the size of next - to - leading - order ( nlo ) contributions .
these have recently been calculated and a report on the results has been given at this meeting @xcite .
the nlo contribution is parametrically of order @xmath2 , where @xmath3 is the strong interaction coupling constant .
interestingly , the effect of small angle radiation gets enhanced at nlo , while the contributions at larger angles - of the same order in @xmath3 - get suppressed at nlo , owing in part to an interplay between enhanced scattering rates and a modification of the phase space .
the net effect , shown in the left panel of figure [ phot_nlo ] , is a modest `` k - factor '' , which deviates from unity by at most 20 - 30% at @xmath4 .
bearing in mind the caveats inherent to a perturbative analysis , this nlo study implies that the phenomenology that relies on those photon rates need not be redone .
the techniques developed there , however , will be of use for other finite temperature applications .
other communications on electromagnetic observables included a proposal to observe photons originating from qcd jet conversion by first triggering on a reconstructed jet , and then searching for the back - scattered photon correlated with this hard parton @xcite .
this technique should avoid contamination from the direct photon sources without an underlying hard process , and suppress fragmentation photons which populate low values of @xmath5 .
recent dilepton calculations were also shown which highlighted the effects of a finite shear viscosity coefficient on lepton - pair observables @xcite .
the dilepton yields were shown to be somewhat insensitive to the viscosity corrections , but the dilepton elliptic flow - as quantified by @xmath6 - shows greater sensitivity .
the absolute value of the predicted dilepton @xmath7 does remain small , which is consistent with the different hydrodynamic calculations of photon elliptic flow @xcite .
these , however , are at odds with current measurements by the phenix collaboration .
the theoretical results for dilepton yields were in accord with recent measurements by the star collaboration ( to be shown and discussed further below ) , and relied on vector spectral densities that were also found to be consistent with na60 data @xcite .
the data there appear to be consistent with a scenario where the charm quarks loose energy during the viscous evolution , see figure [ phot_nlo ] , right panel .
all channels of energy loss will need to be considered before a final assessment is made .
potential , and with the singlet free - energy , @xmath8 .
( left panel ) @xmath9 mev , ( right panel ) @xmath10 mev .
this plot is from ref .
@xcite.,title="fig:",scaledwidth=39.0% ] potential , and with the singlet free - energy , @xmath8 .
( left panel ) @xmath9 mev , ( right panel ) @xmath10 mev .
this plot is from ref .
@xcite.,title="fig:",scaledwidth=39.0% ] in what concerns non - perturbative techniques , advances in lattice qcd calculations were reported at the meeting .
one result highlights progress in the study of quarkonium suppression as a signal of the quark - gluon plasma ( qgp ) .
it addresses some of the ambiguities in extracting a static quark potential at finite temperatures , by identifying the static @xmath11 energy using a spectral decomposition of the temporal wilson loop @xcite . that work shows , on figure [ qqpot ] , that the singlet free energy and the actual potential are only similar at high temperatures , and at that temperatures commensurate with that of the crossover transition the potential deviates only slightly from it zero temperature behavior .
this step forward should impact some of the phenomenological studies presented at the meeting @xcite ( some of these results will be shown in sections to follow ) . still in the context of the in - medium modifications of quarkonium , another lattice communication reported on developments in the determination of the _ dynamical _ properties of the charmonium spectral density @xcite . by extracting the spectral profiles through the maximum entropy method ,
one can obtain the transverse component of the vector channel , and the pseudo scalar channel , both of which are shown in figure [ spectral ] . a clear momentum dependence is observed in both vector and pseudo - scalar spectral densities at the temperatures considered here , and dissolution is observed at @xmath12 in this quenched approximation .
clearly , calculations at other temperatures are needed to complete the description of the fate of charmonium moving through the heat bath , and to make contact with results obtained from using effective field theory or ads / cft techniques .
finally , results of calculations of @xmath13 suppression were shown , where the short - distance @xmath11 production cross - sections were evaluated using non - relativistic qcd ( nrqcd ) , and the longer distance in - medium effects were incorporated through formation and dissociation times for the quarkonium states .
thermal effects , neglected in the bound state wave function , appear necessary to interpret the current measurements , especially at the lhc @xcite .
effects of collisional , radiative , and net energy loss are shown separately .
( right panel ) the individual and combined effect of the collisional and radiative energy loss on @xmath14 meson @xmath15 , plotted together with alice results .
both plots are from @xcite.,title="fig:",scaledwidth=35.0% ] gev .
effects of collisional , radiative , and net energy loss are shown separately .
( right panel ) the individual and combined effect of the collisional and radiative energy loss on @xmath14 meson @xmath15 , plotted together with alice results .
both plots are from @xcite.,title="fig:",scaledwidth=40.0% ] the properties of heavy quarks in the hadronic medium - either cold or hot - were the subject of several talks reporting new results .
these included an attempt to calculated cold nuclear matter effects from first principles , and to estimate gluon saturation effects in charmonium production in d+a and a+a collisions @xcite .
the results of a study on the evolution of heavy quarks in an evolving viscous medium were also shown .
this evolution relied on langevin dynamics modified to also include radiative energy loss @xcite ( usually neglected ) , calculated in the higher - twist formalism @xcite .
the time - dependence of heavy quark energy profile is especially relevant to lepton pair studies , as charm quark decays - for example - form an irreducible dilepton background , as discussed further in subsequent sections .
the left panel of figure [ eloss ] shows the evolution in time of the energy spectrum of a 15 gev heavy quark , in a static qcd `` brick '' with @xmath16 mev .
clearly , the initial delta function is quickly broadened and shifted in energy , the evolution is inherently non - linear , and the radiative losses are important .
this point is driven home in the right panel , where both effects are needed to achieve a satisfactory understanding of lhc data of @xmath14-meson @xmath15 . in summary ,
some fundamental theoretical results were shown at this meeting , which reflected advances in our understanding of fundamental processes , as well as in the detailed modelling of relativistic nuclear collisions .
there were many new results shown for heavy flavor measurements , including those for @xmath14 and @xmath17 mesons , and their decay leptons .
it is found that at rhic , the charm cross section follows the number of binary collisions scaling from @xmath18 , to d+au , to au+au collisions at @xmath19 gev @xcite .
nuclear modification factors @xmath15 of @xmath20 at 200 gev au+au collisions show a strong @xmath21 dependence @xcite .
they increase as a function of @xmath21 , reach a maximum value at @xmath22 gev/@xmath23 and then decrease . at lhc , the @xmath15 of charmed mesons
are measured for @xmath24 gev/@xmath23 in pb+pb collisions at 2.76 tev @xcite . for @xmath25 gev/@xmath23 ,
the observed suppression factors of charmed mesons are similar to those of pions at rhic and lhc .
results from non - prompt @xmath13 measurements indicate that @xmath26 is larger than @xmath27 in central pb+pb collisions @xcite .
more precise measurements together with advanced theoretical calculations should enable us to understand the color and/or flavor dependence of energy loss in greater depth .
in addition , the @xmath14 meson shows a significant positive elliptic flow ( @xmath7 ) for @xmath28 gev/@xmath23 @xcite .
these measurements indicate that heavy flavor quarks interact with the medium strongly , which leads to a similar magnitude of suppression as for pions and generates a significant flow .
heavy quarkonia are thought to be an excellent probe to study color screening features in a hot , dense medium .
different quarkonium states are predicted to dissolve at different temperatures , owing to their different binding energies @xcite .
precise measurements of @xmath21 distributions of quarkonia at different centralities , collision systems and energies should serve as a thermometer of the qgp . however , in addition to color screening features , there are also many effects playing an important role such as gluon dissociation , heavy quark recombination , hot - wind dissociation , formation time , and cold nuclear matter effects @xcite
. collisions from asymmetric systems could provide additional information on cold nuclear matter effects .
it is found that in 200 gev cu+au collisions , @xmath15 of @xmath13 in the au - going direction is larger than that in the cu - going direction . a model taking into account cold nuclear matter effects including parton distribution functions and the @xmath13 break - up cross section qualitatively describes the feature , but under - predicts the magnitude of the suppression @xcite .
the additional suppression suggests hot , dense medium effects .
@xmath15 as a function of centrality in forward rapidity at rhic and lhc .
( upper - right panel ) : @xmath13 @xmath15 as a function of @xmath21 .
( bottom - left panel ) : high @xmath21 @xmath13 @xmath15 as a function of centrality .
( bottom - right panel ) : @xmath13 @xmath7 as a function of @xmath21 .
the normalization uncertainties are not shown .
model comparisons are shown as curves and bands.,title="fig:",scaledwidth=35.0% ] @xmath15 as a function of centrality in forward rapidity at rhic and lhc .
( upper - right panel ) : @xmath13 @xmath15 as a function of @xmath21 .
( bottom - left panel ) : high @xmath21 @xmath13 @xmath15 as a function of centrality .
( bottom - right panel ) : @xmath13 @xmath7 as a function of @xmath21 .
the normalization uncertainties are not shown .
model comparisons are shown as curves and bands.,title="fig:",scaledwidth=35.0% ] @xmath15 as a function of centrality in forward rapidity at rhic and lhc .
( upper - right panel ) : @xmath13 @xmath15 as a function of @xmath21 .
( bottom - left panel ) : high @xmath21 @xmath13 @xmath15 as a function of centrality .
( bottom - right panel ) : @xmath13 @xmath7 as a function of @xmath21 .
the normalization uncertainties are not shown .
model comparisons are shown as curves and bands.,title="fig:",scaledwidth=35.0% ] @xmath15 as a function of centrality in forward rapidity at rhic and lhc .
( upper - right panel ) : @xmath13 @xmath15 as a function of @xmath21 .
( bottom - left panel ) : high @xmath21 @xmath13 @xmath15 as a function of centrality .
( bottom - right panel ) : @xmath13 @xmath7 as a function of @xmath21 .
the normalization uncertainties are not shown .
model comparisons are shown as curves and bands.,title="fig:",scaledwidth=35.0% ] figure [ fig:1 ] shows systematic comparisons of @xmath13 suppression patterns at rhic and lhc . in central collisions , @xmath13 is less suppressed in pb+pb at 2.76 tev for @xmath29 than in au+au collisions at 200 gev for @xmath30 @xcite .
a similar feature is observed for mid - rapidity comparisons between lhc and rhic energies .
comparison to model calculations indicates that recombination plays a more significant role at lhc @xcite .
furthermore , the @xmath21 dependence of @xmath13 suppression @xcite , the flow pattern of @xmath13 as a function of @xmath21 @xcite , and the high @xmath21 @xmath13 suppression pattern as a function of centrality @xcite , reported at lhc and rhic , can be described consistently by model calculations incorporating color screening and recombination features .
we note that substantial suppression observed at high @xmath21 points to the color screening feature since the recombination and cold nuclear matter effects are negligible there .
a cleaner probe to study the color screening feature is @xmath31 since the @xmath32 recombination is small .
measurements at cms show that @xmath33 is 0.4 - 0.5 while @xmath34 is 0.1 in central pb+pb collisions at 2.76 tev @xcite . at 200 gev
, @xmath35 is about 0.4 in central au+au collisions @xcite . considering feed - down contributions [ @xmath3650% for @xmath37 ,
measurements at rhic and lhc indicate that @xmath38 is completely melted and that @xmath39 is strongly suppressed in central a+a collisions . in the future
, different @xmath31 states should be measured at rhic with the muon telescope detector ( mtd ) upgrade at star and sphenix upgrade at phenix @xcite . in light of rhic and lhc precise quarkonium measurements ,
we are in the era to study color screening features of hot , dense medium .
w and z bosons and high energy direct or isolated photons are ideal probes to study the initial nuclear wave function effect since they are produced at initial impact and go through the medium with minimum interactions . at 2.76 tev , the w and z boson production is found to follow number of binary collisions scaling from @xmath18 to pb+pb collisions @xcite .
the isolated photon production also follows binary scaling for @xmath40 gev/@xmath23 at mid - rapidity @xcite . at 200 gev , @xmath15 for direct photons
is consistent with unity within uncertainties at @xmath41 gev/@xmath23 in au+au collisions at mid - rapidity @xcite .
these measurements all suggest that initial - state effects are small for electro - weak probes .
in addition to high energy photon measurements , lower energy photons are used to study thermal radiation . for @xmath42 gev/@xmath23 , phenix measured direct photon yields from di - electron measurements and found an excess in 0 - 20% au+au over @xmath18 at @xmath19 gev , which is exponential in @xmath21 with slope parameters 221 mev @xcite . at lhc ,
an excess of direct photon yield in 0 - 40% pb+pb collisions above @xmath18 was reported at @xmath43 2.76 tev , exponential in @xmath21 with slope parameters 304 mev @xcite .
if indeed the excess is from the qgp phase , the measurements at rhic and lhc would indicate that the initial temperature for the qgp evolution is as high as 300 - 600 mev @xcite . on the other hand ,
@xmath7 has been measured for direct photons and found to be substantial in the range @xmath42 gev/@xmath23 in central 0 - 20% au+au collisions at @xmath19 gev @xcite .
model calculations @xcite for qgp thermal photons in this kinematic region significantly under - predict the observed @xmath7 , though if a significant contribution from the hadronic sources at later stages is added , the excess of the spectra and the observed @xmath7 at @xmath42 gev/@xmath23 are described reasonably well @xcite . it has been proposed that dilepton @xmath7 measurements as a function of @xmath21 in different mass regions would enable us to probe the properties of medium from a hadron - gas - dominated to a qgp - dominated scenario @xcite . at qm2012 ,
star reported the di - electron @xmath7 measurements from 200 gev au+au collisions . within uncertainties ,
the data are compatible with simulations containing contributions from known hadronic sources without qgp or hadron gas thermal radiation @xcite .
however , with more statistics , the di - electron @xmath7 measurement will provide additional sensitivity to study the thermal radiation from the different phases .
we now come to measurements of dilepton spectra . at rhic ,
the phenix experiment observed a significant enhancement in the @xmath44 data above the expectation from hadronic sources for @xmath45 gev/@xmath46 for @xmath47 gev/@xmath23 in 200 gev au+au collisions @xcite .
models @xcite that successfully describe the sps dilepton data consistently fail to describe the phenix data . with the time - of - flight detector upgrade ,
star reported the di - electron spectra in 200 gev au+au collisions at qm2011 and the low - mass excess was not as significant as what phenix observed in 0 - 80% and 0 - 10% collisions @xcite .
further comparisons point to the fact that the discrepancy between star and phenix is in collisions in the 0 - 20% centrality class only . at this conference , phenix reported the di - electron results from the hadron blind detector from 20 - 92% au+au collisions at 200 gev , which are consistent with those in the previous publication @xcite .
star reported the di - electron spectra from 19.6 , 39 , and 62.4 gev au+au collisions @xcite .
a broadened spectral function @xcite , which describes sps dilepton data , consistently accounts for the star low mass excess at 19.6 , 62.4 and 200 gev , as shown in fig .
[ fig:2 ] .
see also ref .
is there a connection to chiral symmetry restoration ?
chiral condensates , which have a direct connection to spectral functions , strongly depend on total baryon density and temperature . to establish a link between dilepton measurements and chiral symmetry restoration , we need more theoretical efforts and more precise measurements over a broad beam energy range from fair , rhic and lhc . in the future ,
e - muon correlations from the mtd upgrade will measure the correlated charm contribution @xcite .
this should enable us to obtain the qgp thermal radiation signal using the intermediate - mass dilepton measurements at rhic .
this edition of the quark matter conferences has seen a stream of results unprecedented in their quality and in their quantity , being presented by the theory community , and by the experimental collaborations working at rhic and at the lhc .
calculations have brought us significantly closer to an _ ab - initio _ modelling of relativistic heavy ion collisions . the precise charm and bottom measurements will constrain heavy flavor dynamics in hot , dense medium .
precise quarkonium measurements put us into the era of studying and characterizing color screening properties .
thermal photon and dilepton measurements from a broad beam energy range enable us to study the fundamental properties of qgp , of chiral symmetry restoration , and will provide stringent tests of the dynamical evolution scenarios of relativistic nuclear collisions .
+ * acknowledgment * the work of cg is supported in part by the natural sciences and engineering research council of canada , and the work of of lr is supported in part by the u. s. department of energy under contract no .
de - ac02 - 98ch10886 .
p. b. arnold , g. d. moore and l. g. yaffe , jhep * 0112 * , 009 ( 2001 ) .
d. teaney , these proceedings .
r. fries , these proceedings .
g. vujanovic , these proceedings .
r. chatterjee and d.k .
srivastava , .
j. ruppert , c. gale , t. renk , p. lichard and j. i. kapusta , phys .
lett . * 100 * , 162301 ( 2008 ) .
a. bazavov and p. petreczky , these proceedings .
r. rapp , these proceedings .
heng - tong ding , these proceedings .
r. sharma , these proceedings .
k. tuchin , these proceedings .
s. cao , these proceedings .
a. majumder , .
w. xie , david tlusty ( star ) , these proceedings .
z. conesa del valle ( alice ) , these proceedings .
a. grelli ( alice ) , these proceedings . c. mironov , m. jo ( cms ) , these proceedings .
d. caffarri ( alice ) , these proceedings .
a. d. frawley , t. ullrich and r. vogt , .
r. hollis , m. rosati ( phenix ) , these proceedings .
adare _ et al .
r. arnaldi , i.ch .
arsene , e. scomparin ( alice ) , these proceedings .
adare _ et al .
_ , . l. adamczyk _
et al . _ ,
arxiv : 1208.2736 . h. yang ( alice ) , these proceedings . b. trzeciak ( star ) , these proceedings .
d. moon ( cms ) , these proceedings .
cms collaboration , arxiv : 1208.2826
. l. ruan _
et al . _ , ; http://drupal.star.bnl.gov/star/system/files/mtd@[email protected] .
et al . _ , arxiv : 1207.6378 . l. benhabib , r. granier de cassagnac , g. roland , g. stephans ( cms ) , these proceedings .
j. dolejsi , i. grabowska - bold , p. steinberg , b. wosiek ( atlas ) , these proceedings .
s. afanasiev _
_ , arxiv:1205.5759 .
et al . _ , . m. wilde ( alice ) , these proceedings .
adare _ et al .
_ , arxiv : 1105.4126 . h. van hees , c. gale , and r. rapp , .
r. chatterjee _ et al .
x. dong , f. geurts , b. huang ( star ) , these proceedings .
x. cui ( star ) , qm2012 poster contribution .
adare _ et al .
r. rapp , j. wambach , and h. van hees , arxiv:0901.3289 .
o. linnyk _
et al . _ , . h. xu _
et al . _ , .
l. adamczyk _
et al . _ , ; l. ruan _ et al .
. j. zhao _
e. atomssa , i. tserruya , ( phenix ) , these proceedings .
b. huang , p. huck ( star ) , qm2012 poster contribution . l. ruan _ et al .
_ , arxiv:1207.7043 . | we summarize and discuss some of the experimental and theoretical results on heavy flavor , quarkonia , and electro - weak probes presented at _
quark matter 2012_. # 1#2#3#4#1 * # 2 * , # 3 ( # 4 ) |
antonov in 1962 @xcite discovered an instability that later became known as antonov s instability or gravothermal catastrophe @xcite .
it signified the beginning of the study of thermodynamics of self - gravitating systems @xcite and the nowadays rapidly developing field of statistical mechanics with long - range interactions @xcite .
+ in the original formulation of antonov s instability @xcite , the system consists of particles ( stars ) bound by a spherical shell with insulating and perfectly reflecting walls .
the particles interact only with newtonian gravity ( no cosmological constant or any relativistic effects are present ) and the number of particles and the energy of the system are constant , i.e. the system is studied in the microcanonical ensemble .
it was found that the entropy of the system has no global maximum and that for @xmath0 no equilibrium states exist
. for @xmath1 there exist local entropy maxima , i.e. metastable states , only for density contrast below the critical value @xmath2 , where @xmath3 , @xmath4 are the central and edge densities , respectively . + in gravity
the ensembles are not equivalent . in the microcanonical ensemble , stable equilibria with negative specific heat
this negative specific heat region is replaced by a phase transition in the canonical ensemble @xcite .
the antonov system in the canonical ensemble ( with no cosmological constant ) , is studied in detail by chavanis @xcite .
he found that the free energy does not have a global minimum and equilibria exist only for temperatures greater than the critical dimensionless temperature @xmath5 which corresponds to dimensionless inverse temperature @xmath6 .
these equilibria are stable only for density contrast less than the critical value @xmath7 ( see ref .
@xcite , as well ) .
+ we studied the gravothermal instability with a cosmological constant @xmath8 in the microcanonical ensemble in ref .
it was found that for a fixed energy , the maximum radius ( which equals @xmath9 with no @xmath8 ) is increasing with increasing @xmath8 , while for @xmath10 there appears a second critical radius which is decreasing and the two radii merge at some marginal @xmath8 .
this was identified as a reentrant behaviour ( see ref .
@xcite for details on reentrant phase transitions in statistical systems with long - range interactions ) . in the present work ,
the system is placed in a heat bath and the walls are non - insulating , i.e. we study the canonical ensemble .
in contrast to the microcanonical one , we find that the critical radius @xmath11 , for a fixed temperature and mass , is decreasing with increasing cosmological constant .
in addition the instability occurs for @xmath12 in the canonical ensemble , in contrast to the microcanonical one .
this is evident in case @xmath13 , since then there exist no equilibria if @xmath14 for a fixed temperature and mass .
a reentrant phase transition appears in the canonical ensemble , when one examines the system with respect to the temperature for a fixed radius .
there appear two critical temperatures for @xmath10 . for the intermediate temperature values , no thermodynamic equilibria are possible .
in the newtonian limit , the poisson equation is modified @xcite in the presence of a cosmological constant @xmath8 as : @xmath15 with @xmath16 .
thus , the gravitational potential can be decomposed to two parts @xmath17 where @xmath18 is the newtonian and @xmath19 the cosmological potential , that are given by @xmath20 for convenience we call the newtonian limit of de sitter ( @xmath21 ) and anti - de sitter ( @xmath22 ) spaces @xcite just ds and ads , respectively . + we consider a self - gravitating gas of @xmath23 particles with unity mass inside a spherical non - insulating shell inside a heat bath and restrict only to spherical symmetric configurations .
we work in the mean field approximation , for which the @xmath23-body distribution function is replaced by the one body distribution function @xmath24 .
the boltzmann entropy is defined as @xmath25 .
the helmholtz free energy is equal to @xmath26 .
equivalently one can work with the massieu function @xcite @xmath27 that gives @xmath28 the maximization of @xmath29 with constant @xmath30 and the maximization of @xmath31 with constant @xmath32 ( and constant @xmath33 in both cases ) , with respect to perturbations @xmath34 , are the same @xcite to first order in @xmath34 , i.e. give the same equilibria , described by the maxwell - boltzmann distribution function @xmath35 where @xmath36 this is proved in @xcite without a cosmological constant , and holds in the presence of @xmath8 as well , with the difference being that @xmath37 does not satisfy poisson equation now , but equation ( [ eq : poisl ] ) .
the two ensembles have the same equilibria . however , the second variations of @xmath29 and @xmath31 are different .
therefore , what is different in the two ensembles is the turning point at which an instability sets in .
this causes a great qualitative difference for the two ensembles .
+ introducing the dimensionless variables @xmath38 , @xmath39 and @xmath40 , and using equation ( [ eq : rho ] ) , equation ( [ eq : poisl ] ) becomes @xmath41 called the emden-@xmath8 equation .
let us call @xmath42 the value of @xmath43 at @xmath44 . in order to generate the series of equilibria needed to study the stability of the system , the emden-@xmath8 equation has to be solved with initial conditions @xmath45 , keeping @xmath33 constant and for various values of the parameters @xmath46 , @xmath47 , @xmath3 .
this is a rather complicated problem , since , unlike @xmath13 case , while solving for various @xmath48 , mass is not automatically preserved , because of the mass scale @xmath49 that @xmath8 introduces .
a suitable @xmath50 value has to be chosen at each @xmath48 .
we define the dimensionless mass @xmath51 where @xmath52 is the mean density of matter .
calling @xmath42 the value of @xmath43 at @xmath44 , equation ( [ eq : mdef ] ) can also be written as @xmath53 where @xmath54 is the dimensionless inverse temperature .
it can be calculated by integrating the emden-@xmath8 equation , to get : @xmath55 we developed an algorithm to solve equation ( [ eq : emdenl ] ) for various values of @xmath50 , @xmath48 keeping @xmath56 fixed . from equation ( [ eq : mdef ] ) it is clear that solving for various fixed @xmath56 can be interpreted as solving for various @xmath46 and/or @xmath44 for a fixed @xmath33 .
+ we find that in ds case there exist multiple series of equilibria for a given cosmological constant , while in ads case there is just one series likewise @xmath13 case .
studying the series @xmath57 for both ads and ds , we determined the critical values of radii and temperature beyond which there are no equilibria .
these critical values are plotted in figures [ fig : rcr ] and [ fig : tcr ] , with respect to @xmath46 , where the unshaded region is the region of instability in both figures .
+ in figure [ fig : rcr ] the radius denoted @xmath58 is the minimum radius for which thermodynamic equilibria exist in the canonical ensemble .
@xmath59 is the radius of the homogeneous solution which exists only in ds and is defined by equation @xmath60 , i.e. @xmath61 . in the shaded region of figure [ fig : rcr ] there
can always be found metastable states . for ads case ( @xmath22 )
, these have always monotonically decreasing density ( @xmath62 ) and suffer a transition to unstable equilibria at a critical density contrast @xmath63 whose value depends on @xmath46 .
for ds case and for @xmath64 , the metastable states suffer a transition to unstable equilibria likewise ads , but the density is not necessarily monotonic , neither does @xmath62 always hold .
for each @xmath46 there exists a tower of solutions with qualitative ( and quantitative of course ) different density functions . in the region @xmath65 ,
all metastable states have @xmath66 .
the ones , whose density is a monotonic function of @xmath44 do not suffer any transition to unstable equilibria , while all others series have a turning point .
+ in figure [ fig : tcr ] we see that in ads the critical temperature is decreasing for increasing cosmological constant , while in ds there appears a reentrant phase transition . for small positive values of @xmath46 ,
compared to the mean density @xmath52 , the system can not stabilize for low temperatures .
it needs higher temperature values to obtain a significant pressure gradient to balance gravity . as @xmath46 is increasing , the cosmological repelling force increases , enforcing the total outward pushing force and hence , enabling the system to stabilize at lower temperatures .
point @xmath67 , denotes the marginal value @xmath68 for which a dynamically static equilibrium is allowed .
this limiting configuration corresponds to a singular solution at which all mass is concentrated at the edge .
it can easily be calculated : @xmath69 @xmath70 @xmath71 @xmath72 @xmath73 . for @xmath74 ,
solutions with @xmath75 ( static dynamical equilibria ) do exist and as @xmath46 increases the mass is allowed to be closer to the center .
all thermodynamic equilibria below the lower critical branch ( low temperatures ) have increasing density ( @xmath66 ) . for fixed @xmath44 , @xmath33 and a fixed @xmath46 with @xmath76 ,
as we increase the temperature beginning from zero , we pass from a region of metastable states with @xmath66 to a region of no equilibria , at a first critical temperature @xmath77 .
we have to increase the temperature even more to reach a new region of metastable states at a second critical temperature @xmath78 . for @xmath79 the metastable states have decreasing density @xmath62 if @xmath80 .
this is a reentrant phase transition similar to the ones found in other statistical systems with long - range interactions @xcite .
the critical temperature w.r.t . @xmath46 for fixed @xmath33 , @xmath44 . ] the critical temperature w.r.t .
@xmath46 for fixed @xmath33 , @xmath44 . | we present here how the gravothermal or antonov s instability , which was originally formulated in the microcanonical ensemble , is modified in the presence of a cosmological constant and in the canonical ensemble .
in contrast to the microcanonical ensemble , there is a minimum , and not maximum , radius for which metastable states exist .
in addition this critical radius is decreasing , and not increasing , with increasing cosmological constant .
the minimum temperature for which metastable states exist is decreasing with increasing cosmological constant , while above some positive value of the cosmological constant , there appears a second critical temperature . for lower temperatures than the second critical temperature value ,
metastable states reappear , indicating a typical reentrant phase transition .
the two critical temperatures merge when the cosmological density equals one half the mean density of the system . |
a study on the heavy - to - light exclusive processes plays a complementary role in determination of the fundamental parameters of the standard model and in developing the qcd theory .
and there is an increasing demand for more reliable qcd calculations of the heavy - to - light form factors .
we have done a consistent analysis of the @xmath11 transition form factor in refs.@xcite , which shows that the results from the pqcd approach , the lattice qcd approach and the qcd light cone sum rules ( lcsrs ) are complementary to each other and by combining the results of those three approaches , one can obtain an understanding of the @xmath11 transition form factor in the whole physical regions .
it is argued that by applying the @xmath0 factorization approach @xcite , where the transverse momentum dependence for both the hard scattering part and the non - perturbative wavefunction , the sudakov effects and the threshold effects are included , one can regulate the endpoint singularity from the hard scattering part effectively and derive a more reliable pqcd result of the b - meson decays .
furthermore , by applying the b - meson wave functions up to next - to - leading fock state , ref.@xcite calculated the @xmath11 transition form factor up to @xmath12 and also made some discussions on the reasonable regions for the two phenomenological b - meson wave function parameters @xmath13 and @xmath14 , where @xmath13 is the effective mass of b meson that determines the b - meson s leading fock state behavior and @xmath14 is a typical parameter that determines the broadness of the b - meson transverse distribution .
since both pion and kaon are pseudo - scalar mesons , it will be interesting to give a consistent pqcd analysis of the @xmath15 transition form factor up to order @xmath16 based on the results of @xmath11 transition form factor . in the literature ,
the @xmath1 transition form factor has been studied under several approaches @xcite .
a pqcd calculation has been done in ref.@xcite , which can be roughly treated as a leading - order estimation @xmath17 since some of the power suppressed terms both in the hard scattering amplitude and the b - meson wave function have been neglected .
the @xmath1 transition form factor has also been analyzed by several groups under the qcd lcsr approach @xcite , where some extra treatments to the correlation function either from the b - meson side or from the kaonic side are adopted to improve their lcsr estimations .
new sum rule for the @xmath1 form factor is derived by expanding the correlation function near the light cone in terms of b - meson distributions @xcite , in which the contributions of the quark - antiquark and quark - antiquark - gluon components in the b - meson are taken into account . while in ref.@xcite an improved lcsr approach that had been raised in ref.@xcite was adopted to eliminate the contributions from the most uncertain kaonic twist-3 wave functions and to enhance the reliability of sum rule calculations of the @xmath18 form factor .
a systematic qcd lcsr calculation has been done in ref.@xcite by including the one - loop radiative corrections to the kaonic twist-2 and twist-3 contributions , and the kaonic leading - order twist-4 corrections .
it can be found that the main uncertainties in estimation of the @xmath1 transition form factor come from the b - meson wave function and the kaonic twist-2 and twist-3 wave functions . in doing the pqcd calculations on the b - meson decays ,
an important issue is whether we need to take both the two wavefunctions @xmath3 and @xmath4 into consideration or simply @xmath3 is enough ? by taking the frequently used first type definition of @xmath19 and @xmath20 , where @xmath21 are defined in ref.@xcite , it can be found that @xcite their distribution amplitudes have quite different endpoint behaviors even under the wandzura - wilczek ( ww ) approximation @xcite , such difference may be strongly enhanced by the hard scattering kernel .
for example , the ratio between the contributions of @xmath4 and @xmath3 is about @xmath22 @xcite for the @xmath11 from factor in the large recoil regions .
so the contribution from @xmath4 under the above definition can not be neglected and it is needed to suppress the big contribution coming from @xmath3 so as to obtain a reasonable total contributions . to derive more accurate estimation , ref.@xcite raised the second type definition of @xmath23 and @xmath24 , under which the contribution from @xmath4 is of order @xmath25 to that of @xmath3 @xcite . for convenience , in the following
, we shall adopt the second type definition of @xmath3 and @xmath4 to do our calculation .
then one may ask is it enough to give a @xmath16 estimation with @xmath3 and @xmath4 under the ww approximation ?
as has been pointed out in ref.@xcite , the 3-particle fock states contributions to the b - meson wave function can be estimated by attaching an extra gluon to the internal off - shell quark line , and then @xmath26 power suppression is induced in comparison to that of the ww - part b - meson wave functions .
recently , the b - meson light - cone wave functions have been investigated up to next - to - leading order in fock state expansion in the heavy quark limit @xcite .
it was shown that by using the relations between 2- and 3- particle wave functions derived from the qcd equations of motion and the heavy - quark symmetry , one can give a constraint on the transverse momentum dependence of the b - meson wave function , whose distribution tends to be a hyperbola - like curve other than a simple delta function that is derived under the ww approximation .
these results provide us a chance to give a consistent pqcd analysis of the @xmath1 form factor up to order @xmath16 .
another issue we need to be more careful is about the kaonic wave functions .
the distribution amplitude ( da ) for the twist-2 wave function @xmath27 has been deeply studied , e.g. by the light - front quark model @xcite , the lcsr approach @xcite and the lattice calculation @xcite and etc . in ref.@xcite ,
the qcd sum rule for the diagonal correlation function of local and nonlocal axial - vector currents is used , in which the contributions of condensates up to dimension six and the @xmath28-corrections to the quark - condensate term are taken into account .
the first gegenbauer moment @xmath29 of the twist-2 da derived there , i.e. @xmath30 , is consistent with that of the lattice calculations @xcite , so we shall constrain @xmath29 within this range when constructing a model for @xmath27 . as for the twist-3 wave function @xmath8 , the calculations of it has more uncertainty than that for the leading twist , e.g. its da moments in refs.@xcite are quite different from each other , where the da moments in refs.@xcite are derived by using the qcd light - cone sum rules and the moments in ref.@xcite are derived based on the effective chiral action from the instanton . under the pqcd approach , according to our experience on the @xmath11 transition form factor @xcite and the pion electro - magnetic form factor @xcite , it can be found that for a twist-3 wave function with a better endpoint behavior other than the asymptotic one , the twist-3 s contributions are indeed power suppressed to the leading twist s contribution that favor the conventional power counting rules . in the present paper
, we shall adopt two groups of da moments @xcite together with the brodsky - huang - lepage ( bhl ) prescription @xcite to construct a model for @xmath8 , and then make a discussion on its uncertainty to the @xmath1 transition form factor .
the @xmath7-breaking effects shall also be included for constructing the kaonic wave functions .
the purpose of the paper is to reexamine the @xmath1 transition form factor in the pqcd @xmath0 factorization approach up to @xmath31 . under the @xmath0 factorization approach , the full transverse momentum dependence ( @xmath0-dependence ) for both the hard scattering part and the non - perturbative wave function ,
the sudakov effects and the threshold effects are included to cure the endpoint singularity .
furthermore , we shall analyze the power suppressed contributions from both the wave functions and the hard scattering amplitude and then give a consistent analysis of the form factor up to @xmath16 , which have not been considered in the literature .
in section ii , we give the calculated technology for the form factor in the large recoil regions .
also we present the model wave functions of the kaon with better endpoint behavior in the same section , which are constructed based on the bhl prescription @xcite and the da moments obtained in ref.@xcite . in section iii
, we give our numerical results .
conclusion and a brief summary are presented in the final section .
the @xmath1 transition form factors @xmath32 and @xmath33 are defined as follows : @xmath34 f_+^{b\to k}(q^2)+ \frac{m_b^2-m_k^2}{q^2 } q_{\mu}f_0^{b\to k}(q^2)\ ] ] where @xmath35 should be equal to @xmath36 so as to cancel the poles at @xmath37 .
the amplitude for the @xmath1 transition form factor can be factorized into the convolution of the wave functions for the respective hadrons with the hard - scattering amplitude . in the large recoil regions ,
the @xmath1 transition form factor is dominated by a single gluon exchange in the lowest order . in ref.@xcite , we have done a consistent analysis of the @xmath11 transition form factor within the @xmath0 factorization approach , where the power suppressed terms up to @xmath16 have been kept explicitly in the hard scattering amplitude .
the interesting reader may refer to ref.@xcite for more details . should be changed to @xmath38 , in eq.(5 ) the factor @xmath39 $ ]
should be changed to @xmath40 $ ] and in eq.(7 ) @xmath41 should be changed to @xmath42 . ] more specifically , for the present case , one needs to know the momentum projection for the matrix element of the kaon and b meson in deriving the hard scattering amplitude . by keeping the transverse momentum dependence in the wave function
, the momentum projection for the matrix element of the kaon has the following form , @xmath43 where @xmath44 is the kaon decay constant and @xmath45 is the phenominological parameter @xmath46 , which is a scale characterized by the chiral perturbation theory .
@xmath47 is the twist-2 wave function , @xmath48 and @xmath49 are twist-3 wave functions , respectively .
@xmath50 , @xmath51 and @xmath52 are two null vectors that point to the plus and the minus directions , respectively . and the momentum projection for the matrix element of the b meson can be written as @xcite : @xmath53 \gamma_5\right\}_{\alpha\beta}\ , \ ] ] where @xmath54 is the momentum fraction for the light spectator quark in the b meson and @xmath55 .
the four - component @xmath56 in eq.([projectorb ] ) is defined through , @xmath57 with @xmath58 . by including the sudakov form factors and the threshold resummation effects , one can obtain the formulae for the @xmath1 transition form factors @xmath32 and @xmath33 in the transverse configuration @xmath59-space , which can be simply obtained from ref.@xcite by changing the pion wave functions to the present case of kaon and by changing @xmath3 and @xmath4 to the second type definition as described in the introduction . in pqcd approach ,
the parton transverse momenta @xmath60 are not negligible around the endpoint region .
the relevant sudakov factors from both @xmath61 and the threshold resummation @xcite can cure the endpoint singularity which makes the calculation of the hard amplitudes infrared safe , and then the main contribution comes from the perturbative region . also it is necessary to keep the transverse momentum dependence in the wave functions to derive a more reliable estimation in pqcd . in principle , the bethe - salpeter formalism @xcite and the discretized light cone quantization approach @xcite could determine the hadronic wave functions , but in practice there are many difficulties in getting the exact wave functions at present .
the bhl prescription @xcite , which connects the equal - time wave function in the rest frame and the wave function in the infinite momentum frame , provides a useful way to use the approximate bound state solution of a hadron in terms of the quark model as the starting point for modeling the hadronic wave function .
so in the present paper , we will adopt the bhl prescription for constructing the kaonic wave functions .
while for the @xmath6-meson wave function , they have been investigated up to next - to - leading order in fock state expansion in the heavy quark limit in ref.@xcite , which shall be adopted to do our discussions .
a simple model has been raised in ref.@xcite for the b - meson wave functions @xmath62 and @xmath63 , which keep the main features caused by the 3-particle fock states and whose transverse momentum dependence are still the like - function of the off - shell energy of the valence quarks but shall broaden the transverse momentum dependence under the ww approximation to a certain degree . and in the compact parameter @xmath64-space , it reads @xcite @xmath65 j_{\delta-1 } [ \kappa ] + ( 1-\delta)\gamma[2-\delta ] j_{1-\delta}[\kappa]\big)\left ( \frac{\kappa}{2 } \right)^{1-\delta}\ ] ] and @xmath66 j_{\delta-1 } [ \kappa ] + ( 1-\delta)\gamma[2-\delta ] j_{1-\delta}[\kappa]\big)\left ( \frac{\kappa}{2 } \right)^{1-\delta } , \label{newmodel2}\ ] ] where @xmath67 , @xmath68 and @xmath69 .
the factor @xmath70 is introduced to ensure their fourier transformation , i.e. @xmath71 , satisfy the normalization , @xmath72 .
it can be found that both @xmath62 and @xmath63 have the same transverse momentum dependence and only two phenomenological parameters @xmath13 and @xmath14 are introduced .
@xmath13 is the effective mass of b meson , @xmath73 , which determines the b meson s leading fock state behavior .
@xmath14 is a typical parameter that determines the broadness of the b - meson transverse distribution in comparison to the ww - like one .
the ww - like b - meson wave functions in the compact parameter @xmath64-space can be found in ref.@xcite .
and a direct comparison shows that when @xmath74 , the transverse momentum dependence of the b - meson wave function in eqs.([newmodel1],[newmodel2 ] ) returns to a simple @xmath14-function , which is the same as that of the b - meson wave function under the wandzura - wilczek approximation @xcite . according to the definitions , we have @xmath75 , @xmath76 and @xmath77 .
next , we construct the kaonic twist-2 wave function based on its first gegenbauer moment @xmath78 and on the bhl prescription @xcite . the first gegenbauer moment @xmath78 has been studied by the light - front quark model @xcite , the lcsr approach @xcite and the lattice calculation @xcite and etc . in ref.@xcite ,
the qcd sum rule for the diagonal correlation function of local and nonlocal axial - vector currents is used , in which the contributions of condensates up to dimension six and the @xmath28-corrections to the quark - condensate term are taken into account .
the moments derived there are close to that of the lattice calculation @xcite , so we shall take @xmath79 to determine the model wave function @xmath27 .
based on the bhl prescription , we take the twist-2 wave function of kaon as @xmath80\times \frac{a_k}{x(1-x ) } \exp \left[-\beta_k^2 \left(\frac{k_\perp^2+m_q^2}{x}+ \frac{k_\perp^2+m_s^2 } { 1-x}\right)\right],\label{wavek}\end{aligned}\ ] ] where @xmath81 , @xmath82 is the gegenbauer polynomial . in comparison to the pion wave function ( e.g. @xcite )
, it can be found that the @xmath7 symmetry is broken by a non - zero @xmath83 and by the mass difference between the @xmath84 quark and @xmath85 ( or @xmath86 ) quark in the exponential factor .
the @xmath7 symmetry breaking in the lepton decays of heavy pseudoscalar mesons and in the semileptonic decays of mesons have been studied in ref.@xcite . for definiteness
, we take the conventional values for the constitute quark masses : @xmath87 and @xmath88 .
the parameters @xmath89 , @xmath83 and @xmath90 can be determined by the value of @xmath91 together with the normalization condition : @xmath92 and the constraint @xmath93 @xcite , where the average value of the transverse momentum square is defined as @xmath94 the parameter @xmath95 in the model wave function stands for some hadronic scale that is of order @xmath96 . for clarity
, we set @xmath97 gev .
the da @xmath98 is defined as @xmath99 .
the first gegenbauer moment @xmath100 of refs.@xcite can be defined as @xmath101 ^ 2 } \ ; , \ ] ] where @xmath102 other than @xmath98 should be adopted , since in refs.@xcite @xmath103 stands for the momentum fraction of @xmath84-quark in the kaon ( @xmath104 ) , while in the present paper we take @xmath103 as the momentum fraction of the light @xmath105-(anti)quark in the kaon ( @xmath106 ) . or @xmath107 in connection to the hard scattering part .
this will cause errors when the @xmath7-symmetry is broken . ]
based on the above discussions , we can obtain the values for @xmath89 , @xmath83 and @xmath90 : @xmath108\;,\ ; \beta_k \cong 0.877{\rm gev}^{-1 } , \ ] ] where the values of @xmath89 and @xmath90 are almost constant , i.e. their changes @xmath109 and @xmath110 are less than @xmath111 by varying @xmath100 within the range of @xmath112 $ ] .
more specifically , for the case of @xmath113 , we have @xmath114 as will be seen that the contributions from twist-3 wave function @xmath115 is less important in comparison to that of @xmath116 and @xmath117 , which is similar to the case of @xmath11 transition from factor @xcite .
so basing on the bhl prescription , we directly take the twist-3 wave function @xmath118 of kaon as @xmath119,\label{waves}\end{aligned}\ ] ] where @xmath120 can be determined by its normalization condition , i.e. @xmath121 . as for the twist-3 wave function @xmath117 ,
its da s asymptotic behavior , @xmath122 , so its endpoint singularity is much more serious .
then the transverse momentum dependence of @xmath117 is much more important than that of @xmath116 and @xmath115 in order to cure the endpoint singularity .
one can construct @xmath117 in the following form , @xmath123\times\nonumber\\ & & \frac{a_p}{x(1-x ) } \exp \left[-\beta_k^2 \left(\frac{k_\perp^2+m_q^2}{x}+ \frac{k_\perp^2+m_s^2 } { 1-x}\right)\right],\end{aligned}\ ] ] where @xmath103 stands for the light quark @xmath105 s momentum fraction , @xmath124 and @xmath125 are gegenbauer polynomials and the coefficients @xmath126 , @xmath127 and @xmath128 can be determined by its da moments .
the da @xmath129 is defined as @xmath130 . with its parameters determined by the two groups of da moments @xcite . the solid line and the dashed line are for @xmath131 and @xmath132 respectively . for comparison ,
the big dotted line and the dash - dot line are for @xmath133 @xcite and @xmath134 @xcite respectively .
the dotted line is the asymptotic behavior of @xmath122.,scaledwidth=50.0% ] to discuss the uncertainty caused by @xmath8 , we take two groups of da moments that have been obtained in refs.@xcite to determine the coefficients @xmath126 , @xmath127 and @xmath128 , where the moments in ref.@xcite are derived by using the qcd light - cone sum rules and the moments in ref.@xcite are derived based on the effective chiral action from the instanton : @xmath135 : & & \langle x^0\rangle^k_p=1,\;\;\langle x^1\rangle^k_p=0.06124,\;\ ; \langle x^2\rangle^k_p=0.36757 , \\
{ \rm group \;\;2\;\;\;\;}[27 ] : & & \langle x^0\rangle^k_p=1,\;\;\langle x^1\rangle^k_p=0.00678,\;\ ; \langle x^2\rangle^k_p=0.35162.\end{aligned}\ ] ] here the moments are defined as @xmath136 with @xmath137 .
it should be noted that the moments defined in ref.@xcite are for @xmath138 other than @xmath129 , since in these references @xmath103 stands for the momentum fraction of @xmath84-quark in the kaon ( @xmath139 ) , while in the present paper @xmath103 stands for the momentum fraction of the light quark @xmath105 in the kaon ( @xmath106 ) . taking the above two groups of da moments for @xmath140 , the parameters of @xmath117 can be determined as , @xmath141 the distribution amplitudes for these two group of parameters are shown in fig.([daphip ] ) , where @xmath131 is determined by group 1 parameters and @xmath132 is determined by group 2 parameters respectively . for comparison
, we also draw the distributions derived in ref.@xcite in fig.([daphip ] ) , i.e. @xmath133 stands for the da obtained in ref.@xcite and @xmath134 stands for that of ref.@xcite .
one may observe that different from @xmath133 and @xmath134 , both @xmath131 and @xmath132 are double humped curves and are highly suppressed in the endpoint region .
such feature is necessary to suppress the endpoint singularity coming from the hard - scattering kernel and then to derive a more reasonable results for the twist-3 contributions to the @xmath1 form factor .
it is more convenient to transform the kaon wave functions into the compact parameter @xmath142-space , which can be done with the help of the fourier transformation @xmath143 where @xmath144 stands for @xmath27 , @xmath8 and @xmath145 , respectively .
the upper edge of the integration @xmath146 is necessary to ensure that the wave function is soft enough @xcite . after doing the fourier transformation
, we obtain the kaonic wave functions in the compact parameter @xmath142-space : @xmath147 \exp \left[-\beta_k^2\left(\frac{m_s^2}{1-x}+ \frac{m_q^2 } { x}\right ) \right ] \nonumber\\ & & \times\int_0^{1/b_k } \exp\left(\frac{-\beta_k^2 k_\perp^2 } { x(1-x)}\right)j_0(b_k k_\perp ) k_\perp dk_\perp \;,\nonumber\\ \psi_{\sigma}(x , b_k)&=&2\pi a_{\sigma } \exp \left[-\beta_k^2\left(\frac{m_s^2}{1-x}+ \frac{m_q^2 } { x}\right ) \right ] \int_0^{1/b_k } \exp\left(\frac{-\beta_k^2 k_\perp^2 } { x(1-x)}\right)j_0(b_k k_\perp ) k_\perp dk_\perp\nonumber\end{aligned}\ ] ] and @xmath148 \exp \left[-\beta_k^2\left(\frac{m_s^2}{1-x}+ \frac{m_q^2 } { x}\right ) \right]\nonumber\\ & & \times \int_0^{1/b_k } \exp\left(\frac{-\beta_k^2 k_\perp^2 } { x(1-x)}\right)j_0(b_k k_\perp ) k_\perp dk_\perp \;. \nonumber\end{aligned}\ ] ]
in the numerical calculations , we adopt @xmath149 as for the phenominological parameter @xmath46 , which is a scale characterized by the chiral perturbation theory , we take its value to be @xmath150 gev . in the following , we first discuss the properties of @xmath151 and @xmath33 that are calculated up to @xmath31 , i.e. to show how @xmath32 and @xmath152 are affected by the b - meson wave function and the kaonic wave functions .
the b - meson wave functions @xmath3 and @xmath4 up to next - to - leading order fock state expansion depend on two phenomenological parameters @xmath13 and @xmath14 .
an estimate of @xmath13 using qcd sum rule approach gives @xmath153 @xcite . by comparing the pqcd results of the @xmath11 form factor with the qcd lcsr results and the lattice qcd calculations
, ref.@xcite shows that @xmath154 . as for the value of @xmath14
, it has been pointed out that if the contribution from the b - meson three - particle wave function is limited to be within @xmath155 of that of the ww - like wave function within the energy region of @xmath156 $ ] , then the value of @xmath14 should be restricted within the region of @xmath157 $ ] @xcite .
for clarity , we take the same regions as obtained from the @xmath158 case @xcite for both @xmath13 and @xmath14 , i.e. @xmath159 { \rm gev}$ ] and @xmath160 $ ] , to study the form factors @xmath32 and @xmath152 in the large and intermediate energy regions . furthermore , according to the discussion in the last section , the remaining uncertainty of the kaonic twist-2 wave function @xmath27 is caused by the value of @xmath29 , cf .
eq.([psika1 ] ) .
there we take @xmath161 @xcite to do our discussion . as for the twist-3 wave function @xmath8
, we take two groups of parameters as shown in eqs.([group1],[group2 ] ) to do the calculation .
next , we compare the @xmath16 result of the form factor with the leading order one that is of order @xmath17 and is calculated by using the ww - like b - meson wave function , and also make a comparison with the lcsr results of ref.@xcite in the large and intermediate energy regions . through comparison , preferable values for the undetermined parameters can be found .
the @xmath1 transition form factors @xmath162 and @xmath163 have been studied within the framework of qcd lcsr @xcite , especially at @xmath37 , it shows @xmath164 , \ ] ] e.g. when @xmath165 , @xmath166 .
more generally , @xmath162 and @xmath167 can be parameterized in the following form @xcite : @xmath168 where @xmath169 contains the contributions to the form factor from the asymptotic da and all higher - twist effects from three - particle quark - quark - gluon matrix elements , @xmath170 contains the contribution from the higher gegenbauer term of da that is proportional to @xmath91 , @xmath171 and @xmath172 respectively . here the factorization scale @xmath173 should be taken as @xmath174 , since the functions @xmath175 are determined with @xmath176 @xcite .
the explicit expressions of @xmath175 can be found in table v and table ix of ref.@xcite .
for the gegenbauer moments @xmath177 and @xmath178 , we take their preferred values : @xmath179 and @xmath180 @xcite . while for @xmath181
, it equals to @xmath182 with the help of qcd evolution .
transition form factors @xmath32 ( left ) and @xmath33 ( right ) with @xmath183 and @xmath165 .
the dash - dot line , the dashed line and the dotted line stand for @xmath184 gev , @xmath185 gev and @xmath186 gev respectively . for comparison ,
the solid line comes from the qcd lcsr result as shown in eq.([sumruleapp ] ) and the fuscous shaded band shows its theoretical error @xmath187.,title="fig:",scaledwidth=50.0% ] transition form factors @xmath32 ( left ) and @xmath33 ( right ) with @xmath183 and @xmath165 .
the dash - dot line , the dashed line and the dotted line stand for @xmath184 gev , @xmath185 gev and @xmath186 gev respectively . for comparison ,
the solid line comes from the qcd lcsr result as shown in eq.([sumruleapp ] ) and the fuscous shaded band shows its theoretical error @xmath187.,title="fig:",scaledwidth=50.0% ] transition form factors @xmath32 ( left ) and @xmath33 ( right ) with @xmath188 gev and @xmath165 .
the dotted line , the dashed line and the dash - dot line stand for @xmath189 , @xmath190 and @xmath191 respectively . for comparison ,
the solid line comes from the qcd lcsr as shown in eq.([sumruleapp ] ) and the fuscous shaded band shows its theoretical error @xmath187.,title="fig:",scaledwidth=50.0% ] transition form factors @xmath32 ( left ) and @xmath33 ( right ) with @xmath188 gev and @xmath165 . the dotted line , the dashed line and the dash - dot line stand for @xmath189 , @xmath190 and @xmath191 respectively . for comparison ,
the solid line comes from the qcd lcsr as shown in eq.([sumruleapp ] ) and the fuscous shaded band shows its theoretical error @xmath187.,title="fig:",scaledwidth=50.0% ] first , we discuss the properties of @xmath32 and @xmath33 caused by the b - meson wave function . for such purpose , we fix the kaonic wave functions by setting @xmath192 and by using the group 1 parameters for @xmath8 .
we show the @xmath1 transition form factors @xmath32 and @xmath33 with @xmath193 in fig.([lambda ] ) , where @xmath13 varies within the region of @xmath194 $ ] . for comparison , we show the qcd lcsr result with @xmath165 and its theoretical error ( @xmath195 ) @xcite by a fuscous shaded band in fig.([lambda ] ) .
the results show that the @xmath1 transition form factors will decrease with the increment of @xmath13 . and
the best fit of the qcd lcsr result at @xmath37 shows that @xmath196 .
moreover , we show @xmath32 and @xmath33 with @xmath197 gev in fig.([delta ] ) , where @xmath14 varies within the region of @xmath157 $ ] .
the results show that the @xmath1 transition form factors will increase with the increment of @xmath14 .
it can be found that when setting @xmath165 , and by varying @xmath14 within the region of @xmath157 $ ] and @xmath13 within the region of @xmath198 $ ] , @xmath199 runs within the region of @xmath200 $ ] .
since the best agreement between the pqcd result and the qcd lcsr result at @xmath37 is obtained around @xmath201 and @xmath202 , we shall always take @xmath203 and @xmath204 to do our following calculations if not specially stated .
transition form factors @xmath32 ( left ) and @xmath33 ( right ) with @xmath188 gev and @xmath183 .
the dotted line , the dashed line and the dash - dot line stand for @xmath205 , @xmath206 and @xmath207 respectively.,title="fig:",scaledwidth=50.0% ] transition form factors @xmath32 ( left ) and @xmath33 ( right ) with @xmath188 gev and @xmath183 .
the dotted line , the dashed line and the dash - dot line stand for @xmath205 , @xmath206 and @xmath207 respectively.,title="fig:",scaledwidth=50.0% ] transition form factors @xmath32 and @xmath33 with @xmath208 , @xmath183 and @xmath165 . the dash - dot and the dashed lines are for @xmath8 with group 1 parameters eq.([group1 ] ) , group 2 parameters eq.([group2 ] ) respectively . for comparison ,
the solid lines come from the qcd lcsr with @xmath165 @xcite .
, width=278 ] second , we discuss the properties of @xmath32 and @xmath33 caused by the twist-2 wave function @xmath27 , i.e. by the value of @xmath29 . for such purpose
, we fix the b - meson wave functions by setting @xmath204 and @xmath203 and by using the group 1 parameters for @xmath8 .
we show the @xmath1 transition form factors @xmath151 and @xmath33 in fig.([a1k ] ) with @xmath209 , @xmath206 and @xmath207 respectively . it can be found that the form factors shall be increased with the increment of @xmath29 , which agree with the observation of ref.@xcite . furthermore ,
since the contribution from @xmath8 is sizable to that of @xmath27 , it is necessary to make a discussion on its uncertainty to the @xmath1 transition form factor .
fig.([twistp ] ) shows @xmath32 and @xmath152 with two groups of parameters for @xmath8 .
the results are very close to each other due to the close shape of their @xmath140 as shown in fig.([daphip ] ) , e.g. around the region of @xmath210 the difference between them is less than @xmath211 .
so by taking proper transverse momentum dependence for the wave function @xmath8 , where we have taken the bhl prescription for its transverse momentum dependence , the uncertainties from its distribution amplitude @xmath140 can be reduced .
transition form factor @xmath32 with fixed @xmath212 , @xmath183 and @xmath165 .
the left diagram is for the different kaon twist structures , @xmath27 , @xmath8 and @xmath145 .
the right diagram is for the different b meson structures , @xmath3 , @xmath4 and @xmath213.,title="fig:",width=278 ] ( a ) transition form factor @xmath32 with fixed @xmath212 , @xmath183 and @xmath165 .
the left diagram is for the different kaon twist structures , @xmath27 , @xmath8 and @xmath145 .
the right diagram is for the different b meson structures , @xmath3 , @xmath4 and @xmath213.,title="fig:",width=278 ] ( b ) finally , in order to get a deep understanding of the @xmath1 transition form factor , we discuss the contributions from different parts of the b - meson wave function or the kaon wave function , correspondingly .
here we take @xmath32 to do our discussions and the case of @xmath33 can be done in a similar way . for convenience ,
we set @xmath203 , @xmath204 , @xmath165 and by using the group 1 parameters for @xmath8 . when discussing the contribution from one of the kaon wave function structures , the contribution from all the b - meson wave function structures
are summed up , and vice versa .
fig.([kaonwave]a ) shows the contributions from the different twist structures of the kaon wave function , i.e. @xmath27 , @xmath8 and @xmath145 ( the contributions from the terms involving @xmath214 are included in @xmath145 ) , respectively .
one may observe that the contribution from @xmath8 is comparable to that of @xmath27 , e.g. its contribution is about @xmath215 of that of @xmath27 at @xmath216 , and the contribution from @xmath145 is small .
fig.([kaonwave]b ) presents the contributions from @xmath3 , @xmath4 and @xmath213 respectively .
it can be found that the contribution from @xmath4 is about @xmath217 in comparison to that of @xmath3 in the region of @xmath218 $ ] , while the contribution from @xmath213 is negligible in comparison to that of @xmath3 and @xmath4 .
so the contribution from @xmath4 should be included for a consistent estimation to the next leading order . as a comparison
, it can be found that under the leading order estimation the contribution from @xmath4 is only about @xmath219 in comparison to that of @xmath3 at @xmath37 , which agree with the rough order estimation that the contribution from @xmath4 is of order @xmath17 .
so as to the leading order estimation @xmath17 , the contribution from @xmath4 is usually neglected in the literature .
such difference of @xmath4 s contribution between the leading order estimation and the next - to - leading order estimation is mainly due to the fact that the transverse momentum dependence of the b - meson wave functions are merely a delta function under the ww - approximation ( the leading - order estimation ) , while it shall be broadened to a certain degree according to the value of @xmath14 by taking into account the 3-particle fock states contributions ( the next - to - leading order estimation ) , cf .
fig.(2 ) of ref.@xcite .
so qualitatively , the contributions from @xmath220 shall be raised to a certain degree for the next - to - leading order case , due to the less suppression of the end - point region ( @xmath221 ) from the transverse momentum distributions than that of the leading order case .
and then the naive order estimation for the contribution of @xmath4 is no longer correct , and the contributions from @xmath3 and @xmath4 are both important in the next - to - leading order calculation .
the ww - like b - meson wave functions in the compact parameter @xmath64-space can be found in ref.@xcite .
taking the ww - like wave functions and cutting off the power suppressed terms in the hard scattering amplitude , we can obtain the leading order results ( @xmath17 ) for the form factors @xmath32 and @xmath33 .
strictly , one should cut off the contribution from @xmath4 to obtain the leading order estimation , since @xmath4 is power suppressed in comparison to @xmath3 .
however for easy comparison with the results in the literature , e.g. ref.@xcite , we keep @xmath4 in the leading order estimation . for convenience , we take @xmath203 , @xmath204 , @xmath165 and by using the group 1 parameters for the wave function @xmath8 to do a comparison of the leading order results with the total results that include the contributions up to order @xmath16 . it can be found that the leading order results are smaller than the total results by about @xmath222 in the large recoil region , e.g. at @xmath37 , the leading order @xmath223 .
one may observe that a larger leading order estimation has been obtained in ref.@xcite , which shows @xmath224 .
we argue that the present leading order estimation is more reliable , and the larger value of @xmath225 derived in ref.@xcite is mainly due to the following two reasons : 1 ) even though the sudakov and threshold resummation factors shall kill the endpoint singularity of the process @xcite , the transverse momentum dependence of kaonic wave functions are still important to give a more reliable pqcd estimation , which is similar to the cases of @xmath226 form factor @xcite and the pion electromagnetic form factor @xcite . in ref.@xcite
the transverse momentum dependence of kaonic wave functions are lacking , i.e. the distribution amplitude other than the wave function is used . while in our present calculation , the bhl - prescription is adopted for the kaonic transverse momentum dependence . as for the wave function @xmath227 , such transverse momentum dependence will results in a double humped da @xmath140 as shown in fig.([daphip ] ) and then it shall give more effective suppression in the end - point region than the one used in ref.@xcite .
in fact , it can be found that the contributions from the end point region shall always be overestimated without taking the transverse momentum into the twist-3 wave function @xmath227 .
can be found in ref.@xcite . ] furthermore , by taking proper transverse momentum dependence for the wave function @xmath8 , the uncertainties from its distribution amplitude @xmath140 can be reduced as has been discussed in sec.iii.a ; 2 ) the distribution amplitude of @xmath27 with a much bigger value of @xmath228 , i.e. @xmath229 , is adopted by ref.@xcite . since the form factors increases with the increment of @xmath29 , a larger value of @xmath228 shall increase the form factors .
furthermore , by varying @xmath13 within the region of @xmath230 $ ] , the uncertainty caused by @xmath13 is the biggest and is of order @xmath26 .
while by varying @xmath14 within the region of @xmath157 $ ] , the uncertainty caused by @xmath14 is smaller and are of order @xmath231 .
this can be qualitatively explained as that @xmath13 is the characteristic parameter that determines the leading fock - state behavior of the @xmath6-meson wave functions , while @xmath14 is the characteristic parameter that determines the higher fock - state s behavior of the @xmath6-meson wave functions .
the uncertainties from @xmath91 and @xmath27 are less than @xmath232 in the large recoil region .
the @xmath1 transition form factor have been analyzed by several groups under the qcd lcsr approach @xcite .
new sum rule for @xmath1 is derived from the correlation functions expanded near the light cone in terms of b - meson distributions @xcite , in which the contributions of the quark - antiquark and quark - antiquark - gluon components in the b - meson are taken into account .
it has been found that the @xmath1 transition form factor in the large recoil region does not receive contributions from the 3-particle b - meson da s .
one may observe that if substituting the b - meson das , which are derived by doing the integration over @xmath64 in eqs.([newmodel1],[newmodel2 ] ) , into the formulae of ref.@xcite , then one can obtain the same results as that of ref.@xcite , since our b - meson das are close to the exponential model wave functions adopted in ref.@xcite .
furthermore , one may observe that the result of @xmath233 under the condition of @xmath234 agrees well with our present pqcd estimation .
secondly , a systematic qcd lcsr calculation has been done in ref.@xcite by including the one - loop radiative corrections to the twist-2 and twist-3 contributions , and leading - order twist-4 corrections .
some comparison of their results with our present one can be found in figs.([lambda],[delta],[a1k ] ) , which also shows a good agreement within reasonable errors .
for example , from eq.([sumruleapp ] ) it can be found that the uncertainty of form factor caused by @xmath29 within the region of @xmath112 $ ] is less than @xmath235 , which is consistent with our present result as shown in fig.([a1k ] ) .
in this paper , we have examined the @xmath1 transition form factor in the pqcd approach up to order @xmath16 , where the transverse momentum dependence for the wave function , the sudakov effects and the threshold effects are included to regulate the endpoint singularity and to derive a more reasonable result .
we have confirmed that the pqcd approach can be applied to calculate the @xmath1 transition form factor in the large recoil regions .
we emphasize that the transverse momentum dependence for both the b meson and the kaon is important to give a better understanding of the @xmath1 transition form factor .
fig.([kaonwave]a ) shows that the contribution from the pionic twist-3 wave function @xmath8 is sizable in comparison to that of @xmath27 , and the contribution from @xmath145 is small .
while fig.([kaonwave]b ) shows that by using the b - meson wave functions up to next - to - leading order fock state expansion , the contributions from @xmath3 and @xmath4 are important . in refs.@xcite , we have shown that the results from the pqcd approach , the lattice qcd approach and the qcd lcsrs are complementary to each other and by combining the results of those three approaches , one can obtain an understanding of the @xmath11 transition form factor in the whole physical regions . and the best fit of the pqcd results with that of the qcd lcsr results in the large recoil region can be obtained by taking @xmath236 $ ] and @xmath160 $ ] @xcite . in the present paper , we show that within the regions of @xmath237 $ ] , @xmath160 $ ] and @xmath238 $ ] , the pqcd results on the @xmath1 form factor in the large recoil region also agree well with that of the qcd lcsr results @xcite .
our present pqcd results in some sense is more reliable than the lcsr calculations due to the fact that by taking the transverse momentum dependence properly for the wave functions the soft endpoint singularity have been effectively suppressed , e.g. as is shown in sec.iii.a the difference between the two models for @xmath8 is less than @xmath211 in the large recoil region , while for the lcsr approach large uncertainty comes from the kaonic twist-3
da @xmath140 that is not too well - known . by running the parameters within the above regions , we obtain @xmath9 .
finally , to illustrate the @xmath7-breaking effects , we calculated the ratio with the help of the @xmath226 results in ref.@xcite : @xmath239=1.13\pm0.02 $ ] , which favors a small @xmath7-breaking effects .
this work was supported in part by the natural science foundation of china ( nsfc ) and by the grant from chongqing university .
this work was also partly supported by the natural science foundation of chongqing under the grant no .
8562 and the national basic research programme of china under grant no .
we thank prof.c.d .
lu for helpful comments .
+ s.j .
brodsky , t. huang and g.p .
lepage , in _ particles and fields-2 _ , proceedings of the banff summer institute , banff , alberta , 1981 , edited by a.z .
capri and a.n .
kamal ( plenum , new york , 1983 ) , p143 ; g.p .
lepage , s.j .
brodskyk t.huang , and p.b .
mackenize , _ ibid .
_ , p83 ; t. huang , _ in proceedings of xxth international conference on high energy physics _ , madison , wisconsin , 1980 , edited by l.durand and l.g .
pondrom , aip conf.proc.no .
69(aip , new york , 1981 ) , p1000 . h.c .
pauli and h.a .
bethe , phys.rev.d 32 , 1993(1985 ) ; * 32 * , 2001(1985 ) ; t. eller , h.c .
pauli and s.j .
brodsky , _ ibid .
_ * 35 * , 1493(1987 ) ; s.j .
brodsky , h.c .
pauli and s.s .
pinsky , phys.rep.*301 * , 299(1998 ) . | in the paper , we apply the @xmath0 factorization approach to deal with the @xmath1 transition form factor @xmath2 in the large recoil regions .
the b - meson wave functions @xmath3 and @xmath4 that include the three - particle fock states contributions are adopted to give a consistent pqcd analysis of the form factor up to @xmath5 .
it has been found that both the wave functions @xmath3 and @xmath4 can give sizable contributions to the form factor and should be kept for a better understanding of the @xmath6 meson decays .
then the contributions from different twist structures of the kaon wavefunction are discussed , including the @xmath7-breaking effects . a sizable contribution from the twist-3 wave function @xmath8
is found , whose model dependence is discussed by taking two group of parameters that are determined by different distribution amplitude moments obtained in the literature . it is also shown that @xmath9 and @xmath10=1.13\pm0.02 $ ] , which are more reasonable and consistent with the light - cone sum rule results in the large recoil regions . + * pacs numbers : * 12.38.aw , 12.38.bx , 13.20.he , 14.40.aq # 1#1 # 1 _ |
as was first observed by oso-3 ( @xcite ) , the dominant feature of the high - energy sky is the intense emission from the galactic plane .
later the complete sas-2 ( @xcite ) and cos - b ( @xcite ) data gave evidence for a correlation between the emission and the spatial structures of the galaxy .
the intensity distribution and the spectral form of the emission have led to the consensus that the diffuse radiation is primarily produced by interactions between galactic cosmic ray particles and the interstellar medium , and to a small extent by unresolved galactic point sources ( @xcite ) .
the egret observations of the magellanic clouds have shown that cosmic ray nucleons in the energy range below 100 gev are almost certainly galactic ( @xcite ) , while the observations made with the osse and comptel instruments aboard cgro have provided strong evidence that cosmic ray electrons are galactic ( @xcite , see also @xcite ) .
therefore the diffuse galactic emission tells us about the propagation of cosmic rays from their sources to the interaction regions and thus complements the direct particle measurements by balloon and satellite experiments . the greater sensitivity and spatial and energy resolution of egret compared with sas-2 and cos - b permit a much more detailed analysis of the diffuse galactic emission than was possible with the earlier experiments .
the spatial and spectral distribution of the diffuse emission within @xmath0 of the galactic plane have recently been compared with a model calculation of this emission which is based on realistic interstellar matter and photon distributions and dynamical balance ( @xcite ) , i.e. cosmic rays having the same spectrum and composition everywhere in the galaxy and having an intensity which follows the surface density of thermal gas convolved with a gaussian with dispersion @xmath1 ( @xcite ) . the distribution of the total intensity above 100 mev agrees surprisingly well with the model predictions . however , at higher energies above 1 gev the model systematically underpredicts the intensity .
if the model is scaled up by a factor 1.6 , the model prediction and the observed intensity above 1 gev agree well .
thus the model displays a deficit of @xmath238% of the total observed emission which depends , if at all , only weakly on location . at energies above 1 gev around 90% of the model intensity
is due to @xmath3-decay , i.e. hadronic processes , and only 10% is due to interactions of electrons .
there are a number of possible explanations for this deficit : a miscalibration of egret could cause an overestimation of the intensity above 1 gev .
this possibility is highly unlikely .
point sources generally show power - law spectra without spectral hardening above 1 gev .
it would require an extreme level of cosmic conspiracy for a calibration error to mimic a general power - law behavior in the spectra of cosmic sources .
the kinematics of @xmath3 production may be poorly understood .
detailed monte - carlo calculations have shown ( @xcite ) , that models based on the current knowledge of particle interactions do not give very different results for the @xmath3 spectra than do simple isobar plus scaling descriptions ( @xcite ) .
it is unlikely that the cosmic ray nucleon spectrum in the solar vicinity is softer than that elsewhere in the galaxy .
the local cosmic ray spectrum samples sources within a few kpc in distance and a few times 10@xmath4 years in time .
since the observed deficit appears to be independent of galactic longitude , the sources of cosmic rays within a few kpc from the sun would have to be different from those in the inner galaxy and those in the outer galaxy .
we have also tested and verified that the uncertainties in the local interstellar cosmic ray spectrum below a few gev are by far not sufficient to account for the deficit .
the uncertainties arising from our limited knowledge of the cosmic ray nucleon spectrum and the nucleon - nucleon interaction kinematics can be estimated to be on the order of a few percent .
there may be unresolved point sources which contribute strongly at higher energies .
the only known class of objects with appropriate spectra is pulsars .
based on the properties of the six identified pulsars it has been found that unresolved pulsars would indeed contribute mainly between 1 gev and 10 gev ( @xcite ) .
however , to account for all the deficit it is required that more than 30 pulsars be detectable by egret as point sources .
this can be compared with less than 10 unidentified sources which are not variable ( @xcite ) and which show pulsar - like spectra ( @xcite ) .
also , the latitude distribution of the emission from unresolved pulsars is inconsistent with that of the observed emission .
unresolved pulsars will contribute 6 - 10% of the observed intensity above 1 gev and around 3% in the energy band between 100 mev and 1 gev ( @xcite ) , and thus they can account only for a small fraction of the high - energy deficit . all in all the effects described above can account only for a small fraction of the deficit , or can add only small systematic uncertainties . in this paper
we will investigate whether the remaining deficit of 30 - 35% may be caused by inverse compton emission of high - energy electrons .
leptonic processes contribute only around 10% of the intensity above 1 gev in the model of hunter et al .
( 1997 ) , which corresponds to @xmath56% of the total observed intensity .
thus the leptonic contribution would have to be increased to 35 - 40% of the total observed emission to explain all the deficit .
the galactic distribution of cosmic ray electrons is intimately linked to that of their sources .
the recently - found evidence of x - ray synchrotron radiation from the four supernova remnants sn1006 ( @xcite ) , rx j1713.7 - 3946 ( @xcite ) , ic443 ( @xcite ) , and cas a ( @xcite ) supports the hypothesis that galactic cosmic ray electrons are accelerated predominantly in snr .
x - ray synchrotron radiation implies tev emission from the comptonization of the microwave background at a flux level which depends only on the average magnetic field strength ( @xcite ) , and indeed the detection of the remnant sn1006 at tev energies has been announced recently ( @xcite ) .
interestingly , there is no clear observational proof that the nuclear component of cosmic rays is accelerated likewise in snr .
the acceleration of cosmic ray nucleons in snr should lead to observable flux levels at tev energies ( @xcite ) , however with a spectrum different from that of the leptonic emission .
the generally tight upper limits for tev emission from the nearest snr ( @xcite ) are in conflict with simple shock acceleration models for cosmic ray nucleons in snr .
most of the radio synchrotron spectra of snr can be well represented by power - laws with indices around @xmath6 , corresponding to electron injection indices of @xmath7 ( @xcite ) .
this is in accord with predictions based on models of particle acceleration ( @xcite ) .
however , it is different from the electron injection spectral index of @xmath8 which has been inferred from the locally observed electron spectrum ( @xcite ) and which subsequently has been used in the model of galactic emission of hunter et al .
the contribution of cosmic ray electrons to the galactic spectrum at high energies depends strongly on their injection spectral index .
if the acceleration cut - off energy is high enough , the leptonic emission may even dominate at tev - pev energies ( @xcite ) .
a change in the electron injection index by @xmath9 could increase the inverse compton emissivities at a few gev by an order of magnitude or more .
let us suppose that for some reason the local cosmic ray electron spectrum is different from the average electron spectrum in the galaxy .
then the following scenario appears viable : the bulk of cosmic ray electrons is accelerated in snr with an injection index around @xmath7 .
the leptonic emission at a few gev would be much stronger than in the hunter et al .
model , and it may explain a substantial fraction of the discrepancy between their model and the observed spectra .
so if we would find a mechanism or an effect which would cause the local electron spectrum to be different from the galactic average , we may in a second step reassess the spectra produced by galactic cosmic rays without having to assume an electron injection index of @xmath8 .
it has been noted before that the spatial distribution of cosmic ray sources affects the locally observable spectra ( @xcite ) .
as far as electron acceleration in snr is concerned , there is no evidence that the star formation activity and thus the snr production rate in the solar vicinity is significantly less than the galactic average . in the next section
we will show that for cosmic ray electrons , unlike nucleons , the local spectra above a certain energy can deviate from the galactic average , even if the spatial distribution of cosmic ray accelerating snr is homogeneous .
this is a result of the discrete nature of snr both in space and in time .
we will use this finding in the third section to model the high - energy excess as result of inverse compton emission of cosmic ray electrons , albeit with a harder injection spectrum than conventionally assumed .
the spectrum of cosmic ray electrons in terms of the contributions of discrete sources like supernova remnants ( snr ) has been discussed before ( @xcite ) .
these authors have investigated the case of continuously active sources and have concluded that one needs sources situated within a few hundred parsecs of the solar system , in order that the energy losses of electrons do not induce a cutoff in the energy spectrum . since the required number of active sources exceeds that of supernova remnants by an order of magnitude , snr were found unlikely to be the only source of cosmic ray electrons between 1 gev and 1 tev . in that paper
the diffusion coefficient had been assumed independent of energy . with the usual energy dependence @xmath10 some of the statements in the cowsik and lee paper
would have to be relaxed . in this section
we will also consider the finite lifetime of snr or other possible cosmic ray accelerators together with the random distribution of snr in space and time .
the latter induces a time dependence in the local electron spectrum at higher energies , which stems from the poisson fluctuations in the number of snr within a certain distance and within a certain time interval .
as we will see , the discreteness of sources does not simply cause a cutoff in the electron spectrum , but makes it time variable and thus unpredictable beyond a certain energy .
since effects of the discreteness of sources show up only at higher particle energy , we may describe the propagation of cosmic ray electrons at energies of 1 gev to 1 tev by a simplified transport equation @xmath11 where we consider continous energy losses by synchrotron radiation and inverse compton scattering , an energy - dependent diffusion coefficient @xmath12 , and a source term @xmath13 .
the green s function for this problem can be found in the literature ( @xcite ) .
@xmath14 where @xmath15 in case of discrete sources the injection term @xmath13 is a sum over all sources . for an individual source showing up at time @xmath16 and injecting for a time period @xmath17 we can write @xmath18 without loss of generality we can set @xmath19 and obtain @xmath20 where @xmath21 and @xmath22 is the distance between source and observer .
@xmath23 is the contribution to the local electron spectrum provided by a single source ( snr ) at distance @xmath22 which is ( was ) injecting electrons for a time period @xmath17 starting at @xmath16 .
the local spectrum of electrons can now be obtained by summing the contributions from all individual sources . in our case
the distribution of snr is , for ease of exposition and computation , assumed to be a homogeneous disk of radius @xmath24kpc and half - thickness @xmath25kpc .
other choices for the spatial distribution of snr do not impose serious changes in the results , as long as the distribution is not structured on sub - kpc scales .
the numerical procedure is as follows : for each volume element @xmath26 the expected number of snr injecting cosmic rays within the time interval @xmath27 is @xmath28 where @xmath29 is the total volume of the source distribution and @xmath30years is the inverse of the supernova rate ( @xcite ) .
@xmath27 has to be the maximum lookback time in our model which is the sum of the snr lifetime @xmath17 and the electron energy loss time @xmath31 at 1 gev , the lowest energy considered here . at a given time
the actual number of snr in that volume element is a poissonian random number with mean @xmath32 , and each of the snr has a birth date which is a random number uniformly distributed within @xmath33 $ ] .
the final electron spectrum is then derived by summing over the contributions of the individual snr per volume element and summing over all relevant volume elements .
we calculate 400 such ` random ' spectra and thus derive the distribution of possible spectra and their spread . in fig.1
we show the resulting range of local electron spectra compared with the observed spectra ( @xcite ) .
the energy density of the ambient photon fields plus that of the perpendicular component of the magnetic field strength is in total taken to be 3.5 ev/@xmath34 .
the changes in the compton cross section in the klein - nishima regime of optical and near - infrared photon fields are neglected .
the diffusion coefficient is @xmath35 at 1 gev and increases with energy to the power a=0.6 , in accord with results for a two - dimensial diffusion model fit to the local spectra of 13 primary and secondary cosmic ray nuclei at rigidities between 1 gv and 10@xmath36 gv ( @xcite ) . at these energies electrons and nuclei
will scatter off the same turbulence , except for the helicity , and thus their diffusion behaviour may be expected to be similar .
the injection spectral index of electrons is s=2.0 , which corresponds to the mean synchrotron spectral index of individual snr ( @xcite ) .
the solar modulation of electrons at a few gev energy has a strong effect on the observed spectrum .
note the difference between the spectrum observed when the modulation level was high ( @xcite ) and the spectrum observed during the passage of ulysses over the solar south pole ( @xcite ) .
we have crudely approximated the effect of solar modulation using the force - field approach ( @xcite ) with @xmath37 .
thus we probably overestimate the modulation in case of the ulysses data and underestimate in case of the golden et al . data .
the spectra at a given time are not necessarily smooth .
there is no preference for the usual broken power - laws or power - laws with exponential cut - offs .
in fact the individual spectra are bumpy above @xmath2 50 gev and some display step - like features . as an example we show a particular spectrum as dash - dotted line in fig.1 , which is slightly on the low side between 10 gev and 100 gev , where it suffers further softening before it abruptly hardens at 300 gev .
below 10 gev the local electron spectrum is well determined . between 10 gev and 100 gev
it varies with time by a factor of 2 or 3 , and above 100 gev is completely unpredictable .
changes in the absolute numbers for the diffusion coefficient and the radiative energy losses do not change the basic behaviour , but can shift the transition between weak and strong variability to lower or higher energies . if the energy dependence of the diffusion coefficient is weaker , i.e. @xmath38 , the transition between weak and strong variability will be faster , and vice versa ( a slower transition for higher powers than a=0.6 ) . for comparison
we have indicated the result for an energy dependence of the diffusion coefficient , @xmath39 .
as shown in fig.1 , the high - energy data for the local electron flux are in accord with an injection spectral index of s=2.0 , though in a model with steady injection and a smooth source distribution these data would require an injection index of around 2.4 ( @xcite ) .
concerning the distribution of high - energy electrons , the galaxy would look like swiss cheese , with holes and regions of higher density . in the line - of - sight integrals , which are relevant for comparison with the egret data
, the averaging over holes and high - density regions will give the same result as a model with steady injection , however with source index of 2.0 instead of 2.4 . at higher latitudes
the line - of - sight will be so short that regions of low or high electron density will be resolved .
the leptonic spectra in direction of the galactic poles should be relatively soft since the line - of - sight integral of the emissivity will be dominated by the soft local spectrum .
the absolute electron flux is reproduced if each snr provides an energy input of @xmath40 in the form of electrons , which is one thousandth of the canonical value of @xmath41 for the kinetic energy input per supernova .
taken over a lifetime of @xmath42 the corresponding power of @xmath43 is less than the x - ray luminosity of sn1006 alone .
the time variability of the high energy cosmic ray spectrum will not be related or even be synchronous to variability in the flux of low - energy cosmic ray nucleons , which can be traced by cosmogenic instable isotopes in sediments ( @xcite ) or meteorites ( @xcite ) . a few notes should be added .
we have taken supernova explosions to be completely independent of each other .
one may expect some level of correlation in ob associations and snobs , which would make the basic effect of time dependence even more dramatic , since the ob associations and snobs would act as single sources with longer lifetime , but much smaller frequency of occurrence .
another important point is that we have assumed that all electron sources produce the same spectrum . in reality
this does nt need to be the case .
some snr will produce electrons with harder spectrum , and another group of snr will provide softer spectra . it may be that the spectral form depends on the age of an snr .
in fact the radio data show that snr do have different synchrotron spectra ( @xcite ) .
if we take the electron injection spectral index of an individual snr not as a fixed number , but a random variable following some probability function , the time - averaged spectrum ( the dotted line in fig.[acct3 ] ) will get a positive curvature .
the level of time variability on the other hand will get larger .
the shaded regions in fig.1 , in which the spectrum is contained for 68% , respectively 95% , of the time , extend beyond those for fixed injection index . a final note concerns secondary positrons and electrons .
these particles are generated subsequent to interactions of cosmic ray nucleons with ambient gas , and hence the effect discussed does not apply and the local spectrum of secondary electrons will not vary . thus the observed positron fraction will also exhibit variability anticorrelated with that of the primary electron spectrum .
if we are indeed living in a hole in the distribution of high - energy electrons then the positron fraction above , say , 20 gev will be above the level expected in steady injection models , if the gas density within @xmath441 kpc from the sun is not also sub - average .
this may explain the observed positron fraction in that energy range , which is indeed slightly above the model predictions ( @xcite ) .
we have seen that the discreteness of sources of cosmic ray electrons causes a strong variability of the local electron spectrum at higher energies .
therefore the high energy electron spectrum does not prescribe our choice of electron injection spectrum in propagation models .
if we consider emission in the galactic plane , the line - of - sight integral of the emissivity will correspond to an averaging over the different variability states , and hence the intensity calculated with time - dependent models will not differ significantly from the results of steady - state models .
the latter are much easier to compute and thus preferrable , but they will in general not be able to reproduce the local high energy electron spectrum correctly .
on the other hand , in the absence of reliable data on the position and age of all nearby snr it is also impossible to calculate the time - dependent local electron spectrum precisely , we can only infer the level of variability .
therefore we feel that , if the acceleration of electrons occurs predominantly in snr , the emission in the galactic plane can be sufficiently well described with steady - state models , provided the model is * not required to fit the local electron data above @xmath4430 gev*. in the next section we will discuss a steady - state model for the diffuse leptonic emission for an injection index of 2.0 .
there is a richness of literature on the topic of electron propagation in the galaxy , including analytical solutions for the one- or two - dimensional diffusion and diffusion - convection problem ( @xcite ) .
these solutions can be well described in their basic behavior by the concept of the catchment sphere ( @xcite ) .
the energy losses prevent electron propagation farther from their source than a critical distance @xmath45 which is defined by equality of the time scales for transport and for energy losses , so that the spatial dependence of the green s function of the problem is basically a function which is a constant for distances less than @xmath45 and which is zero beyond @xmath45 .
if the transport is governed by diffusion and if the energy loss terms do not strongly depend on location , then @xmath45 will not strongly depend on direction , and thus a source of cosmic rays at the position @xmath46 would fill a sphere of radius @xmath45 with cosmic rays . hence we may separate the spatial problem and the energy problem , and approximate the solution to the spatial problem by a gaussian function for the catchment sphere .
the gaussian function is exact at higher energies where radiative energy losses dominate , but is a crude approximation at very low energies , where ionization and coulomb losses are important .
we will include escape as a catastrophic loss term , which limits @xmath45 to some maximum value .
this approximates the effect of a sudden increase of the diffusion coefficient at a certain height above the galactic plane ( @xcite ) .
we regard this a better description than a finite boundary with density and density gradient set to zero at @xmath47 , a few kpc above the galactic plane , since for a physical escape solution the density outside the diffusion region relates to that inside as @xmath48 and thus it is definitely not zero . with a diffusion coefficient @xmath49 , @xmath50 being the kinetic energy
, we have @xmath51 where @xmath52 we can write the differential number density of cosmic rays at position @xmath53 coming from a source at @xmath46 as @xmath54 where @xmath55 @xmath56 is the energy loss term and @xmath57 is the source spectrum at position @xmath46 .
physically @xmath58 is a propagator and it can be treated like a green s function . given the spatial distribution of sources @xmath59 we obtain the cosmic ray spectrum at any position @xmath53 as @xmath60 our method thus enables us to calculate the three - dimensional distribution of electrons resulting from an arbitrary three - dimensional distribution of sources .
the individual snr may accelerate electrons with slightly different spectra .
this would result in a positive curvature of the composite injection spectrum ( @xcite ) . to demonstrate the effect of a possible dispersion of the injection spectral index in the electron sources , we assume that the injection indices for individual snr follow a normal distribution @xmath61 at the energy @xmath62 .
radio spectral index measurements at a few ghz , corresponding to @xmath63gev , indicate @xmath64 ( @xcite ) .
then the source spectrum of primary electrons is @xmath65 for @xmath66 this reduces to the conventionally assumed single power - law .
the energy losses due to ionization and coulomb interactions , bremsstrahlung and adiabatic cooling , and synchrotron and inverse compton emission are well described by @xmath67\ ] ] where we have used the following abbreviations @xmath68 @xmath69 is the neutral gas density , @xmath70 the density of ionized gas , @xmath71 the bulk velocity of electrons , and @xmath72 the energy density of the magnetic field and the ambient radiation field in @xmath73 .
we assume the magnetic field strength to be constant over the total volume of the galaxy with @xmath74 .
as we shall see later , this value leads to synchrotron emission consistent with observations .
the interstellar radiation field can be calculated from the respective emissivities for optical and infrared emission , and the microwave background emission ( @xcite and recent updates @xcite ) .
the distribution of ionized gas has been modelled on the basis of pulsar data ( @xcite ) .
the derivation of the distribution of neutral matter will be discussed below . for the propagation calculation
all parameters of the interstellar medium are averaged over a scale of 1 kpc to mimic the average environmental conditions of a cosmic ray electron during its life time .
a galactic wind is assumed to operate in the galactic halo , such that adiabatic cooling outside the disk provides similar energy losses as bremsstrahlung inside the disk , i.e. the energy loss terms can be written independent of the spatial location within a catchment sphere .
note that the energy loss terms for neighboring catchment spheres may be different , since they are averages over different volumes .
the radial extent of the galaxy is taken to be 16.5 kpc .
note that the computer time consumption scales as radial extent squared .
the calculation of the bremsstrahlung and inverse compton emissivities is described elsewhere ( @xcite ) .
the energy loss time scale , which determines the radius of the catchment sphere @xmath45 , can be understood as an ensemble average @xmath75 where @xmath57 is the electron injection spectrum .
this average age can be approximated by @xmath76 to better than @xmath77 accuracy except for the lowest energies . for @xmath78 , corresponding to @xmath79 , the average age @xmath80 is larger than the e - folding energy loss time scale . here
the energy losses will also depend explicitely on @xmath45 since ioniziation and coulomb interactions occur only in the gas disk .
for ease of computation we will assume @xmath81 to be constant at these low energies , as if ionization and coulomb interactions would not occur .
this means we overestimate @xmath45 but underestimate @xmath82 .
the two effects work in opposite directions but may not balance each other .
we want to keep in mind that we are using a crude approximation at low electron energies , which for this paper however will have an impact only on the bremsstrahlung spectra at @xmath83 .
we neglect secondary electrons in our model .
the locally observed fraction of secondary electrons is of order 10% ( @xcite ) , but the fraction may be strongly dependent on position and on the propagation behavior of particles ( @xcite ) .
there are basically two arguments which may allow us to neglect the secondaries .
firstly , because of the energy dependence of the cosmic ray secondary - to - primary ratios , secondary electrons will have a production spectrum which is softer than the production spectrum of cosmic ray nucleons by @xmath84 , at least above 1 gev electron energy .
thus secondary electrons will have a softer spectrum than primary electrons , so that their contribution to the diffuse galactic emission can not lead to a hardening of the spectrum , irrespective of the flux .
the second argument concerns the luminosity .
since the production cross sections for charged and for neutral pions are of the same order , and the electrons take only about two thirds of the pion energy , the source power supplied to secondary electrons is linked to the hadronic luminosity .
only a fraction of the source power is channeled back into emission , since synchrotron radiation takes away some energy . due to the kinematical low - energy cut - off at @xmath2100 mev in the secondary production spectrum and the decrease of hadronic luminosity with energy above 1 gev , the secondary contribution to leptonic emission above 100 mev
will always be limited to less than 10% of the luminosity due to @xmath3-decay ( @xcite ) and thus be negligible in this energy range .
the three - dimensional distribution of thermal material in the galaxy has been determined by deconvolution of h@xmath85 surveys for atomic gas and co surveys for molecular gas with the rotation curve of @xcite .
the h@xmath85 surveys include the leiden - greenbank survey ( @xcite ) , the weaver - williams survey ( @xcite ) , the maryland - parkes survey ( @xcite ) , the high - latitude parkes survey ( @xcite ) , and the heiles - habing survey ( @xcite ) .
the co data are taken from the columbia survey ( @xcite ) as updated ( @xcite ) .
all these data are publicly available from either the adc or the cds data bank .
it should be noted that none of these surveys is stray - light corrected .
the level of uncertainty in the rotation curve , the position of the sun , and non - circular motions of gas is high , so that our deconvolution should be taken as a model rather than as a datum .
the general procedere in the deconvolution process is as follows : the rotation curve is used to calculate the relation between distance and line - of - sight velocity , which is then transformed into a probability distribution for distance on the basis of the actual velocity resolution of the surveys and a turbulent velocity dispersion of 10 km / sec for co and 25 km / sec for h@xmath85 .
this approach tends to smear out the gas distribution along the line - of - sight , but relaxes most of the forbidden velocity problem .
the near - far ambiguity towards the inner galaxy is resolved by dividing the intensity according to the amplitudes of gaussian probability functions for the distribution of gas normal to the galactic plane .
these gaussian probability functions are the convolutions of the local gas distribution functions and the spatial resolution function of the particular survey .
the effective scale heights of gas on the far side of the galaxy are thus systematically larger than the local values . as a result
the gas tends to be more evenly distributed over the galaxy than in other derivations ( @xcite ) and the deconvolved gas distribution on the far side of the galaxy will be slightly smeared out normal to the galactic plane .
this has no impact on the calculation of the emission since the column density of gas is always preserved , and it has also no impact on the cosmic ray propagation since our algorithm is based on the gas surface density , which is also preserved .
the h@xmath85 data are scaled under the assumption of a constant spin temperature of @xmath86k .
the obvious absorption features in the direction of the galactic center have been replaced by a linear interpolation between the neighboring velocity bins .
the distribution of h@xmath85 normal to the galactic plane is assumed to be a gaussian of dispersion @xmath87 kpc , where @xmath22 is the galactocentric radius and @xmath88 is a heavyside function , with an offset according to the warping of the h@xmath85 disk ( @xcite ) .
the co - data are scaled with an x - factor of 1.25 , which is the mean of the best fit values in published papers on egret data analysis ( @xcite ) . in the inner kpc of the galaxy
the x - factor is reduced to 25% of its nominal value to account for the higher excitation temperature and different metallicity ( @xcite ) .
the noise in the co spectra is preserved in the deconvolution process to keep the line - of - sight integral of the density unchanged .
the vertical co distribution is assumed to be a gaussian of dispersion @xmath89 kpc , where @xmath22 is the galactocentric radius and @xmath88 is a heavyside function ( @xcite ) .
the position of the sun is assumed to be located 8.5 kpc from the galactic center and 15 pc above the plane ( @xcite ) . for a region of @xmath90 towards the galactic center and towards the anticenter
the kinematical resolution is insufficient , and the data have been edited by hand .
the distribution of gas in these two regions is basically an interpolation between the results for the adjacent regions . in the anticenter region
any excess over this interpolation has been evenly distributed over all distances , while in the galactic center region any excess is attributed to the galactic center .
the distribution of gas in the galactic plane is shown in fig.[gas ] .
the true distribution of snr in the galaxy is not well known as a result of selection effects and the absence of a proper distance measure to the remnants . in many papers ( @xcite )
the cosmic ray distribution in the galaxy has been estimated on the basis of a functional form for the snr distribution which fits the data for 116 remnants ( @xcite ) .
one of the general findings in these studies is that the overall cosmic ray distribution is too steep to explain the gradual slope of the emissivity over the galactic radius ( @xcite ) .
here we use a revised functional form for the snr distribution ( @xcite ) , which fits the data for 194 remnants @xmath91 where @xmath928.5 kpc denotes the distance between the sun and the galactic center .
the vertical distribution of snr is taken to be box - shaped with a half thickness of 150 pc .
our model allows us to use true 3d source distributions .
we know that the galaxy has structure in the form of spiral arms , a bulge and so forth .
we have thus folded a spiral arm model ( @xcite ) with the radial snr distribution as described above to investigate the influence of spiral arm structure . in the georgelin and georgelin model
the galaxy has four symmetric arms with pitch angle of around 12@xmath93 .
we have rescaled their model to @xmath928.5 kpc .
each spiral arm is described by a gaussian of dispersion 500 pc , i.e. a fwhm of 1177 pc .
the spiral arm model is normalized in azimuth , so that the integral @xmath94 of that model yields unity , and then folded with the radial snr distribution .
the normalization is required to preserve the radial distribution of snr .
a face - on view of the resultant source distribution is shown in fig.[src ] .
in this section we show results for two choices of the spatial distribution of sources , the pure snr distribution and the snr distribution folded with a model of the spiral arms in the galaxy .
we will also show results for two choices of injection index , at first a fixed index s=2.0 for all sources , and then a normal distribution of indices with mean 2.0 and dispersion 0.2 .
we will not vary any other parameter in the propagation model , and stick to the best fit values given by webber , lee and gupta ( 1992 ) . in a forthcoming paper
we will extend our model to nucleons and determine the propagation parameters self - consistently which fit the data and the local spectra of primary and secondary cosmic rays . in this paper
our emphasis is to show that an injection index of around s=2.0 for cosmic ray electrons is sufficient to explain the observed spectrum of diffuse high - energy , while leaving all other parameters unchanged .
we have shown in sec.2 that if electrons are accelerated in snr , their local spectra above @xmath2 30 gev will strongly depend on time , so that the direct electron measurements above this energy may deviate from the average electron spectrum . as a result
, we are not required to choose the electron injection index according to the directly observed electron spectrum above @xmath2 30 gev .
below a few gev , on the other hand , the locally observed spectra are strongly affected by solar modulation , for which we do not have reliable models , so that only over roughly one decade in energy does the local electron spectrum provide clear data .
the data from direct electron measurements compared with the steady - state spectra of our model are shown in fig.[local_both ] .
the model spectra fit the data reasonably well in the relevant energy range up to 30 gev .
as an example we show in fig.[hunt_snr ] the spectra of the bremsstrahlung and inverse compton emission in the direction of the inner galaxy for the case of sources with injection indices following a normal distribution of mean s=2.0 and dispersion @xmath95=0.2 .
the figure includes the results of our model , the observed spectrum , and a template of the @xmath3-decay spectrum ( @xcite ) . at around 5 gev
the intensity due to leptonic processes is higher than that due to hadronic @xmath3 decay .
our model assumes that the power - law behaviour of the electron injection spectra persists to @xmath220 tev .
the high - energy spectrum will rather directly reflect structure in the electron source spectra .
if for example the true source spectrum would deviate from a simple power - law above 1 tev , the inverse compton spectrum would show corresponding features above 50 gev .
when we consider the latitude distribution of the intensity above 1 gev , which is done in fig.[plotb_both ] , we find a high level of agreement .
the fraction of the total diffuse intensity , which is due to leptonic emission , is almost constant between b=0@xmath93 and b=10@xmath93 .
it goes up from @xmath26% in the hunter et al .
model , @xmath210% in our model for an injection index of s=2.4 , to 30% - 48% for an injection index of s=2.0 , enough to explain all the excess .
we can compare the longitude distribution of the observed diffuse emission above 1 gev to that of the leptonic contribution in our model .
this is done in fig.[plotl_both ] .
it is obvious that the galactocentric gradient of the leptonic emission in our model is stronger than in the data .
this is the case for all cosmic ray propagation models which are based on the snr distribution ( @xcite ) . while towards the inner galaxy the fraction of the total diffuse intensity , which is due to leptonic emission , is around 35% - 52% , far enough to explain all the excess , towards the outer galaxy the leptonic contribution in our model accounts only for roughly two thirds of the excess .
the snr distribution of case and bhattacharya has a zero at galactocentric radius r=0 which causes the double peak structure in the model intensity towards the inner galaxy .
such double peak structure is not visible in the data , so that this effect may result more from the specific choice of mathematical function in the fit of the snr distribution than astrophysical reality .
we can generally say that a flatter snr distribution would beget a better harmony of the longitude distribution of observed emission and model .
interestingly , one study of the snr distribution , which was not based on a specific form of the radial profile , indicates very long radial scale lengths up to 9 kpc ( @xcite ) .
we have therefore investigated the impact of the fit uncertainties in the snr distribution . within 1 @xmath96 in both parameters
the snr distribution may be @xmath97 here we show the results for the spiral arm model only . for similar local electron
spectra the leptonic contribution to the diffuse galactic emission above 1 gev can be 33% -47% while the center / anticenter contrast in the model agrees with the observed one to better than 25% . as can be seen in fig.[plotb_flat ] , the latitude distribution of the leptonic contribution varies little .
the longitude distribution of the leptonic contribution in fig.[plotl_flat ] shows that the overall gradient is reasonably well reproduced , but the double hump structure towards the inner galaxy remains as well as a general overprediction around @xmath98 .
this double hump structure however is a consequence of the mathematical function chosen to fit the snr distribution and not a consequence of the snr distribution itself .
a further constraint on our model is the synchrotron flux towards the north galactic pole ( ngp ) .
available data at 408 mhz ( @xcite ) , at 820 mhz ( @xcite ) , and at 1420 mhz ( @xcite ) can be corrected for zero level uncertainties , the contributions of the microwave background and unresolved extragalactic sources ( @xcite ) , and contributions from the coma cluster ( @xcite ) .
here we do not use data at frequencies below 100 mhz , which we expect to be affected by free - free absorption . in fig.[synngp ] we compare the synchrotron intensity predicted by our model with the data in direction of the ngp .
it can be seen that for a total magnetic field strength of 10@xmath95 g there is good agreement . in our model the fwhm of the z - distribution of synchrotron emission at 1420 mhz
is @xmath21.1 kpc , which is the value typically found for edge - on galaxies ( @xcite ) .
in this paper we have investigated whether cosmic ray electrons can be responsible for the recently observed high intensity of diffuse galactic emission above 1 gev .
models based on the locally observed cosmic ray spectra underpredict the observed intensity by nearly 40% ( @xcite ) .
one feature of these models is the relatively soft electron injection spectral index of s=2.4 ( @xcite ) , which is required to account for the local electron spectrum above 50 gev .
the recent detection of non - thermal x - ray synchrotron radiation from the four supernova remnants sn1006 ( @xcite ) , rx j1713.7 - 3946 ( @xcite ) , ic443 ( @xcite ) , and cas a ( @xcite ) supports the hypothesis that galactic cosmic ray electrons are accelerated predominantly in snr .
we have shown in this paper that , if this is indeed the case , the local electron spectra above 30 gev are variable on time scales of a few hundred thousand years .
this variability stems from the poisson fluctuations in the number of snr in the solar vicinity within a certain time period . while the electron spectra below 10 gev are stable , the level of fluctuation increases with electron energy , and above 100 gev the local electron flux is more or less unpredictable . with that time variability in mind
we have seen that an electron injection index of s=2.0 is consistent with the data of the direct particle measurements if snr are the dominant source of cosmic ray electrons .
in fact , both the radio spectra of individual snr ( @xcite ) and the hard spectrum of the inverse compton emission at high latitudes ( @xcite ) would better harmonize with an injection index of s=2.0 instead of s=2.4 .
we have then presented a three - dimensional steady - state diffusion model for cosmic ray electrons , based on the propagation parameters which have been derived from similar models for cosmic ray nucleons . while being entirely consistent with the local electron electron flux up to @xmath5 30 gev energy and with the radio synchrotron spectrum towards the north galactic pole , the leptonic contribution to the diffuse galactic emission above 1 gev in the galactic plane increases from @xmath26% in the model of hunter et al . , @xmath2 , 10% in our model for an injection index s=2.4 , to 30 - 48% for an injection index s=2.0 depending on the assumed spatial distribution of snr and depending on whether some dispersion of injection spectral indices is allowed .
an electron injection index s=2.0 can therefore explain the bulk of the observed excess over the predictions of the hunter et al .
while the latitude distribution of the leptonic emission is fully consistent with that of the observed emission , we find that the longitude distribution deviates from the observed one . in our model
the contrast between the galactic center and the anticenter is stronger than in the data .
a similar effect can be found in all cosmic ray propagation models which are based on the snr distribution ( @xcite ) .
note that in the hunter et al .
model only the emissivity spectrum is taken according to a propagation calculation ( @xcite ) , while the spatial distribution of the emissivity is scaled according to the distribution of thermal gas .
we have seen that the fit uncertainties in the snr distribution of case and batthacharya allow us to use a flatter profile , which leads to a better agreement between model and data in the longitude distribution ( see fig.[plotl_flat ] ) .
thus the gradient problem may be simply the result of an inappropriate choice of radial snr distribution or lack of error propagation , respectively .
this flatter profile would also harmonize better with the results of li et al .
( 1991 ) , who found very large radial scale lengths of up to 9 kpc for the galactic distribution of snr .
other possible sources of systematic errors are discussed below .
this gradient problem is unlikely to be caused by additional thermal matter .
very cold ( 3@xmath93k ) molecular gas has recently been discussed as candidate for baryonic dark matter ( @xcite ) .
if organised in small clumps ( @xcite ) the probability of finding absorption features in the spectra of bright background objects would be small and the clumps would easily evade detection ( for a review see @xcite ) .
however , the thermal gas mainly affects the bremsstrahlung and only indirectly the inverse compton emission , which dominates above 1 gev energy , and thus has little influence on the gradient in the total leptonic emission above 1 gev .
if our model for the interstellar radiation field were wrong , then it would have only a limited influence on the gradient , since the galaxy acts as a fractional calorimeter at high electron energies ( @xcite ) , which channels the electron source power directly into synchrotron and inverse compton emission with a ratio corresponding to that of the energy densities in the magnetic field and the photon field .
any radial variation of the magnetic field strength will be directly reflected in the center / anticenter contrast of the synchrotron intensity , so that the radio surveys would constrain the parameter space here .
it is our personal view that the uncertainties in the radial distribution of snr are large enough so that the overly strong gradient in our model may simply be the result of an inappropriate choice of snr distribution .
the artificial double peak structure towards the inner galaxy is an example of systematic effects which arise from possibly ill - defined fits to the snr distribution .
our findings indicate a potential problem in the determination of the extragalactic background ( @xcite ) .
the standard method uses a linear regression analysis of observed intensities and model predictions ( with the model of hunter et al . ) to extrapolate to zero galactic intensity . if at higher energies the inverse compton emission is indeed much stronger than assumed by hunter et al .
, then a large fraction of it will be attributed to the extragalactic emission .
our steady - state model predicts an intensity of high - latitude inverse compton emission above 1 gev at a level of @xmath99 of the extragalactic background intensity , while hunter et al .
assume much smaller values .
the intensity difference between the inner galaxy and the outer galaxy at medium latitudes ( @xmath100 ) would be around 10% of the extragalactic background intensity , depending on the choice of model for the electron source distribution ( snr distribution ) .
on the other hand , if we are indeed living in a region of temporarily low flux of high - energy electrons , we would also expect the intensity of inverse compton emission towards the galactic poles to be less that the steady - state value , so that the true level of galactic intensity at high latitudes is difficult to assess .
the spectrum of inverse compton emission , [email protected] in our model , is somewhat harder than the [email protected] of the extragalactic background .
it has been noted before that the average spectrum of the identified agn is softer than that of the background ( @xcite ) , which indicates a problem with the idea that the background is mainly due to unresolved agn .
now , since the presently determined background spectrum may be substantially contaminated with hard inverse compton emission , the true background spectrum may be softer than s=2.1 , which would probably relieve the spectral discrepancy with the average spectrum of resolved agn ( @xcite ) .
anyway , the systematic uncertainty in the spectral index of the extragalactic background emission is much higher than the statistical uncertainty . in a forthcoming paper
we shall discuss the distribution of synchrotron emission in more detail .
we shall also use a truly three - dimensional calculation of the interstellar photon fields based on cobe / firas data .
finally we shall discuss cosmic ray nucleons in parallel with electrons to derive the propagation parameters self - consistently .
the egret team gratefully acknowledges support from the following : bundesministerium fr bildung , wissenschaft , forschung und technologie ( bmbf ) , grant 50 qv 9095 ( mpe ) ; nasa cooperative agreement ncc 5 - 93 ( hsc ) ; nasa cooperative agreement ncc 5 - 95 ( su ) ; nasa contract nas5 - 96051 ( ngc ) ; and nasa contract nas5 - 32490 ( usra ) .
allen g.e .
et al . , 1997 , , 487 , l97 arimoto n , sofue y. , tsujimoto t. , 1996 , , 48 , 275 barwick s.w .
et al . , 1997 , , 482 , l191 berkey g.b . , shen c.s . , 1969 , physical review , 188 , 1994 berkhuijsen e.m . , 1972 , , 5
, 263 bertsch d.l .
et al . , 1993 , , 416 , 587 blandford r. , eichler d. , 1987 , , 154 , 1 bloemen h. , 1989 , , 27 , 469 bloemen h. et al . , 1993 , , 267 , 372 bonino g. , 1996 , proc .
24th icrc , il nuovo cimento vol .
19c , n.6 , p.865 brecher k. , burbidge g.r . , 1972 , , 174 , 253 buckley j.h .
et al . , 1998 , , 329 , 639 burton w.b . , 1985 , , 62 , 365 burton w.b
, 1976 , , 14 , 275 burton w.b . ,
liszt h.s . , 1983 , , 52 , 63 case g. , bhattacharya d. , 1996 , , 120 , c437 cappelaro e. et al .
, 1997 , , 321 , 431 chen a. , dwyer j. , kaaret p. , 1996
, , 463 , 169 cleary m.n . , heiles c. , and haslam c.g.t , 1979 , , 36 , 95 clemens d.p . , 1985 , , 295 , 422 combes f. , pfenniger d. , 1997 , , 327 , 453 cowsik r. , lee m.a . : 1979 , , 228 , 297 dame t.m .
et al . , 1987 , , 322 , 706 dermer c.d . , 1986 , , 157 , 223 digel s.w . , dame t.m . , 1997 , private communication digel s.w . et al . , 1996 , , 463 , 609 digel s.w . , hunter s.d . , mukherjee r. , 1995 , , 441 , 270 dogiel v.a . , uryson a.v . , 1988 , , 197 , 335 drury l. oc . , aharonian f.a
, vlk h.j . , 1994 , , 287 , 959 esposito j.a .
et al . , 1998 , , submitted fazio g.g .
, stecker f.w .
, wright j.p . , 1966 , , 144 , 611 ferrando p. et al . , 1996 , , 316 , 528 fichtel c.e .
et al . , 1975 , , 198 , 163 gleeson l.j . , axford w.i . , 1968 , , 154 , 1011 georgelin y.m . , georgelin y.p
, 1976 , , 49 , 57 ginzburg v.l .
, syrovatskii s.i . : 1964 , the origin of cosmic rays , pergamon press golden r.l .
et al . , 1994 , , 436 , 769 golden r.l .
et al . , 1984 , , 287 , 622 green d.a . , 1995 , a catalogue of galactic snr , mullard rao , cambridge , uk hammersley p.l .
et al . , 1995 , , 273 , 206 haslam c.g.t .
et al . , 1982 , , 47 ,
1 heiles c. , habing h.j .
, 1974 , , 14 , 557 hummel e. , 1991 , iau symposium 144 , ed . h. bloemen , kluwer , p.257 hunter s.d .
et al . , 1997 , , 481 , 205 hunter s.d .
, 1994 , , 436 , 216 keohane j.w .
et al . , 1997 , , 484 , 350 kerr f.j .
et al . , 1986 , , 66 , 373 kocharov g.e . , 1996 ,
24th icrc , il nuovo cimento vol .
19c , n.6 , p.883 kodaira k. , 1974 , , 26 , 255 koyama k. et al . , 1997 , , 49 , l7 koyama k. et al . , 1995 , , 378 , 255 kraushaar w. et al . , 1972 , , 177 , 341 lerche i. , schlickeiser r. , 1981 , , 47 , 33 lerche i. , schlickeiser r. , 1982 , , .
201 , 1041 lerche i. , schlickeiser r. , 1982 , , 107 , 148 lessard r.w .
, 1995 , 24th icrc , 2 , 475 li z. , wheeler j.c . , bash f.n . ,
jefferys w.h . , 1991 , , 378 , 93 mayer - hasselwander h. et al . , 1982 , , 105 , 164 mchargue l.r . , damon p.e . , donahue d.j . , 1995 ,
lett . , 22 , 659 mclaughlin m.a .
et al . , 1996 , , 473 , 763 merck m. et al . , 1996 , , 120 , c465 mori m. , 1997 , , 478 , 225 pfenniger d. , combes f. , 1994 , , 285 , 94 pfenniger d. , combes f. , martinet l. , 1994 , , 285 , 79 pohl m. , 1996 , , 307 , l57 pohl m. , 1994 , , 287 , 453 pohl m. , schlickeiser r. , 1990 , , 234 , 147 pohl m. , kanbach g. , hunter s.d . , jones b.b . , 1997a , , 491 , 159 pohl m. , hartman r.c .
, jones b.b . , sreekumar p. , 1997b
, , 326 , 51 porter t.a . ,
protheroe r.j . , 1997 , j. phys .
g , 23 , 1765 reich p. , reich w. , 1986 , , 63 , 205 reich p. , reich w. , 1988 , , 74 , 7 schlickeiser r. , pohl m. , ramaty r. , skibo j.g .
, 1997 , proc .
4th compton symposium , eds . c.d .
dermer and j.d .
kurfess , aip conf .
410 , p.449 schlickeiser r. , 1982 , , 106 , l5 schlickeiser r. , sievers a. , thiemann h. , 1987 , , 182 , 21 skibo j.g . , 1993 , phd dissertation , university of maryland sodroski t.j .
et al . , 1995 , , 452 , 262 sonnett c.p .
et al . , 1987 , nature , 330 , 458 sreekumar p. et al . , 1993 , , 70 , 127 sreekumar p. et al . , 1998 , , 494 , 523 stecker f.w . , jones f.c . , 1977 , , 217 , 843 strong a.w . , 1997 , private communication strong a.w .
, mattox j.r . , 1996 , , 308 , l21 strong a.w . , 1995 , , 76 , 205 strong a.w .
et al . , 1988 , , 207 , 1 taira t. et al . , 1993 , 23rd icrc , 2 , 128 tang k.k . , 1984 , , 278 , 881 tanimori t. et al . , 1997 ,
iauc 6706 taylor j.h .
, cordes j.m . , 1993 , , 411 , 674 valle j.p . , 1995 , , 454 , 119 weaver h.f . , williams d.r.w . , 1973 , , 8,1 webber w.r . , lee m.a .
, gupta m. , 1992 , , 390 , 96 webster a.s . , 1970 , , 5 , 189 youssefi g. , strong a.w . , 1991 , 22nd icrc , 1 , 129 | the recently observed x - ray synchrotron emission from four supernova remnants ( snr ) has strengthened the evidence that cosmic ray electrons are accelerated in snr .
we show , that if this is indeed the case , the local electron spectrum will be strongly time - dependent , at least above roughly 30 gev .
the time dependence stems from the poisson fluctuations in the number of snr within a certain volume and within a certain time interval .
as far as cosmic ray electrons are concerned , the galaxy looks like actively bubbling swiss cheese rather than a steady , homogeneously filled system .
our finding has important consequences for studies of the galactic diffuse gamma - ray emission , for which a strong excess over model predictions above 1 gev has been reported recently . while these models were relying on an electron injection spectrum with index 2.4 chosen to fit the local electron flux up to 1 tev we show that an electron injection index of around 2.0 would a ) be consistent with the expected poisson fluctuations in the locally observable electron spectrum and b ) explain the above mentioned gamma - ray excess above 1 gev .
an electron injection index around 2 would also correspond to the average radio synchrotron spectrum of individual snr .
we use a three - dimensional propagation code to calculate the spectra of electrons throughout the galaxy and show that the longitude and latitude distribution of the leptonic gamma - ray production above 1 gev is in accord with the respective distributions for the gamma - ray excess .
we finally point out that our model implies a strong systematic uncertainty in the determination of the spectrum of the extragalactic gamma - ray background . |
the nonlinear optical ( nlo ) properties on conjugated polymers have always been received wide interest for study.@xcite experimentally observed nonlinear phenomenon in conducting polymers , such as photoinduced absorption , bleaching , photoluminescence , ultrafast optical process and exciton behaviors,@xcite have already encouraged the theorists for the possible explanations.@xcite in the early 90s , the large third order susceptibility ( @xmath3 ) of trans - polyacetylene ( pa ) have been observed experimentally@xcite and have been received wide explanations theoretically.@xcite for a long time , the origin of the experimentally observed two - photon absorption peak ( tpa)@xcite receives persistent discussions in theory.@xcite based on the static current - current correlation ( @xmath1)@xcite and followed by keldysh green function methods@xcite , tpa could be interpreted through the analytical forms under both ssh@xcite and tlm models@xcite for infinite chains .
besides the unnecessary application of nonequilibrium diagram techniques in a stable optical response problem,@xcite however , the various numerical approaches from dipole formalism(@xmath4 ) , such as the results by yu and su@xcite from butcher - bloembergen - shen dipole formula,@xcite the results by wu and sun@xcite from genkins - mednis approach@xcite and the results by shuai and brdas from sum - over - state(sos ) orr - ward formalism,@xcite have shown that the tpa is no longer existing if the damping in the energy is considered.@xcite therefore , the two - photon cusp under the current schemes is considered to be a van hove singularity@xcite caused by the singular density of states ( dos ) on the fermi surface in one - dimensional ( @xmath5 ) systems . the fact of the nonresonant property in two - photon cusp is also noticed by wu in his current formulations.@xcite however , besides the difficulty in explaining this tpa under current formulas , there exists another obvious difficulty in theory apparent zero frequency divergence ( zfd ) in the definition of the current formalism.@xcite this problem actually was realized by dakhnovskii and pronin in their computations based on ssh model.@xcite unfortunately , the results of @xmath3 obtained from ssh model returns tlm model if the process of linearization procedures is made under @xmath1 scheme.@xcite because of the apparent divergence difficulties in current formulas@xcite , practically , intense studies based on dipole - dipole ( @xmath0 ) correlation have been carried out to obtain the physical pictures of tpa in @xmath3 of trans - polyacetylene .
on one hand , the weakly correlated and single electron theories are applied in the numerical dipole computations , thus , the finite size effect@xcite and the lifetime assumption@xcite are seemingly good in generating a peak at exact position of tpa ; on the other hand , the strongly correlated electron theories have also been applied to short chains , soos and ramesesha obtained the tpa based on pariser - parr - pople ( ppp ) models , but it is shifted too low in frequency and has too small an intensity@xcite relative to the experimental peak.@xcite recent work by guo and mazumdar shows that this tpa could be interpreted as a three - photon resonance based on extended hubbard hamiltonian.@xcite whether or not the weakly correlated theory or even the single electron models such as ssh and tlm models are still suitable in describing tpa still remains as a question . meanwhile , in obtaining the qualitatively correct result , it was suggested by etemad and soos that the attention should be made to the @xmath3 frequency dependences rather than magnitudes.@xcite it is certainly reasonable in physical point of view since the experiments are more sensitive to detect frequency dependences .
thus , some typical approximations are made in the dipole formula .
substitution of the dipole operator @xmath6 by the polarization operator @xmath2 , namely , the dipole approximation@xcite has been extensively applied in the nonlinear optical formula , however , @xmath7 computations have shown both quantitatively and qualitatively different results between @xmath2 and dipole operator @xmath6.@xcite whether the problem caused by the dipole approximation still existing in @xmath3 is still under our concern .
furthermore , the static dipole formulas@xcite and static current formulas@xcite are conventionally considered to be equivalent.@xcite however , as we pointed out in a recent work,@xcite it has zfd problem in the static current formalism when the gauge phase factor is ignored , the static formalism under two gauges should not return the same results under the exact same basis of unperturbed wave functions.@xcite whether this nonequivalence between @xmath1 and @xmath0 correlations existing in the specific case of @xmath3 of trans - polyacetylene is another interest topic .
although the qualitative features have already been reported in the numerical solutions based on @xmath0 correlation,@xcite the exact analytical solutions from the dipole formulas of @xmath3 have not been obtained previously .
the analytical results , if obtained , will be very helpful for us to directly check all above concerns and give more information . fortunately , both ssh and tlm models are single electron models and can be exactly solvable for the nonlinear susceptibilities under perturbated schemes , although the efforts will be no doubt paid to cumbersome computations . after choosing the long wavelength approximation which are conventionally used previously,@xcite we could directly compared those analytical results between two gauges for both models .
the paper is organized as follows , we will first discuss the ssh hamiltonian in sec.ii.a .
the analytical form of @xmath3 by the dipole - dipole ( dd ) correlation is obtained and compared with the analytical form by @xmath1 correlation for ssh models in sec.ii.b . in sec.ii.c
, we will obtain the results under the polarization operator @xmath2 , the qualitative and quantitative difference between the results under @xmath6 and @xmath2 are outlined . in sec.iii , we will obtain the exact analytical forms under @xmath0 correlation for tlm models and directly compared with those under @xmath1 correlation .
the discussions on the possible implications of this work are made in sec.iv and a brief conclusion is made in sec.v .
based on periodic tight - binding - approximations , the ssh hamiltonian@xcite is given by : @xmath8 ( \hat{c}_{l+1,s}^{\dag}\hat{c}_{l , s}^{}+\hat{c}_{l , s}^{\dag}\hat{c}_{l+1,s})^ { } , \label{hssh}\end{aligned}\ ] ] where @xmath9 is the transfer integral between the nearest - neighbor sites , @xmath10 is the gap parameter and @xmath11 creates(annihilates ) an @xmath12 electron at site @xmath13 with spin @xmath14 . in continuum limitation , above ssh model will give the tlm model.@xcite for the ssh model , each site is occupied by one electron . under the @xmath0 correlation ,
the interaction hamiltonian is expressed by @xmath15 , to provide the periodic property and to avoid the ill - definition of position operator @xmath16 , we should express the position operator @xmath16 under the bloch states : @xmath17 , where @xmath18 and @xmath19 are band index and crystal momentum correspondingly , @xmath20 is the periodic function under the translation of lattice vector.@xcite thus , we obtain : @xmath21 and @xmath22 where @xmath23 is unit cell volume .
we change hamiltonian eq.([hssh ] ) into the momentum space by applying the following consecutive transformations : @xmath24 and @xmath25 with @xmath26 where @xmath27 ^ 2+\left [ \delta sin(ka ) \right]^2 } , \label{ek}\end{aligned}\ ] ] and @xmath28 and @xmath29 are the excitations of electrons in the conduction band and the valence band with momentum @xmath30 and spin @xmath14 , @xmath31 and @xmath32 are odd and even position defined by @xmath33 if we choose the spinor description @xmath34= @xmath35 , @xmath36 and apply the long wavelength approximation,@xcite ssh hamiltonian including @xmath37 in momentum space is described by : @xmath38 where @xmath39 and @xmath40 by eq .
( [ r ] ) , the dipole operator @xmath6 could be obtained as follows:@xcite @xmath41 where @xmath42 is the coefficient related to the interband transition between the conduction and valence bands in a unit cell @xmath43 and the second term in eq.([d ] ) is related to the intraband transition,@xcite @xmath44 is the electric charge and @xmath45 are the pauli matrixes .
we have already omitted the relative distortion @xmath46(@xmath47 ) in this computation since the contribution of @xmath46 is quite small according to the linear case.@xcite within the semiclassical theory of radiation,@xcite the electrical field is treated classically and the third order susceptibility @xmath3 is described by:@xcite @xmath48 ^ 3 \int d{\bf r}_1 d{\bf r}_2 d{\bf r}_3 \int dt_1 dt_2 dt_3 \nonumber \\ & & \int d{\bf r } dt\ , e^{-i { \bf k \cdot r}+ i \omega t } \langle \hat{t } \hat{{\bf d } } ( { \bf r},t )
\hat{{\bf d}}({\bf r}_1,t_1 ) \hat{{\bf d}}({\bf r}_2,t_2 ) \hat{{\bf d } } ( { \bf r}_3,t_3 ) \rangle , \label{dd}\end{aligned}\ ] ] where @xmath49 and t is the time - ordering operator with @xmath50 is a dipole operator .
the third - harmonic generation ( thg ) is defined by setting @xmath51 . following similiar procedures as we have done in the linear computations,@xcite by eq.([d ] ) and eq.([dd ] ) , we obtained the following expression for @xmath52@xmath53 @xmath54 : @xmath55 where @xmath56 is defined in eq.([d ] ) and @xmath57 is the number of chains in unit cross area and the polymer chains are assumed to be oriented , and green function @xmath58 defined as following:@xcite @xmath59 with @xmath60 .
after tedious derivation , we obtained the following analytical expressions for the third - harmonic generation for the ssh models @xmath61 : @xmath62 \right ] \nonumber \\ & -&\frac{\beta(k)}{(2\omega_k-3\omega)}\frac{\partial}{\partial k}\left [ \frac{1}{(2\omega_k-2\omega)}\frac{\partial}{\partial k } \left [ \frac{\beta(k)}{(2\omega_k-\omega ) } \right ] \right ] \biggr\ } , \label{ssh}\end{aligned}\ ] ] the above expression eq.([ssh ] ) are almost identical to eq.(a13 ) in agrawal _ et al . _
s work@xcite except one more term ( last term in ( a13 ) ) , and with the following substitution is applied in eq.(a13 ) : @xmath63 eq.([ssh ] ) is also quite similar to eq.(11 ) in wu and sun s result@xcite based on genkins - mednis approach@xcite with the above substitution ( [ st ] ) made and if the symmetric procedures on the negative frequency term should be applied . for infinite chains , we could separate two full derivative terms of @xmath30 from eq.([ssh ] ) as follows : @xmath64 \right ] \nonumber \\ & -&\frac{\partial}{\partial k}\left[\frac{\beta(k)}{(2\omega_k-3\omega ) ( 2\omega_k-2\omega)}\frac{\partial}{\partial k } \left [ \frac{\beta(k)}{(2\omega_k-\omega ) } \right ] \right ] \biggr\},\end{aligned}\ ] ] if we consider following two facts in actual optical process : ( i ) .
the velocity on the fermi surface @xmath65 .
the life time of the states , thus , we should include the damping in the process , that is , @xmath66 and @xmath67 .
the above full derivative term @xmath68 after integral of @xmath30 then is 0 for the infinite chains .
eq.([ssh ] ) can be simplified as following : @xmath69^{\frac{1}{2 } } } \biggl\{\frac{37 - 24(1+\delta^2)x^2 + 12\delta^2x^4}{8 x^8(x^2-z^2 ) } \nonumber \\ & + & \frac{9\left [ 243 - 216(1+\delta^2)x^2 + 188\delta^2 x^4 \right ] } { 8 x^8(x^2-(3 z)^2 ) } \biggr\ } \nonumber \\ & = & \chi_0^{(3 ) } \frac{5}{1024z^8}\biggl\ { -336e(\frac{\pi}{2},\sqrt{1-\delta^2 } ) + 120z^2\delta^2f(\frac{\pi}{2},\sqrt{1-\delta^2 } ) \nonumber \\ & + & \frac{8z^4}{5}\left[(-12 + 7\delta^2 - 12\delta^4)e(\frac{\pi}{2 } , \sqrt{1-\delta^2})+6(1+\delta^2)\delta^2f(\frac{\pi}{2},\sqrt{1-\delta^2 } ) \right ] \nonumber \\ & + & 9(37 - 24(1+\delta^2)z^2 + 12\delta^2z^4)g(z ) + ( 3 - 24(1+\delta^2)z^2 + 188\delta^2z^4)g(3z ) \biggr\ } , \label{sshdd}\end{aligned}\ ] ] and @xmath70 where @xmath71 , @xmath72 , @xmath73 and @xmath74 .
@xmath75 are first , second and third kind of complete elliptical integrals.@xcite there is exactly no two photon cusp in eq.([sshdd ] ) .
this result is consistent with yu and su , shuai and brdas results .
and it is quite different from the expression obtained from @xmath1 correlation , which shows the following form for @xmath76 of ssh models:@xcite @xmath77g(z ) -8[1 - 4z^2(1+\delta^2)+16z^4\delta^2]g(2z ) \nonumber \\ & + & [ 3 - 24z^2(1+\delta^2)+188z^4\delta^2]g(3z ) -8\delta^2z^4e(\frac{\pi}{2},\sqrt{1-\delta^2})\ } , \label{sshjj}\end{aligned}\ ] ] where @xmath78 , same as the constant as defined in eq.([sshdd ] ) and g(z ) is defined by eq.([gz ] ) . we should point out that the elliptical form of eq.([sshjj ] ) @xcite is the exact the same as the integral form of eq.(11 ) in wu and sun s result,@xcite which was derived from the incomplete genkins - mednis approach.@xcite the disappearance of zfd under static current schemes is somewhat puzzling or simply should be understood as coincidence for the specific ssh model .
as we pointed out in a recent work,@xcite the ignorance of gauge phase factor in the static current formula will cause zfd in the susceptibilities and non - equivalent result from the dipole formula .
the derivation in the previous current work@xcite does not consider the gauge phase factor s effect , it will cause zfd even for linear case if following a straightforward computations.@xcite obviously , we observe the qualitatively difference for results under @xmath0 ( eq.[sshdd ] ) ) and @xmath1 ( eq.([sshjj ] ) ) correlation especially for @xmath79 and @xmath80 .
the inequivalence could be understood by the ignore of the gauge phase factor in optical response theory.@xcite let @xmath81 in the integral of eq.([sshdd ] ) , the comparison graph of absolute values between eq.([sshdd ] ) and eq.([sshjj ] ) for the ssh models are plotted in fig.1 with the following parameters chosen for trans - polyacetylene:@xcite @xmath82 , @xmath83 , @xmath84 , @xmath85 and @xmath86 , we have @xmath87 and @xmath88 esu .
-10pt from the graph , we find another very obvious peak at @xmath80 at a ratio of @xmath89 of the peak at @xmath90 , which is also different comparing with the previous theoretical computations.@xcite however , this peak has not been reported by the experiments because it is out of the scanning range of photon energy.@xcite in a recent work , we discussed the different results caused by polarization operator @xmath2 in linear response@xcite .
the polarization operator @xmath2 is extensively applied in the nlo theory.@xcite based on ssh model , we can do a comparison of @xmath91 between @xmath6 and @xmath2 .
the polarization operator @xmath2 is defined in real space as : @xmath92 if the chain region @xmath13 is going from @xmath93 to @xmath94 , we obtain the unit polarization @xmath95 from the total polarization @xmath96 in the momentum space : @xmath97 substituting @xmath6 with @xmath2 in eq.([dd ] ) , we obtain the @xmath98 for ssh models as following for infinite chains : @xmath99^{\frac{1}{2 } } } \biggl\{\frac{1}{x^2-z^2}-\frac{9}{x^2-(3z)^2}\biggr\}\end{aligned}\ ] ] the above expressions lacks tpa . however , the magnitude is too small compared with results under @xmath6 .
( @xmath100 from @xmath2 is about @xmath101 of @xmath102 from @xmath6 if @xmath87 for trans - polyacetylene . )
another peak at @xmath103 will shown up through @xmath104 , this peak corresponding to the transition from the bottom of valence band to the top of conduction band .
this peak in infinite chains seems not to agree with ` unklapp enhancement ' in the solid states.@xcite while in @xmath0 computations , this peak at @xmath105 will not be very obvious .
the comparison on the absolute value between @xmath98 and @xmath91 will be shown in fig.2 . from the graph ,
the different shapes of curves are quite obvious .
besides the possible ill - definition of polarization operator @xmath2,@xcite both the qualitative and quantitative features obtained from the @xmath2 are quite different even for nonlinear optical response theory
. thus it will not be quite suitable in using @xmath2 to obtain the equivalent results as @xmath6 . -10pt
the results under tlm model for the @xmath0 correlation can be obtained by setting @xmath106 and @xmath107 in eq.([sshdd]).@xcite since we have the following properties for the complete elliptical integrals:@xcite @xmath108 where @xmath109 we will obtain the following expressions for thg under tlm models for @xmath0 correlation , @xmath110 where @xmath111 defined in eq.([sshdd ] ) .
the @xmath112 is plotted in fig.3 . and no singular property such as van hove singular shows in @xmath113 under @xmath0 correlation . in @xmath1 correlation
, the result of @xmath114 obtained by wu as follows:@xcite @xmath115 the comparison between our result(@xmath0 ) and wu s result ( @xmath1)@xcite on absolute value of @xmath112 is plotted in fig.4 .
if the zero frequency limitation is made , let @xmath116 , we obtain the different values for the zero frequency limit for both @xmath0 and @xmath1 correlations : @xmath117 the nonequivalent results between @xmath0 and @xmath1 correlation are quite obvious even for the static limit .
the disappearance of zfd could be understood as coincidence.@xcite -10pt -10pt
the analytical forms eq.([sshdd ] ) and eq.([tlmdd ] ) show that there will be exactly no two - photon peak or even no van hove singularity under @xmath0 correlation .
they are qualitatively different from the expressions eq.([sshjj ] ) and eq.([tlmjj ] ) derived from @xmath1 correlation .
the results with no tpa peak under single electron models are certainly reasonable in physical pictures , which is also agrees with the previous arguments on this problem.@xcite the nonequivalence between two gauges also provided some evidence that the gauge phase factor in the optical response theory can not be ignored , and the static current expressions@xcite can not be directly used and return the exact same results as @xmath0 correlation . on one hand , @xmath0 correlation should be much more suitable to study the nonlinear susceptibilities than @xmath118 correlation.@xcite on the other hand , based on the exact expression of dipole operator @xmath6 , we find that polarization operator @xmath2 can not be directly applied in periodic systems where the electrons are certainly not bounded .
and @xmath2 will returns the qualitatively different curves from the rigorous results . from the results we obtained in this paper
, it seems that the direct application of single electron models may not be suitable in explaining for tpa .
it also implies that the size effect@xcite , lifetime considerations@xcite , exciton effects@xcite , quantum fluctuations or disorders should be included to explain this non - trivial tpa .
both ssh and tlm models still can be served as the good basis to include all those interactions .
the analytical solutions to the thg for both ssh and tlm models obtained under the @xmath0 correlation computations , showing nonexistence of the two - photon peak exactly .
the nonequivalent results between @xmath0 and @xmath1 correlations provide an evidence for the possible explanations through gauge phase factor.@xcite in actual practical computations , @xmath0 will be much more suitable for obtaining reasonable results than @xmath1 correlation , however , the replacement of @xmath6 by the polarization operator @xmath2 will result in qualitatively different results especially in periodic systems .
very helpful discussions with professor j.l .
birman , professor y.r .
shen , professor z.g .
shuai , professor z.g .
soos , professor c.q .
wu and dr . z.g .
yu are acknowledged .
this work was supported by chemistry department , new york university , the project @xmath119 and the national natural science foundation of china ( 59790050 , 19874014 ) . , edited by j. messier , f. kajzar , and p. prasad , nato advanced study institute series vel .
e194 ( kluwer , dordrecht , 1991 ) . , edited by j.l .
brdas and r.r .
chance , nato advanced study institute series vol .
e182 ( kluwer , dordrecht , 1990 ) . , edited by a.j .
heegers , j. orenstein and d.r .
ulrich , mrs symposia proceedings no .
109 ( materials research society , pittsburgh , 1988 ) .
z.g.soos , d. mukhopadhyay , a. painelli and a. girlando , in _ handbook of conducting polymers _ , edited by t.a .
skotheim , r.l .
elsenbaumer and j.r .
reynolds ( marcel dekker , inc . , 1998 ) , p.165 .
l. lauchlan , s. etemad , t.c .
chung , a.j .
heeger and a.g .
mcdiarmid , phys . rev .
b * 24 * , 3701 ( 1981 ) .
j. orenstein and g.l .
baker , phys .
* 49 * , 1043 ( 1982 ) .
z. vardeny , e. ehrenfreund and o. brafman , mol .
. cryst . * 117 * , 245 ( 1985 ) .
hayes , i.d.w .
samuel , and r.t .
phillips , phys .
b * 52 * , r11569 ( 1995 )
harrison , g.h .
hayes , r.t .
phillips , and r.h .
friend , phys . rev
. lett . * 77 * , 1881 ( 1996 ) .
garstein , m.j .
rice , and e.m .
conwell , phys .
b * 55 * , 1683 ( 1995 ) . j.l . brdas , j. cornil , and a.j .
heeger , adv .
* 8 * , 447 ( 1996 ) .
a. khler , d.a .
dos santos , d. beljonne , z. shuai , j.l .
brdas , a.b .
holmes , a. kraus , k. mllen , and r.h .
friend , nature , vol . * 392 * , 903 ( 1998 ) .
fann , s. benson , j.m.j .
madey , s. etemad , g.l .
baker , and f. kajzar , phys .
62 * , 1492 ( 1989 ) .
a.j . heeger , d. moses , and m. sinclair , synth . met .
* 17 * , 343 ( 1987 ) .
weikang wu , phys .
lett . * 61 * , 1119 ( 1988 ) .
y.i . dakhnovskii and k.a .
pronin , synth . met .
* 54 * , 295 ( 1993 ) .
z.g . soos and s. ramasesha , j. chem .
phys . * 90 * , 1067 ( 1989 ) .
j. yu , b. friedman , p.r .
baldwin , and w.p .
su , phys .
b * 39 * , 12814 ( 1989 ) .
j. yu and w.p .
su , phys .
b * 44 * , 13315 ( 1991 ) .
c.q . wu and x. sun , phys .
b * 41 * , 12845 ( 1990 )
. c.q . wu and x. sun , phys .
b * 42 * , r9736 ( 1990 ) .
z. shuai and j.l .
brdas , phys .
b * 44 * , r5962 ( 1991 ) .
f.c . spano and z.g .
soos , j. chem .
phys . * 99 * , 9265 ( 1993 ) .
f. guo , d. guo , and s. mazumdar , phys .
b * 49 * , 10102 ( 1994 ) .
keldysh , zh .
* 47 * , 1515 ( 1964 ) [ sov .
jetp * 20 * , 1018 ( 1965 ) ] .
schrieffer , and a.j .
heeger , phys .
lett . * 42 * , 1698 ( 1979 ) ; phys .
b * 22 * , 2099 ( 1980 ) .
h. takayama , y.r .
lin - liu , and k. maki , phys . rev .
b * 21 * , 2388 ( 1980 ) .
butcher and d. cotter , _ the elements of nonlinear optics _ ( cambridge university press , 1990 )
. n. bloembergen , _ nonlinear optics _ ( w.a . benjamin , inc , 1977 ) y.r .
shen , _ the principles of nonlinear optics _ ( john wiley @xmath120 sons , inc , 1984 ) .
v.m . genkins and p. mednis , sov .
jetp * 27 * , 609 ( 1968 ) g.p .
agrawal , c. cojan , and c. flytzainis , phys .
b * 17 * , 776 ( 1978 ) .
b.j . orr and j.f .
ward , mol . phys . * 20 * , 513 ( 1971 ) .
f. gebhard , k. bott , m. scheidler , p. thomas , and s.w .
koch , phil . mag .
b , vol . * 75 * , no.1 , 1 - 12 ( 1997 ) .
s. etemad and z.g .
soos in _ nonlinear optical spectroscopy in advances in spectroscopy _ , vol .
* 19 * , edited by r.j.h .
clark and hester ( wiley , 1991 ) .
s. tretiak , v. chernyak , and s. mukamel , phys .
lett . * 77 * , 4656 ( 1996 ) .
minzhong xu and xin sun , to be submitted , see part i. j. callaway , _ quantum theory of the solid state _
, second edition ( academic press , inc , 1991 ) , p.483 .
i.s . gradshteyn and i.m .
ryzhik , _ table of integrals , series , and products _
( academis press , inc , 1980 ) , p.904 . | the analytical solutions for the third - harmonic generation ( thg ) on infinite chains in both su - shrieffer - heeger ( ssh ) and takayama - lin - liu - maki ( tlm ) models of trans - polyacetylene are obtained through the scheme of dipole - dipole ( @xmath0 ) correlation .
they are not equivalent to the results obtained through static current - current ( @xmath1 ) correlation or under polarization operator @xmath2 .
the van hove singularity disappears exactly in the analytical forms , showing that the experimentally observed two - photon absorption peak ( tpa ) in thg may not be directly explained by the single electron models . |
for three decades considerable effort has been spent in obtaining samples of agn to investigate their cosmological evolution .
this evolution is seen as a change in the luminosity function with redshift . in recent years
, galaxy redshift surveys and x ray cluster surveys have become sufficiently deep that the evolution of galaxies and clusters of galaxies can be examined in the same way as that of agn .
quantifying and understanding the evolution of these populations is essential to modern cosmology .
a number of methods exist for demonstrating and quantifying the cosmological evolution of a population , eg the @xmath1 method of schmidt ( 1968 ) or by comparison with monte carlo simulations ( la franca & cristiani 1997 ) .
the simplest , most intuitive method is to construct a binned differential luminosity function in a number of redshift intervals . for flux limited samples ( which are the norm in evolution studies ) , this is usually done using the @xmath0 method ( eg maccacaro et al .
1991 , ellis et al . 1996 ) .
however , the @xmath0 method introduces a significant error for objects close to the flux limit , and hence is _ not _ the optimum method for constructing binned differential luminosity functions .
we present a simple , alternative method which we show is superior to @xmath0 .
this paper is laid out as follows .
section [ sec : method ] describes both the @xmath0 method , and our new method , for constructing binned differential luminosity functions .
the relative merits of the two methods are compared in section [ sec : advantages ] , and in section [ sec : sim ] a monte carlo simulation is used to demonstrate the improvement that our method offers over the @xmath0 method . in section [ sec : binmod ] we present a technique for comparing binned and model luminosity functions .
we apply the two different methods to construct binned luminosity functions of the large bright qso survey sample in section [ sec : lbqs ] . in section [ sec : lbqsmod ] this sample is compared to one of the model luminosity functions and evolution laws from boyle et al .
( 1991 ) using both an averaged luminosity function and using the method given in section [ sec : binmod ] .
our conclusions are presented in section [ sec : conclusions ] .
we define the differential luminosity function @xmath2 , of any extragalactic population , as the number of objects per unit comoving volume per unit luminosity interval , i.e. @xmath3 where @xmath4 is the number of objects of luminosity @xmath5 found in comoving volume @xmath6 at redshift @xmath7 .
we assume @xmath2 is a continuous function over the range of redshift and luminosity for which it is defined .
the differential luminosity function is often defined as a function of logarithmic luminosity or ( by optical astronomers ) magnitude rather than luminosity .
the method described here is equally applicable to luminosity functions defined in this way , with the appropriate substitution of ` log @xmath8 ' or ` @xmath9 ' where we have used ` @xmath5 ' .
the @xmath0 method was originally proposed ( schmidt 1968 ) to measure space density ( @xmath10 ) ; the method was generalised for samples with multiple flux limits in avni & bahcall ( 1980 ) . if @xmath4 objects have luminosities and redshifts in the interval @xmath11 around the bin centre @xmath12 : @xmath13 where @xmath14 is the survey volume in which object @xmath15 with luminosity @xmath16 could have been detected and remained in the redshift bin @xmath17 .
it is an unbiased estimator of the space density ( felten 1976 ) ; the main advantage of this method ( compared with simply dividing the number of objects found by the average volume searched ) is that it takes account of the fact that more luminous objects are detectable over a larger volume than the less luminous objects .
the @xmath0 method is used to estimate @xmath2 by dividing the space density by the luminosity bin width .
we will refer to the approximation of @xmath2 obtained using this method as @xmath18 : @xmath19 the luminosity function as defined in equation [ eq : phi1 ] is related to the expected number of objects found in any region @xmath20 of the volume - luminosity plane by @xmath21 where @xmath22 is the bottom of the redshift interval @xmath17 and @xmath23 is the highest redshift possible for an object of luminosity l to be detected and remain in the redshift interval @xmath24 .
angled brackets denote expectation value .
if @xmath2 changes little over @xmath11 , then @xmath25 it follows that if @xmath4 objects have been found over some volume - luminosity region @xmath26 , @xmath27 where @xmath28 is our binned estimate of the luminosity function .
consider a single bin , corresponding to the redshift - luminosity interval @xmath29 around the point ( @xmath30 ) , in a binned luminosity function constructed from a flux limited sample of objects .
we assume that @xmath31 and @xmath17 are sufficiently small and/or that @xmath2 is a sufficiently slowly varying function of @xmath5 and @xmath7 that our single estimate of @xmath2 is appropriate over the whole interval .
some portion of the region @xmath32 may represent objects which are fainter than the survey flux limit .
this situation is illustrated in fig .
[ fig : redlum ] where the curve @xmath33 represents the flux limit of the survey . the shaded region in fig .
[ fig : redlum]a is that region of the volume luminosity plane in the interval @xmath34 which has been surveyed .
this shaded area corresponds to the double integral in equation [ eq : small ] . by definition ( equation [ eq : approx ] )
, @xmath28 gives a good estimate to @xmath35 .
the shaded region in fig .
[ fig : redlum]b has an area equal to ( @xmath36 ) for object i , represented by the black spot .
this area is clearly _ not _ the same as that of the surveyed region ( the shaded region in fig .
[ fig : redlum]a ) and hence there is no guarantee that @xmath18 will give a good estimate of @xmath2 . however , there are two specific cases in which @xmath18 will always give a good estimate of @xmath2 : \1 ) the entire redshift , luminosity interval @xmath29 corresponds to objects brighter than the flux limit of the survey , i.e. in fig .
[ fig : redlum ] the curve @xmath33 passes above the @xmath29 region . in this case ,
@xmath37 and @xmath38 \2 ) @xmath31 is very small : @xmath39 note that even in these limiting cases , @xmath18 does not give a better estimate to @xmath2 than @xmath28 ; instead @xmath18 and @xmath28 have the same value in case 1 and converge to the same value in case 2 .
case 1 generally applies for luminosity functions of objects which are much brighter than the flux limits .
hence for luminosity functions in a number of redshift intervals , the two methods will produce the same results for the highest luminosity bins of any given redshift shell . at the lowest luminosities in each redshift shell , where objects are close to the survey flux limit and the shaded region of volume - luminosity space in fig.[fig :
redlum]a is much smaller than the rectangular @xmath40 region , the two methods are likely to give the most different results . in all cases ,
@xmath28 can be expected to give as good or better estimates for @xmath2 than @xmath18 , and is therefore the better estimator . from the definition of @xmath28 in section [ sec : approx ] it is easily shown that the statistical uncertainty , @xmath41 , is given by : @xmath42 where @xmath43 is the uncertainty on @xmath4 objects and can be calculated from poisson or gaussian statistics as appropriate .
the statistical uncertainty on @xmath18 is harder to estimate , because each object makes a different statistical contribution .
the uncertainty is typically estimated by the following expression : @xmath44 ( eg marshall 1985 , boyle et al .
1988 ) but this assumes gaussian statistics and therefore is not appropriate for bins containing few objects .
hence the uncertainty on @xmath28 is easily calculated for any number of objects within a bin , but the uncertainty on @xmath18 can only be properly estimated when there are many sources per bin . the comparison so far and fig .
[ fig : redlum ] has assumed the simple case of a survey with a single flux limit .
however , many surveys have different flux limits for different parts of the survey area , and most luminosity function investigations combine different surveys with different flux limits . indeed , combining large area shallow surveys and small area deep surveys is currently the only practical way to obtain a luminosity function which spans a wide range of luminosities at a wide range of redshifts
. the most efficient means to combine areas with different flux limits is to assume that each object could be found in any of the survey areas for which it is brighter than the corresponding flux limit .
this is coherent addition of samples in the language of avni & bahcall ( 1980 ) . in this case
, it is no longer true that only the lowest luminosity objects in each redshift shell are close to the flux limit , because objects of higher luminosity are close to the flux limit of the shallower survey areas .
the problem with @xmath18 therefore affects luminosity bins that correspond to any of the survey flux limits , although the severity of the problem is watered down by the deeper survey areas .
many surveys also have a _ maximum _
flux limit beyond which sources are too bright to have been selected . in this case @xmath22 in equations [ eq : phi1 ] - [ eq : approxerror ] , as well as @xmath45 , is a function of l. as an illustration , if the curve @xmath33 in fig .
[ fig : redlum ] were to represent a maximum ( rather than a minimum ) flux limit , objects would have to lie on the other side of the curve to be included in the survey . if the object ( black dot ) in fig .
[ fig : redlum ] were shifted straight upwards to lie above the black line , it is now the shaded area of fig .
[ fig : redlum]a subtracted from @xmath46 which corresponds to the double integral in equation [ eq : small ] , and the shaded area of fig .
[ fig : redlum]b subtracted from @xmath46 . which corresponds to ( @xmath47 ) .
just as for the minimum flux limit ( section [ sec : which ] ) , these two quantities are different , and there is no guarantee that @xmath18 will give a good estimate of @xmath2 .
therefore , for surveys with both bright and faint flux limits , @xmath18 is likely to be inferior to @xmath28 at the highest as well as the lowest luminosity bins in each redshift shell .
in this section we use a sample simulated by a monte carlo method to demonstrate the improvement that @xmath28 offers over the traditional @xmath18 .
the simulation was performed using a two power law model x ray luminosity function which is unchanging with redshift ( i.e. no evolution ) and a single flux limit .
the monte carlo simulation produced a source list of @xmath48 sources .
binned luminosity functions were produced for this simulated data in a range of redshift intervals using both methods .
these are shown in fig .
[ fig : bothsim ] , @xmath28 on the left and @xmath18 on the right .
the model luminosity function is shown as a dashed line .
the two methods are in exact agreement for the high luminosity points of each redshift interval , as expected from sec [ sec : which ] .
however , for the lowest luminosity bins of the @xmath49 and @xmath50 redshift intervals @xmath18 is significantly smaller than the input model , while @xmath28 is a good representation of the input model . for the lowest luminosity bin of the @xmath51 redshift shell
@xmath18 is larger than the input model .
we therefore see that for objects close to the flux limit @xmath18 can be too small or it can be too large ; in simulations we have performed using different flux limits and luminosity functions @xmath18 is more frequently too small than too large .
the luminosity function is frequently modelled as some analytical function ( eg a broken power law or a schechter function ) .
models are sometimes fitted to the binned luminosity function and tested for goodness of fit using @xmath52 .
more often the fit is made to the unbinned data by maximum likelihood ( crawford , jauncey & murdoch 1970 ) and models are tested for overall goodness of fit using a one or two dimensional kolmogorov smirnov test ( press et al .
however , even in this latter case , comparison between the binned and model luminosity function is the usual recourse to find out _ why _ models are rejected , i.e. at which redshifts and luminosities the data and model are inconsistent . in this comparison
, a problem can arise that the model is a continuous function and ( for a steep luminosity function ) can span a rather large range of values within any one luminosity bin .
similarly , any model luminosity function which is evolving with redshift can have a large range of values for a single luminosity over a redshift interval ( although this latter effect can be overcome if the evolution is known or assumed a - priori , eg mathez et al .
1996 , kassiola & mathez 1990 ) .
the binned luminosity function could be compared to the model luminosity function evaluated at some arbitrary ` middle point ' of each luminosity / redshift bin ; alternatively some weighted average of the model could be produced for each luminosity redshift bin .
a more valid statistical approach is to compute the expectation value of the binned luminosity function from the model , for each redshift / luminosity bin . in this method ,
a single , unique , model value is produced for each redshift / luminosity bin , which can be compared to the corresponding binned luminosity function data point .
this allows the binned luminosity function to be compared with the model using a statistical goodness of fit test such as @xmath52 , and/or enables the fitting of a model to the binned luminosity function .
this approach is illustrated in this section using the monte carlo simulated sample of section [ sec : sim ] .
first , the formulas for expectation value are given .
the expectation value of @xmath18 is given by : @xmath54 where @xmath55 and @xmath56 are respectively the minimum and maximum luminosities of objects within the luminosity bin and @xmath57 is the probability density corresponding to an object of luminosity @xmath5 .
@xmath57 is given by @xmath58 where @xmath22 is the lower limit of the redshift shell and @xmath23 is the smaller of the maximum detectable redshift of an object of luminosity l and the top of the redshift shell .
@xmath59 is given by equation [ eq : expectation ] .
hence @xmath60 from equations [ eq : expectation ] and [ eq : approx ] , the expectation value of @xmath28 is given by : @xmath62 [ sec : expsim ] in fig .
[ fig : bothexp ] @xmath64 and @xmath65 are shown with @xmath18 and @xmath28 of the monte carlo simulated sample of section [ sec : sim ] .
unsurprisingly , the binned luminosity function of the monte carlo simulated sample is consistent with the expectation luminosity function derived from the input model .
note that the systematic differences between @xmath18 and @xmath2 ( the turn down of @xmath18 at low luminosities in fig .
[ fig : bothsim ] ) are reproduced by @xmath64 .
hence it is in principle possible to compare and/or fit a model @xmath2 to a binned @xmath18 provided that the model is first transformed into @xmath64 .
however , it is simpler to evaluate equation [ eq : modpredest ] for arbitrary @xmath66 than equation [ eq : modpredva ] , so we are again led to the conclusion that @xmath28 is superior to @xmath18 .
to demonstrate the difference between binned luminosity functions produced using @xmath28 and @xmath18 , we now apply these two methods to real data . for this
we have chosen the large bright qso survey ( hereafter lbqs , hewett , foltz & chaffee 1995 ) , the current largest single sample of qsos available .
this is ideal because as samples become larger , the systematic problems with @xmath18 become larger than the statistical uncertainty associated with the data .
note that the lbqs has a maximum flux limit as well as minimum flux limits ( see section [ sec : multiple ] ) ; these are given by hewett et al .
( 1993 ) .
@xmath28 and @xmath18 are shown for the lbqs in fig .
[ fig : lbqs ] for @xmath67 and h@xmath68=50 km s@xmath69 mpc@xmath69 .
k - correction was performed using the composite spectrum of cristiani & vio ( 1990 ) .
the contrast is striking : the @xmath18 binned luminosity function gives the misleading impression that evolution is complex and highly luminosity dependent while @xmath28 shows a much simpler picture of the evolution ; the latter could be described approximately by pure luminosity evolution .
we now compare the binned luminosity function @xmath28 of the lbqs with a model luminosity function undergoing pure luminosity evolution . the model chosen is the @xmath67 model from boyle et al .
( 1991 ) , which is a two power law luminosity function described by : @xmath70(\alpha + 1)}\ ] ] @xmath71(\beta + 1 ) } \}^{-1}\ ] ] where @xmath72 mag@xmath69 mpc@xmath73 , @xmath74 and @xmath75 .
the model is subject to pure luminosity evolution such that @xmath76=-22.5 - 8.625 log ( 1+@xmath7 ) , until @xmath77 after which @xmath76 remains constant at @xmath78 . the binned luminosity function is shown in fig .
[ fig : lbqs7wins2 ] ; the different redshift shells have been placed in separate panels for clarity . for comparison
the model is shown as both a volume - weighted average over each redshift shell ( dashed line ) , and as a set of expectation values @xmath65 calculated using equation [ eq : modpredest ] ( solid stepped line ) .
the two arrows near the bottom of each plot show the position of the ` break ' luminosity of the model luminosity function at the top and bottom of the redshift shell .
the advantage of comparing the binned luminosity function to expectation values is well demonstrated by the low luminosity bins of the upper four panels of fig .
[ fig : lbqs7wins2 ] , in which the continuous volume weighted luminosity function makes the data look far more discrepant than it really is . the model from boyle et al .
( 1991 ) has already been compared to luminosity function of the lbqs , by hewett et al .
these authors used cumulative 1/va luminosity functions in the same redshift shells as used here ; these were compared to the model predictions using kolmogorov smirnov tests in each redshift shell .
their most important conclusion was that at @xmath79 the luminosity function systematically steepens relative to the model with increasing @xmath7 , implying that pure luminosity evolution is no longer a viable model for qso evolution .
similar claims for other surveys have been made more recently by goldschmidt & miller ( 1998 ) , la franca & cristiani ( 1997 ) and by hawkins & veron ( 1995 ) .
the binned luminosity function of fig .
[ fig : lbqs7wins2 ] does not support this conclusion . instead
, significant discrepancy between data and model at @xmath79 is always for fainter absolute magnitudes , at or below @xmath76 , and is always in the sense that the model overpredicts the number of low luminosity sources . to quantify this , we calculated the @xmath52 separately for bins brighter , and for bins fainter @xmath76 in the four redshift shells at @xmath79 .
these are tabulated in table [ tab : chisq ] ; the lowest and highest luminosity bins of the @xmath49 redshift shell and the highest luminosity bin of the @xmath80 redshift shell have not been included because they contain too few objects for @xmath52 to be appropriate . at absolute magnitudes brighter than @xmath76
the model is acceptable at 95% , overall and for each redshift shell individually . at fainter absolute magnitudes the model is significantly deviant for all but the @xmath80 redshift shell , and it is notable that in this @xmath80 shell the model shows the largest deficiency relative to the data at high luminosities .
although the discrepancy between data and model at low luminosities is sufficient to reject the model at @xmath81 , there is no strong evidence that the luminosity function steepens relative to the model as redshift increases .
ccccc & & + redshift shell&@xmath82&p@xmath83&@xmath82&p@xmath83 + & & & & + @xmath49&2.89/2&0.24&58.70/3&1.1@xmath84 + @xmath50&3.61/2&0.16&14.72/3&2.1@xmath85 + @xmath80&3.72/2&0.16&0.68/2&0.71 + @xmath86&3.30/2&0.19&19.21/2&6.7@xmath87 + total & 13.52/8&0.10&93.31/10&1.2@xmath88 + & & & & + + +
we demonstrate that the @xmath0 method can lead to systematic errors when used to produce binned differential luminosity functions .
this problem is most significant for objects close to their parent sample s flux limit(s ) . as a result
the lowest luminosity bins of @xmath0 luminosity functions which are split by redshift can be unrepresentative , distorting the apparent evolution of extragalactic populations .
a new method for constructing binned luminosity functions , which does not have this problem , is presented .
the improvement of this new method over the @xmath0 method is demonstrated using a monte carlo simulated sample of objects .
this new method also has the advantages that statistical uncertainty is easily estimated even when there are few objects per bin .
we also present a practical method for comparing binned and model luminosity functions ( by eye or by statistical test ) which resolves the problem that the model luminosity function has many values within one luminosity - redshift bin .
we demonstrate the difference between the @xmath0 method and the new method for constructing binned luminosity functions with the large bright qso survey sample .
evolution appears complex and highly luminosity dependent when the @xmath0 method is used , but relatively simple in the binned luminosity function produced using the new method .
we also use the lbqs , along with a model luminosity function and evolution law from boyle et al .
( 1991 ) , to demonstrate the advantages of our method for comparing model and binned luminosity functions .
we show that the model is inconsistent with the data at low luminosities , but unlike hewett et al .
( 1993 ) we do not find strong evidence that the luminosity function steepens relative to the model as redshift increases .
we would like to thank p. hewitt for supplying the lbqs source list in electronic form and for very useful comments that have improved this paper .
fjc thanks the dges for partial financial support , under project pb95 - 0122 . | we show that binned differential luminosity functions constructed using the @xmath0 method have a significant systematic error for objects close to their parent sample s flux limit(s ) .
this is particularly noticeable when luminosity functions are produced for a number of different redshift ranges as is common in the study of agn or galaxy evolution .
we present a simple method of constructing a binned luminosity function which overcomes this problem and has a number of other advantages over the traditional @xmath0 method .
we also describe a practical method for comparing binned and model luminosity functions , by calculating the expectation values of the binned luminosity function from the model . binned luminosity functions produced by the two methods are compared for simulated data and for the large bright qso survey ( lbqs ) .
it is shown that the @xmath0 method produces a very misleading picture of evolution in the lbqs .
the binned luminosity function of the lbqs is then compared to a model two power law luminosity function undergoing pure luminosity evolution from boyle et al .
( 1991 ) .
the comparison is made using a model luminosity function averaged over each redshift shell , and using the expectation values for the binned luminosity function calculated from the model .
the luminosity function averaged in each redshift shell gives a misleading impression that the model over predicts the number of qsos at low luminosity even when model and data are consistent .
the expectation values show that there are significant differences between model and data : the model overpredicts the number of low luminosity sources at both low and high redshift .
the luminosity function does not appear to steepen relative to the model as redshift increases .
= -0.25 in = -0.5 in = 6.75 in = -0.25 in = 9.50 in galaxies : luminosity function - galaxies : evolution - galaxies : quasars |
the purpose of the present work is the analysis of paramagnetic effects of magnetic field ( zeeman term ) on metastability in the zero - bandwidth limit of the extended hubbard model with pair hopping interaction @xcite ( i.e. the limit of the so - called penson - kolb - hubbard ( pkh ) model ) .
the pkh model is one of the conceptually simplest effective models for studying superconductivity of the narrow band systems with short - range , almost unretarded pairing @xcite .
the model includes a nonlocal pairing mechanism that is distinct from on - site interaction in the attractive hubbard model and that is the driving force of pair formation and also of their condensation .
because of the complexity of the pkh model there are no exact solutions for that model . in this paper
we present the exact results for the pkh model with .
we extend our investigations of the model to the case of finite external magnetic field @xmath0 and concentrate on metastable phases and phase separations in the case of .
our starting point is the model with the hamiltonian given by @xmath6 where , , , and @xmath7 is @xmath8-component of the total spin at @xmath9 site . @xmath10 and @xmath11 denote the creation and annihilation operators , respectively , of an electron with spin at the site @xmath9 , which satisfy standard fermion anticommutation relations : @xmath12 where @xmath13 is the kronecker delta .
@xmath3 is the on - site density interaction , @xmath4 is the intersite charge exchange interaction ( pair hopping ) between nearest neighbours , @xmath14 is the external magnetic field , and @xmath1 is the chemical potential , determining the concentration of electrons by the formula : @xmath15 with .
@xmath16 denotes the average value of the operator @xmath17 in the grand canonical ensemble , and @xmath18 is the total number of lattice sites .
indicates the sum over nearest - neighbour sites @xmath9 and @xmath19 independently .
we also introduce , where @xmath8 is a number of the nearest - neighbour sites .
it is important to mention that model ( [ row : ham1 ] ) on the alternate lattices exhibits two symmetries .
the first one is a symmetry between ( @xmath20-pairing , ss , @xmath21 ) and ( @xmath22-pairing , @xmath22s , , @xmath23 is half of the smallest reciprocal lattice vector ) cases in the absence of the field conjugated with the ss order parameter @xmath24 .
thus in the following we restrict our analysis to the case only .
notice that in the presence of finite single electron hopping the symmetry is broken in the general case @xcite .
the second one is the particle - hole symmetry @xcite , thus the phase diagrams obtained are symmetric with respect to half - filling and they will be presented only in the range and . model ( [ row : ham1 ] ) has been intensively investigated for @xcite as well as for @xcite ( in particular , in the context of the phase separation @xcite ( for , ) and metastable phases @xcite ( for ) ) .
the analysis has been performed within a variational approach ( va ) , which treats the @xmath3 term exactly and the intersite @xmath4 interaction within the mean - field approximation ( mfa ) ( which is a rigorous treatment of the @xmath4 term in the limit of infinite dimensions @xcite ) . as a result for the thermodynamic limit ,
one gets two equations for @xmath2 and @xmath24 , which are solved self - consistently .
equations for the energy and other thermodynamical properties are derived explicitly in refs .
@xcite . in the superconducting ( ss )
phase , whereas in the nonordered ( no ) phase . for
fixed @xmath2 , model ( [ row : ham1 ] ) can exhibit also the phase separation ( ps : ss / no ) , which is a state with two coexisting domains ( ss and no ) with different electron concentrations , @xmath25 and @xmath26 .
the free energy of the ps state is derived in a standard way , using maxwell s construction ( e.g. @xcite ) .
it is important to find homogeneous solutions for all local minima ( even very low ones ) with respect to @xmath27 of grand canonical potential @xmath28 ( or free energy @xmath29 ) if system is considered for fixed @xmath1 ( or @xmath2 ) .
the solution ( of the set of two self - consistent equations for @xmath2 and @xmath24 ) is related to a metastable phase if it corresponds to a ( local ) minimum of @xmath30 ( or @xmath31 ) with respect to @xmath27 and the stability condition ( system with fixed @xmath2 ) is fulfilled .
otherwise , the phase is unstable .
a stable ( homogeneous ) phase is a metastable phase with the lowest free energy ( among all metastable phases and phase separated states ) .
let us distinguish six regions which can occur on the phase diagrams : ( 1 ) only the no phase is stable ; ( 2 ) only the ss phase is stable ; ( 3 ) no(ss ) in the region of the no phase stability the ss phase is metastable ; ( 4 ) ss(no ) the no phase is metastable in the region of the ss phase stability ; ( 5 ) ps(no , ss ) the ps state has a lowest energy , the both homogeneous phases are metastable , and the ss phase has a higher energy than the no phase ; ( 6 ) ps(ss , no ) the same as in region ( 5 ) , but here the no phase has a higher energy than the ss phase . above denotations are used ( interchangeably ) in figs .
[ rys : gs][rys : pshvskt ] .
( a ) and @xmath2 ( b ) ( ) .
details in text in sec .
[ sec : results : gs ] ( cf . also fig .
[ rys : kbtvsmin]).,title="fig : " ] ( a ) and @xmath2 ( b ) ( ) .
details in text in sec .
[ sec : results : gs ] ( cf . also fig .
[ rys : kbtvsmin]).,title="fig : " ] the ground state ( gs ) phase diagrams are presented in fig .
[ rys : gs ] .
notice that metastable phases can occur only at and the boundaries in fig .
[ rys : gs ] of the metastable phases occurrence are the extensions from infinitesimally small , formally . at one phase ( state )
can be stable only . at the gs the discontinuous ss no transition occurs at ( for fixed ) whereas the continuous ss no transition occurs at and .
the ps state ( ss / no ) stability region is determined by conditions : and ( ) . at ( )
the discontinuous ss no transition occur for .
the extension ( to the gs ) of the discontinuous transition line between metastable phases ( ss and no ) is located at ( for fixed @xmath2 ) .
the boundaries for the regions of the metastability of homogeneous phases close to the gs are located at : for the no phase , and ( , any @xmath2 ) ; for the ss phase , and ( and any @xmath2 ) . notice that for the homogeneous ss phase the condition is fulfilled at ( in particular in the ranges of the ps state occurrence ) , whereas in the no phase @xcite .
let us point out that for the discontinuous transition between two no phases with @xmath33 ( empty / full - filled ) and @xmath34 ( half - filled mott state ) occurs at and , but it does not exist for any .
in fact , the homogeneous no phase for is degenerated with the ps state in which two domains of the no phase ( with and for or and ) exists .
this degeneration is removed for any and such a ps state does not exist at . ( a ) and @xmath2 ( b ) .
dotted , solid and dashed lines indicate first - order , second - order and `` third - order '' boundaries , respectively .
dashed - doted lines indicate the boundaries of metastable phase occurrence ( names of metastable phases in brackets ) .
@xmath36 denotes tricritical point .
details in text in sec .
[ sec : results : finite].,title="fig : " ] ( a ) and @xmath2 ( b ) . dotted , solid and dashed lines indicate first - order , second - order and `` third - order '' boundaries , respectively .
dashed - doted lines indicate the boundaries of metastable phase occurrence ( names of metastable phases in brackets ) .
@xmath36 denotes tricritical point .
details in text in sec .
[ sec : results : finite].,title="fig : " ] the complete phase diagram of the model has been determined in @xcite .
the system analysed shows very interesting multicritical behaviour including tricritical points .
depending on the values of model parameters , the system can exhibit not only the homogeneous phases ( ss and no ) , but also the phase separated states ( ps : ss / no ) .
all transition temperatures and the ss phase metastability boundary are decreasing functions of @xmath5 and @xmath37 @xcite . only the no phase metastability boundary can exhibit non - monotonic behaviour @xcite .
let us start the discussion of the behaviour of the system for the case .
as an example , the phase diagrams for and are shown in fig .
[ rys : kbtvsmin ] . the ss no transition with increasing temperature can be second - order ( continuous change of @xmath24 , the transition temperature decreases with increasing @xmath38 and @xmath39 ) as well as first - order ( discontinuous change of @xmath24 , the transition temperature increases with increasing @xmath38 ) .
it is rather obvious that the regions of the metastable phases occurrence are present near the first - order transition ( for fixed @xmath40 ) , i.e. above the transition temperature the ss phase is metastable ( region ( 3 ) ) , whereas below the transition temperature the no phase is metastable ( region ( 4 ) ) , fig .
[ rys : kbtvsmin](a ) .
if the system is analysed for fixed
@xmath2 , the first - order transition line ( for fixed @xmath1 ) splits into two `` third - order '' lines ( and ) @xcite .
both `` third - order '' transition temperatures increase with increasing @xmath39 ( fig .
[ rys : kbtvsmin](b ) ) . at this transition
a size of one domain in the ps state decreases continuously to zero at the transition temperature . in the region of the ps state occurrence ( where the ps state has the lowest energy @xmath41 )
the first - order transition between two metastable ( homogeneous ) phases is present ( the transition temperature increase with increasing @xmath39 ) . below this line
the energy of the no phase is the highest ( i.e. , region ( 6 ) ) , whereas above the line the energy of the ss phase is the highest ( i.e. , region ( 5 ) ) , cf . also .
[ rys : pshvskt ] .
the line of the first order transition between stable phases for fixed @xmath40 ( metastable phases for fixed @xmath2 ) ends at and ( ) .
one metastable phase ( ss or no ) can also be present in the regions of homogeneous phases ( no or ss , respectively ) stability for fixed @xmath2 ( where the ps state does not exist ) , fig .
[ rys : kbtvsmin](b ) . vs. @xmath42 phase diagrams for and ( as labelled ) .
denotations as in fig .
[ rys : kbtvsmin ] .
labels ( 1)(6 ) defined at the beginning of sec . [ sec : results ] . dashed - dotted - dotted lines ( panels ( a),(b ) ) schematically denote location of the no phase metastability boundaries ( which can occur for other values of @xmath5 and @xmath2 than shown ) .
details in text in sec .
[ sec : results : finite].,title="fig : " ] vs. @xmath42 phase diagrams for and ( as labelled ) .
denotations as in fig .
[ rys : kbtvsmin ] .
labels ( 1)(6 ) defined at the beginning of sec . [ sec : results ] . dashed - dotted - dotted lines ( panels ( a),(b ) ) schematically denote location of the no phase metastability boundaries ( which can occur for other values of @xmath5 and @xmath2 than shown ) .
details in text in sec .
[ sec : results : finite].,title="fig : " ] vs. @xmath42 phase diagrams for and ( as labelled ) .
denotations as in fig .
[ rys : kbtvsmin ] .
labels ( 1)(6 ) defined at the beginning of sec . [ sec : results ] .
dashed - dotted - dotted lines ( panels ( a),(b ) ) schematically denote location of the no phase metastability boundaries ( which can occur for other values of @xmath5 and @xmath2 than shown ) .
details in text in sec .
[ sec : results : finite].,title="fig : " ] let us discussed the behaviour of the system with increasing for . in fig .
[ rys : pshvskt ] we present a few exemplary @xmath37 vs. @xmath42 phase diagrams obtained for and fixed @xmath2 ( ) .
one can distinguish five essentially different cases . * for small @xmath2 ( ) , the tricritical @xmath36-point ( associated with a change of the transition order ) is present on the phase diagram ( fig .
[ rys : pshvskt](a ) ) . with increasing @xmath2 the @xmath42-coordinate of the @xmath36-point increases , whereas its @xmath37-coordinate decreases .
at the @xmath36-point is located at .
* for higher @xmath2 ( ) all five boundaries remaining on the phase diagrams ( i.e. two of `` third - order '' , one of first - order between metastable phases , and two of metastable phase occurrence ) end at and do not have any points in common ( fig .
[ rys : pshvskt](b ) ) .
these five boundaries move towards lower values of @xmath37 and @xmath42 with increasing @xmath2 and finally three of them vanish continuously at , whereas two other lines ( i.e. the ps no `` third - order '' line and the ss phase metastability boundary ) are fixed at for . as the first one , the boundary of the no phase metastability vanishes at . * for
there is not any region , where the no phase is unstable ( region ( 2 ) does not occur on the diagram ) . * for the region of the ss phase occurrence does not exist and for low temperatures only the ps state is stable ( the ss phase can be metastable only , region ( 4 ) is not present on the diagram , fig .
[ rys : pshvskt](c ) ) . * at the boundary of the ss no first - order transition between metastable phase disappears and for below the line only region ( 5 ) occurs , which shrinks towards lower @xmath43 with increasing @xmath2 . at the ps state
does not occur , there is no transitions with increasing @xmath42 ( the no phase has always the lowest energy ) and the ss phase is metastable in a certain range of model parameters @xmath37 and @xmath42 .
notice that the boundary of the ss phase metastability is independent of @xmath2 ( and @xmath40 ) and this line is a projection of the tricritical line ( on the @xmath37@xmath42 plane in the case of the @xmath37 vs. @xmath42 diagrams ) . for small @xmath2 , in the presence of @xmath36-point , the boundary does not exist for low @xmath37 ( fig .
[ rys : pshvskt](a ) ) . for smaller values of not all behaviours ( i)(v ) discussed above and presented in figs .
[ rys : pshvskt](a - c ) occur and four new types of phase diagrams can appear .
two of them are ( i ) and ( ii ) , which structures are similar to these of ( i ) and ( ii ) , respectively .
the only difference is that the both ends of the no phase metastability boundary are located at ( in particular , the end at lower @xmath43 is for ) and the no phase is unstable ( region ( 2 ) ) only at some ( cf .
lower dashed - dotted - dotted lines in figs .
[ rys : pshvskt](a , b ) ) .
moreover , the boundary of the no phase metastability can be non - monotonic ( as in the cases ( i ) and ( ii ) ) and can end at and ( as in the cases ( i ) and ( ii ) ) .
thus two more new types : ( i ) and ( ii ) can occur , cf .
upper dashed - dotted - dotted lines in fig .
[ rys : pshvskt](a , b ) . in general
, there is a continuous change from ( i ) and ( ii ) cases to ( i ) and ( ii ) cases , respectively .
in particular , the following sequences of structures of diagrams with increasing occur ( some of them in very narrow ranges of model parameters ) : * for : only case ( i ) ; * for : ( i ) , ( i ) , and ( i ) ( because the ss no transition is always second - order at for , cf . fig .
[ rys : pshvskt](a ) ) ; * for : ( i ) , ( i ) , ( i ) , and ( ii ) ( for the @xmath36-point can occur at ) ; * for : ( i ) , ( i ) , ( ii ) , and ( ii ) ( the @xmath36-point at moves toward lower @xmath2 with increasing @xcite ) * for : ( i ) , ( i ) , ( ii ) , ( ii ) , and ( iii ) ( for the no phase is metastable for any @xmath43 near for @xcite and case ( iii ) appears ) ; * for : ( i ) , ( ii ) , ( ii ) , ( ii ) , and ( iii ) ; * for : ( i ) , ( ii ) , ( ii ) , and ( iii ) ( no re - entrance of region ( 2 ) for with increasing @xmath43 , the boundary of the no phase metastability is monotonic at for @xcite ) . on the contrary to the case discussed previously , in above cases ( a)(g ) `` third - order '' and first - order ( metastable ) boundaries connect together at . for and the first - order ss no transition between stable phases is present ( cf .
fig . 5 in @xcite ) and the metastable phases exists in the neighbourhood of this transition ( cf . also fig . 1 in @xcite ) .
at the ps state does not occur .
the no metastability boundary vanishes continuously to a point at ( ) ( the change from ( ii ) to ( iii ) ) or at ( ) ( the change from ( ii ) to ( iii ) ) . for
only the no phase is stable .
there is no metastable phases and no transitions with increasing @xmath42 .
notice that determined regions of metastable phases occurrence can be smaller than they actually are due to finite numerical accuracy of local minima finding .
this issue can be crucial for determination of ( ii ) case occurrence , in particular for .
the superconductivity with extremely short coherence length and the phase separation phenomenon involving ss states are very current topics ( for a review see @xcite and references therein ) .
it is worthwhile to notice that metastable and unstable states as well as phase separation have been found in many physical systems experimentally and theoretically .
note that the temperature dependence of the upper critical field in unconventional superconductors has a positive curvature in coincidence with results of fig .
[ rys : pshvskt ] ( cf .
ps no `` third - order '' line ) .
obviously the macroscopic ps state founded is different from the abrikosov - shubnikov mixed state in type - ii superconductors
@xcite , e.g. no vortex lattice , no magnetic flux quantization , etc . the results presented in this paper are an extension of our previous investigations of model ( [ row : ham1 ] ) to the case of involving the consideration of metastable phases and phase separation . model ( [ row : ham1 ] ) can be considered as a relatively simple , effective model of a superconductor with local electron pairing @xcite .
moreover , the knowledge of the exact results for the limit of the pkh model can be used as a starting point for a perturbation expansion in powers of the hopping @xmath44 and provides a benchmark for various approximate approaches analysing the corresponding finite bandwidth models . in the model considered the external magnetic field only acts on the spin through the zeeman term ( the paramagnetic effect ) . at arbitrary the system is spin polarized with non - zero magnetization if .
a curious issue of the orbital contribution through the pair hopping ( the diamagnetic effect ) @xcite is left for future investigations .
nevertheless , in materials with heavy electron mass ( narrow bands ) or multiple small fermi pockets the paramagnetic effect becomes crucial . for interacting fermions on non - rotating optical lattices also only paramagnetic effect can occur .
the interplay and competition between superconductivity and intersite magnetic @xcite or density - density @xcite interactions is a very interesting problem .
some results concerning the interplay of these interactions with the pair hopping term for have been presented in @xcite .
the author is indebted to professor stanisaw robaszkiewicz for very fruitful discussions during this work and careful reading of the manuscript .
the work has been financed by national science centre ( ncn , poland ) as a research project under grant no .
dec-2011/01/n / st3/00413 and a doctoral scholarship no .
dec-2013/08/t / st3/00012 .
the author also thanks the european commission and the ministry of science and higher education ( poland ) for the partial financial support from the european social fund operational programme `` human capital''pokl.04.01.01 - 00 - 133/09 - 00``__proinnowacyjne ksztacenie , kompetentna kadra , absolwenci przyszoci _ _ '' as well as the foundation of adam mickiewicz university in pozna for the support from its scholarship programme .
kapcia , phd thesis , adam mickiewicz university in pozna , pozna , march 2014 , http://hdl.handle.net/10593/10857 .
a. hui , s. doniach , _ phys .
b _ * 48 * , 2063 ( 1993 ) .
doi : 10.1103/physrevb.48.2063 a. ptok , m. m. maka , m. mierzejewski , _ j. phys .
: condens . matter _ * 21 * , 295601 ( 2009 ) .
doi : 10.1088/0953 - 8984/21/29/295601 a. ptok , m. m. maka , m. mierzejewski , _ phys . rev .
b _ * 84 * , 094526 ( 2011 ) .
doi : 10.1103/physrevb.84.094526 | we present the analysis of paramagnetic effects of magnetic field ( @xmath0 ) ( zeeman term ) in the zero - bandwidth limit of the extended hubbard model for arbitrary chemical potential @xmath1 and electron density @xmath2 .
the effective hamiltonian considered consists of the on - site interaction @xmath3 and the intersite charge exchange term @xmath4 , determining the hopping of electron pairs between nearest - neighbour sites .
the model has been analyzed within the variational approach , which treats the on - site interaction term exactly and the intersite interactions within the mean - field approximation ( rigorous in the limit of infinite dimensions ) . in this report
we focus on metastable phases as well as phase separated ( ps ) states involving superconducting ( ss ) and nonordered ( no ) phases and determine their ranges of occurrence for ( ) in the presence of magnetic field .
our investigations of the general case for arbitrary @xmath5 show that , depending on the values of interaction parameters ( for fixed @xmath2 ) , the ps state can occur in higher fields than the homogeneous ss phase ( field - induced ps ) .
moreover , a first - order ss no transition occurs between metastable phases and these metastable phases can exist inside the regions of the ps state stability .
such behaviour is associated with the presence of tricritical line on the phase diagrams of the system . |
the recent precision measurement of the muon anomaly @xmath6 at the brookhaven national laboratory @xcite has led to renewed interest in accurate measurements of the cross section for @xmath9 annihilation into hadrons .
contributions to the photon spectral functions due to quark loops , are not calculable for low hadronic mass states because of the failure of perturbative qcd in such conditions .
a very clean way out has been known for a long time @xcite .
the imaginary part of the hadronic piece of the spectral function is connected by unitarity to the cross section for @xmath0hadrons .
a dispersion relation can thus be derived , giving the contribution to @xmath6 as an integral over the hadronic cross section multiplied by an appropriate kernel .
an example of a first complete estimate of the correction was given in 1985 , @xmath10=707(6)(17),-10 , @xcite .
the process @xmath0 contributes 500 out of the 700 value above .
the cross section for @xmath0 becomes negligible above 1 gev .
the most recent measurements of ( @xmath0 ) in the energy interval 610@xmath11961 mev come from cmd-2 at novosibirsk . they claim a systematic error of 0.6% with a statistical error of 0.7% @xcite .
their data have been used most recently together with @xmath12 and data up to 3 gev , in an attempt to produce a firm prediction for comparison with the bnl result@xcite .
there is unfortunately a rather strong disagreement between the @xmath13 value obtained using @xmath0 data and @xmath12 decay data , after isospin corrections .
finally the @xmath0 based result disagrees by 3 with the bnl @xcite measurements .
there are thus many reasons for new measurements of the annihilation cross section into two pions .
initial state radiation , isr , is a convenient mechanism which allows studying @xmath0 hadrons , over the entire energy range from 2@xmath14 to @xmath15 , the center of mass energy of the collision . in the case of interest
it is potentially vitiated by the possibility of final state radiation . for a photon radiated before annihilation of the pair ,
the system energy is @xmath16 , thus one measures the coupling to pions of an off - mass shell photon of mass m ( ) . for a photon radiated by the final state pions , the virtual photon coupling to the pair has a mass @xmath15 . by just counting powers of @xmath17 , the relative probability of isr and fsr
are equal .
this requires very careful estimates of the two processes in order to be able to use the reaction @xmath0 to extract ( @xmath0 ) .
the karlsruhe group has given much attention to this problem and has also developed the eva and phokhara monte carlo programs @xcite - @xcite which are fundamental to our analysis . in the following
we will refer to the invariant mass squared of the system as @xmath18 .
in general the and cross section are related through : @xmath19 integrating the cross section above , for @xmath20 and @xmath21 we get @xmath22 eq . [ eq : h ] defines the radiator function @xmath23 . in the following we will drop the variable @xmath24 which is a constant in the present work . the radiator function @xmath25 used in our analysis is obtained from the phokhara monte carlo program .
our present analysis is based on the observation of ref.@xcite , that for small polar angle of the radiated photon , the isr process vastly dominates over the fsr process . at lowest order
the cross section diverges as @xmath26 , just like ( @xmath0 ) .
we limit ourselves in the following to studying the reaction @xmath0 with @xmath27 or @xmath28 . for small @xmath29 ,
the di - pion system recoiling against a small angle photon will results in one or both pions being lost also at small angle .
we are therefore limited to measuring ( ) for @xmath30550 mev
. we will be able to investigate the cross section near threshold , as soon as next to leading order calculation for fsr become available .
this is of great importance , since there are no recent , good measurements of ( ) at low mass , which weigh strongly in the estimate of @xmath13 .
the kloe detector @xcite consists of only two elements : a large , precision drift chamber and an electromagnetic calorimeter .
the detector design was optimized for @xmath31 study in neutral kaon decays , for k mesons produced in the decay of @xmath32 mesons almost at rest , @xmath3312.5 mev / c .
the measurements reported in the following rely heavily on the drift chamber .
its large dimensions , 4 m diameter , use of he and a magnetic field of 0.52 t allow measuring momenta and direction of charged particles with very high accuracy .
because of our choice of accepting only events with @xmath34 , the photon does not reach the calorimeter .
the two relevant parameters , @xmath18 and @xmath35 are obtained from the reconstructed tracks of events with two opposite charge particles forming a vertex near the interaction point .
the quality of the detector is fully described by the mass squared resolution and the resolution in measuring the production angle of single tracks or the angle of the vector sum of the two track momenta .
we find @xmath360.2% for @xmath37=500 mev , without much change over the range studied and @xmath3810 mrad , with @xmath39 and @xmath40 .
the momentum resolution is @xmath41 for large angle tracks .
the calorimeter still plays an important role in the measurement .
its exceptional timing resolution , @xmath42 ps and high segmentation provide particle identification allowing us to distinguish pion from electrons and muons , as discussed later .
it is also used in the luminosity measurement , the kloe detector is operated at , the frascati , running at @xmath43 mev .
beams cross in at an angle of 25 mrad , resulting in a momentum of the @xmath32 mesons of 12.5 mev / c along the @xmath44-axis .
we use a coordinate system where the @xmath44-axis points to the center of the colliders , the @xmath45-axis bisects the two beam lines and the @xmath46-axis is vertical
. the precise values of @xmath15 and @xmath33 is measured on a run by run basis using bhabha events , to accuracies significantly better than 100 kev .
the rms machine energy spread is 350 kev , measured with pairs in kloe .
the excellent kloe resolution allows us to efficiently selected events , without photon detection , because of their clean signature : two tracks from the interaction point .
we assume that the event is of the type @xmath0@xmath47 , with @xmath48 and @xmath49 . the undetected photon polar angle
, @xmath50 is the angle with the @xmath45-axis of the vector -*p*+-*p*- .
the fiducial volume for events has been chosen in order to maximize the acceptance for events due to isr .
the main background processes are @xmath0@xmath0 and radiative bhabha scattering . at low mass
the final state @xmath51 becomes important but requiring small @xmath35 removes most of this final state .
we rely otherwise on kinematics and particle identification , i d , in order to reject the mentioned backgrounds to a negligible level . from studies with eva ,
confirmed with phokhara , a good compromise for the fiducial volume is @xmath34 and @xmath52 for the polar angle of both tracks at the interaction point .
these requirements ensure abundant statistics for the measurement and contamination of fsr hard radiation well below the 1% level .
the efficiency for the events falling inside the fiducial cuts is contained in the value of the radiator , obtained from phokhara .
we note that no detected photon is required , nor are events rejected if photons are present .
the two tracks must form a vertex close to the interaction point .
we require that the vertex coordinates @xmath53 satisfy @xmath54 cm and @xmath55 cm .
we also require @xmath56 or @xmath57 mev / c to remove spiraling tracks .
90 to 95% of the events survive these requirements .
this result is obtained from monte carlo studies and examination of @xmath0 decays .
a fraction of radiative bhabha events , , survives kinematical selection , see sec .
[ sec : kincut ] below , contributing a non - negligible background . in order to reject these events ,
we have developed a particle i d method , using approximate likelihood estimators which are effective in distinguishing pion and electrons .
the likelihood is based on time - of - flight vs momentum comparison and on the shape and magnitude of the energy deposits in the calorimeter by the charged particles .
correlations , which are significant , between used variables are ignored .
two likelihood functions have been built using actual data : one , @xmath58 to test the pion hypothesis and one for @xmath59 , since the energy deposition of the two particles is different , in particular at these energies .
this is not necessary for electrons . ]
( using @xmath60 events ) and the other , @xmath61 to test the electron hypothesis ( using bhabha events ) .
we require that candidate pion tracks have a greater likelihood of being pions than of being electrons .
we use as discriminating variable @xmath62 and we accept as pions particles with @xmath63 , see fig .
[ likelihood ] .
the efficiency for pions is 98% and 3% of the electrons survive rejection .
we require that only one of the two particles satisfy the pion i d requirement ( or ) achieving an efficiency of 100% with some very small residual background .
requiring that both tracks are pions ( and ) leads to a 5% loss .
we use the and for evaluating backgrounds , see later . for an @xmath64 final state lorentz invariance
allows us to compute the mass of @xmath44 , which we call @xmath65 from @xmath66 for @xmath67 , the value of @xmath65 peaks at @xmath68 thus allowing selection of signal events .
the density distribution of the two track events in the [ @xmath69 plane is very effective for separating signal from background . the distribution is shown in fig .
[ fig : mtrkvsq2 ] together with the selection cut defined by : @xmath70 , all units in mev .
events populate a curve in the plane .
the signal events above the region @xmath71 mev are due to additional isr or fsr radiation .
the cut on @xmath65 rejects a fraction of events with additional radiation .
the implications are discussed in section 4.4 .
data were taken in july through december 2001 , for an integrated luminosity of =140.7 pb . after fiducial volume and selection cuts we find 1.5,6 , events .
[ fig : nraw ] gives the mass squared distribution of the events in bins of 0.01 gev in @xmath18 .
the @xmath72 peak and the @xmath72-@xmath73 interference structure are clearly visible , demonstrating the excellent mass resolution of the kloe detector . to obtain the cross section
, we subtract the residual background from this spectrum and divide by the selection efficiency and the integrated luminosity : @xmath74 the [ @xmath69 plane population of signal and background events is illustrated in fig .
[ fig : mtrkvsq2 ] .
the amount of background in the signal region is obtained by fitting the @xmath65 spectrum of the selected events , in slices of @xmath18 with the @xmath65 spectra for signal , @xmath75 and @xmath76 events obtained from monte carlo simulation .
an example of such a fit is shown in fig .
[ fig : bkg1 ] , left .
background from @xmath77 events appears at higher @xmath65 values and the missing mass , @xmath78 , peaks at @xmath79 .
the @xmath77 background count is obtained by fitting the @xmath80 distribution with the shapes obtained from the monte carlo simulation .
an example of such a fit is shown in fig .
[ fig : bkg1 ] , right .
the result of the fits is shown in fig .
[ fig : bkg2 ] which shows the relative contribution of the different backgrounds to the observed signal .
the shape of the background distribution is well reproduced by the monte carlo simulation ensuring that systematic uncertainties are smaller than the fit errors . other possible contributions , such as the reaction @xmath0 with the electrons going down the beam pipe , have been studied and shown to be negligible given our event selection criteria .
the integrated luminosity is measured using large angle bhabha ( lab ) events , with the kloe detector itself .
the effective bhabha cross section at large angles ( @xmath81 ) is about @xmath82 nb .
this cross section is large enough so that the statistical error on the luminosity measurement is negligible .
the number of lab candidates , @xmath83 , is counted and normalized to the effective bhabha cross section , obtained by monte carlo : @xmath84 the precision of this measurement depends on the correct inclusion of higher order terms in computing the bhabha cross section .
we use two independent bhabha event generators : bhagenf ( @xcite , @xcite ) and babayaga ( @xcite ) . for each generator a systematic error of @xmath85 is quoted by the authors .
the two result for the luminosity agree to better than 0.2% .
lab events are selected with cuts on variables which are well simulated by the kloe monte carlo simulation .
the electron and positron polar angle cuts , 55@xmath86125 , is based on the calorimeter clusters , while the energy cuts , @xmath87 mev , is based on drift chamber information .
the background from @xmath88 , @xmath89 and @xmath77 events is well below @xmath90 and is subtracted .
all selection efficiencies ( trigger , emc cluster , dc tracking ) are @xmath91 and are well reproduced by the detector simulation program .
we also obtain excellent agreement between the experimental distributions ( @xmath92 , @xmath93 ) and those obtained from monte carlo simulation , see fig . [
fig : bhabha ] .
finally , corrections are applied on a run - by - run basis for fluctuations in the center - of - mass energy of the machine and in the detector calibrations .
the experimental uncertainty in the acceptance due to all these effects is 0.4% .
we assign a total systematic error for the luminosity of @xmath94 .
an independent check of the luminosity measurement is obtained using @xmath0 events .
we find agreement to within @xmath95 .
these efficiencies have been obtained from control samples of , and events selected without requiring the vertex .
the efficiencies have been parameterized as a function of single track parameters ( charge , momentum and angle ) combined using monte carlo simulation .
the largest source of systematic error in this procedure arises from the difference between the kinematics of the control samples and the signal . .
the efficiency for the @xmath96 cut has been evaluated from data by accepting one track and studying the distribution of the other one .
the systematic error introduced by the @xmath96 cut is negligible , since we use the or , at least on particle is a pion , for which the efficiency is 100% for events .
the efficiency for the @xmath65 cut is obtained as a by - product of the residual background evaluation ; the result of the fit provides the efficiency in each @xmath18 bin .
however , this efficiency depends upon the way the monte carlo simulation treats multi - photon processes .
the @xmath65 efficiency has been obtained with our reference monte carlo simulation , which uses the phokhara version 1.0 generator .
phokhara simulates the full nlo qed correction for annihilation with the emission of additional photons from the initial state . in order to estimate the systematic effect due to the generator we have repeated the exercise using a different generator , babayaga ver.3.5 , the same generator used in the luminosity measurement . in this generator
, isr is treated using the parton shower approach .
the resulting value for the @xmath65 efficiency differs from that evaluated with phokhara by @xmath95 .
an additional contribution , which is treated separately , is the emission of both an isr and fsr photon .
this is discussed in the next section .
the contribution of the systematic error for each efficiency to the total error is shown in table [ tab : syseff1 ] .
.list of systematic uncertainties from the three sources : experimental , theoretical and ignoring fsr . [ cols="<,^ " , ] in order to obtain @xmath97 as a function of the true @xmath18 value , we unfold the mass resolution from the measured @xmath18 distribution . the measured value of @xmath98 is related to the true value by a resolution matrix * g*(@xmath99 ) .
this matrix has been generated by monte carlo simulation .
the resolution matrix is nearly diagonal with @xmath100 of the events in the diagonal , @xmath101 in each of the first once - off - diagonal elements and only @xmath102 in the remaining elements . the matrix is graphically illustrated in fig .
[ fig : matrix ] , left .
unfolding is obtained by applying the inverse matrix * g*@xmath103 to the data .
the result is shown in fig .
[ fig : matrix ] , right , which shows the difference between the reconstructed and the generated spectra with and without the application of the unfolding procedure
. this procedure does not affect the value of @xmath6 , since the correlation is short range , @xmath104 of the events being in the same or the adjacent bins .
the cross section for @xmath105 , after applying the corrections described above , is shown in fig . [
fig : sigppg ] . in order to get the @xmath106 cross section ,
the radiator function @xmath107 is needed .
the radiator is obtained from phokhara , setting @xmath108 and _ switching off _ the vacuum polarization of the intermediate photon in the generator .
the radiator @xmath109 is also shown in fig.[fig : sigppg ] , left .
part of the multiphoton events are removed by the @xmath110 cut .
the effect of this cut is evaluated using phokhara 1.0 which , however , does not include soft radiation in the final state accompanying one hard photon from the initial state .
an preliminary estimate done with phokhara - ii , shows that this effect is smaller than 2% .
we _ do not correct _ here for fsr photons and we include in the theoretical systematic error the full value ( 2% ) due to this effect . in order to obtain the pion form factor , or the _ bare _ cross section ,
vacuum polarization effects must be subtracted .
this can be done by correcting the cross section for the running of @xmath17 , see refs .
@xcite , @xcite , as follows : @xmath111 we have used an approximate , but quite adequate for the purpose estimate of @xmath112@xcite .
[ fig : sigppg ] , right , shows the resulting cross section .
we have used the _ bare _ cross section to evaluate of @xmath13 in the region covered by the cmd-2 experiment ( @xmath113 ) . the resulting value ( in 10@xmath114 units )
is : @xmath115 to be compared with : @xmath116 the statistical error is negligible .
the systematic error is at the moment at the @xmath117 value , but there is still room to improve it down to @xmath118 . clearly the dominating error is the one we have conservatively assigned to the fsr radiation effect .
this error will be substantially reduced once the new phokhara version will be inserted into the kloe mc .
the central value of our result in the region @xmath113 is slightly larger than the one obtained by cmd-2 .
the discrepancy is , at the moment , not significant ( 0.5 @xmath119 ) .
this difference is , however , not equally distributed in @xmath18 , as summarized below .
, based on the values tabulated in ref .
@xcite ] @xmath120 our data differ from the cmd-2 results mostly below the @xmath72-peak .
however , for the mass squared range @xmath121 , our data confirm the discrepancy between @xmath12 data and @xmath0 results , which is 10 - 15% .
we thank the da@xmath8ne team for their efforts in maintaining low background running conditions and their collaboration during all data - taking . we want to thank our technical staff : g. f. fortugno for his dedicated work to ensure an efficient operation of the kloe computing center ; m. anelli for his continous support to the gas system and the safety of the detector ; a. balla , m. gatta , g. corradi and g. papalino for the maintenance of the electronics ; m. santoni , g. paoluzzi and r. rosellini for the general support to the detector ; c. pinto ( bari ) , c. pinto ( lecce ) , c. piscitelli and a. rossi for their help during shutdown periods .
this work was supported in part by doe grant de - fg-02 - 97er41027 ; by eurodaphne , contract fmrx - ct98 - 0169 ; by the german federal ministry of education and research ( bmbf ) contract 06-ka-957 ; by graduiertenkolleg ` h.e .
phys . and part
. astrophys . ' of deutsche forschungsgemeinschaft , contract no .
gk 742 ; by intas , contracts 96 - 624 , 99 - 37 ; and by tari , contract hpri - ct-1999 - 00088 . | we have measured the cross section ( @xmath0 ) with the kloe detector at , at an energy @xmath1 gev . from the dependence of the cross section on @xmath2 , where @xmath3 is the energy of the photon radiated from the initial state , we extract ( @xmath0 ) for the mass range @xmath4 gev@xmath5 . from our result we extract the pion form factor and the hadronic contribution to the muon anomaly , @xmath6 . kloe note n@xmath7 189 july 2003 -1 cm , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , |
the magnetophonon resonance ( mpr ) effect occurs when two landau levels are a phonon energy apart which leads to a resonant scattering due to emission or absorption of phonons . since the pioneering work by gurevich and firsov [ 1 ]
, this effect has been extensively studied in bulk [ 2 - 4 ] as well as low - dimensional semiconductor systems.[5 - 10 ] the resonant character makes it a powerful spectroscopic tool .
magnetophonon resonances have been used to obtain information on band structure parameters , such as the effective mass and the energy levels , and on the electron - phonon interaction .
the vast majority of work on mpr was done on the transport properties of semiconductors , usually the magnetoresistance , which inevitably involves a complicated averaging of scattering processes .
the oscillations in the magnetoresistance are the results of a combination of scattering and broadening processes which can lead to a quite complicated dependence of the resonance amplitudes on doping , sample structure , carrier concentration , and temperature .
however , the mpr can also be observed directly through a study of the electron cyclotron resonance ( cr ) linewidth and effective mass , i.e. , the so called optically detected mpr ( odmpr ) .
this was demonstrated by barnes _
[ 11 ] for a two - dimensional ( 2d ) semiconductor system formed in gaas / algaas heterojunctions .
the odmpr allows one to make quantitative measurements of the scattering strength for specific landau levels and yields direct information on the nature of the electron - phonon interaction in semiconductors . in this work
, we extend the theory for odmpr to three - dimensional ( 3d ) systems and present the first theoretical study of the magnetophonon resonances in the frequency - dependent conductivity in bulk polar semiconductors .
our calculations show strong oscillations of both the linewidth and the effective mass in a 3d system of gaas which indicate that the odmpr should also be observed experimentally in bulk polar semiconductors .
the present paper is organized as follows . in sec .
ii , we present our theoretical formulation of the problem .
the numerical results and discussions are given in sec .
iii , and we summerize our results in sec .
magnetophonon resonance is essentially a single - particle effect and , consequently , can be treated as a one - polaron problem .
we consider a polar semiconductor in a uniform magnetic field * b * directed along the @xmath2-axis .
the system under consideration can be described by the following hamiltonian , @xmath3 with @xmath4 and @xmath5 where @xmath6 is the bare electron effective mass , the vector potential @xmath7 is choosen in the symmetrical coulomb gauge , @xmath8 ( @xmath9 ) the momentum ( position ) operator of the electron , @xmath10 @xmath11 the creation ( annihilation ) operator of an optical - phonon with wave vector @xmath8 and energy @xmath12 .
the electron - phonon interaction hamiltonian @xmath13 is given by the frhlich interaction hamiltonian @xmath14 where @xmath15 and @xmath16 is the electron - lo - phonon coupling constant .
first , we calculate the optical aborption spectrum of the polaron in the presence of a magnetic field from which we are able to investigate the polaron cr spectrum and the corresponding mpr effects . for convenience
we use units such that @xmath17 within linear response theory the frequency - dependent magneto - optical absorption spectrum for cyclotron resonance [ 12 - 14 ] is given by @xmath18^{2}+[{\rm im}\sigma ( z)]^{2 } } } , \label{absorption}\ ] ] where @xmath19 is the unperturbed electron cyclotron frequency , @xmath20 is the so called memory function , and @xmath21 where @xmath22 is a broadening parameter .
notice that @xmath22 is introduced semi - empirically to remove the divergence of the landau level density of states .
we take @xmath23 as a constant . for the magneto - optical absorption spectrum in the faraday ( active - mode ) configuration , which corresponds to the cyclotron resonance experiments
, the memory function is given by [ 13 ] @xmath24 with @xmath25\rangle , \ ] ] where @xmath26 , and the correlation function is given by @xmath27\rangle = [ 1+n(\omega _ { lo})]e^{-i\omega _ { lo}t}s^{\ast } ( -\vec{q},t)-n(\omega _ { lo})e^{-i\omega _ { lo}t}s(\vec{q},t),\ ] ] where @xmath28 is the number of lo - phonons and @xmath29 is the space fourier transform of the electron density - density correlation function . in eq .
( 10 ) @xmath30 , where @xmath31 is the boltzmann constant . for a weak electron - lo - phonon coupling system , i.e. @xmath32
, the density - density correlation function is calculated for a free electron in a magnetic field which is given by @xmath33 where @xmath34 and @xmath35 , \ ] ] from the above equations , we obtain the memory function for @xmath36 .
the calculation proceeds along the lines of a similar calculation which was presented in ref . [ 13 ] .
the results for the memory function are @xmath37 ^{-(n+n^{\prime } ) } } { n!n^{\prime } ! } } \int_{0}^{\infty } { \frac{dx}{x}}% e_{n+n^{\prime } + 1}\left ( { \frac{x^{2}}{\omega _ { c}\tanh ( \beta \omega _ { c}/2)}}\right ) \nonumber
\\ & & \times \{\exp \left ( { \frac{\beta \omega _ { nn^{\prime } } } { 2}}\right ) \{2{% d\left ( { \frac{\sqrt{\beta } x}{2}}+{\frac{\sqrt{\beta } } { 2x}}\omega _ { nn^{\prime } } \right ) } \nonumber \\ { } & { } & -{d\left ( { \frac{\sqrt{\beta } x}{2}}+{\frac{\sqrt{\beta } } { 2x}}% ( \omega _ { nn^{\prime } } + \omega ) \right ) } -{d\left ( { \frac{\sqrt{\beta } x}{2}}% + { \frac{\sqrt{\beta } } { 2x}}(\omega _ { n^{\prime } n}-\omega ) \right ) } \ } \nonumber \\ & & + \exp \left ( -{\frac{\beta \omega _ { n^{\prime } n}}{2}}\right ) \{2{d\left ( { % \frac{\sqrt{\beta } x}{2}-\frac{\sqrt{\beta } } { 2x}\omega _ { n^{\prime } n}}% \right ) } \nonumber \\ & & { -d\left ( { \frac{\sqrt{\beta } x}{2}-\frac{\sqrt{\beta } } { 2x}}(\omega _ { n^{\prime } n}-\omega ) \right ) } -{d\left ( { \frac{\sqrt{\beta } x}{2}-\frac{% \sqrt{\beta } } { 2x}}(\omega _ { n^{\prime } n}+\omega ) \right ) } \}\},\end{aligned}\ ] ] and @xmath38 ^{-(n+n^{\prime } ) } } { n!n^{\prime } ! } } \nonumber \\ & & \times \int_{0}^{\infty } { \frac{dx}{x}}e_{n+n^{\prime } + 1}\left ( { \frac{% x^{2}}{\omega _ { c}\tanh ( \beta \omega _ { c}/2)}}\right ) \nonumber \\ & & \times \left [ \exp \left ( -{\frac{\beta x}{4}}-{\frac{\beta } { 4x}}(\omega _ { nn^{\prime } } -\omega ) \right ) + \exp \left ( -{\frac{\beta x}{4}}+{\frac{% \beta } { 4x}}(\omega _ { n^{\prime } n}-\omega ) \right ) \right ] , \end{aligned}\ ] ] where , @xmath39 , @xmath40 is the dawson integral function , and @xmath41 in the case of non zero broadening , i.e. @xmath42 the calculation is more tedious .
we obtain the following results for the memory function , @xmath43 , \ ] ] and @xmath44 , \ ] ] with @xmath45 ^{-(n+n^{\prime } ) } } { n!n^{\prime } ! } } % \int_{0}^{\infty } { \frac{dx}{x}}e_{n+n^{\prime } + 1}\left ( { \frac{x^{2}}{% \omega _ { c}\tanh ( \beta \omega _ { c}/2)}}\right ) \nonumber \\ & & \times \{\exp \left ( { \frac{\beta \omega _ { nn^{\prime } } } { 2}}\right ) \{{%
d\left ( { \frac{\sqrt{\beta } x}{2}}+{\frac{\sqrt{\beta } } { 2x}}\omega _ { nn^{\prime } } \right ) } \nonumber \\ { } & { } & { -{\frac{\sqrt{\pi } } { 4}}}\left [ { { \rm im}w\left ( { \frac{\sqrt{\beta } % x}{2}}+{\frac{\sqrt{\beta } } { 2x}}(\omega _ { nn^{\prime } } + \omega + i\gamma ) \right ) } + { \rm im}w{\left ( { \frac{\sqrt{\beta } x}{2}}+{\frac{\sqrt{\beta } } { % 2x}}(\omega _ { n^{\prime } n}-\omega
+ i\gamma ) \right ) } \right ] \ } \nonumber \\ & & + \exp \left ( -{\frac{\beta \omega _ { n^{\prime } n}}{2}}\right ) \{{d\left ( { % \frac{\sqrt{\beta } x}{2}-\frac{\sqrt{\beta } } { 2x}\omega _ { n^{\prime } n}}% \right ) } \nonumber \\ & & { -{\frac{\sqrt{\pi } } { 4}}}\left [ { { \rm im}w\left ( { \frac{\sqrt{\beta } x}{2}% -\frac{\sqrt{\beta } } { 2x}}(\omega _ { n^{\prime } n}-\omega -i\gamma ) \right ) } + % { \rm im}w{\left ( { \frac{\sqrt{\beta } x}{2}-\frac{\sqrt{\beta } } { 2x}}(\omega _ { n^{\prime } n}+\omega -i\gamma ) \right ) } \right ] \}\},\end{aligned}\ ] ] and @xmath46 ^{-(n+n^{\prime } ) } } { % n!n^{\prime } ! } } \int_{0}^{\infty } { \frac{dx}{x}}e_{n+n^{\prime } + 1}\left ( { % \frac{x^{2}}{\omega _ { c}\tanh ( \beta \omega _ { c}/2)}}\right ) \nonumber \\
\times \{\exp \left ( { \frac{\beta \omega _ { nn^{\prime } } } { 2}}\right ) & & [ { \rm % re}w\left ( { \frac{\sqrt{\beta } x}{2}}+{\frac{\sqrt{\beta } } { 2x}}(\omega _ { nn^{\prime } } + \omega + i\gamma ) \right ) \nonumber \\ & & -{\rm re}w\left ( { \frac{\sqrt{\beta } x}{2}}+{\frac{\sqrt{\beta } } { 2x}}% ( \omega _ { nn^{\prime } } -\omega + i\gamma ) \right ) ] \nonumber \\ + \exp \left ( -{\frac{\beta \omega _ { n^{\prime } n}}{2}}\right ) & & [ { \rm re}% w\left ( { \frac{\sqrt{\beta } x}{2}}-{\frac{\sqrt{\beta } } { 2x}}(\omega _ { n^{\prime } n}-\omega -i\gamma ) \right ) \nonumber \\ & & -{\rm re}w\left ( { \frac{\sqrt{\beta } x}{2}}-{\frac{\sqrt{\beta } } { 2x}}% ( \omega _ { n^{\prime } n}+\omega -i\gamma ) \right ) ] \},\end{aligned}\ ] ] where @xmath47 is the complex error function .
in this section , we present our numerical results on the magneto - optical aborption spectra and study the magnetophonon resonant effects . as an example , we apply our theory to the semiconductor gaas where @xmath48 .
first , we show numerical results for @xmath49 k and level broadening parameter @xmath36 . due to the importance of the memory function in the absorption spectrum , we plot the real and imaginary parts of the memory function in fig .
1(a ) and 1(b ) , respectively , as a function of frequency at three different magnetic fields .
we see that , at @xmath50 ( @xmath51 ... ) , re@xmath52 exhibits a jump while im@xmath53 diverges logarithmically .
the discontinuity of re@xmath52 and the divergency in im@xmath54 reflects the resonant coupling between the state @xmath55 and landau level @xmath56 the stronger this coupling , the larger the discontinuity in re@xmath52 .
actually , the real part of the memory function re@xmath52 is responsible for the shift in the observed cr energy which is due to the electron - phonon interaction . while , the imaginary part leads to a broadening of the spectrum which is a result of scattering .
when im@xmath57 like in a 2d system , the absorption is a @xmath58 function and its position is determined by the equation @xmath59 .
1(b ) shows that for a bulk system im@xmath52 is non zero which reflects the 3d character of the electron states . the scattering which is mainly in the direction parallel to the magnetic field results in a finite im@xmath52 and consequently a finite linewidth even for @xmath60 . in fig .
1(c ) , we show the corresponding magneto - optical absorption spectra .
the position of the aborption peaks corresponds to the cyclotron resonance frequency @xmath61 at which the cyclotron resonance occurs .
we notice an asymmetric double peak structure around @xmath62 for @xmath63 ( the solid curve ) and the absorption becomes zero at @xmath64 the zeros in the absorption spectrum are a consequence of the divergences in im@xmath54 and which can be traced back to the divergent nature of the density of states .
the double peak structure is a consequence of the magnetophonon resonance which leads to an anticrossing behavior in the cr spectrum .
when the cr frequency @xmath65 deviates from @xmath66 this splitting becomes very weak and difficult to observe in the aborption spectrum . as we will see below , however
, the magnetophonon resonance will strongly affect the linewidth of the magneto - optical aborption and the cr mass . from the dash and the dotted curves
, we notice that the absorption peak appears at a frequency @xmath67 due to the polaron effect which shifts the cyclotron frequency to lower frequencies . the latter is often interpreted as an increase of the cyclotron mass , i.e. @xmath68 . in fig .
2 , we show the absorption spectrum around ( a ) @xmath69 and ( b ) @xmath70 .
the double peak structure disappears when @xmath71 deviates from @xmath72 .
the aborption spectra also demonstrate clearly a nonlinear magnetic field dependence of the peak position and linewidth around @xmath73 fig .
3 demonstrates the effect of the broadening parameter @xmath74 on the absorption spectrum .
notice that , with increasing @xmath22 : i ) the double peak structure disappears for @xmath75 , ii ) the zero in the absorption spectrum disappears when @xmath76 , and iii ) the position of the absorption peak shifts to higher frequency .
this indicates that the anticrossing behavior in the cr spectrum will be difficult to be observed experimently at @xmath63 due to broadening effects which are a consequence of scattering on e.g. impurities and acoustical phonons .
as soon as the polaron cr frequency @xmath61 is determined from the postion of the magneto - optical absorption peak , the cr mass of the polaron is obtained by @xmath77 the numerical results of the polaron cr mass and the fwhm ( full width at half maximum ) for @xmath78 are plotted as a function of the unperturbed cr frequency at different temperatures in figs .
4(a ) and 4(b ) , respectively .
one observes that the polaron cr mass is an oscillating function of the magnetic field .
4(b ) shows that the fwhm of the polaron magneto - optical absorpion spectrum reaches a local maximum at @xmath79 where the polaron mass has an inflection point .
this result demonstrates the derivative - like relation between the polaron cr mass and the linewidth which are due to the fact that the real and imaginary part of the memory function are related to each other through a kramers - kronig relation .
one finds that , for temperature @xmath80 k , the resonance grows rapidly with increasing @xmath81 .
this effect can lead to a direct measurement of the optical - phonon scattering rate .
we notice also an overall increase of the linewidth with temperature , but an overall decrease of the effective mass when @xmath82 k. the resonant position is slightly larger than the non - interacting resonant condition @xmath83 and is almost independent of temperature .
a detailed analysis indicates that , at @xmath84 and @xmath85 the peak position both in the fwhm and in the derivative of the cr mass is at @xmath86 and @xmath87 , respectively .
experimentally this position determines the so called fundamental field @xmath88 which is an important quantity to study the effective mass , nonparabolicity of the energy band , as well as the lo - phonon frequency .
the linewidth is a direct measure of the lifetime of the state .
notice that the conventional mpr occurs in the resistivity , which is given by @xmath89 im@xmath90 .
but odmpr is related to both the real and imaginary part of the memory function which occurs for @xmath91 and is a dynamical mpr .
5 shows the cr mass oscillation amplitude at @xmath92 and @xmath93 as a function of temperature . with increasing temperature , the number of phonons increases and consequently , the oscillation amplitude increases . on
the other hand , the background electron - phonon scattering ( coupling ) increases which results in a suppression of the oscillation amplitude .
6 shows an activation plot of the amplitude of the resonant peak in the fwhm at @xmath92 and @xmath93 as a function of @xmath94 .
we find that , for the resonance around @xmath84 , the linewidth can be described rather well by the exponential law @xmath95 for @xmath96 k , while that around @xmath97 can be described by @xmath98 for @xmath99 k. this exponential behavior can be understood as follows .
mpr is proportional to the number of lo - phonons which are present and therefore should increase as @xmath100 . on the other hand , thermal broadening of the landau levels , which is proportional to @xmath101 , will diminish the resonant structure in @xmath102 .
thus this contribution decreases the resonant character and consequently we expect that @xmath103 which agrees with the exponential laws found for @xmath84 and @xmath97 .
we have extended the theory for odmpr to three - dimensional ( 3d ) systems and present the first detailed theoretical study of the magnetophonon resonance in the frequency - dependent conductivity of electrons in bulk gaas . in comparison to the corresponding 2d systems ,
the theoretical obtained amplitudes for the oscillations of both the linewidth and the effective mass in a 3d system are for gaas predicted to be about half of those in 2d .
therefore , we believe that odmpr can also be observed experimentally in bulk polar semiconductors . out
numerical results indicate that the amplitude of the resonant peak in the fwhm can be described by an exponential law for not too large temperatures .
j. van royen , j. de sitter , l.f .
lemmens , and j.t .
devreese , physica b * 89 * , 101 ( 1977 ) ; j. van royen , j. de sitter , and j.t .
devreese , phys .
b * 30 * , 7154 ( 1984 ) ; j.p .
vigneron , r. evrard , and e. kartheuser , phys .
b * 18 * , 6930 ( 1978 ) .
d. schneider , c. brink , g. irmer , and p. verma , physica b * 256 - 258 * , 625 ( 1998 ) ; d. schneider , k. pricke , j. schulz , g. irmer , and m. wenzel , in _ proceedings of the 23rd international conference on the physics of semiconductors _ , eds .
m. scheffler and r. zimmermann ( world scientific , singapore , 1996 ) , p. 221 .
w. xu , f. m. peeters , j. t. devreese , d. r. leadley , and r. j. nicholas , int . j. mod .
b * 10 * , 169 ( 1996 ) ; d. r. leadley , r.j .
nicholas , j. singleton , w. xu , f.m .
peeters , j.t .
devreese , j.a.a.j .
perenboom , l. van bockstal , f. herlach , j.j .
harris , and c.t .
. lett . * 73 * , 589 ( 1994 ) . | magnetophonon resonances are found for @xmath0 with @xmath1 .... in the polaron cyclotron resonance ( cr ) linewidth and effective mass of bulk polar semiconductors .
the cr mass and the linewidth are obtained from the full polaron magneto - optical absorption spectrum which are calculated using the memory function technique .
the amplitude of the resonant peak in the linewidth can be described by an exponential law at low temperature . |
star - forming rings or `` pseudorings '' are common to early and intermediate - type spiral galaxies ( cf . athanassoula & bosma 1985 ; buta , purcell , & crocker 1995 ; buta & combes 1996 ) , including our own milky way galaxy ( cf .
gusten & mezger 1982 ; clemens , sanders , & scoville 1988 ; waller 1990a ) .
such ring - like accumulations of gas and associated starbirth activity may have helped to build the inner parts of many primeval disk galaxies ( friedli & benz 1995 ) , as exemplified by the recent discovery of a starburst - ringed galaxy at @xmath4 ( colley , tyson , & turner 1996 ; tyson et al .
the formation and maintenance of these starburst rings are often attributed to orbital resonances with rotating bar or `` oval '' asymmetries in the stellar disks ( cf .
combes 1994 , byrd et al .
1994 ; combes et al . 1995 ; buta & combes 1996 and references therein ) . however , other dynamical mechanisms including gravitational instabilities ( elmegreen 1992 , 1994 ; kenney & jogee 1997 ) , outward propagating star formation ( walker , lebofsky , & rieke 1988 ; waller , gurwell , & tamura 1992 ) , and even radially - driven pileups from nuclear outbursts ( waller et al .
1992 ; tenorio - tagle et al . 1997 ) may play significant roles in orchestrating some of the starburst rings that are observed . as the closest early - type spiral galaxy of low inclination ,
m94 ( ngc 4736 ) has received concentrated attention from both observers and theorists . this ( r)sa(r)ab - type galaxy ( de vaucouleurs et al .
1991 ) is noted for its inner ring of ongoing starburst activity ( r @xmath5 45@xmath1 ) , oval stellar distribution at intermediate radius ( r @xmath5 220@xmath1 ) ( cf .
mulder & van driel 1993 ; mulder 1995 ; mollenhoff , matthias , & gerhard 1995 ) , and outer stellar ring near its de vaucouleurs radius ( r@xmath6 @xmath5 330@xmath1 ) . * figure 1 * ( extracted from the digital sky survey using the skyview utility [ mcglynn , scollick , & white 1996 ] ) shows the outermost portions of m94 , highlighting the oval disk and outer pseudoring .
the inner starburst ring is a prominent source of h@xmath0 , h i , and co emission ( smith et al . 1991 ; mulder & van driel 1993 ; gerin , casoli , & combes 1991 ) .
the discovery of compact thermal & nonthermal radio sources in the ring ( duric & dittmar 1988 ) indicates the presence of dense h ii regions and young snrs .
the ring s velocity field can be described by a combination of circular rotation with velocities of order 200 km / s and residual non - circular motions of order 15 km / s ( mulder 1995 ) to 25 km / s ( buta 1988 ) , depending on the adopted inclination and major axis position angle .
interior to the ring , the bright bulge and inner disk show twisted isophotes at red and near - ir wavelengths , indicative of a weak bar - like distortion ( beckman et al .
1991 ; shaw et al . 1993
; mollenhoff et al . 1995 ) .
fir and co observations interior to the ring reveal a rich ism with gas surface densities exceeding that of the ring ( smith & harvey 1994 ; garman & young 1986 ; gerin et al .
1991 ; wong & blitz 2000 ) .
optical spectroscopy of the nuclear region yields liner - type emission lines along with absorption lines from the circumnuclear stellar population , consistent with an early main - sequence stellar turnoff ( a4a7 ) and corresponding age of @xmath2500 myr ( pritchett 1977 ; keel 1983 ; taniguchi et al .
. further support for a young central population comes from nir spectroscopy which shows deep co absorption bands from red giant and asymptotic giant branch stars of similar age ( walker et al .
these authors have proposed an outward propagating mode of star formation , whereby ngc 253 , m82 , m94 , and m31 represent increasingly evolved versions of the same starbursting sequence .
although the kinematics of the ring show very little evidence for outward _ expanding motions _
( contrary to prior claims of bulk expansion [ van der kruit 1974 ; 1976 ] ) , they also do not preclude a scenario for _ radially propagating _ star formation .
other investigators have modeled the inner and outer rings in terms of _ resonant dynamics _ mediated by bar or `` oval '' potentials interior to the rings ( gerin et al .
1991 ; shaw et al . 1993 ; mollenhoff et al . 1995 ; mulder & combes 1996 ) , with the observed non - circular motions resulting from dispersion orbits near the lindblad resonances ( buta 1988 ) . in this paper
, we present and discuss new observational clues to the dynamical mechanisms governing the star formation in m94 .
the ultraviolet images obtained by the ultraviolet imaging telescope , in particular , reveal hitherto unrecognized patterns of recent star formation whose presence lends further support to the hypothesis of galaxy evolution via bar - mediated resonances .
the various imaging and spectroscopic observations and reductions are described in section 2 .
the resulting fuv , nuv , h@xmath0 , r , and i - band emission morphologies are presented and compared in section 3 . radial intensity profiles and other photometric comparisons
are discussed in section 4 .
uv and optical spectroscopy of the inner disk and nucleus is presented in section 5 .
kinematic properties and inferred dynamical scenarios are considered in section 6 .
our summary of the wavelength - dependent morphological , spectro - photometric , and dynamical properties of m94 appears in section 7 , wherein evolutionary implications are discussed . in the following sections , we assume a distance to m94 of 4.6 ( 75/h@xmath7 ) mpc , based on the galaxy s recession velocity of 345 km / s with respect to the local group ( sandage & tammann 1981 ) .
the corresponding spatial scale is 22.3 pc / arcsec .
unresolved sources imaged by the hst / foc provide additional constraints on the distance , as discussed in section 4 .
we adopt a nominal inclination of 40@xmath8 and major - axis position angle of 120@xmath8 , while recognizing that both of these quantities may vary significantly with radius and with measuring technique ( e.g. morphological vs. kinematic determinations ) ( bosma , van der hulst , & sullivan 1977 ; buta 1988 ; mulder & van driel 1993 ; mulder 1995 ; mollenhoff et al .
1995 ; wong & blitz 2000 ) .
a log of the ultraviolet and visible imaging is presented in * table 1*. a listing of complementary uv and visible spectra is shown in * table 2*. the ultraviolet imaging telescope ( uit ) imaged m94 in the far - ultraviolet ( @xmath9 = 1521 , @xmath10 = 354 ) on 1995 march 12 as part of the 16-day spacelab/_astro-2 _ mission aboard the space shuttle endeavour .
this wide - field telescope images 40-arcmin fields of view at @xmath23 arcsec resolution . in the case of m94 , the uit image represents the only extant uv image of the entire galaxy ( see * figure 2a * ( plate xxx ) ) .
the 1040-sec exposure was obtained with a dual - stage image intensifier with csi photocathodes and was recorded on carbon - backed iiao kodak film .
after processing of the film and scanning of the emulsion , the resulting digitized `` density image '' was fog - subtracted , flat - fielded , linearized to `` exposure units , '' and calibrated to flux units using iue observations of standard stars ( cf .
stecher et al .
1992 , 1997 ; waller et al .
1995 and references therein ) .
correction for image distortion produced by the magnetically focused image intensifiers was carried out according to the procedures described by greason et al .
the resulting corrections amounted to a few arcsec in the field center ( which includes all of m94 s fuv emission ) increasing to 1020 arcsec near the edge of the 40 arcmin field of view .
astrometry was tied to 10 compact knots evident in both the fuv and b - band images ( see next subsection ) .
positions in the resulting distortion - corrected image are good to @xmath23 arcsec , and the spatial resolution is of similar magnitude . the hubble space telescope s faint object camera ( foc ) imaged the center of m94 in the near - ultraviolet ( @xmath9 = 2300 , @xmath10 = 500 ) on 18 , july 1993 before the optical repair mission as part of a uv imaging survey of 110 large nearby galaxies ( maoz et al .
1995 , 1996 ) . after standard stsci pipeline processing
, the 596 sec nuv exposure has a @xmath11 field of view at 0.022@xmath1/pixel ( see * figure 2a * ( plate xxx ) ) .
the spherical aberration of hst s primary mirror resulted in a point - spread function ( psf ) featuring a sharp core with fwhm @xmath5 0.05@xmath1 and about 15% of the total light surrounded by an extensive halo of several arcsec radius containing most of the energy .
following maoz et al .
( 1996 ) , our calibration of the detected foc counts into flux densities assume a conversion of @xmath12 erg s@xmath13 @xmath14 @xmath13/count s@xmath13 , while noting that the photflam conversion in the image header is @xmath15 erg s@xmath13 @xmath14 @xmath13/count s@xmath13 .
flux uncertainties are estimated at @xmath25% over large areas increasing to @xmath220% for compact sources ( maoz et al .
1995 , 1996 , and references therein ) . a wide - field ( @xmath16 ) b - band image was obtained with the palomar 5-m telescope and tek3 ccd camera ( 1024 @xmath17 1024 pix ) on 1994 february 11 under hazy skies .
this 600-sec exposure is saturated in the central 2 arcmin , but contains high s / n detections of the oval disk and parts of the outer ring .
astrometry of this image is tied to the positions of several foreground stars , as measured on a corresponding image in the digitized sky survey ( stsci 1994 ) .
groundbased h@xmath0 , r , and i - band images of m94 were obtained with the now - retired kpno 0.9-m telescope and rca-1 ccd camera ( 508 @xmath17 316 pix ) on 17 , february 1986 .
these images have @xmath18 fields of view at 0.86@xmath1/pix .
sky conditions varied from photometric to hazy , yielding psfs of about 2@xmath1 ( fwhm ) .
astrometry is tied to 5 foreground stars that are common to the kpno ( h@xmath0 , r , i - band ) and palomar ( b - band ) images . for calibration purposes ,
spectrophotometric standard stars ( bd26@xmath82606 & hd84937 ) were imaged before and after the target imaging .
subtraction of the red continuum from the h@xmath0-band image was carried out with the r - band image according to the formulations in waller ( 1990b ) , whereby corrections were made for the @xmath238% [ n ii]@xmath196548 , 6584 contribution to the total h@xmath0 @xmath20 [ n ii ] line emission and resulting @xmath215% contamination of the h@xmath0 image after transmission by the 36 bandwidth h@xmath0 filter .
a pure red - continuum image was also produced , by removing the contaminating h@xmath0 + [ n ii ] line emission from the r - band image ( waller 1990b ) ( see * figure 2a * ( plate xxx ) ) .
photometry of the h@xmath0 emission from the @xmath21 field yields a total flux of 9.9 @xmath17 10@xmath22 erg @xmath14 s@xmath13 , 14% higher than that determined by kennicutt & kent ( 1983 ) within a 7@xmath23-diameter aperture ( after correcting for a 38% [ n ii ] contamination within their 100 bandwidth ) .
the starburst ring ( r = 30@xmath1 @xmath24 60@xmath1 ) is measured to have f(h@xmath0 ) = 9.6 @xmath17 10@xmath22 erg @xmath14 s@xmath13 , which is nearly 1.8 times higher than that obtained by smith et al .
( 1991 ) from the 75 bandwidth image of pogge ( 1989 ) . some of these discrepancies can be attributed to the varying bandpasses and corresponding uncertainties in the [ n ii ] emission being transmitted , vagaries in the continuum subtraction , problematic h@xmath0 _ absorption _ produced by a - type stars in the inner disk ( see section 4.2 ) , and final calibration ( smith et al .
because of hazy conditions during the i - band exposure , calibration of the i - band image was done by bootstrapping to the i - band radial intensity profiles resulting from the photoelectric photometry of munoz - tunon et al .
( 1989 ) and the @xmath25 color profiles of beckman et al .
( 1991 ) .
fuv ( 1200 to 2000 ) and nuv ( 2000 to 3200 ) spectra of m94 s inner disk were obtained from the international ultraviolet explorer ( iue ) archive .
the iue data were accessed via the iue data analysis center at nasa / gsfc and are representative of the new spectral image processing system ( newsips ) .
* table 2 * lists the image numbers , dates , exposure times , nominal positions and roll angles of the low - resolution fuv ( swp ) and nuv ( lwp & lwr ) iue spectra . here ,
swp , lwp , and lwr respectively refer to iue s short - wavelength prime , long - wavelength prime , and long - wavelength redundant cameras .
* figure 3 * shows the iue apertures ( @xmath26 ) on grey - scale images of the fuv emission , where the mapping is based on the nominal positions and roll angles of the fuv ( swp ) and nuv ( lwp & lwr ) observations .
although the accuracy of the nominal positions is of order @xmath2710@xmath1 , * figure 3 * indicates that the apertures were most likely sampling the inner disk rather than the starburst ring . *
table 2 * notes which apertures are filled with disk ( d ) and/or nuclear ( n ) emission , based on visual inspection of the overlays .
the limiting spectral resolution of these data is about 6 .
groundbased visible spectra ( 3800 to 7500 ) were obtained from the lick 1-m nickel telescope and image dissecting spectrograph ( ids ) .
the summed spectrum of the central region ( 8.1 arcsec circular aperture ) is a mean - flux average of 64 min in blue and red grating settings , with 16 min in an intermediate setting to ensure the overlap area was well calibrated .
the spectrum has been rebinned to 2.5- pixels from the original , which still oversamples the resolution of @xmath2810 fwhm .
we note that adjacent pixels in the ids spectrum are not statistically independent , yielding detections similar to those of a non - centroiding photon counter which spreads single photons across several output pixels .
we complement this composite spectrum with analysis of the image - dissector scanner data presented by keel ( 1983 ) and discuss a new high - resolution spectrum , obtained with the 2.5-m isaac newton telescope ( int ) on la palma , using the intermediate - dispersion spectrograph with image photon - counting system ( ipcs ) detector .
the 1 " slit was oriented approximately along the major axis ( pa 135@xmath29 ) for this 1000-second exposure .
there is useful signal over 45 spatial increments of 0.6@xmath1 each .
we consider equivalent widths of the lines along with spectral slopes and discontinuities of the continuum emission .
* figure 2a * ( plate xxxx ) shows the dramatically different morphologies that are detected at ultraviolet and visible wavelengths .
the inner disk and bulge component that is so prominent at r - band completely disappears in the fuv image .
the oval disk at intermediate radius also has no fuv counterpart . instead
, the fuv image is characterized by ( a ) an extended and elongated nucleus , ( b ) a diffuse inner disk , ( c ) a bright inner ring at the perimeter of the inner disk ( r = 48@xmath1 = 1.1 kpc ) , and ( d ) two 500-pc size knots exterior to the ring ( r = 130@xmath1 = 2.9 kpc ) .
the insert in * figure 2a * ( plate xxxx ) contains the hst / foc pre - costar image of the nuclear region ( maoz et al .
1995 , 1996 ) .
this near - uv image shows a marginally - resolved nucleus ( fwhm @xmath5 0.1@xmath1 ) embedded in a bright core of @xmath21@xmath1 ( 22 pc ) diameter along with a faint `` mini - bar '' that has a total length of 20@xmath1 ( 450 pc ) at a position angle of approximately 30@xmath8 .
low - level ripples in the emission ( which are included in the estimate of the bar s total length ) are evident 2@xmath1 , 5@xmath1 and 9@xmath1 to the sw of the nucleus .
maoz et al .
( 1995 ) attribute these features to bow shocks or tidal arms resulting from a recent merger event . however , we see that the ripples overlap with the larger concentric arcs that are evident in the uit / fuv image at projected radii of @xmath29@xmath1 and 15@xmath1 ( see next subsection ) .
similar ripples are also present in an archival hst / wfpc2 v - band image of this region with connections to larger spiral arcs . to the north of the nucleus
are two point sources whose flux densities are consistent with those of single b - type supergiant stars ( see section 4 ) .
the uit / fuv emission from the inner disk includes concentric fuv arcs to the sw of the nucleus which can be traced for @xmath2@xmath30 radians ( @xmath2700 pc ) .
these arcs are of low contrast and , because of the noise characteristics of the uit imaging , are not amenable to typical contrast enhancement techniques ( e.g. median filtering ) .
however , they do show up as enhancements in the radial distribution of intensities ( see * figure 2b * ) .
typical spacing between the arcs is 9@xmath1 ( 200 pc ) , with additional features at 15@xmath1 and 34@xmath1 .
the arcs themselves appear to show some substructure at the limits of resolution .
contrary to a merging scenario ( maoz et al .
1995 ) , which would seem to be precluded by the lack of significant tidal effects beyond the inner few hundred parsecs , these features more likely indicate an orbital dynamic at work in the inner disk . at an interpolated orbital velocity of 140 km / s and corresponding shear rate of -1360 km@xmath13 kpc@xmath31 ( see section 6 ) ,
such features could have been differentially swept out over a timescale of @xmath220100 myr .
this estimated timespan is consistent with a population of late b - type stars whose main - sequence lifetimes are of similar duration .
the uit / fuv image of the inner disk also shows a brighter arc to the north , just inside the starburst ring at a radius of 40@xmath1 ( see * figure 2b * ) , along with widespread diffuse emission at a level amounting to @xmath215% the mean surface brightness of the ring .
the nature of these resolved and unresolved fuv components is uncertain , although some clues can be obtained from the longer wavelength imagery . at h@xmath0
the inner disk appears _ in absorption _ with respect to the continuum emission from the underlying population of stars .
the strongest absorbers at h@xmath0 are the photospheric atmospheres of b and a - type stars , whose temperatures are sufficiently cool for hydrogen to remain neutral and sufficiently warm for significant population of the h balmer ( n=2 ) electronic energy level ( cf .
mihalas 1978 ) . in the uit / fuv image ,
the diffuse light from the inner disk probably arises from these same stars . indeed , _
the uit / fuv image represents the first view of this young stellar population , unconfused by the longer - wavelength emission from the older inner disk and bulge components_. * figure 4 * compares a spatially - filtered r - band image with an @xmath25 color image of the inner disk . the spatially - filtered r - band image ( * fig .
4a * ) was created by median smoothing the r - band image over a @xmath32 window and then subtracting the smoothed image from the original .
the resulting fine - scale structure includes the nuclear `` mini - bar '' previously noted by mollenhoff et al .
( 1995 ) and mulder ( 1995 ) , along with a `` dark '' spiral arc of diminished emissivity to the west and other flocculent spiral structures associated with the starburst ring .
the archival hst / wfpc2 v - band image of this region resolves the flocculent structure into spiral dust lanes of high pitch angle which cross through the nearly circular ring .
the @xmath25 color image ( * fig . 4b * ) shows reddening along the western spiral arc seen in ( * fig . 4a * ) and in another arc to the ne , along with blue knots in the starburst ring .
_ none of the fine - scale structures and reddened features in the optical inner disk have counterparts in the uit / fuv image . _ therefore , neither emitting nor absorbing structures , nor reddening dust lanes in the _ old disk _ can be defining the observed concentric fuv arcs to the sw and ne .
we infer that the fuv arcs are recently - generated structures of emitting b and a - type stars and/or scattering clouds of low dust optical depth .
we further speculate that the emitting and/or scattering fuv sources , being relatively young , are distributed in a much thinner disk than that associated with the optical fine - scale structures and reddened features , thus explaining the morphological differences between these two wavelength regimes .
further insights on the stellar content of the inner disk can be gained from the iue spectroscopy presented in section 6 .
the inner starburst ring in m94 is the single dominant emission feature in the fuv , h@xmath0 , and radio continuum bands .
rectification of the galaxy to its nominal face - on orientation ( pa = 120@xmath8 @xmath33 [ mulder & van driel 1993 ] ) shows that the ring is almost perfectly circular , with a mean radius of 47@xmath1 ( 1.1 kpc ) and fwhm of 21@xmath1 ( 0.49 kpc ) .
the latter measurement is based on doubling the measured hwhm of the emission _ beyond _ the mean radius , thereby excluding the diffuse contribution interior to the ring .
adopting an inclination of 35@xmath8 , as in the dispersion orbit model of buta ( 1988 ) , would yield a finite but small ellipticity of 0.063 , with a 16@xmath8 offset between the kinematic line of nodes and projected major axis .
we are not able to further constrain the ring s orientation , and so shall continue to assume an inclination of 40@xmath8 , based on the h i study of mulder & van driel ( 1993 ) .
a comparison of the fuv and h@xmath0 morphologies reveals strong similarities . *
figure 5a * shows the h@xmath0 emission , and * figure 5b * shows the ratio of h@xmath0 and fuv intensities in the inner disk and ring .
except for the spiral - like ridges of enhanced h@xmath0 emission to the wnw and ese , very little coherent variation in this ratio is evident in the ring . in particular , the intensity ratios do not show any radial displacement in the fuv emission relative to that at h@xmath0 . from this lack of radial structuring in the h@xmath0/fuv ratio , we can conclude that _ little evidence is found for outward or inward propagating starburst activity _ , as explained below .
a morphological comparison of h@xmath0 and fuv emitting regions provides a useful means of tracking sequential patterns of star formation in disk galaxies .
the h@xmath0 emission is dominated by the most massive and short - lived stars ( m @xmath34 20 m@xmath35 ) , while the fuv emission arises mostly from less massive , longer- lived stars ( 20 @xmath28 m / m@xmath35 @xmath28 2.5 ) . for an evolving star cluster with an imf typical of local star - forming regions , the h@xmath0 emission typically reaches a maximum within a million years of the star - forming episode , and
quickly decays with an e - folding timescale of about 3 myr .
by contrast , the fuv emission reaches a maximum at about 3 myr ( due to the onset of b - type supergiants ) , decaying to 1/e in yet another 3 myr .
the h@xmath0/fuv ratio is seen to decrease by 2 orders of magnitude after 10 myr ( e.g. hill et al .
subsequently , the h@xmath0 emission will vanish , while the fuv declines slowly for the next few 100 myrs .
in the outward propagating starburst scenario , the h@xmath0 emission would concentrate where the propagating wavefront is located on the outer perimeter of the fuv - emitting ring .
inward propagating scenarios would have the h@xmath0 emission interior to the fuv ring .
spatial displacements of h@xmath0 and fuv emission have been observed across the spiral arms of m51 ( oconnell 1997 ; petit et al .
1996 ) , m74 ( marcum et al . 1997 ) , and ngc 4258 ( courtes et al . 1993 ) .
these displacements have been interpreted as the result of spiral density waves concentrating gas along the insides of the spiral arms and subsequent migration of the evolving clusters past the spiral wave fronts .
given the resolution of our images ( @xmath2100 pc ) and characteristic timescale between h@xmath0 and fuv maxima ( 3 myr ) , any residual propagation of starburst activity would have to proceed at a speed less than 35 km / s to avoid detection .
implications of this propagation speed limit are discussed in section 7 .
the arm - like enhancements in the h@xmath0/fuv ratio correspond to similar features in a recent co mapping with the bima interferometer ( wong et al .
these arm - like extensions away from the ring have further analogues at b - band , where a complex spiral pattern is evident ( see next subsection ) .
one of the most remarkable aspects of m94 s fuv emission morphology is the pair of knots on diametrically opposite sides of the nucleus .
both knots have projected radii of approximately 130@xmath1 ( 2.9 kpc ) .
a line drawn through the knots intersects the nucleus at a position angle of 105@xmath8 , intermediate between the pas of the inner starburst ring ( 127@xmath8 ) and the oval disk ( 95@xmath8 ) ( mollenhoff et al .
the two knots are roughly 20@xmath1 ( 500 pc ) in size , with the eastern knot showing the more complex structure .
only the eastern knot shows significantly in our continuum - subtracted h@xmath0 image ( * figure 5 * ) .
other much smaller knots of fuv emission appear at low s / n to the sw , ssw , ene , and nne .
the sw , ssw , and ene knots have faint counterparts in the light of h@xmath0 , while all of these features have counterparts at b - band ( see below ) . * figure 6a * and * figure 6b * respectively show the b - band image before and after spatial filtering .
the filtering in this case involves median smoothing over a @xmath36 window and then dividing the original image by this smoothed image .
median normalized spatial filtering
_ reveals fine - scale structure over a wide range of surface brightnesses ( e.g. waller et al .
1998 ) . here , it highlights the starburst ring and bisymmetric knots as significant enhancements above the disk at b - band .
the filtered image also reveals many flocculent spiral arms outside of the ring , whose relation with the knots is somewhat ambiguous .
the western knot appears part of a dominant arm with a `` shingled '' morphology connecting to the ring in the northeast .
the eastern knot appears to be associated with several arms , including one which would be the symmetric counterpart to the arm that links the western knot to the ring .
the marginally detected fuv knots to the sw , ssw , ene , and nne all show enhanced b - band emission from associations of massive stars among the myriad spiral arms . in summary , the bi - symmetric fuv knots have b - band counterparts which appear to be part of the complex spiral arm structure at these radii .
their symmetric prominence on opposite sides of the nucleus , however , requires a dynamical explanation that is spatially more specific than that of the arms .
similarly , the starburst ring can be regarded as an especially bright , tightly wound component of the overall spiral pattern , whose prominence also indicates special dynamical circumstances .
bi - symmetric knots or plumes have been noted in other ringed - barred galaxies , including ngc 1433 ( buta 1986 ) , ngc 7020 ( buta 1990 ) , ngc 7098 ( buta 1995 ) , and ic 4214 ( buta & combes 1996 ) . as discussed in these papers ,
the symmetric features are likely tracing dynamical resonances that are connected with the general ring - bar phenomenon .
from the uit and hst / foc images , circular aperture photometry of the nucleus out to a radius of 5@xmath1 yields @xmath37 = 14.7 mag and @xmath38 = 13.5 mag , respectively .
the resulting @xmath39 color of 1.2 mag is significantly redder than those derived from iue spectra of the inner disk ( see next section ) . between 5@xmath1
10@xmath1 radii , the non - nuclear emission is significantly bluer than those derived using iue , with @xmath40 .
the latter colors indicate the presence of late b and hotter stars , depending on the amount of reddening ( fanelli et al .
estimates of the reddening have been derived from the visible colors ( smith et al .
1991 ; see also next subsection ) , visible spectra and subsequent modelling of the stellar populations ( taniguchi et al .
1996 ; pritchett 1977 ) .
the resulting estimates of visual extinction are 0.51.0 mag in the nuclear region , corresponding to uv color excesses of @xmath41 = 0.10.2 mag .
therefore , the corrected @xmath42 color of the nucleus could be close to 1.0 mag , or the equivalent of a stellar population with a late a - type stellar cutoff ( fanelli et al .
1992 ) .
the non - nuclear point sources in the hst / foc near - uv image of the nuclear region provide helpful checks on the distance to m94 .
for the brightest point source , maoz et al .
( 1996 ) lists a monochromatic flux of @xmath43(2300 ) = 2.6 @xmath17 10@xmath44 erg s@xmath13 @xmath14 @xmath13 ( m@xmath45(2300 ) = 17.9 mag ) .
this source is unresolved at a resolution of 0.1@xmath1 and corresponding linear scale of @xmath462.2 pc at the assumed distance .
although an extremely compact star cluster can not be ruled out , this exceptional uv source is most likely dominated by a single hot supergiant star . in the absence of extinction ,
a b0 - 2 ia - o supergiant star has @xmath47
@xmath5 @xmath4810.7 mag ( fanelli et al .
1992 ) , which would imply a distance modulus of ( @xmath49 ) = 28.6 mag and corresponding distance of 5.2 mpc .
assuming nuclear color excesses of @xmath50 = 0.150.27 mag ( taniguchi et al . 1996 ) and galactic - type extinction law , the corresponding near - uv extinctions would be @xmath51 @xmath5 12 mag , and the revised distances would be 3.32.1 mpc , respectively .
the adopted distance of 4.6 mpc can thus be regarded as a reasonable estimate , given the uncertainties in stellar spectral type and extinction . the second source , 1-arcsec to the e , is 6 times ( 1.9 mags ) fainter consistent with it being an a0 - 2 ia - o supergiant at this distance , if subject to the same amount of extinction .
were the galaxy significantly closer , more abundant oba - type giant and main - sequence stars would be resolved which is not the case .
* figure 7 * shows radial profiles of the fuv surface brightness and cumulative flux .
these profiles were derived from annular - averaged photometry using elliptical annuli consistent with the adopted position angle ( 120@xmath8 ) and inclination ( 40@xmath8 ) . since galactic extinction is negligible towards m94 ( de vaucouleurs et al .
1991 ) , no correction was made . the resulting surface brightness profile ( * fig . 7a * ) is dominated by light from the nucleus , inner disk , starburst ring , and an exponentially - declining disk that is punctuated by an enhancement from the bi - symmetric knots at @xmath52 . the exponentially declining component between 50 and 80-arcsec radii
has an e - folding scalelength of only @xmath53 ( 223 pc ) , similar to that found at r and i - bands ( cf .
munoz - tunon et al .
the fuv emission beyond the bi - symmetric knots also shows a steep decline with an estimated scalelength of @xmath54 ( 446 pc ) , significantly shorter than the 70@xmath1 scalelength measured at r and i - bands in the outer disk ( munoz - tunon et al .
1989 ) . the cumulative flux profile ( * fig . 7b * ) indicates that the half - light radius matches that of the starburst ring , and that more than 80% of the total fuv emission is contained within a radius of @xmath55 ( 1.3 kpc ) .
the total fuv flux from m94 is f@xmath45(1520 ) = 6.02 @xmath17 10@xmath56 erg @xmath14 s@xmath13 @xmath13 , corresponding to m(fuv ) = 9.45 mag , or m(fuv ) = @xmath57 mag at the adopted distance of 4.6 mpc .
the corresponding fuv luminosity of 1.5 @xmath17 10@xmath58 erg s@xmath13 @xmath13 is the photometric equivalent of 2 @xmath17 10@xmath59 orion nebulae ( bohlin et al .
1982 ) or about 90 30-doradus regions ( as measured on an uit / fuv image of 30 dor out to a radius of 5 arcmin [ 67 pc ] ) , before correcting for the extinction in these sources . adopting a salpeter imf with
lower and upper mass limits of 0.1 and 100 m@xmath35 , respectively , yields an uncorrected star formation rate of 0.15 m@xmath35/yr .
this sfr estimate assumes continuous star formation and includes the strong contribution of b supergiant stars to the overall luminosity .
an overall visual extinction of 1 mag would increase the global luminosity and inferred sfr by about a factor of 11 ( hill et al .
1997 ) . at this rate
, it would have taken @xmath210 gyrs to transform the dynamical mass of 1.6 @xmath17 10@xmath60 m@xmath35 that is present within @xmath55 of the nucleus(garman & young 1986 ) into the dominant stellar component that we see today . as a check on these starbirth estimates
, we note that the measured h@xmath0 flux of 9.6 @xmath17 10@xmath22 erg @xmath14 s@xmath13 converts to a luminosity of l(h@xmath0 ) = 2.45 @xmath17 10@xmath61 erg / s , or only 5600 equivalent orion nebulae .
the origin of this discrepancy is deferred to the next subsection .
assuming case b recombination and multiplying the h@xmath0 luminosity by @xmath62 yields a photoionization rate of n@xmath63 = 1.8 @xmath17 10@xmath64 photons / sec , and corresponding star formation rate of 0.22 m@xmath35/yr .
the similarity of the fuv and h@xmath0-based sfrs , _ before correcting for extinction _ , suggests either that insignificant obscuration is present in the photometrically dominant starburst ring , or that significant obscuration exists with other mechanisms making up for the greater attenuation of the fuv emission relative to the longer - wavelength h@xmath0 emission .
such mechanisms include ( 1 ) an excess contribution of non - ionizing b - type stars to the total fuv emission , ( 2 ) a reduction of the total h@xmath0 emission due to h@xmath0 absorption by the atmospheres of b and a - type stars , and ( 3 ) the absorption of euv photons before they ionize the gas and induce h@xmath0 emission .
likely euv absorbers include the nebular dust associated with the h ii regions as well as the various metal species in the o - type stellar atmospheres themselves ( cf .
hill et al .
1997 ; waller , parker , & malumuth 1996 ) .
we conclude in the next subsection that modest obscuration plus non - ionizing contributions to the total fuv emission best explain the observed levels of h@xmath0 and fuv emission .
stellar absorption at h@xmath0 is probably less than a 10% effect overall , given h@xmath0 emission equivalent widths of order 100 in the ring , and stellar absorption equivalent widths peaking at less than 10 .
the global ( total ) @xmath65 color of m94 is 1.21 mag , which is characteristic of early - type disk galaxies with circumnuclear starburst activity ( e.g. ngc 1068 , ngc 3351 [ waller et al .
the @xmath25 colors shown in * figure 4b * do not vary as much as those found by beckman et al .
( 1991 ) in their photometric study .
bootstrapping our measured intensity ratio in the nuclear region to a color of @xmath25 = 0.45 mag as reported by beckman et al .
( 1991 ) , we obtain colors that range from @xmath25 = 0.3 mag in the rings starburst knots to @xmath25 = 0.5 mag in the arcs interior to the ring .
beckman et al .
( 1991 ) obtain much bluer colors of order 0.1 mag near the ring .
we attribute this discrepancy to our having used the h@xmath0-band image to remove contaminating line emission from the r - band image ( waller 1990b ) , thereby reducing the red flux from the knots in the ring by 20% .
comparison of the contaminated and uncontaminated r - band images confirm our attribution .
if the reddened arcs are caused by dust , the corresponding color excess relative to that of the starburst knots would amount to @xmath66 mag , or the equivalent visual extinction of roughly 0.6 mag in excess of the knots . allowing for bluer stellar populations in the knots
could reduce the estimate of excess extinction in the arcs to negligible levels , while inclusion of some internal reddening in the knots would increase the total estimated extinction by @xmath281 mag .
such low estimates for the extinction are similar to the spectroscopic results obtained by pritchett ( 1977 ) and by taniguchi et al .
( 1996 ) in their studies of the nuclear region , where color excesses of 0.27 mag and 0.15 mag , corresponding to @xmath67 mag and 0.48 mag , were respectively obtained .
photometry of the fuv and h@xmath0 emission from the starburst ring ( between projected radii of 30@xmath1 and 60@xmath1 ) yields a mean h@xmath0/fuv flux ratio ( in equivalent width units ) of 22 . by way of comparison
, we note that this flux ratio is somewhat lower than that found in the orion nebula ( 38 [bohlin et al . 1982 ] ) , within the range of flux ratios evident in m33 s giant h ii regions ( 19 to 69 [ parker , waller , & malumuth 1996 ] ) , and close to the mean of m51 s wide - ranging ghrs ( @xmath68 [ hill et al .
1997 ] ) . a model of steady - state star formation with a salpeter - type imf yields a somewhat lower ratio of 16 , while a modeled 2 myr - old starburst would match our result ( hill et al . 1995
[ see their fig .
such a short burst timescale is probably untenable , thus indicating some need to correct our fuv flux for extinction . for 1 mag of visual extinction ,
the resulting `` corrected '' h@xmath0/fuv flux ratio would be lower by a factor of @xmath25 , leading to a modeled burst age of 6 myr close to the age limit of the ionizing o stars . based on these considerations alone , we surmise that the fuv and h@xmath0 emission from the starburst ring are subject to no more than 1 mag of visual extinction ( on average ) .
we also can conclude that the similarity in uncorrected sfrs based on the global fuv and h@xmath0 fluxes is probably coincidental .
after correction for 0.51.0 mag of visual extinction , the h@xmath0 flux would be increased by a factor of 1.42.1 , while the fuv emission would increase by a factor of 3.311 . in other words ,
_ the fuv emission is tracing an additional non - ionizing population i component in the inner disk that is missed at h@xmath0 and , after correction for extinction , is contributing to the overall calculated rate of star formation_. the bi - symmetric fuv knots to the west and east of the ring have fluxes of @xmath69 and @xmath70 erg @xmath14 s@xmath13 @xmath13 , respectively .
this translates to the equivalent of 100 and 127 orion nebulae , respectively .
only the eastern knot shows significant h@xmath0 emission , with a flux of 2.57 @xmath17 10@xmath56 erg @xmath14 s@xmath13 and hence an h@xmath0/fuv flux ratio of 67 about twice that of the orion nebula thus indicating greater fuv obscuration than is present in orion or a hotter ionizing cluster .
examination of the spatially - filtered b - band image ( * figure 6b * ) reveals dark spiral - arm features at the position of the eastern knot which may correspond to obscuring dust clouds .
the western knot shows similar structures to one side .
the lack of significant h@xmath0 emission from this knot is difficult to explain with dust , and is most likely due to the presence of an older ( @xmath71 myr ) , non - ionizing stellar population .
* figure 8a * shows an average of 5 iue / swp spectra of the extranuclear emission from the inner disk .
based on the aperture mapping shown in * figure 3a * some emission from the nucleus may be present , but at low levels .
the wavelengths of commonly observed fuv absorption and emission lines are indicated on the spectrum for comparison .
although the averaged spectrum is of generally low s / n , several features can be identified . of these ,
the most prominent are the absorption blends of s ii ( 1250 - 1259 ) and si ii ( 1260,1265 ) , the p - cygni profile of s iv ( 1394,1443 ) , the absorption blends of fe iii ( 1601 - 1611 ) , al iii ( 1600 - 1612 ) , c ii ( 1720 - 1722 ) and al ii ( 1719 - 1725 ) , and part of the broad absorption complex of fe iii ( 1891 - 1988 ) .
the low - ionization lines , in particular , are most characteristic of b and a - type stars .
the strength of the blueshifted s iv absorption feature relative to that of c iv ( 1550 ) along with the strong si ii absorptions at 1260,1265 and 1527,1533 indicate a composite spectral type later than b3 but earlier than b8 ( fanelli et al .
1992 ; kinney et al . 1993 ; walborn , parker , & nichols 1995 ) . the absorptions at 1470 , 1780 , and 1790 remain unidentified . in emission
, there is some evidence for c iv ( 1548 - 1551 ) , he ii ( 1640 ) , n iii ] ( 1730,1750 ) , and c iii ] ( 1909 ) much of which may be the result of liner activity being scattered by the ism in the inner disk .
the c iii ] ( 1909 ) line emission may be responsible for filling in part of the broad stellar absorption complex of fe iii ( 1891 - 1988 ) . * figure 8b * shows an average of 2 iue / lwp spectra of the disk .
the most prominent spectral features are the absorption lines of fe ii ( 2609 , 2750 ) , mg ii ( 2800 ) , and mg i ( 2852 ) .
comparison with stellar spectra from the iue spectral atlas ( wu et al .
1991 ) shows that mg ii ( 2800 ) is unusually weak relative to mg i ( 2852 ) ; the line ratios in this spectral range being consistent with light dominated by late g - type stars , based on this cursory comparison .
however , the spectrum is too blue and the amplitude of the 2800 break too small for this to be the whole story .
comparison with the nuclear spectra ( iue lwr 12221,12238 as listed in table 2 ) shows evidence for a spectrally smooth blue component off the nucleus proper .
also , the mg ii 2800 break is roughly 50% smaller than that evident in the nuclear spectrum ( see below ) .
all the absorption features are broader in the off - nuclear spectrum , because the light there almost uniformly fills the aperture ( so centering shifts are not an issue ) .
the off - nuclear spectrum shown here is not only bluer but shows a diminished 2800 break amplitude .
measured as a ratio of flux f@xmath45 above and below 2800 , the ratio is 2.5 at nucleus and 1.8 away from it . using the spectral - break indices developed by fanelli et al .
( 1992 ) , we obtain for the off - nucleus averaged spectrum ( mg ii 2800 mg i 2852 ) = 0.31 mag , ( 2609/2660 ) = 0.45 mag , and ( 2828/2921 ) = 0.31 mag all of which are consistent with a @xmath72 color of 0.20.3 mag , or the equivalent of a late a - type main sequence star or late a to early f - type giant / supergiant star .
the off - nuclear continuum is likewise flatter in the swp range , with little trace of the emission - like features around 1900 seen in the nuclear spectrum ( see kinney et al . 1993 ) .
these properties can be accounted for , if recent star formation ( few times @xmath73 years ago ) contributes relatively more light off the nucleus than on it , consistent with an aging burst that is spatially more extended than the centrally - concentrated stars of the older disk and bulge .
* figure 8c * shows the spectrum of optical emission from the central d = 8.1 arcsec , as obtained with the lick 1-m telescope and ids .
absorption lines include the ca ii k&h lines at 3934 and 3968 , h@xmath74 ( 4101 ) , ch g - band ( 4300 ) , h@xmath75 ( 4340 ) , he ii ( 4686 ) , h@xmath76 ( 4860 ) , mg i ( 5170 ) , fe i ( 5270 ) , na d ( 5890,5896 ) , h@xmath0 ( 6563 ) , and atmospheric absorption bands of o@xmath77 ( 6867 ) and h@xmath77o ( 7186 ) .
emission lines are restricted to [ n ii ] ( 6584 ) and [ s ii ] ( 6731 )
. the h@xmath76 equivalent width of about 5.4 is huge for an old population typical of spiral bulges , even before any correction for line emission .
there has been star formation in this region not long ago .
the 4000- break is also suppressed , indicating the effective age is much younger than for typical early - spiral bulges . following the dressler and shectman ( 1987 ) definition ( ratio of f@xmath78 between 3950 - 4050 and 3750 - 3850 ) , the observed break has f@xmath78 = 1.6 , whereas values of 1.92.0 are more usual . from the ids
/ ipcs on the 2.5-m isaac newton telescope , we see that the nucleus ( inner 1 " or so ) is bluer at optical wavelengths than its immediate surroundings .
misalignment in the ipcs detector is less than 0.1 pixel , so this is not an obvious instrumental effect .
this nuclear blueing effect is substantial , amounting to about 30% over the 44005200 range ( more or less equivalent to @xmath79 mags , where the mean ( @xmath80 ) color of the bulge is 0.91 mags , which is already bluer than ordinary sa / sb bulges [ keel & weedman 1978 ] ) . outside this area ,
the color along the slit is quite constant out to 15@xmath1 from the nucleus , where the signal begins to die out .
this effect has no observed counterpart in the iue spectra , because of their courser spatial resolution .
the prominent h@xmath76 absorption line in the nuclear int / ipcs spectrum has fwhm 1820 , consistent with values seen in mid - to - late a stars .
thus , the data are consistent with seeing the main - sequence turnoff near this spectral type , rather than the supergiant dominance of a much younger population giving superficially similar spectral features ( as seen in the starbursting nucleus of ngc 4569 [ keel 1996 ] ) .
the optical spectra confirm our imaging result that h@xmath0 is in absorption with respect to the continuum from the underlying population of a and b - type stars .
by contrast , the neighboring line of [ n ii ] ( 6584 ) is seen in emission in the lick / ids spectrum . as previously noted , h@xmath76 is also dominated by photospheric absorption .
the ids / ipcs spectrum shows some h@xmath75 emission at least at the nucleus , that is not swallowed by the stellar absorption line .
* figure 9a * shows the rotation curve based on the h i observations of mulder and van driel ( 1993 ) , where key morphological features are noted .
this curve is qualitatively similar to that obtained from a recent interferometric mapping of co ( wong & blitz 2000 ) but is on - average 12% lower , due to the adoption of a 40@xmath8 inclination compared to their 35@xmath8 .
* figure 9b * shows the corresponding radial profiles of h i and h@xmath77 gas , the latter being derived from the co observations of gerin et al .
( 1991 ) .
the molecular gas component clearly dominates the inner disk s ism and may continue to prevail at higher radii , where the co emission has yet to be measured . the plotted extrapolation beyond the last reliable measurement of co emission
is intended to provide an _ upper limit _ on the total gas and thus a lower limit on the q stability index ( see below ) .
the interferometric measurements by wong & blitz ( 2000 ) yield even lower extrapolated surface densities of gas consistent with our extrapolated upper limit .
moreover , fir measurements with iras and the kao ( smith et al . 1991 ) reveal negligible dust emission beyond 60 arcsec radius further corroborating the gas upper limit used here .
the h i profile , though of lower amplitude , shows enhancements at the radii of the starburst ring and the bi - symmetric knots .
* figure 9c * shows the corresponding radial profile of the gravitational stability index ( q ) . here , we have considered the simplest case of the gaseous stability , without any coupling with the stellar component , such that @xmath81 where the epicyclic frequency @xmath82 is closely linked to the rotation curve v(r ) through @xmath83 and where @xmath84 is the gas velocity dispersion a quantity which is not well constrained but is probably of order 510 km / s . here ,
we leave @xmath84 as an unknown .
the resulting radial profile of q/@xmath84 out to the limits of the co observations shows a shallow minimum at the radius of the starburst ring .
although this minimum is too broad and shallow to explain the more discrete and prominent starburst ring , it may help to explain the relatively young ( @xmath85 yr ) stellar population pervading the inner disk .
* figure 10 * shows the orbital resonance diagram that results from the rotation curve along with the radii of key morphological features . a single wave pattern speed of 35 km s@xmath13 kpc@xmath13 ( where d = 4.6 mpc ) would place the nuclear bar inside the inner - inner lindblad resonance ( iilr ) , the starburst ring between the iilr & outer inner lindblad resonance ( oilr ) , the bisymmetric knots at the 4:1 `` ultra - harmonic '' resonance , the oval disk terminating at co - rotation ( cr ) , and the outer pseudo - ring at the outer lindblad resonance ( olr ) .
alternatively , a higher pattern speed of 56 km s@xmath13 kpc@xmath13 ( as modeled by mulder & combes ( 1996 ) ) would move the starburst ring just outside the oilr , the bisymmetric knots to the cr radius , and the edge of the oval disk to the olr .
this latter model , however , fails to account for the outer ring without invoking additional patterns .
further support for the proposed sequence of resonances comes from specific ratios of the corresponding radii ( cf .
buta 1986 ) . given a flat rotation curve ,
the modeled ratio of olr and uhr radii is r(olr)/r(uhr ) = 2.6 .
if these olr and uhr resonances are respectively traced by the outer pseudoring and bi - symmetric knots , then we obtain a ratio of radii equaling 2.54 closely matching the modeled ratio .
the theoretical ratio of olr and cr radii , being r(olr)/r(cr ) = 1.7 , is also well matched by the observed relative dimensions of the outer ring and oval disk , where a ratio of 1.5 is obtained .
admittedly , morphological tracers and rotation curves are insufficient to discriminate between these and other possible kinematic patterns ( cf .
buta & combes 1996 ) .
further progress on constraining the resonant dynamics in m94 will require analysis of the complete velocity field in the disk ( cf .
westpfahl & adler 1996 ; canzian & allen 1997 ) .
a complete h i mapping with the vla has been made recently , whose spectral and spatial resolution would be sufficient to derive a complete velocity field and its associated resonant states ( d. westpfahl , private communication ) .
until such an analysis is carried out , we think that the resonant state diagrammed in * figure 10 * best explains the most features with the fewest conditions .
through uv - optical imaging and spectroscopy , we have found new evidence for bar - mediated resonances as the primary drivers of evolution in m94 at the present epoch .
our observational results include evidence for ( 2 . ) an inner disk with diffuse fuv emission in concentric arcs that do not match the fine - scale structures or reddened structures at visible wavelengths . since @xmath86 is observed in absorption here , the uit / fuv image represents the first view of this non - ionizing but relatively young disk component .
uv - optical colors and spectral indices in the nucleus and inner disk that indicate b and a - type stars in the presence of modest extinction ( a@xmath87 @xmath281 mag ) along with some liner activity from the nucleus itself .
( 4 . ) a 2.2 kpc diameter starbursting ring at the perimeter of the inner disk that is bright at fuv , h@xmath0 , and radio - continuum wavelengths .
the level of starbirth activity in this inner ring rivals the levels observed in starbursting irregular galaxies such as ngc 1569 and ngc 4449 . the inferred star formation rate within the ring and inner disk amounts to 1.5 m@xmath35 yr@xmath13 sufficient to build up the stellar mass of the inner disk and bulge in @xmath210@xmath60 yr .
two 500-pc size fuv - emitting knots exterior to the ring on diametrically opposite sides of the nucleus .
the bisymmetric knots and starburst ring appear to be especially prominent parts of a complex spiral arm structure , as revealed in a spatially - filtered b - band image .
the starburst ring , bi - symmetric knots , oval disk , and outer pseudo - ring as signposts of resonant dynamics in the disk of m94 .
more specifically , the radii of these features match those of various orbital resonances , given a pattern speed of 35 km s@xmath13 kpc@xmath13 at our adopted distance and inclination .
these orbital resonances are most likely driven by some combination of the nuclear mini - bar and oval distortion in the disk .
( 8 . ) a shallow minimum of gravitational stability at the radius of the starburst ring that extends inward into the inner disk .
although too broad to explain the discrete starburst ring , the shallow minimum may help to explain the 10@xmath88-yr old stellar population interior to the ring .
although we can set a limit on the speed of outward or inward propagating star formation in the ring , we can not preclude the existence of such a mode . at a propagation speed of 35 km
/ s , a wave initiated in the nucleus could traverse the inner disk to the radius of the current starburst ring in only 31 myr .
therefore , it is possible that the @xmath89-yr old stellar population detected in the inner disk is the result of such an outward propagating wave .
the striking difference in emission morphologies at fuv and red wavelengths provides further support for the starburst ring being a transient phenomenon which does not persist at any one radius for very long . either these resonant phenomena come and go , as the oval distortions undergo secular evolution , or their operating radii migrate in response to other dynamical influences on the stars and gas ( cf .
combes 1994 ; combes et al . 1995 ; friedli & benz 1995 ) .
otherwise , one must invoke strong radial inflows of stars from the starburst ring to populate the inner disk and bulge , a feat requiring unusual circumstances
e.g. mergers .
the results reported herein may have important implications with regard to observations of the most distant observable galaxies . at redshifts of 15 ,
the 2-kpc diameter starburst ring in m94 would subtend angles of only ( 0.7@xmath1 1.0@xmath1)h@xmath7/75 in an einstein - de sitter universe ( q@xmath7 = 1/2 ) and ( 0.3@xmath1 0.2@xmath1)h@xmath7/75 in an open ( milne ) universe ( q@xmath7 = 0 ) ( cf .
narlikar 1983 ) .
the uv - bright nuclear rings evident in ngc 1097 , ngc 1317 , ngc 1433 , ngc 1512 , ngc 2997 , ngc 4321 , and ngc 5248 ( maoz et al . 1995 ;
maoz et al . 1996 ; kuchinsky et al . 2000 ; marcum et al .
2000 ) would subtend even smaller angles at the same redshifts .
moreover , nuclear rings tend to have higher fuv surface brightnesses than their larger counterparts the inner ring in m94 being a remarkable exception . therefore , some of the `` core - halo '' morphologies that are evident at high - redshift in the restframe fuv ( cf .
giavalisco et al .
1997 ) may , in fact be marginally - resolved representations of galaxies with starburst rings in their centers .
gravitationally - lensed galaxies are fortuitously magnified , enabling resolutions of their structure at high s / n .
an important precedent in this regard is the gravitationally - lensed `` pretzel galaxy '' which lies behind the galaxy cluster 0024 + 1654 at an estimated redshift of 1.2 1.8 ( colley et al . 1996 ; tyson et al . 1997 ) .
detailed reconstructions of the multiply - lensed galaxy show a clear annular morphology on a scale of several kpc .
if m94 and other nearby ringed galaxies can be used as current - epoch analogues , the `` pretzel galaxy '' and perhaps other marginally - resolved `` core - halo '' galaxies at high redshift may represent youthful inner disks and bulges growing under the organizing influence of oval or bar asymmetries(friedli & benz 1995 ; waller et al .
conversely , if evidence for starburst rings at high redshift proves to be sparse , then massive inner disks featuring ring - bar dynamics have yet to form in most systems , or starbursting bulges are masking their presence .
we thank david adler , gene byrd , francoise combes , daniel friedli , and david westpfahl for generously providing consultation on the dynamics of ringed - barred spiral galaxies .
whw is grateful to john huchra and the oir division of the harvard - smithsonian center for astrophysics for their kind hospitality during his visiting appointment at the cfa .
whw also thanks eric murphy and christine winslow , tufts undergraduates who helped craft some of the graphics .
uit research is funded through the spacelab office at nasa headquarters under project number 440 - 51 .
we are deeply grateful to the crew of sts-67 and the many people who helped make the _
astro-2 _ mission a success .
whw acknowledges partial support from nasa s astrophysics data program ( 071 - 96adp ) .
giavalisco , m. , et al .
1997 , in the ultraviolet universe at low and high redshift : probing the progress of galaxy evolution , eds . w. h. waller , m. n. fanelli , j. e. hollis , & a. c. danks ( new york : aip press ) , p. 221
marcum , p. m , et al .
1997 , in the ultraviolet universe at low and high redshift : probing the progress of galaxy evolution , eds .
w. h. waller , m. n. fanelli , j. e. hollis , & a. c. danks ( new york : aip press ) , p. 88
oconnell , r. w. 1997 , in the ultraviolet universe at low and high redshift : probing the progress of galaxy evolution , eds .
w. h. waller , m. n. fanelli , j. e. hollis , & a. c. danks ( new york : aip press ) , p. 11
tyson , j. a. , kochanski , g. , & dellantonio , i. 1997 , in the ultraviolet universe at low and high redshift : probing the progress of galaxy evolution , eds . w. h. waller , m. n. fanelli , j. e. hollis , & a. c. danks ( new york : aip press ) , p. 204
waller , w. h. , et al .
1997 , in the ultraviolet universe at low and high redshift : probing the progress of galaxy evolution , eds .
w. h. waller , m. n. fanelli , j. e. hollis , & a. c. danks ( new york : aip press ) , p. 39 waller , w. h. , varosi , f. , boulanger , f. , & digel , s. w. 1998 , in new horizons from multi - wavelength sky surveys , iau symp .
179 , eds .
b. j. mclean , d. a. golombek , j. e. hayes , & h. e. payne b. lasker & m. postman ( dordrecht : kluwer ) , p. 194 , ( astro - ph/9612233 ) | the dynamic orchestration of starbirth activity in the starburst - ringed galaxy m94 ( ngc 4736 ) is investigated using images from the ultraviolet imaging telescope ( fuv - band ) , hubble space telescope ( nuv - band ) , kitt peak 0.9-m telescope ( h@xmath0 , r , and i bands ) , and palomar 5-m telescope ( b - band ) , along with spectra from the international ultraviolet explorer and lick 1-m telescopes .
the wide - field uit image shows fuv emission from ( a ) an elongated nucleus , ( b ) a diffuse inner disk , where h@xmath0 is observed in _ absorption _ , ( c ) a bright inner ring of h ii regions at the perimeter of the inner disk ( r = 48@xmath1 = 1.1 kpc ) , and ( d ) two 500-pc size knots of hot stars exterior to the ring on diametrically opposite sides of the nucleus ( r = 130@xmath1 = 2.9 kpc ) .
the hst / foc image resolves the nuv emission from the nuclear region into a bright core and a faint 20@xmath1-long `` mini - bar '' at a position angle of 30 deg .
optical and iue spectroscopy of the nucleus and diffuse inner disk indicates a @xmath210@xmath3 yr - old stellar population from low - level starbirth activity blended with some liner activity .
analysis of the h@xmath0 , fuv , nuv , b , r , and i - band emission along with other observed tracers of stars and gas in m94 indicates that most of the star formation is being orchestrated via ring - bar dynamics involving the nuclear mini - bar , inner ring , oval disk , and outer ring .
the inner starburst ring and bi - symmetric knots at intermediate radius , in particular , argue for bar - mediated resonances as the primary drivers of evolution in m94 at the present epoch .
similar processes may be governing the evolution of the `` core - dominated '' galaxies that have been observed at high redshift .
the gravitationally - lensed `` pretzel galaxy '' ( 0024 + 1654 ) at a redshift of @xmath21.5 provides an important precedent in this regard . |
in the chiral su(2 ) limit ( @xmath8 ) the @xmath0 decay amplitude is precisely known to order @xmath9@xcite .
the amplitude is @xmath10 in the low energy chiral counting .
it is determined entirely by the anomaly induced on the divergence of the axial current @xmath11 by the electromagnetic interaction , and is expressed in terms of the only two available quantities the fine structure constant @xmath9 and the pion decay constant @xmath12the decay width in this limit is thus given by @xmath13 .
the explicit breaking of chiral @xmath14 symmetry induced by non - zero u- and d - quark masses generates corrections to the chiral limit result , and it is the purpose of the present work to evaluate these corrections as well as to understand their origin . in order to achieve this goal it it is crucial to perform the analysis in the extended framework of three flavors supplemented by the @xmath1 expansion in order to include explicitly the @xmath4 degree of freedom . in this three flavor framework
the corrections turn out to be of two types : * those due to isospin breaking ( mixing corrections ) that are proportional to @xmath15 or to @xmath16 , both giving contributions to the decay amplitude that , according to the counting defined in the next section , are @xmath10 , _
i.e. _ , of the same order as the leading term , and * those proportional to @xmath17 that are of subleading order@xmath2and which stem from different sources , as shown below .
the inclusion of such corrections is crucial for a prediction of the @xmath18 width at the @xmath19 level , which is the level of theoretical precision required by the forthcoming dramatic improvement expected in the experimental measurement of the @xmath20 width via the primakoff effect .
the primex experiment at jefferson lab @xcite is aiming at a measurement with an error about @xmath21 , which is several times smaller than the @xmath22 uncertainty in the current world - average value @xmath23 @xcite .
however , this quoted experimental uncertainty is open to question as can be seen by the large dispersion of the experimental results which suggests that the errors of the individual experiments have been underestimated .
indeed , a direct measurement gives @xmath24@xcite , a production experiment in @xmath25 collisions yields @xmath26@xcite , while primakoff effect experiments , all dating back to the early seventies , give disparate values : a large number@xmath27 @xcite , and two that are consistent with the current world average@xmath28 @xcite and @xmath29 @xcite .
this unsatisfactory experimental situation together with the rather precise theoretical prediction derived in the present work clearly lend great significance to the upcoming primex measurement . within the two - flavor framework ,
wherein the strange quark is integrated out , any corrections to the amplitude are @xmath2 and reside entirely in @xmath30 @xcite or in the @xmath2 odd - intrinsic parity chiral lagrangian@xcite , also known as the @xmath2 wess - zumino ( wz ) lagrangian . at leading order ,
the ensuing theoretical prediction ( taking @xmath31 ) is @xmath32 , a result that agrees well with the experimental world average within its generous error .
the analysis within @xmath33 , however , does not provide insight on the origin of the @xmath34 wz contribution just mentioned .
such an insight _ can _ be gained by instead performing the analysis in the _
three_-flavor framework , as has been shown by moussallam@xcite .
in particular , he pointed out that the primary corrections to the @xmath20 width result from the leading order isospin breaking effects mentioned above in i ) , which stem from the @xmath35-induced mixing between the pure isospin state @xmath20 and pure @xmath36 states @xmath7 and @xmath4 ( to be denoted below by @xmath37 , @xmath38 and @xmath39 respectively )
. the present work confirms that such isospin breaking corrections persist as the dominant effect when next to leading order ( nlo ) corrections are included . in this work
then the @xmath40 decay rate is evaluated to nlo within @xmath41 chiral perturbation theory wherein the @xmath4 meson is included consistently by means of the @xmath1 expansion since in the large @xmath42 limit the @xmath4 becomes a goldstone boson .
such a framework was recently developed by herrera - sikldy , latorre , pascual and taron@xcite and by kaiser and leutwyler@xcite , who showed that a simultaneous chiral and @xmath1 expansion leads to an effective theory for the pseudoscalar nonet that is not only internally consistent but is also very useful in practice , as the present work shows .
the chief result of our paper is that the @xmath40 width is enhanced by about @xmath43 from the lowest order chiral anomaly prediction , a result expected to hold within an uncertainty of @xmath44 after nlo contributions are included .
the magnitude of this enhancement agrees with that obtained in the analysis of moussallam @xcite , where the nlo corrections were not implemented in a consistent fashion , as done in the present work .
the decay amplitudes of @xmath20 , @xmath7 and @xmath4 into two photons can be obtained from the ward identities satisfied by the three axial vector currents @xmath45 ( @xmath46 ) , where @xmath47 are @xmath5 generators ( @xmath48 being the @xmath49 generator ) normalized via @xmath50 . in the presence of the strong and electromagnetic interactions , the divergence of the axial vector current
is given by : @xmath51 where @xmath52 is the quark mass matrix and @xmath53 is the electric charge operator .
here @xmath54 , @xmath55 being the electromagnetic field tensor , and similarly @xmath56 , @xmath57 being the gluon field , and the ellipsis denotes terms irrelevant to this work . the two - photon amplitudes can be obtained by considering the matrix elements @xmath58 where @xmath59 , @xmath60 , and @xmath61 .
if @xmath62 denotes the total momentum of the final two - photon state , then in the limit of small @xmath63 eq . [ div2 ] admits a low energy expansion and can be expressed as : @xmath64 where @xmath65 are the two - photon amplitudes , the ellipsis denotes contributions from excited mesons as well as from the continuum , all being of nlo or higher , and the mass eigenstates @xmath66 that correspond to the physical @xmath20 , @xmath7 and @xmath4 are given by @xmath67 where the mixing matrix that diagonalizes the mass matrix is parametrized in terms of euler angles @xmath68 , @xmath69 and @xmath70 @xmath71 where @xmath72 and @xmath73 .
here , for small mixing , the projection of the physical @xmath20 onto @xmath38 is given by the angle @xmath74 , the projection of the physical @xmath7 onto @xmath39 is approximately given by @xmath70 ( @xmath75 can therefore be identified with the well known @xmath76 mixing angle as it is customarily defined ) , and the projection of the physical @xmath20 onto @xmath39 is given by @xmath77 .
the nlo_i.e .
_ , @xmath2corrections in eq .
[ div3 ] reside in the terms displayed explicitly through their dependence on the masses and decay constants , as well as in pieces that stem from continuum and excited states . in works that preceded that of moussallam @xcite , such as references@xcite and @xcite , such mixing corrections as well as the nlo effects of the latter kind were disregarded .
ignoring such effects implies that only nlo corrections which are absorbed into the @xmath20 decay constant remain , as it was shown in @xcite ; in that case and taking @xmath31 , the predicted @xmath20 width is the previously mentioned @xmath78 .
as shown here , however , disregarding mixing in particular constitutes a very poor approximation . in the presence of mixing the pseudoscalar decay constants form a @xmath79 matrix defined by the matrix elements of the axial vector currents , which connect the pseudoscalar mesons to the vacuum : @xmath80 indeed this decay - constant matrix contains all that is needed to calculate the three two - photon amplitudes , except the contributions stemming from the @xmath10 wz lagrangian , and its evaluation is the centerpiece of the present work .
as already mentioned in the introduction , the present analysis includes the @xmath4 as an explicit degree of freedom , which , in order to be consistently implemented in an effective theory , requires the validity of the @xmath1 expansion , wherein , taking the chiral @xmath81 limit , @xmath82 .
thus , in the framework of the @xmath1 expansion @xmath83 should be considered as `` small , '' and its explicit inclusion becomes consistent with having a simultaneous low energy chiral expansion .
this explicit inclusion of the @xmath4 in the low energy expansion implies that @xmath83 must count as a quantity of @xmath84 , which in turn implies that @xmath1 should be counted as a quantity of the same order .
indeed , a consistent effective theory can be formulated with such a counting scheme@xcite , and it is interesting to note that @xmath1 and the magnitude of @xmath36 breaking are comparable in size in the real world . taking the chiral limit@xmath85in eq . [ div3 ] and neglecting the electromagnetic piece_i.e .
_ , the @xmath86 term , equating the residues at @xmath87 lead to the well known relation @xmath88 with @xmath89 being the @xmath4 mass in the chiral limit . here
the lowest order result @xmath90 was used , where @xmath91 is the pion decay constant in the chiral limit . in the large @xmath42 limit
, @xmath91 scales as @xmath92 , while in the chiral limit the ratio @xmath93 is equal to unity up to corrections of order @xmath1 . on the other hand , for non - vanishing quark masses equating the residues in eq .
[ div3 ] yields @xmath94 where @xmath95 .
as is well known , the @xmath63 dependence of the lhs appears first at @xmath2 @xcite , which would affect the two - photon amplitudes at @xmath96 , _
i.e. _ , beyond the accuracy needed in this work .
thus , it is consistent to use eq .
[ pseudoscalar ] at @xmath95 to represent the lhs in the entire low @xmath63 domain . at lo@xmath10
then , eqns .
[ div3 ] , [ metap ] , and [ pseudoscalar ] yield immediately the result for the two - photon amplitudes @xmath97 at this order the decay constant matrix is simply given by @xmath98 where @xmath99 is the mixing matrix obtained from the @xmath84 mass formulas .
of course , the result of eq .
[ loamplitudes ] coincides with the result obtained by means of the @xmath10 wz term including explicitly the singlet pseudoscalar .
the purpose of carrying out the above ward identity analysis is , however , to make more transparent the origin and structure of the higher order corrections and to set the stage for the inclusion of nlo modifications .
the leading order mass formulas are obtained from the @xmath84 chiral lagrangian .
the @xmath5 field is parametrized by the unitary matrix : @xmath100 where @xmath101 at lo .
the @xmath84 lagrangian with the standard definitions of covariant derivatives and sources @xmath102 @xcite is given by : @xmath103 and the mass matrix in the @xmath37 , @xmath38 , @xmath39 sector of interest resulting from @xmath104 is : @xmath105 using the leading order mass formulas_e.g .
_ , @xmath106 , with @xmath107and extracting isospin breaking from the @xmath108 mass difference via dashen s theorem to eliminate the em contributions , a best fit to the masses yields a singlet mass @xmath89 of approximately 850 mev , and the euler angles : @xmath109 , @xmath110 , and @xmath111 .
this fit yields then @xmath112 , which is substantially larger than the value @xcite @xmath113 that arises in the limit @xmath114 , when only octet degrees of freedom are included .
the lo mass matrix , however , gives a poor result for the masses .
in particular the @xmath7 mass is too low by almost 50 mev , a problem that is generic at lo in the low energy and @xmath1 expansions@xcite . using the relations @xmath115 which connect the decay amplitudes and associated widths , the fitted parameters at lo lead to a @xmath0 decay width of 8.08 ev , which is @xmath43 larger than the leading order result @xmath116 obtained in the two - flavor framework wherein mixing effects are moved to nlo .
it should be noted , however , that the two - photon widths of the @xmath7 and @xmath4 predicted in this leading order fit are too large , the first being 22% and the second 20% larger than the corresponding experimental values .
one of the chief reasons for this disagreement is that @xmath36 breaking in the pseudoscalar decay constants is not included at lo .
the @xmath76 mixing angle @xmath75 turns out to be @xmath117 at lo , and will be reduced by almost a factor of two when nlo corrections are included .
the magnitude of the observed lo enhancement of the @xmath20 width is in line with the ratio of isospin breaking versus @xmath118 ( @xmath119 ) and versus @xmath89 ( @xmath120 ) , and is therefore not surprising .
it is important to note that the corrections due to mixing with @xmath7 and with @xmath4 are of the same sign and of similar magnitude .
this point is the primary reason why the @xmath4 _ must _ be explicitly included for a full understanding of the mixing effects .
although @xmath121 mixing is smaller than the @xmath122 mixing , the bare singlet state has an intrinsic two - photon amplitude larger by a factor @xmath123 that compensates for the smaller mixing .
overall , however , the lo fit is poor , and dramatic improvement results when the nlo corrections are included , as shown in the following section .
at nlo the amplitudes receive corrections of two types those that affect the decay constants and mixing angles , and those that stem from the presence of excited states and which are included in the @xmath2 wz lagrangian .
the first type of correction requires the determination of the masses and decay constants to nlo and can be obtained in the standard fashion by calculating the two - point functions of axial vector currents , where the relevant diagrams are shown in fig .
1 . up to @xmath84 and @xmath124
such two - point functions require only the effective lagrangians @xmath104 and @xmath125 .
chiral loop corrections are @xmath126 , and therefore beyond the precision of the present calculation , but such loop corrections will be calculated merely as a means to estimate the size of possible contributions from terms of that order and also as a test on the practical validity of the @xmath1 expansion . the lowest order lagrangian has already been given in eq .
[ l2 ] , while the next to leading order lagrangian @xmath125 has the form given by @xcite @xmath127 where only the terms relevant to this work are included . in the presence of the @xmath36 singlet axial vector source field @xmath128 , @xmath129 . at this point
it is important to note that the singlet axial current has non - vanishing anomalous dimension @xcite , which implies that some low energy constants ( lecs ) as well as the singlet field @xmath39 must be renormalized ( the corresponding renormalized quantities will thus depend on the qcd renormalization scale @xmath130 ) @xcite .
since the renormalization of the axial current is subleading in @xmath1 , such dependence appears first at the level of @xmath125 lecs .
it has been found in particular that the lecs @xmath131 , @xmath132 , @xmath133 and @xmath134 , all of which are subleading in @xmath1 , must be renormalized@xcite , implying that the values of these lecs will depend on the value of the scale @xmath130 .
other quantities such as @xmath135 and the singlet @xmath39 field must be renormalized as well and depend on @xmath130 through the renormalization factor @xmath136 associated with the singlet axial current .
it is very convenient to make use of the asymptotic freedom of qcd to set @xmath130 arbitrarily large and give the values of the lecs in that limit .
indeed , the renormalization factor of the axial current @xmath136 evolves to a fixed point that can be taken to be @xmath137 as @xmath138 .
all quantities given in the following that have a dependence on @xmath130 are then taken in this limit .
it is well known that the low energy constants @xmath139 and @xmath140 are @xmath141 , while @xmath142 and @xmath143 are subleading and @xmath144 @xcite and are needed in order to renormalize the one - loop contributions from @xmath104 . with @xmath4 as an explicit degree of freedom , @xmath145 is also subleading in @xmath1 .
the renormalized pieces of the subleading lecs are therefore set to zero at the chosen chiral renormalization scale @xmath146 in our analysis .
on the other hand , the lecs @xmath131 and @xmath132 are @xmath147 and the corresponding terms in the lagrangian are @xmath148 and thus must be included in the calculation .
the low energy constant @xmath131 provides an @xmath149 correction to the @xmath4 decay constant , while both @xmath131 and @xmath132 affect entries in the mass matrix involving the @xmath4 at order @xmath150 .
finally , the terms involving the lecs @xmath133 and @xmath134 are of @xmath151 which is beyond the order of the present calculation and therefore their renormalized pieces are set to vanish as well .
the renormalized lecs are defined in the usual @xmath152 renormalization scheme@xcite : @xmath153 the @xmath154-functions @xmath155 associated with the lecs @xmath156 , and @xmath157 associated with @xmath158 that result from the chiral one - loop calculation are given by @xcite : @xmath159 , @xmath160 , @xmath161 , @xmath162 , @xmath163 , @xmath164 , @xmath165 , @xmath166 , and @xmath167 .
the two - point functions of axial currents can be written in momentum space as @xmath168 where the term explicitly shown contains the light pseudoscalar poles and @xmath169 is the propagator of the corresponding mass eigenstate . from the location of the low energy poles and the residues of the two - point function
the light pseudoscalar decay constants and masses are extracted , yielding @xmath170 the corresponding decay constants are given by : @xmath171 where the following definitions were made : @xmath172 and , @xmath173[\lambda^b,\lambda^d]\rangle\nonumber\\ { \cal{m}}^{abcd}&=&\frac{1}{2}\sum_{perm.\{\sigma\ } } \langle{\cal{m}}_q\lambda^{\sigma_a } \lambda^{\sigma_b}\lambda^{\sigma_c}\lambda^{\sigma_d}\rangle.\end{aligned}\ ] ] throughout , the terms whose renormalized lecs are set to vanish have not been displayed explicitly .
it is interesting to note that @xmath174 does not receive any loop contributions from the singlet pseudoscalar mode , implying that @xmath175 is also free of such contributions .
it is useful at this point to give the explicit expressions of the masses and decay constants at nlo disregarding the chiral logarithms . for the decay constants
the above expressions lead to @xmath176 while for the masses , @xmath177 where @xmath178 .
the second class of nlo corrections can be grouped into a single term contained in the @xmath2 odd - intrinsic parity wz lagrangian@xcite @xmath179 ( there exists a second term @xcite that , upon the explicit inclusion of the singlet pseudoscalar meson , becomes subleading in @xmath1 and is therefore neglected . )
the low energy constant @xmath180 has vanishing @xmath154-function and its value can be estimated by means of a qcd sum rule for the general three - point function involving the pseudoscalar density and two vector currents and saturating the spectral function in the hadronic sector with the states indicated in fig .
( 2 ) , yielding@xcite : @xmath181 here the @xmath182 contribution is determined by the masses and decay constants of the vector mesons ( the vector meson mass is naturally taken to be @xmath183 ) and is represented by fig . ( 2b ) , while the contribution proportional to @xmath184 is determined by excited pseudoscalars , such as the @xmath185 , and is represented by fig .
this latter contribution can be estimated within a model @xcite and its expected size is at most one third of the magnitude of the vector meson contribution .
since this is similar to the level of uncertainty expected in the sum rule evaluation , the @xmath184-piece will be disregarded henceforth .
as shown by the numerical analysis below , the effects on the @xmath20 width due to the @xmath186 with @xmath180 as estimated above are of similar magnitude to the rest of the nlo corrections and in the range of 0.5 % . at this point
the nlo two - photon amplitudes can be explicitly given : @xmath187 and the term proportional to @xmath180 can be obtained in two equivalent ways , either by determining the contribution from @xmath188 to the matrix elements @xmath189 in eq . [ div3 ] , or by directly calculating the contribution to @xmath190 due to that effective lagrangian .
note that the contribution from @xmath188 has the same scaling in @xmath42 as the leading one . in eq .
[ nlodecay ] the factor @xmath191 can be expressed using the lo mass formulas , namely @xmath192 at nlo the extraction of the ratio @xmath193 that characterizes the size of isospin breaking should be improved by including nlo corrections to dashen s theorem . over time
several works have shown that the corrections are sizeable . from the early works of donoghue , holstein and wyler @xcite and bijnens @xcite , and more recent works @xcite ,
it is well established that the mass difference @xmath194 left after subtracting the em contribution is larger than the one predicted by dashen s theorem . while dashen s theorem predicts @xmath195 , after the corrections have been implemented it is estimated that @xmath196 .
the mass difference is slightly smaller if the chiral logarithms are neglected following the approach of this work . in such a case @xmath197 .
these corrections to dashen s theorem translate naturally into a smaller value for r , as shown in the analysis that follows .
there is additional evidence that r is overestimated by applying dashen s theorem , and that comes from the decay @xmath198 . at lowest order the decay rate , which is proportional to @xmath199 , is found to be a mere 66 ev @xcite , which is almost a factor of four smaller than the experimental value of @xmath200 ev . at nlo in the chiral expansion the rate is increased to @xmath201 ev @xcite , while dispersive analyses @xcite give @xmath202 ev , which is still substantially below the experimental value .
clearly one way to make up for the difference is to increase @xmath203 .
because of the large uncertainties in both the experimental and theoretical side it is difficult to be precise , but it seems that the increment implied by the violations to dashen s theorem mentioned above is in line with the enhancement required to explain the observed @xmath204 width . in principle
a precise measurement of the @xmath20 width could provide an independent determination of @xmath205 .
however , as shown by the results of the present analysis , the @xmath20 width is affected by the corrections to dashen s theorem only at the level of 0.5% , which is unfortunately well below the experimental error of 1.4% aimed at by primex and about the same as the 0.6% uncertainty due to the 0.3% error in the the experimental value of @xmath206 @xcite
. there are nine low energy constants to be determined@xmath91 , @xmath207 ( @xmath208 ) , @xmath89 , @xmath139 , @xmath140 , @xmath131 , and @xmath132 and these can be found by solving for the observables : @xmath209 , @xmath210 , @xmath211 , @xmath212 , @xmath213 , @xmath214 , @xmath194 ( which as mentioned before , is 5.25 mev at lo , while at nlo and disregarding ( including ) the chiral logarithms in the corrections to dashen s theorem is 6.47 mev ( 6.97 mev ) , @xmath215 , and @xmath216 .
note that the tenth lec @xmath180 can not at this stage be extracted phenomenologically and thus its value is taken according to the estimate made above .
in table i the second column displays the lo results , and the next three columns display three different nlo fits , namely : * nlo#1 includes terms of @xmath2 and @xmath151 in the decay amplitude _i.e .
_ chiral logarithms are omitted .
* nlo#2 includes chiral logarithms , which are @xmath217 , and the renormalization scale @xmath146 is set equal to @xmath218 .
* nlo#3 is identical to nlo#1 but sets @xmath219 _i.e . _
excludes the @xmath2 wz contributions .
.results for the lecs for several different fits , the lo fit and three nlo fits as explained in the text .
the lecs that depend of the qcd renormalization scale @xmath130 correspond to the limit @xmath138 .
although all the lecs are the renormalized ones , only those in the nlo fit # 2 depend on the chiral renormalization @xmath146 that is taken to be equal to @xmath218 .
[ cols="^,^,^,^,^",options="header " , ] it should be noted that the mass of the @xmath4 in the chiral limit and at nlo in @xmath1 is given by @xmath220 mev , which is slightly high leaving not enough room for the piece linear in the quark masses . this linear contribution is suppressed by the rather large value of @xmath132 which leads to a very small value of @xmath221 .
this cancellation between the leading and subleading in @xmath1 contributions seems to indicate some difficulty with the @xmath1 expansion for the masses .
ther first manifestation of this problem is of course in the problem found with the @xmath7 and @xmath4 masses at lo .
although this problem has a minor impact on @xmath20 width , it certainly deserves further study .
it should be noted that the mass difference @xmath222 has been given as input in fit # 2 since its value emerges as too large if left unconstrained .
the reason why it is too large can be traced back to the chiral logarithms generated by the @xmath4 .
it seems , therefore , that requiring the subleading renormalized lecs to vanish is not such a good approximation when such chiral logarithms are included .
table ii lists the associated predictions for various quantities of interest , in particular the @xmath20 width .
it is evident from the nlo fits that @xmath223 is rather stable and always within @xmath224 of the leading order result .
this is within the expected range of the nlo corrections .
as shown by comparison of the first and third nlo fits , the correction from the @xmath2 wz lagrangian @xmath186 reduces the @xmath20 width by @xmath225 , a magnitude in line with the fact that such a correction is controlled by the ratio @xmath226 .
the chiral - logarithm contributions to the amplitudes as shown by fit # 2 , provide an increase of order @xmath225 to the @xmath20 width . since these are subleading corrections of @xmath217 to the decay amplitude
, they are somewhat larger than the @xmath227 expected from the ratio @xmath228 that determines them . this problem is similar to the one with the pion mass difference just mentioned ; in this case the @xmath4 loops affect the mixing angles producing a larger than expected correction to the rate . indeed
, turning off the chiral logarithms generated by the @xmath4 the @xmath20 width is essentially identical to the result in fit # 1 .
it is important to note that the mixing angles are substantially modified at nlo : in particular , the @xmath229 mixing angle is found to be @xmath230 in the three fits , which is @xmath231 smaller than the lo result of @xmath232 , but still larger than that obtained at lo within @xmath36 and given in eqn .
[ su3mixing ] .
the latter is chiefly a consequence of the corrections to dashen s theorem .
the @xmath233 mixing @xmath234 goes from @xmath235 at lo to approximately @xmath236 at nlo ; finally , the @xmath76 mixing angle is dramatically reduced to about @xmath237 from its lo value of @xmath238 . in view of these substantial corrections to the mixing angles ,
the stability of the @xmath20 width is nontrivial : besides the corrections due to the @xmath2 wz lagrangian , the decay amplitude is determined by the decay constant matrix @xmath239 , which is affected by the mixing of states as well as by the nlo corrections contained in the decay constant matrix @xmath240 given in eqn .
it turns out that the entries in @xmath241namely @xmath242affecting the @xmath20 amplitude remain stable well within the natural size of the nlo corrections . in order to asses the theoretical uncertainty of the analysis of the @xmath20 width
, an estimate of the magnitude of em corrections should also be given .
such corrections are of order @xmath243 , which puts them in the @xmath244 range .
note also that the value of @xmath245 being used is that of @xmath206 which has an em correction .
this correction has been studied @xcite and is given by @xmath246 where the low energy constants that determine the coefficient @xmath247 can be estimated in a resonance saturation model that gives @xmath248 , thus leading to the estimate @xmath249 which is well within the experimental uncertainty in the value of @xmath206 , and implies a correction to the @xmath20 width consistent with the range mentioned above .
it seems , therefore , safe to neglect electromagnetic corrections at the level of precision of this work .
although the @xmath76 complex is not the primary focus of this work , the analysis carried out illuminates crucial aspects of this system . at
lo the description is rather poor , in particular because the two - photon widths depart quite substantially from the experimental world averages .
indeed , @xmath250 versus the world average experimental value of @xmath251 , and @xmath252 vs @xmath253 .
these latter disagreements are mostly due to the large @xmath76 mixing angle that results from the lo mass formulas , and the fact that at lo all decay constants are set to be equal to @xmath12 . at nlo
the scheme that emerges is the one already found in other works @xcite , where the mixing angle of the pure @xmath5 states is in the proximity of @xmath237 rather than the @xmath254 obtained in lo , and where the decay constant matrix can be parametrized by means of two angles and two decay constants @xcite .
following the conventions and notation in@xcite , the present analysis gives ( quantities are denoted in boldface not to be confused with quantities defined heretofore in the text ) : @xmath255 mev , @xmath256 mev , @xmath257 @xmath258 , and the angles @xmath259 and @xmath260 for the three nlo fits are respectively @xmath261 , @xmath262 and @xmath263 .
the difference @xmath264 turns out to be between @xmath265 and @xmath266 , to be compared with the @xmath265 obtained in @xcite ( in a nlo estimate in that reference a value of @xmath267 is obtained , which departs substantially from the one of the current analysis ) .
there exist numerous studies of the @xmath76 complex .
it makes sense only to compare results with those using the two - angle scheme @xcite .
although some of the analyses in these references are purely phenomenological , these results are in general in good agreement with the results obtained in this work .
the quark mass ratios obtained in the different fits deserve comment .
the ratio @xmath268 is in good agreement with the standard value @xmath269 obtained in su(3 ) @xcite ; in fit # 2 it is a couple of standard deviations smaller , most likely because the chiral logarithms included do not represent the full nnlo contributions . in all
, this is not surprising as the assumption that the low energy constants that are subleading in @xmath1 can be disregarded is one of the important assumptions in the extraction of the standard ratios .
however , a comment is in order : it is observed that using lo mass formulas in the nlo results _i.e_. , expressing quark masses in terms of meson masses squared in the nlo terms leads to a fit that is less stable and generates large corrections to the quark mass ratios .
the ratio @xmath205 is smaller here than the standard value @xmath270 , and this is simply because the corrections to dashen s theorem have been included .
the values for @xmath205 in fits # 1 and # 3 are about one standard deviation smaller than the standard value , while in # 2 the chiral logarithms involving the @xmath4 loops give a substantial reduction ( when these are turned off @xmath205 is similar to the result in the other fits ) .
an interesting observation is that setting @xmath271 leads to an inconsistent fit and a value for the @xmath20 width of 8.5 ev .
the decay rate for @xmath272 has been calculated within a combined chiral and @xmath1 expansion . at leading order in the expansion ,
the isospin - breaking induced mixing of the pure @xmath5 states increases the size of @xmath273 from the value @xmath274 ev predicted by the lowest order chiral anomaly by more than 4.5% .
this effect is largely due to the fact that the contributions from mixing with the @xmath38 and @xmath39 add constructively , and are of similar magnitude .
however , at lo the resulting @xmath275 widths are found to be too large , and in general the fit is quite poor .
there is a clear need then for the nlo calculation , both in order to improve the results in the @xmath76 sector and to test the stability of the enhancement of @xmath273 observed at lo .
the nlo calculation reveals that the lo result for @xmath273 is quite robust , being modified by the nlo corrections by less than 1% .
this stability is , however , non - trivial . indeed ,
as already noted , at nlo the mixing angles are substantially affected the mixing angles @xmath276 and @xmath234 are reduced by 10 to 30% ( a more dramatic reduction of roughly 50% results for the mixing angle @xmath75 ) . the @xmath20 width ,
however , is only slightly affected because the effects that ultimately determine the corrections to the amplitude are in the decay constant matrix @xmath239 shown in eq .
this matrix is affected by the mixing , and also receives nlo corrections that reside in @xmath175 , and apparently the nlo modifications to the mixings are partially compensated by the latter corrections in the case of the entries relevant to the @xmath20 .
the primary source of theoretical uncertainty in the present calculation of @xmath273 resides in the value of @xmath205 , which has an uncertainty of about 15% . using the empirical formula resulting from the results in table ii , @xmath277 , the uncertainty in @xmath205 translates into an error in the @xmath20 width of @xmath278 .
other sources of uncertainty are the nnlo corrections , of which the chiral - logs are an example , and which should be expected to be in the range of @xmath227 , and also em corrections which according to a straightforward estimate should be in a similar range .
all this together implies an uncertainty of less than @xmath224 in the theoretical prediction of @xmath273 .
note that there is an overall uncertainty in @xmath273 due to the 0.3% error in @xmath206 .
the fact that the theoretical prediction for @xmath273 shows the 4.5% enhancement that can be experimentally observed , and the fact that the experimental result with the smallest quoted error @xcite lies more than three standard deviations below that prediction lend great significance to the upcoming primex measurement .
it is evident from the above analysis that no predictions for the @xmath7 and @xmath4 two - photon widths can be made .
rather , these quantities are inputs , and their precise values do not affect in any dramatic way the @xmath20 width . in a more complete study , wherein the analysis is extended to additional processes such as @xmath279 , a more precise experimental knowledge of such widths would be necessary .
a more extensive analysis would also illuminate the @xmath1 corrections encoded in the lecs @xmath131 and @xmath132 and help determine them more precisely .
as this manuscript was being completed an analysis by ananthanarayan and moussallam @xcite was posted where the electromagnetic corrections are studied in detail .
their result for the @xmath20 rate is in agreement with the results in this work .
related work is also being completed by kaiser and leutwyler @xcite .
we thank j. f. donoghue for his collaboration during the early stages of this work and for useful discussions .
one of us ( jlg ) would like to especially thank h. leutwyler for enlightening discussions , as well as a. gasparian , j. gasser , p. minkowski and a. nyffeler , and the institut fr theoretische physik of the university of bern for the kind hospitality and support while part of this work was completed .
this work was partially supported by the national science foundation through grants # phy-9733343 ( jlg ) and # phy-9801875 ( brh ) , by the department of energy through contracts de - fc02 - 94er40818 ( amb ) and de - ac05 - 84er40150 ( jlg ) , and by a sabbatical leave support from the southeastern universities research association ( jlg ) .
e.g. _ , j. f. donoghue , e. golowich , and b. r. holstein , dynamics of the standard model , cambridge univ .
press , new york ( 1992 ) a. gasparian et .
al . , conceptual design report for primex , url : www.jlab.org / primex .
particle data group ( c. caso _ et al .
_ ) , eur . phys .
j. * c15 * , ( 2000 ) 1 .
h. w. atherton et al .
lett . * b158 * ( 1985 ) 81 .
d. williams et al . , phys .
* d38 * ( 1988 ) 1365 .
g. bellettini et al , nuovo cim .
* 66a * ( 1972 ) 20 . v. i. kryshkin , jetp * 30 * ( 1970 ) 1037 .
a. browman et al , phys . rev .
33 * ( 1974 ) 1400 .
j. bijnens , a. bramon and f. cornet phys .
* 61 * ( 1988 ) 1453 .
j. f. donoghue , b. r. holstein and y .- c .
r. lin , phys .
55 * , 2766 ( 1985 ) .
r. akhoury and a. alfakih , ann .
phys . * 210 * ( 1991 ) 81 .
b. moussallam , phys .
* d51 * ( 1995 ) 4939 .
herrera - sikldy , j. i. latorre , p. pascual and j. taron , nucl .
* b497 * ( 1997 ) 345 , and phys .
* b419 * ( 1998 ) 326 .
r. kaiser and h. leutwyler , in non - perturbative methods in quantum , field theory , a. schreiber , a. g. williams and a. w. thomas , editors , world scientific ( singapore 1998 ) , hep - ph/9806336 .
r. kaiser and h. leutwyler , eur .
j. * c17 * ( 2000 ) 623 .
j. gasser and h. leutwyler , nucl .
* b250 * ( 1985 ) 465 . h. georgi , phys . rev .
* d49 * ( 1994 ) 1666 .
j. kodaira , nucl .
* b165 * ( 1980 ) 129 .
+ d. espri and r. tarrach , z. phys . *
c16 * ( 1982 ) 77 . m. knecht and a. nyffeler , eur .
j. * c21 * ( 2001 ) 659 .
riazuddin and fayyazuddin , phys .
* d 37 * ( 1988 ) 149 . j. f. donoghue , b. r. holstein , and d. wyler , phys .
* d47 * , ( 1993 ) 2089 .
j. bijnens , phys .
* b306 * , ( 1993 ) 343 .
j. bijnens and j. prades , nucl .
* b490 * , ( 1997 ) 239 .
+ b. moussallam , nucl .
phys . * b504 * , ( 1997 ) 381
. h. osborn and d. j. wallace , nucl .
* b20 * , ( 1970 ) 23 .
cronin , phys .
rev . * 161 * , ( 1967 ) 1483 .
j. kambor et al . , nucl .
* b465 * , ( 1996 ) 215 .
a. anisovich and h. leutwyler , phys .
b375 * , ( 1996 ) 335 .
r. urech , nucl .
* b433 * ( 1995 ) 234 . + r. baur and r. urech ,
* b499 * ( 1997 ) 319 .
feldmann and p. kroll , eur .
j. * c5 * ( 1998 ) 327 .
feldmann , p.
kroll and b. stech , phys .
* d58 * ( 1998 ) 114006 . th .
feldmann , p. kroll and b. stech , phys .
* b449 * ( 1999 ) 339 .
feldmann , int .
j. mod .
* a15 * ( 2000 ) 159 .
feldmann and p. kroll , hep - ph/0201044 .
+ r. escribano and j. m. frre , phys .
* b459 * ( 1999 ) 288 .
+ f. de fazio and m. r. pennington , jhep * 07 * ( 2000 ) 051 .
+ a. bramon , r. escribano and m. d. scadron , phys .
* b503 * ( 2001 ) 271 . h. leutwyler , hep - ph/0011049 and references therein .
b. ananthanarayan and b. moussallam , hep - ph/02052702 h. leutwyler , private communication . | the @xmath0 decay width is analyzed within the combined framework of chiral perturbation theory and the @xmath1 expansion up to @xmath2 and @xmath3 in the decay amplitude . the @xmath4 is explicitly included in the analysis .
it is found that the decay width is enhanced by about 4.5% due to the isospin - breaking induced mixing of the pure @xmath5 states .
this effect , which is of leading order in the low energy expansion , is shown to persist nearly unchanged at next to leading order .
the chief prediction with its estimated uncertainty is @xmath6 .
this prediction at the 1% level makes the upcomming precision measurement of the decay width even more urgent .
observations on the @xmath7 and @xmath4 can also be made , especially about their mixing , which is shown to be significantly affected by next to leading order corrections . |
we show that the cohomology of ( bi)algebras always carries a lie bracket ( which we call the _ intrinsic bracket _ ) , induced by an @xmath0-structure on the corresponding cochain complex .
we also discuss the master equation related to this @xmath0-structure . by a _
( bi)algebra _ we mean an algebra over a certain @xmath4-linear @xmath2 .
therefore a ( bi)algebra is given by a homomorphism of @xmath5 , where @xmath6 denotes the endomorphism of a @xmath4-vector space @xmath7 .
observe that this notion encompasses not only `` classical '' algebras ( associative , commutative associative , lie , & c. ) but also various types of bialgebras ( @xmath3-bialgebras , lie bialgebras , infinitesimal bialgebras , & c. ) .
let us recall that a _ minimal model _ of a @xmath4-linear @xmath2 is a differential ( non - negatively ) graded -linear prop @xmath8 together with a homology isomorphism @xmath9 such that ( i ) the @xmath10 is free and ( ii ) the image of the degree @xmath11 differential @xmath12 consists of decomposable elements of @xmath10 ( the minimality condition ) , see @xcite for details .
it is not our aim to discuss in this paper the existence and uniqueness of minimal models , nor the methods how to construct such models explicitly .
let us say only that for a large class of operads and these minimal models can be constructed using the koszul duality @xcite .
let us emphasize that instead of a minimal model of @xmath2 we may use in the following constructions any cofibrant ( in a suitable sense ) resolution of @xmath2 .
but since explicit minimal models of @xmath2 exist in all cases of interest we will stick to minimal models in this note .
this will simplify some technicalities .
assume we are given a homomorphism @xmath5 describing a @xmath2-algebra @xmath13 .
to define its cohomology , we need to choose first a minimal model @xmath14 of @xmath2 .
the composition @xmath15 makes @xmath6 an @xmath16-module ( in the sense of @xcite ) , one may therefore consider the graded vector space of derivations @xmath17 . for @xmath18 define @xmath19 .
it follows from the obvious fact that @xmath20 , implied by the triviality of the differential in @xmath2 , that @xmath21 is again a derivation , so @xmath22 is a well - defined endomorphism of @xmath17 which clearly satisfies @xmath23 .
we conclude that @xmath17 is a non - positively graded vector space equipped with a differential of degree @xmath11 .
finally , let @xmath24 be the suspension of the graded vector space @xmath17 with reversed degrees .
the differential @xmath22 induces on @xmath25 a degree @xmath26 differential denoted by @xmath27 .
the _ cohomology of @xmath13 with coefficients in itself _
is then defined by @xmath28 see @xcite . for `` classical '' algebras
, the cochain complex @xmath29 agrees with the `` standard '' constructions .
thus , for associative algebras , ( [ tel - aviv ] ) gives the hochschild cohomology , for associative commutative algebras the harrison cohomology , for lie algebras the chevalley - eilenberg cohomology , & c. more generally , for algebras over a quadratic koszul operad the above cohomology coincides with the triple cohomology .
therefore nothing dramatically new happens here .
this situation changes if we consider ( bi)algebras over a general @xmath2 . to our best knowledge ,
( [ tel - aviv ] ) is the only definition of a cohomology of ( bi)algebras over . as we argued in @xcite
, it governs deformations of these ( bi)algebras .
let @xmath30 be the minimal model of the @xmath31 for @xmath3-bialgebras constructed in @xcite and @xmath13 a bialgebra given by a homomorphism @xmath32
. then @xmath33 is isomorphic to the gerstenhaber - schack cochain complex @xmath34 and ( [ tel - aviv ] ) coincides with the gerstenhaber - schack cohomology @xcite , @xmath35 see section [ gs ] for details .
the result announced in the abstract follows from the following : [ main ] let @xmath13 be a ( bi)algebra over a @xmath2 .
then there exist a graded lie algebra bracket on the cohomology @xmath36 induced by a natural @xmath0-structure @xmath37 on the defining complex @xmath29 .
we will see in section [ 1bis ] and also in section [ 3 ] that the @xmath0-structure of theorem [ main ] can be given by explicit formulas that involve the differential @xmath12 of the minimal model @xmath16 .
it will also be clear that this @xmath0-structure uses all the information about the minimal model @xmath16 of @xmath2 and that , vice versa , the minimal model @xmath16 can be reconstructed from the knowledge of this @xmath0-structure .
therefore the brackets @xmath38 can be understood as massey products that detect the homotopy type of the @xmath2 .
since the minimal model @xmath16 is , by definition , free on a @xmath39-bimodule @xmath40 , @xmath41 , it is graded by the number of generators .
this means that @xmath42 , where @xmath43 is spanned by `` monomials '' composed of exactly @xmath44 elements of @xmath40 .
the minimality of @xmath12 is equivalent to @xmath45 .
the differential @xmath12 is called _ quadratic _ if @xmath46 .
[ main1 ] if the differential of the minimal model @xmath16 of @xmath2 used in the definition of the cohomology ( [ tel - aviv ] ) is quadratic , then the higher brackets @xmath47 of the @xmath0-structure vanish , therefore @xmath29 forms an ordinary dg lie algebra with the bracket @xmath48:= l_2(-,-)$ ] . the minimal model of a quadratic koszul operad @xmath1
is given by the cobar construction on its quadratic dual @xmath49 and is therefore quadratic .
thus , for algebras over such an operad , the complex @xmath50 is a lie algebra whose bracket coincides with the classical intrinsic bracket given by identifying this complex with the space of coderivations of a certain cofree nilpotent @xmath49-coalgebra , see ( * ? ? ?
* section ii.3.8 ) .
the similar observation is true also for various types of `` bialgebras '' defined over quadratic ( in a suitable sense ) , such as lie bialgebras @xcite , infinitesimal bialgebras @xcite and @xmath51bialgebras @xcite .
in contrast , as we will see in section [ gs ] , the gerstenhaber - schack complex of an @xmath3-bialgebra carries a fully fledged @xmath0-algebra structure .
.3em * relation to previous results . * as indicated in the above paragraph , it is well - known that , for an algebra @xmath13 over a quadratic koszul operad @xmath1 , the cochain complex @xmath50 is a dg - lie algebra with the structure given by a generalization of schlessinger - stasheff s intrinsic bracket @xcite . for algebras over a general operad ,
an @xmath0-generalization of this structure was obtained by van der laan @xcite as follows .
van der laan noticed that , for each homotopy cooperad ( in an appropriate sense ) @xmath40 and for each operad @xmath52 , the @xmath39-module @xmath53 , where @xmath54 , is a homotopy operad ( again in an appropriate sense ) , which generalizes the convolution operad of @xcite .
he also proved that , for each homotopy operad @xmath55 , the `` total space '' @xmath56 has an induced @xmath0-structure which descents to an @xmath0-structure on the symmetrization @xmath57 .
therefore , for @xmath52 and @xmath40 as above , the graded vector space @xmath58 is a natural @xmath0-algebra . on the other hand ,
let @xmath59 be a minimal model of @xmath1 .
van der laan observed that the differential @xmath12 makes the @xmath39-module of generators @xmath40 a homotopy cooperad and that , for @xmath60 , @xmath61 combining the above facts , he concluded that @xmath25 is a natural @xmath0-algebra and proved that the map @xmath62 defining the @xmath1-algebra @xmath13 determines a maurer - cartan element @xmath63 .
he then constructed the @xmath0-structure on @xmath50 as the @xmath64-twisting , in the sense recalled in section [ sh ] , of the @xmath0-algebra given by the identification ( [ zavody_ve_dvore_zacaly_a_porad_prsi ] ) .
we were recently informed about an on - going work @xcite whose central statement proves the existence of an @xmath0-structure on the space of @xmath65-graded extended morphisms from a free dg to an arbitrary from which laan s arguments and their generalization to follow . the methods of this article are independent of the above mentioned papers . while our approach is not very conceptual , it is straightforward and immediately produces , from a given differential in the minimal model , explicit formulas for the induced @xmath0-structure . .3em * relation to derived spaces of algebra structures .
* as argued in @xcite , for an operad @xmath1 and a _ finite - dimensional _ vector space @xmath66 , there exists a scheme @xmath67 parameterizing @xmath1-algebra structures on @xmath66 .
it is characterized by the property that for each commutative dg - algebra @xmath68 , morphisms @xmath69 are in bijection with @xmath70-algebra structures on the dg-@xmath68-module @xmath71 .
let @xmath72 be a free resolution of the operad @xmath1 .
then @xmath73 was interpreted , in ( * ? ? ?
* section 3.2 ) , as a smooth dg - scheme in the sense of @xcite , representing a right - derived space @xmath74 of @xmath1-actions on @xmath66 in a suitable derived category of dg - schemes .
it can be easily seen , using methods of ( * ? ? ?
* section 3.5 ) , that if @xmath75 is the minimal model of the operad @xmath1 and @xmath76 describes a @xmath1-algebra @xmath13 with the underlying vector space @xmath66 , then the components of the dg - tangent space at @xmath77 \in \alg { \calf}w$ ] , @xmath78 , can be described as @xmath79 } \alg { \calf}w \cong \ccp^{n+1}(v;v),\ \mbox { for } n \geq 0\ ] ] ( we used a different degree convention than @xcite ) .
the existence of an @xmath0-structure on @xmath80 would then follow from general properties of dg - schemes and is in fact equivalent to specifying a local coordinate system at the smooth point @xmath77 $ ] of @xmath73 ( * ? ? ?
* proposition 2.5.8 ) .
on the other hand , let @xmath16 be a free dg- and @xmath81 a homomorphism .
denote by @xmath82 the @xmath83 considered as an @xmath16-module with the action induced by the homomorphism @xmath84 .
we will prove in theorem [ e ] of section [ 2bis ] that the desuspended space @xmath85 of derivations has a natural @xmath0-structure . by the definition ( [ joshua_v_praze ] ) of @xmath86 ,
theorem [ main ] follows from theorem [ e ] by taking @xmath16 a minimal model of the @xmath2 , @xmath87 and @xmath88 the composition @xmath89 , where @xmath5 describes the algebra @xmath13 and @xmath90 is the map of the minimal model . in the light of (
* proposition 2.5.8 ) , theorem [ e ] translates to the statement that for each free dg- @xmath16 and each homomorphism @xmath81 , the space @xmath91 forms a smooth dg - scheme .
the derived scheme @xmath74 of @xcite is then , for @xmath2 the operad @xmath1 and @xmath66 a finite - dimensional vector space , the specialization of this construction at the point represented by @xmath92 and @xmath93 .
it this sense , the results of the present paper are meta - versions of constructions in ( * ? ? ?
* section 3.2 ) that completely avoid all assumptions required by the ` classical ' geometry , namely the fact that the target of the map @xmath84 is the endomorphism of a finite - dimensional vector space .
the present paper thus finishes the program to find a `` universal variety of structure constants '' formulated in @xcite . .3em
* the master equation .
* let @xmath68 be a `` classical '' algebra over a quadratic koszul operad @xmath1 ( associative , commutative associative , lie , & c. ) , so that the cochain complex @xmath94 is a graded dg - lie algebra ( proposition [ main1 ] ) .
one usually shows that an element @xmath95 , represented by a bilinear map ( or by a collection of bilinear maps ) , is a deformation of the @xmath1-algebra structure @xmath68 if and only if it solves the `` classical '' master equation @xmath96 $ ] . for a ( bi)algebra @xmath13 over a general @xmath2 , the cochain complex @xmath97 forms only an @xmath0-algebra , but we will prove , in section [ sh ] , that solutions @xmath98 of the `` quantum '' master equation @xmath99 are deformations of @xmath13 .
this means that the @xmath0-structure of theorem [ main ] represents an @xmath0-version of the deligne groupoid for deformations of @xmath13 , see @xcite for the terminology .
we will see in section [ 3 ] how this observation applies to @xmath3-bialgebras .
although the sum ( [ master1 ] ) is infinite , we will see that , in situations considered in this paper , it converges . .3em
* acknowledgment : * the author would like to express his thanks to the mathematics department of bar ilan university for a very stimulating and pleasant atmosphere .
it was during his visit of this department , sponsored by the israel academy of sciences , that the present work started .
also the suggestions and comments of ezra getzler , jim stasheff , bruno vallette and sasha voronov regarding the first draft of this paper were extremely helpful .
proposition [ podzim ] and most of section [ sh ] is basically only a generalization of the material on pages 21 - 23 of @xcite .
my thanks are also due to s. merkulov and b. vallette for sharing their on - going work @xcite with me .
.3em * outline of the paper .
* in the following section we indicate the idea behind the @xmath0-structure of theorem [ main ] . a rigorous proof of this theorem is then contained in sections [ 2 ] and [ 2bis ] . in short section [ sh ]
we discuss master equations in @xmath0-algebras . in the last section
how constructions of this paper together with the description ( * ? ? ?
3.1 ) of the minimal model of the bialgebra give an explicit @xmath0-structure on the gerstenhaber - schack complex @xmath34 .
in this section we explain the idea behind the @xmath0-structure of theorem [ main ] and indicate why the @xmath0-axioms are satisfied .
we believe that this section will help to understand the concepts , but we do not aim to be rigorous here , see also the remark at the end of this section .
formal constructions and proofs based on the equivalence between symmetric brace algebras and pre - lie algebras @xcite are then given in sections [ 2 ] and [ 2bis ] .
we need to review first some definitions and facts concerning and their derivations . given a @xmath2 and a @xmath2-module @xmath100 @xcite , then a _ degree @xmath101 derivation _
@xmath102 is a map of @xmath39-bimodules @xmath103 which is a degree @xmath101 derivation ( in the evident sense ) with respect to both the horizontal and vertical compositions in the @xmath2 and the @xmath2-module @xmath100 .
an equivalent definition is the following . for each @xmath101 , the @xmath2-module structure on @xmath100
induces the obvious structure on the direct sum @xmath104 of the @xmath39-bimodule @xmath2 and the @xmath101-fold desuspension of the @xmath39-bimodule @xmath100 .
a degree @xmath101 map @xmath105 of @xmath39-bimodules is then a degree @xmath101 derivation if and only if @xmath106 is a homomorphism .
the equivalence of the above two definitions of derivations can be easily verified directly .
we denote by @xmath107 the graded vector space of derivations @xmath108 . if @xmath109 , we write simply @xmath110 instead of @xmath111 .
[ jita ] let @xmath41 be the free generated by a @xmath39-bimodule @xmath40 and @xmath100 an @xmath16-module .
then there is a canonical isomorphism @xmath112 given by restricting a derivation @xmath113 onto the space @xmath114 of generators . in ( [ jitka2 ] )
, @xmath115 denotes the space of linear bi - equivariant maps of @xmath39-bimodules .
the _ proof _ follows from the interpretation ( [ jitka3 ] ) of derivations as homomorphisms and the standard universal property of free .
let us look more closely at the structure of the free @xmath116 generated by a @xmath39-bimodule @xmath40 .
as explained in @xcite , the components of this are the colimit @xmath117 taken over the category @xmath118 of directed @xmath119-graphs without directed cycles and their isomorphisms . in ( [ jsem_zvedav_jestli_se_ta_druha_jitka_ozve ] )
, @xmath120 denotes the vector space of all decorations of vertices of @xmath121 by elements of @xmath40 , see ( * ? ? ?
* section 8) for precise definitions .
therefore elements of the free @xmath116 can be represented by sums of @xmath40-decorated directed graphs . to simplify the exposition
, we accept the convention that @xmath122 ( with or without a subscript ) will denote an @xmath40-decorated graph , and @xmath121 ( with or without a subscript ) the underlying un - decorated graph .
if @xmath122 is such an @xmath40-decorated graph , we denote by @xmath123 the corresponding decoration of a vertex @xmath124 of the underlying un - decorated graph . for bi - equivariant linear maps @xmath125 , homomorphism @xmath126 and _
distinct _ vertices @xmath127 , we denote by @xmath128 \in \end_v(g)\ ] ] the @xmath6-decorated graph whose vertices @xmath129 , @xmath130 , are decorated by @xmath131 and the remaining vertices @xmath132 by @xmath133 .
see figure [ figu1 ] .
( 170.00,170.00)(0.00,0.00 ) ( 90.00,10.00)(0.00,0.00)@xmath134 ( 90.00,160.00)(0.00,0.00)@xmath134 ( 130.00,80.00)(0.00,0.00)@xmath135 ( 70.00,40.00)(0.00,0.00)@xmath135 ( 50.00,70.00)(0.00,0.00)@xmath135 ( 130.00,130.00)(0.00,0.00)@xmath136 ( 100.00,60.00)(0.00,0.00)@xmath137 ( 73.00,100.00)(0.00,0.00)@xmath138 ( 50.00,120.00)(0.00,0.00)@xmath139 ( 140.00,150.00)(0,1)20.00 ( 60.00,150.00)(0,1)20.00 ( 50.00,150.00)(0,1)20.00 ( 140.00,0.00)(0,1)20.00 ( 60.00,0.00)(0,1)20.00 ( 50.00,0.00)(0,1)20.00 ( 95.00,85.00)(130.00,130.00 ) the structure of @xmath6 determines the contraction @xmath140 along @xmath121 ( * ? ? ? * section 8) . applying this contraction to ( [ stanu_se_ul_instruktorem ? ] ) , we obtain a linear map @xmath141 ) \in \end_v(m , n ) = \lin(\otexp vn,\otexp vm).\ ] ] let us show , after these preliminaries , how the @xmath0-braces of theorem [ main ] can be constructed .
assume that , as in the introduction , @xmath5 is a @xmath2-algebra and @xmath14 a minimal model of @xmath2 .
recall that @xmath84 denotes the composition @xmath142 .
assume that @xmath41 for some @xmath39-bimodule @xmath40 .
it follows from definition ( [ joshua_v_praze ] ) and isomorphism ( [ jitka2 ] ) that @xmath143 for @xmath144 , represent the value @xmath145 of the differential as a sum of @xmath40-decorated @xmath119-graphs , @xmath146 with a finite set of summation indices @xmath147 .
let @xmath148 correspond , under isomorphism ( [ piji_kavu ] ) , to a cochain @xmath149 , @xmath130 , and define @xmath150 by @xmath151),\ ] ] where @xmath152 runs over all @xmath44-tuples of distinct vertices of the underlying graph @xmath153 of the @xmath40-decorated graph @xmath154 .
the overall sign in the right hand side , defined later in ( [ dnes_jsem_byl_na_sipcich ] ) , plays no role in this section .
the linear map @xmath155 determines , by ( [ piji_kavu ] ) , an element @xmath156 , which is precisely the @xmath44-th @xmath0-bracket of theorem [ main ] . observe that ( [ nabla ] ) makes sense also for @xmath157 when it reduces to @xmath158 where @xmath159 is the @xmath40-decorated graph whose underlying graph is @xmath153 and all vertices @xmath160 are decorated by @xmath161 .
this clearly means that @xmath162 , therefore @xmath163 . since @xmath20
, this implies that @xmath164 .
it is equally simple to verify that @xmath165 coincides with the differential @xmath166 in @xmath25 .
let us explain why formula ( [ nabla ] ) indeed defines an @xmath0-structure .
it is not difficult to see that @xmath167 , @xmath168 , have the appropriate symmetry . to understand why the @xmath0-axiom recalled in ( [ jestli_jsem_se_nezblaznil ] ) below is satisfied , expand the equation @xmath169 where @xmath170 is the @xmath40-decorated graph obtained as follows . for @xmath171 ,
let @xmath172 where @xmath173 are @xmath40-decorated graphs indexed by a finite set @xmath174 .
the graph @xmath170 is then given by replacing the @xmath40-decorated vertex @xmath160 of @xmath154 by the @xmath40-decorated graph @xmath175 . by ( [ 21 ] ) ,
@xmath176)\ ] ] for arbitrary @xmath125 and @xmath168 .
this summation can be further refined as @xmath177),\ ] ] where @xmath178 runs over all @xmath179-unshuffles with @xmath180 and the rightmost summation is restricted to @xmath44-tuples of distinct vertices @xmath181 such that @xmath182 are vertices of the subgraph @xmath183 and @xmath184 are vertices of the complement of @xmath185 in @xmath186 .
it is obvious that for such @xmath178 and @xmath152 , the graph @xmath187 is obtained from the @xmath6-decorated graph @xmath188 by replacing the vertex @xmath160 by the @xmath6-decorated graph @xmath189 , see figure [ zatim_pocitac_drandi ] .
= 1.000000pt ( 210.00,290.00)(-10,0 ) ( -20,150)(0.00,0.00)[r]@xmath190 ( 107,124)(0.00,0.00)@xmath191 ( 60.00,140.00)(0.00,0.00)@xmath192 ( 195.00,260.00 ) ( 0.00,0.00)[lb ] @xmath193 ( 90.00,100.00)(0.00,0.00)@xmath134 ( 120.00,90.00)(0,1)20.00 ( 70.00,90.00)(0,1)20.00 ( 60.00,90.00)(0,1)20.00 ( 120.00,210.00)(0,1)20.00 ( 90.00,160.00)(100.00,100.00 ) ( 108,182)(0.00,0.00)@xmath191 ( 20.00,230.00)(0.00,0.00)@xmath191 ( 80.00,70.00)(0.00,0.00)@xmath191 ( 170.00,110.00)(0.00,0.00)@xmath191 ( 30.00,80.00)(0.00,0.00)@xmath194 ( 160.00,70.00)(0.00,0.00)@xmath195 ( 142,204)(0.00,0.00)[lb]@xmath196 ( 160.00,170.00)(0.00,0.00)@xmath197 ( 104,150.00)(0.00,0.00)@xmath198 ( 60.00,190.00)(0.00,0.00)@xmath199 ( 80.00,220.00)(0.00,0.00)@xmath134 ( 90.00,20.00)(0.00,0.00)@xmath134 ( 90.00,270.00)(0.00,0.00)@xmath134 ( 70.00,210.00)(0,1)20.00 ( 60.00,210.00)(0,1)20.00 ( 180.00,0.00)(0,1)40.00 ( 50.00,0.00)(0,1)40.00 ( 30.00,0.00)(0,1)40.00 ( 180.00,250.00)(0,1)40.00 ( 50.00,250.00)(0,1)40.00 ( 30.00,250.00)(0,1)40.00 ( 105.00,145.00)(210.00,210.00 ) therefore one can reinterpret the right hand side of ( [ uz_mne_z_ty_kvety_hrabe ] ) as @xmath200 with @xmath201 , which is the axiom of @xmath0-algebras .
we are sure that the reader has already realized at which points we were not precise .
first , we did not say what is a decoration of a graph .
second , our formulas ( [ porad_odkladam_ty_testy ] ) and ( [ 23 ] ) for the differential assumed choices of representatives of decorated graphs , and a rigorous proof of ( [ jestli_jsem_se_nezblaznil ] ) would require assumptions about the compatibility of these choices .
we also ignored signs .
namely the compatibility assumption would make a rigorous version of the above arguments very complicated .
in theorem [ ji ] of this section we prove that , for a free @xmath41 and for an arbitrary @xmath83 , the space @xmath202 of derivations of @xmath16 with values in the coproduct @xmath203 admits a natural pre - lie algebra structure .
observe that if @xmath83 is the trivial , then @xmath204 and theorem [ ji ] is an analog of the classical theorem about the existence of a pre - lie structure on the space of ( co)derivations of a ( co)free algebra ( * ? ? ?
* section ii.3.9 )
. we will also study how this pre - lie structure behaves with respect to some natural maps induced by a homomorphism @xmath205 ( lemma [ nahravam_bacha ] ) .
recall that if @xmath206 and @xmath207 are represented as quotients of free , @xmath208 , @xmath209 , their coproduct @xmath210 is the quotient @xmath211 , where @xmath212 denotes the ideal generated by @xmath213 .
the following technical proposition will be useful in the sequel .
[ jitb ] given @xmath206 , @xmath207 and a @xmath214-module @xmath100 , there is a canonical isomorphism @xmath215 which sends @xmath216 into the direct sum @xmath217 of restrictions . in the right hand side of ( [ jitka1 ] ) , the @xmath218-module structure on @xmath100
is induced from the @xmath210-structure by the inclusion @xmath219 , @xmath220 .
the _ proof _ follows from the representation ( [ jitka3 ] ) of derivations as homomorphisms and the universal property of coproducts .
the last thing we need to observe before coming to the main point of this section is that , given a homomorphism @xmath221 of @xmath2-modules and a derivation @xmath222 , the composition @xmath223 of @xmath39-bimodule maps is a derivation in @xmath224 .
the correspondence @xmath225 therefore induces the ` standard ' map @xmath226 suppose that @xmath16 and @xmath83 are .
the central object of this section will be the graded vector space @xmath227 , where the coproduct @xmath228 is considered as an @xmath16-module with the structure induced by the canonical inclusion @xmath229 .
we need to introduce , for the purposes of the next section , three maps @xmath68 , @xmath13 and @xmath230 that relate @xmath227 with other spaces of derivations .
suppose that @xmath83 is equipped with a homomorphism @xmath231 .
it then makes sense to consider @xmath232 , where @xmath82 denotes @xmath83 with the @xmath16-module structure given by the homomorphism @xmath84 .
the map @xmath84 also induces a homomorphism @xmath233 by @xmath234 and @xmath235 . by the definition of @xmath82
, @xmath236 can be considered as a map of @xmath16-modules which in turn induces the standard map ( [ to_jsem_zvedav_na_jitku ] ) @xmath237 similarly , the canonical inclusion @xmath238 induces an inclusion of vector spaces @xmath239 from this moment on , we suppose that @xmath16 is the _
free @xmath41 generated by a @xmath39-bimodule @xmath40_. let @xmath240 be the canonical inclusion and denote by @xmath68 the composition @xmath241 with the isomorphisms given by proposition [ jita ] .
the three maps introduced above can be organized into the diagram @xmath242 ' '' '' we will use the inclusion @xmath13 to identify @xmath243 with a subspace of @xmath227 .
the following simple lemma will be useful .
[ pred_odjezdem_do_srni ] the linear maps defined in ( [ a])([c ] ) above satisfy @xmath244 by definition , for @xmath245 , @xmath246 , because @xmath247 by the definition of @xmath236 .
since each derivation in @xmath232 is , by proposition [ jita ] , determined by its restriction to the space of generators , this proves ( [ jedna ] ) .
similarly , for @xmath248 , @xmath249 , again by the definition of @xmath236 , which proves ( [ dve ] ) . before we formulate the next statement , we observe that proposition [ jitb ] implies @xmath250 in words , each derivation @xmath251 can be uniquely extended into a derivation @xmath252 characterized by @xmath253 .
let @xmath254 be two derivations .
then the commutator of the composition @xmath255 : = \w \phi \circ \psi - ( -1)^{|\phi||\psi| } \cdot \w \psi \circ\phi\ ] ] is again a derivation , and the assignment @xmath256 $ ] makes @xmath227 a graded lie algebra . it is straightforward to check that @xmath257 $ ] defined in ( [ jitka_zvala ] ) has the derivation property with respect to both the vertical and horizontal compositions .
the rest of the lemma is obvious .
let us look more closely at the isomorphism @xmath258 that follows from proposition [ jita ] .
it sends @xmath251 into the restriction @xmath259 . in the opposite direction , to each @xmath260 there exists a unique extension @xmath261 characterized by .
[ ji ] the graded vector space @xmath262 is a natural graded pre - lie algebra , with the structure operation @xmath263 given by @xmath264 where @xmath265 is the extension of the composition @xmath266 _ proof _ of theorem [ ji ] is straightforward , but since this theorem is a central technical tool of this section , we give it here . for the ease of reading ,
we omit in this proof the @xmath267 denoting the extension of derivations of @xmath227 into derivations of @xmath268 . by definition (
* section 2 ) , @xmath269 is a pre - lie product if the associator @xmath270 is ( graded ) symmetric in @xmath271 and @xmath272 . by ( [ jitka ! ] ) , this associator can be written as @xmath273 where @xmath274 since @xmath275 is a derivation belonging to @xmath227 , it is determined by is restriction to @xmath40 . by ( [ jit ] ) , clearly @xmath276 the antisymmetry of @xmath275 in @xmath271 and @xmath272 is then equivalent to the antisymmetry of the restrictions to @xmath40 , @xmath277 which is , by ( [ jitula ] ) , the same as @xmath278 where we , of course , omitted the overall factor @xmath279 . using the bracket ( [ jitka_zvala ] ) and moving @xmath280 to the right ,
the above display can be rewritten as @xmath281 - \e(\psi \circ \phi|_e ) + ( -1)^{|\phi||\psi|}\e(\phi \circ \psi|_e ) \right\ } \circ\theta|_e = 0\ ] ] so it is enough to prove that @xmath282 - \e(\psi \circ \phi|_e)+ ( -1)^{|\phi||\psi|}\e(\phi \circ \psi|_e ) = 0.\ ] ] since the left hand side is an element of @xmath227 , it suffices to prove that it vanishes when restricted to generators , that is @xmath282|_e - \psi \circ \phi|_e + ( -1)^{|\phi||\psi| } \phi \circ \psi|_e = 0,\ ] ] which immediately follows from the definition ( [ jitka_zvala ] ) of the bracket .
the last statement in this section relates the @xmath269-product of theorem [ ji ] with the maps @xmath68 and @xmath13 .
[ nahravam_bacha ] let @xmath68 and @xmath230 be the maps defined in ( [ a ] ) and ( [ c ] ) .
then for each @xmath283 , @xmath284 by proposition [ jita ] , it suffices to prove the restrictions of the above equalities onto the space @xmath40 of generators of @xmath41 . by definition ,
@xmath285 because @xmath286 .
this proves the left equation of ( [ pisu_v_ratajich ] ) .
similarly , by definition @xmath287 let us prove that @xmath288 in @xmath289 . by proposition [ jitb ] this means to verify that @xmath290 the first equation is obvious because @xmath291 by the definition of the tilde - extension .
the second equality is established by @xmath292 where we used the definition ( [ a ] ) of the map @xmath230 and the equality @xmath293 proved in lemma [ pred_odjezdem_do_srni ] .
this finishes the proof of the right equation of ( [ pisu_v_ratajich ] ) .
in this section we prove theorem [ main ] and proposition [ main1 ] of the introduction and formulate some technical statements which will guarantee the convergence of the master equation . according to @xcite , the pre - lie algebra product @xmath269 on @xmath202
whose existence we proved in theorem [ ji ] generates unique _
symmetric braces_. this means that for each @xmath294 , @xmath295 , there exists a ` brace ' @xmath296 such that @xmath297 these braces satisfy the axioms recalled in the appendix a on page , where we also indicate how these braces are generated by @xmath269 .
the following statement generalizes lemma [ nahravam_bacha ] .
[ opravu_se_na_nem_neda_pracovat ] let @xmath68 and @xmath230 be the maps defined in ( [ a ] ) and ( [ c ] ) .
then for each @xmath298 , @xmath299 , @xmath300 and @xmath301 before we prove the lemma we notice that ( [ monteverdi ] ) for @xmath302 ( with @xmath303 ) says that @xmath304 which is , by ( [ dnes_to_zkusim_nainstalovat ] ) , the same as @xmath305 , which we recognize as the first equality in ( [ pisu_v_ratajich ] ) .
similarly , ( [ claudio ] ) for @xmath302 means that @xmath306 which is , again by ( [ dnes_to_zkusim_nainstalovat ] ) , the same as @xmath307 , the second equality in ( [ pisu_v_ratajich ] ) .
therefore lemma [ opravu_se_na_nem_neda_pracovat ] indeed generalizes lemma [ nahravam_bacha ] .
we prove ( [ monteverdi ] ) by induction .
for @xmath302 it is ( [ jitulka ] ) .
assume we have already proved ( [ monteverdi ] ) for all @xmath308 and prove it for @xmath309 . by axiom ( [ axiom ] ) of symmetric braces , @xmath310 where the sum in the right hand side runs over all unshuffles @xmath311 of the set @xmath312 .
the sign @xmath313 is not important for the purposes of this proof , because all terms in the right hand side are zero , by induction .
this establishes ( [ monteverdi ] ) for @xmath309 .
equation ( [ claudio ] ) will also be be proved by induction .
for @xmath302 it is ( [ pincakova ] ) .
suppose we have already established ( [ claudio ] ) for all @xmath308 and prove it for @xmath309 . by ( [ axiom ] ) , @xmath314 with the sum in the right hand side taken over the set ( [ jarka_pred_chvili_volala . ] ) and @xmath313 an appropriate sign .
the sum is zero by ( [ monteverdi ] ) while second term equals @xmath315 by induction . using ( [ axiom ] ) , we can write @xmath316 where the summation range and @xmath313 are the same as in ( [ hribecek ] ) .
the sum in the right hand side is zero by ( [ monteverdi ] ) , while the second term equals , by ( [ axiom ] ) , @xmath317 where @xmath318 the first term in ( [ mam_hlad ] ) equals @xmath319 by ( [ pincakova ] ) , while the second term equals , by induction , @xmath320 since , by ( [ pincakova ] ) , @xmath321 for @xmath322 , ( [ xx ] ) equals @xmath323 which is @xmath324 by induction .
we therefore established that @xmath325 by ( [ axiom ] ) , the right hand side of the above display equals @xmath326 , which establishes ( [ claudio ] ) for @xmath309 . in the following important definition , @xmath230 is the map introduced in ( [ a ] ) . recall also that we use the inclusion @xmath13 defined in ( [ b ] ) to identify @xmath243 with a subspace of @xmath202 .
[ zase_mne_jakobi_boli_hlava ] .8em for @xmath327 and @xmath328 , define the derivation @xmath329 \in \der(\sfm,\cale_\beta)$ ] by @xmath330 : = c\left(\phi \langle af_1,\ldots , af_n \rangle\right),\ ] ] where @xmath331 in the r.h.s.is the symmetric brace in @xmath332 .
the following proposition shows that @xmath333 behaves as a left module over the symmetric brace algebra @xmath334 .
[ e_napsala ] the brace @xmath335 $ ] is graded symmetric in @xmath336 .
moreover , for each @xmath337 with @xmath338 , @xmath339 = \hskip 2em } \\ & & \hskip 2em \nonumber \sum \epsilon \cdot \phi [ \phi_1 [ f_{i_1 ^ 1},\ldots , f_{i_{t_1}^1}],\ldots , \phi_n [ f_{i_1^m},\ldots , f_{i_{t_m}^m } ] , f_{i_1^{m+1}},\ldots , f_{i_{t_{m+1}}^{m+1}}],\end{aligned}\ ] ] where the sum is taken over all unshuffle decompositions @xmath340 of @xmath341 and where @xmath313 is the koszul sign of the corresponding permutation of @xmath342 .
the graded symmetry of the brace ( [ brace ] ) immediately follows from the definition .
let us prove ( [ 7 ] ) .
we have @xmath343 = c\left ( \phi\langle \rada \phi 1 m \rangle \langle\rada { af_1}{af_n}\rangle \right)\ ] ] which can be , using ( [ axiom ] ) , expanded into @xmath344 where @xmath313 and the sum is the same as in ( [ axiom ] ) . by ( [ claudio ] ) ,
this equals @xmath345 which , by definition ( [ brace ] ) of the braces , can be rewritten as @xmath346.\end{aligned}\ ] ] at this point we need to observe that , by ( [ dve ] ) , @xmath347 ( recall that we identified @xmath348 with its image @xmath349 ) .
therefore the assumption @xmath350 for @xmath351 implies that we may assume , in the sum ( [ prohlidkou_jsem_prosel ! ] ) , that all @xmath352 , because , if @xmath353 , @xmath354 since @xmath355 by ( [ jedna ] ) , the term in ( [ prohlidkou_jsem_prosel ! ] ) equals @xmath356,\ ] ] with the same summation as in ( [ 7 ] ) . by the definition ( [ brace ] ) of the braces , this is precisely the right hand side of ( [ 7 ] ) . as usual , @xmath357 ( resp .
@xmath358 ) denotes the suspension ( resp .
desuspension ) of a graded vector space @xmath66 .
we use the same symbols to denote also the corresponding maps @xmath359 and @xmath360 . in the following theorem , @xmath361 will be elements of @xmath362 and @xmath363 [ e ] let @xmath364 be a degree @xmath11 derivation such that @xmath365 and @xmath366 .
then the formula @xmath367\ ] ] defines on the suspension @xmath368 a structure of an @xmath0-algebra @xcite .
observe first that @xmath365 is , by ( [ dnes_to_zkusim_nainstalovat ] ) , equivalent to @xmath369 .
expanding @xmath370\ ] ] using ( [ 7 ] ) we obtain @xmath371,\rada { \desusp \hskip -.2em f}{\sigma(i+1)}{\sigma(n)}]\ ] ] with @xmath178 running over all @xmath372-unshuffles with @xmath373 and @xmath374 the koszul sign of the permutation @xmath375 substituting for @xmath376 and @xmath377 from ( [ zpatky_v_praze ] ) gives @xmath378 where @xmath379 is as ( [ jestli_jsem_se_nezblaznil ] ) .
we recognize ( [ dnes_mne_strasne_bolela_zada ] ) as the defining equation for @xmath0-algebras , see ( * ? ? ?
* definition 2.1 ) .
_ proof of theorem [ main ] _ easily follows from theorem [ e ] applied to the situation when @xmath380 is a minimal model @xmath381 of the @xmath2 , @xmath382 and @xmath93 as in section [ intro ] . the condition @xmath383 is implied by the minimality of the differential @xmath12 .
the following proposition compares the braces defined above with the constructions of section [ 1bis ] .
[ maminka_si_zlomila_nohu ] the @xmath0-structure of theorem [ main ] has the form ( [ nabla ] ) of section [ 1bis ] .
let , in this proof , an @xmath384-decorated graph means a graph with vertices decorated either by @xmath40 or by @xmath83 .
we use the convention that @xmath385 with a subscript will denote an @xmath384-decorated graph , and @xmath386 with the same subscript the underlying un - decorated graph . for such @xmath385 ,
let @xmath387 be the set of @xmath40-decorated vertices of @xmath385 .
we start the proof by giving an explicit formula for the operation @xmath269 introduced in theorem [ ji ] .
let @xmath388 as in ( [ jitka ! ] ) .
it follows from ( [ jsem_zvedav_jestli_se_ta_druha_jitka_ozve ] ) and the definition of the free product that , for @xmath144 , @xmath389 is the summation @xmath390 of @xmath384-decorated @xmath119-graphs over a finite indexing set @xmath391 .
the derivation property of @xmath271 implies that @xmath392,\ ] ] where the second summation runs over @xmath393 and @xmath394 $ ] denotes @xmath395 with the decoration @xmath123 of @xmath396 changed to @xmath397 .
let us prove inductively that the symmetric brace induced by @xmath269 satisfies @xmath398,\ ] ] where @xmath399 runs over distinct elements of @xmath400 , @xmath401 $ ] denotes @xmath402 with the decoration @xmath403 of @xmath129 changed to @xmath404 , and @xmath405 for @xmath406 . since , by ( [ dnes_to_zkusim_nainstalovat ] ) , @xmath407 , ( [ zatim_nevim ] ) holds for @xmath408 by ( [ kdy_se_nit_zivota_pretrhne ? ] ) . before continuing , we rewrite the right hand side of ( [ zatim_nevim ] ) into a sum of @xmath384-decorated graphs . to this end
, we introduce a notation which will be useful also later in the proof .
let , for @xmath406 , @xmath409 where @xmath410 are @xmath384-decorated graphs and @xmath411 a finite set .
for a subset @xmath412 , let @xmath413 denote the @xmath384-decorated graph obtained from @xmath402 by replacing , for each @xmath414 , the @xmath403-decorated vertex @xmath129 by the decorated graph @xmath410 .
with this notation , @xmath415 = \sum_{\rada t1n } \kr { \{\rada 1n\}},\ ] ] where @xmath416 runs over @xmath417 , @xmath406 , so we may rewrite the right hand side of ( [ zatim_nevim ] ) as @xmath418 = \hskip 10em } \\ & & \nonumber \hskip 10em \sum_{s \in r_\xi } \sum_{v_1,\ldots , v_n } \sum_{t_1,\ldots , t_n } ( -1)^{\epsilon_n } \cdot \kr{\{\rada 1n\}}. \end{aligned}\ ] ] suppose we have proved ( [ zatim_nevim ] ) for all @xmath44 , @xmath419 .
consider the equation @xmath420 which follows from axiom ( [ axiom ] ) of symmetric braces ; in the last term @xmath421 let us analyze the first term in the right hand side of ( [ zase_jsem_nachlazeny - je_to_mozne ? ] ) . by the induction assumption , @xmath422.\ ] ] with the notation above , @xmath423 = \sum_{\rada t1{n-1 } } \kr { \{1,\ldots , n-1\}},\ ] ] where @xmath416 runs over @xmath411 , @xmath424 , therefore @xmath425,\end{aligned}\ ] ] where @xmath426 runs over @xmath40-decorated vertices of @xmath427 . since clearly @xmath428 the right hand side of ( [ dnes_je_mne_zase_trochu_lip ] ) breaks into @xmath429 components , @xmath430 where @xmath431\ ] ] with the superscript @xmath432 meaning that @xmath426 runs over @xmath433 , and @xmath434,\ 1 \leq i \leq n-1,\ ] ] where @xmath435 means that @xmath426 runs over @xmath436 .
it is immediately clear that , for @xmath437 , @xmath438 so @xmath439 which is , by ( [ psano_v_bonnu ] ) , the right hand side of ( [ zatim_nevim ] ) .
it is equally clear that , for @xmath440 , @xmath441 = ( -1)^{\omega } \cdot \kr { \{1,\ldots , i-1,i+1,\ldots ,
n-1\}}^{\{v_i\}}[\phi_i\langle \phi_n \rangle],\ ] ] therefore @xmath442 \\ & = & \hskip -.2em \sum_{s \in r_\xi } \sum_{v_1,\ldots , v_{n-1 } } { \sum_{v_n}}^{(i ) } ( -1)^{\epsilon_n + \omega } \upsilon_s^{\{\rada v1{n-1}\}}[\phi_1,\ldots,\phi_i \langle \phi_n \rangle,\ldots,\phi_{n-1 } ] , \end{aligned}\ ] ] which equals @xmath443 , by induction .
we recognize this expression as one of the remaining terms in the right hand side of ( [ zase_jsem_nachlazeny - je_to_mozne ? ] ) , taken with the minus sign . assembling the above results
, we obtain ( [ zatim_nevim ] ) .
let @xmath444 be as in ( [ porad_odkladam_ty_testy ] ) and @xmath445 , for @xmath406 . by ( [ zatim_nevim ] ) , @xmath446(\xi).\ ] ] applying , as in definition [ zase_mne_jakobi_boli_hlava ] , the map @xmath230 on this identity
, we get a formula for @xmath447 $ ] which agrees , modulo signs , to the right hand side ( [ nabla ] ) .
the sign factor is induced by ( de)suspensions .
proposition [ maminka_si_zlomila_nohu ] has several important implications .
let us formulate first a corollary that implies proposition [ main1 ] ; the notation is the same as the one introduced in the paragraph preceding this proposition .
[ r ] let @xmath448 be such that @xmath449
. then @xmath450 in particular , if @xmath451 , then @xmath452 for @xmath453 .
if @xmath449 , then all graphs in ( [ porad_odkladam_ty_testy ] ) have @xmath454 vertices , so the summation in ( [ nabla ] ) is empty for @xmath455 . in this paper
we write several formulas containing infinite sums .
their convergence will be guaranteed by the following property of an -algebra @xmath456 : @xmath457 in ( [ v ] ) , @xmath458 denotes the component of @xmath459 in @xmath460 .
there are other conditions that can guarantee the convergence of our formulas , as the nilpotency ( * ? ? ?
* definition 4.2 ) , but -algebras considered in this paper may not be nilpotent .
[ conv ] the -structure of proposition [ main1 ] satisfies ( [ v ] ) . in this case
@xmath461 for each @xmath462 define a @xmath463-@xmath464-invariant subspace @xmath465 as @xmath466 since @xmath467 is finite , there are @xmath463-@xmath464-invariant subspaces @xmath468 , @xmath469 , such that @xmath470 the same reasoning as in the proof of corollary [ r ] shows that the subspaces @xmath471 satisfy ( [ v ] ) .
let @xmath472 and denote by @xmath473 the restriction of @xmath474 to the space of generators of @xmath41 , @xmath475 .
let @xmath476 be the extension of @xmath477 into a homomorphism .
[ podzim ] let @xmath478 be the -structure corresponding to the trivial @xmath2-algebra . under the above notation
, one has the following equality of elements of @xmath479 : @xmath480 since @xmath481 corresponds to the trivial @xmath2-structure , the map @xmath84 in ( [ stanu_se_ul_instruktorem ? ] ) is zero and the decorated graph @xmath482 $ ] may be nontrivial only if @xmath122 has precisely @xmath44 vertices , all decorated by @xmath477 .
therefore ( [ nabla ] ) , with @xmath483 , describes the @xmath44-th homogeneous component of the extension of @xmath477 into a homomorphism @xmath484 composed with @xmath485 .
the signs in ( [ je_to_trochu_lepsi ] ) are induced by ( de)suspensions .
in this short section we indicate how the methods of this paper generalize to cohomology of strongly homotopy algebras and how `` curved '' @xmath0-algebras naturally arise in this context . we will consider @xmath0-algebras @xmath486 with possibly nontrivial @xmath487 .
these generalized @xmath0-algebras can be defined by allowing @xmath488 for @xmath157 in ( * ? ? ?
* definition 2.1 ) ; we leave the details for the reader .
axiom ( 2 ) of @xcite for @xmath489 gives @xmath490 and for @xmath302 @xmath491 therefore @xmath165 need not be a differential if @xmath492 .
we will call such an -algebra _ curved _ and @xmath493 its _ curvature_. if @xmath164 we say that @xmath494 is _
flat_. flat -algebras are thus ordinary -algebras without the @xmath493 term .
-algebras with @xmath495 are sometimes called _
minimal_. the following statement is ( * ? ? ?
* theorem 2.6.1 ) , slightly generalized by allowing curved @xmath0-algebras .
[ zavody ] let @xmath496 be an -algebra satisfying ( [ v ] ) and @xmath497 an y element
. then @xmath498 with @xmath499 is an -algebra satisfying ( [ v ] ) whose curvature @xmath500 equals @xmath501 the proof is a direct verification , see @xcite . let us remark that there is another sign convention for -algebras used for example in @xcite related to the one introduced in @xcite and used in this paper by @xmath502 , @xmath503 . in this convention , all terms in the above sums
have the @xmath504 sign .
we will call @xmath505 the _ @xmath64-twisting _ of @xmath494 . observe that , if @xmath494 is flat and @xmath64 satisfies the master equation ( [ master1 ] ) in @xmath494
, then @xmath505 is an ordinary flat @xmath0-algebra .
proposition [ zavody ] then defines the classical twisting of an @xmath0-algebra by a maurer - cartan element , see for example ( * ? ? ?
* lemma 4.4 ) or ( * ? ? ?
* proposition 4.4 ) .
we will also need the following elementary lemma whose proof is straightforward . [
kousek_od_zirce ] suppose that , under assumptions of proposition [ zavody ] , @xmath66 is equipped with a degree @xmath26 differential @xmath101 such that all operations @xmath506 of the algebra @xmath494 are chain maps ( this , in particular , means that @xmath507 ) .
suppose moreover that @xmath497 satisfies @xmath508 then @xmath509 , where @xmath510 are as in proposition [ zavody ] , is a flat -algebra .
let @xmath2 be a @xmath4-linear .
strongly homotopy @xmath2-(bi)algebras are , by definition , algebras over the minimal model @xmath381 of @xmath2 .
let @xmath511 be the -(bi)algebra whose all structure operations are trivial and @xmath512 the flat -algebra constructed in theorem [ main ] corresponding to @xmath511 .
the minimality @xmath513 of @xmath481 follows from the minimality of @xmath381 .
assume that @xmath7 is graded , with a degree @xmath26 differential @xmath101 . slightly abusing the notation
, we will denote by @xmath101 also the induced differential on @xmath6 .
it is immediately clear that all operations @xmath514 are chain maps .
the following proposition is a version of lemma 5.11 of @xcite .
[ pisu_v_aute_ve_dvore ] there is a one - to - one correspondence between elements @xmath515 satisfying @xmath516 in @xmath481 and strongly homotopy -(bi)algebra structures on @xmath7 .
let @xmath517 be , as in proposition [ podzim ] , the ( unique ) homomorphism extending @xmath518 .
using ( [ je_to_trochu_lepsi ] ) , one easily sees that ( [ 44 ] ) is equivalent to @xmath519 which means that @xmath520 is a homomorphism of dg- defining a strongly homotopy @xmath2-algebra .
the interpretation of homotopy structures in terms of solutions of the ( generalized ) maurer - cartan equation was found by van der laan for homotopy @xmath1-algebras @xcite .
the generalization to homotopy ( bi)algebras over a given here was independently found by merkulov - vallette @xcite .
we are finally ready to analyze the structure of the deformation complex of a strongly homotopy ( bi)algebra .
let @xmath521 be as in proposition [ pisu_v_aute_ve_dvore ] .
the curvature of the @xmath64-twisting @xmath522 of @xmath481 equals @xmath523 .
assumptions of lemma [ kousek_od_zirce ] are satisfied , therefore @xmath524 is a flat -structure which induces a lie bracket on the cohomology @xmath525 of the strongly homotopy ( bi)algebra @xmath13 corresponding to @xmath64 .
let us illustrate the above analysis on - ( strongly homotopy associative ) algebras @xcite .
they are algebras over the minimal model @xmath526 of the operad @xmath3 for associative algebras .
it immediately follows from the description of @xmath526 given for instance in ( * ? ? ?
* example 4.8 ) that @xmath527 the only nontrivial operation of the flat algebra @xmath481 is the bilinear bracket @xmath528 given by @xmath529,\ ] ] where @xmath530 , @xmath531 , the subscript @xmath429 denotes the component in @xmath532 and @xmath48 $ ] is the gerstenhaber bracket of hochschild cochains @xcite .
equation ( [ 44 ] ) for @xmath533 , expanded into homogeneous components , reads @xmath534,\ n \geq 2,\ ] ] which we easily recognize as the axiom for -algebras in the form ( * ? ? ?
* section 1.4 ) .
the @xmath64-twisting @xmath505 of @xmath481 equals @xmath535 , with the curvature @xmath536 @xmath537 given by @xmath538\ ] ] and @xmath539 as in ( [ znechuceni ] ) . in the `` flattened '' algebra @xmath540 @xmath541
is the differential on the cochain complex defining the cohomology of the -algebra determined by @xmath64 with coefficients in itself ( * ? ? ?
* section 2.2 ) .
[ 3 ] in this section we show how to construct , by applying methods of section [ 1bis ] to the differential of the minimal model @xmath542 for the bialgebra @xmath31 , an explicit lie bracket on the gerstenhaber - schack cohomology of a bialgebra .
formulas for the differential @xmath543 of @xmath544 are given in ( * ? ? ?
recall that a ( @xmath3-)bialgebra @xmath13 is a vector space @xmath7 with a _ multiplication _ @xmath545 and a _ comultiplication _ ( also called a _ diagonal _ ) @xmath546 .
the multiplication is associative : @xmath547 the comultiplication is coassociative : @xmath548 and the usual compatibility relation between @xmath549 and @xmath550 is assumed : @xmath551 where @xmath552 is defined by @xmath553 for @xmath554 .
compatibility ( [ compatibility ] ) of course expresses the fact that @xmath555 where @xmath556 and the dot @xmath557 in the right hand side denotes the multiplication induced on @xmath558 by @xmath549 .
let us recall that in the definition of the gerstenhaber - schack cohomology @xcite one considers the bigraded vector space @xmath559 where @xmath560 it will be useful to introduce the _ biarity _ of a function @xmath561 as the couple @xmath562 . for each @xmath563 ,
the iterated diagonal @xmath564}:= ( \delta \ot { \iden}^{\ot ( q-2)})\circ(\delta \ot { \iden}^{\ot ( q-1 ) } ) \circ \cdots \circ \delta : v \to \otexp vq\ ] ] induces on @xmath565 the structure of a @xmath566-bimodule , by @xmath567}(u ) \cdot ( \otrada v1q),\ \mbox { and } \\ ( \otrada v1q)u & : = & ( \otrada v1q ) \cdot \delta^{[q]}(u),\end{aligned}\ ] ] where @xmath557 denotes the multiplication induced on @xmath565 by @xmath549 .
therefore it makes sense to define @xmath568 to be the hochschild differential of the algebra @xmath566 with coefficients in the @xmath566-bimodule @xmath565 .
the `` cohochschild '' differential @xmath569 is defined in dual manner .
it turns out that @xmath570 forms a bicomplex shown in figure [ fig1 ] .
( 20,15)(-3,0 ) ( 0,1 ) ( 2.5,0 ) ( 0,0)(0,0)@xmath571 ( 10,0)(0,0)@xmath572 ( 20,0)(0,0)@xmath573 ( 30,0)(0,0)@xmath574 ( 2.5,5 ) ( 0,0)(0,0)@xmath575 ( 10,0)(0,0)@xmath576 ( 20,0)(0,0)@xmath577 ( 30,0)(0,0)@xmath578 ( 2.5,10 ) ( 0,0)(0,0)@xmath579 ( 10,0)(0,0)@xmath580 ( 20,0)(0,0)@xmath581 ( 30,0)(0,0)@xmath582 ( 5.5,0)(0,0)(10,0)4 ( 1,0)3(-1.2,1)(0,0)@xmath583 ( 5.5,5)(0,0)(10,0)4 ( 1,0)3(-1.2,1)(0,0)@xmath583 ( 5.5,10)(0,0)(10,0)4 ( 1,0)3(-1.2,1)(0,0)@xmath583 ( 2.5,1)(0,0)(0,5)3 ( 0,1)3(-1,1.2)(0,0)@xmath584 ( 12.5,1)(0,0)(0,5)3 ( 0,1)3(-1,1.2)(0,0)@xmath584 ( 22.5,1)(0,0)(0,5)3 ( 0,1)3(-1,1.2)(0,0)@xmath584 ( 32.5,1)(0,0)(0,5)3 ( 0,1)3(-1,1.2)(0,0)@xmath584 the _ gerstenhaber - schack _ cohomology of @xmath13 with coefficients in @xmath13 is the cohomology of its ( regraded ) total complex @xmath585 where @xmath586 let us compare now the gerstenhaber - schack cohomology with the cohomology @xmath587 recalled in ( [ tel - aviv ] ) , where @xmath31 is the for bialgebras . to this end , we need to review some facts about the minimal model of @xmath31 . for a generator @xmath588 of biarity @xmath119 ( @xmath429 ` inputs ' and @xmath589 ` outputs , ' @xmath590 ) , let @xmath591 \otimes
\bfk \cdot \xi \otimes \bfk[\sigma_n]$ ] , with the obvious mutually compatible left @xmath463- right @xmath464-actions .
the minimal model @xmath592 of @xmath31 is of the form @xmath593 , where @xmath594 with @xmath595 see ( * ? ? ?
* theorem 16 ) .
the differential @xmath543 of @xmath544 is explicitly determined by ( * ? ? ?
3.1 ) . while general free are complicated objects , the fact that the @xmath39-bimodule @xmath596 is a direct sum of regular representations implies the following relatively simple description of @xmath597 .
let us agree that , in this section , by directed a graph we mean a finite , not necessary connected , graph @xmath121 such that * each edge @xmath598 is equipped with a direction and there are no directed cycles ( wheels ) in @xmath121 .
the direction of edges determines at each vertex @xmath599 of a directed graph @xmath121 a disjoint decomposition @xmath600 of the set of edges adjacent to @xmath160 into the set @xmath601 of incoming edges and the set @xmath602 of outgoing edges .
we will also assume that * for each vertex @xmath599 , the sets @xmath601 and @xmath602 are linearly ordered .
the pair @xmath603 is called the _ biarity _ of @xmath160 .
a vertex @xmath124 is _ binary _ if @xmath604 .
next assumption we impose on the graphs in this section is that * @xmath121 has no vertices of biarity @xmath605 or @xmath606
. vertices of biarity @xmath607 are called the _ input vertices _ and vertices of biarity @xmath608 the _ output vertices _ of @xmath121 .
we finally denote by @xmath609 the set of isomorphism classes of directed graphs @xmath121 satisfying ( i)(iii ) above such that * the input vertices of @xmath121 are labeled by @xmath610 and the output vertices by @xmath611 . with this notation , @xmath612 the degree @xmath11 differential @xmath543 on @xmath597 is of course uniquely determined by its values @xmath613 ,
@xmath614 , on the generators of @xmath596 .
we will see more explicitly below how these values look , now just write @xmath615 where @xmath616 , @xmath617 is a finite indexing set depending on @xmath614 , and @xmath618 .
it follows from the results of ( * ? ? ?
* section 4 ) that the minimal model of @xmath31 is the cellular chain complex of a sequence of finite polytopes whose faces are indexed by directed graphs that we may in ( [ je_to_takova_glegla ] ) assume @xmath619 . in particular , the minimal model of @xmath31
is defined over the integers ! this has to be compared to a similar argument proving the integrality of the minimal model of the operad for associative algebras based on the existence of the associahedra ( * ? ? ?
* example 4.8 ) . in the same manner , any derivation @xmath620
is uniquely determined by specifying , for each @xmath621 , multilinear maps @xmath622 this defines an isomorphism @xmath623 which we use to identify @xmath624 with @xmath625 .
let us inspect how the differential @xmath626 of the cochain complex @xmath624 acts on some @xmath627 .
if we denote by @xmath628 the component of @xmath629 in @xmath630 , then @xmath631,\ ] ] where @xmath632 $ ] is an element of @xmath633 which is nontrivial only if the vertex @xmath634 has biarity @xmath635 and if all remaining vertices of @xmath153 are binary . in this case
@xmath632 $ ] is obtained by decorating the vertex @xmath160 with @xmath636 , vertices of biarity @xmath637 with the multiplication @xmath549 , vertices of biarity @xmath638 with the comultiplication @xmath550 , and then performing the compositions indicated by the graph @xmath153 .
we leave as an exercise to show that formula ( [ psano_v_koline ] ) indeed follows from the definition of @xmath27 with @xmath639 recalled in section [ intro ] .
it turns out ( see appendix b on page ) that @xmath640 only if @xmath641 or @xmath642 and that @xmath643 this shows that the cochain complexes @xmath34 and @xmath644 are isomorphic and that our cohomology @xmath587 coincides with the gerstenhaber - schack cohomology of @xmath13 .
the @xmath0-structure on @xmath645 announced in the abstract is determined by an obvious generalization of ( [ psano_v_koline ] ) .
the components @xmath646 of the bracket @xmath647 can be computed as @xmath648,\end{aligned}\ ] ] where @xmath649 is as in ( [ dnes_jsem_byl_na_sipcich ] ) and in the second summation we assume that all vertices @xmath650 are mutually different .
the multilinear map @xmath651 \in \lin(\otexp vn,\otexp vm)$ ] is nonzero only if @xmath652 for @xmath653 and if all other remaining vertices of @xmath153 are binary . when it is so , then @xmath654 $ ] is defined by decorating the vertices @xmath129 with @xmath655 , @xmath130 , vertices of biarity @xmath637 with @xmath549 , vertices of biarity @xmath638 with @xmath550 , and then performing the compositions indicated by the graph @xmath153 . before we give examples of these @xmath0-brackets , we need to recall the calculus of fractions introduced in @xcite .
a fraction is a special type of a composition of elements of a defined using a restricted class of permutations which we we need to recall first . for @xmath656 and @xmath657 ,
let @xmath658 be the permutation given by @xmath659 where @xmath660 is such that @xmath661 .
permutations of this form are called _
special permutations_. an example is the permutation @xmath662 in ( [ compatibility ] ) .
another example is @xmath663 with the convention that the ` flow diagrams ' should be read from the bottom to the top .
let @xmath2 be a .
let @xmath664 , @xmath665 , @xmath666 , @xmath667 and @xmath668 .
then the _ @xmath669-fraction _ is defined as @xmath670 if @xmath671 or @xmath672 , the @xmath669-fractions are just the ` operadic ' compositions : @xmath673 as a more complicated example , consider @xmath674 and @xmath675 .
then -1.3em @xmath676{$d$ } } \put(10.00,10.00){\makebox(0.00,0.00)[l]{$c$ } } \put(50.00,60.00){\makebox(0.00,0.00)[l]{$b$ } } \put(10.00,60.00){\makebox(0.00,0.00)[l]{$a$ } } \put(60.00,50.00){\line(0,-1){30.00 } } \put(40.00,0.00){\line(0,1){20.00 } } \put(70.00,0.00){\line(-1,0){30.00 } } \put(70.00,20.00){\line(0,-1){20.00 } } \put(40.00,20.00){\line(1,0){30.00 } } \put(40.00,70.00){\line(0,-1){20.00 } } \put(70.00,70.00){\line(-1,0){30.00 } } \put(70.00,50.00){\line(0,1){20.00 } } \put(40.00,50.00){\line(1,0){30.00 } } \put(50.00,50.00){\line(-1,-1){30.00 } } \put(20.00,50.00){\line(1,-1){30.00 } } \put(0.00,0.00){\line(0,1){20.00 } } \put(30.00,0.00){\line(-1,0){30.00 } } \put(30.00,20.00){\line(0,-1){20.00 } } \put(0.00,20.00){\line(1,0){30.00 } } \put(10.00,40.00){\line(0,-1){20.00 } } \put(10.00,50.00){\line(0,-1){10.00 } } \put(30.00,70.00){\line(-1,0){30.00 } } \put(30.00,50.00){\line(0,1){20.00 } } \put(0.00,50.00){\line(1,0){30.00 } } \put(0.00,70.00){\line(0,-1){20.00 } } \end{picture}}\hskip 2mm.\ ] ] similarly , for @xmath677 and @xmath678 , -1.3em @xmath679{$v$ } } \put(50.00,10.00){\makebox(0.00,0.00)[l]{$u$ } } \put(10.00,10.00){\makebox(0.00,0.00)[l]{$z$ } } \put(90.00,60.00){\makebox(0.00,0.00){$y$ } } \put(20.00,60.00){\makebox(0.00,0.00){$x$ } } \put(110.00,20.00){\line(-1,0){30.00 } } \put(110.00,0.00){\line(0,1){20.00 } } \put(80.00,0.00){\line(1,0){30.00 } } \put(80.00,20.00){\line(0,-1){20.00 } } \put(40.00,0.00){\line(0,1){20.00 } } \put(70.00,0.00){\line(-1,0){30.00 } } \put(70.00,20.00){\line(0,-1){20.00 } } \put(40.00,20.00){\line(1,0){30.00 } } \put(0.00,0.00){\line(0,1){20.00 } } \put(30.00,0.00){\line(-1,0){30.00 } } \put(30.00,20.00){\line(0,-1){20.00 } } \put(0.00,20.00){\line(1,0){30.00 } } \put(60.00,20.00){\line(1,1){30.00 } } \put(20.00,20.00){\line(2,1){60.00 } } \put(30.00,50.00){\line(2,-1){60.00 } } \put(20.00,50.00){\line(1,-1){30.00 } } \put(100.00,50.00){\line(0,-1){30.00 } } \put(70.00,70.00){\line(0,-1){20.00 } } \put(110.00,70.00){\line(-1,0){40.00 } } \put(110.00,50.00){\line(0,1){20.00 } } \put(70.00,50.00){\line(1,0){40.00 } } \put(0.00,70.00){\line(0,-1){20.00 } } \put(40.00,70.00){\line(-1,0){40.00 } } \put(40.00,50.00){\line(0,1){20.00 } } \put(0.00,50.00){\line(1,0){40.00 } } \put(10.00,50.00){\line(0,-1){30.00 } } \end{picture}}\hskip 2pt.\ ] ] to use formula ( [ dnes_prijede_ten_american ] ) , we need of course to know the differential @xmath543 in the minimal model @xmath680 which determines the graphs @xmath153 . in @xcite , @xmath543 was described as a perturbation , @xmath681 , of its -part @xmath682 .
let us therefore give some formulas for the unperturbed part @xmath682 first .
if we denote @xmath683 and @xmath684 , then @xmath685 . if @xmath686 , then @xmath687 . with the obvious , similar notation ,
@xmath688 and here is the differential @xmath543 in its full beauty : @xmath689 , @xmath690 , @xmath691 , @xmath692 , # 1#2 @xmath693 let us finally see what formula ( [ dnes_prijede_ten_american ] ) tells us in some concrete situations . for @xmath694 , the component @xmath695 is calculated as @xmath696 = ( -1)^{|f| } \sum_{u , v } \left(\zbbzb - \bzbbz\right)^{\{u , v\ } } [ f , g],\ ] ] where @xmath697 .
expanding the right - hand side , we obtain @xmath698 which we easily recognize as the `` classical '' gerstenhaber bracket @xmath699 $ ] of bilinear cochains . the component @xmath700 is given by @xmath701 = ( -1)^{|f| } \sum_{u , v } \left(\dvojiteypsilon - \mfrac{\jednadva \hskip .2em \jednadva}{\dvajedna \hskip .2em \dvajedna } \right)^{\{u , v\ } } [ f , g ] \\ & = & -\sum_{u , v } \dvojiteypsilon^{\{u , v\ } } [ f , g ] + \sum_{u , v } \mfrac{\jednadva \hskip .2em
.2em \dvajedna}^{\{u , v\ } } [ f , g].\end{aligned}\ ] ] the first term of the last line is zero , while the second one expands to @xmath702 this provides the first example of a `` non - classical '' bracket . in terms of elements , @xmath703 for @xmath704 , where the standard sweedler notation for the coproduct is used .
an example of a triple bracket is given by @xmath705 slightly more complicated is : @xmath706 we believe that the reader already understands the ideas behind our definitions and that he or she can easily construct other examples of brackets using explicit formulas for the differential @xmath543 ( * ? ? ?
notice that the bottom row @xmath707 of the bicomplex in figure [ fig1 ] is the hochschild complex of the algebra @xmath708 with coefficients in itself . for arbitrary @xmath709 ,
the component @xmath710 of @xmath711 coincides with the classical gerstenhaber bracket of the hochschild cochains @xcite while the components @xmath712 , @xmath713 , of the higher brackets are trivial . in this sense
our construction extends the classical gerstenhaber bracket of hochschild cochains .
let @xmath714 be a vector space @xmath7 considered as a bialgebra with trivial product and coproduct .
let @xmath715 , @xmath716 and @xmath717 .
finally , let @xmath718 be the @xmath0-structure on the gerstenhaber - schack complex @xmath719 constructed above .
clearly @xmath720 .
let us verify directly that , as predicted by proposition [ pisu_v_aute_ve_dvore ] , the element @xmath721 satisfies the master equation ( [ 44 ] ) if and only if @xmath722 forms a bialgebra . from degree reasons ,
the only possibly nontrivial components of @xmath723 are @xmath724 these values are determined by @xmath725 , @xmath726 and @xmath727 .
looking at these values we see that @xmath728 only for @xmath729 or @xmath730 .
the components of @xmath731 are @xmath732 the components of @xmath733 are @xmath734 the above calculations make the claim obvious .
the material of this appendix is taken almost word - by - word from @xcite .
a symmetric brace algebra is a graded vector space @xmath66 together with a collection of degree @xmath735 multilinear braces @xmath736 that are graded symmetric in @xmath737 and satisfy the identities @xmath738 and @xmath739 where the sum is taken over all unshuffle decompositions @xmath740 of @xmath341 and where @xmath313 is the koszul sign of the permutation @xmath741 of elements of @xmath66 . for elements
@xmath742 of an arbitrary symmetric brace algebra @xmath66 , put @xmath743 .
one easily proves that then @xmath744 is a graded pre - lie algebra in the sense of ( * ? ? ?
* section 2 ) .
vice versa , higher brackets @xmath745 of an arbitrary symmetric brace algebra are , for @xmath746 , determined by their ` pre - lie part ' @xmath747 .
for instance , axiom ( [ axiom ] ) implies that @xmath748 can be expressed as @xmath749 the same axiom applied on @xmath750 can then be clearly interpreted as an inductive rule defining @xmath751 in terms of @xmath752 , with @xmath753 . as proved in @xcite , _ an arbitrary _ pre - lie algebra determines in this way a _ unique _ symmetric brace algebra .
let us emphasize that this statement is not obvious .
first , axiom ( [ axiom ] ) interpreted as an inductive rule is ` overdetermined . ' for example , @xmath754 can be expressed both from ( [ axiom ] ) applied to @xmath755 and also from ( [ axiom ] ) applied to @xmath756 , and it is not obvious whether the results would be the same .
second , even if the braces are well - defined , it is not clear whether they satisfy the axioms of brace algebras , including the graded symmetry .
in this appendix we indicate how to prove that the gerstenhaber - schack differential is the same as the differential given by formula ( [ je_to_takova_glegla ] ) that involves the minimal model @xmath16 of the bialgebra @xmath31 .
this can in principle also be achieved by analyzing the explicit minimal model described in @xcite , but we sketch out a procedure that requires much less information .
we will explain it for the case of bialgebras , but it will be clear how to generalize it to algebras over an arbitrary .
let us start by rewriting ( [ je_to_takova_glegla ] ) in a way that will make it clear which part of the minimal model it really needs . as we already remarked , we know that the minimal model of the bialgebra @xmath16 is of the form @xmath757 , where @xmath588 are , for @xmath614 defined in ( [ sm_je_hovado ] ) , generators of biarity @xmath119 and degree @xmath758 .
consider the free @xmath759 where @xmath588 are the generators of the @xmath16 , @xmath760 is a generator of biarity @xmath606 placed in degree @xmath735 , and @xmath761 are , for @xmath614 , the ` suspended ' generators of @xmath16 , i.e. @xmath762 is of biarity @xmath119 and of degree @xmath763 .
introduce finally the degree @xmath26 derivation @xmath764 by @xmath765 we use @xmath766 to equip @xmath767 with the differential @xmath768 given by @xmath769 } \circ \xi^m_n - \xi^m_n \circ \varphi^{[n ] } - \dd(\psfb(\xi^m_n)),\ \mbox { for $ ( m , n ) \in i$.}\end{aligned}\ ] ] where @xmath770 } : = \sum_{0 \leq j \leq k-1 } ( \id^{\ot j } \ot \varphi \ot \id^{\ot ( k - j-1)}),\ k \geq 1,\ ] ] with @xmath771 the unit and @xmath772 denoting the horizontal composition . to prove that @xmath773
, one needs to observe that @xmath774 } \circ x - x \circ \varphi^{[n ] } - \dd(\psfb(x))\ ] ] for each @xmath775 ; notice that ( [ hajzl_ruskej ] ) is ( [ hovado_rusky ] ) with @xmath776 . then @xmath777 } \circ \xi^m_n - \xi^m_n \circ \varphi^{[n ] } - \dd(\psfb(\xi^m_n))\right ) \\ & = & \varphi^{[m ] } \circ \psfb(\xi^m_n ) - \psfb(\xi^m_n ) \circ \varphi^{[n ] } - \psfb(\dd(\psfb(\xi^m_n ) ) ) = 0,\end{aligned}\ ] ] where the vanishing in the second line follows from ( [ hovado_rusky ] ) applied to @xmath778 .
the vanishing @xmath779 is obvious .
representing @xmath780 as in ( [ je_to_takova_glegla ] ) , the rightmost term @xmath781 in ( [ hajzl_ruskej ] ) can be written as @xmath782 with @xmath783 the decorated graph obtained from @xmath153 by replacing the vertex @xmath160 of biarity , say , @xmath784 ( which is by definition decorated by @xmath785 ) by the vertex of the same biarity decorated by @xmath786 .
loosely speaking , @xmath783 is obtained by raising the degree of @xmath160 by one .
consider the @xmath31-submodule @xmath793 of the coproduct @xmath790 generated by @xmath794 and @xmath760 .
it is spanned by monomials in @xmath795 containing precisely one @xmath762 or @xmath760 .
it is clear that @xmath796 is @xmath791-stable and that it in fact represents the free @xmath31-module generated by @xmath794 and @xmath760 . on the other hand , for arbitrary @xmath784
, the structure of @xmath6 determines an @xmath6-module action on the suspension @xmath797 .
given a bialgebra @xmath798 , the corresponding map @xmath32 composed with this action determines a @xmath31-module structure on @xmath797 . by the freeness of @xmath793 , each @xmath799 specifies the @xmath31-module map @xmath800 that satisfies @xmath801 using ( [ hajzl ] ) , one can rewrite ( [ psano_v_koline ] ) as @xmath802 this equality has an important consequence which we formulate as in the rest of this appendix we show that the dg- @xmath804 can be described without the knowledge of the minimal model @xmath16 of the bialgebra @xmath31 .
we say that a degree @xmath735 map @xmath805 is a _ derivation _ of a bialgebra @xmath806 if @xmath807 let @xmath808 be the describing structures consisting of a bialgebra @xmath13 and a derivation @xmath809 of @xmath13 .
consider the homomorphism @xmath810 such that @xmath811 coincides with the canonical inclusion @xmath812 , @xmath813 for @xmath614 and @xmath814 is the generator for the derivation .
let us check that @xmath815 defined in this way is a dg - map .
the equation @xmath816 is clear .
the vanishing of for @xmath817 follows from degree reasons .
it remains to show that . by
( [ hajzl_ruskej ] ) , @xmath818}))= \vartheta \circ \mu - \mu \circ ( \vartheta \ot \id + \id \ot \vartheta),\ ] ] where we denoted by the same symbols operations on @xmath7 and the corresponding generators in @xmath808 ; we are sure this will not lead to a confusion here .
we conclude that @xmath819 vanishes because @xmath809 is a @xmath549-derivation .
the vanishing of @xmath820 can be proved in the same manner .
the central statement of this section is : the above proposition says that @xmath822 is _ a @xmath31-free minimal model _ of the @xmath808 . by the @xmath31-freeness
we mean that @xmath804 is obtained by adding free generators to @xmath31 .
minimality means that the image of the differential @xmath791 consists of decomposable elements of the augmented @xmath804 see @xcite for a definition of indecomposables in augmented . assuming uniqueness of minimal models , _ any _ @xmath31-free minimal model of @xmath808 will be isomorphic to @xmath804 .
one candidate for such a model can be constructed by expanding formulas for the gerstenhaber - schack differential into diagrams ; denote the -free dg- obtained in this way by @xmath823 .
methods developed in @xcite then can be used to prove that the canonical map @xmath824 is a homology isomorphism .
an analog of ( [ 67 ] ) for @xmath823 then identifies @xmath626 with the gerstenhaber - schack differential .
let us prove ( i ) first . consider the grading @xmath828 of @xmath767 defined by @xmath829 and decompose @xmath768 as @xmath830 , where @xmath831 raises the grading by one and @xmath832 preserves it .
one can easily check that @xmath833 and that @xmath834 , therefore @xmath835 is a bicomplex .
it is not difficult to prove that @xmath836 with the ideal @xmath837 generated by the relations @xmath838}\circ
\xi^m_n \circ \varphi^{[n ] } = 0,\ ( m , n ) \in i,\ ] ] which say that @xmath760 is a derivation with respect to all @xmath588 , see ( [ svine_ruska ] ) for the notation .
it follows from ( [ 45 ] ) that the @xmath839-term of the spectral sequence associated to this bicomplex equals @xmath840 where @xmath12 coincides with the minimal model differential @xmath841 on the @xmath842-generators and @xmath843 .
we conclude that @xmath844 , so the associated spectral sequence degenerates at the @xmath845-level from degree reasons and @xmath846 .
one can easily see that the latter isomorphism is induced by @xmath847 .
this proves ( i ) . to prove
( ii ) , one introduces the grading @xmath848 of @xmath767 by formulas @xmath849 and a similar grading @xmath850 of @xmath804 by @xmath851 since @xmath852 preserves these gradings , it preserves also the filtrations @xmath853 and induces the map @xmath854 of the induced spectral sequences .
clearly @xmath855 where @xmath12 coincides with the differential @xmath841 of the minimal model @xmath16 of @xmath31 on the @xmath842-generators and is trivial on the remaining ones , so @xmath856 .
on the other hand , clearly @xmath857 , therefore @xmath858 , too .
in fact , it is not hard to see that @xmath859 is an isomorphism of complexes .
a standard spectral sequence argument then finishes our proof of ( ii ) .
m. aguiar .
infinitesimal bialgebras , pre - lie and dendriform algebras . in j.
bergen , editor , _ hopf algebras , proceedings from the international conference , depaul university , chicago , il , usa _ ,
volume 237 of _ lect
. notes pure appl
_ , pages 133 .
marcel dekker , 2004 . | we show that there exists a lie bracket on the cohomology of any type of ( bi)algebras over an operad or a , induced by an @xmath0-structure on the defining cochain complex , such that the associated @xmath0-master equation captures deformations .
this in particular implies the existence of a lie bracket on the gerstenhaber - schack cohomology @xcite of a bialgebra that extends the classical intrinsic bracket @xcite on the hochschild cohomology , giving an affirmative answer to an old question about the existence of such a bracket .
we also explain how the results of @xcite provide explicit formulas for this bracket .
we assume a certain familiarity with operads and , see @xcite .
the reader who wishes only to know how the intrinsic bracket on the gerstenhaber - schack cohomology looks might proceed directly to section [ gs ] which is almost independent on the rest of the paper and contains explicit calculations .
we also assume some knowledge of the concept of strongly homotopy lie algebras ( also called @xmath0-algebras ) , see @xcite .
we will make no distinction between an operad @xmath1 and the @xmath2 generated by this operad .
this means that for us operads are particular cases of . as usual
, bialgebra will mean a hopf algebra without ( co)unit and antipode . to distinguish these bialgebras from other types of `` bialgebras '' we will sometimes call them also @xmath3-bialgebras .
all algebraic objects will be defined over a fixed field @xmath4 of characteristic zero although , surprisingly , our constructions related to @xmath3-bialgebras make sense over the integers . |
the original motivation for the study of kaon - pair production in the @xmath8 reaction near threshold was the investigation of the structure of the scalar mesons @xmath9 or @xmath10 @xcite .
such measurements were initially performed by the cosy-11 collaboration at several different excess energies below the @xmath1-meson production threshold @xcite .
however , their results showed that scalar meson production can not in fact be the dominant driving mechanism in kaon pair production @xcite and that the data can be explained without the explicit inclusion of the @xmath11 .
furthermore , they showed that the @xmath4 and @xmath12 invariant mass spectra were strongly distorted , presumably by the @xmath4 final state interaction ( fsi ) @xcite .
this was most apparent in the ratio of the differential cross sections in terms of the @xmath4 and @xmath13 invariant masses .
the @xmath0 reaction was also investigated with higher statistics above the threshold for the production of the @xmath1 meson , mainly with the aim of investigating the properties of that meson @xcite . after removing the @xmath1 contribution in the spectra
, it was clear that the @xmath4 and @xmath12 distributions in the non-@xmath1 data were both strongly influenced by the @xmath4 interaction @xcite .
it has been suggested that this is connected with the production of the @xmath14 excited hyperon @xcite , which might be treated as a @xmath15 quasi - bound state with a width that overlaps the @xmath15 threshold @xcite .
this idea was put on a quantitative footing by assuming that the @xmath14 was formed through the decay @xmath16 @xcite . the strength and details of the @xmath15 interaction are clearly important elements in the interpretation of possible kaon nuclear systems , such as the deeply bound @xmath12 states @xcite .
in addition to the @xmath4 fsi , and one between the two protons , the data also showed an enhancement at low @xmath5 invariant masses with some possible structure at the @xmath6 threshold @xcite .
though the effects are small , they might be influenced by the @xmath9 or @xmath10 scalar mesons .
however , the investigation of this region was hampered by the need to separate the non-@xmath1 from the @xmath1 contribution and the fact that the data were spread over a very wide range of @xmath5 invariant masses .
measurements below the @xmath1 threshold can provide useful information on these interesting fsi effects without suffering the distortion of the @xmath1 meson .
however , the limited statistics in the low - energy cosy-11 data @xcite are insufficient for detailed studies .
previous measurements of the @xmath0 reaction were carried out at the cosy - anke magnetic spectrometer at @xmath17 , 67 , and 108 mev @xcite , where the @xmath1 threshold is at @xmath18 mev . here
the excess energy is defined as @xmath19 , where @xmath20 is the total center - of - mass energy and @xmath21 and @xmath22 are the particle masses in the final state . because of the limited acceptance of this spectrometer , an ansatz has to be made regarding the distribution of events over the four - body phase space in order to convert count rates into cross sections .
this was done assuming that the distortions were the products of those present in the two - particle subsystems .
all the anke non-@xmath1 data seemed to be consistent with an _ effective _ scattering length of @xmath23 fm with no obvious influence of an energy dependence associated with an effective range term .
the dominance of the imaginary part is not unexpected because of the strong couplings to the @xmath24 and @xmath25 channels but , due to the presence of two other final - state particles , this parameter is not necessarily an intrinsic feature of the isolated @xmath4 system .
the anke measurements at three excess energies also showed some enhancement at low @xmath5 invariant masses but with at least a break of slope at the @xmath6 threshold . a combined analysis of all the results in this region @xcite shows that the data can be understood in terms of a final state interaction involving both @xmath5 elastic scattering plus a contribution from the @xmath26 charge exchange .
although suggestive , the data are not sufficient to draw firm conclusions .
in this paper we present much more precise @xmath0 differential cross section data at a beam energy of @xmath27 gev ( @xmath28 mev ) obtained using the cosy - anke spectrometer . with high statistics on the reaction below the @xmath1-meson threshold
, we can study the effects of the final state interactions in the @xmath4 and @xmath5 systems in greater details .
the paper is organized as follows .
we first describe the experimental setup and data analysis in sec [ experiment ] . given that the procedures involved are similar to those employed at higher energies @xcite
, this can be quite brief .
the fitting of the phenomenological parametrization to the raw @xmath0 data in order to make acceptance corrections is also described here .
the resulting differential cross sections and total cross section for the @xmath0 reaction are presented in sec [ results ] , followed by our conclusions in sec [ conclusions ] .
the measurement of the @xmath0 reaction was performed at an internal target station of the cooler synchrotron ( cosy ) of the forschungszentrum jlich @xcite .
the anke spectrometer @xcite , which consists of three dipole magnets , registers positively and negatively charged ejectiles in the side detection systems , with the fast positively charged particles being detected in the forward detector .
particle identification relies on time - of - flight measurements @xcite from start and stop counters , and momentum information obtained from the multiwire proportional chambers .
missing - mass distribution in the @xmath29 reaction at @xmath27 gev .
the hatched histogram shows the cuts imposed for the selection of the non - detected proton .
the solid line , which is a second - order polynomial fit , was used to estimate the background contribution under the proton peak . ] in close - to - threshold production experiments , the total cross section changes very rapidly with small changes in the excess energy .
the proton beam energy , @xmath30 gev , was therefore determined very precisely through a careful monitoring of the schottky spectra @xcite .
the resulting value of the excess energy with respect to the @xmath31 production threshold , @xmath28 mev , is well below the nominal @xmath1 threshold .
however it should be noted that this is an average value , since the beam energy decreases by up to 4.6 mev through the course of a machine cycle due to the interaction with the target .
this effect was also investigated in the simulation .
the experiment relied on a triple - coincidence , involving the observation of a @xmath5 pair in the side detectors and a fast proton in the forward detector .
the @xmath0 reaction was then identified by requiring that the missing mass of the @xmath32 system be consistent with that of a proton . in the analysis , a @xmath33 ( @xmath34 mev/@xmath35 )
cut was applied on the missing - mass distribution of the selected @xmath32 events , as shown in fig .
[ mm2.57 ] .
the fraction of misidentified events inside the cut window around the proton mass was estimated to be about 5% , which was subtracted from the peak using weighted data from the side bands , as parameterized by the solid line .
any ambiguity in this procedure , which is less than 3% , is one source of systematic uncertainty .
reaction at @xmath28 mev .
vertical bars represent the statistical uncertainties and horizontal ones the bin widths .
the individual panels are ( a ) the cosine of the polar angle of the @xmath36 in the @xmath5 reference frame , ( b ) the polar angle of the kaon pairs in the overall c.m .
frame , ( c ) the polar angle of the emitted proton in the @xmath37 reference frame relative to the beam direction , ( d ) the polar angle of the proton in the @xmath37 reference frame relative to the direction of the kaon pair , ( e ) the proton momentum in the @xmath37 reference frame , and ( f ) the @xmath32 invariant mass distribution . ] after identifying clean @xmath0 events in anke , acceptance corrections must be performed in order to evaluate differential cross sections .
the simple ansatz used on data taken above the @xmath1 meson production threshold tried to take into account the influence of final state interactions in the various two - particle subsystems @xcite .
this ansatz , which is also the basis for the current simulation , assumes that the overall enhancement factor @xmath38 is the product of enhancements in the @xmath37 , @xmath5 , and @xmath4 systems : @xmath39 where @xmath40 , @xmath41 , @xmath42 , and @xmath43 are the magnitudes of the relative momenta in the @xmath37 , the two @xmath4 , and the @xmath5 system , respectively .
it is believed that the @xmath13 interaction might be weakly repulsive and , if so , its neglect would be interpreted as extra attraction in the @xmath4 system .
the fsi enhancement in the @xmath4 case was calculated in the scattering length approximation , @xmath44 and the best fit to the higher energy data @xcite was found with a purely imaginary effective scattering length , @xmath45 fm .
the proton - proton enhancement factor was derived from the jost function @xcite .
the enhancement factor in the @xmath5 system takes into account elastic @xmath5 scattering plus the charge - exchange @xmath26 @xcite .
the seven degrees of freedom required to describe the unpolarized @xmath31 final state were chosen to be four angles , the @xmath46 and @xmath47 invariant masses , and the relative momentum of the protons in the @xmath37 system @xcite .
distributions in these seven variables were generated inside the anke acceptance and compared with the experimental data , some of which are shown in fig .
the best fit to the data was achieved with @xmath48 fm , which is significantly larger in magnitude than the starting value of @xmath49 fm .
the uncertainty in the real part is large and strongly correlated with the imaginary part . to allow easy comparison with the analysis of the higher energy data @xcite ,
the effective scattering length was taken to be purely imaginary .
the @xmath50 scattering lengths for isospin - one and zero were taken as in our previous work @xcite and the ratio of the @xmath51 and @xmath52 production amplitudes of @xmath53-wave @xmath54 pairs was parameterized as @xmath55 .
the best fit was obtained with @xmath56 and @xmath57 , which are consistent with our earlier evaluation @xcite based on the above @xmath1 threshold data .
the resulting descriptions of the experimental data in fig .
[ raw ] are very good and certainly sufficient for evaluating the acceptance corrections .
the luminosity needed in the analysis was determined with an overall systematic uncertainty of 9% by measuring @xmath37 elastic scattering in the forward detector @xcite .
this was checked by simultaneous studies of the beam current and schottky spectra @xcite , which could fix the absolute luminosity with a systematic uncertainty of 6% . within these uncertainties
the two methods agreed but , in order to be coherent with our previous work , the luminosity extracted from the @xmath37 elastic scattering data was used in the final analysis .
invariant mass distribution for the @xmath0 reaction at @xmath28 mev .
the dotted curve shows the four - body phase space simulation whereas the inclusion of the final state interactions through eq . gives the dashed curve for @xmath49 fm and the red solid curve @xmath58 fm .
the dot - dashed curve was obtained by considering only the @xmath37 and @xmath4 final state interactions with @xmath58 fm . ] the differential cross section for the @xmath0 reaction at an excess energy @xmath28 mev is shown in fig .
[ i m ] as a function of the @xmath5 invariant mass . also shown are simulations based on a four - body phase space and this distorted by the final state interactions in the @xmath5 , @xmath37 , and @xmath4 systems within the product ansatz of eq . .
this was done separately with effective scattering lengths of @xmath49 fm and @xmath58 fm .
the most striking features in the data are the strength near the @xmath5 threshold and the dip at @xmath59 gev/@xmath35 , which corresponds precisely to the @xmath60 production threshold @xcite .
this is compelling evidence for a cusp effect coming from the @xmath61 transitions . to investigate this phenomenon in greater detail ,
the @xmath5 invariant mass distribution was divided by a simulation where only the final state interactions in the @xmath37 and @xmath4 , with @xmath58 fm , were considered .
the best fit to the data shown in fig . [ enhance ] is achieved with a contribution from the isospin - zero channel that is about three times stronger than the isospin - one .
this finding is consistent with our earlier result @xcite .
the deviations apparent in figs .
[ i m ] and [ enhance ] at high @xmath5 invariant masses might be connected with the approximations made in our coupled - channel model @xcite .
invariant mass distribution in the @xmath0 reaction to a simulation that includes only @xmath4 and @xmath37 final state interactions ( shown by the dot - dashed curve in fig .
[ i m ] ) .
in addition to the current data ( solid circles ) , weighted averages of previous measurements ( open squares and circles ) are also presented .
the solid curve represents the best fit in a model that includes elastic @xmath5 fsi and @xmath61 charge - exchange @xcite .
the best fits neglecting charge exchange and including only this effect are shown by the dashed and dot - dashed curves , respectively . ]
previous analyses of the @xmath0 reaction at different excess energies @xcite have all shown a strong preference for low values of the @xmath4 and @xmath12 invariant masses , @xmath62 and @xmath63 . to study this further ,
we have evaluated differential cross sections in terms of these invariant mass distribution and also their ratios : @xmath64 reaction with respect to the invariant masses of @xmath13 ( upper panel ) and @xmath4 ( middle panel ) , and their ratio @xmath65 ( lower panel ) .
the dotted histograms represent the four - body phase - space simulations , whereas the red solid and dashed black ones represent estimations based on eq .
that take into account @xmath4 , @xmath37 and @xmath5 final state interactions with @xmath58 fm and @xmath49 fm , respectively . ]
reaction with respect to the invariant masses of @xmath66 ( upper panel ) and @xmath12 ( middle panel ) , and their ratio @xmath67 ( lower panel ) .
the conventions for the theoretical estimates are as in fig .
[ ratio_pk ] ] the corresponding experimental data and simulations are shown in figs . [ ratio_pk ] and [ ratio_ppk ] .
both @xmath65 and @xmath67 display the very strong preferences for lower invariant masses seen in the earlier data .
the low mass enhancements in figs .
[ ratio_pk]c and [ ratio_ppk]c clearly indicate once again that the @xmath0 reaction can not be dominated by the undistorted production of a single scalar resonance @xmath68 or @xmath69 . within a four - body phase space simulation
both ratios should be constant and equal to one and such a simulation also fails to describe the individual @xmath70 and @xmath71 distributions . whereas the inclusion of a @xmath4 fsi with an effective scattering length @xmath49 fm improves the situation , it overestimates the data in the high invariant mass regions for both @xmath65 and @xmath67 . with the larger effective scattering length @xmath58 fm , these ratios , as well as the individual distributions , can be well reproduced . within the product ansatz of eq .
the @xmath4 final state interaction effectively becomes stronger at lower excess energies .
this illustrates the limitations of this simple ansatz to the complex four - body dynamics .
although the @xmath4 elastic final state interaction describes well the vast bulk of the data shown in figs .
[ ratio_pk ] and [ ratio_ppk ] , it is interesting to note that there seems to be a small but significant deviation between the @xmath4 data and simulation in fig .
[ ratio_pk]b at low invariant masses .
since the @xmath7 threshold is at 1437.2 gev/@xmath35 , this suggests that the data in this region might also be influenced by @xmath72 channel coupling . due to the low statistics , the cosy-11 data at 10 and 28 mev @xcite can not distinguish between predictions based on effective scattering lengths of @xmath49 fm and @xmath58 fm .
this illustrated for the @xmath65 ratio in fig .
[ ratio_10_28 ] but this lack of sensitivity is equally true for @xmath67 . for the @xmath0 reaction measured by cosy-11 at ( a ) @xmath73 mev and ( b ) 28 mev @xcite , respectively .
the dotted istograms represent the four - body phase - space simulations , whereas the red solid and dashed ones represent the theoretical calculations taking into account @xmath4 , @xmath37 and @xmath5 final state interactions with @xmath58 fm and @xmath74 fm , respectively . ] the acceptance - corrected invariant mass distribution shown in fig .
[ i m ] was used to determine the value of the total cross section , @xmath75 nb , where the errors reflect the statistical and systematical uncertainties , respectively .
the systematic effects considered here arise from the background subtraction , acceptance correction , tracking efficiency correction , and luminosity determination .
the total cross section result is plotted in fig .
[ txs ] along with previous measurements from disto @xcite , cosy-11 @xcite , and anke @xcite .
the new point seems high compared to the cosy-11 result at @xmath76 mev but it should be noted that this value was already increased by 50% compared to that originally quoted .
this was achieved through a re - analysis of the data that included @xmath37 and @xmath4 final state interactions with @xmath49 fm @xcite .
it is therefore possible that a further increase might follow if the @xmath5 final state interaction were included in the evaluation of the acceptance corrections as well as an effective scattering length of @xmath58 fm was used instead .
there seems to be less sensitivity to the fsi assumptions at @xmath73 mev , where the original re - analysis only increased the total cross section by 20% .
reaction as a function of excess energy @xmath77 .
the present result ( cross ) is shown together with earlier experimental data taken from disto ( triangle ) , anke ( circles ) , cosy-11 ( squares ) .
the dotted line shows the four - body phase space simulation , whereas the solid line represent the simulations with @xmath49 fm .
the predictions of a one - boson exchange model are represented by the dashed line @xcite . ]
it is clear from fig .
[ txs ] that the four - body phase space can not reproduce the energy dependence of the total cross section . with the inclusion of the @xmath37 , @xmath5 , and @xmath4 fsi , with an effective scattering length of @xmath49 fm , the data above the @xmath1 threshold can be described well but those at lower energy are significantly underestimated .
an increase in the value of @xmath78 might help in this region but the coincidence of strong effects in different two- or even three - body channels must also bring the factorization assumption of eq . into question .
the dashed line , which represents a calculation within a one - boson exchange model @xcite , also underestimates the near - threshold data .
this model includes energy dependent input derived from fits to the @xmath79 total cross sections , though it does not include the @xmath37 final state interaction .
the production of @xmath5 pairs has been measured in the @xmath0 reaction channel at an excess energy of @xmath28 mev .
even taking its 4.3 mev/@xmath35 width into account , this is well below the central @xmath1-meson threshold at 32.1 mev .
the reaction was identified in anke through a triple coincidence of a @xmath5 pair and a forward - going proton , with an additional cut on the @xmath32 missing - mass spectrum .
the high statistics and low excess energy allow us to produce a detailed @xmath5 invariant mass distribution below the @xmath1 threshold .
the distortion of both the @xmath4 and @xmath12 spectra , which are even stronger than in our higher energy data , can be explained quantitatively within the product ansatz of eq . with an effective @xmath4 scattering length @xmath80 fm .
this is to be compared with the @xmath74 fm obtained from the analysis of data measured above the @xmath1 production threshold .
a full treatment of the dynamics of the four - body @xmath31 channel is currently impractical . as a consequence ,
an energy dependence of @xmath78 is possible because this is merely an effective parameter within a very simplistic description of the four - body final state interaction .
the strong @xmath4 final state interaction may be connected with the @xmath14 in the production process and it has been suggested @xcite that the production of non-@xmath1 kaon pairs proceeds mainly through the associated production @xmath81 .
this would also lead to deviations from the simple product ansatz for the final state interactions , not least because an attraction between the @xmath14 and the proton would involve three final particles .
our results show a very strong preference for low @xmath12 masses and this effect seems to be even more marked than in the higher energy data @xcite .
although this might be connected with the ideas of a @xmath12 deeply bound state @xcite , it must be stressed that our data were measured far above threshold .
they should not therefore be taken as necessarily implying that the @xmath82 will bind with two protons .
there is strong evidence for a cusp effect arising from the @xmath83 transitions .
our analysis within a coupled - channel description suggests that , with the values of the @xmath54 scattering lengths used , the production of isospin - zero @xmath54 pairs dominates .
though this is consistent with results extracted from data taken above the @xmath1 threshold @xcite , there is clearly room for some refinement in the model . on the other hand , the structure of the @xmath4 invariant mass distribution of fig .
[ ratio_pk ] in the 1437 mev/@xmath35 region suggests that there might be important coupling also between the @xmath4 and @xmath7 systems .
it is evident that the interactions in the four - body @xmath31 final state are extremely complex .
nevertheless , the energy dependence of the total cross section can be well described above the @xmath1 threshold by introducing the effects of the @xmath37 , @xmath5 and @xmath4 final state interaction with an effective scattering length of @xmath49 fm . this would , however , have to be increased to have any hope of fitting the lower energy data .
further theoretical work is required to clarify the reaction mechanisms .
we would like to thank the cosy machine crew for providing the good conditions that were necessary for this work .
we are also grateful for the continued assistance of other members of the anke collaboration .
discussions with p. moskal were very helpful .
this work was supported in part by the us department of energy under contract no .
de - fg02 - 03er41231 and cosy ffe .
99 w. oelert , _ proceedings of the workshop on meson production , interaction , and decay , cracow , 1991 _ , ed .
a. magiera , w. oelert , and e. grosse ( world scientific , singapore , 1991 ) p. 199 .
m. wolke , ph.d .
thesis , westflische wilhelms - universitt mnster ( 1998 ) .
c. quentmeier _ et al .
_ , phys .
b * 515 * , 276 ( 2001 ) .
p. winter _
_ , phys .
b * 635 * , 23 ( 2006 ) .
f. balestra _ et al .
_ , phys .
c * 63 * , 024004 ( 2001 ) .
y. maeda _ et al .
_ , phys .
c * 77 * , 015204 ( 2008 ) .
a. dzyuba _ et al .
_ , phys .
b * 668 * , 315 ( 2008 ) .
q. j. ye _ et al .
_ , phys .
c * 85 * , 035211 ( 2012 ) . c. wilkin , acta phys .
* 2 * , 89 ( 2009 ) .
r. h. dalitz and s. f. tuan , phys .
* 2 * , 425 ( 1959 ) .
j. j. xie and c. wilkin , phys .
c * 82 * , 025210 ( 2010 ) .
t. yamazaki _ et al .
_ , phys .
* 104 * , 132502 ( 2010 ) .
r. maier _ et al .
_ , nucl .
instrum .
a * 390 * , 1 ( 1997 ) .
s. barsov _ et al .
_ , nucl .
instrum .
meth . phys
. res . a * 462 * , 364 ( 2001 ) .
m. hartmann _ et al .
_ , int .
j. mod .
a * 22 * , 317 ( 2007 ) .
m. hartmann _ et al .
_ , phys .
lett . * 96 * , 242301 ( 2006 ) .
m. bscher _ et al .
_ , nucl .
instrum .
a * 481 * , 378 ( 2002 ) .
h. j. stein _ et al .
_ , phys .
st accel .
beams * 11 * , 052801 ( 2008 ) .
m. silarski _ et al .
_ , phys .
c * 80 * , 045202 ( 2009 ) .
a. sibirtsev , w. cassing , and c. m. ko , z. phys .
a * 358 * , 101 ( 1997 ) .
m. agnello _ et al .
_ , phys .
lett.*94 * , 212303 ( 2005 ) .
t. yamazaki and y. akaishi , phys .
c * 76 * , 045201 ( 2007 ) .
shevchenko , a. gal , j. mare , and j. rvai , phys .
c * 76 * , 044004 ( 2007 ) . | the @xmath0 reaction was measured below the @xmath1 threshold at a beam energy of 2.568 gev using the cosy - anke magnetic spectrometer . by assuming that the four - body phase space is distorted only by the product of two - body final state interactions
, fits to a variety of one - dimensional distributions permit the evaluation of differential and total cross sections .
the shapes of the distributions in the @xmath2 and @xmath3 invariant masses are reproduced only if the @xmath4 interaction is even stronger than that found at higher energy .
the cusp effect in the @xmath5 distribution at the @xmath6 threshold is much more clear and some evidence is also found for coupling between the @xmath4 and @xmath7 channels .
however , the energy dependence of the total cross section can not be reproduced by considering only a simple product of such pair - wise final state interactions . |
recent advances in the molecules manipulation have made it possible to measure and characterize molecular properties at a single molecule level .
one of basic characteristics is the extension - force curves ( efcs)@xcite .
these curves have provided lots of interesting and useful physical information about studied molecules , going from the detailed elastic properties@xcite to complex structure transitions@xcite .
on theoretical side , many kinds of models have been constructed to character and explain the recorded various efcs of different molecules @xcite . except computer simulations , e.g. , molecular dynamic or monte carlo sampling ,
the calculation of the end - to - end distance distributions ( eedds ) of the force stretched molecules is the center problem in using statistical mechanical method . in principle
, eedds can be obtained by partition function . but
two questions must be faced firstly : one is what physical interactions should be taken into account ; the other is what mathematical technique is needed to solve the eedds .
it is not easy to describe physical interactions in complex molecules , such as polyelectrolytes or proteins . while specific mathematical technique is not always useful in different molecular systems .
in contrast to traditional mind , in this paper we try to extract eedds from the recorded efcs in experiments using the maximum - entropy ( or least - biased ) method ( mem ) .
our motivations are that , first , to our knowledge , little concern about eedds has been given in previous force stretching models .
although eedds of simpler molecules may be simpler enough , it is no reason to assume that they are still simple for complex molecules , such as secondary structure rna ; second , because the eedds are the results of interplay between intra - molecule and force , they can be seen as primitive examining for more realistic physical models .
in addition , our studies also show that the eedd at vanishing force calculated by mem keeps almost all characteristic of the exact distribution without force .
therefore this method may provide a possible way to directly measure " eedds by force spectroscopy .
the organization of this paper is as follows . we first , in sec
. [ memintrod ] , briefly review the maximum entropy method . in sec .
[ momentsefc ] , the basic relations between the distance moments and efcs are demonstrated . in sec .
[ testmem ] , mem is examined by reconstructing eedds of three force stretching chain models : gaussian chain , free - joined chain and self - avoiding chain on 2-dimension lattice .
we also show in sec .
[ applicationrna ] that the method is capable of resolving eedds of complex chain molecules stretched by force . as an illustration , the model of force unzipping double - stranded chain molecules is used to provide exact efcs and eedds@xcite ; these results are necessary to calculate and compare eedds solved by mem . section .
[ conclusion ] is our conclusion .
given a finite set of the moments of a distribution function , how to construct the function is an old mathematical problem .
the mem has been proved to be useful in this problem@xcite . from a normalized distribution function @xmath2 on the interval @xmath3 ,
the power moments are calculated as @xmath4 on the other hand , given a set of @xmath5 moments , from @xmath6 to @xmath7 , it is not always possible to find a positive , well - behave function @xmath2 that will have these moments .
the necessary and sufficient conditions for the existence of a function @xmath2 with a set of @xmath5 moments on interval @xmath3 are the hausdorff relations@xcite : @xmath8 the mem offers a definite procedure for the construction of the approximate distribution @xmath9 based on @xmath5 moments as the following form@xcite : @xmath10.\end{aligned}\ ] ] the @xmath11 are a set of @xmath5 constants determined by the @xmath5 known @xmath12 .
this involves a straightforward nonlinear iterative procedure that usually converges rapidly@xcite .
in general , a real distribution is not always defined on interval @xmath3 .
hence the first step in using mem is to convert the distribution to a function on this interval@xcite . given that the power moments of the original distribution @xmath13 are @xmath14 , and the lower and upper bounds are designated as @xmath15 and @xmath16 respectively .
defining the extent of the distribution @xmath17 first , shift the moments @xmath14 to interval @xmath18 by @xmath19 then scale these moments @xmath20 to interval @xmath3 by @xmath21 thus mem can be used to calculate distribution @xmath2 .
conversely , if the approximate distribution @xmath2 is solved from moments @xmath14 , the distribution can be first rescaled from interval @xmath3 to @xmath18 by the change of variable y @xmath22 then shift the distribution @xmath23 to interval @xmath24 by @xmath25
assuming that one end of a chain consisting of @xmath26-links is fixed at origin , and external force @xmath27 is exerted on the other end , where unit vector @xmath28 is along @xmath29-axis .
let @xmath30 be the probability distribution function that the end - to - end vector of the force stretched chain is @xmath31 .
then the power moments of component @xmath32 distribution @xmath33 are calculated by @xmath34 in order to illustrate expressions more seriously and explicitly , distribution function of the ideal chains is used@xcite @xmath35\int{\cal d}[{\bf r}(s)]\delta^{3 } \left({\bf r}- \int_{0}^{l}ds{\bf v}(s)\right)\nomb\\ & & \times\exp\left[-\frac{1}{k_bt}\int_{0}^{l}ds \rho_e({\bf r},s ) + \frac{f}{k_bt}{\bf z}_0\cdot \int_{0}^{l}ds{\bf v}(s ) \right],\end{aligned}\ ] ] where @xmath36 is boltzmann constant , @xmath37 is temperature , @xmath38 is arclength of the chain , vector " @xmath39 describes the local state at arclength point @xmath40 , e.g. , in the case of a flexible gaussian chain , @xmath41 is a three - dimensional position vector ; while in case of a wormlike chain , @xmath41 is the unit tangent vector@xcite .
vector @xmath42 is also different according to concrete chain model , e.g. , @xmath43 for gaussian chain , and @xmath44 or the tangent vector for wormlike chain .
the normalization factor @xmath45 $ ] is @xmath46&=&\int d{\bf r}\int{\cal d}[{\bf r}(s)]\delta\left({\bf r}- \int_{0}^{l}ds{\bf v}(s)\right)\nomb\\ & & \times\exp\left[-\frac{1}{k_bt}\int_{0}^{l } ds\rho_e({\bf r},s ) + \frac{f}{k_bt}{\bf z}_0\cdot \int_{0}^{l } ds{\bf v}(s)\right]\nomb\\ & = & \int{\cal d}[{\bf r}(s)]\exp\left[-\frac{1}{k_bt}\int_{0}^{l}ds \rho_e({\bf r},s ) + \frac{f}{k_bt}{\bf z}_0\cdot \int_{0}^{l}ds{\bf v}(s ) \right].\end{aligned}\ ] ] replacing eq .
[ eeddfunction ] into eq .
[ momentexp ] and performing @xmath47 integral we have @xmath48\int{\cal d}[{\bf r}(s ) ] \left({\bf z_0}\cdot \int_{0}^{l}ds{\bf v}(s)\right)^m\nomb\nomb\\ & & \times\exp\left[-\frac{1}{k_bt}\int_{0}^{l}ds \rho_e({\bf r},s ) + \frac{f}{k_bt}{\bf z}_0\cdot \int_{0}^{l}ds{\bf v}(s ) \right].\end{aligned}\ ] ] it is easy to prove that eq .
[ eedintegrate ] can be rewritten as @xmath49}\frac{\partial^m}{\partial{f^m}}{{\cal q}[f]}.\end{aligned}\ ] ] the first moment is just the average extension @xmath50 recorded in experiments as a given force @xmath51 , @xmath52}\frac{\partial}{\partial{f}}{\cal q}[f].\end{aligned}\ ] ] we can alternatively relate the partition function @xmath45 $ ] to derivatives of @xmath50 with respect to @xmath51 as follows : @xmath53 and so on , where @xmath54.\end{aligned}\ ] ] according to eq .
[ rzderiveq ] , the moments @xmath55 can be solved in terms of derivatives of @xmath50 with respect to force @xmath51 as follows : @xmath56 the above relations show that if one has the first @xmath0 derivatives of @xmath50 , then this is enough information to calculate the first @xmath1 moments of the distribution @xmath33 . these derivatives of @xmath50 can be obtained by expanding the extension @xmath50 in a taylor series about the reference force @xmath57 as @xmath58 where @xmath59 in general , no analytical @xmath50 is used in real situation ; only the efcs are recorded in experiments .
all the derivatives have to be calculated by numerical methods . before beginning the next section
, we clarify our procedure : first calculate different derivatives of @xmath50 from efcs by numerical method ; then use eq .
[ rzderive ] to obtain necessary power moments of @xmath33 ; and finally , apply mem presented in sec .
[ memintrod ] to construct approximate eedds .
in this section , the mem is examined with three force stretching chain models which have different statistical properties : gaussian chain , free - joined chain and excluded - volume ( ev ) chain on two dimension .
the main reason to choice these model is that their eedds with forces have exact expressions .
eedds and efcs of the models are solved by statistical mechanical methods firstly .
then seeing the obtained efcs as experiment data , approximate distributions is computed by mem according to procedure mentioned in above section .
distributions solved by two methods are compared finally . for each chain model
, eedds at three nonzero forces are calculated respectively . in addition , distributions at zero force are also solved . because that eedd without force is important in polymer research , such as the calculation of root - mean - square end - to - end distance , it is interesting to see whether the distributions calculated by mem at zero force can keep main characteristic of the exact eedd . though our method only solve @xmath60 , the length distribution @xmath61 could be obtained from the numerical relations provided by domb et al .
early@xcite . as the simplest @xmath26-link chain model
, the energy with force @xmath27 in gaussian model@xcite is expressed as @xmath62 where @xmath63 is effective bond length , @xmath41 is position vector . using path integral method@xcite ,
the eedd of gaussian chain stretched by force is derived as @xmath64 correspondingly , the component @xmath32 distribution in @xmath28 direction can be integrated by @xmath65 the extension versus force then is calculated as @xmath66 as an illustration , we choice @xmath67 and plot the function in fig
. [ gaussplotefc ] . this function will be viewed as efcs measured " in experiments . .
the three arrows point out forces in which corresponding eedds are calculated by mem . ] before applying mem , firstly expand eq .
[ gaussefc ] in taylor series about force @xmath57 as @xmath68 or @xmath69 and @xmath70 .
three moments can be obtained at any given force @xmath57 through eq .
[ rzderive ] directly .
then approximate distributions are solved by mem .
three distributions calculated by mem and their comparing with exact eq . [ gaussrzeed ] at forces @xmath71 are shown in fig .[gaussploteed ] .
considering that the extension is linear with force , three moments approximation is used in this model .
obviously , mem can precisely reconstruct distributions of the gaussian chain .
in fact , because the approximation function @xmath72 is just the gaussian distribution , it is not unexpected that mem reconstruct eedds of gaussian chain perfectly .
in addition , eedd at @xmath73 is the same with distributions at nonzero forces , since the first - order derivate of @xmath50 is constant at any force , [ cols="^,^ " , ]
in this paper , contrary to the traditional mind , we calculate end - to - end distance distribution of force stretching chain molecules from the measured efcs by using mem .
because the method is independent of polymer energy formula or structure details , it provide a useful and simple way to detect the real physical information about complex molecules .
many results , such as the important role played by ev interactions , single- or multipeak in eedds can be obtained from the simple efcs .
it is interesting to see whether these results can be found in real stretching experiments . | using the maximum - entropy method , we calculate the end - to - end distance distribution of the force stretched chain from the moments of the distribution , which can be obtained from the extension - force curves recorded in single - molecule experiments .
if one knows force expansion of the extension through the @xmath0th power of force , it is enough information to calculate the @xmath1 moments of the distribution .
we examine the method with three force stretching chain models , gaussian chain , free - joined chain and excluded - volume chain on two - dimension lattice .
the method reconstructs all distributions precisely .
we also apply the method to force stretching complex chain molecules : the hairpin and secondary structure conformations .
we find that the distributions of homogeneous chains of two conformations are very different : there are two independent peaks in hairpin distribution ; while only one peak is observed in the distribution of secondary structure conformations .
our discussion also shows that the end - to - end distance distribution may discover more critical physical information than the simpler extension - force curves can give . |
graphene , a one - atom thick layer of carbon atoms , has been a topic of intense study since it s experimental realization@xcite in 2004 .
interest in graphene results , in particular , from the prediction that carriers in it behave as massless , chiral dirac fermions , moving in a two - dimensional ( 2d ) plane and described by a dirac - type hamiltonian .
this model predicts unusual electronic properties such as the gapless electronic spectrum , the perfect transmission at normal incidence through any potential barrier , i.e. , the klein paradox @xcite which was recently addressed experimentally@xcite , the zitterbewegung recently verified @xcite , etc .
, see ref . for a recent review . motivated by all these properties
, condensed matter physicists started extending the known properties of a two - dimensional electron gas ( 2deg ) in semiconductor materials to the relativistic 2d fermions ( dirac electrons ) in graphene .
another particularly interesting system to consider is the application of a periodic potential to graphene , that is , a superlattice ( sl ) , which under special conditions leads to collimation of electron beams@xcite . as found recently in ref . , using a tight - binding formalism , the dispersion relation for such a sl can reveal extra dirac points at the fermi level @xcite .
close to the fermi level the electronic properties of graphene are well described by the massless 2d dirac equation .
in two recent studies , ref . and , an exact condition was found for the emergence of extra dirac points ( zero modes ) in the presence of a sinusoidal or square - wave sl potential . however , both studies are not able to describe the character and spatial distribution in @xmath0 space of these new dirac points , as they expand the spectrum for small @xmath1 . in this work
we describe under which condition this is possible and also where these extra dirac points arise in the electronic structure of massless dirac fermions in single - layer graphene when a square - wave periodic potential is applied .
further , we analytically investigate the anisotropic renormalization of the group velocities at these new dirac points , and find that the degree of the renormalization depends on the parameter @xmath2 , which is linear in the barrier height @xmath3 and period @xmath4 of the sl , in the sense that , @xmath5 holds for @xmath6 such that an extra dirac point arises while for very high values @xmath6 we have @xmath7 .
moreover , we also consider the case of unequal barrier and well widths , not treated previously , that results in a qualitatively different electronic spectrum .
we organize the paper as follows . in sec .
[ sec2 ] we introduce our model . in sec .
[ sec3 ] we investigate the emergence of the extra dirac points , approximate the implicit dispersion relation for small energies , and take a closer look at the group velocity near the extra dirac points .
further we investigate the influence of the features of the spectrum on the density of states and conductivity .
finally , we make a summary and concluding remarks in sec .
we describe the electronic structure of an infinitely large flat graphene flake by the nearest - neighbour , tight - binding model and consider solutions with energy and wave vector close to the k point .
the relevant hamiltonian is @xmath8 , with @xmath9 the @xmath10 unit matrix .
explicitly @xmath11 is given by @xmath12 where @xmath13 is the momentum operator and @xmath14 the fermi velocity .
in the presence of a one - dimensional ( 1d ) square - wave potential @xmath15 , such as the one shown in fig .
1 , the equation @xmath16 admits solutions of the form @xmath17 with @xmath18 @xmath19^{1/2}$],@xmath20 , @xmath21 , @xmath22 , and @xmath23 ; the parameters @xmath24 and @xmath25 are in units of inverse length .
we consider an infinite number of periodically spaced barriers , as shown in fig .
[ fig2_1 ] , with unit cell length @xmath4 and barrier ( well ) width @xmath26 ( @xmath27 ) .
it is convenient to introduce the dimensionless variables @xmath28 , @xmath29 , @xmath30 , @xmath31 , @xmath32 , @xmath33 , and @xmath34 .
the wave function of this periodic system is a bloch function and the transfer matrix @xmath35 pertinent to it leads to an expression for the dispersion relation , see appendix [ app1 ] . for @xmath36 and @xmath37 as in eq .
( [ eq2a_4 ] ) , the transfer matrix @xmath35 can be written as @xcite @xmath38 then the dispersion relation becomes @xmath39 with @xmath40 given by @xmath41\ ] ] and @xmath42^{1/2 } , \quad \lambda = [ { \varepsilon}_b^2 - k_y^2]^{1/2}. \end{aligned}\ ] ] writing the rhs of eq .
( [ eq2a_2 ] ) explicitly gives @xmath43 although this derivation is only correct for @xmath36 , eq .
( [ eq2a_5 ] ) is also valid beyond this limitation , see ref . .
( [ eq2a_5 ] ) it can be seen that the dispersion relation possesses the symmetry property @xmath44 for @xmath45 .
the asymmetric spectrum is not unexpected because the symmetry of the potential about the fermi level is lost for @xmath46 .
for @xmath47 we have @xmath48 where @xmath49 and @xmath50 .
for this interesting case , the potential possesses particle - hole symmetry and the extra dirac points originate at the fermi level ; we will show their arrangement , in @xmath0 space , in sec .
[ sec3a ] . in fig .
[ fig2](a ) the spectrum resulting from eq .
( [ eq2a_6 ] ) for equal barrier and well widths , i.e. , for @xmath51 , is plotted for @xmath52 . as can be seen ,
the spectrum is symmetric about the fermi level ; there are two extra dirac points on both sides of the main dirac point , and their velocities are renormalized .
the anisotropic behaviour of the new dirac cones is clearer in the projection of the conduction band shown in fig . [
fig2](b ) .
further details about the renormalization of the velocities will be given in sec .
[ sec3c ] . + for unequal barrier and well widths the spectrum is shown in fig . [ fig10](a ) for @xmath53 and @xmath54 .
the spectrum is no longer symmetric about the fermi level , the two extra dirac points are shifted in energy relative to the main point and their velocities are renormalized .
the location of the extra dirac points will be investigated in sec .
in order to find the location of the dirac points we assume @xmath55 , @xmath56 , and @xmath57 in eq .
( [ eq2a_6 ] ) . then eq . ( 8) becomes @xmath58 \,\sin^2 \lambda/2,\ ] ] which has solutions for @xmath59 or @xmath60 .
for the first possibility @xmath61 is the only solution and corresponds to the usual dirac point .
the second possibility leads to @xmath62 with @xmath63 , because @xmath64 makes the denominator @xmath65 vanish and does not lead to a solution .
for @xmath62 we have @xmath66 where we reinserted the dimensions after the second equality sign . as such
, eq . ( [ eq3_3 ] ) describes the spatial arrangement of the extra dirac points along the @xmath1 axis .
also , it clearly shows how many points we have at particular values of @xmath6 , namely @xmath67 , and where they are located in * k * space .
each time @xmath6 becomes a multiple of @xmath68 a new pair of dirac points is generated for @xmath61 .
the condition @xmath63 gives us a threshold value of @xmath69 for the emergence of the first pair .
the integer @xmath70 denotes the @xmath70th extra dirac point , so the outer extra dirac points have @xmath71 as they are generated first . in fig .
[ fig3](a ) we show slices of the sl spectrum along @xmath1 for @xmath55 and @xmath72 .
the solid red , dash - dotted green , and dashed blue dashed curves correspond to barrier widths @xmath73 , and @xmath74 respectively .
the thin black line is the curve on which the extra dirac points , on the left of the main one at @xmath75 , are located for various @xmath26 . in fig .
[ fig3](b ) we show slices of the spectrum along @xmath1 for @xmath55 .
the solid red , dot - dot - dashed black , dashed green , and dash - dotted blue curves are for different values of the barrier height such that @xmath76 , and @xmath77 respectively . for values of @xmath78 which are multiples of @xmath79 ,
new dirac points are generated .
interestingly , if new extra points are to arise , the dispersion becomes almost flat along the @xmath1 axis at the dirac point , i.e , collimation occurs .
we will come back to this issue in sec .
[ sec3c ] . _ unequal well and barrier widths_. we return to the more general case of unequal well and barrier widths for which @xmath46 .
it is more difficult to locate the extra dirac points which no longer occur at the fermi level as seen from the green and blue curves in fig .
[ fig3](a ) showing slices of spectra from eq .
( [ eq2a_5 ] ) for @xmath55 . by means of the symmetry @xmath44 for @xmath45 , we know the complementary plots for @xmath80 . as can be seen , the extra dirac points shift mainly down ( up ) in energy as @xmath26 increases ( decreases ) . to find their coordinates @xmath81 we assume @xmath82=@xmath83 and @xmath84@xcite .
this gives ( appendix [ app2 ] ) @xmath85^{1/2 } , \end{aligned}\ ] ] where @xmath70 and @xmath86 are integers .
this method also shows higher and lower crossing points if @xmath87 . in fig .
[ fig3](a ) the extra dirac points on the left , obtained with this method , are indicated by open circles and the thin black curve shows their trajectory in @xmath88 space as the width @xmath26 varies . for a particular @xmath6
there is a minimal width @xmath26 ( and a corresponding maximal width @xmath80 ) below ( above ) which the various extra dirac points disappear . in fig .
[ fig3](a ) the `` @xmath89dirac cones '' at these crossing points for @xmath90 are not only reshaped with a renormalized anisotropic velocity but , as shown by the blue dashed curve , the `` @xmath89extra dirac point '' is not at a local minimum ( maximum ) of the conduction ( valence ) band . as the purpose is to have a closer look at the behaviour of the extra dirac points and we can not prohibit @xmath1 from being large , we expand eq .
( [ eq2a_6 ] ) for small energies , up to second order in @xmath24 , and obtain the following explicit dispersion relation @xmath91}{k_y^4 a \sin a + a^2 u^4/16 - 2 k_y^2 u^2 \sin^2 ( a/2 ) } } \right]^{1/2},\ ] ] with @xmath92^{1/2}$ ] .
if we only need the behaviour of the spectrum near the k point ( for small @xmath93 and @xmath1 ) , it suffices to make an expansion for small @xmath24 and @xmath1 in eq .
( 8) , up to third order in products of @xmath24 and @xmath1 since this is the first order with an energy dependence .
the result is @xmath94 then we solve for the energy @xmath24 and obtain @xmath95^{1/2}.\ ] ] in fig .
[ fig6](a ) we show @xmath24 from eq .
( [ eq6_1 ] ) and compare it with the exact dispersion relation , for @xmath55 , following from eq .
( [ eq2a_6 ] ) .
the expansion ( [ eq6_1 ] ) is rather good for low energies near the extra dirac points ; accordingly , we will use eq .
( [ eq6_1 ] ) to further assess their behaviour .
the spectrum in the low - energy range consists of two kinds of valleys , one near the main dirac point and the other near the extra dirac points . near the original dirac point
the spectrum is almost linear , perpendicular to the barriers , and zero parallel to them , whereas near the extra dirac points the situation can be reversed depending on the height of the barriers .
_ group velocity along the @xmath96-axis at the dirac point .
_ to compare with the collimation found by park _
et al._@xcite , we notice that in fig .
[ fig3](b ) , for the solid red , dashed green , and dash - dotted blue curves , corresponding to barrier heights which are multiples of @xmath68 , the dispersion becomes more flat for small @xmath1 . from eq .
( [ eq5_2 ] ) we could already expect that , to order @xmath97 , the @xmath1 dependence disappears for these values of @xmath6 .
further , if we expand eq .
( [ eq6_1 ] ) in powers of @xmath1 we obtain @xmath98 k_y^3 + o(k_y^5 ) , \end{split}\ ] ] which is linear in @xmath1 , for small @xmath1 , and the velocity becomes @xmath99 in fig .
[ fig6f ] the velocities of the dirac point , in the @xmath96 and @xmath100 directions , are given by the @xmath101 curves .
for @xmath102 we have @xmath103 which is cubic in @xmath1 for small @xmath1 . if @xmath70 and consequently @xmath6 become larger the dispersion gets flatter . in fig .
[ fig6](b ) we plot eq .
( [ eq_approxfig6 ] ) for @xmath69 and @xmath104 as dashed curves , which correspond , respectively , to the zoomed - in plots of the solid red and dashed green curves of fig .
[ fig3](b ) , shown here as blue and red curves .
_ group velocity along the @xmath100-axis at the extra dirac points .
_ the dispersion relation ( [ eq6_1 ] ) for the @xmath1 values of the extra dirac points , determined by @xmath105^{1/2}$ ] , gives us an idea of how dispersionless the spectrum near these points is along the @xmath96-direction .
if @xmath106 exists , eq .
( [ eq6_1 ] ) becomes @xmath107 and the partial derivative@xcite of @xmath24 with respect to @xmath93 is @xmath108 this means that for smaller @xmath70 ( the most distant extra dirac points ) the group velocity along the @xmath96-direction is strongly suppressed .
further , as @xmath109 must hold in order for @xmath110 to be real , @xmath111 is smaller than @xmath112 at @xmath55 . only for the special values @xmath113 , for which new dirac points appear
, we have @xmath114 .
meanwhile the dispersion in the @xmath1 direction is also of interest .
first , let us take @xmath55 and expand the dispersion relation ( [ eq6_1 ] ) for @xmath115 .
to first order in this difference we obtain @xmath116 ( k_y - { k_y}_{j,\pm}).\ ] ] this gives the velocity @xmath117 at the extra dirac points @xmath118
/ u^2.\ ] ] since the coordinates of the extra dirac points should be real , @xmath119 is positive and smaller than @xmath120 and we have @xmath121 ( the outer dirac points , for @xmath71 , show the largest @xmath117 ) .
this entails that both @xmath122 and @xmath117 are renormalized at the new dirac points . a plot of the velocities of the extra dirac points , in the @xmath96 and @xmath100 directions , given by eqs ( [ eq_vx ] ) and ( [ eq_vy ] ) , is shown in fig .
[ fig6f ] .
as seen , for the extra dirac points , @xmath123 , shown by the dashed red curves , starts from @xmath124 and decreases to zero with increasing @xmath6 while @xmath125 ( dash - dotted blue curves ) , starts from zero and approaches @xmath124 slowly for large @xmath6 .
the thin solid black curves show the velocities for @xmath71 , obtained numerically from eq .
( [ eq2a_6 ] ) .
as can be seen , the two curves match almost perfectly . _ density of states .
_ at zero temperature the density of states ( dos ) d(e ) is given by @xmath126 with @xmath127 the energy .
we show the dos in fig .
[ fig5 ] , for @xmath128 ( solid red curve ) and @xmath129 ( dashed blue curve ) , as well as the dos for graphene without any sl potential ( dash - dotted black curve ) ; the latter is given by @xmath130 , with @xmath131 the amount of states per unit area and
l is the period of the sl .
the dos shows an oscillating behaviour .
the dips in it are located at the crossing points in the energy bands for @xmath61 ( @xmath132 ) , while the peaks marked by a star are ascribed to the saddle points between the crossing points for @xmath133 and to the minima of the energy bands at the edge of the brillouin zone , @xmath134 , marked by a cross . for
@xmath135 the dos ( dashed blue curve ) does not vanish at @xmath56 nor is it symmetric about this energy . , and @xmath128 and @xmath136 is shown by , respectively , the solid red and dashed blue curve .
stars and crosses placed near the peaks of the solid red curve ( for negative energies ) are ascribed , respectively , to saddle points in the spectrum , for @xmath55 , and to minima for @xmath137 .
the dos without a sl potential is shown by the dash - dotted black curve.,height=151 ] _ conductivity tensor @xmath138 . _
the diffusive dc conductivity @xmath139 for the sl system can be readily calculated from the spectrum if we assume a nearly constant relaxation time @xmath140 .
it is given by@xcite @xmath141 with @xmath142 the area of the system , @xmath143 the energy band index , @xmath144 and @xmath145 $ ] the equilibrium fermi - dirac distribution function ; @xmath146 is the fermi energy and @xmath147 . in figs . [ fig8](a ) and [ fig8](b ) we show , respectively , @xmath148 and @xmath149 for a sl with @xmath54 , and the temperature dependence is given by @xmath150 ( in dimensionless units ) .
the solid red and dashed blue curves correspond to @xmath128 and @xmath129 , respectively .
the dash - dotted black curve shows the conductivity at zero temperature and in the absence of a sl potential , @xmath151 with @xmath152 and @xmath153 .
+ notice that @xmath148 is an oscillating function of the fermi level and recovers a quasi linear behaviour similar to that in graphene without a sl potential when @xmath154 is well above the barrier height , as shown in the inset , i.e. with the x axis displaced over the potential height or well depth , i.e. , @xmath155 . on the average @xmath149 increases with @xmath154 and approaches the result without a sl for large energies . the oscillations in both @xmath148 and @xmath149 result from the motion of the fermi level through the different sl minibands .
notice that for @xmath129 the conductivities are asymmetric with respect to electron and hole conduction .
in both cases , @xmath156 and @xmath129 , @xmath148 shows dips at @xmath157 , where energy band crossings occur in the spectrum for @xmath61 . in the former case
the dos has dips occurring at the same energy values that are dominated by the same crossings . in the latter case
we see that , unlike the dos , @xmath148 is almost unaffected by the extra dirac points for low energies since the spectrum is almost flat near these points .
similarly , for @xmath129 we see that the minimum in @xmath149 is located at @xmath158 , that is , the energy value for which the two extra dirac points occur in the spectrum . in figs .
[ fig8](c ) and [ fig8](d ) we show , respectively , @xmath148 and @xmath149 , for a sl with @xmath156 , for different potential heights , such that @xmath159 , and @xmath150 .
notice that the conductivity @xmath149 , in the low - energy range ( @xmath160 ) , is lower than that in the absence of a sl potential while its slope increases as the potential barriers become higher .
this is due to the extra dirac points , that appear for larger potential heights , near which the velocity is larger along the @xmath100-axis .
notice that for @xmath160 we have @xmath161 as a result of the inequality @xmath5 near the dirac point .
we investigated the appearance of zero modes , touching points at the fermi level ( extra dirac points ) in the spectrum of single - layer graphene in the presence of a 1d superlattice ( sl ) .
the system was described by a dirac - type hamiltonian , and the sl barriers were square . in the general case of unequal well and barrier widths , there is no particle - hole symmetry and the extra dirac points are no longer located at the fermi level .
we obtained an analytical expression for the position of the crossing points in the spectrum .
the extra `` @xmath89dirac cones '' that appear at the various crossing points are reshaped , i.e. , they are no longer circular symmetric and the slope is renormalized . for fixed height of the barriers
, we found lower and upper bounds for the barrier and well widths for the occurrence of these extra dirac cones . for a sl with equal well and barrier widths we complemented the investigations of refs . and , which numerically demonstrated the emergence of extra dirac points ( zero modes ) .
in doing so we found a simple analytical expression for the spatial distribution of these points in @xmath0 space as well as a threshold value of the potential strength for their appearance .
further , we approximated the dispersion relation for energies close to the fermi energy and found an explicit expression for the @xmath0 space behaviour of the extra dirac points at the fermi level . using this expression we showed how the group velocities at the various extra dirac points are renormalized in the @xmath96 and @xmath100 directions .
we also quantified how dispersionless the spectrum is in the neighbourhood of a dirac point along the y - direction and the emergence of new points at which the conduction and the valence bands touch each other .
finally , we obtained numerically the density of states ( dos ) , which exhibits an interesting oscillatory behaviour and is reflected in the conductivity of the system .
we found that the dips in the dos , for symmetric sls , are located at the touching points in the spectrum for @xmath162 , i.e. , for @xmath132 . for asymmetric sls
these dips persist but extra dips due to the extra dirac points arise .
the conductivity @xmath148 was found to have dips at the same values for @xmath154 as the dos while the main features of @xmath149 in the low - energy range are due to the spectrum near the extra dirac points .
we notice in passing that the influence of velocity renormalization on transport was not studied in refs . and nor the modification of the extra dirac points for unequal well and barrier widths .
this work was supported by imec , the flemish science foundation ( fwo - vl ) , the belgian science policy ( iap ) , the brazilian council for research ( cnpq ) , and the canadian nserc grant no . ogp0121756 .
the wave functions in the regions before and after the barrier , labeled , respectively , by @xmath71 and @xmath163 , can be written as @xmath164 , with @xmath165 since the wave function of the entire periodic system is a bloch function and the transfer matrix @xmath35 connects the regions before and after the barriers , we have @xmath166 with @xmath93 the bloch wave vector . from these boundary conditions we extract the relation ( @xmath167^{1/2}$ ] )
@xmath168 for nontrivial solutions of eq .
( [ eqapp1_3 ] ) , the determinant of @xmath169 must be zero , i. e. , @xmath170 evaluating the determinant gives the dispersion relation @xmath171 with @xmath172 .
suppose a solution @xmath173 of the dispersion relation ( [ eq2a_5 ] ) is known for which the derivative @xmath174 at a certain @xmath1 value is undefined ; then this @xmath1 value can be a crossing point .
the condition for such a solution is @xmath175 , and @xmath84 , which entails @xmath176 with @xmath70 and @xmath86 integers .
explicitly we obtain @xmath177w_w^2 = ( j \pi)^2,\\ & \big[({\varepsilon}- u ( 1-w_b))^2 - k_y^2\big]w_b^2 = ( ( j + 2 m ) \pi)^2 . \end{aligned}\ ] ] subtracting the second equation from the first one in ( b2 ) gives @xmath178 from which the corresponding value of the energy @xmath24 can be extracted . substituting this value in the first of eqs .
( [ eqapp2_2 ] ) one obtains @xmath179^{1/2}. \end{aligned}\ ] ] 10 k. s. novoselov , a. k. geim , s. v. morozov , d. jiang , y. zhang , s. v. dubonos , i. v. grigorieva , and a. a. firsov , science * 306 * , 666 ( 2004 ) . o. klein , z. phys . * 53 * , 157 ( 1929 ) .
m. i. katsnelson , k. s. novoselov , and a. k. geim , nature physics * 2 * , 620 ( 2006 ) ; j. m. pereira jr . , p. vasilopoulos , and f. m. peeters , appl .
. lett . * 90 * , 132122 ( 2007 ) .
n. stander , b. huard , and d. goldhaber - gordon , phys .
* 102 * , 026807 ( 2009 ) ; a. f. young and p. kim , nature phys . * 5 * , 222 ( 2009 ) .
j. schliemann , d. loss , and r.m .
westervelt , phys .
lett . * 94 * 206801 ( 2005 ) ; w. zawadzki , phys . rev .
b * 72 * 085217 ( 2005 ) ; r. winkler , u. zulicke , and j. bolte , phys .
b * 75 * 205314 ( 2007 )
. a. h. castro neto , f. guinea , n. m. r. peres , k. s. novoselov , and a. k. geim , rev .
* 81 * , 109 ( 2009 ) .
park , l. yang , y .- w .
son , m. l. cohen , and s. g. louie , nature phys . * 4 * , 870 ( 2008 ) .
park , y .- w .
son , l. yang , m. l. cohen , and s. g. louie , nano lett . *
8 * , 2920 ( 2008 ) .
y. p. bliokh , v. freilikher , s. savelev , and f. nori , phys .
b * 79 * , 075123 ( 2009 ) .
j. h. ho , y. h. chiu , s. j. tsai , and m. f. lin , phys .
b * 79 * , 115427 ( 2009 ) . in ref .
9 a nearest - neighbour , tight - binding method was applied with a large rectangular super - cell along a sinusoidal potential as basis .
extra touching points at the fermi level were observed and , together with the group velocity in their vicinity , their influence on the density of states was demonstrated .
h . park , y .- w .
son , l. yang , m. l. cohen , and s. g. louie , phys . rev . lett . * 103 * , 046808 ( 2009 ) .
l. brey and h. a. fertig , phys .
103 * , 046809 ( 2009 ) .
d. j. griffiths and c. a. steinke , am .
j. phys . , * 69 * , 137 ( 2001 ) .
m. barbier , f. m. peeters , p. vasilopoulos , and j. m. pereira jr .
b * 77 * , 115446 ( 2008 ) . the implicit function theorem can not be applied if the gradient ( jacobian ) of the dispersion relation is zero .
this occurs because of the crossing of the bands at these points : from the form of the derivatives , too long to write down , we can see that they are both zero if @xmath180 while @xmath181 . m. charbonneau , k. m. van vliet , and p. vasilopoulos , j. math . phys .
* 23 * , 318 ( 1982 ) . | we investigate the emergence of extra dirac points in the electronic structure of a periodically spaced barrier system , i.e. , a superlattice , on single - layer graphene , using a dirac - type hamiltonian . using square barriers allows us to find analytic expressions for the occurrence and location of these new dirac points in * k * space and for the renormalization of the electron velocity near them in the low - energy range . in the general case of unequal barrier and
well widths the new dirac points move away from the fermi level and for given heights of the potential barriers there is a minimum and maximum barrier width outside of which the new dirac points disappear .
the effect of these extra dirac points on the density of states and on the conductivity is investigated . |
this talk is based on ref .
@xcite and on earlier work with stan brodsky , summarized in ref .
i would like to thank the organizers of the trento international light - cone workshop for their hospitality and for setting up such a well - organized meeting .
the research of one of m.k . was supported in part by a grant from the united states - israel binational science foundation ( bsf ) , jerusalem , israel , and by the basic research foundation administered by the israel academy of sciences and humanities .
j. r. ellis and m. karliner , arxiv : hep - ph/0108259 .
m. karliner , _ `` a possible resolution of the @xmath114 puzzle '' , _ hep - ph/0108106 , invited talk at @xmath0 _ physics at intermediate energies _ ,
slac , 30 apr 2 may 2001 .
a. antonelli _ et al .
_ , nucl . phys .
b * 517 * , 3 ( 1998 ) .
a. antonelli _ et al . _ ,
b * 334 * , 431 ( 1994 ) .
d. bisello _ et al . _
[ dm2 collaboration ] , z. phys .
c * 48 * , 23 ( 1990 ) .
g. bardin _ et al .
_ , nucl .
b * 411 * , 3 ( 1994 ) . v. l. chernyak and a. r. zhitnitsky , phys .
rept . * 112 * , 173 ( 1984 ) .
, see also t. h. skyrme , proc .
a * 260 * ( 1961 ) 127 .
t. h. skyrme , nucl .
* 31 * ( 1962 ) 556 .
e. witten , nucl .
b * 160 * , 57 ( 1979 ) .
g. s. adkins , c. r. nappi and e. witten , nucl .
b * 228 * , 552 ( 1983 ) .
i. zahed and g. e. brown , phys .
rept . * 142 * , 1 ( 1986 ) .
j. schechter and h. weigel , hep - ph/9907554 .
j. j. verbaarschot , t. s. walhout , j. wambach and h. w. wyld , nucl .
a * 461 * , 603 ( 1987 ) .
h. m. sommermann , r. seki , s. larson and s. e. koonin , phys .
d * 45 * , 4303 ( 1992 ) .
y. lu and r. d. amado , phys .
c * 52 * , 2158 ( 1995 ) [ nucl - th/9505021 ] , m. a. halasz and r. d. amado , phys .
d * 63 * , 054020 ( 2001 ) [ hep - ph/0009260 ] and references therein .
r. d. amado , private communication .
s. anderson _
[ cleo collaboration ] , phys .
d * 56 * , 2485 ( 1997 ) [ hep - ex/9701013 ] .
m. artuso _ et al . _
[ cleo collaboration ] , phys .
d * 50 * , 5484 ( 1994 ) .
l3 collaboration , note 2566 , _
@xmath115 formation in two - photon collisions at lep _
, june 2000 , submitted to ichep-2000 in osaka . | we discuss the puzzling experimental results on baryon - antibaryon production in @xmath0 annihilation close to the threshold , in particular the fact that @xmath1 .
we discuss an interpretation in terms of a two - step process , via an intermediate coherent isovector state serving as an intermediary between @xmath0 and the baryon - antibaryon system .
we provide evidence that the isovector channel dominates both @xmath2 and from @xmath3 annihilation at rest , and show that the observed ratio of @xmath4 can be understood quantitatively in this picture .
experimental data from the fenice collaboration @xcite indicate that @xmath5 is relatively large close to threshold .
their data may be compared with earlier data on @xmath6 @xcite and also with data on the time - reversed reaction @xmath7 , for which more precise data are available @xcite . as seen in fig .
[ fig : nnbar ] , the combined data indicate that @xmath8 when @xmath9 gev .
averaging over the available data on both the direct and time - reversed reactions , which are very consistent , and ignoring any possible variation with energy , we find : @xmath10 in other words , _ close to the @xmath11 threshold , the timelike form factor of the proton is somewhat smaller than that of the neutron ! _ the fact that this ratio is less than unity requires confirmation , but even equal cross sections for @xmath12 and @xmath13 would be quite surprising , in view of the the conventional perturbative picture of baryon - antibaryon production in @xmath0 annihilation , as shown in fig .
[ ee - fig ] .
we recall that the ratio of the cross sections for the corresponding @xmath14-channel processes @xmath15 should be infinite at zero momentum transfer , where the form factor simply measures the total proton and neutron charges , corresponding to a coherent sum over the electromagnetic charges of their constituent quarks .
it is also believed that the ratio ( [ exptalratio ] ) should be large at high momentum transfers . in a naive perturbative description of @xmath0 annihilation into baryons , the virtual time - like
photon first makes a ` primary ' @xmath16 pair , which is then dressed by two additional quark - antiquark pairs that pop out of the vacuum .
this dressing is thought to be a perturbative qcd process at high momentum transfers , which does not distinguish between the @xmath17 and @xmath18 quarks , since gluon couplings are flavor - blind .
thus , in this conventional perturbative picture , the only difference between the production rates of proton and neutron is through the different electric charges of the primary @xmath16 pairs .
the total perturbative cross section is obtained by superposing the amplitudes with different primary @xmath16 pairs and squaring the result : @xmath19 where @xmath20 denotes the amplitude at @xmath21 for making the baryon @xmath22 with a given primary flavor @xmath23 , which is determined by the baryon wave functions .
since the wave functions of the baryon octet have a mixed symmetry , the amplitudes @xmath20 tend to be highly asymmetric in specific models .
for example , in the chernyak - zhitnitsky proton wave function @xcite , the @xmath17 quark dominates , i.e. , @xmath24 , @xmath25 and similarly @xmath26 , @xmath27 .
in such a limiting case we have @xmath28 while this is an extreme case , on general grounds we expect that the @xmath17 contribution dominates in the proton and the @xmath18 in the neutron , so that @xmath29 at large momentum transfers .
we find it puzzling that the experimental ratio ( [ exptalratio ] ) is apparently below unity when @xmath30 gev@xmath31 , whereas the ratio should be much larger than unity at both larger ( timelike ) and smaller ( spacelike ) momentum transfers . clearly , the mechanism at work here is qualitatively different from those responsible for the above intuition .
the lack of a conventional theoretical explanation is part of the motivation for the proposed new asymmetrical @xmath32 high - statistics collider at slac for the regime @xmath33 gev @xcite .
this machine will yield high - precision data on baryon production in @xmath32 annihilation at threshold , providing a check on the fenice data and an accurate benchmark for testing possible theoretical explanations .
the first thing one must realize is that even though @xmath34 , the process is highly nonperturbative .
this is because the ` extra ' kinetic energy available to the quarks is very small .
our approach @xcite to this puzzle is based on thinking about the time - reversed processes : @xmath35 .
these may be viewed as two - step processes , with a coherent meson state serving as an intermediate state , as shown schematically in fig .
[ pions1-fig ] .
one possible motivation for this picture might be provided by the skyrme model @xcite , according to which baryons appear as solitons in a purely bosonic chiral lagrangian .
this model is formally justified as a low - energy approximation to large-@xmath36 qcd @xcite , and is known to provide a good description of many low - energy properties of baryons : see @xcite for reviews .
skyrmion - anti - skyrmion annihilation provides @xcite-@xcite a fairly accurate description of low - energy baryon - antibaryon annihilation . just after the skyrmion and anti - skyrmion touch
, they ` unravel ' each other , and a coherent classical pion wave emerges as a burst that takes away energy and baryon number as quickly as causality permits .
a specific parametrization of the initial pion configuration is @xcite : @xmath37 where @xmath38 is the profile of the chiral field , @xmath39 $ ] , @xmath40 is a range parameter , @xmath41 is chosen so that the total energy is that of the @xmath11 pair , and the form of @xmath38 guarantees that the pion configuration has zero net baryon number .
this crude model has been shown @xcite to reproduce satisfactorily the inclusive single - pion spectrum in @xmath42 annihilation at rest and the branching ratios for multi - pion final states .
the details of this specific configuration are unimportant for our purposes : what is important is that the data are not inconsistent with such a model .
indeed , although the skyrme model provides some motivation for our approach , it is not even essential for our purpose .
what is important is that _ a single intermediate state should dominate _ the two - step @xmath35 process .
this could , for example , equally well be a single intermediate @xmath43 resonant meson state . to be more precise ,
since @xmath11 annihilation is a strong - interaction process , one must consider separately the @xmath44 and @xmath45 channels .
accurately stated , our key assumption is that both of these channels are dominated by single states .
these might be some excited @xmath46 and @xmath47 mesons , for example , just as well as coherent pion configurations ( [ amado ] ) with @xmath44 and @xmath45 . with this picture in mind
, we write the @xmath48 @xmath35 annihilation amplitudes as @xmath49 , where the overall phase is irrelevant , @xmath50 and @xmath51 are relatively real , and @xmath52 is the relative phase between the @xmath44 and @xmath45 amplitudes .
we then have @xmath53 it is apparent from ( [ ratio ] ) that @xmath54 if either @xmath55 or _
vice versa_. remarkably , _ there is evidence from both @xmath56 and @xmath11 annihilations that @xmath57 final states dominate by large factors_. the clearest evidence comes from @xmath58 , where it is found by measuring final states with even and odd numbers of pions respectively that @xmath59 as seen in fig . [
fig : pions ] .. the cross section in fig .
[ fig : pions ] corresponds to the final state @xmath60 ; for the total @xmath61 one should multiply it by 1.5 @xcite .
] at these energies , we expect most final states created by @xmath62 to contain @xmath63 pairs , so that ( [ eeratio ] ) corresponds to the non-@xmath64 initial states we expect to dominate in @xmath11 annihilation
. the value ( [ eeratio ] ) is similar to that found at lower energies , where @xmath65 , in agreement with naive quark models .
the fact that the ratio @xmath66 continues to be large at higher energies is consistent with ideas of generalized vector meson dominance .
the data from @xmath67 annihilations are less clear .
theoretically , there are various calculations and other suggestions that the cross sections for @xmath67 annihilations into @xmath44 and @xmath45 may be similar . experimentally , several initial states contribute , including @xmath68s@xmath69 , @xmath70s / d@xmath71 and various p waves , but we are interested only in the @xmath43 @xmath70s / d@xmath71 initial states .
it is in principle possible to distinguish different initial states by comparing annihilations in gas and liquid , as has been done in the analysis of ozi - violating final states , but we are unaware of a comparable analysis of multi - pion final states . because the initial state is a mixture with different @xmath72 , it is not possible to separate @xmath44 from @xmath45 simply by counting pions , as was the case in @xmath56 annihilation .
the most convincing experimental information known to us comes from an analysis of @xmath73 . by comparing the rates for @xmath74 and @xmath75 , it has been possible to extract @xcite @xmath76 which is comparable to the corresponding ratio ( [ eeratio ] ) in @xmath77 annihilation
the fact that @xmath44 dominates over @xmath45 in the ratio ( [ nnbarratio ] ) is consistent with the hypothesis that most of the @xmath78 final states are created by @xmath79 and @xmath80 pairs in the initial @xmath67 state , with a @xmath64 pair popping out of the vacuum .
the ratio ( [ nnbarratio ] ) would be small if primary @xmath64 pairs dominated .
we now use the experimental information on the dominance by the @xmath44 channel in both @xmath56 ( [ eeratio ] ) and @xmath70s / d@xmath71 annihilation ( [ nnbarratio ] ) in a quantitative analysis of the @xmath81 production @xmath82 ratio ( [ ratio ] ) . defining @xmath83 ,
we can rewrite ( [ ratio ] ) as @xmath84 it is clear that the zero - momentum - transfer limit @xmath85 is obtained in the limit @xmath86 , and that the high - momentum - transfer limit @xmath87 is obtained in the limit @xmath88 . in order to estimate @xmath89 , our assumption of dominance in each isospin channel by a single state ( either a coherent multi - pion state ( [ amado ] ) or a generalized vector meson @xmath90 ) tells us that @xmath91 inserting the experimental indications ( [ eeratio],[nnbarratio ] ) into ( [ factorization ] ) , we estimate that @xmath92 the top end of this range seems to us quite conservative , whereas the lower end surely requires more justification from @xmath11 annihilation data . in the following numerical analysis
, we keep @xmath93 general , but focus extra attention on the limits @xmath94 and @xmath95 .
it is apparent from ( [ newratio ] ) that @xmath96 is possible for any value of @xmath93 , for a restricted range of the relative phase @xmath97 .
however , the allowed range of @xmath52 is extended if @xmath93 is small .
it is easy to see that @xmath82 lies in a narrow range @xmath98 around unity if @xmath52 falls within the following range : @xmath99 it is apparent that @xmath96 for all values of @xmath52 if @xmath93 is small , as suggested ( [ epsilon ] ) by the available data on @xmath56 and @xmath11 annihilation .
the quantitative behaviour of @xmath82 as a function of @xmath100 and @xmath101 is shown in fig .
[ fig : threed ] . displayed explicitly is the region of the @xmath102 plane where @xmath82 falls within the experimental range ( [ exptalratio ] ) .
we see that this range favours @xmath103 , whatever the value of @xmath93 .
fig .
[ fig : twod ] displays projections of fig .
[ fig : threed ] for the two limiting values @xmath94 and @xmath95 . the allowed range ( [ exptalratio ] ) of @xmath82 and the corresponding ranges of @xmath52
are also shown .
we conclude that the _ a priori _ puzzling large experimental value of the ratio @xmath104 can be understood qualitatively .
this is relatively easy if the @xmath105 amplitude dominates over the @xmath45 amplitude , as suggested by the available data on @xmath56 and @xmath11 annihilation and our assumption of dominance by a single coherent state in each isospin channel .
the specific range ( [ exptalratio ] ) can be understood quantitatively if the @xmath44 and @xmath45 amplitude have a large relative phase @xmath52 .
we are not in a position to judge the plausibility of such a large value of @xmath52 , from either an experimental or a theoretical point of view .
it would be interesting to make tests of this possibility .
the disagreement between the naive theoretical prediction and experiment is striking again .
we comment finally on the surprisingly large value of the ratio @xmath106 observed by the cleo collaboration @xcite ( see also @xcite-@xcite for related experimental work ) . as shown in fig .
[ cleo - gg - fig ] , cleo find that close to threshold , which seems analogous to fenice result for the @xmath107 ratio . as illustrated by fig .
[ gg - fig ] , for a given quark flavor , the perturbative amplitude for baryon - antibaryon production in the photon - photon reaction scales like the quark charge squared , compared with the linear dependence of the amplitudes on the quark charge in the @xmath0 case discussed earlier .
thus one might naively expect the ratio @xmath108 to be even smaller than the corresponding perturbative prediction for @xmath109 in @xmath0
. it would be interesting to approach this puzzle from a point of view similar to that adopted in this paper .
however , the situation in @xmath110 collisions is more complicated , because of the wider range of possible spin and isospin states . also , the information available on the isospin and spin decomposition is sparse compared with that in @xmath0 annihilation , which we used above .
data for @xmath111 close to threshold might cast light on the @xmath112 puzzle , but are not yet available . in ref .
@xcite we have proposed a simple model that is able to accommodate the surprisingly large observed value of the ratio @xmath113 .
our suggestion is based on a simple two - step approach , in which a single intermediate state with @xmath44 dominates over @xmath45 .
this dominant intermediate state could be motivated by a skyrmion - anti - skyrmion picture , or could be some excited @xmath46 resonance .
our model could be tested by further measurements of the ratios of different isospin amplitudes in @xmath56 and @xmath11 annihilation , and suggests a relatively large phase difference between @xmath44 and @xmath45 amplitudes .
we look forward to more experimental data bearing on these issues , for example from a new low - energy @xmath56 collider @xcite . |
the problem of `` scaling up for high dimensional data and high speed data streams '' is among the `` ten challenging problems in data mining research''@xcite .
this paper is devoted to estimating entropy of data streams using a recent algorithm called _ compressed counting ( cc ) _
this work has four components : ( 1 ) the theoretical analysis of entropies , ( 2 ) a much improved estimator for cc , ( 3 ) the bias and variance in estimating entropy , and ( 4 ) an empirical study using web crawl data .
while traditional data mining algorithms often assume static data , in reality , data are often constantly updated .
mining data streams@xcite in ( e.g. , ) 100 tb scale databases has become an important area of research , e.g. , @xcite , as network data can easily reach that scale@xcite .
search engines are a typical source of data streams@xcite .
we consider the _ turnstile _ stream model@xcite .
the input stream @xmath2 , @xmath3 $ ] arriving sequentially describes the underlying signal @xmath4 , meaning @xmath5 = a_{t-1}[i_t ] + i_t,\end{aligned}\ ] ] where the increment @xmath6 can be either positive ( insertion ) or negative ( deletion ) .
for example , in an online store , @xmath7 $ ] may record the total number of items that user @xmath8 has ordered up to time @xmath9 and @xmath6 denotes the number of items that this user orders ( @xmath10 ) or cancels ( @xmath11 ) at @xmath12 . if each user is identified by the ip address , then potentially @xmath13 .
it is often reasonable to assume @xmath14\geq 0 $ ] , although @xmath6 may be either negative or positive . restricting @xmath14\geq 0 $ ] results in the _ strict - turnstile
_ model , which suffices for describing almost all natural phenomena .
for example , in an online store , it is not possible to cancel orders that do not exist . *
compressed counting ( cc ) * assumes a _ relaxed strict - turnstile _
model by only enforcing @xmath14\geq0 $ ] at the @xmath12 one cares about . at other times
@xmath15 , @xmath16 $ ] can be arbitrary .
the @xmath0th frequency moment is a fundamental statistic : @xmath17^\alpha .
\end{aligned}\ ] ] when @xmath18 , @xmath19 is the sum of the stream .
it is obvious that one can compute @xmath19 exactly and trivially using a simple counter , because @xmath20 = \sum_{s=0}^t i_s$ ] .
@xmath21 is basically a histogram and we can view @xmath22}{\sum_{i=1}^d a_t[i ] } = \frac{a_t[i]}{f_{(1)}}\end{aligned}\ ] ] as probabilities .
a useful ( e.g. , in web and networks@xcite and neural comptutations@xcite ) summary statistic is * shannon entropy * @xmath23}{f_{(1)}}\log \frac{a_t[i]}{f_{(1 ) } } = -\sum_{i=1}^d p_i\log p_i.\end{aligned}\ ] ] various generalizations of the shannon entropy exist .
the rnyi entropy@xcite , denoted by @xmath24 , is defined as @xmath25^\alpha}{\left(\sum_{i=1}^d a_t[i]\right)^\alpha } = -\frac{1}{\alpha-1}\log \sum_{i=1}^dp_i^\alpha\end{aligned}\ ] ] the tsallis entropy@xcite , denoted by @xmath26 , is defined as , @xmath27 as @xmath1 , both rnyi entropy and tsallis entropy converge to shannon entropy : @xmath28 thus , both rnyi entropy and tsallis entropy can be computed from the @xmath0th frequency moment ; and one can approximate shannon entropy from either @xmath24 or @xmath26 by using @xmath29 .
several studies@xcite ) used this idea to approximate shannon entropy .
network traffic is a typical example of high - rate data streams .
an effective and reliable measurement of network traffic in real - time is crucial for anomaly detection and network diagnosis ; and one such measurement metric is shannon entropy@xcite .
the _ turnstile _ data stream model ( [ eqn_turnstile ] ) is naturally suitable for describing network traffic , especially when the goal is to characterize the statistical distribution of the traffic . in its empirical form
, a statistical distribution is described by histograms , @xmath14 $ ] , @xmath30 to @xmath31 .
it is possible that @xmath32 ( ipv6 ) if one is interested in measuring the traffic streams of unique source or destination .
the distributed denial of service ( * ddos * ) attack is a representative example of network anomalies .
a ddos attack attempts to make computers unavailable to intended users , either by forcing users to reset the computers or by exhausting the resources of service - hosting sites . for example , hackers may maliciously saturate the victim machines by sending many external communication requests
. ddos attacks typically target sites such as banks , credit card payment gateways , or military sites . a ddos attack changes the statistical distribution of network traffic .
therefore , a common practice to detect an attack is to monitor the network traffic using certain summary statics .
since shannon entropy is a well - suited for characterizing a distribution , a popular detection method is to measure the time - history of entropy and alarm anomalies when the entropy becomes abnormal@xcite .
entropy measurements do not have to be `` perfect '' for detecting attacks .
it is however crucial that the algorithm should be computationally efficient at low memory cost , because the traffic data generated by large high - speed networks are enormous and transient ( e.g. , 1 gbits / second ) .
algorithms should be real - time and one - pass , as the traffic data will not be stored@xcite .
many algorithms have been proposed for `` sampling '' the traffic data and estimating entropy over data streams@xcite , in high - speed networks , anomaly events including network failures and ddos attacks may not always be detected by simply monitoring the traditional traffic matrix because the change of the total traffic volume is sometimes small .
one strategy is to measure the entropies of all origin - destination ( od ) flows@xcite .
an od flow is the traffic entering an ingress point ( origin ) and exiting at an egress point ( destination ) .
@xcite showed that measuring entropies of od flows involves measuring the intersection of two data streams , whose moments can be decomposed into the moments of individual data streams ( to which cc is applicable ) and the moments of the absolute difference between two data streams .
the recent work@xcite was devoted to estimating the shannon entropy of msn search logs , to help answer some basic problems in web search , such as , _ how big is the web ?
_ the search logs can be viewed as data streams , and @xcite analyzed several `` snapshots '' of a sample of msn search logs .
the sample used in @xcite contained 10 million @xmath33query , url , ip@xmath34 triples ; each triple corresponded to a click from a particular ip address on a particular url for a particular query .
@xcite drew their important conclusions on this ( hopefully ) representative sample .
alternatively , one could apply data stream algorithms such as cc on the whole history of msn ( or other search engines ) .
a workshop in nips03 was denoted to entropy estimation , owing to the wide - spread use of shannon entropy in neural computations@xcite .
( http://www.menem.com/~ilya/pages/nips03 ) for example , one application of entropy is to study the underlying structure of spike trains . because the elements , @xmath14 $ ] , are time - varying
, a nave counting mechanism requires a system of @xmath31 counters to compute @xmath35 exactly ( unless @xmath36 ) .
this is not always realistic .
estimating @xmath35 in data streams is heavily studied@xcite .
we have mentioned that computing @xmath19 in _ strict - turnstile _
model is trivial using a simple counter .
one might naturally speculate that when @xmath29 , computing ( approximating ) @xmath35 should be also easy .
however , before _ compressed counting ( cc ) _
, none of the prior algorithms could capture this intuition .
cc improves _
symmetric stable random projections_@xcite uniformly for all @xmath37 as shown in figure [ fig_comp_var_factor ] in section [ sec_cc ] . however , one can still considerably improve cc around @xmath36 , by developing better estimators , as in this study .
in addition , no empirical studies on cc were reported .
@xcite applied _ symmetric stable random projections _ to approximate the moments and shannon entropy . the nice theoretical work @xcite provided the criterion to choose the @xmath0 so that shannon entropy can be approximated with a guaranteed accuracy , using the @xmath0th frequency moment .
* * we prove that using rnyi entropy to estimate shannon entropy has ( much ) smaller bias than using tsallis entropy . *
when data follow a common zipf distribution , the difference could be a magnitude . *
* we provide a much improved estimator .
* cc boils down to a statistical estimation problem .
the new estimator based on _ optimal quantiles _ exhibits considerably smaller variance when @xmath38 , compared to @xcite .
* * we demonstrate the bias - variance trade - off in estimating shannon entropy * , important for choosing the sample size and how small @xmath39 should be . *
* we supply an empirical study .
* section [ sec_data ] illustrates what entropies are like in real data .
section [ sec_limit ] includes some theoretical studies of entropy .
the methodologies of cc and two estimators are reviewed in section [ sec_cc ] .
the new estimator based on the _ optimal quantiles _ is presented in section [ sec_quantile ] .
we analyze in section [ sec_entropy_est ] the biases and variances in estimating entropies .
experiments on real web crawl data are presented in section [ sec_exp ] .
since the estimation accuracy is what we are interested in , we can simply use static data instead of real data streams .
this is because at the time @xmath12 , @xmath40^\alpha$ ] is the same at the end of the stream , regardless whether it is collected at once ( i.e. , static ) or incrementally ( i.e. , dynamic ) .
ten english words are selected from a chunk of web crawl data with @xmath41 pages .
the words are selected fairly randomly , except that we make sure they cover a whole range of sparsity , from function words ( e.g. , a , the ) , to common words ( e.g. , friday ) to rare words ( e.g. , twist ) .
the data are summarized in table [ tab_data ] .
l l l l l l l + word & nonzero & @xmath42 & @xmath43 & @xmath44 & @xmath45 & @xmath46 + + twist & 274 & 5.4873 & 5.4962 & 5.4781 & 6.3256 & 4.7919 + rice & 490 & 5.4474 & 5.4997 & 5.3937 & 6.3302 & 4.7276 + friday & 2237 & 7.0487 & 7.1039 & 6.9901 & 8.5292 & 5.8993 + fun & 3076 & 7.6519 & 7.6821 & 7.6196 & 9.3660 & 6.3361 + business & 8284 & 8.3995 & 8.4412 & 8.3566 & 10.502 & 6.8305 + name & 9423 & 8.5162 & 9.5677 & 8.4618 & 10.696 & 6.8996 + have & 17522 & 8.9782 & 9.0228 & 8.9335 & 11.402 & 7.2050 + this & 27695 & 9.3893 & 9.4370 & 9.3416 & 12.059 & 7.4634 + a & 39063 & 9.5463 & 9.5981 & 9.4950 & 12.318 & 7.5592 + the & 42754 & 9.4231 & 9.4828 & 9.3641 & 12.133 & 7.4775 + [ tab_data ] + figure [ fig_entropy ] selects two words to compare their shannon entropies @xmath42 , rny entropies @xmath24 , and tsallis entropies @xmath26 .
clearly , although both approach shannon entropy , rny entropy is much more accurate than tsallis entropy .
this section presents two lemmas , proved in the appendix .
lemma [ lem_bias ] says rnyi entropy has smaller bias than tsallis entropy for estimating shannon entropy .
[ lem_bias ] @xmath47 lemma [ lem_bias ] does not say precisely how much better . note that when @xmath48 , the magnitudes of @xmath49 and @xmath50 are largely determined by the first derivatives ( slopes ) of @xmath24 and @xmath26 , respectively , evaluated at @xmath48 .
lemma [ lem_limit ] directly compares their first and second derivatives , as @xmath48 .
[ lem_limit ] as @xmath1 , @xmath51 lemma [ lem_limit ] shows that in the limit , @xmath52 , verifying that @xmath24 should have smaller bias than @xmath26 . also , @xmath53 .
two special cases are interesting . in this case ,
@xmath54 for all @xmath8 .
it is easy to show that @xmath55 regardless of @xmath0 .
thus , when the data distribution is close to be uniform , rnyi entropy will provide nearly perfect estimates of shannon entropy . in web and nlp applications@xcite ,
the zipf distribution is common : @xmath56 .
, @xmath57 .
the curves largely overlap and hence we do not label the curves .
, width=153 ] figure [ fig_ratio ] plots the ratio , @xmath58 . at @xmath59 ( which is common@xcite ) ,
the ratio is about @xmath60 , meaning that the bias of rnyi entropy could be a magnitude smaller than that of tsallis entropy , in common data sets .
compressed counting ( cc ) assumes the _ relaxed strict - turnstile _ data stream model .
its underlying technique is based on _ maximally - skewed stable random projections_. a random variable @xmath61 follows a maximally - skewed @xmath0-stable distribution if the fourier transform of its density is@xcite @xmath62 where @xmath63 , @xmath64 , and @xmath65 .
we denote @xmath66 . the skewness parameter @xmath67 for general stable distributions ranges in @xmath68 $ ] ; but cc uses @xmath65 , i.e. , * maximally - skewed*. previously , the method of _ symmetric stable random projections_@xcite used @xmath69 .
consider two independent variables , @xmath70 .
for any non - negative constants @xmath71 and @xmath72 , the `` @xmath0-stability '' follows from properties of fourier transforms : @xmath73 note that if @xmath69 , then the above stability holds for any constants @xmath71 and @xmath72 .
we should mention that one can easily generate samples from a stable distribution@xcite .
conceptually , one can generate a matrix @xmath74 and multiply it with the data stream @xmath21 , i.e. , @xmath75 .
the resultant vector @xmath76 is only of length @xmath77 .
the entries of @xmath78 , @xmath79 , are i.i.d .
samples of a stable distribution @xmath80 . by property of fourier transforms ,
the entries of @xmath76 , @xmath81 @xmath82 to @xmath77 , are i.i.d .
samples of a stable distribution @xmath83_j = \sum_{i=1}^d r_{ij } a_t[i]%\\ \sim s\left(\alpha,\beta=1,f_{(\alpha ) } = \sum_{i=1}^d a_t[i]^\alpha\right),\end{aligned}\ ] ] whose scale parameter @xmath35 is exactly the @xmath0th moment . thus , cc boils down to a statistical estimation problem . for real implementations , one should conduct @xmath84 incrementally .
this is possible because the _ turnstile _ model ( [ eqn_turnstile ] ) is a linear updating model .
that is , for every incoming @xmath2 , we update @xmath85 for @xmath82 to @xmath77 . entries of @xmath78 are generated on - demand as necessary .
@xcite commented that , when @xmath77 is large , generating entries of @xmath78 on - demand and multiplications @xmath86 , @xmath82 to @xmath77 , can be prohibitive .
an easy `` fix '' is to use @xmath77 as small as possible , which is possible with cc when @xmath29 . at the same @xmath77 , all procedures of cc and _ symmetric stable random projections _
are the same except the entries in @xmath78 follow different distributions .
however , since cc is much more accurate especially when @xmath29 , it requires a much smaller @xmath77 at the same level of accuracy .
cc boils down to estimating @xmath35 from @xmath77 i.i.d .
samples @xmath87 .
@xcite provided two estimators .
the asymptotic ( i.e. , as @xmath88 ) variance is @xmath89 as @xmath1 , the asymptotic variance approaches zero . which is asymptotically unbiased and has variance @xmath90 @xmath91 is defined only for @xmath92 and is considerably more accurate than the geometric mean estimator @xmath93 .
the two estimators for cc in @xcite dramatically reduce the estimation variances compared to _
symmetric stable random projections_. they are , however , are not quite adequate for estimating shannon entropy using small ( @xmath77 ) samples .
we discover that an estimator based on the _ sample quantiles _ considerably improves @xcite when @xmath29 .
given @xmath77 i.i.d samples @xmath94 , we define the @xmath95-quantile is to be the @xmath96th smallest of @xmath97 . for example , when @xmath98 , then @xmath99th quantile is the smallest among @xmath97 s . to understand why the quantile works ,
consider the normal @xmath100 , which is a special case of stable distribution with @xmath101 .
we can view @xmath102 , where @xmath103 .
therefore , we can use the ratio of the @xmath95th quantile of @xmath81 over the @xmath95-th quantile of @xmath104 to estimate @xmath105 . note that @xmath35 corresponds to @xmath106 , not @xmath105 .
assume @xmath94 , @xmath82 to @xmath77 .
one can sort @xmath97 and use the @xmath96th smallest @xmath97 as the estimate , i.e. , @xmath107 denote @xmath108 , where @xmath109 .
denote the probability density function of @xmath61 by @xmath110 , the probability cumulative function by @xmath111 , and the inverse by @xmath112 .
the asymptotic variance of @xmath113 is presented in lemma [ lem_q_var ] , which follows directly from known statistics results , e.g. , ( * ? ? ?
* theorem 9.2 ) .
[ lem_q_var ] @xmath114 we can then choose @xmath115 to minimize the asymptotic variance .
we denote the optimal quantile estimator by @xmath116 .
the optimal quantiles , denoted by @xmath117 , has to be determined by numerically and tabulated ( as in table [ tab_oq ] ) , because the density functions do not have an explicit closed - form .
we used the * fbasics * package in * r*. we , however , found the implementation of those functions had numerical problems when @xmath118 and @xmath119 .
table [ tab_oq ] provides the numerical values for @xmath120 , @xmath121 ( [ eqn_w ] ) , and the variance of @xmath122 ( without the @xmath123 term ) . .in order to use the optimal quantile estimator , we tabulate the constants @xmath120 and @xmath121 . [ cols="<,<,<,<",options="header " , ] [ tab_oq ] figure [ fig_comp_var_factor ] ( left panel ) compares the variances of the three estimators for cc . to better illustrate the improvements , figure [ fig_comp_var_factor ] ( right panel ) plots the ratios of the variances .
when @xmath124 , the _ optimal _ quantile reduces the variances by a factor of 70 ( compared to the _ geometric mean _
estimator ) , or 20 ( compared to the _ harmonic mean _
estimator ) .
this section analyzes the biases and variances in estimating shannon entropy .
also , we provide the criterion for choosing the sample size @xmath77 .
we use @xmath125 , @xmath126 , and @xmath127 to denote generic estimators .
since @xmath125 is ( asymptotically ) unbiased , @xmath129 and @xmath130 are also asymptotically unbiased .
the asymptotic variances of @xmath129 and @xmath130 can be computed by taylor expansions : @xmath131 we use @xmath132 and @xmath133 to denote the estimators for shannon entropy using the estimated @xmath129 and @xmath130 , respectively .
the variances remain unchanged , i.e. , @xmath134 however , @xmath132 and @xmath133 are no longer ( asymptotically ) unbiased , because @xmath135 the @xmath136 biases arise from the estimation biases and diminish quickly as @xmath77 increases .
however , the `` intrinsic biases , '' @xmath137 and @xmath138 , can not be reduced by increasing @xmath77 ; they can only be reduced by letting @xmath0 close to 1 .
the total error is usually measured by the mean square error : mse = bias@xmath139 + var .
clearly , there is a variance - bias trade - off in estimating @xmath42 using @xmath24 or @xmath26 .
the optimal @xmath0 is data - dependent and hence some prior knowledge of the data is needed in order to determine it .
the prior knowledge may be accumulated during the data stream process .
experiments on real data ( i.e. , table [ tab_data ] ) can further demonstrates the effectiveness of compressed counting ( cc ) and the new _ optimal quantile _ estimator .
we could use static data to verify cc because we only care about the estimation accuracy , which is same regardless whether the data are collected at one time ( static ) or dynamically .
we present the results for estimating frequency moments and shannon entropy , in terms of the normalized mses .
we observe that the results are quite similar across different words ; and hence only one word is selected for the presentation .
figure [ fig_rice_f ] provides the normalized mses ( by @xmath140 ) for estimating the @xmath0th frequency moments , @xmath35 , for word rice : * the errors of the three estimators for cc decrease ( to zero , potentially ) as @xmath1 .
the improvement of cc over _ symmetric stable random projections _ is enormous . *
the optimal quantile estimator @xmath122 is in general more accurate than the geometric mean and harmonic mean estimators near @xmath141 .
however , for small @xmath77 and @xmath142 , @xmath122 exhibits bad behaviors , which disappear when @xmath143 . *
the theoretical asymptotic variances in ( [ eqn_f_gm_var ] ) , ( [ eqn_f_hm_var ] ) , and table [ tab_oq ] are accurate .
+ figure [ fig_rice_hr ] provides the mses from estimating the shannon entropy using the rnyi entropy , for word rice : * using _ symmetric stable random projections _ with @xmath0 close to 1 is not a good strategy and not practically feasible because the required sample size is enormous .
* there is clearly a variance - bias trade - off , especially for the _ geometric mean _ and _ harmonic mean _ estimators .
that is , for each @xmath77 , there is an `` optimal '' @xmath0 which achieves the smallest mse .
* using the _ optimal quantile _ estimator does not show a strong variance - bias trade - off , because its has very small variance near @xmath36 and its mses are mainly dominated by the ( intrinsic ) biases , @xmath137 .
+ figure [ fig_rice_ht ] presents the mses for estimating shannon entropy using tsallis entropy .
the effect of the variance - bias trade - off for geometric mean and harmonic mean estimators , is even more significant , because the ( intrinsic ) bias @xmath138 is much larger .
web search data and network data are naturally data streams .
the entropy is a useful summary statistic and has numerous applications , e.g. , network anomaly detection . efficiently and accurately computing the entropy in large and frequently updating data streams , in one - pass , is an active topic of research .
a recent trend is to use the @xmath0th frequency moments with @xmath29 to approximate shannon entropy .
we conclude : * we should use rnyi entropy to approximate shannon entropy .
using tsallis entropy will result in about a magnitude larger bias in a common data distribution . *
the _ optimal quantile _
estimator for cc reduces the variances by a factor of 20 or 70 when @xmath144 , compared to the estimators in @xcite . * when _ symmetric stable random projections _ must be used
, we should exploit the variance - bias trade - off , by not using @xmath0 very close 1 .
@xmath145 and @xmath146 . note that @xmath147 , @xmath148 if @xmath92 and @xmath149 if @xmath142 . for @xmath150 , @xmath151 always holds , with equality when @xmath152 .
therefore , @xmath153 when @xmath92 and @xmath154 when @xmath142 .
also , we know @xmath155 . therefore , to show @xmath156 , it suffices to show that both @xmath26 and @xmath24 are decreasing functions of @xmath157 .
taking the first derivatives of @xmath26 and @xmath24 yields @xmath158 to show @xmath159 , it suffices to show that @xmath160 .
taking derivative of @xmath161 yields , @xmath162 i.e. , @xmath163 if @xmath92 and @xmath164 if @xmath142 . because @xmath165 , we know @xmath160 .
this proves @xmath166 . to show @xmath167
, it suffices to show that @xmath168 , where @xmath169 note that @xmath170 and hence we can view @xmath171 as probabilities . since @xmath172 is a concave function , we can use jensen s inequality : @xmath173 , to obtain @xmath174
as @xmath1 , using lhopital s rule @xmath175^\prime}{\left[(\alpha-1)^2\right]^\prime}\\\notag = & \lim_{\alpha\rightarrow1 } \frac{- ( \alpha-1)\sum_{i=1}^d p_i^\alpha \log^2p_i}{2(\alpha-1)}=-\frac{1}{2}\sum_{i=1}^d p_i \log^2 p_i.\end{aligned}\ ] ] note that , as @xmath48 , @xmath176 but @xmath177 . again , applying lhopital s rule yields the expressions for @xmath180 and @xmath181 @xmath182^\prime}{\left[(\alpha-1)^2\sum_{i=1}^dp_i^\alpha\right]^\prime}\\\notag = & \lim_{\alpha\rightarrow 1 } \frac{\left[\sum_{i=1}^dp_i^\alpha\log p_i \log \sum_{i=1}^dp_i^\alpha - ( \alpha-1)\sum_{i=1}^d p_i^\alpha \log^2 p_i\right]}{\left[2(\alpha-1)\sum_{i=1}^dp_i^\alpha + ( \alpha-1)^2\sum_{i=1}^dp_i^\alpha\log p_i \right]}\\\notag = & \lim_{\alpha\rightarrow 1 } \frac{\left(\sum_{i=1}^dp_i^\alpha\log p_i\right)^2/\sum_{i=1}^dp_i^\alpha - \sum_{i=1}^d p_i^\alpha \log^2 p_i + \text{negligible terms}}{2\sum_{i=1}^dp_i^\alpha + \text{negligible terms}}\\\notag = & \frac{1}{2}\left(\sum_{i=1}^dp_i\log p_i\right)^2 -\frac{1}{2}\sum_{i=1}^d p_i \log^2 p_i\end{aligned}\ ] ] | the shannon entropy is a widely used summary statistic , for example , network traffic measurement , anomaly detection , neural computations , spike trains , etc .
this study focuses on estimating shannon entropy of data streams .
it is known that shannon entropy can be approximated by rnyi entropy or tsallis entropy , which are both functions of the @xmath0th frequency moments and approach shannon entropy as @xmath1 . * * _ compressed counting ( cc)_**@xcite is a new method for approximating the @xmath0th frequency moments of data streams .
our contributions include : * we prove that rnyi entropy is ( much ) better than tsallis entropy for approximating shannon entropy .
* we propose the _ optimal quantile _ estimator for cc , which considerably improves the estimators in @xcite . *
our experiments demonstrate that cc is indeed highly effective in approximating the moments and entropies .
we also demonstrate the crucial importance of utilizing the variance - bias trade - off . |
in the recent years , the statistical physics techniques have been successfully applied in the description of socioeconomic phenomena . among the studied problems we can cite opinion dynamics , language evolution , biological aging , dynamics of stock markets , earthquakes and many others @xcite .
these interdisciplinary topics are usually treated by means of computer simulations of agent - based models , which allow us to understand the emergence of collective phenomena in those systems .
recently , the impact of nonconformity in opinion dynamics has atracted attention of physicists @xcite .
anticonformists are similar to conformists , since both take cognizance of the group norm .
thus , conformers agree with the norm , anticonformers disagree . on the other hand
, we have the independent behavior , where the individual tends to resist to the groups influence . as discussed in @xcite , independence is a kind of nonconformity , and it acts on an opinion model as a kind of stochastic driving that can lead the model to undergo a phase transition . in fact , independence plays the role of a random noise similar to social temperature @xcite . in this work
we study the impact of independence on agents behavior in a kinetic exchange opinion model . for this purpose
, we introduce a probability @xmath3 of agents to make independent decisions .
our analytical results and numerical simulations show that the model undergoes a phase transition at critical points @xmath5 that depend on another model parameter , related to the agents flexibility .
this work is organized as follows . in section 2
we present the microscopic rules that define the model and in section 3 the numerical and analytical results are discussed .
finally , our conclusions are presented in section 4 .
our model is based on kinetic exchange opinion models ( keom ) @xcite .
a population of @xmath6 agents is defined on a fully - connected graph , i.e. , each agent can interact with all others , which characterizes a mean - field - like scheme .
in addition , each agent @xmath7 carries one of three possible opinions ( or states ) , namely @xmath8 , @xmath1 or @xmath2 .
the following microscopic rules govern the dynamics : 1 .
an agent @xmath7 is randomly chosen ; 2 . with probability @xmath3
, this agent will act independently . in this case , with probability @xmath9 he / she chooses the opinion @xmath10 , with probability @xmath11 he / she adopts the opinion @xmath8 and with probability @xmath11 he / she chooses the opinion @xmath12 ; 3 . on the other hand , with probability
@xmath4 we choose another agent , say @xmath13 , at random , in a way that @xmath13 will influence @xmath7 .
thus , the opinion of the agent @xmath7 in the next time step @xmath14 will be updated according to @xmath15\,,\ ] ] where the sign function is defined such that @xmath16 . in the case where the agent @xmath7 does not act independently , the change of his / her state occur according to a rule similar to the one proposed recently in a keom @xcite .
notice , however , that in ref .
@xcite two randomly chosen agents @xmath7 and @xmath13 interact with competitive couplings , i.e. , the kinetic equation of interaction is @xmath17 $ ] . in this case
, the couplings @xmath18 are random variables presenting the value @xmath1 ( @xmath0 ) with probability @xmath19 ( @xmath20 ) . in other words ,
the parameter @xmath19 denotes the fraction of negative interactions . in this case , the model of ref .
@xcite undergoes a nonequilibrium phase transition at @xmath21 . in the absence of negative interactions ( @xmath22 )
, the population reaches consensus states with all opinions @xmath0 or @xmath1 .
thus , our eq .
( [ eq1 ] ) represents the keom of ref .
@xcite with no negative interactions , and the above parameter @xmath9 can be related to the agents flexibility @xcite . in this case , for @xmath23 ( no independence ) all stationary states will give us @xmath24 , where @xmath25 is the order parameter of the system , @xmath26 and @xmath27 denotes a disorder or configurational average taken at steady states . the eq .
( [ eq2 ] ) defines the `` magnetization per spin '' of the system .
we will show by means of analytical and numerical results that the independent behavior works as a noise that induces a phase transition in the keom in the absence of negative interactions .
the three states considered in the model can be interpreted as follows @xcite .
we have a population of voters that can choose among two candidates a and b. thus , the opinions represent the intention of an agent to vote for the candidate a ( opinion @xmath0 ) , for the candidate b ( opinion @xmath1 ) , or the agent may be undecided ( opinion @xmath2 ) . in this case ,
notice that there is a difference among the undecided and independent agents .
an agent @xmath7 that decide to behave independently ( with probability @xmath3 ) can make a decision to change or not his / her opinion based on his / her own conviction , whatever is the his / her current state @xmath28 ( decided or undecided ) .
in other words , an interaction with an agent @xmath13 is not required . on the other hand ,
an undecided agent @xmath7 can change his / her opinion @xmath28 in two ways : due to an interaction with a decided agent @xmath13 ( following the rule given by eq . ( [ eq1 ] ) , with probability @xmath4 ) or due to his / her own decision to do that ( independently , with probability @xmath3 ) . regarding the independent behavior , one can consider the homogeneous case ( @xmath29 ) and the heterogeneous one ( @xmath30 ) .
these cases will be considered separately in the next section .
one can start studying the homogeneous case @xmath29 . in this case
, we have that all probabilities related to the independent behavior , namely @xmath9 and @xmath11 , are equal to @xmath31 .
thus , the probability that an agent @xmath7 chooses a given opinion @xmath0 , @xmath1 or @xmath2 independently of the opinions of the other agents is @xmath32 . for the analysis of the model , we have considered the order parameter @xmath25 defined by eq .
( [ eq2 ] ) , as well as the susceptibility @xmath33 and the binder cumulant @xmath34 @xcite , defined as @xmath35 ( a ) , order parameter @xmath25 ( b ) and susceptibility @xmath33 ( c ) as functions of the independence probability @xmath3 for the homogeneous case ( @xmath29 ) and different population sizes @xmath6 . in the inset we exhibit the corresponding scaling plots .
the estimated critical quantities are @xmath36 , @xmath37 , @xmath38 and @xmath39 .
results are averaged over @xmath40 , @xmath41 , @xmath42 and @xmath43 samples for @xmath44 and @xmath45 , respectively.,title="fig:",scaledwidth=45.0% ] ( a ) , order parameter @xmath25 ( b ) and susceptibility @xmath33 ( c ) as functions of the independence probability @xmath3 for the homogeneous case ( @xmath29 ) and different population sizes @xmath6 . in the inset
we exhibit the corresponding scaling plots .
the estimated critical quantities are @xmath36 , @xmath37 , @xmath38 and @xmath39 .
results are averaged over @xmath40 , @xmath41 , @xmath42 and @xmath43 samples for @xmath44 and @xmath45 , respectively.,title="fig:",scaledwidth=45.0% ] + ( a ) , order parameter @xmath25 ( b ) and susceptibility @xmath33 ( c ) as functions of the independence probability @xmath3 for the homogeneous case ( @xmath29 ) and different population sizes @xmath6 . in the inset
we exhibit the corresponding scaling plots .
the estimated critical quantities are @xmath36 , @xmath37 , @xmath38 and @xmath39 .
results are averaged over @xmath40 , @xmath41 , @xmath42 and @xmath43 samples for @xmath44 and @xmath45 , respectively.,title="fig:",scaledwidth=45.0% ] notice that the binder cumulant defined by eq .
( [ eq4 ] ) is directly related to the order s parameter _ kurtosis _
@xmath46 , that can be defined as @xmath47 .
the initial configuration of the population is fully disordered , i.e. , we started all simulations with an equal fraction of each opinion ( @xmath31 for each one ) .
in addition , one time step in the simulations is defined by the application of the rules defined in the previous section @xmath6 times . in fig .
[ fig1 ] we exhibit the quantities of interest as functions of @xmath3 for different population sizes @xmath6 .
all results suggest the typical behavior of a phase transition . in order to estimate the transition point ,
we look for the crossing of the binder cumulant curves for the different sizes @xcite . from fig .
[ fig1 ] ( a ) , the estimated value is @xmath48 , which agrees with the analytical prediction @xmath49 [ see eq .
( [ qc_sym ] ) of the appendix ] .
in addition , in order to determine the critical exponents associated with the phase transition we performed a finite - size scaling ( fss ) analysis .
we have considered the standard scaling relations , @xmath50 that are valid in the vicinity of the transition .
thus , we exhibit in the insets of fig .
[ fig1 ] the scaling plots of the quantities of interest ( @xmath34 , @xmath25 and @xmath33 ) .
our estimates for the critical exponents are @xmath37 , @xmath38 and @xmath39 .
notice that the critical probability , @xmath49 , presents the same value of the critical fraction of negative interactions ( @xmath21 ) of the keom of ref . @xcite .
in addition , the critical exponents are the same in the two formulations of the model .
thus , the inclusion of the independent behavior with equal probabilities ( i.e. , @xmath29 ) produces a similar effect to the introduction of negative interactions in the keom of ref .
@xcite . as a function of @xmath3 for @xmath51 and typical values of @xmath9 .
one can see that the transition points depend on @xmath9 .
the inset shows the region near @xmath52 .
results are averaged over @xmath43 simulations.,scaledwidth=45.0% ] one can also consider the general case where @xmath30 . in this case , for an agent that act independently , the probabilities to choose the three possible opinions are different . as in the previous subsection
, we started all simulations with an equal fraction of each opinion . in fig .
[ fig2 ] we show the order parameter as a function of @xmath3 for typical values of @xmath9 and population size @xmath51 .
one can see that the phase transition occurs for all values of @xmath9 exhibited in fig .
[ fig2 ] , and the critical points depend on @xmath9 , i.e. , we have @xmath53 . furthermore , another interesting result that one can see in fig . [ fig2 ] is that for @xmath54 the order parameter goes exactly to @xmath52 at @xmath55 , presenting no finite - size effects as the other curves do , as can be easily seen in the inset of fig .
this fact can be easily understood .
indeed , for @xmath54 all agents that behave independently choose the opinion @xmath56 .
thus , for a sufficiently large value of @xmath3 all agents will change independently to @xmath56 . in this case , eq .
( [ eq2 ] ) give us an order parameter @xmath52 .
this qualitative discussion can be confirmed by analytical considerations ( see the appendix ) . of samples ( over @xmath42 simulations ) that reaches the @xmath2 consensus as a function of @xmath3 for @xmath54 and typical population sizes @xmath6 ( main plot ) .
in the inset it is exhibited the corresponding scaling plot .
the best collapse of data was obtained for @xmath57 and @xmath58.,scaledwidth=45.0% ] thus , the case @xmath54 is special , because all agents change their opinions to @xmath56 for a sufficient large value of the parameter @xmath3 . indeed ,
if all agents are in the @xmath56 state , the evolution equation ( [ eq1 ] ) , when applied ( with probability @xmath4 ) , does not change the opinions to @xmath0 or @xmath1 anymore , which means that the system is in an absorbing state .
this fact , together with the absence of finite - size effects for the order parameter defined in eq .
( [ eq2 ] ) , suggests that one can not apply the scaling relations ( [ eq5 ] ) - ( [ eq8 ] ) for @xmath54 . in this case , it is better to analyze other quantity as an order parameter , as was done , for example , for the 2d sznajd model @xcite .
thus , following @xcite , we performed several simulations of the system for @xmath54 and we measured the fraction @xmath59 of samples that reached the absorbing state with all opinions @xmath56 as a function of @xmath3 .
the result is exhibited in fig .
[ fig3 ] for typical values of @xmath6 , and in this case this order parameter strongly depends on the system size .
considering scaling relations in a similar way as in ref .
@xcite , i.e. , plotting @xmath59 as a function of the variable @xmath60 , one obtains @xmath61 , in agreement with the previous discussion , and @xmath62 .
the corresponding data collapse is exhibited in the inset of fig .
[ fig3 ] . as above discussed
, the numerical results suggest that critical points @xmath5 depend on @xmath9 .
this picture is confirmed by the analytical solution of the model , which give us ( see eq .
( [ qcs ] ) of the appendix ) @xmath63 \ , .\ ] ] notice that the above solution give us @xmath64 , and the exact result for @xmath54 is @xmath65 , which agrees with the above discussion .
we performed a fss analysis based on eqs .
( [ eq5 ] ) - ( [ eq8 ] ) in order to obtain the critical points and the critical exponents for other values of @xmath66 . in fig .
[ fig4 ] the eq .
( [ eq9 ] ) is plotted together with all numerical estimates of @xmath67 .
one can see that the numerical results agree very well with the analytical prediction .
in addition , the critical exponents are the same for all values of @xmath66 , i.e. , we have @xmath37 , @xmath38 and @xmath39 , which indicates a universality on the order - disorder frontier of the model , except on the `` special '' point @xmath54 . versus @xmath9 , separating the ordered and the disordered phases .
the symbols are the numerical estimates of the critical points @xmath5 , whereas the full line is the analytical prediction , eq .
( [ eq9 ] ) .
the open ( blue ) circle denotes the special case @xmath54 , as discussed in the text .
the error bars determined by the fss analysis are smaller than data points.,scaledwidth=50.0% ]
in this work we introduce the mechanism of independence in a three - state ( @xmath0 , @xmath1 and @xmath2 ) kinetic exchange opinion model . in the absence of negative interactions ,
this model always evolve to ordered ( consensus ) states .
our results show that independence acts as a noise , inducing a nonequilibrium phase transition in the model , and that the critical points depend on the agents flexibility @xmath9 .
the numerical simulations suggest that we have the same critical exponents for all values of @xmath66 , i.e. , we have @xmath37 , @xmath38 and @xmath39 , which indicates a universality on the order - disorder frontier of the model .
this is an expected result , due to the mean - field character of the interactions . on the other hand , the case @xmath54 is special , and the system undergoes a phase transition to an absorbing state with all agents in the undecided state @xmath56 .
following the lines of refs .
@xcite , we computed the critical values of the probability @xmath3 .
we first obtained the matrix of transition probabilities whose elements @xmath68 furnish the probability that a state suffers the shift or change @xmath69 .
let us also define @xmath70 , @xmath71 and @xmath72 , the stationary probabilities of each possible state . in the steady state , the fluxes into and out from a given state must balance . in particular , for the null state , one has @xmath73
moreover , when the order parameter vanishes , it must be @xmath74 .
finally , let us define @xmath75 , with @xmath76 , the probability that the state shift per unit time is @xmath46 , that is , @xmath77 .
in the steady state , the average shift must vanish , namely , @xmath78
+ r(1)-r(-1)=0 \,.\ ] ] for the more general case considering the flexibility parameter @xmath9 , the elements @xmath68 of the transition matrix are @xmath79 first , one can consider the homogeneous case @xmath29 . in this case , the above elements @xmath68 are simplified , and the null state condition ( [ nullstate ] ) give us @xmath80 ( disorder condition ) .
thus , the null average shift condition ( [ nullshift ] ) , together with the above disorder condition , leads to @xmath81 for the more general case , the conditions ( [ nullstate ] ) and ( [ nullshift ] ) lead to a second - order equation for the variable @xmath5 , @xmath82 which give us two distinct solutions , namely @xmath83 \ , .\end{aligned}\ ] ] although both solutions are mathematically valid , the solution @xmath84 leads to @xmath85 in the disordered phase , and consequently @xmath86 and @xmath87 . on the other hand ,
the solution @xmath88 is physically acceptable because it leads to @xmath89 as well as @xmath89 and @xmath89 , satisfying the normalization condition @xmath90 .
thus , the physically valid analytical solution for the general model is given by @xmath88 .
in particular , we have @xmath49 for @xmath29 [ which agrees with eq .
( [ qc_sym ] ) ] and @xmath91 for @xmath54 .
in addition , it can be shown that the null state condition ( [ nullstate ] ) for @xmath54 give us the solution @xmath92 in the disordered phase , and then @xmath93 .
this explains the result @xmath52 for @xmath94 observed in figs .
[ fig2 ] and [ fig3 ] , that was discussed in section 3 .
the author acknowledges financial support from the universidade federal fluminense , brazil . | in this work we study the critical behavior of a three - state ( @xmath0 , @xmath1 , @xmath2 ) opinion model with independence .
each agent has a probability @xmath3 to act as independent , i.e. , he / she can choose his / her opinion independently of the opinions of the other agents .
on the other hand , with the complementary probability @xmath4 the agent interacts with a randomly chosen individual through a kinetic exchange .
our analytical and numerical results show that the independence mechanism acts as a noise that induces an order - disorder transition at critical points @xmath5 that depend on the individuals flexibility . for a special value of this flexibility the system undergoes a transition to an absorbing state with all opinions @xmath2 .
keywords : social dynamics , collective phenomenon , computer simulation , phase transition |
in a recent paper @xcite we pointed out that the @xmath0-qubit pauli group ( for @xmath8 ) is a rich source of both observables - based and projectors - based parity proofs " of the kochen - specker ( ks ) theorem @xcite .
we refer to the proofs as parity proofs because , in either the observables - based or projectors - based versions , they exploit the concept of parity to achieve their ends .
the purpose of this paper is to give examples of both types of proofs for @xmath9- and higher qubit systems and to point out several of their properties that we did not discuss earlier in our work on 2- and 3-qubit systems @xcite .
more precisely , the goals of this paper are the following : + ( 1 ) we explain what we mean by an observables - based ks proof and show how it can be depicted in the form of a diagram from which it is obvious by inspection .
the two best known examples of such proofs are a 2-qubit proof due to peres @xcite and mermin @xcite and a 3-qubit proof due to mermin @xcite based on earlier work by greenberger , horne and zeilinger @xcite . in ref .
@xcite we presented several examples of @xmath2- and @xmath3-qubit proofs of this kind and in ref .
@xcite we indicated that we had found a large number of @xmath9- and higher qubit proofs but gave few details . here
we give examples of 4-,5- and 6-qubit proofs that convey a feeling for the wide variety of possibilities that open up as one goes to a larger number of qubits .
we should stress that we consider only _ critical _ proofs , i.e. , ones that can not be reduced to smaller proofs by omitting some subset of qubits and/or observables in them , but that despite this restriction the number of possibilities still grows very rapidly as one goes to a larger number of qubits .
+ ( 2 ) we show how any observables - based ks proof can be used to construct a system of projectors and bases from which a large number of projectors - based parity proofs of the ks theorem can be obtained .
the simplest example of this procedure is provided by the 2-qubit peres - mermin square , whose nine observables give rise to a system of projectors and bases that yield a total of @xmath10 projectors - based parity proofs @xcite . in recent years
a number of other examples of projectors - based parity proofs have been found in four @xcite and eight @xcite dimensions .
a major point of this paper is that every observables - based ks proof , based on a subset of observables of the pauli group , gives rise to a system of projectors - based parity proofs , and we give a simple algorithm for making this transition .
we illustrate this algorithm in the particular case of a 4-qubit observables - based proof and show how the @xmath11 associated projectors - based proofs can be obtained with practically no effort ( once the system of projectors and bases within which they are embedded has been set up ) .
in addition to the fact that they are easy to generate , the projectors - based proofs are also easy to check , since only simple counting is called for . like the observables - based proofs from which they are derived ,
the projectors - based proofs are critical in the sense that they can not be whittled down to smaller proofs by omitting some subset of their bases . because each observables - based proof gives rise to a large number of projectors - based proofs , the variety and quantity of the latter are vastly greater than those of the former .
+ ( 3 ) it is interesting to ask if there are any infinite classes of observables - based proofs that apply to all numbers of qubits from two up .
we have discovered several such classes that we call the star , the wheel , the whorl and the kite ( with the names reflecting the shapes of the associated diagrams ) .
the most complex of these families is the kite , and we give a detailed discussion of it in this paper .
all the members of this family can be represented by diagrams having the form of a kite , with the body consisting of nine observables arranged on a 3 x 3 grid and the tail consisting of a string of observables of arbitrary length . for any number of qubits ,
a suitable choice of observables ( and often more than one ) can be placed on the framework of the kite to yield a ks proof .
the @xmath2-qubit peres - mermin square can be regarded as a kite without a tail , and the higher qubit proofs of this family involve tails of increasing lengths .
an interesting feature of this family is that all its members give rise to projectors - based ks proofs involving just nine bases ( or experimental contexts ) , and so are the most compact proofs of this type known in @xmath6 dimensions for all @xmath7 .
+ the next three sections are devoted to a discussion of the above three points .
we then comment on the significance of our results and their relation to other work .
although this paper builds upon our earlier work @xcite , it is written to be self contained and requires no familiarity with that work .
an observables - based proof of the ks theorem for a system of @xmath0 qubits consists of a subset of observables of the @xmath0-qubit pauli group that forms a number of commuting sets of a special kind . the proof is conveniently displayed in the form of a diagram in which the observables are represented as points ( or actually as letters within circles centered at points ) and the special commuting sets as lines ( which could be straight or curved ) joining the points .
an observable is represented by a sequence of letters , each of which can be one of @xmath12 or @xmath13 ( these being the pauli and identity operators of a qubit ) .
for example , @xmath14 represents a 5-qubit observable that is the tensor product of the observables @xmath15 and @xmath16 of the individual qubits .
every special commuting set in any of our proofs has the property that the product of all the observables in it is either + * * i * * or @xmath17**i * * , where * i * is the identity operator in the space of all the qubits
. sets with product + * * i * * are shown by thin lines and sets with product @xmath17**i * * by thick lines in our diagrams . any diagram representing an observables - based ks proof has the following two properties : ( a ) each observable lies at the intersection of an even number of lines , and ( b )
the total number of thick lines is odd .
these properties guarantee that the diagrams provide proofs of the ks theorem . to see why , note that the eigenvalues of any @xmath0-qubit observable are @xmath18 and that a noncontextual hidden variables theory is required to assign the value @xmath19 or @xmath20 to each of the observables in such a way that the product of the values assigned to the observables on a thin ( or a thick ) line equals @xmath19 ( or @xmath20 ) .
however properties ( a ) and ( b ) rule out such a value assignment be the product of the values assigned to the observables in the commuting set indexed by @xmath21 and consider the product @xmath22 taken over all the commuting sets .
property ( b ) requires that @xmath23 but property ( a ) requires that @xmath24 , so there is a contradiction . ] and so prove the ks theorem .
+ a number of @xmath0-qubit proof - diagrams are shown in figs.1 - 4 for @xmath25 and @xmath26 .
the names we have given to the diagrams are whimsical and merely try to capture their shapes .
we have also attached a symbol to each diagram that should help the reader pick out the commuting sets in it ( particularly in the case of the more complicated diagrams ) .
the left half of each symbol lists the number of observables of each multiplicity ( with the multiplicities as subscripts ) and the right half lists the number of commuting sets of each size ( with the sizes as subscripts ) .
for example , the symbol @xmath27-@xmath28 for the 4-qubit star of fig.1 indicates that it contains 12 observables of multiplicity 2 ( i.e. , that each occur twice among its commuting sets ) and that there are four commuting sets of four observables and one each of five and three observables .
the sum of the products of each number with its subscript in the left half of the symbol must equal the similar sum of products on the right , and this can be used as a quick check on the consistency of the symbol . for a symbol to represent a valid ks proof , all the subscripts in its left half must be even ( it is also necessary that the number of commuting sets with product @xmath17**i * * be odd , but this fact is not made evident in the symbol and can only be checked by looking at the diagram . ) + [ cols="^,^ " , ] [ kite7_16 ]
this paper has established the following results : + 1 .
the hierarchy of @xmath0-qubit pauli groups ( for @xmath8 ) contains many subsets of observables that provide parity proofs of the ks theorem .
we restrict our attention to proofs that are irreducible ( i.e. that can not be reduced to simpler proofs by omitting some subset of observables and/or qubits in them ) and unitarily inequivalent . with these caveats ,
the 2-qubit group gives rise to only two distinct types of proofs ( the peres - mermin square and a more complicated structure we call the whorl " , which are shown as figs.1 and 2 of ref .
@xcite ) , but the 3-qubit group leads to many more @xcite and the possibilities increase rapidly as one goes upwards in the number of qubits . it would be an interesting problem to make a more systematic inventory of the proofs at a given value of @xmath0 and to get a feeling for how this number grows with @xmath0 .
there are two devices we have introduced in this paper that could assist with this task : the first is a diagrammatic representation of each proof ( from which it can be verified by inspection ) and the second is a symbol that captures the important features of the observables and special commuting sets in the proof .
however we should caution that neither of these devices suffices to pin down a proof uniquely since there are distinct proofs that can be accommodated on the same diagrammatic skeleton , as well as inequivalent proofs that share the same symbol ( see fig.7 of ref .
@xcite for an example ) . even in the absence of a detailed knowledge of the terrain
, one can state quite confidently that the pauli group ( particularly as one goes to a larger number of qubits ) abounds in a great variety of structures that can be used to give transparent demonstrations of quantum contextuality .
the second major point of this paper is that every observables - based ks proof can be used to generate a system of projectors and bases from which a a large number of projectors - based ks proofs can be obtained .
the algorithm for generating these proofs is simple , and the proofs themselves are easy to validate because only a simple parity check is called for .
we have introduced a detailed symbol for a projectors - based proof that describes the projectors and bases in it , both as a way of summarizing the key aspects of the proof and to draw attention to the wide variety of proofs that can coexist within the same framework of pure and hybrid bases provided by an observables - based proof .
since the observables - based proofs are themselves very numerous , and each spawns a large number of projectors - based proofs ( typically thousands ) , the quantity and variety of the latter proofs vastly outstrip those of the former .
we should add that the symbols we have introduced for the projectors - based proofs , though useful and informative , do not serve to pin them down uniquely since we have found many examples of inequivalent proofs that are described by the same symbol .
among all the observables - based and related projectors - based proofs , there is a simple class that is worth singling out for special mention : it is the one in which each observable occurs in exactly two commuting sets and any two commuting sets have at most one observable in common .
one knows in this case , even before one has set up the basis table , that one will find exactly @xmath5 projectors - based proofs .
we have discovered several infinite families of observables - based proofs ( that we term the star , the wheel , the whorl and the kite ) whose members yield ks proofs for all numbers of qubits from two up .
the kite is the most complex of these families , and we have given a detailed discussion of it in this paper .
this family has two remarkable features .
the first is that for a system of @xmath0 qubits it is always possible to find commuting sets of considerably less than @xmath0 observables that can be used to construct a ks proof , thus leading to greater economy in the design of experimental tests based on this class of proofs .
and the second is that _ any _ kite proof always yields a projectors - based proof involving only nine bases ( or experimental contexts ) .
these proofs are generalizations of the classic 18 - 9 proof of cabello et al @xcite in four dimensions and hold in all dimensions of the form @xmath6 , for @xmath8 .
it is an open question whether there are any proofs involving less than nine bases in these dimensions ( we believe the answer is no ) .
an even more basic question is whether there are any projectors - based parity proofs @xcite .
] at all in any even dimension not of the form @xmath6 .
again we suspect that the answer is no , but we do not have a proof of this conjecture ( and would find it fascinating if someone came up with a counterexample ) .
+ there is currently a great interest in contextuality and nonlocality and their experimental tests .
in one interesting development , it has been shown that any projectors - based ks proof can be converted into an inequality for testing quantum contextuality @xcite and a number of experimental tests of such inequalities have actually been carried out @xcite .
a number of works have shed light on contextuality in a manner that bypasses the ks theorem @xcite . on the formal side ,
a connection between ks proofs and logical bell inequalities " has been made in ref .
@xcite and the question of state - independent contextuality for identical particles has been explored in ref . @xcite .
although the present work concentrates entirely on qubits , it should be pointed out that ks proofs for qudits ( i.e. , @xmath29-state systems ) have been explored in ref . @xcite . among the practical applications of contextuality that are receiving attention are quantum key distribution @xcite , quantum error correction @xcite , random number generation @xcite , parity oblivious transfer @xcite and the design of relational databases @xcite .
this is not a complete survey , of course , but should convey a feeling for the broader context in which the results reported here may be of interest .
e.p.specker , _ dialectica _ * 14 * , 239 ( 1960 ) ; s. kochen , e.p.specker : _ j. math .
mech . _ * * 17 * * , 59 ( 1967 ) .
see also j.s .
bell : _ rev .
phys . _ * * 38 * * , 447 ( 1966 ) . reprinted in j.s .
bell , _ speakable and unspeakable in quantum mechanics _ , ( cambridge university press , cambridge , uk , 1987 )
. d.m.greenberger , m.a.horne and a.zeilinger in _
bell s theorem , quantum theory and conceptions of the universe _ , edited by m.kafatos ( kluwer , dordrecht , 1989 ) ; d.m.greenberger , m.a.horne , a.shimony and a.zeilinger : _ am . j. phys . _
* 58 * , 1131 ( 1990 ) .
m. pavii , n.d .
megill and j.p .
merlet : _ phys .
a _ * * 374 * * , 2122 ( 2010 ) ; m. pavii , j.p .
merlet , b.d .
mckay and n.d .
megill : _ j. phys .
a _ * * 38 * * , 1577 ( 2005 ) ; m. pavii , j.p .
merlet and n.d .
megill : _ the french national institute for research in computer science and control research reports _ * * rr-5388 * * ( 2004 ) .
g. kirchmair , f. zhringer , r. gerritsma , m. kleinmann , o. ghne , a. cabello , r. blatt , c.f .
roos : _ nature _ * * 460 * * , 494 ( 2009 ) ; h. bartosik , j. klep , c. schmitzer , s. sponar , a. cabello , h. rauch , y. hasegawa : _ phys .
lett . _ * * 103 * * , 040403 ( 2009 ) ; e. amselem , m. rdmark , m. bourennane , a. cabello : _ phys . rev
. lett . _ * * 103 * * , 160405 ( 2009 ) ; o. moussa , c.a .
ryan , d.g .
cory , r. laflamme : _ phys .
lett . _ * * 104 * * , 160501 ( 2010 )
. a.a .
klyachko , m.a .
can , s. biniciolu and a.s .
shumovsky : _ phys .
lett . _ * * 101 * * , 020403 ( 2008 ) ; y.c . liang , r.w .
spekkens , h.m .
wiseman : _ phys .
rep . _ * * 506 * * , 1 ( 2011 ) ; p. badziag , i. bengtsson , a. cabello , h.h .
granstrm and j .- .
larsson : _ found phys _ * * 41 * * , 414 ( 2011 ) ; s.yu and c.h.oh , _ phys .
lett . _ * * 108 * * , 030402 ( 2012 ) ; s.yu and c.h.oh , arxiv:1112.5513 ( 2011 ) . | we present a number of observables - based proofs of the kochen - specker ( ks ) theorem based on the @xmath0-qubit pauli group for @xmath1 , thus adding to the proofs that have been presented earlier for the @xmath2- and @xmath3-qubit groups .
these proofs have the attractive feature that they can be presented in the form of diagrams from which they are obvious by inspection .
they are also irreducible in the sense that they can not be reduced to smaller proofs by ignoring some subset of qubits and/or observables in them .
a simple algorithm is given for transforming any observables - based ks proof into a large number of projectors - based ks proofs ; if the observables - based proof has @xmath4 observables , with each observable occurring in exactly two commuting sets and any two commuting sets having at most one observable in common , the number of associated projectors - based parity proofs is @xmath5 .
we introduce symbols for the observables- and projectors - based ks proofs that capture their important features and also convey a feeling for the enormous variety of both these types of proofs within the @xmath0-qubit pauli group .
we discuss an infinite family of observables - based proofs whose members include all numbers of qubits from two up , and show how it can be used to generate projectors - based ks proofs involving only nine bases ( or experimental contexts ) in any dimension of the form @xmath6 for @xmath7 . some implications of our results are discussed .
example.eps gsave newpath 20 20 moveto 20 220 lineto 220 220 lineto 220 20 lineto closepath 2 setlinewidth gsave .4 setgray fill grestore stroke grestore |
the scuba half degree survey ( shades , dunlop 2005 , mortier et al .
2005 ) was originally designed with the aim of characterizing the star - formation history ( hughes et al .
2002 ) and clustering properties ( van kampen et al .
2005 ) of the bright - end of the luminous dust - enshrouded galaxy population . to achieve these goals we mapped two regions of the sky centered on the lockman hole east ( lh ) and subaru xmm - newton deep field ( sxdf ) with the submillimetre common - user bolometer array ( scuba , holland et al .
1999 ) . with a proposed 1@xmath5 sensitivity of 2 mjy at 850@xmath0 m
the complete survey was predicted to identify a statistically robust sample of @xmath6 galaxies , with sufficient radio to fir ancillary data to help identify optical / ir counterparts and derive spectroscopic / photometric redshifts .
this redshift information is essential for determining the star formation and clustering properties for the whole population of ultraluminous dust - enshrouded galaxies .
scuba was de - commissioned in mid-2005 having covered @xmath7 per cent of the originally - proposed area of the shades .
paper i of this series ( mortier et al .
2005 ) describes the survey motivation , strategy and the philosophy adopted for the analysis .
paper ii ( coppin et al . 2006 ) presents the catalogue and number counts derived from the 850@xmath0 m sources .
paper iii ( ivison et al . 2007 ) describes the identification of radio and mid - ir counterparts of these sources .
this paper ( iv ) constructs the redshift distribution derived from the radio - mm - fir photometry of the shades sources based on a compilation of the 850@xmath0 m and 450@xmath0 m scuba data ( coppin et al .
2006 ) , 1.4ghz very large array photometry ( ivison et al . 2007 ) and other previously published mm to fir photometric observations towards these fields .
a study of the mid - ir to optical properties of the shades population , and further constraints on the photometric redshifts of the sources , will be published elsewhere ( clements et al .
2007 , dye et al .
2007 , serjeant et al . 2007 )
. a spectroscopic study of a sub - sample of shades sources with identified optical / ir counterparts ( blain et al . 2007 ) will also provide an important comparison of spectroscopic and photometric redshifts .
the cosmological parameters adopted throughout this paper are @xmath8 kms@xmath9mpc@xmath9 , @xmath10 , @xmath11 .
despite having mapped only @xmath12 per cent of the planned 0.5 sq . degree area , shades remains the largest extragalactic sub - mm survey to date .
the difficulties of following - up such large areas at other wavelengths , and hence the inhomogeneity of the multi - wavelength data , implies that the same photometric redshift technique can not be applied to all sources .
this section has been divided in two subsections : the first ( 2.1 ) deals with the consideration of 850@xmath0 m and 1.4ghz photometry which is available for all sources , and the use of the sub - mm radio spectral - index as a diagnostic of redshift ; and the second ( 2.2 ) describes the inclusion of additional photometry at 70 to 450@xmath0 m which is sufficiently sensitive to place important constraints on the radio - mm - fir photometric redshifts for only a few tens of sources . in both subsections
we make a brief introduction to the techniques used , the estimated uncertainties found when comparing photometric and spectroscopic redshifts for similar sub - mm galaxies , the results from the application of the techniques to individual shades sources , and the combined redshift distributions derived for the entire shades population .
one of the simplest redshift - indicators for the sub - mm galaxy population is that formed by the ratio of the flux densities at 1.4ghz and 850@xmath0 m .
these wavebands trace the tight correlation between radio continuum emission , which is dominated by synchrotron radiation from supernova remnants , and thermal emission from warm dust heated by young stars ( helou , soifer & rowan - robinson 1985 , condon 1992 , yun , reddy & condon 2001 ) .
this redshift indicator was systematically studied by carilli & yun ( 1999 , 2000 ) , and has been subsequently revised for different sub - mm galaxy sub - populations ( dunne , clements & eales 2000 , rengarajan & takeuchi 2001 ) .
the 1.4ghz to 850@xmath0 m flux - density ratio , or a spectral index derived from it , increases monotonically with redshift , with some degeneracy due to the variety of radio synchrotron - slopes and mm dust - emissivity indices present in the ism of those local galaxies used to define the relationship . additionally there exists a level of degeneracy between the temperature of the dust generating the rest - frame fir luminosity ( and hence sub - mm flux ) and the redshift .
regardless , by adopting a library of local galaxy templates , and accepting the intrinsic dispersion in their seds , the 1.4ghz to 850@xmath0 m flux - density ratio still provides a crude but useful estimation of the redshift .
this indicator becomes relatively insensitive to redshift beyond @xmath13 , as the 850@xmath0 m filter starts to sample the flattening of the spectral energy distribution ( sed ) towards the rest - frame fir peak , whilst still providing a powerful discriminant between low - redshift ( @xmath14 ) and high - redshift ( @xmath15 ) objects .
we shall discuss the 1.4ghz/850@xmath0 m spectral index following two different prescriptions : ( a ) the single - template maximum - likelihood technique originally designed by carilli & yun ( 1999 , 2000 ) , denoted as @xmath16 ; and ( b ) a maximum likelihood technique which simultaneously fits the 20 local templates of starbursts , ulirgs and agn used by aretxaga et al .
( 2003 , 2005 ) , denoted as @xmath17 .
the success of any photometric - redshift technique is measured by the accuracy with which it can predict the individual redshifts for a sample of representative galaxies with known redshifts , which have not been used to define the method .
aretxaga , hughes & dunlop ( 2006 ) have previously assessed the accuracy of the above two 1.4ghz/850@xmath0 m photometric redshift indicators . based on this study ,
we show in figure1 a comparison of spectroscopic and photometric - redshifts for 58 sub - mm / mm selected galaxies , complemented with a few objects selected at optical / fir wavelengths , which have published optical / ir or co spectroscopic redshifts and accompanying radio - fir photometry
. we will refer to this dataset as the ` comparison sample ' hereafter .
this comparison study shows that the @xmath18 prescription has a mean accuracy @xmath19 over the whole redshift interval , when one selects a robust sub - sample of objects with unambiguous optical / ir / radio counterparts and spectroscopic redshifts derived from the identification of two or more spectral lines .
for the same robust sample of objects , @xmath20 has systematically larger errors , @xmath21 .
this sample does not include powerful radio - loud agn , for which the template seds used in the photometric redshift analysis are not appropriate .
the r.m.s . of the relation is @xmath22 .
restricting the analysis only to those galaxies with co spectroscopic redshifts , the measured accuracy is @xmath23 and has an r.m.s of 0.8 at @xmath24 .
the precision degrades as the redshift increases , as expected from the 1.4ghz/850@xmath0 m spectral index , which flattens beyond @xmath25 ( carilli & yun 2000 ) , leading to a measured @xmath26 at @xmath27 .
using all objects with published photometry and spectroscopic redshifts , regardless of whether the associations that lead to the spectroscopic redshift are unambiguous or not , the overall accuracy over the @xmath28 regime degrades to @xmath29 ( see aretxaga et al .
2006 , figure 1 ) .
the radio counterparts adopted for the photometric redshift calculations of shades sources are the secure sample detected within 8 arcsecs of the sub - mm position ( ivison et al .
2007 ) , with a chance - association probability between the radio and sub - mm source of @xmath30 .
we have accepted some additional counterparts when a robustly - detected radio source is still within 10 arcsec of the sub - mm centroid and , additionally , a 24@xmath0 m counterpart is associated with this radio identification .
these extra radio counterparts are marked in the notes provided for each sub - mm source ( see tables [ tab : cy_lh ] and [ tab : cy_sxdf ] ) , where we have calculated the corresponding @xmath31-value of the radio association , which remains lower than 0.08 .
the 34 radio sources adopted as counterparts of shades galaxies in the lh field , and the 35 radio sources in the sxdf have a combined chance association @xmath32 , and thus we expect to have incorrectly associated @xmath33 of the shades sub - mm sources with a projected radio source . for both techniques
the error bars of the photometric redshifts were derived by bootstrapping on the reported photometric and calibration errors ( coppin et al .
2006 , ivison et al .
2007 ) , and are defined as the 68% confidence interval of the resulting redshift probability distribution .
the photometric error distributions used for the 850@xmath0 m photometry were derived by de - boosting the measured flux densities of the scuba sources .
a de - boosting correction is necessary to provide a more accurate estimate of the flux of low s / n blank - field sources in sub - mm surveys , where the counts are typically very steep and faint galaxies can be statistically boosted above the nominal detection threshold .
these errors are often non - gaussian ( see figure 5 in coppin et al 2006 ) .
the 1.4ghz flux densities do not need to be de - boosted , since this correction is dependent on the area defined by the search radius in identifying the source . in the case of finding associations within @xmath34 arcsec radius around a known object , this is negligible .
the error distributions for the 1.4ghz flux densities were assumed to be gaussian . in the case of @xmath35 , the error estimated by carilli & yun ( 2000 ) , to allow for a difference in templates ,
is added in quadrature to the errors derived by bootstrapping the photometry .
the probability distribution calculated for each source in this manner typically have a single peak , which broadens as the most probable redshift of the source increases .
figure 2 shows an example of a typical solution derived from the use of 850@xmath0 m and 1.4ghz photometry .
we have assembled the best estimates of photometric redshift for each source in the lh and sxdf shades fields ( figure 3 ) .
these are identified by the modes of the individual probability distributions , @xmath36 , which , by definition , are the redshifts with the highest probability values .
the galaxies that are not detected with confidence at radio - wavelengths , _
i.e. _ 26 out of the 60 galaxies in the lh , and 25 out of the 60 in the sxdf , have very flat individual redshift probability distributions in our computations , and hence we quote only the lower - limits to their redshift , which are defined as their 90% lower confidence - limits ( tables 1 and 2 ) .
we incorporate these objects into the population distribution by adding flat probability distributions between their calculated lower 90% confidence limits and @xmath37 , and alternatively between their calculated lower 90% confidence limits and @xmath38 , or only at their lower redshift - limits if these indicate @xmath15 .
these two alternative priors illustrate how the resulting redshift distributions ( that include shades galaxies without radio detections ) can be biased high and low .
figure [ fig : zcumcy_a ] shows the final photometric redshift distribution using the 1.4ghz/850@xmath0 m spectral index , both for the full shades sample and for the lh and sxdf fields separately .
the redshift distribution of sxdf sources peaks at slightly lower redshifts ( median @xmath39 ) compared to the distribution of lh sources ( median @xmath40 ) . the low - redshift tail ( @xmath41 ) is also slightly more prominent in the sxdf than in the lh ( 6 vs. 1 sources among the radio - detected sample ) .
the difference in shape of the two distributions of radio - detected sources can be measured using the two - tailed kolmogorov - smirnov ( k - s ) test , which gives a 3 per - cent chance that they are drawn from the same parent distribution .
furthermore , the mean - redshifts of the two distributions are significantly different at the 99.7 per cent level , according to a mann - whitney @xmath42-test .
this difference can be attributed to differences in the noise levels of the radio maps , although some intrinsic variations between the fields are expected ( see 3 ) . for those objects with more than one redshift estimate ( due to the ambiguity in their counterparts ) ,
we have produced alternative population distributions , with or without their inclusion , and with different combinations of possible counterparts .
the results do not significantly change the final combined distribution .
all figures presented in this paper include the primary radio association if there are multiple options .
the shades fields have been targeted by other fir / sub - mm / mm and radio surveys . in this subsection
we describe the extra constraints on the photometric redshifts that can be derived from these additional complementary data for a few tens of shades sources .
photometric redshifts with modest precisions ( @xmath43 to 0.5 ) have been obtained in the past few years using a combination of spectral indices between the radio and mm - wavelength regimes and the fir spectral peak .
this information has been exploited by several groups using a wide array of fitting - techniques and seds ( e.g. yun & carilli 2002 , hughes et al .
2002 , aretxaga et al .
2003 , 2005 , wiklind 2003 , hunt & maiolino 2005 , laurent et al . 2006 ) .
there remain , however , degeneracies imposed by the choice of multiple sed templates , with fir emission peaks distributed over a range of wavelengths , which can limit the precision of the derived redshifts ( e.g. blain , barnard & chapman 2003 ) .
we have previously developed a radio mm fir technique based on monte - carlo simulations , that take into account constraining prior information such as the number counts of sub - mm galaxies , the favoured luminosity / density evolution up to @xmath44 , and the lensing amplification of a certain field ( hughes et al .
2002 , aretxaga et al .
2003 , 2005 ) .
we only offer a brief summary of this technique here .
a catalogue of 60@xmath0 m luminosities and redshifts for mock galaxies is generated from an evolutionary model for the @xmath45 m luminosity function that fits the observed 850@xmath0 m number - counts ( e.g. luminosity evolution @xmath46 for @xmath47 , and no evolution at @xmath15 ) and covers a simulated area of 10 sq .
template seds are drawn at random , without regard to their intrinsic luminosity , from a library of 20 local starbursts , ulirgs and agn , to provide fir
radio colours for the mock galaxies .
the seds cover a wide - range of fir luminosities ( @xmath48 ) and temperatures ( @xmath49 ) .
the flux densities of the mock galaxies include both photometric and calibration errors , consistent with the quality of the observational data for each sub - mm galaxy detected in a particular survey .
we reject from the catalogue those mock galaxies that do not respect the detection thresholds and upper - limits of the particular sub - mm galaxy under analysis .
the redshift probability distribution of an individual sub - mm galaxy is then calculated as the normalized distribution of the redshifts of the mock galaxies in the reduced catalogue , weighted by the likelihood of identifying the colours and flux densities of each mock galaxy with those of the sub - mm galaxy in question .
this technique will be denoted @xmath50 in the discussion that follows .
the validity of the results derived from this technique is limited by the assumption that the seds of high-@xmath51 sub - mm galaxies are similar to the local analogues , adopted as templates , which are scaled in luminosity and shifted in redshift . while this might seem a naive approach , all the templates used in the calculations that follow offer a good description of the radio mm fir photometry of scuba galaxies , including 350@xmath0 m observations ( laurent et al .
2006 , kovcs et al 2006 ) , with known spectroscopic redshifts and un - ambiguous multi - wavelength counterparts ( aretxaga et al . 2005 ) .
there are however a few sub - mm galaxies which do not match any of the templates we use in this paper at their published redshifts ( see figure 4 in aretxaga et al .
2006 ) . in these examples
their radio emission is higher than that implied by the radio fir correlation , possibly due to accretion activity , or their fir emission peaks at wavelengths longer than those of the templates used in this study at the adopted redshift .
we describe the redshift solutions for these galaxies as ` catastrophic ' and this might be indicative of incompleteness in the library of sed templates used as analogues .
there is still sufficient debate in the literature , however , about the nature and/or the ambiguity of the multi - wavelength counterparts to these sub - mm sources , from which the redshifts are derived , to justify their exclusion from a robust comparison sample ( see aretxaga et al .
2005 , 2006 , laurent et al .
2006 , kovcs et al .
2006 for a detailed complementary discussion on these galaxies ) . in order to estimate the accuracy of the @xmath50 technique we use the full sed information of a robust sub - sample of 11 sub - mm galaxies , out of the comparison sample of 58 galaxies considered in
2.1.1 , which have detections in three or more bands . furthermore
, the same galaxies have undisputed identifications of their optical / ir / radio counterparts and spectroscopic redshifts derived from the measurement of two or more spectral lines .
we derive a mean accuracy for this sub - sample of @xmath52 and an r.m.s .
@xmath53 over the whole redshift interval ( figure 4 ) .
using all objects with published photometry , regardless of whether the spectroscopic redshift derived from the optical associations is ambiguous or not , the overall accuracy over the @xmath28 regime degrades to @xmath54 , with an r.m.s . of 0.80 ( see figure 3 in aretxaga et al .
a few significant outliers which remain in the correlation are discussed by aretxaga et al .
( 2005 ) and kovcs et al .
( 2006 ) . within the small sub - sample of study ,
the accuracy is independent of redshift .
if we restrict the use of photometry to 450@xmath0 m upper limits combined with the 1.4ghz and 850@xmath0 m detections for the comparison sample of galaxies ( adopting simulated 450@xmath0 m upper limits , when necessary , to mimic a shallower survey at this wavelength ) , we find a mean accuracy of @xmath55 and an r.m.s .
of 0.7 for the robust sample .
this result is especially relevant for the photometric redshift calculations of the shades sources in [ sec : pzallsed ] , the majority of which have similarly sparsely - sampled photometry . considering only the complete sample with robust and tentative
spectroscopic redshifts the mean accuracy degrades slightly to @xmath56 , and the r.m.s .
scatter is 0.90 .
table 3 summarizes the most recent photometric redshifts calculated with the monte carlo technique for shades sources with additional photometry published in the literature .
in contrast , and for completeness , tables 4 and 5 list the photometric redshifts derived only from the combination of the shades 450/850@xmath0 m and the 1.4ghz photometry using two approaches : the monte carlo based technique , @xmath50 described above , and a non - prior maximum likelihood fit to the same 20 seds used for the first method that includes survival analysis ( isobe , feigelson & nelson 1986 ) to incorporate the non - detections into the maximum likelihood formalism , @xmath57 .
this second technique is introduced to provide a comparison of how the priors affect the redshift estimation of the sources and the final combined redshift distribution of shades galaxies .
while most of the photometric redshifts derived from the two methods are similar , the pure survival analysis produces a few high-@xmath51 catastrophic results in the robust comparison sample ( 2 out of 11 ) .
the overall reliability of the maximum likelihood technique is @xmath58 .
these high-@xmath51 catastrophic solutions get suppressed by the mc technique due to the weighting priors that disfavour high-@xmath51 solutions for these sources , since , if they were typical of the sub - mm population , they would overproduce the 850@xmath0 m number counts under the assumed luminosity evolution model .
although we give the values of photometric redshifts with and without priors in tables 4 and 5 , we will now continue the analysis of the complete shades sample using only the mc solutions , since they have been shown to perform better against the comparison sample . for 5 sources in table5 , sxdf850.5 , 21 , 28 , 77 and 119
, we also include complementary photometry at 70 and 160@xmath0 m from the spitzer legacy survey swire ( lonsdale et al .
2003 , surance et al .
2007 ) that are used to derive mid - ir counterparts to the shades sources ( clements et al . 2007 ) .
the remainder of the shades sources are not significantly detected ( @xmath59 ) in the spitzer catalogues , and the noise properties of the swire maps , providing @xmath60 upper - limits of @xmath6123 mjy and 160 mjy at 70 and 160@xmath0 m , respectively ( afonso - luis et al . 2007 ) , do not further constrain the photometric redshifts .
figure 5 shows the single - peaked redshift probability distribution derived for one of the sources that have the most complete photometric data .
although the majority of the sources show similar probability distributions , there are sometimes secondary peaks ( see examples in aretxaga et al .
2003 ) . nevertheless , it is always the primary redshift - peak that defines the solutions given in tables 3,4,5 .
figure 6 shows our final photometrically - derived redshift distribution for the shades sources , using our best available estimate for the redshift of each source ( i.e. @xmath50 taken from table 3 for those sources with the most complete photometry , and from tables 4 and 5 for the remainder ) the distribution of radio - identified shades sources clearly peaks in the bin @xmath62 , with a 50 per cent interquartile interval @xmath63 .
we incorporate the non - radio detected sources ( lower panel in figure 6 ) into the population distribution in two alternative ways , to serve as examples of how much these sources could alter the final population distribution : ( a ) approximating their individual probability distributions as flat distributions between their lower 90% confidence limit and @xmath37 ; and ( b ) as flat distributions between their lower 90% confidence limit and @xmath38 , or at their lower limit if this lies at @xmath15 .
solution ( a ) is actually derived from the adopted non - informative ( flat ) prior for the photometric - redshift calculations .
this creates a high-@xmath51 tail which is a reflection of the adopted range for the flat redshift distributions which are unconstrained by the photometry .
solution ( b ) is biased against high-@xmath51 , by imposing a maximum redshift for the radio - undetected sample which is lower than the redshift of the peak of the radio - detected sample .
this radio - undetected sample could be composed of colder sub - mm galaxies than those found in the template library , or they could have the same template shapes as those adopted in the photometric redshift analysis , and still be undetected at the depth of the present radio surveys .
regardless , in these alternative solutions , the mode of the population remains at @xmath62 , with at least 50 per cent of the galaxies in the interquartile range @xmath64 . in order to consider the effect of the objects with more than one redshift estimate ( due to ambiguity in their radio - counterparts ) ,
we have produced alternative population distributions .
for instance , figure 6 shows the combination of the first entries for each source in tables 3 , 4 , 5 .
the introduction of the second tabulated values instead of the first ones , for those sources with ambiguous associated photometry , produces an alternative distribution which is indistinguishable ( with a 99.96 per - cent probability ) , via a kolmogorov - smirnov ( k - s ) test , from the one represented here . as in the case of the photometric redshift distribution derived from the 1.4ghz/850@xmath0 m spectral index , the distribution of redshifts derived from the full sed analysis of sxdf sources peaks at slightly lower redshifts ( median @xmath39 ) than that of lh sources ( median @xmath65 ) , and its low - redshift tail ( @xmath41 ) is also more prominent .
these differences in the distributions are statistically - significant , as indicated by a k - s test at a level of 98.9% .
a mann - whitney u - test shows that their mean - redshifts differ at the 99.997% level .
shades was designed with the objective of constraining the redshift distribution and clustering properties of the sub - mm galaxy population , an exercise which van kampen et al .
( 2005 ) demonstrated could discriminate between galaxy formation models . with @xmath6140 per cent of the survey completed
before scuba was de - commissioned in the summer 2005 , shades has provided 120 robust sources .
the radio - mm - fir photometry assembled for the survey favours a nearly gaussian redshift distribution of the population peaking at @xmath62 , albeit still with the possibility of a high - redshift tail remaining .
the photometric redshift distribution of the radio - detected sub - mm galaxies is qualitatively similar to the optical spectroscopic redshift distribution published by chapman et al .
( 2003 , 2005 ) who followed - up a sample of sub - mm galaxies derived from various surveys .
the agreement is perhaps not surprising , given that the photometric redshifts of the comparison sample have shown a relatively good agreement with the spectroscopic redshifts published in the literature ( aretxaga et al .
2006 , 2.1.1 , 2.2.1 ) .
furthermore , the majority of the sub - mm sources that have spectroscopic redshifts are drawn from scuba surveys of similar depths to shades .
the high - redshift correction applied to the measured spectroscopic redshift distribution , suggested by chapman et al .
( 2005 ) to account for the bias introduced by non - detection of the higher - redshift radio counterparts that provide candidates for optical spectroscopic follow - up , also falls within the range of photometric redshift estimations we have derived for shades sources that are not detected at radio - wavelengths .
these sources provide the high - redshift ( @xmath66 ) tail of figure 6 , and could in fact be placed anywhere above @xmath67 , even producing secondary peaks .
our calculations do not support the existence of a substantial low - redshift ( @xmath68 ) tail within the luminous sub - mm radio - detected population sampled by shades . at first sight
this might appear to be in conflict with the results of pope et al .
( 2005 , 2006 ) who found that @xmath69% of the sub - mm sources found in the scuba imaging of the goods - north field may lie at @xmath68 .
if the goods - north 850@xmath0 m catalogue is restricted to sources with de - boosted flux densities @xmath70mjy , however , then the proportion of robustly identified sub - mm galaxies which lie at @xmath71 drops to 6% .
this is entirely consistent with the results found here for shades galaxies .
thus , these results may be providing further evidence that the peak of the redshift distribution of sub - mm sources is positively correlated with sub - mm flux - density / luminosity , consistent with the apparently anti - hierarchical nature of star - formation history reported in several other recent studies ( e.g. heavens et al .
furthermore field - to - field variations in the spatial distribution of the large - scale structure can provide a simple explanation for the differences between the redshift distributions of sub - mm sources derived from the individually - mapped contiguous - areas ( typically @xmath72 sq . degrees ) taken from the current generation of scuba surveys .
our analysis also suggests that only a modest fraction of sub - mm galaxies could be hiding in the optical redshift - desert at @xmath73 during spectroscopic searches for shades sources with robust radio counterparts .
the photometric redshift probability density distributions of radio - detected shades sources using the 1.4ghz/850@xmath0 m index or the full radio mm fir sed information contain @xmath74 % and @xmath75 of sources in this redshift desert regime , respectively . the difference between the redshift distribution of sub - mm sources in the sxdf and lh fields is entirely consistent with the different properties of the 1.4-ghz maps as discussed by ivison et al .
the lh data have a higher vla resolution than those in sxdf , and the lh data are also deeper , although the coverage is less uniform .
there is clearly potential for systematic differences between radio measurements in the lh and sxdf .
for an extended source in the lh ( of which there are several ivison et al .
2002 ) , a larger fraction of emission on scales larger than the synthesized beam will be resolved away than for similar cases in sxdf . moreover
, the lh data will suffer greater significant bandwidth smearing and , although the appropriate correction has been made to the measured flux densities , some faint sources will be lost below the radio - detection threshold and may receive misleadingly low flux - density limits .
these effects can be viewed as a systematic flux calibration offset with consequences as severe as those encountered in optical / infrared photometric - redshift estimation .
while random 1.4ghz calibration uncertainties of 5 per cent have been accounted for in the estimation of the photometric redshifts , a systematic flux - density offset could shift the redshift distribution significantly . in order to explore this possibility
, we have applied a 10 per cent flux increase to the lh photometry and recalculated the photometric redshifts .
the combined redshift distribution shifts its peak by @xmath76 , and consequently the mean values of the sxdf and lh distributions are more consistent , increasing from 0.3 per cent ( 2.1.3 ) to a 7 per - cent probability , according to a mann - whitney @xmath42-test .
some intrinsic variation on the distribution of redshifts between the fields is however to be expected ( see below ) .
van kampen et al .
( 2005 ) studied four different galaxy formation models that yielded different redshift distributions and clustering properties for the sub - mm population expected to be found in a survey of the depth and area covered by shades : ( @xmath77 ) a hydrodynamical model ( muanwong et al .
2002 ) , that follows the evolution of dark - matter , gas , star - like particles and galaxy fragments , that has been coupled with the analytical form for redshift distribution of baugh , cole & frenk ( 1996 ) ; ( @xmath78 ) a simple merger model that identifies sub - mm galaxies with major mergers of massive galaxies ; ( @xmath79 ) a phenomenological model ( van kampen 2004 ) , which is based on @xmath80-body simulations that identify the sites of major - mergers and has two modes of star formation , quiescent and bursting ; and ( @xmath81 ) a stable clustering model ( gaztaaga & hughes 2001 ) .
figure 7 represents the theoretical redshift distributions of shades galaxies found in these models .
this figure has been complemented with ( @xmath82 ) a semi - analytical model for the joint formation and evolution of spheroids and qsos ( granato et al .
2004 , silva et al .
2005 ) ; and ( @xmath83 ) an alternative semi - analytic model of galaxy formation for sub - mm galaxies ( baugh et al .
2005 ) .
furthermore , to enable a more accurate discrimination between the above predictions , all the galaxy formation models in figure7 account for the incompleteness of sources in the shades catalogue ( coppin et al .
the models have also been convolved with a representative radio - mm - fir photometric precision of @xmath84 , which is intermediate between the measured uncertainties derived for the two techniques used in this paper .
we have made a comparison , via a k - s test , of the observed redshift probability density distributions with those predicted from the above models . in each case a k - s statistic has been calculated that accommodates the 1@xmath5 uncertainty in the median redshift of the models due to field - to - field variations . a study of 25 simulations made for each of the four models analyzed by van kampen et al .
( 2005 ) demonstrate that the mean redshift of @xmath85 shades - like sub - mm galaxies varies by @xmath86 ( r.m.s . )
@xmath87 . in part
these shifts can be explained by poisson noise ( estimated @xmath88 from the simulations ) .
the models show , however , that there could also be a significant component in the field - to - field variations that arises from intrinsic redshift differences due to varying amounts of groups or proto - clusters of galaxies along the line - of sight .
thus the differences found between the lh and sxdf areas , and between these and smaller , deeper surveys like goods - n , could be partially explained by this effect .
the results of the k - s test suggest that only model(@xmath82 ) is close to being formally acceptable , with an 87% probability for the model to agree with the measured probability density distribution that includes shades sources with and without radio - detections according to solution ( a ) . with only a small shift ( @xmath89 ) in the distribution , model(@xmath82 ) also qualitatively reproduces ( @xmath90 probability of similarity ) the photometric - redshift distribution of the radio - detected shades galaxies .
the shades sources in our analysis that are not detected at radio wavelengths have very flat redshift probability distributions , which simply places them at @xmath91 , and hence these shades sources could also produce a secondary peak in the redshift distribution . in the ranking of similarities of measured and model - distributions , models(@xmath77 ) and ( @xmath79 ) , @xmath92% probability , have double peaks and are broader than the observed distributions .
a different prior , that optimizes the redshift - distribution of the shades sources without radio - detections , could bring them closer to a level of formal - acceptance . finally models(@xmath81 ) , ( @xmath83 ) and ( @xmath78 ) are all rejected with probabilities of @xmath93 of being consistent with the range of solutions depicted in figure7 .
the catalogues of redshifts presented in tables 1 to 5 are an initial step towards characterizing the fir luminosities and star formation rates of the shades population .
the available photometry in the fir peak regime ( 70450@xmath0 m ) , however , is not deep enough to fully constrain the seds of most shades sources at these wavelengths .
one viable approach is to use the 20 seds in our local template catalogue to derive the corresponding fir luminosities from the 850@xmath0 m flux densities , bearing in mind that the lack of constraints at short wavelengths will dominate the errors in luminosity estimation over those of redshift ( e.g. hughes et al . 2002 ) .
alternatively , one could use the 1.4ghz radio flux density to deduce fir luminosities via the radio - fir luminosity correlation that characterizes the sub - mm galaxy population , since this now has been extended to @xmath94 ( kovcs et al .
this latter approach has the advantage of providing mean fir luminosities which are accurate for the bulk of the population , reducing the uncertainties in luminosity primarily to the accuracy of the photometric redshifts .
however , the normalization of the relation might be shifted from the local iras correlation , and this could affect the fir luminosities derived , and the comparison of these to nearby galaxies . regardless of this complication , the observed 1.4ghz flux densities have been converted to rest - frame 1.4ghz flux densities using a mean synchrotron radio slope of index @xmath95 , and the monochromatic 1.4ghz luminosity has been inferred using the photometric redshift solution for each source . this is converted to fir luminosity using the linear relationship @xmath96 ( kovcs et al .
2006 ) . for each source
we have considered the effect of the uncertainties in redshift , 1.4ghz flux and the reported scatter in the sub - mm galaxy fir - radio correlation ( kovcs et al .
2006 ) by bootstrapping 1000 times on the measured errors .
for the 69 radio - detected shades sources in the lh and sxdf fields the median fir luminosity is @xmath97 , with a high luminosity tail that extends to @xmath98 ( see figure [ fig : lum ] ) .
the distribution of luminosities for sources in both the lh and sxdf fields are similar .
the effect of non - radio - detected sources , which have very unconstrained and possibly high redshifts ( 2.1 ) , has also been considered by scaling a simple grey - body of temperature @xmath99 k and emissivity index @xmath100 , the average of a parametrized sed of the short sub - mm wavelength detected scuba galaxies that define the radio - fir correlation at @xmath101 ( kovcs et al .
2006 ) , to the observed 850@xmath0 m flux density .
the errors in the 850@xmath0 m flux density have been taken into account by bootstrapping on the inferred de - boosted distributions of 850@xmath0 m flux densities for each source , and the redshift has been selected at random between their 90 per cent lower - limits ( tables 4 , 5 ) and , arbitrarily , @xmath37 .
the resulting combined distribution has a median fir luminosity of @xmath97 , the same as the distribution of radio - detected sources .
if one returns to the alternative approach of using the combination of 20 seds with the de - boosted 850@xmath0 m flux densities to derive the fir luminosities , one derives a broader luminosity distribution than that depicted in figure 8 , reflecting the wide variety of acceptable sed templates , with a median luminosity that is increased by @xmath7 per cent .
the evolution of the global star formation rate ( sfr ) density traced by shades sources is shown in figure [ fig : sfr ] . the conversion from fir luminosity to star formation rate
was performed using a constant of @xmath102 /(yr@xmath9 ) , which is constrained to @xmath103 per cent uncertainty ( kennicutt 1998 ) .
we have multiplied the contribution of each source to the star formation density by the inverse of the shades survey completeness function at the appropriate flux density ( coppin et al .
redshift - space was binned into six intervals , and a monte carlo was performed to assign each galaxy to a redshift bin according to its expected photometric redshift error .
the fir luminosities have been estimated from the radio - fir correlation or with a single sed which is considered to be representative of the sub - mm galaxy population ( as in 3.2 ) .
the error bars in sfr density are the result of the uncertainties in photometry , seds and redshift , and are computed as the standard deviation traced by 1000 monte carlo simulations .
the evolution in the sfr density traced by the radio - detected sources shows a clear peak at @xmath104 and a slow decline at both low and high redshifts .
a good description of the sfr density traced by this population is given by @xmath105 $ ] .
the contribution of non - radio - detected shades sources to the sfr history may be significant at high - redshifts compared to the radio - detected shades sources , as indicated in figure 9 by the empty black diamonds ( solution ( a ) in figure 6 ) .
since the redshifts of these sources are not well constrained , this should be considered as only a possible evolutionary history , awaiting confirmation by better multi - wavelength data to improve the constraints on the redshifts of the radio - undetected shades sources .
the effect of placing all the radio - undetected sources between their lower redshift - limits and @xmath38 or at their lower limits if they are at @xmath15 ( solution ( b ) in figure 6 ) is shown by the grey empty diamonds in figure 9 , which understandably broadens the peak of star formation density to lower-@xmath51 . the levels of star formation deduced for shades sources are consistent with those derived for other samples of radio - detected sub - mm galaxies at @xmath106 ( chapman et al .
2005 ) . at higher - redshifts , however , the sfr densities implied in this photometric - redshift study exceed the extrapolations of chapman et al
. eventually spectroscopic measurements of shades sources at @xmath107 , via millimetre - wavelength observations of molecular co - lines , or optical spectroscopy with increased sensitivity , will provide a definitive measurement of the sfr density of obscured galaxies in the high - redshift universe . in order to estimate the contribution to the global star formation rate density of @xmath108 m sources that are fainter than those detected by shades ,
we have adopted the @xmath45 m luminosity function with a pure luminosity evolution that follows @xmath109 at @xmath110 , and then maintains a constant level for @xmath111 .
this evolutionary form provides an adequate description of the 850@xmath0 m and 1.1 mm counts ( e.g. scott et al .
2002 , greve et al . 2004 ) and is also supported up to @xmath112 by the evolution of the luminosity function of 24@xmath0m selected galaxies ( le floch et al .
2005 ) . by implementing this luminosity - function correction
the sfr density increases up to a maximum factor of 2 at @xmath15 .
the contribution of shades galaxies to the global star formation rate density of the universe is comparable to the contribution of starbursts selected at optical / uv wavelengths at @xmath113 before the latter are corrected for dust extinction .
it is important to recall that the @xmath114sfr factor we have adopted could be in error by @xmath103 per cent , and that sed differences could also account for an increase of @xmath7 per cent .
these uncertainties have not been carried into the estimation of error - bars in figure 9 .
if we complete the luminosity function of shades galaxies towards lower luminosities , the fir star formation rate traced by ultraluminous starbursts is still a factor of 1.2 to 2 lower than that of optical / uv starbursts that have been corrected for intrinsic dust extinction .
the recent demonstration that the contribution of luminous and ultraluminous ir galaxies dominates the sfr density at @xmath115 ( le floch et al .
2005 ) suggests that even if we correct for the incomplete sampling of the sub - mm galaxy luminosity function , a bright shades sub - mm survey could be missing the integrated contribution of dusty starbursts to the global sfr by a large factor ( as high as @xmath116 ) , and thus dusty starbursts could indeed prove to be a significant mode of the star formation of the universe ( blain et al .
1999 ) .
the contribution of the fainter ( @xmath117mjy ) 850@xmath0 m sub - mm galaxy population to the star formation history of the universe at @xmath118 remains unconstrained at present , since detecting faint sub - mm galaxies has been restricted to a few strongly - lensed fields ( e.g. smail , ivison & blain 1997 ) and extremely - deep confusion - limited pencil observations ( e.g. hughes et al . 1998 ) .
our estimations presented in figure 9 should therefore be considered an educated estimate of how ultraluminous ir - submm galaxies trace the star formation history of the universe .
accurate measurements of the surface density and redshift distribution of the entire sub - mm galaxy population ( and the faintest galaxies in particular ) that contribute the complete sub - mm to fir extragalactic background require deeper and larger mm and sub - mm surveys than are currently possible . the anticipated continuum and spectroscopic surveys with scuba-2 ( holland et al .
2006 ) , the large millimetre telescope ( lmt , serrano et al .
2006 ) , and the atacama large millimeter array ( alma , beasley , murowinski & tarenghi 2006 ) , for example , will provide suitable data .
in the meantime , however , it is still possible that the redshift distribution of the more populous and fainter ( possibly extremely high - redshift or alternatively less luminous ) sub - mm galaxies is significantly different to those galaxies identified in the shades survey .
* we have derived the photometric redshift distribution of shades sources with de - boosted 850@xmath0 m flux densities @xmath119 mjy towards the lockman hole ( lh ) and the subaru / xmm newton deep field ( sxdf ) using rest - frame radio to fir photometry . the redshift distribution of the radio - detected sub - mm sources peaks at @xmath120 with 50 per cent of the population between redshifts 1.8 and 3.1 . *
the combined redshift distribution of shades sources with robust radio counterparts , @xmath85 per cent of the population , has a distribution which is qualitatively consistent with the distribution of rest - frame uv - optical spectroscopic redshifts published by chapman et al .
* we find a small ( @xmath121 ) , but significant , difference between the peaks of the photometric - redshift distributions of the lh and sxdf , which can be attributed to differences in the sensitivities of their respective radio maps .
intrinsic field - to - field redshift variance is also expected , and is characterized according to a variety of models ( @xmath122 ) .
this drives us to the conclusion that the incomplete area ( @xmath123 sq .
arcmins ) observed by scuba , despite being the largest sub - mm survey to date , may still be too small to be a representative sample of the bright sub - mm galaxy population . *
the complete redshift distribution of all shades sources , including those sub - mm sources without detections at radio wavelengths ( for which we adopt a variety of possibilities that describe their unconstrained distribution of redshifts ) still maintains the peak ( mode ) of the bright sub - mm galaxy redshift distribution at @xmath124 .
we have considered a variety of priors that describe the unconstrained redshift distributions of the sub - mm sources without radio detections . in the most extreme solutions , distributing these sources with equal probability between their lower redshift - limit and @xmath38 or @xmath37 , the bulk of the sub - mm population ( 50 per cent interquartile ) lies in the range @xmath125 or @xmath126 , respectively . * the combined shades lh and sxdf redshift probability - density distribution is compatible , within the uncertainties of our analysis , with the semi - analytical model for the joint formation of spheroids and qsos of granato et al .
( 2004 ) and silva et al .
if sources detected only at 850@xmath0 m are also introduced into the redshift probability - density , with other priors than those illustrated here , then the hydrodynamical model of muanwong et al .
( 2002 ) and phenomenological model of van kampen ( 2004 ) and van kampen et al .
( 2005 ) could also be in agreement with the observations . these compatible models , which are physically quite distinct , predict different clustering properties for the shades galaxies that could allow further discrimination between them ( van kampen et al . 2005 ) .
a detailed study of the clustering properties of shades galaxies will be the topic of a further paper . * the bright shades galaxies contribute to the sfr density of the universe with @xmath127 to 0.03 yr@xmath9 mpc@xmath128 in the redshift interval @xmath129 , and reach the levels of the dust - uncorrected lyman break galaxy population ( giavalisco et al .
2004 ) . the sfr density of dust - enshrouded starburst galaxies traced by ultraluminous shades galaxies , and completing the luminosity function to lower luminosity galaxies , is estimated to be a factor of 2 larger .
this is still a factor of 1.2 to 2 lower than the optical / uv - selected starburst galaxy samples that include the latest dust - correction estimates .
the current shades survey and complementary multiwavelength data , however , can not characterize the bulk of the rest - frame fir emission arising from these lower luminosity galaxies . a more statistically - complete measurement of the universal history of star - formation from powerful dusty , optically - obscured galaxies awaits the commissioning of future large - aperture single - dish and interferometric submillimetre and millimetre telescopes targetting suitable extragalactic fields that have the necessary multi - wavelength ancillary data .
ia and dhh acknowledge partial support from conacyt grants 39548-f and 39953-f and a royal society academia mexicana de ciencias 2005 exchange program .
this work has been also partially funded by pparc .
kc acknowledges additional partial support from nserc , irs from the royal society , ajb from the gordon and betty moore foundation , ms from a gary mccue fellowship through the center of cosmology at uc irvine , and cs from a pparc advanced fellowship .
gaztaaga e. & hughes d.h . , 2001 , in lowenthal j.d .
, hughes d.h .
, eds . , proc .
umass / inaoe conf .
, deep millimeter surveys : implications for galaxy formation and evolution . world scientific pub . | we present the redshift distribution of the shades galaxy population based on the rest - frame radio - mm - fir colours of 120 robustly detected 850@xmath0 m sources in the lockman hole east ( lh ) and subaru xmm - newton deep field ( sxdf ) . the redshift distribution derived from the full sed information
is shown to be narrower than that determined from the radio submm spectral index , as more photometric bands contribute to a higher redshift accuracy .
the redshift distribution of sources derived from at least two photometric bands peaks at @xmath1 and has a near - gaussian distribution , with 50 per cent ( interquartile range ) of sources at @xmath2 .
we find a statistically - significant difference between the measured redshift distributions in the two fields ; the sxdf peaking at a slightly lower redshift ( median @xmath3 ) than the lh ( median @xmath4 ) , which we attribute to the noise - properties of the radio observations .
we demonstrate however that there could also be field - to - field variations that are consistent with the measured differences in the redshift distributions , and hence , that the incomplete area observed by shades with scuba , despite being the largest sub - mm survey to date , may still be too small to fully characterize the bright sub - mm galaxy population .
finally we present a brief comparison with the predicted , or assumed , redshift distributions of sub - mm galaxy formation and evolution models , and we derive the contribution of these shades sources and the general sub - mm galaxy population to the star formation - rate density at different epochs .
[ firstpage ] surveys galaxies : evolution cosmology : miscellaneous infrared : galaxies submillimetre |
if the up and down quark masses are zero and the strange quark mass is not , the transition between quark - gluon plasma and hadronic matter may be first or second order at zero baryon chemical potential .
if the up and down quark masses are small enough , there may be a phase transition at sufficiently large baryon chemical potential .
this phase transition is predicted to be in the same universality class as liquid - gas phase transitions and the 3d ising model .
diverse studies suggest that there is a curve of first - order phase transition in the @xmath0-@xmath1 plane that terminates in a second - order phase transition at some critical point @xmath2 .
the location of this chiral critical point has been estimated using various effective field theory models , such as the namu jona - lasinio model @xcite-@xcite , a composite operator model @xcite , a random matrix model @xcite , a linear @xmath3 model @xcite , an effective potential model @xcite , and a hadronic bootstrap model @xcite , as well as various implementations of lattice qcd @xcite-@xcite .
reviews have been written by stephanov @xcite and mohanty @xcite .
this subject is of experimental interest because collisions between heavy nuclei at medium to high energy , such as at the future facility for antiproton and ion research ( fair ) , or in future low energy runs at the relativistic heavy ion collider ( rhic ) , may provide experimental information on the phase diagram in the vicinity of a critical point .
one characteristic signature would be large fluctuations on an event by event basis @xcite-@xcite .
the goal of this work @xcite is to understand the equation of state of qcd near the chiral critical point and some of its implications for high energy heavy ion collisions . among the requirements are the incorporation of critical exponents and amplitudes and to have sensible limits as temperature @xmath4 and baryon chemical potential @xmath5
this is accomplished by parametrizing the helmholtz free energy density as a function of @xmath1 and baryon density @xmath6 in an appropriate fashion .
perhaps the closest work that addressed some of these issues blended a parameterization of the 3d ising model equation of state into quark and hadron equations of state @xcite .
these two parameterizations can perhaps be viewed as alternatives which provide some idea as to the range of uncertainty in how to describe matter near the chiral critical point .
numerous studies suggest that the curve in the @xmath1 vs. @xmath0 plane separating the quark - gluon and hadron phases is approximately quadratic .
@xmath7 see figure 1 . the parameters @xmath8 mev and @xmath9 mev were chosen to match theoretical estimates and experimental data for the transition between the two phases at @xmath10 and @xmath11 , respectively .
in this example the critical point is chosen to have the temperature @xmath12 mev . along the solid portion of the curve
the transition is first order .
it terminates at a second order transition at the critical point .
the dashed portion of the curve indicates that there is no true thermodynamic phase transition separating the phases , only a rapid crossover . .
the parameterization is given in the text . ] .
the parameterization is given in the text . ]
we will need the values of the pressure @xmath13 , energy density @xmath14 , baryon density @xmath15 , and entropy density @xmath16 at the critical point .
the parametrization of the equation of state near the critical point , to be given below , is normalized to these values .
let us assume that the high energy density equation of state can be parameterized as @xmath17 the dimensionless coefficients @xmath18 are adjusted to match a free gas of gluons and 2.5 flavors of massless quarks ( the 1/2 to approximately take into account the mass of the strange quark ) .
the dimensionless coefficients @xmath19 are adjusted to reproduce lattice results near the crossover when @xmath20 @xcite-@xcite , and to make the pressure a constant along the critical curve when ( [ critcurve ] ) is inserted .
the resulting equation of state for @xmath20 and @xmath21 is shown in figure 2 .
given the helmhotz free energy density @xmath22 the other thermodynamic quantites follow uniquely from it , as summarized below .
@xmath23 these satisfy the thermodynamic identity @xmath24 .
it is useful to define the dimensionless variables @xmath25 and @xmath26 which are a measure of distance from the critical point in temperature and density . near the critical point
the heat capacity @xmath27 has the divergent behavior @xmath28 while the thermal conductivity @xmath29 and baryon susceptibility @xmath30 have the divergent behavior @xmath31^{-1 } \rightarrow \left\ { \begin{array}{ll } \kappa_- ( -t)^{-\gamma } & \mbox{when $ t \rightarrow 0 ^ -$ } \\ \kappa_+ t^{-\gamma } & \mbox{when $ t \rightarrow 0^+$ . } \end{array } \right.\end{aligned}\ ] ] along the coexistence curve the density difference between the higher density liquid phase ( quarks and gluons ) and the lower density gas phase ( hadrons ) goes to zero as @xmath32 along the critical isotherm the pressure behaves as @xmath33 here @xmath34 , @xmath35 , @xmath36 and @xmath37 are critical exponents .
these exponents are related by @xmath38 and @xmath39 .
mean field theories normally give @xmath40 , @xmath41 , @xmath42 , and @xmath43 .
typical fluids are measured to have @xmath44 , @xmath45 , @xmath46 , and @xmath47 @xcite .
the 3d ising model has @xmath48 , @xmath49 , @xmath50 , and @xmath51 @xcite , which are the values used here . furthermore
, the critical amplitudes are related by @xmath52 and @xmath53 , which is universal to all theories within the same class .
see @xcite .
the challenge now is to parametrize @xmath54 so that it yields the correct critical behavior near @xmath55 and still gives sensible results away from the critical point .
mean field theories normally result in a taylor series expansion in integral powers of @xmath56 with coefficients that depend on @xmath57 .
the simplest way to incorporate the correct critical behavior is with a modified expansion of the form @xmath58 the power @xmath59 .
higher powers of @xmath56 could be included but they would not affect the critical behavior and so are discarded here for simplicity .
the coefficient functions must have the following behavior .
@xmath60 @xmath61 @xmath62 here @xmath63 and @xmath64 are smooth functions of @xmath57 with @xmath65 .
they are parametrized so that @xmath22 has sensible limits at @xmath11 and at @xmath66 .
the coefficient @xmath67 is assumed to be constant and proportional to @xmath13 ; for definiteness we shall take @xmath68 mev/@xmath69 .
results for the critical curves , latent heat , thermal conductivity , and heat capacity are shown in figures 3 to 6 .
they are evaluated along the coexistence curve while for @xmath70 it is evaluated at the critical density . when divided by the entropy density at the critical point
the results are nearly independent of the choice of critical temperature . ]
they are evaluated along the coexistence curve while for @xmath70 it is evaluated at the critical density . when divided by the entropy density at the critical point
the results are nearly independent of the choice of critical temperature . ]
in any system of finite size there will be thermal fluctuations .
these fluctuations become large as the critical point is approached .
fluctuations can be studied with landau theory @xcite .
the question is : what is the probability to find a system of volume @xmath71 with a baryon density different from the equilibrium one ?
the expansion away from equilibrium states is determined by the thermodynamic potential @xmath72 v .\ ] ] here @xmath73 . along the coexistence curve @xmath74 . the relative probability to be at a density other than the equilibrium one , along the coexistence curve , is @xmath75 where @xmath76 v .\ ] ] for purposes of illustration the volume is taken to be 400 @xmath69 .
the value of 400 @xmath69 is about as large as one can imagine for high energy nuclear collisions . considering that the critical density is estimated to be about @xmath77 baryons/@xmath69
, this would mean that about 300 baryons participate in the fluctuation .
that is a substantial fraction of the total in au+au , pb+pb , or in u+u collisions .
smaller volumes would have even larger fluctuations .
the thermodynamic potential is shown in figure 7 for several different temperatures .
phase coexistence occurs below @xmath78 ; the two phases are located at the minima of the potential .
they are not very deep , indicating that fluctuations do not cost much free energy .
the potential is so flat because @xmath79 and because the power @xmath80 is so large .
the shallowness of the potential is reflected in the probabilities , as shown in figure 8 .
the probability to find the system with a density anywhere between @xmath81% of the critical density is greater than 50% .
fluctuations are large ! they would be even larger for smaller volume systems .
an equation of state valid in the vicinity of the chiral critical point has been constructed .
it incorporates correct values of the critical exponents and amplitudes .
since only certain properties of the equation of state are universal , there is some freedom to vary the noncritical functional dependence on temperature and density .
work on extending the equation of state to a wider range of @xmath1 and @xmath0 is underway .
the landau theory of fluctuations away from equilibrium states was used to determine the magnitude of the fluctuations one might expect in heavy ion collisions .
these fluctuations are quite large , partly due to finite volume effects but mostly because the critical exponent @xmath37 is much larger than in mean field theories .
this flattens the landau free energy as a function of density away from the equilibrium densities and so decreases the cost to fluctuate away from them .
it will require careful thought as to how to incorporate fluctuations near the critical point into dynamical simulations of heavy ion collisions .
what is the appropriate way to describe the transition in heavy ion collisions ?
is it nucleation @xcite , spinodal decomposition @xcite , or something else ?
what are the best experimental observables and can they be measured at rhic and/or fair ?
the future of this topic is exciting !
this work was supported by the us department of energy ( doe ) under grant no .
de - fg02 - 87er40328 .
99 a. barducci , r. casalbuoni , s. de curtis , r. gatto , and g. pettini , phys . lett . * b231 * , 463 ( 1989 ) ; phys . rev .
d * 41 * , 1610 ( 1990 ) ; a. barducci , r. casalbuoni , g. pettini , and r. gatto , _ ibid .
_ * 49 * , 426 ( 1994 ) . | the helmholtz free energy density is parametrized as a function of temperature and baryon density near the chiral critical point of qcd .
the parametrization incorporates the expected critical exponents and amplitudes .
an expansion away from equilibrium states is achieved with landau theory .
this is used to calculate the probability that the system is found at a density other than the equilibrium one .
such fluctuations are predicted to be very large in heavy ion collisions . |
one of the greatest unsolved mysteries of the terascale is the origin of electroweak symmetry breaking ( ewsb ) . within the usual description of the standard model ( sm )
, a weakly coupled higgs boson performs this task .
however , it has still not been experimentally verified whether electroweak symmetry is broken by such a higgs mechanism , by strong dynamics @xcite , or by something else .
this is one of the crucial questions particle physicists hope to answer in the upcoming experiments at the large hadron collider ( lhc ) at cern .
experiments have already been able to put some constraints on theoretical ideas about ewsb . in theories involving ewsb by strong dynamics
, the scale @xmath1 at which new physics enters can be guessed from the scale at which massive gauge boson scattering becomes non - unitary .
a simple estimate gives a value of ~4m_w / g ~1.8 tev , which is disfavored by precision electroweak constraints ( pec ) @xcite .
thus , strong dynamics would seem to be largely ruled out as the source of ewsb .
however , a new class of models , termed `` higgsless '' @xcite , have been able to raise the scale of strong dynamics , allowing agreement with pec @xcite .
realistic higgsless models contain new tev - scale weakly coupled states accessible at the lhc . among those ,
there are new massive vector bosons ( mvb ) , heavy cousins of the @xmath2 , @xmath3 and @xmath4 of the sm , which are of primary interest .
it is those states that delay unitarity violation and hence allow the scale @xmath1 to be raised @xcite .
unfortunately , the details of the fermion sector of the theory are highly model - dependent . for instance
, initial higgsless models did not allow sufficient change in @xmath1 to agree with pec @xcite , and modifications of the fermion sector were necessary .
however , the basic mechanism by which @xmath1 is raised is identical in all `` higgsless '' models , even regardless of the number of underlying dimensions @xcite .
it is this mechanism that was studied in @xcite , focusing on its collider signatures . in this talk
, we will review the analysis of ref .
@xcite , and present some new results , relevant for the international linear collider ( ilc ) . in sec .
[ sec : sumrules ] we derive a set of sum rules which should be obeyed by the couplings between the new mvbs and the sm @xmath5 gauge bosons .
we identify discovery signatures of the new mvbs at the lhc which only rely on the couplings guaranteed by the sum rules , and compare to the sm higgs search signals . in sec .
[ sec : lhc ] we discuss the lhc reach for charged mvbs and methods for testing the sum rules of sec . [ sec : sumrules ] in order to identify the `` higgsless '' origin of the mvb resonances . in sec .
[ sec : ilc ] we discuss the corresponding higgsless phenomenology at the ilc .
= 0.7 pt ( 100,100)(0,0 ) ( 15,80)(35,50)26 ( 15,20)(35,50)26 ( 35,50)(65,50)26 ( 65,50)(85,80)26 ( 65,50)(85,20)26 ( 50,10)[c](a ) ( 50,60)[c]@xmath2 ( 5,80)[c]@xmath2 ( 5,20)[c]@xmath3 ( 95,80)[c]@xmath2 ( 95,20)[c]@xmath3 ( 100,100)(0,0 ) ( 15,80)(50,65)26 ( 15,20)(50,35)26 ( 50,65)(50,35)26 ( 50,65)(85,20)29 ( 50,35)(85,80)29 ( 50,10)[c](b ) ( 35,50)[c]@xmath2 ( 5,80)[c]@xmath2 ( 5,20)[c]@xmath3 ( 95,80)[c]@xmath2 ( 95,20)[c]@xmath3 ( 100,100)(0,0 ) ( 15,80)(85,20)213 ( 15,20)(85,80)213 ( 50,10)[c](c ) ( 5,80)[c]@xmath2 ( 5,20)[c]@xmath3 ( 95,80)[c]@xmath2 ( 95,20)[c]@xmath3 ( 100,100)(0,0 ) ( 15,80)(50,65)26 ( 15,20)(50,35)26 ( 50,65)(50,35)2 ( 50,65)(85,80)26 ( 50,35)(85,20)26 ( 50,10)[c](d ) ( 40,50)[c]@xmath6 ( 5,80)[c]@xmath2 ( 5,20)[c]@xmath3 ( 95,80)[c]@xmath2 ( 95,20)[c]@xmath3 ( 100,100)(0,0 ) ( 15,80)(35,50)26 ( 15,20)(35,50)26 ( 35,50)(65,50)26 ( 65,50)(85,80)26 ( 65,50)(85,20)26 ( 50,10)[c](e ) ( 50,62)[c]@xmath7 ( 5,80)[c]@xmath2 ( 5,20)[c]@xmath3 ( 95,80)[c]@xmath2 ( 95,20)[c]@xmath3 ( 100,100)(0,0 ) ( 15,80)(50,65)26 ( 15,20)(50,35)26 ( 50,65)(50,35)26 ( 50,65)(85,20)29 ( 50,35)(85,80)29 ( 50,10)[c](f ) ( 35,50)[c]@xmath7 ( 5,80)[c]@xmath2 ( 5,20)[c]@xmath3 ( 95,80)[c]@xmath2 ( 95,20)[c]@xmath3 consider the elastic scattering process @xmath8 . in the absence of the higgs boson , this process receives contributions from the three feynman diagrams shown in figs . [ wzwz](a)(c ) .
the resulting amplitude contains terms which grow with the energy @xmath9 of the incoming particle as @xmath10 and @xmath11 and ultimately cause violation of unitarity at sufficiently high energies . in the sm , both of these terms are precisely cancelled by the contribution of the higgs exchange diagram in fig .
[ wzwz](d ) . in the higgsless theories , on the other hand , the diagram of fig .
[ wzwz](d ) is absent , and the process instead receives additional contributions from the diagrams in figs .
[ wzwz](e ) and [ wzwz](f ) , where @xmath7 denotes the charged mvb of mass @xmath12 .
the index @xmath13 corresponds to the kk level of the state in the case of a 5d theory , or labels the mass eigenstates in the case of a 4d deconstructed theory .
remarkably , the @xmath10 and @xmath11 terms can again be exactly cancelled by the contribution of the mvbs , provided that the following sum rules are satisfied @xcite : & = & ^2 + _ i ( ^(i))^2 , 2(-^2)(^2+^2 ) + ^2 & = & _ i ( ^(i))^2 . here
@xmath14 ( @xmath15 ) is the @xmath2-boson ( @xmath3-boson ) mass and the notation for the triple and quartic gauge boson couplings is self - explanatory . in 5d theories
, these equations are satisfied exactly if all the kk states , @xmath16 , are taken into account .
this is not an accident , but a consequence of the gauge symmetry and locality of the underlying theory .
while this is not sufficient to ensure unitarity at all energies ( the increasing number of inelastic channels ultimately results in unitarity violation ) , the strong coupling scale can be significantly higher than the naive estimate .
for example , in the warped - space higgsless models @xcite unitarity is violated at the scale @xcite _ nda ~ , which is typically of order 510 tev . in 4d models ,
the number of mvbs is finite and the second of the sum rules is only satisfied approximately ; however , our numerical study of sample models indicates that the violation of the sum rule has to be very small ( at the level of 1% ) to achieve an adequate improvement in @xmath1 .
considering the @xmath17 scattering process yields sum rules constraining the couplings of the neutral mvbs @xmath18 ( with masses denoted by @xmath19 ) @xcite : & = & ^2 + ^2 + _ i ( ^(i))^2 , 4 ^ 2 & = & 3 .
considering other channels such as @xmath20 ( see fig .
[ wwzz ] ) and @xmath21 does not yield any new sum rules .
the presence of multiple mvbs , whose couplings obey eqs . , , is a generic prediction of the higgsless models .
= 0.7 pt ( 100,100)(0,0 ) ( 15,80)(50,65)26 ( 15,20)(50,35)26 ( 50,65)(50,35)26 ( 50,65)(85,80)26 ( 50,35)(85,20)26 ( 50,10)[c](a ) ( 35,50)[c]@xmath2 ( 5,80)[c]@xmath2 ( 5,20)[c]@xmath2 ( 95,80)[c]@xmath3 ( 95,20)[c]@xmath3 ( 100,100)(0,0 ) ( 15,80)(50,65)26 ( 15,20)(50,35)26 ( 50,65)(50,35)26 ( 50,65)(85,20)29 ( 50,35)(85,80)29 ( 50,10)[c](b ) ( 35,50)[c]@xmath2 ( 5,80)[c]@xmath2 ( 5,20)[c]@xmath2 ( 95,80)[c]@xmath3 ( 95,20)[c]@xmath3 ( 100,100)(0,0 ) ( 15,80)(85,20)213 ( 15,20)(85,80)213 ( 50,10)[c](c ) ( 5,80)[c]@xmath2 ( 5,20)[c]@xmath2 ( 95,80)[c]@xmath3 ( 95,20)[c]@xmath3 ( 100,100)(0,0 ) ( 15,80)(35,50)26 ( 15,20)(35,50)26 ( 35,50)(65,50)2 ( 65,50)(85,80)26 ( 65,50)(85,20)26 ( 50,10)[c](d ) ( 50,62)[c]@xmath6 ( 5,80)[c]@xmath2 ( 5,20)[c]@xmath2 ( 95,80)[c]@xmath3 ( 95,20)[c]@xmath3 ( 100,100)(0,0 ) ( 15,80)(50,65)26 ( 15,20)(50,35)26 ( 50,65)(50,35)26 ( 50,65)(85,80)26 ( 50,35)(85,20)26 ( 50,10)[c](e ) ( 35,50)[c]@xmath22 ( 5,80)[c]@xmath2 ( 5,20)[c]@xmath2 ( 95,80)[c]@xmath3 ( 95,20)[c]@xmath3 ( 100,100)(0,0 ) ( 15,80)(50,65)26 ( 15,20)(50,35)26 ( 50,65)(50,35)26 ( 50,35)(85,80)29 ( 50,65)(85,20)29 ( 50,10)[c](f ) ( 35,50)[c]@xmath22 ( 5,80)[c]@xmath2 ( 5,20)[c]@xmath2 ( 95,80)[c]@xmath3 ( 95,20)[c]@xmath3 our study of the collider phenomenology in the higgsless models will focus on the vector boson fusion processes .
these processes are attractive for two reasons .
firstly , the production of mvbs via vector boson fusion is relatively model - independent , since the couplings are constrained by the sum rules , .
this is in sharp contrast with the drell - yan production mechanism @xcite , which dominates for the conventional @xmath23 and @xmath24 bosons but is likely to be suppressed for the higgsless mvbs due to their small couplings to fermions , as needed to evade pec @xcite . in the following ,
unless specified otherwise , we shall assume that the mvbs have no appreciable couplings to sm fermions .
secondly , if enough couplings and masses can be measured , these processes can provide a _
test _ of the sum rules , probing the mechanism of partial unitarity restoration .
. indicates that the first mvb should appear below @xmath25 tev , and thus be accessible at the lhc . for @xmath26 ,
the sum rules imply an inequality ^(1 ) .
this bound is quite stringent ( @xmath27 for @xmath28 gev ) .
also , convergence of the sum rules requires @xmath29 the combination of heavier masses and lower couplings means that the heavier mvbs may well be unobservable , so that only the @xmath30 states can be studied . the `` saturation limit '' , in which there is only a single set of mvbs whose couplings saturate the sum rules , is likely to provide a good approximation to the phenomenology of the realistic higgsless models . in this limit , the partial width of the @xmath26 is given by @xmath31 given the couplings of the mvbs to the sm @xmath2 and @xmath3 , we can now predict ( at the parton level ) the size of the new physics signals in the various channels of vector boson fusion . fig .
[ fig : parton ] provides an illustration for the case of @xmath32 and @xmath33 .
we show the expected signal for either a sm higgs boson of mass @xmath34 gev , or the corresponding mvb @xmath30 of mass @xmath35 gev in the saturation limit
. the sum rules ( [ sumz ] ) govern the signal in the @xmath32 channel shown in the left panel of fig .
[ fig : parton ] .
however , the @xmath36 final state is difficult to observe over the sm backgrounds at the lhc : in the dilepton channel there is no resonance structure , while the jetty channels suffer from large qcd backgrounds .
it is therefore rather challenging to test the sum rules ( [ sumz ] ) .
notice that even if a @xmath36 resonance is observed , without a test of the sum rules ( [ sumz ] ) , its interpretation is unclear , since the sm higgs boson is _ also _ expected to appear as a @xmath36 resonance ( see the left panel in fig .
[ fig : parton ] ) .
we shall therefore concentrate on the @xmath33 channel , in which the higgsless model predicts a series of resonances as in fig .
[ wzwz](e ) , while in the sm the amplitude is unitarized by the @xmath37-channel diagram of fig .
[ wzwz](d ) and has no resonance ( see the right panel in fig .
[ fig : parton ] ) .
conventional theories of ewsb by strong dynamics may also contain a resonance in this channel , but it is likely to be heavy ( @xmath38 tev for qcd - like theories ) and broad due to strong coupling . in contrast , the mvb resonance is very narrow , as can be seen from fig .
[ fig : parton ] and eq .
( [ width ] ) .
in fact it is almost a factor of 20 narrower than a sm higgs boson of the same mass .
this is primarily due to the vector nature of the mvb and our assumption that it only has a single decay channel .
we therefore conclude that a resonance in the @xmath33 channel would be a smoking gun for the higgsless model ( for alternative interpretations involving extended higgs sectors , see @xcite and references therein ) .
finally , the @xmath39 channel is a good discriminator as well , since it will exhibit a resonance for the case of the sm but not the higgsless models ( see fig . [ wwzz ] ) .
a comparison of the resonant structure of the three vector boson fusion final states is shown in table [ tab : res ] .
.comparison of the resonance structure of the sm and higgsless models in different vector boson fusion channels . [ cols="^,^,^,^",options="header " , ] [ tab : res ]
at the lhc , the vector boson fusion processes will occur as a result of @xmath5 bremsstrahlung off quarks .
the typical final state for such events includes two forward jets in addition to a pair of gauge bosons .
the production cross section of @xmath26 in association with two jets is shown by the solid line in the left panel of fig .
[ fig : lhc ] . to estimate the prospects for the charged mvb search at the lhc
, we require that both jets be observable ( we assume jet rapidity coverage of @xmath40 ) , and impose the following lower cuts on the jet rapidity , energy , and transverse momentum : @xmath41 gev , @xmath42 gev .
these requirements enhance the contribution of the vector boson fusion diagrams relative to the irreducible background of the non - fusion @xmath43 sm process as well as @xmath44 drell - yan process .
the `` gold - plated '' final state @xcite for this search is @xmath45 + , with the additional kinematic requirement that two of the leptons have to be consistent with a @xmath3 decay .
we assume lepton rapidity coverage of @xmath46 .
the @xmath47 invariant mass , @xmath48 , can be reconstructed using the missing transverse energy measurement and requiring that the neutrino and the odd lepton form a @xmath2 .
the number of `` gold - plated '' events ( including all lepton sign combinations ) in a 300 fb@xmath49 lhc data sample , as a function of @xmath48 , is shown in fig .
[ fig : lhc ] , for the sm ( dotted ) , the higgsless model with @xmath28 gev ( blue ) , and two `` unitarization '' models : pad ( red ) and k - matrix ( green ) @xcite ( for details , see @xcite ) .
the higgsless model can be easily identified by observing the mvb resonance : for the chosen parameters , the dataset contains @xmath50 @xmath51 events .
the irreducible non - fusion sm background is effectively suppressed by the cuts : the entire dataset shown in fig .
[ fig : lhc ] contains only @xmath52 such events .
we therefore estimate the discovery reach for @xmath26 resonance by requiring 10 signal events after cuts .
the efficiency of the cuts for @xmath53 is in the range @xmath54 .
we then find that with @xmath55 of data , corresponding to 1 year of running at low luminosity , the lhc will probe the higgsless models up to @xmath56 gev , while covering the whole preferred range up to @xmath57 tev requires @xmath58 .
note , however , that one should expect a certain amount of reducible background with fake and/or non - isolated leptons .
once the @xmath26 resonance is discovered , identifying it as part of a higgsless model requires testing the sum rules ( [ sumw ] ) by measuring its mass @xmath59 and coupling @xmath60 .
the coupling can be determined from the total @xmath26 production cross section @xmath61 .
however , we are observing the @xmath26 resonance in an exclusive channel , which only yields the product @xmath62 .
a measurement of the total resonance width @xmath63 would remove the dependence on the unknown branching fraction @xmath64 .
however , the accuracy of this measurement is severely limited by the poor missing energy resolution .
even though a higgsless origin of the resonance can be ruled out if the value of @xmath60 , inferred with the assumption of @xmath65 , violates the bound ( [ bound ] ) , the lhc alone will not be able to settle the issue and precise measurements at the ilc appear to be necessary for the ultimate test of the theory .
unlike traditional technicolor , higgsless models offer new discovery opportunities for a lepton collider with a center - of - mass energy in the sub - tev range .
from we have seen that the masses of the new mvbs are expected to be below 1 tev , and they can be produced at the ilc through the analogous vector boson fusion process by bremsstrahlung of @xmath2 s and @xmath3 s off the initial state @xmath66 and @xmath67 .
the @xmath30 production cross - sections for vector boson fusion @xmath68 and @xmath69 as well as associated production @xmath70 are shown in the left panel of fig .
[ fig : ilc ] .
the horizontal lines correspond to the total cross - section of the continuum sm background .
we see that for a large range of @xmath30 masses , ilc searches appear promising , already at the level of total number of events , before cuts and efficiencies . furthermore , because of the cleaner environment of the ilc , one can now use the dominant hadronic decay modes of the @xmath2 and @xmath3 , and easily reconstruct the invariant mass of the @xmath30 resonance , which provides an extra handle for background suppression ( see the right panel in fig .
[ fig : ilc ] ) .
further detailed studies are needed to better evaluate the ilc potential for testing the generic predictions ( [ sumw ] ) and ( [ sumz ] ) of the higgsless models .
99 s. dimopoulos and l. susskind , nucl .
b * 155 * , 237 ( 1979 ) ; l. susskind , phys .
d * 20 * , 2619 ( 1979 ) ; s. weinberg , phys . rev .
d * 19 * , 1277 ( 1979 ) .
m. e. peskin and t. takeuchi , phys .
lett . * 65 * , 964 ( 1990 ) ; phys .
d * 46 * , 381 ( 1992 ) .
c. csaki _ et al .
_ , phys .
d * 69 * , 055006 ( 2004 ) [ arxiv : hep - ph/0305237 ] . c. csaki , c. grojean , l. pilo and j. terning , phys .
lett . * 92 * , 101802 ( 2004 ) [ arxiv : hep - ph/0308038 ] .
y. nomura , jhep * 0311 * , 050 ( 2003 ) [ arxiv : hep - ph/0309189 ] . c. csaki
_ et al . _ ,
d * 70 * , 015012 ( 2004 ) [ arxiv : hep - ph/0310355 ] .
g. cacciapaglia , c. csaki , c. grojean and j. terning , phys .
d * 71 * , 035015 ( 2005 ) [ arxiv : hep - ph/0409126 ] ; r. foadi , s. gopalakrishna and c. schmidt , phys .
b * 606 * , 157 ( 2005 ) [ arxiv : hep - ph/0409266 ] ; c. schwinn , phys .
d * 71 * , 113005 ( 2005 ) [ arxiv : hep - ph/0504240 ] ; g. cacciapaglia _ et al .
_ , arxiv : hep - ph/0505001 .
r. sekhar chivukula , d. a. dicus and h. j. he , phys .
b * 525 * , 175 ( 2002 ) [ arxiv : hep - ph/0111016 ] .
r. barbieri , a. pomarol and r. rattazzi , phys .
b * 591 * , 141 ( 2004 ) [ arxiv : hep - ph/0310285 ] ; g. burdman and y. nomura , phys .
d * 69 * , 115013 ( 2004 ) [ arxiv : hep - ph/0312247 ] ; g. cacciapaglia , c. csaki , c. grojean and j. terning , phys .
d * 70 * , 075014 ( 2004 ) [ arxiv : hep - ph/0401160 ] .
h. davoudiasl , j. l. hewett , b. lillie and t. g. rizzo , phys .
d * 70 * , 015006 ( 2004 ) [ arxiv : hep - ph/0312193 ] .
h. davoudiasl , j. l. hewett , b. lillie and t. g. rizzo , jhep * 0405 * , 015 ( 2004 ) [ arxiv : hep - ph/0403300 ] ; r. barbieri , a. pomarol , r. rattazzi and a. strumia , nucl .
b * 703 * , 127 ( 2004 ) [ arxiv : hep - ph/0405040 ] ; j. l. hewett , b. lillie and t. g. rizzo , jhep * 0410 * , 014 ( 2004 ) [ arxiv : hep - ph/0407059 ] .
r. foadi , s. gopalakrishna and c. schmidt , jhep * 0403 * , 042 ( 2004 ) [ arxiv : hep - ph/0312324 ] ; r. casalbuoni , s. de curtis and d. dominici , phys .
d * 70 * , 055010 ( 2004 ) [ arxiv : hep - ph/0405188 ] ; r. s. chivukula _ et al .
_ , phys . rev .
d * 70 * , 075008 ( 2004 ) [ arxiv : hep - ph/0406077 ] ; h. georgi , phys .
d * 71 * , 015016 ( 2005 ) [ arxiv : hep - ph/0408067 ] ; m. perelstein , jhep * 0410 * , 010 ( 2004 ) [ arxiv : hep - ph/0408072 ] ; h. georgi , arxiv : hep - ph/0508014 .
a. birkedal , k. matchev and m. perelstein , phys .
* 94 * , 191803 ( 2005 ) [ arxiv : hep - ph/0412278 ] .
m. papucci , arxiv : hep - ph/0408058 .
e. asakawa and s. kanemura , arxiv : hep - ph/0506310 .
a. dobado , m. j. herrero and j. terron , z. phys .
c * 50 * ( 1991 ) 205 ; z. phys .
c * 50 * ( 1991 ) 465 .
j. bagger _
et al . _ , phys .
d * 49 * , 1246 ( 1994 ) [ arxiv : hep - ph/9306256 ] ; phys .
d * 52 * , 3878 ( 1995 ) [ arxiv : hep - ph/9504426 ] .
see also m. s. chanowitz , arxiv : hep - ph/0412203 and references therein . | we investigate the signatures of the recently proposed higgsless models at future colliders .
we focus on tests of the mechanism of partial unitarity restoration in the longitudinal vector boson scattering , which do not depend on any higgsless model - building details .
we study the lhc discovery reach for charged massive vector boson resonances and show that all of the preferred parameter space will be probed with @xmath0 of lhc data .
we also discuss the prospects for experimental verification of the higgsless nature of the model at the lhc .
in addition , in this talk we present new results relevant for the discovery potential of higgsless models at the international linear collider ( ilc ) . |
the quantum many - body problem has been perfectly clarified a long time ago for the single - reference approaches .
its logics is given by the linked cluster theorem,@xcite which relies on a perturbative expansion and its diagrammatic representation .
a panoply of computational methods satisfying the basic size - consistency requirements are available .
one of them is the mller - plesset order - by - order perturbative expansion @xcite which suffers from its slow convergence ( and in some cases high - order divergences @xcite ) .
other methods lead to non - linear equations coupling the coefficients of the vectors which are singly or doubly excited with respect to the single reference @xmath0 . among them
one may quote the coupled electron pair approximation ( cepa ) formalism @xcite which eliminates unlinked diagrams from a singles and doubles configuration interaction ( sdci ) and introduces part of ( in cepa-2 @xcite ) or all ( in the @xmath1 sdci version @xcite ) of the exclusion principle violating diagrams .
the coupled cluster method ( ccm @xcite ) has a better standard since it is an elegant and generalizable formalism .
this method gives an exponential structure to the wave operator @xmath2 which transforms a reference function @xmath0 in the exact solution @xmath3 , @xmath4 , @xmath5 . in practice @xmath6 in
limited to a certain set of excitations .
the simplest version , called ccsd , introduces single and double excitations on the top of @xmath0 and takes into account the connected effect of the quadruples .
it does notes treat correctly the effect of the triples , which has to be incorporated perturbatively in the ccsd(t ) formalism.@xcite + the present work returns to a low - order semi - perturbative development from @xmath0 .
it leads to a system of equations fixing the coefficients of the singly and doubly excited vectors , which * does not require to assume an exponential structure of the wave - operator , * introduces the effect of deviations from additivity of excitation energies ( deviations which is implicitly supposed to be negligible in ccsd ) , * and introduces some high - order effects through epv corrections to the excitation energies .
the method is applied to a few infinite periodic spin lattices governed by heisenberg hamiltonians which only consider interactions between nearest neighbors . in these problems
if all bonds are identical , the solutions of our self - consistent perturbative ( scp ) method lead to simple polynomial equations .
the results , compared to the exact solutions , are of surprising quality .
the method is then employed to identify phase transitions in two different 2-d problems .
let us call @xmath0 a suitable single reference supposed to be a relevant approximation to the ground state of the system governed by an hamiltonian @xmath7 and @xmath8 vectors orthogonal to @xmath0 . in the intermediate normalization @xmath9
the exact energy is given by the eigenequation relative to @xmath0 , @xmath10 @xmath11 where @xmath12 .
the hamiltonians is at most bi - electronic .
one may call @xmath13 and @xmath14 the mono - electronic and bi - electronic parts of the hamiltonian @xmath15 .
one may label @xmath16 ( holes ) the mono - electronic function occupied in @xmath17 and @xmath18 ( particles ) those which are not occupied in @xmath17 . + in the following we shall refer to an epstein - nesbet@xcite zero - order hamiltonian which is the diagonal part of the hamiltonian in the basis of the n - electronic vectors @xmath19 @xmath20 the perturbation operator @xmath21 @xmath22 is zero - diagonal in this basis @xmath23 the hamiltonian being at most bi - electronic , it appears from eq .
2 that the knowledge of the coefficient of the singly and doubly excited vectors which interact with @xmath17 is sufficient to give the ground state energy . we shall call first generation @xmath24 the set of vectors @xmath25 interacting with @xmath17 , @xmath26 and @xmath27 the operators creating the vectors @xmath25 from @xmath17 , @xmath28 these operators may be written as @xmath29 or @xmath30 where @xmath31 is the kroneker symbol . the coefficients of the first generation vectors
can be estimated from the eigenequations relative to them . for @xmath32 the equation @xmath33
can be written using eq . 2 as @xmath34 to determine the coefficients @xmath35 of the first generation vectors , it is sufficient to have an estimate of the coefficient of vectors @xmath36 of the second generation , i.e. , those which interact with the vectors of @xmath37 .
among the vectors @xmath36 belonging to the second generation one may distinguish those which are obtained from @xmath32 by the actions of operators @xmath38 possible on @xmath0 ( type 1 ) , and the others which are obtained from @xmath32 by operators @xmath39 different from the @xmath40 ( type 2 ) .
+ if one calls @xmath41 the sum of the operators @xmath38 and of their adjoints @xmath42 the operators @xmath43 @xmath44 is the sum of all the operators changing one or two mono - electronic functions which are different from @xmath38 operators .
notice that @xmath45 . with the hamiltonians considered hereafter the second generation vectors @xmath36
are either of type 1 or of type 2 , i.e. , are generated either as @xmath46 or as @xmath47 as pictured in fig .
( 80,33 ) ( 50,5)@xmath0 ( 40,15)@xmath48 ( 60,15)@xmath8 ( 50,25)@xmath36 ( 50,7)(-1,1)8 ( 52,7)(1,1)8 ( 42,17)(1,1)8 ( 60,17)(-1,1)7.5 ( 31,15)@xmath49 ( 68,15)@xmath32 ( 78,15)@xmath50 ( 73,25)@xmath51 ( 70,17)(71.5,21)(73,25 ) ( 77,25)(78.5,21)(80,17 ) ( 53.5,6)(61,10.25)(68.5,14.5 ) ( 48.5,6)(40.25,10.25)(32,14.5 ) ( 54,6)(66,10.5)(78,15 ) ( 66,20.5)@xmath52 ( 79,20.5)@xmath53 ( 61.5,11)@xmath54 ( 35.5,8)@xmath55 ( 53.5,26)(61,21.75)(68.5,17.5 ) ( 48.5,26)(40.25,21.75)(32,17.5 ) ( 60,23)@xmath55 ( 36,22)@xmath54 ( 70,9)@xmath56 ( 43,10)@xmath57 ( 52,12)@xmath58 ( 43,20)@xmath58 ( 52,18)@xmath57 ( 1,5)reference ( 1,15)first generation ( 1,25)second generation ( 47,30)(type 1 ) ( 70,30)(type 2 ) regarding the _ vectors of type 1 _ , the operators @xmath38 are in general possible on @xmath32 , but some of them are impossible ( @xmath59 ) .
let call @xmath60 ( exclusion principle violating correction ) the quantity@xcite @xmath61 noting that if @xmath62 , and if @xmath38 is a double excitation @xmath63 .
introducing eq .
11 the eq .
8 can be written @xmath64 a first order evaluation of @xmath65 is of course the epstein - nesbet@xcite one @xmath66 introducing @xmath67 corrections in the energy denominator @xmath68 incorporates an infinite summation of diagrams.@xcite one may use a second order perturbation theory to evaluate the coefficients @xmath69 and @xmath70 .
+ the vector of type 1 @xmath71 can also be reached from @xmath0 by other sets of excitations @xmath72 ( cf fig .
the second order expansion of the wave function tells us @xmath73 where @xmath74 and @xmath75 runs on the couples @xmath76 but different from the couple @xmath77 .
an infinite summation of diagrams would lead to the introduction of an @xmath67 correction in the energy denominator @xmath78 and to replace @xmath79 by @xmath80 .
@xmath81 now taking benefit of ( eq .
14 ) i.e. , replacing @xmath82 ( and @xmath83 ) by products @xmath84 in eq .
17 one obtains an expression closer to the cc expansion : @xmath85 the cc theory writes @xmath86 which ignores the possible deviation of the energy denominators from additivity i.e. , assumes + @xmath87 . using the compact notation @xmath88 , @xmath89 , one obtains @xmath90 notice that if there is only one route to create @xmath91 from @xmath0 the last sum disappears .
if the excitation energies ( eventually including @xmath92 ) are additive , the full term vanishes in agreement with the cancellation of disconnected diagrams .
this will be the case in the applications to spin lattices with nearest - neighbor interactions when excitations @xmath27 and @xmath38 will concern remote bonds .
+ regarding the _ vectors of type2 _ , the operators @xmath93 which lead from @xmath32 to @xmath94 have to be treated perturbatively as follows .
analogously to eq .
17 one obtains @xmath95 if @xmath39 is a bi - electronic operator , @xmath96 and the eigenequation relative to @xmath32 will be @xmath97 this will be our basic equation .
it is somewhat related to the usual coupled cluster equations .
the similarity and differences appear clearly if ones considers the case of an scf @xmath0 .
then the vectors of the first generation are all the doubles , hence the method should be compared to a ccd expansion .
the differences are the following : * the effect of the quadruples in ccd does not take care of possible deviations from the additivity of the denominators and the sum over @xmath98 reduces to @xmath99 we take into account the deviation from additivity of energy denominators , * we take care of the coupling with the singles and triples through the last summation .
hence our method would be similar to a ccd(s , t ) .
* and our ( semi ) perturbative treatment of the single and triples includes @xmath67 corrections in the energy denominators .
these corrections also appear in the effect of deviations from additivity of multiple excitation energies .
most of these effects were considered in previous works,@xcite which were expressed in an intermediate hamiltonian formalism , requiring an iterative dressing of small matrices .
the set of coupled non linear equations ( eq .
22 ) is much clearer .
as will be shown in this section the method is of an extreme simplicity when applied * to periodic lattices of spins , ( and electrons , @xcite ) * ruled by model hamiltonians which only introduce short range interactions ( usually between nearest - neighbor atoms ) * provided that one starts from a strongly localized wave function , product of atomic or bond orbitals . the usual implementation of the coupled cluster method are governed by a hierarchy of excitations .
they introduce for instance all single and double excitations . here
we adopt a different strategy , since we follow the genealogy of the generation of @xmath3 from @xmath0 under the successive applications of @xmath7 .
due to the localized character of @xmath0 and the short range nature of the operator in @xmath7 , the @xmath100-generation vectors will involve short range excitations only , and in periodic systems one has a limited number of types of @xmath100-generation excitations . under these conditions the number of coefficients to determine is extremely reduced , and their values are obtained by solving a set of coupled polynomial equations , which may be done on a pocket calculator .
we shall take a few examples illustrating the method and its performances , comparing the so - calculated values to the exact ( or highly accurate ) ones .
section a will illustrate the derivation of the scp equations for simple lattices , and will compare the resulting cohesive energies to benchmarks values .
section b will employ the scp method to study two phase transition phenomena taking place in 2-d lattices .
the ruling hamiltonian will introduce only nearest - neighbor interactions @xmath101 between adjacent atoms .
this hamiltonian puts to zero the energy of the upper multiplet if all bonds are equal , taking @xmath0 as the nel function ( spin alternation between adjacent atoms ) , the zero order energy per atom is @xmath102 .
( 80,15 ) ( 5,4)(1,0)60 ( 66,3)@xmath0 ( 10,5)@xmath103 ( 30,5)@xmath103 ( 50,5)@xmath103 ( 20,5)@xmath104 ( 40,5)@xmath104 ( 60,5)@xmath104 ( 35,1)@xmath105 there is only one type of vectors in @xmath24 , the one obtained from a spin exchange @xmath27 on a bond @xmath105 , of coefficient @xmath106 , coupled by @xmath107 with @xmath0 ( @xmath108 ) hence @xmath109 ( 80,15 ) ( 5,4)(1,0)60 ( 66,3)@xmath32 ( 10,5)@xmath103 ( 20,5)@xmath104 ( 30,5)@xmath104 ( 40,5)@xmath103 ( 50,5)@xmath103 ( 60,5)@xmath104 ( 30.8,4.3)(1,0)10 ( 35,1)@xmath105 ( 23,1)@xmath110 ( 15,1)@xmath111 ( 46,1)@xmath112 ( 55,1)@xmath98 ( 25,3)@xmath113 ( 45,3)@xmath113 a simple second - order estimate of the energy would lead to @xmath114 , @xmath115 , much larger than the exact value @xmath116 . including
the @xmath92 in the denominator would give @xmath117 which is already a better estimate .
there is no coupling between the vectors @xmath32 and @xmath118 .
the second generation vectors are obtained by two spin exchanges on bonds @xmath105 and @xmath98 .
if the bonds @xmath105 and @xmath98 are far the doubly excited vectors @xmath119 have a coefficient @xmath120 since * there is no other couple of @xmath100-generation excitation @xmath76 , * the excitation energies are additive .
the only exceptions concern the double exchanges @xmath121 and @xmath122 , on second - neighbor bonds . for them
@xmath123 , @xmath124 .
( 80,15 ) ( 5,4)(1,0)60 ( 10,5)@xmath104 ( 20,5)@xmath104 ( 40,5)@xmath104 ( 30,5)@xmath103 ( 50,5)@xmath103 ( 60,5)@xmath103 ( 15,3)@xmath113 ( 55,3)@xmath113 ( 25,1)@xmath105 ( 45,1)@xmath98 their coefficient can be evaluated according to ( eq .
18 ) as @xmath125 and the final equation is @xmath126 its solution is @xmath127 hence @xmath128 , which deviates by @xmath129 from the exact value @xmath130 . here
@xmath131 , @xmath132 , where @xmath106 is the coefficient relative to a spin exchange on a bond . here @xmath133 , @xmath134 .
the coefficients of the doubly spin exchanged vectors are @xmath135 , except for * spin exchanges on the same plaquette .
+ ( 70,40 ) ( 5,15)(1,0)30 ( 5,25)(1,0)30 ( 15,5)(0,1)30 ( 25,5)(0,1)30 ( 15,15.3)(1,0)10 ( 15,25.3)(1,0)10 ( 10,14)@xmath113 ( 30,14)@xmath113 ( 10,24)@xmath113 ( 30,24)@xmath113 ( 14,10)@xmath113 ( 14,30)@xmath113 ( 24,10)@xmath113 ( 24,30)@xmath113 ( 20,12)@xmath98 ( 20,26)@xmath105 ( 11,20)@xmath136 ( 26,20)@xmath137 ( 37,29)since ( 47,29)@xmath138 ( 37,19)and ( 45,21)@xmath139 ( 45,16)@xmath140 ( 33,5)hence ( 43,5)@xmath141 + there are two such double excitations for a bond @xmath105 . * spin exchanges on other second neighbor bonds ( the number of which is 14 for a given bond @xmath105 ) . + ( 70,40 ) ( 5,15)(1,0)20 ( 15,5)(0,1)20 ( 25,25)(1,0)20 ( 35,15)(0,1)20 ( 25,5)(0,1)30 ( 15,15.3)(1,0)10 ( 25,25.3)(1,0)10 ( 15,25)(1,0)10 ( 25,15)(1,0)10 ( 20,24)@xmath113 ( 30,14)@xmath113 ( 10,14)@xmath113 ( 40,24)@xmath113 ( 14,10)@xmath113 ( 14,20)@xmath113 ( 24,10)@xmath113 ( 24,30)@xmath113 ( 34,20)@xmath113 ( 34,30)@xmath113 ( 20,12)@xmath105 ( 30,22)@xmath98 ( 47,31)@xmath142 ( 44,16)@xmath143 ( 44,5)@xmath144 the final equation is @xmath145 the solution of which is @xmath146 , @xmath147 to be compared with the best monte carlo estimate @xmath148 . the error is @xmath149 .
our estimate may be compared with classical cc evaluations @xcite which give [ cols="^,^,^,^ " , ] .
( @xmath150 ) qmc , @xcite scp results : ( ... ) from xy @xmath0 , ( - ) from nel @xmath151 .
the thick symbols identify the relevant domains from @xmath0 and @xmath151 respectively . ]
* quantum monte carlo calculations @xcite * real space renormalization group with effective interactions @xcite * coupled cluster expansions @xcite * dressed cluster method ( dcm ) .
@xcite the two first techniques do not introduce any reference function @xmath0 .
the two last ones imply a choice of such reference functions either as a zero - order function for the coupled cluster method , or as a bath in which a finite cluster is embedded in the dcm .
our approach belongs to the same family .
the equations derived from @xmath152 or from @xmath153 are given in the appendix a. one may see that they coincide for @xmath154 , which appears as the critical value of the parameter .
2 gives the evolution of the cohesive energy @xmath155 and @xmath156 calculated from both references and a comparison with the most accurate results .
one sees * that the derivatives of the energies in @xmath154 are not identical @xmath157 which confirms the @xmath100-order character of the phase transition , * the agreement of the here - calculated cohesive energies with the best qmc calculations in good , the error being lower than @xmath158 in the @xmath159 domain .
the 1/5-depleted square lattice has received interest in the context of the study of a real material , namely the @xmath160 lattice .
precise studies of that material indicated that @xmath161-neighbor interactions have to be considered .
nevertheless the physics of the simple lattice , built of plaquettes and octogons , or of plaquettes connected by bonds ( cf fig .
3 ) , appeared interesting by itself .
there are two types of spin interactions , namely in the plaquette ( @xmath162 ) and in the bonds between the plaquettes ( @xmath163 ) .
early studies did not give evidence of phase transitions , but perturbative expansions @xcite and quantum monte carlo calculations @xcite indicate the existence of three phases , namely a gapped plaquette phase for @xmath164 , a gapped dimer phase for @xmath165 and a gapless nel - type phase in the intermediate regime .
we have reexamined this problem with the scp equations starting from three different references @xmath0 , namely : 1 . a product of bond singlets on the inter - plaquette bonds which is expected to be relevant in the @xmath166 regime 2 .
the nel function , which should be valid in the @xmath167 region 3 . a product of bond singlets on intra - plaquette bonds , relevant in the @xmath168 regime .
( 60,55 ) ( 2,26.97)(4.825,29.795)(7.65,32.62 ) ( 2,26.97)(4.825,24.14)(7.65,21.31 ) ( 7.65,32.62)(10.48,29.795)(13.31,26.97 ) ( 7.65,21.31)(10.48,24.14)(13.31,26.97 ) ( 7.65,32.62)(7.65,37.62)(7.65,42.62 ) ( 7.65,21.31)(7.65,16.31)(7.65,11.31 ) ( 7.65,42.62)(4.825,45.445)(2,48.27 ) ( 7.65,42.62)(10.48,45.445)(13.31,48.27 ) ( 2,48.17)(4.825,51.095)(7.65,53.92 ) ( 7.65,53.92)(10.48,51.095)(13.31,48.27 ) ( 7.65,11.31)(4.825,8.48)(2,5.65 ) ( 7.65,11.31)(10.48,8.48)(13.31,5.65 ) ( 2,5.65)(4.825,2.825)(7.65,0 ) ( 13.31,5.65)(10.48,2.825)(7.65,0 ) ( 13.31,48.27)(18.31,48.27)(23.31,48.27 ) ( 23.31,48.27)(26.135,51.095)(28.96,53.92 ) ( 23.31,48.27)(26.135,45.445)(28.96,42.62 ) ( 28.96,53.92)(31.785,51.095)(34.61,48.27 ) ( 28.96,42.62)(31.785,45.445)(34.61,48.27 ) ( 34.61,48.27)(39.61,48.27)(44.61,48.27 ) ( 44.61,48.27)(47.435,51.095)(50.26,53.92 ) ( 44.61,48.27)(47.435,45.445)(50.26,42.62 ) ( 50.26,53.92)(53.085,51.095)(55.91,48.27 ) ( 50.26,42.62)(53.085,45.445)(55.91,48.27 ) ( 13.31,5.65)(18.31,5.65)(23.31,5.65 ) ( 23.31,5.65)(26.135,8.48)(28.96,11.31 ) ( 23.31,5.65)(26.135,2.825)(28.96,0 ) ( 28.96,11.31)(31.785,8.48)(34.61,5.65 ) ( 28.96,0)(31.785,2.825)(34.61,5.65 ) ( 34.61,5.65)(39.61,5.65)(44.61,5.65 ) ( 44.61,5.65)(47.435,8.48)(50.26,11.31 ) ( 44.61,5.65)(47.435,2.825)(50.26,0 ) ( 50.26,11.31)(53.085,8.48)(55.91,5.65 ) ( 50.26,0)(53.085,2.825)(55.91,5.65 ) ( 50.26,11.31)(50.26,16.31)(50.26,21.31 ) ( 50.26,42.62)(50.26,37.62)(50.26,32.62 ) ( 28.96,11.31)(28.96,16.31)(28.96,21.31 ) ( 28.96,42.62)(28.96,37.62)(28.96,32.62 ) ( 13.31,26.97)(18.31,26.97)(23.31,26.97 ) ( 23.31,26.97)(26.135,29.795)(28.96,32.62 ) ( 23.31,26.97)(26.135,24.14)(28.96,21.31 ) ( 28.96,32.62)(31.785,29.795)(34.61,26.97 ) ( 28.96,21.31)(31.785,24.14)(34.61,26.97 ) ( 44.61,26.97)(39.61,26.97)(34.61,26.97 ) ( 44.61,26.97)(47.435,29.795)(50.26,32.62 ) ( 44.61,26.97)(47.435,24.14)(50.26,21.31 ) ( 50.26,32.62)(53.085,29.795)(55.91,26.97 ) ( 50.26,21.31)(53.085,24.14)(55.91,26.97 ) ( 32,30)@xmath162 ( 17,28)@xmath163 the corresponding equations are given in appendix b. the results appear in fig .
4 . one should remark that our scp equations never diverge , even when applied in paradoxical regimes , for instance when one enters in the @xmath169 regime from inter - plaquette bonds .
ratio according to the scp method .
( - - ) from inter - plaquette bond singlets @xmath0 , ( ... ) from nel @xmath151 , and ( ) from plaquette singlets @xmath170 .
the thick symbols identify the relevant domains from the various reference functions .
the arrows indicate the phase transition critical values . ]
the expansion from nel tends to an erroneous ( @xmath171 instead of @xmath172 ) value for @xmath173 ( isolated plaquettes ) . using the perturbative evaluation of the coefficient of the doubly permuted vectors ( eq .
17 instead of eq .
18 ) , one reaches the correct asymptotic value for @xmath173 .
one must concentrate on the estimates of the cohesive energy in the relevant domains . as seen from fig .
4 one notices the existence of three phases . * a dimer phase for @xmath174 * a nel - order phase for @xmath175 * a plaquette phase for @xmath176 . using eq .
17 instead of eq .
18 , the critical ratios remain almost identical .
the method confirms the existence of these three phases and the calculated critical values of the @xmath177 ratios are in good agreement with previously mentioned estimates .
the here presented method is based on perturbative expansion of the wave - function but it is not strictly perturbative .
it is perturbative in the sense that it follows a genealogic generation of the wave function and that it uses ( or refers to ) pertubative estimates of the coefficients of the second generation vectors . + but it is not perturbative in the sense that the coefficients of the first generation vectors are obtained by solving a set of coupled polynomial equations . it may compared to the coupled cluster method ( and in particular to a ccd(s , t ) version ) from which it differs by the consideration of possible deviations from additivity of multiple - excitation energies and by the introduction of @xmath67 corrections in energy denominators .
+ although perfectly applicable to _ ab initio _ calculations , the method has been essentially designed for the study of periodic lattices and model hamiltonians . in such cases , starting from strongly localized zero - order pictures , the method is analytical , it leads to simple polynomial expressions , introducing a very limited number of variables ( one per type of bond in most cases ) , which can be solved on a pocket calculator .
the accuracy of such a simple method is surprising , as illustrated above .
it provides an elegant exploratory tool for the study of spin periodic lattices .
the present work has shown that the method may be employed to identify phase transitions .
the method is of course applicable to hubbard hamiltonians , whose exact solutions are not known ( except for the 1-d regular chain ) and to 2-d or 3-d lattices . since
the method only explores the physics around a given bond ( up to six bonds in each direction ) it may lead to some stimulating qualitative discussions to assess what is local , regional and properly collective in the cohesive energy of delocalized systems .
@xcite further work will study the spin - peierls distortions .
the present method essentially furnishes an estimate of the ground state energy .
it may also provide short range correlation functions .
the localized character of @xmath17 and the fact that the excitation processes playing a role in the low - order processes only concern a few bonds prevent to give estimates of the correlation beyond this limit .
the related coupled cluster formalism has led to specific methods for the evaluation of spectra , namely the equation of motion coupled cluster ( eom - cc ) algorithm .
@xcite future works will analyse the possibility of such an extension in the frame of our development .
1 . equation from nel @xmath178c^2 + 14j \left [ \frac{12\lambda j-14jc}{10\lambda j-13jc}-1\right]c^2 = 0\ ] ] @xmath179 2 .
equation from xy @xmath180c+\frac{j}{2 } \left ( 1+\lambda\right)+2j \left [ 2 \left ( \frac{12j-7j(1+\lambda)c}{8j-6j(1+\lambda)c}\right)-1\right]c^2 \\ \nonumber & & + 14j \left [ \frac{12j-7j(1+\lambda)c}{10j-\frac{13j}{2}(1+\lambda)c}-1\right]c^2 -\frac{j^2(1-\lambda)^2c}{2 \left [ j-\frac{j}{2}(1+\lambda)c\right]}-\frac{j^2(1-\lambda)^2c}{8 \left [ j-\frac{j}{2}(1+\lambda)c\right]}=0\end{aligned}\ ] ] @xmath181\ ] ] 1 .
equation from inter - plaquette bond singlets @xmath184c+\frac{\sqrt{3}}{2}(1+\delta ) -\frac{15(1+\delta)^2c}{4(1-\delta)-\frac{9\sqrt{3}}{2}(1+\delta)c}=0\ ] ] @xmath185\ ] ] 2 .
equations from nel function . the coefficient @xmath106 ( resp .
@xmath186 ) is relative to the spin exchange in the plaquette ( resp . between the plaquettes ) @xmath187c+(1+\delta)+(1+\delta ) \left [ 2\frac{8 - 6(1+\delta)c-4(1-\delta)c'}{4(1-\delta)-4(1-\delta)c'-4(1+\delta)c}-1\right]c^2 \\ \nonumber & & 4(1-\delta ) \left [ \frac{8 - 6(1+\delta)c-4(1-\delta)c'}{4(1+\delta)+2(1-\delta)-4(1+\delta)c-3(1-\delta)c'}-1\right]c^2 \\ \nonumber & & + 2(1+\delta ) \left [ \frac{4 + 4(1+\delta)-7(1+\delta)c-3(1-\delta)c'}{4(1+\delta)+2(1-\delta)-4(1+\delta)c-3(1-\delta)c'}-1\right]cc ' = 0\end{aligned}\ ] ] + @xmath188c'+(1-\delta ) \\
\nonumber & & + 4(1+\delta ) \left [ \frac{4 + 4(1+\delta)-7(1+\delta)c-3(1-\delta)c'}{4(1+\delta)+2(1-\delta)-4(1+\delta)c-3(1-\delta)c'}-1\right]cc ' \\ \nonumber & & 4(1+\delta ) \left [ \frac{8(1+\delta)-8(1+\delta)c-2(1-\delta)c'}{6(1+\delta)-6(1+\delta)c-2(1-\delta)c'}-1\right]c'^2=0\end{aligned}\ ] ] + @xmath189\ ] ] 3 .
equations from intra - plaquette bond singlets .
@xmath106 and @xmath186 concern respectively intra and inter - plaquette diexcitations @xmath190c+\sqrt{3}(1+\delta ) + \frac{2\sqrt{3}(1-\delta ) \left [ \frac{\sqrt{3}}{2}(1-\delta)c+\sqrt{3}(1+\delta)c'\right]}{-4(1+\delta)+2\sqrt{3 } \left [ ( 1-\delta)c'+(1+\delta)c\right]}=0\ ] ] @xmath191c'+\frac{\sqrt{3}}{2}(1-\delta ) \\
\nonumber & & + \frac{3(1-\delta)^2c'+2\sqrt{3}(1+\delta ) \left [ \frac{\sqrt{3}}{2}(1-\delta)c+\sqrt{3}(1+\delta)c'\right]}{-4(1+\delta)+2\sqrt{3 } \left [ ( 1-\delta)c'+(1+\delta)c\right]}=0\end{aligned}\ ] ] @xmath192\right]\ ] ] 9999 k. a. brueckner , phys . rev .
* 100 * , 36 ( 1955 ) .
n. m. hugenholtz , physica .
* 23 * , 481 ( 1957 ) .
j. goldstone , progr , phys .
a * 239 * , 267 ( 1957 ) c. mller , and s. plesset , phys . rev . * 46 * , 618 ( 1934 ) .
j. olsen , o. christiansen , h. koch and p. jorgensen , j. chem .
phys . * 105 * , 5082 ( 1996 ) .
kelly , and a. m. sessler phys
132 * , 2091 ( 1963 ) , phys .
a * 134 * , 1450 ( 1964 ) .
w. meyer , int .
symp . * 5 * , 341 ( 1971 ) , j. chem
. phys . * 58 * , 1017 ( 1973 ) .
r. alrichs , h. lishka , v. staemmler , and w. kutzelnigg , j. chem . phys .
* 62 * , 1225 ( 1975 ) .
j. p. daudey , j. l. heully , and j. p. malrieu , j. chem . phys . * 99 * , 1240 ( 1993 ) . f. coester and h. k@xmath193mmel , nucl .
* 17 * , 477 ( 1960 ) .
j. @xmath194i@xmath195ek , j. chem . phys .
* 45 * , 4256 ( 1966 ) .
g. d. purvis , and r. j. bartlett , j. chem .
* 68 * , 2114 ( 1978 ) .
k. ragavachari , g. w. trucks , j. a. pople , and m. head - gordon , chem .
. lett . * 157 * , 479 ( 1985 ) .
m. urban , j. noga , s. j. cole and r. j. bartlett , j. chem . phys . * 83 * , 4041 ( 1985 ) .
p. s. epstein , phys .
* 28 * , 695 ( 1925 ) .
r. k. nesbet , proc .
london a * 230 * , 312 ( 1955 ) .
claverie , s. diner , and j. p. malrieu , int .
* 1 * , 751 ( 1967 ) .
kelly , , chem .
* 14 * , 142 ( 1966 ) . m. b. lepetit , and j.p .
malrieu , j. chem . phys . * 87 * , 5937 ( 1987 ) . b. miguel , and j.p .
malrieu , phys .
b * 54 * , 1652 ( 1996 ) . b. miguel , m. cousy , and j.p .
malrieu , int .
* 67 * , 115 ( 1998 ) .
j. p. malrieu , and v. robert , j. chem .
phys . * 120 * , 7374 ( 2004 ) .
v. robert , and j. p. malrieu , j. chem . phys . * 120 * , 8853 ( 2004 ) .
r. f. bishop , j. b. parkinson , and y. xian , phys .
b * 43 * , 13782 ( 1991 ) .
r. f. bishop , r. g. hale , and y. xian , phys .
lett * 73 * , 3157 ( 1994 ) .
z. weihong , j. oitmaa , and c. j. hamer , phys .
b * 43 * , 8321 ( 1991 ) .
z. weihong , j. oitmaa , and c. j. hamer , phys .
b * 52 * , 10278 ( 1995 ) .
witte , l.c.l .
hollenberg , and w.h .
zheng , phys .
b * 55 * , 10412 ( 1997 ) . c. j. hamer , z. weihong , and p. arndt , phys . rev .
b * 46 * , 6276 ( 1992 ) .
t. barnes and e. s. swanson , phys .
b * 37 * , 9405 ( 1988 ) .
t. barnes , k. j. cappon , e. dagotto , d. kotchan , and e. s. swanson , phys .
b * 40 * , 8945 ( 1989 ) .
n. trivedi and d. m. ceperley , phys .
b * 41 * , 4552 ( 1990 ) k. j. runge , phys . rev .
b * 45 * , 12292 ( 1992 ) .
u .- j . wiese and h .-
ying , z. phys .
b * 93 * , 147 ( 1994 ) .
b. b. beard and u .- j .
wiese , phys .
lett * 77 * , 5130 ( 1996 ) .
a. w. sandvik , phys .
b * 56 * , 11678 ( 1997 )
. a. w. sandvik and c. j. hamer , phys .
b * 60 * , 6588 ( 1999 ) .
lin , j , s. flynn , and d. d. betts , phys .
b * 64 * , 214411 ( 2001 ) .
m. al hajj , n. guihry , j. p. malrieu , and p. wind , phys .
b * 70 * , 094415 ( 2004 ) .
k. ueda , h. kontani , m. sigrist , and p. a. lee , phys .
lett * 76 * , 1932 ( 1996 ) .
m. troyer , h. kontani , and k. ueda , phys .
lett * 76 * , 3822 ( 1996 ) .
j. f. stanton , and r. j. bartlett , j. chem .
* 98 * , 7029 ( 1993 ) . | the work presents a simple formalism which proposes an estimate of the ground state energy from a single reference function .
it is based on a perturbative expansion but leads to non linear coupled equations .
it can be viewed as well as a modified coupled cluster formulation . applied to a series of spin lattices governed by model hamiltonians the method leads to simple analytic solutions .
the so - calculated cohesive energies are surprisingly accurate .
two examples illustrate its applicability to locate phase transition . |
@xmath0 and @xmath1 bosons are produced in abundance in high energy hadron collisions .
their large mass scale and well known fermion couplings make their production cross sections accurately calculable in perturbative qcd , and their leptonic decay modes provide distinctive experimental signatures that are relatively easy to identify , trigger on and separate from large backgrounds . in many respects
vector boson production has become one of the most important `` standard candles '' in experimental hadron collider physics . at the tevatron ,
measurements of @xmath0 and @xmath1 inclusive cross sections are routinely used to validate detector and trigger performance and stability .
@xmath2 events in particular , due to their negligible backgrounds and trigger redundancy as well as precisely known mass , are the most important samples for the experimental determination of energy and momentum scales and lepton identification efficiencies .
inclusive vector boson cross section measurements are necessary starting points for more detailed measurements , for example differential cross section measurements , exclusive measurements ( e.g. @xmath3 ) , measurements of rare processes ( e.g. diboson production ) and precision measurements of vector boson properties . at the large hadron collider ( lhc )
the use of massive vector bosons as standard candles may be taken one step further with their yields being used to either compute , or provide an alternative definition of , collider integrated luminosities . in the following sections the basic physics of vector boson production at hadron colliders
is reviewed , and the relevant parameters of the cdf and d detectors are stated .
measurements of inclusive and differential production cross sections are discussed in section [ sec : xs ] , as well as some of the particle properties that can be extracted from these measurements .
asymmetry measurements in @xmath1 and @xmath0 production are presented in section [ sec : asym ] .
finally , the status of precision direct measurements of the @xmath0 boson mass and width at the tevatron is given in section [ sec : precision ] .
figure [ fig : w_z_production_schematic ] indicates the phenomenological ingredients required for a complete description of vector boson production in hadron collisions . at leading order , a quark - antiquark pair annihilate to produce a @xmath0 or @xmath1 boson , which subsequently decays to a fermion - antifermion pair .
cross sections are computed as a convolution of partial cross sections ( @xmath4 ) over parton distribution functions ( pdf s ; @xmath5 , where @xmath6 is the fraction of the proton s momentum carried by quark flavor @xmath7 ) : @xmath8 \times { \rm d}\sigma_{q\bar{q}\to w / z\to l\bar{l } } \;\;\ ; { \rm d}x_{p}\ ; { \rm d}x_{\bar{p}}}\;\ ; .\ ] ] the longitudinal momentum ( or rapidity ) distribution of the produced vector bosons depends directly on the pdf s . the rate and kinematic distributions of the decay fermions additionally depend on the couplings and branching ratios , as well as the mass ( @xmath9 ) and width ( @xmath10 ) , of the produced vector boson : @xmath11 \;\ ; , \ ] ] where the parton - parton center of mass energy @xmath12 .
qcd corrections to this process can be divided into perturbative and non - perturbative parts .
perturbative corrections , either in the form of fixed higher order matrix elements or parton showers , modify the vector boson production kinematics - most notably the vector boson transverse momentum ( @xmath13 ) distributions - and give rise to final states with multiple , sometimes high-@xmath13 partons .
non - perturbative effects , sometimes understood as intrinsic-@xmath14 distributions for the colliding partons , are especially important for obtaining a complete description of the low end of the vector boson @xmath13 spectra .
finally , qcd corrections modify the helicity and therefore decay angular distributions of the vector bosons , a small but measurable effect that is important to take into account for precision measurements . by far the most important electroweak corrections to vector boson production and decay are those due to final state photon radiation from charged leptons , where the effect on lepton kinematics and identification is appreciable .
initial state radiation from the colliding quarks and , in the case of @xmath0 production , radiation from the @xmath0 itself , are important to consider in analyses of high-@xmath13 photons produced in association with a vector boson , but have a negligible impact on more inclusive @xmath0 and @xmath1 measurements . or @xmath1 boson , which subsequently decays , in this case to a lepton - antilepton pair .
qcd effects need to be taken into account to give a full description of vector boson production kinematics , as described in the text . ]
cdf and d are general purpose @xmath15 detectors , efficiently triggering on and identifying leptonic decays of @xmath0 and @xmath1 bosons over significant regions of phase space @xcite .
a particular strength of the cdf detector is the excellent central tracking resolution @xmath16 [ @xmath17 , beam constrained ] .
the central electromagnetic calorimeter has an energy resolution @xmath18 .
muon chambers cover the pseudorapidity region @xmath17 and forward calorimeters cover the region out to @xmath19 , important for the accurate reconstruction of missing-@xmath20 .
d has significantly larger muon acceptance with chamber coverage and accurate momentum determination out to @xmath21 , and excellent hermeticity with calorimeter coverage out to @xmath22 .
the performance of the tevatron collider continues to increase , such that at the time of this meeting is has delivered almost @xmath23 @xmath24 to each experiment .
the measurements presented here are based on the first @xmath25 @xmath26 of run ii data .
the measurements of inclusive @xmath0 and @xmath1 production cross - sections in leptonic decay modes provide a benchmark for all analyses of events containing high @xmath13 leptons at the tevatron .
a compilation of both run i and run ii measurements from cdf and d , based on run ii datasets between @xmath27 @xmath26 and @xmath28 @xmath26 , is shown in figure [ fig : xsec_w_z_inclusive ] , where the good agreement with predictions calculated at nnlo in qcd can be observed @xcite .
@xmath29 \includegraphics[height=.35\textwidth]{xsec_w_narrow_range.jpg } & \includegraphics[height=.35\textwidth]{xsec_z_narrow_range.jpg } \\ [ 0.0 cm ] % \mbox{\bf ( a ) } & \mbox{\bf ( b ) this } \end{array}$ ] in electron and muon decay channels , the best measurements are systematically limited at the @xmath30 level , not including an overall @xmath31 uncertainty on the integrated luminosity ( completely correlated for all measurements from a single experiment , and partially correlated between cdf and d ) .
the largest contributions to the non - luminosity systematic derive from pdf uncertainties , and residual experimental uncertainties on lepton identification efficiencies and backgrounds .
interestingly , with non - luminosity systematic uncertainties of @xmath32 , and given the uncertainty on the theoretical cross section predictions of approximately @xmath33 , the cdf measurements are already providing a cross check of the integrated luminosity measured for that experiment .
the ratio of @xmath0 to @xmath1 leptonic cross sections can be written as : @xmath34 with the ratio of inclusive production cross sections taken from a nnlo calculation , the measurement of the @xmath1 leptonic branching ratio from lep and a standard model calculation of the @xmath0 leptonic decay partial width , a measurement of @xmath35 can therefore be interpreted as an indirect determination of the full @xmath0 decay width . the cdf @xmath27 @xmath26 analysis results in a measurement @xmath36 @xmath37 that is already comparable in precision with the world average result .
analyses are currently being developed with the goal of minimizing the systematic uncertainty on the cross section ratio rather than the individual @xmath0 and @xmath1 cross sections , in order to further improve the indirect @xmath0 width determination .
tau decay modes of @xmath0 and @xmath1 bosons are experimentally much more challenging in many respects , including triggering , identification and separation from backgrounds .
however the physics reward is significant , allowing tests of @xmath38 generation lepton universality and establishing a benchmark for many searches , most notably the search for the mssm higgs bosons .
both cdf and d have recently performed cross section measurements for @xmath39 production , where one @xmath40 decays leptonically and the other hadronically @xcite .
the reconstruction of the hadronic tau decay modes involves combining information from both the tracking and calorimetry , with constraints from the tau mass and decay topologies .
figure [ fig : z_tt ] shows the mass distributions from the two analyses , showing good agreement with monte carlo after applying background corrections .
the corresponding cross sections : @xmath41 [ d ; @xmath42 @xmath26 ] @xmath44 [ cdf ; @xmath28 @xmath26 ] agree very well with measurements in the electron and muon channels . in addition
cdf have a measurement of the @xmath0 production cross section in the hadronic tau decay channel using @xmath27 @xmath26 , and a corresponding determination of the ratio @xmath45 .
an improved measurement of this ratio with much higher integrated luminosity will be particularly interesting to compare to measurements from lep ii .
@xmath29 \includegraphics[height=.35\textwidth]{d0_ztt_mass.jpg } & \includegraphics[height=.35\textwidth]{cdf_ztt_mass.png } \\ [ 0.0 cm ] % \mbox{\bf ( a ) } & \mbox{\bf ( b ) } \end{array}$ ] cdf and d have moved on from inclusive cross sections to the measurement of differential cross sections . measuring the drell - yan cross section @xmath46 over as large
a mass range as possible controls an important background to searches for new physics in dilepton channels .
both the mass and the rapidity is a function of energy and longitudinal momentum : @xmath47 differential cross section @xmath48 can , with sufficient integrated luminosity , be used to constrain pdf s and test higher - order qcd .
figure [ fig : xsec_diff ] shows measured drell - yan mass ( @xmath1 boson rapidity ) differential cross sections in muon ( electron ) channels from cdf ( d ) , both with @xmath49 @xmath26 of data .
the agreement with theoretical predictions , after careful unfolding of the data for smearing effects , is very good .
differential cross section measurements for @xmath0 s are harder than for @xmath1 bosons , due to the incomplete kinematic reconstruction of events with neutrinos in the final state . as a first attempt
, the cdf experiment has measured the cross section for @xmath0 production with forward electrons ( @xmath50 ) in @xmath51 @xmath26 of data , using silicon - only small angle tracking and with a different triggering strategy to central leptons .
the ratio of central to forward @xmath0 cross sections is particularly interesting : @xmath52 the prediction for this ratio depends on the choice of pdf set : @xmath53 although not yet significant , there is clearly hope that with more data , this and other differential cross section measurements at the tevatron can yield useful inputs to global pdf fits .
@xmath54 \includegraphics[height=.32\textwidth]{cdf_dsdm_plot_dt_znorm.png } & \includegraphics[height=.32\textwidth]{d0_dsdy_dsdy_e12f05.jpg } \\ % \mbox{\bf ( a ) } & \mbox{\bf ( b ) } \end{array}$ ]
in addition to cross sections , measured asymmetries are particularly useful for the extraction of underlying physics parameters such as electroweak couplings and pdf s .
many systematics cancel in the measurement of asymmetries as opposed to absolute cross sections , and in certain cases ( most notably the search for @xmath55 bosons ) , hints of new or unexpected physics may first turn up as asymmetry rather than rate anomalies .
the drell - yan forward - backward asymmetry compares the fractions of events for which the negatively charged lepton is produced in the forward and backward hemispheres , defined in the parton - parton center of mass system : @xmath56 the result of a recent cdf update to their published @xmath57 analysis @xcite is shown in figure [ fig : asym ] .
the structure around the @xmath1 pole , as well as the measured asymmetry at high mass where indications of new physics may first be expected to be observed , agree well with theoretical predictions . the previous analysis has also used the measured asymmetry to place constrains on the neutral current light quark couplings .
although less precise , the tevatron @xmath57 data , along with data from the analysis of deep inelastic scattering at hera , do break a degeneracy in the determination of light quark couplings at lep , providing an important confirmation of the standard model .
the observed @xmath0 charge asymmetry , defined with respect to the decay lepton direction , is a combined effect of both the underlying @xmath0 production asymmetry and @xmath58 decay asymmetry : @xmath59 the expected @xmath0 charge asymmetry is sensitive to the assumed pdf , in particular the @xmath60 ratio at high-@xmath6 , and benefits from increased statistics with respect to the @xmath1 rapidity measurement .
important experimental issues are forward lepton identification and , crucially , well controlled charge mis - identification rates in order to correct the observed asymmetry for any dilution due to lepton charge mis - assignment .
figure [ fig : asym ] shows a recent @xmath0 charge asymmetry measurement by d in the muon channel , where charge mis - identification rates were kept at the @xmath61 level out to muon pseudorapidities of 2 .
interestingly , the measurement uncertainties are now appreciably smaller than the variation in theoretical predictions that can be generated by varying the input pdf set .
this indicates that this data would already contribute additional constraining power to future global pdf fits .
cdf have measured the asymmetry using electrons in a smaller dataset @xcite .
they have additionally shown that certain lepton kinematic regions are more sensitive to pdf variations than others .
analyses are currently underway to fully exploit all the kinematic information available in @xmath0 events with the goal of unfolding directly back to the underlying @xmath0 production asymmetry .
@xmath62 % \includegraphics[height=.35\textwidth]{cdf_plugw_mtw_final_met25_blessing_new.png } & \includegraphics[height=.32\textwidth]{cdf_afb_afb_blessing_v4.png } & \includegraphics[height=.32\textwidth]{d0_w_asym_asym_e14f07_right.jpg } \\ [ 0.0 cm ] % \mbox{\bf ( a ) } & \mbox{\bf ( b ) } \end{array}$ ]
the precision determination of the @xmath0 mass is one of the most important measurements to be performed at the tevatron .
@xmath63 is a critical input to standard model fits which can constrain the mass of the unobserved higgs boson or , subsequent to a discovery of a higgs boson , may give indications as to what lies beyond the standard model .
the width of the @xmath0 is a less sensitive observable in global electroweak fits , but the direct measurement of the @xmath0 width nevertheless confirms a basic prediction of the standard model and provides a useful cross check of indirect measurements .
the most precise top mass measurements at the tevatron are at the level of @xmath64 .
by contrast the goal for the @xmath63 is a measurement substantially better than @xmath65 .
this highlights the extremely difficult nature of this measurement , requiring exquisite understanding of @xmath0 production and decay , detector response and the effect of backgrounds .
cdf have estimated that in the first @xmath66 @xmath26 of run ii data , they will measure the @xmath0 mass with a precision of at least @xmath67 @xmath68 , better than the combined tevatron run i uncertainty .
d have started with a direct measurement of the @xmath0 width in @xmath69 @xmath26 of run ii data : @xmath70 @xmath37 .
both experiments are finalizing new measurements of the @xmath0 mass , with an ultimate goal for run ii of @xmath71 @xmath68 per experiment .
the cdf and d experiments have completed benchmark measurements of inclusive @xmath0 and @xmath1 production , as well as differential cross section and asymmetry measurements .
these measurements are providing important information on fundamental @xmath0 and @xmath1 properties , as well as parton distribution functions and other aspects of vector boson production physics . with the basic signatures and experimental techniques involved in the measurement of @xmath0 and @xmath1 production well understood ,
attention is now focused on optimizing analyses for sensitivity to pdf s and other physics parameters of interest .
all of this information is being brought to bear in the precision measurement of the @xmath0 mass in run ii , the first results on which are expected soon .
ultimately , measurements of @xmath0 and @xmath1 production at the tevatron will be used as inputs to the accurate modeling of vector boson production at the lhc , crucial for using these processes to help commission and calibrate lhc detectors . in general , vector boson production promises to continue to be one of the most useful standard candles of hadron collider physics .
d. acosta _ et al .
_ ( cdf collaboration ) , phys . rev .
d * 71 * , 032001 ( 2005 ) . v. m. abazov _ et al . _
( d collaboration ) , `` the upgraded d detector '' , hep - physics/0507191 ( submitted to nucl .
instrum .
methods phys .
res . a ) .
d. acosta _ et al .
_ ( cdf collaboration ) , phys . rev
* 94 * , 091803 ( 2005 ) ; hep - ex/0508029 ( submitted to phys .
d ) . v. m. abazov _ et al . _
( d collaboration ) , phys .
d * 71 * , 072004 ( 2005 ) ; a. safonov ( for the cdf collaboration ) , nucl .
. suppl . * 144 * 323 ( 2005 ) ; d. acosta _ et al . _
( cdf collaboration ) , phys .
d * 71 * , 052002 ( 2005 ) .
d. acosta _ et al .
_ ( cdf collaboration ) , phys . rev .
d * 71 * , 051104(r ) ( 2005 ) . | an overview of @xmath0 and @xmath1 production in high energy hadron collisions is given . @xmath0 and @xmath1 cross section and asymmetry measurements from cdf and d are described , with particular emphasis on recent results .
the current status of precision @xmath0 mass and width measurements is reported .
the fundamental physics parameters that can be extracted from these measurements are described , and the relevance of @xmath0 and @xmath1 production studies for the lhc is pointed out .
address = department of physics & astronomy , university college london , london wc1e 6bt , u.k . |
we use the degenerate - band ( multichannel ) kondo lattice model @xmath7where @xmath8 is the band energy , @xmath9 are spin-1/2 operators , @xmath10 is the @xmath11 exchange parameter , @xmath12 are the pauli matrices , @xmath13 is the orbital degeneracy index . for the sake of convenient constructing perturbation theory
, we explicitly include the heisenberg @xmath14 exchange interaction @xmath15 in the hamiltonian , although in fact this interaction is usually the indirect rkky coupling .
the density of states corresponding to the spectrum @xmath8 is supposed to contain a van hove singularity ( vhs ) near the fermi level . in particular
, for the square lattice with the spectrum @xmath16 we have the density of states@xmath17where @xmath18 is the complete elliptic integral of the first kind , the bandwidth is determined by @xmath19 . at @xmath4
the fixed point lies in a weak coupling region which makes possible successful application of perturbation and renormalization group approaches .
we apply the poor man scaling approach @xcite .
this considers the dependence of effective ( renormalized ) model parameters ( effective @xmath11 coupling and spin - fluctuation frequencies ) on the flow cutoff parameter @xmath20 to find the equation for the renormalized coupling parameter @xmath21 we pick out in the sums for the kondo corrections the contribution of intermediate electron states near the fermi level with @xmath22 bearing in mind a nfl - type behavior , we write down the scaling equations to next - leading order by taking into account the corrections of order of @xmath23 .
@xmath24\eta ( -\frac{% \overline{\omega } } { c})\delta c / c \label{ief}\ ] ] where @xmath1 is a characteristic spin - fluctuation energy , @xmath25 is the scaling function satisfying the condition @xmath26 in the simple spin - diffusion approximation ( for a paramagnet ) and spin - wave approximation ( for a magnetically ordered phase ) we have @xcite@xmath27/[(1-x)^{2}+\delta ^{2}]\ } \\
\eta ^{afm}(x ) & = & -(2x^{2})^{-1}\ln [ ( 1-x^{2})^{2}+4\delta ^{2}]\end{aligned}\ ] ] the scaling functions for the ordered phases contain van hove singularities at @xmath28 the renormalizations of magnetic moment and spin - wave fluctuation frequency are given by @xcite@xmath29the latter result holds for all magnetic phases with @xmath30 for the paramagnetic ( pm ) phase , @xmath31 for the ferromagnetic ( fm ) phase , @xmath32 for the antiferromagnetic ( afm ) phase . here
@xmath33 are some averages over the fermi surface ( see ref.@xcite ) , in the approximation of nearest - neighbor at the distance @xmath34@xmath35for the staggered afm ordering @xmath36where @xmath37 for the square and cubic lattice , @xmath38 and @xmath39 are the exchange integrals between nearest and next - nearest neighbors ( @xmath40 in the nearest - neighbor approximation . adopting the definition of the effective ( renormalized ) and bare coupling constant@xmath41we obtain@xmath42g_{ef}^{2}(\xi )
\psi ( \lambda + \psi -\xi ) , ~ \label{sc2}\]]@xmath43where @xmath44 we put in spirit of scaling consideration @xmath45 ) , @xmath46 first we discuss briefly the one - impurity case ( @xmath47 ) . the lowest - order ( one - loop ) scaling calculation according to ( [ ief ] ) yields the divergence of @xmath48 at the kondo temperature@xmath49
\label{perturb}\]](see also refs.@xcite ) on introducing @xmath50(owing to the structure of perturbation theory , this quantity enters also corrections for physical properties like magnetic susceptibility and specific heat ) we have an approximate scaling equation @xmath51g_{ef}^{2}(\xi ) , \]]which is similar to that for @xmath52 in the smooth dos case ( where the factors @xmath53 are absent ) . in the latter case
the two - loop equation corresponding to ( [ sc2 ] ) gives a finite fixed point @xmath54 .
it is known that this point is unphysical ( unreachable ) for @xmath6 , but for @xmath4 the scaling consideration gives a qualitatively correct description @xcite ( the case @xmath5 is marginal , so that additional logarithmic factors occur in the physical properties ) . unlike the smooth dos case , the equation ( [ sc2 ] ) can not be solved analytically even in the one - impurity case , where only asymptotic solution at large @xmath55 can be obtained : @xmath56the second term in brackets can change sign , being positive for large @xmath3 and negative for small @xmath3 , so that occurrence of a maximum in the dependence of @xmath57 is possible .
besides that , the factor in ( [ expansion ] ) is well determined only within the @xmath58-expansion .
thus the solution is rather sensitive to details of approximations .
the above results demonstrate existence of the fixed point@xmath59 the scaling trajectories approaching this by the law @xmath60unlike the power law in the case of smooth dos @xcite . writing down the kondo correction to magnetic susceptibility by analogy with refs.@xcite
, we obtain the scaling equation for the effective magnetic moment@xmath61so that to leading order@xmath62it should be noted that in the non - singular dos case the power - law critical behavior like ( [ pow ] ) takes place in a wide region , including @xmath63 and @xmath64 gan .
thus , unlike the situation of total screening ( where the strong - coupling region can not be described by simple methods ) we have an interpolation description . on taking into account higher orders in @xmath58
one has in the flat band one - impurity case @xcite @xmath65which agrees with the bethe ansatz solution , see ref .
as demonstrated above , the structure of perturbation theory is similar to the non - singular case with the replacement @xmath66 . to establish properly the correspondence with the one - impurity case ( delts ) ,
we may put @xmath67 this yields for @xmath4 correct critical exponents for magnetic susceptibility , specific heat and resistivity .
the important case @xmath5 is more difficult from the theoretical point of view , although the resistivity is still described by the @xmath58 expansion , see @xcite .
we present below numerical results for @xmath6 ( @xmath68 ) and for @xmath69 ( @xmath70 ) ; the latter case may be relevant for ce@xmath71 ion @xcite .
the dependences @xmath72 and @xmath73 from solution of eqs.(sc2 ) , ( [ sc3 ] ) in the paramagnetic phase are shown in fig.1 . during the scaling process
@xmath73 increases according to ( [ sc3 ] ) .
one can see that @xmath72 demonstrates a plateau at @xmath74 , which can be named a quasi - fixed point .
provided that @xmath75 is not too small , at intermediate @xmath55 ( near the plateau ) we can put for rough estimations @xmath76 to obtain @xmath77(@xmath78 ) .
thus a power - law behavior occurs @xmath79which corresponds to the one - impurity nfl behavior ( [ pow ] ) .
the dependence ( [ chi ] ) takes place up to the point @xmath80for
@xmath81 @xmath82 is practically constant since @xmath83 becomes small .
( a ) and @xmath84 ( b ) .
the parameter values are @xmath85 , @xmath86 , @xmath87 ( @xmath88 , solid lines ) and @xmath89 ( @xmath90 , dashed lines ) for the curves from below to above ( when considering the left - hand part of the figure ) respectively , title="fig:",width=316 ] ( a ) and @xmath84 ( b ) .
the parameter values are @xmath85 , @xmath86 , @xmath87 ( @xmath88 , solid lines ) and @xmath89 ( @xmath90 , dashed lines ) for the curves from below to above ( when considering the left - hand part of the figure ) respectively , title="fig:",width=316 ] ( a ) and @xmath84 ( b ) .
the singularity is shifted from the fermi level by @xmath91 .
other parameter values are the same as in fig.1 , title="fig:",width=316 ] ( a ) and @xmath84 ( b ) .
the singularity is shifted from the fermi level by @xmath91 .
other parameter values are the same as in fig.1 , title="fig:",width=316 ] when the dos singularity is shifted from the fermi level by the distance @xmath92 , its influence on the scaling behavior becomes weaker .
a similar effect occurs when the logarithmic singularity is smeared ( e.g. , small electron damping is introduced , @xmath93 ) . at small @xmath92 the scaling behavior
is determined by a combined action of the kondo and van hove singularities .
the influence of the shift is described by the replacement @xmath94 in the singular factors in the scaling equations .
the scaling trajectories for @xmath91 are shown in fig.2 .
the influence of the shift on the dependence @xmath72 is more pronounced than that on @xmath73 ( for the latter , we have some quantitative difference only ) . on the other hand , at not too small @xmath92 and @xmath75 the dos singularity becomes unimportant , so that with increasing @xmath55 we come rapidly to a non - physical fixed point , @xmath95 with large @xmath96 .
since exact divergence of dos is not required , we can suppose that not only strong logaritmic singularities , but also weaker vhs ( e.g. , those in 3d cubic lattices ) may change considerably the scaling behavior .
note that the increase of @xmath73 in paramagnetic phase is much more stronger than for a smooth dos ( cf .
@xcite ) .
the interval of bare coupling constant , where a nfl - like behavior occurs , is very wide : we come to a quasi - fixed point independently of @xmath75 , although this point becomes unstable with increasing @xmath55 . on the other hand , the one - loop scaling yields for finite @xmath3 the nfl behavior in a narrow interval of the bare coupling constant @xmath75
only , since with increasing @xmath75 we come rapidly to strong - coupling regime where @xmath97 the critical value @xmath98 being rather small @xcite . on the contrary , in the two - loop scaling there is no such a critical @xmath75 value at all : @xmath52 remains finite for any @xmath75 in the paramagnetic case .
thus the lowest - order scaling can not describe properly the case of not too small @xmath75 . in the region of the plateau ( @xmath99 const )
, @xmath52 decreases with increasing @xmath55 .
therefore the scaling curves @xmath72 can intersect each other for different @xmath100 however , this feature disappears when introducing small shift of the singularity from the fermi level , the size of the plateau decreasing ( fig .
2 ) . from ( [ pm ] )
we obtain the power - law dependence of magnetic susceptibility @xmath101as demonstrate numerical renormalization group ( nrg ) calculations for the one - impurity kondo model with vhs @xcite , the local magnetic susceptibility @xmath102 ( which just determines spin correlation functions and therefore corresponds to present calculations ) has a slight maximum tending to a constant value with lowering temperature for @xmath6 and demonstrates a power - law nfl behavior for @xmath5 ( unlike logarithmic behavior in the flat - band case ) .
thus the tendency to nfl behavior increases in the presence of vhs .
now we come to the situation of magnetic ordering . of especial interest
is ferromagnetic state : its realization is connected with large density of states at the fermi level , so that peculiarities of the nfl state in the presence of vhs should be also treated . in magnetically ordered phases ,
the behavior for @xmath103 is similar to that in a paramagnet , but the situation for @xmath104 changes since the van hove singularity of @xmath105 at @xmath106 plays an important role . instead of decreasing
, @xmath107 starts to increase at approaching @xmath108 .
( a ) and @xmath84 ( b ) .
the parameter values are @xmath85 , @xmath109 , @xmath110 ( @xmath88 , solid lines ) and @xmath111 ( @xmath90 , dashed lines ) for the curves from below to above respectively , the damping parameter is @xmath112 , title="fig:",width=316 ] ( a ) and @xmath84 ( b ) .
the parameter values are @xmath85 , @xmath109 , @xmath110 ( @xmath88 , solid lines ) and @xmath111 ( @xmath90 , dashed lines ) for the curves from below to above respectively , the damping parameter is @xmath112 , title="fig:",width=316 ] for sufficiently large @xmath75 ( needed to reach the appreciable value during the scaling increase at small @xmath55 ) , provided that @xmath113at @xmath104 the argument of the function @xmath114 in ( [ sc3 ] ) becomes almost constant ( fixed ) , @xmath115 .
thus further behavior is determined by the singularity of the scaling function and is similar to that for smooth dos @xcite .
we have for the frequency and magnetic moment @xmath116 the scaling curves for a ferromagnet are shown in fig.3 ( in the antiferromagnetic case the picture is qualitatively the same , cf . ref .
i16 ) .
( a ) and @xmath84 ( b ) .
the parameter values are @xmath69 ( @xmath117 ) , @xmath118 , @xmath109 , @xmath119 for the curves from below to above respectively , the damping parameter is @xmath120,title="fig:",width=316 ] ( a ) and @xmath84 ( b ) .
the parameter values are @xmath69 ( @xmath117 ) , @xmath118 , @xmath109 , @xmath119 for the curves from below to above respectively , the damping parameter is @xmath120,title="fig:",width=316 ] thus the scaling behavior is changed at some critical value @xmath98 . above the critical value @xmath98 , the qualitative picture of the nearly linear scaling trajectories @xmath73 does not depend on @xmath75 : they are almost parallel and slowly come together . in turn
, there is a value of the damping parameter @xmath121 ( which ia a cutoff for the singularity in the scaling function determining @xmath122 ) , so that for @xmath123 the infinite linear behavior does not occur .
this value , @xmath124 is determined by the values of @xmath125 and @xmath126 for large @xmath121 and @xmath127 the linear dependence of the type ( [ lin ] ) can take place in a restricted region changing the behavior of the type ( chi ) , so that two nfl - like regions are observed ( see fig.4 , @xmath128 is about 0.015 for @xmath69 ) .
for @xmath6 the critical damping is not small : @xmath128 is smaller than 0.1 .
this is favorable for occurrence of the nfl regime ( [ lin ] ) : it can take place in the case of a somewhat pronounced ( even not too sharp ) peak in the scaling function @xmath129 .
of course , the simple model with constant damping can be generalized .
so , the increase of the damping with growing of @xmath52 was considered in the scaling versions of refs.@xcite . as well as in the paramagnetic case ,
a small shift of the singularity from the fermi level results in a change of @xmath72 behavior at large @xmath130 but influences weakly the behavior @xmath73 .
remember again that the quantity @xmath73 determines the temperature dependences of magnetic moment and thermodynamic characteristics ( for the corresponding discussion in the antiferromagnetic case , see refs .
731,i16 ) .
we have demonstrated that the system with van hove singularities in electron spectrum demonstrates non - fermi - liquid behavior in a wide parameter region .
this system is characterized by moderate specific heat , but large magnetic susceptibility .
as follows from ( [ perturb ] ) , the value of the kondo temperature is rather high .
therefore , heavy - fermion behavior is not expressed , but a tendency to magnetic ordering occurs , which is characteristic for weak itinerant magnets with vhs like zrzn@xmath131 . the results of the present paper demonstrate strong renormalization of magnetic moment ( much stronger than in the case of small dos ) .
the results obtained for the scaling behavior are quantitatively reliable for @xmath4 , but should be verified by more strict analytical and numerical methods , including functional renormalization group ( frg ) .
the corresponding problems of higher - order scaling are also discussed in recent works kapustin , kapustin1 . it should be noted that occurrence of giant van hove singularities ( which are important , e.g. , for ferromagnetism of iron ) is intimately connected with intersection of more weak singularities , i.e. with degeneracy of electron bands @xcite .
already in the classical textbook on magnetism mattis such a degeneracy is considered as a key to itinerant ferromagnetism .
the research was carried out within the state assignment of faso of russia ( theme `` quantum '' no .
01201463332 ) .
this work was supported in part by ural branch of russian academy of science ( project no . 15 - 8 - 2 - 9 ) and by the russian foundation for basic research ( project no
. 16 - 02 - 00995 ) . | a renormalization group treatment of the kondo model with logarithmic van hove singularity in the electron density of states is performed within next - leading order scaling , different magnetic phases being considered .
the effective coupling constant remains small , renormalized magnetic moment and spin - fluctuation frequency decreasing by several orders of magnitude .
thus wide non - fermi - liquid behavior regions are found from the scaling trajectories in a broad interval of the bare coupling parameter .
applications to itinerant magnetism theory are discussed .
rare - earth and actinide compounds with anomalous electronic properties ( kondo lattices and heavy - fermion systems ) are studied extensively starting from the middle of 1980s @xcite .
they possess highly unusual electronic properties and magnetism @xcite .
besides heavy - fermion features ( large electronic specific heat ) , they demonstrate the non - fermi - liquid ( nfl ) behavior : logarithmic or anomalous power - law temperature dependences of magnetic susceptibility and specific heat stewart1l.the magnetism of these systems which has both localized and itinerant features .
thus a combined treatment of kondo and itinerant - electron systems becomes usual @xcite main role in the physics of the kondo lattices belongs to the interplay of the on - site kondo screening of magnetic moments and intersite exchange interactions inducing magnetic order .
this idea was developed in a series of the papers @xcite treating the mutual renormalization of two characteristic energy scales : the kondo temperature @xmath0 and characteristic spin - fluctuation frequency @xmath1 .
the corresponding scaling consideration of this renormalization process in the @xmath2 exchange model @xcite yields , depending on the values of bare parameters , both the usual states ( a non - magnetic kondo lattice or a magnetic state with weak kondo corrections ) and the peculiar magnetic kondo - lattice state , including the nfl behavior .
however , the region of the latter depends strongly on the approximations and the concrete model of electron and magnetic structure @xcite .
the nfl behavior can be related to peculiar features of electron and spin fluctuation spectrum , as well as in the itinerant systems ( which are usually described by the hubbard model ) .
it is well known that an important role in the formation of magnetism belongs to peculiarities of bare electron structure .
the case of singular density of states ( dos ) was considered in ref.@xcite within the lowest - order scaling , a considerable increase of nfl region being demonstrated , but the situation did not qualitatively change . on the other hand ,
the nfl behavior occurs naturally in the one - impurity @xmath3-channel kondo model @xcite .
this model , which assumes existence of degenerate electron bands , explains power - law or logarithmic behavior of electronic specific heat and magnetic susceptibility @xcite .
physically , such a behavior is connected with overscreening of impurity spin by conduction electrons in many channels .
the model permits a consistent scaling investigation in the next - leading approximation .
a characteristic feature of this approximation is occurrence of an intermediate fixed point .
this is reasonable for @xmath4 since the fixed point is within the weak - coupling region ( however , the marginal case @xmath5 requires a more accurate consideration ) . on the other hand , for @xmath6
the fixed point is unphysical .
the situation for the lattice is more complicated , especially in the case singular dos , since the singularities change the structure of perturbation theory . in the present paper
we treat the kondo lattice model with the electron spectrum containing a logarithmic density - of - states ( dos ) singularity to compare the results of leading and next - leading order scaling .
this singularity is typical , in particular , for the two - dimensional case .
the scaling equations are discussed in sect.2 .
the numerical results are presented in sect.3 for the for the cases of a paramagnet with smooth spin spectral function and a ferromagnet ( singular spin spectral function ) . |
elastic scattering of light by a conducting medium , and thus the reflection and absorption of a beam of light passing through a slab of material , are largely determined by the interaction of the beam with the material s free electrons . if an external magnetic field is applied to the medium , the lorentz force acts on the conduction electrons , and
if , for definiteness , the electric component of the incident beam is linearly polarized along the @xmath0-direction ( see ) , the transmitted as well as the reflected beam both acquire a component along the @xmath1-direction .
this rotation of the polarization direction in the reflected ( transmitted ) beam is called the kerr ( faraday ) effect .
the direction of the rotation depends only on the direction of the magnetic field , not on the propagation direction of light , such that faraday and kerr effect enable non - reciprocal optical devices@xcite like optical isolators , which allow light to pass in one direction while blocking propagation in the reverse direction .
for the particular case of a beam of light of frequency @xmath2 passing through an atomically thin sheet of monolayer graphene in a constant magnetic field of magnitude @xmath3 , one finds that the polarization is rotated by the faraday angle@xcite @xmath4 under typical experimental conditions , monolayer graphene can rotate the polarization by several degrees=@xmath5=1 , such that @xmath6=@xmath7=@xmath8 radians . in the typical experimental setting ( see for example ref .
@xcite ) , graphene is deposited on a dielectric substrate .
depending on the refractive index of the specific substrate material used , the prefactor @xmath9 in eqn .
( [ eqn : tf_def ] ) is modified , such that the rotation angles reached with graphene - on - sic devices are by a factor of @xmath101.7 smaller than those we show for free - standing graphene in fig .
[ fig : classicplot ] . ] , a large effect considering that faraday isolators based on conventional magneto - optical materials require propagation distances of several millimeters@xcite . neglects terms of higher than linear order in the fine - structure constant , and it is assumed that the magnetic field as well as the propagation direction of the incident light are both perpendicular to the graphene surface .
@xmath11 denotes the off - diagonal element of graphene s conductivity - tensor , such that a measurement of @xmath12 represents an optical ( that is , @xmath13 ) analogue to a dc measurement of the hall current . after passing through graphene in a perpendicular magnetic field .
[ fig : sketch ] ] in the static limit , the dc hall conductivity of graphene is known to show a step - like behavior as a function of the filling factor @xmath14 @xcite , @xmath15 where @xmath14 marks the last filled level right below the fermi energy @xmath16 . for low carrier densities and large magnetic fields , the landau levels around @xmath16
are well separated , and correspondingly , a faraday angle displaying quantized steps as in has been observed at @xmath2=1thz for graphene samples with @xmath16=60mev and @xmath3 up to 7 t @xcite . however , the spacing of graphene s landau levels @xmath17 is non - equidistant@xcite , with the energy difference between adjacent levels getting smaller for higher @xmath18 .
correspondingly , for modest magnetic fields , and with carrier densities in typical graphene - on - substrate samples being far from the neutrality point , the experimentally observed @xmath19@xcite follows the classical off - diagonal drude conductivity@xcite , @xmath20 } , \label{eqn : sxy_drude_int}\end{aligned}\ ] ] which does not resolve the quantized landau levels .
is parametrized in terms of a cyclotron frequency , then @xmath21 with the effective electron mass @xmath22 ( see ref .
@xcite ) leads to . ]
@xmath23 @xmath24 denotes the fermi velocity , @xmath25=@xmath26=2 are factors for the spin- and valley degeneracy , and @xmath27 is a phenomenological scattering time usually employed as a fitting parameter @xcite
. shows @xmath12 versus frequency calculated with for typical values of @xmath16 and @xmath3 , and assuming @xmath27=1.5@xmath2810@xmath29s , the value corresponding to the transport scattering time @xmath30 for a graphene sample with dc mobility @xmath31=10000@xmath32v@xmath33s@xmath33@xcite .
the rotation angle is seen to be maximal at frequencies slightly below @xmath34 .
assuming @xmath35 , which allows for closed cyclotron orbits between collisions , yields @xmath36\bigr\ } \label{eqn : max}\end{aligned}\ ] ] for this maximal value of @xmath12 . from , it is clear that the magnitude of the achievable faraday rotation strongly depends on @xmath27 , that is , on the quality of the specific graphene sample . versus frequency , calculated from the classical drude formula for @xmath37 .
the horizontal line indicates the maximal value of @xmath12 [ see ] at frequencies below the cyclotron frequency @xmath34 .
the scattering time has been set to @xmath27=@xmath38=1.5@xmath2810@xmath29s .
[ fig : classicplot ] ] due to the particular structure of graphene s two - dimensional crystal lattice , which results in a two - component spinor structure of the electron wavefunctions , it is not only the magnitude of disorder that determines carrier dynamics .
rather , a microscopic description of disorder effects in graphene has to distinguish between scalar- and gauge potentials@xcite .
while the former are typically caused by charged impurities , the latter arise due to deformations of graphene s crystal lattice , and their effect on the charge carriers can be described in terms of pseudomagnetic fields@xcite .
pseudomagnetic fields affect carriers near the two inequivalent fermi points @xmath39 and @xmath40 with an opposite sign , such that they either add up to or have to be subtracted from the ( real ) external bias field @xmath3 acting on the carriers@xcite .
as it has been found that the strain - induced pseudomagnetic field in graphene nano - bubbles can lead to a landau quantization of the charge carriers equivalent to field strengths of a few hundred tesla@xcite , it seems possible to engineer the magneto - optical response of graphene via the controlled application of non - homogeneous mechanical strain , if a constant pseudomagnetic field can be produced over a large spatial region@xcite .
apart from such engineered strain distributions , random strain fields are present in most available graphene samples , because graphene membranes , whether supported by a substrate material or in suspended devices , are never completely flat . instead
, they show a rippled structure@xcite , and for clean samples , the pseudomagnetic fields caused by these mesoscopic corrugations have been shown to set the dominant limit on electronic transport@xcite . despite this importance of ripples for graphene s transport properties , experiments on the faraday effect in graphene
do not reveal any valley - dependent interplay between ( potentially large ) gauge fields and the magnetic bias field . instead , the data is found to be fitted well by with an appropriately chosen scattering time @xcite .
it is the aim of the present work to investigate the effect of sample - inherent , random strain configurations on the magneto - optical properties of graphene , and to understand why ripple - induced effective magnetic fields of the form @xmath41 in each valley do not seem to play a role in graphene s faraday effect . to this end , we compare the effect of random ripples on the faraday rotation with that of random scalar scatterers .
in order to relate the scattering parameter @xmath27 appearing in to an underlying microscopic scattering mechanism , we assume random gaussian disorder@xcite . denoting the fourier transform of the scattering potential with @xmath42
, disorder scattering can be introduced in perturbation theory by the correlator @xmath43 , where the average is taken over all possible disorder configurations . as the electron - wavefunction @xmath44 is a spinor of rank four in sublattice ( superscripts @xmath45 and @xmath3 ) and valley - space ( subscripts @xmath39 and @xmath40 ) , @xmath46 the correlator forms a tensor @xmath47_{h i ; j k } , \qquad h , i , j , k \in \{1,2,3,4\ } , \label{eqn : corr}\end{aligned}\ ] ] see b. we want to distinguish between a scalar and a gauge potential @xmath48 , and consider [ eqns : corr_match ] @xmath49 where @xmath50 and @xmath51 are 4@xmath284 matrices .
the scalar potential @xmath52 affects carriers on the @xmath45 and @xmath3 sublattices of graphene and in both @xmath39 and @xmath40 valleys in the same manner .
in contrast , the gauge potential is proportional to @xmath51 , defined as @xmath53 where we use the standard 2@xmath282 pauli matrices @xmath54 acting on the sublattice index , that is , @xmath45 or @xmath3 site @xcite .
thus , @xmath55 affects carriers in each valley with opposite sign .
apart from their different tensor structure , we have set both correlators proportional to a constant @xmath56 of dimensions ( energy)@xmath57@xmath28area , as would be the case for sharply localized , @xmath58-like scattering centers . in a more detailed description ,
gauge as well as scalar disorder potentials are likely to show long - range correlations , resulting in an overall power - law dependence of the correlator in @xcite .
as we will see below , the approximation with a momentum - independent constant @xmath56 greatly facilitates the evaluation of otherwise infrared - divergent loop integrals , while still taking into full account the scalar or gauge nature of the disorder .
further , the block - diagonal potentials in neglect possible inter - valley scattering processes , which have been shown to yield only weak contributions to the resistivity of graphene@xcite . in order to assign a value to the disorder strength @xmath56
, we aim to link it to the dc carrier mobility characteristic for each sample .
we assume the dc conductivity of graphene to be of the drude - form @xmath59 , where @xmath38 is obtained from fermi s rule together with a boltzmann weighting factor @xmath60 which supresses contributions of small - angle scattering events ( @xmath61 ) to the dc conductivity@xcite .
then , the definition of @xmath31 as the electrical conductivity per carrier , @xmath62 , where @xmath18 is the carrier density , leads to @xmath63 transport measurements revealed that @xmath31 is independent of the carrier density@xcite , hence the scattering parameter @xmath38 has to be proportional to @xmath64 and , as a function of @xmath31 , has to fulfill the above relation . the retarded electronic self energy due to scattering at the potential @xmath65 or @xmath66 in born approximation ( see the diagram in a ) is given by @xmath67 importantly , it does not distinguish between a scalar or vector character of the underlying disorder .
the imaginary part of @xmath68 describes the broadening of electronic momentum - eigenstates due to the disorder potential , and defines a quantum scattering time @xmath69 , @xmath70 for white - noise disorder , @xmath69 is known to be smaller than @xmath38 by a factor of 2 @xcite , and form together with the requirement , we arrive at @xmath71 for the disorder strength in a graphene sample with mobility @xmath31 .
from kubo s formula , the hall conductivity of graphene can be expressed as@xcite @xmath72 with @xmath73 the fermi distribution , and the integral kernel @xmath74 resulting from a sum over bubble diagrams , see c : @xmath75 \nonumber\end{aligned}\ ] ] the sum over the indices @xmath76 and @xmath77 realizes different combinations of retarded and advanced greens functions @xmath78 for charge carriers , given in
. disorder corrections can enter either through self - energy insertions ( as shown in a ) , which shift the resonances of the greens functions in the bubble diagrams of c , or through vertex corrections , shown in d , see ref .
@xcite . within the lowest - order born approximation
, we assume the real part of disorder - induced self - energies to be negligible , such that they simply introduce the scattering time @xmath69 of into the greens functions in .
for such a constant damping parameter , the bubble diagrams of have been evaluated in refs .
@xcite , and we only recall the main steps of the calculation . after performing the momentum integration in with the help of relation ( [ eqn : otho_l ] ) , @xmath74 can be expressed in terms of the polarization functions @xmath79 ^ 2 - e_n^2 } \nonumber \\ & \times \frac { \hbar\omega ' + i \gamma \hbar/(2\tau_q ) } { \hbar^2[\omega ' + i \gamma /(2\tau_q)]^2 - e_m^2 } , \label{eqn : pidef}\end{aligned}\ ] ] where the landau levels are given by @xmath80 with @xmath81 the magnetic length .
we arrive at@xcite @xmath82 , \label{eqn : kubo_kernel_clean2}\end{aligned}\ ] ] @xmath25=@xmath26=2 denoting spin and valley degeneracy , respectively .
diagrammatic expressions appearing in the calculation of @xmath37 . *
a ) * self energy @xmath68 , see .
* b ) * disorder correlator @xmath43 , see .
* c ) * integral kernel of .
* d ) * vertex correction @xmath83 , see .
* e ) * resummation of vertex corrections leading to eqn .
( [ eqn : sxy_resum_scalar_3 ] ) . ] apart from self - energy corrections contributing to the scattering time , disorder introduces , in leading order of perturbation theory , a vertex correction .
@xmath84_{h j } = \int \frac{d^2q}{(2\pi)^2 } \ , \bigl [ \langle v_{{\mathbf { q } } - { \mathbf { k } } } \otimes v_{-{\mathbf { q } } + { \mathbf { k } } } \rangle_{\rm dis}\bigr]_{h i;j k } \nonumber \\ & \times \bigl[g^{\beta } ( \omega,{\mathbf { q } } ) \gamma_2 g^{\gamma } ( \omega',{\mathbf { q}})\bigr]_{i k } \label{eqn : vertex_correction_general}\end{aligned}\ ] ] ( see d ) , which replaces the bare velocity operator @xmath85 appearing on the right - hand - side of . for @xmath42 a constant independent of @xmath86 , the integration over the loop momentum in can be performed using , and for the scalar and gauge disorder of , we obtain @xmath87 where the entries @xmath88 and @xmath89 of the diagonal matrix are dimensionless functions of the frequencies @xmath90 and @xmath91 : [ eqns : ab_strain_loop ] @xmath92 such that @xmath93 . resumming the vertex correction in a geometric series as indicated in e yields , for either scalar ( superscript @xmath94 ) or gauge ( superscript @xmath56 ) disorder , [ eqn : sxy_resum_scalar ]
@xmath95 here [ eqns : kres ] @xmath96 as the optical response of doped graphene is due to excitation of carriers close to the fermi level , the sum over landau levels in is dominated by only a few terms around the filling factor @xmath97 . for our parameters ,
the main contributions are found to come from the level @xmath98 and @xmath99 , and we only retain these two terms in the numerical evaluation of . within this approximation , the faraday angle calculated from @xmath100 and @xmath101 is shown in , where the blue curve corresponds to scalar disorder and the red one to gauge disorder . assuming a graphene sample with @xmath31=@xmath102@xmath32v@xmath33s@xmath33 , we set @xmath69=@xmath103=7.5@xmath2810@xmath104 . comparing with the classical expression shown in
, we note that neither type of disorder changes the functional form of @xmath19 : the frequency of maximal rotation remains slightly below @xmath34 of , and hence is determined by the external magnetic field .
this is in good accordance with experimental observations , where the data can be captured well by , if @xmath27 is treated as a fitting parameter .
in contrast , a naive model of gauge disorder introducing a constant effective field @xmath105 for carriers in the @xmath39 and @xmath106 valley , respectively , would shift the effective resonance frequencies in to @xmath107 . concerning the magnitude of the achievable rotation angle , shows a larger value for scalar disorder compared to gauge disorder , which stems from the different contribution of the vertex corrections to @xmath100 and @xmath101 . in particular , upon expanding as [ eqns : kexp ] @xmath108 & + \frac{g}{8\pi^2
( \hbar v_f)^4 } [ ( b^{\beta \gamma}_s)^2 - ( a^{\beta \gamma}_s)^2 ] + \mathcal{o}(g^2 ) \nonumber , \\ k_g^{\beta\gamma}(\omega,\omega ' ) & = \frac{1}{4\pi ( \hbar v_f)^2 } ( b^{\beta \gamma}_s - a^{\beta \gamma}_s ) + \mathcal{o}(g^2 ) , \label{eqn : kgexp}\end{aligned}\ ] ] we note that in the case of gauge disorder , vertex corrections to the simple bubble - diagrams are absent at linear order in @xmath56 .
the simple bubble - diagrams given by the @xmath109 terms of are known to reproduce the classical drude formula of , with @xmath110 .
( see for details . ) consequently , shows the maximal rotation angle in a sample with gauge disorder to be well aproximated by @xmath111 , whereas the classical @xmath112 is restored by the contribution of vertex corrections for scalar disorder in the presence of scalar disorder , how a resummation of self - energy insertions and vertex corrections approximately restores the appearance of the transport scattering time @xmath38 in ] .
calculated from the kubo formula eqn .
( [ eqn : sxy_resum_scalar_3 ] ) , for a graphene sample with @xmath113@xmath32v@xmath33s@xmath33 due to scalar- ( blue curve ) and gauge disorder ( red curve ) , respectively . for comparison ,
the black curve re - displays the classical model shown in .
the scattering time @xmath69 in has been set to 7.5@xmath2810@xmath104s , resulting in a disorder strength @xmath114 .
horizontal dashed lines indicate the maximum angle given by .
the dotted curves show the results obtained by using only terms up to linear order in @xmath56 in the integral kernel of the kubo formula .
[ fig : disorderplot ] ]
the dynamics of charge carriers which simultaneously experience the presence of a pseudomagnetic field @xmath115 as well as an external bias field @xmath3 is governed by an overall effective field @xmath41 , depending on whether the carriers reside near the @xmath39 or @xmath40 point in graphene s brillouin zone .
this results in a different spectrum of landau levels for @xmath39 and @xmath40 carriers@xcite , and , in terms of the classical hall conductivity @xmath37 of , one might therefore expect carriers at @xmath39 and @xmath40 to contribute to the faraday effect with a an effective cyclotron frequency of @xmath116 , respectively .
however , instead of resulting in an approximately constant pseudomagnetic field over an extended area , the effect of sample - inherent ripples rather corresponds to the exposure of graphene to a random magnetic field , varying in direction and magnitude on a length scale of @xmath10100@xcite .
the above calculation of @xmath37 , which regards the random pseudomagnetic field as a sublattice - mixing disorder potential , predicts that the frequency dependence of @xmath12 is similar for both scalar- and gauge- disorder mechanisms .
consequently , experimental data on faraday rotation can be fitted well by assuming either scalar or gauge disorder , if an appropriate @xmath27 is used as a phenomenological fitting parameter .
however , we suggest that in case the quantum scattering time @xmath69 of the specific sample is known , the magnitude of the faraday rotation might allow to differentiate between samples with predominantly gauge disorder and those showing predominantly scalar disorder
. an experimental possibility would be to determine @xmath69 by spectroscopic observation of the width of the relevant landau levels near the fermi energy@xcite , and compare it with the scattering time extracted form the maximum faraday rotation observed in the same sample . while our calculations do not predict a shift of the observable effective @xmath34 due to random strain fields , we wish to emphasize that this does not exclude the possibility to influence the magneto - optical properties of graphene with appropriately engineered macroscopic strain configurations .
the research leading to these results has received funding from the european union seventh framework programme under grant agreement 604391 graphene flagship , from the european research council through grant 290846 , and from the spanish ministry of economy and competitiveness under projects mat2014 - 53432-c5 - 1-r and fis2014 - 207432 .
the authors are grateful for the hospitality of the donostia international physics center ( dipc , san sebastian ) , where part of this work was concluded .
the retarded ( advanced ) propagator for two - dimensional charge carriers in graphene reads@xcite @xmath117 ^ 2-e_n^2 } , \label{eqn : gdef}\end{aligned}\ ] ] where @xmath118 and @xmath17 denotes the energy of the @xmath18th landau level , see .
the numerator of is given by @xmath119\gamma^0 \nonumber \\ & \times \biggl\ { l^{(0)}_n\bigl ( 2 { \mathbf { k}}^2 l_b^2 \bigr ) \bigl[1 - i \gamma^1\gamma^2 \operatorname{sign}(e
b ) \bigr]/2 \nonumber \\ & - ( 1-\delta_{n,0})l^{(0)}_{n-1}\bigl ( 2 { \mathbf { k}}^2 l_b^2
\bigr ) \bigl[1 + i \gamma^1\gamma^2 \operatorname{sign}(e b ) \bigr]/2 \biggr\ } \nonumber \\ & - ( 1-\delta_{n,0})4 \hbar v_f { \mathbf { k}}\cdot\vec{\gamma } l^{(1)}_{n-1}\bigl ( 2 { \mathbf { k}}^2 l_b^2 \bigr ) .
\label{eqn : gndef}\end{aligned}\ ] ] the associated laguerre polynomials @xmath120 appearing in fulfill the orthogonality relation@xcite @xmath121 together with the identity @xmath122 see ref .
@xcite , allows to compute the momentum integrals appearing in and of the main text .
the presence of laguerre polynomials in results from overlap integrals of hermite polynomials @xmath123 , @xmath124 the hermite polynomials , in turn , appear in the electron wavefunctions in the presence of a magnetic field , see ref . @xcite .
the simple bubble - diagrams treated in allow , apart from numerical evaluation of with the kernel @xmath74 of , a further analytical treatment : one of the frequency integrations in can be performed noting that , by construction , @xmath74 is a purely imaginary function of @xmath90 and @xmath91 , and fulfills @xmath125 , such that we can rewrite as @xmath126 where @xmath127 the @xmath128 integration in can be solved by closing the contour in the complex @xmath91-plane , the integrals that appear are of the form @xmath129 [ ( \omega'\pm i/(2\tau))^2-e_n^2 ] } \nonumber \\ & = \frac{\omega\pm\omega+i/(2\tau)}{e_n^2-[\omega\pm\omega+i/(2\tau)]^2 } .
\nonumber\end{aligned}\ ] ] with @xmath130 given by an analytical expression , @xmath37 of can then be evaluated by numerically performing the remaining frequency integration and truncating the sum over landau levels at a suitable cut - off value . in the classical regime ,
the spacing between single landau levels at the fermi energy is much smaller than the fermi energy itself , and @xmath131 . in this limit
, one can proceed to approximate the integrand in and perform the remaining frequency integration .
this yields the classical drude formula of , with @xmath110 , plus an additional interband term which is non - resonant for @xmath132 . for a detailed presentation of the calculational steps ,
we refer the reader to ref . .
35ifxundefined [ 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 10.1038/ncomms2609 [ * * , ( ) ] link:\doibase 10.1038/nphys2927 [ * * , ( ) ] link:\doibase 10.1038/nphoton.2011.270 [ * * , ( ) ] link:\doibase 10.1140/epjb / e2012 - 30685 - 9 [ * * , ( ) ] link:\doibase 10.1103/revmodphys.81.109 [ * * , ( ) ] link:\doibase 10.1038/nature04233 [ * * , ( ) ] link:\doibase 10.1038/nature04235 [ * * , ( ) ] link:\doibase 10.1038/ncomms2866 [ * * , ( ) ] link:\doibase 10.1038/nphys1816 [ * * , ( ) ] @noop _
_ , ed .
( , ) \doibase http://dx.doi.org/10.1063/1.3543633 [ * * , ( ) ] link:\doibase 10.1021/nn403282x [ * * , ( ) ]
link:\doibase 10.1103/physrevb.77.195412 [ * * , ( ) ] link:\doibase 10.1103/physrevb.77.205421 [ * * , ( ) ] link:\doibase 10.1016/j.physrep.2010.07.003 [ * * , ( ) ] link:\doibase 10.1016/j.physrep.2015.12.006 [ * * , ( ) ] link:\doibase 10.1103/physrevb.87.121408 [ * * , ( ) ] link:\doibase 10.1021/nl1018063 [ * * , ( ) ] link:\doibase 10.1126/science.1191700 [ * * , ( ) ] link:\doibase 10.1038/nphys1420 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.115.245501 [ * * , ( ) ] link:\doibase 10.1038/nature05545 [ * * , ( ) ] link:\doibase 10.1103/physrevx.4.041019 [ * * , ( ) ] @noop _ _ , ed .
( , ) link:\doibase 10.1088/0953 - 8984/19/2/026222 [ * * , ( ) ] link:\doibase 10.1103/physrevb.73.245411 [ * * , ( ) ] link:\doibase 10.1103/physrevb.87.195432 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.103.136403 [ * * , ( ) ] link:\doibase 10.1016/j.physrep.2015.02.003 [ * * , ( ) ] , @noop _ _ , ed . , edited by and ( , , ) link:\doibase 10.1103/physrevb.65.245420 [ * * , ( ) ] | a beam of linearly polarized light transmitted through magnetically biased graphene can have its axis of polarization rotated by several degrees after passing the graphene sheet .
this large faraday effect is due to the action of the magnetic field on graphene s charge carriers . as deformations of the graphene membrane result in pseudomagnetic fields acting on the charge carriers ,
the effect of random mesoscopic corrugations ( ripples ) can be described as the exposure of graphene to a random pseudomagnetic field .
we aim to clarify the interplay of these typically sample inherent fields with the external magnetic bias field and the resulting effect on the faraday rotation . in principle
, random gauge disorder can be identified from a combination of faraday angle and optical spectroscopy measurements . |
cadmium zinc telluride ( czt ) has emerged as the detector material of choice for the detection of hard x - rays and soft @xmath1-rays when excellent position and energy resolution is needed and cryogenic cooling is impractical .
most commonly , czt detectors are employed to detect photons in the 10 kev to @xmath01 mev energy range where its high density ( @xmath2 5.76 g/@xmath3 ) and high average atomic number ( @xmath450 ) result in high stopping power and a large cross section for photoelectric interactions .
several astrophysical , nuclear physics , and homeland security applications require the detection of photons with energies of several mev with good position and energy resolution and with a detection efficiency close to 100% .
a detector built from closely packed czt detectors units may perform substantially better than a scintillation detector and may be more compact and more economic than a cryogenically cooled ge detector .
so far czt detectors have mainly been studied in the photoelectric regime ( 10 kev-@xmath0300 kev ) .
two different designs have been widely used , pixellated detectors and single sided strip detectors @xcite .
excellent energy resolutions have been achieved with both designs . for a pixellated detector with 11 @xmath5 11 pixels and a steering grid zhang et al .
@xcite reported 0.8% ( full width half maximum , fwhm ) at 662 kev for single pixel events and 2.3% at 662 kev ( fwhm ) for three pixel events .
the detector size was 1.5 @xmath5 1.5 @xmath5 1 @xmath3 . for a single sided charge - sharing czt strip detectors fwhm energy resolutions of 19.5 kev at 662 kev and 23.7 kev ar 1333 kev were reported @xcite . whereas detectors with n @xmath5
n pixels require n@xmath6 readout channels czt cross - strip detectors require 2 @xmath5 n channels . the penalty for this advantage are ambiguities in matching strips with the right cross - strip counterparts in case of multiple interactions in one detector . in order to achieve good detector efficiency , energy resolution and imaging properties in the `` compton '' regime of several hundred
kev requires to take multiple pixel or multiple strip events into account .
strategies exist to ameliorate this problem somewhat , for further details @xcite .
this paper deals with pixellated czt detectors only .
we use the geant 4.0 code @xcite to study the performance of very thick detectors ( @xmath7 10 cm ) built from closely packed czt detector units with the objective of evaluating the theoretically achievable performance of such czt calorimeters given the electronic properties of present - day czt substrates . as we will describe further below , the main thrust of our study is to evaluate general performance limitations arising from the combination of the location and spatial extent of the charge clouds below individual detector pixels , trapping of electrons and holes and the influence of the weighting potential .
we neglect the performance limitations arising from te - precipitates and other crystal defects .
the good agreement of the simulated detector response and the one experimentally measured shows that electron and hole trapping and the detectors weighting potential indeed limits the performance of current czt detectors . te -
precipitates and other crystal defects seem to be only important for underperforming detector areas .
the calorimeter performance depends on the shape of the 3-d weighting potential and on the number , energy and distribution of secondary particles produced in the calorimeter .
simulations with a particle interaction code like geant 4 and a device simulation code is the most efficient and most accurate way to estimate the performance of such calorimeters .
this paper is structured as follows . in section [ astronuclappl ] ,
we give a brief description of astrophysics and nuclear physics experiments that may use czt calorimeters .
subsequently , in section [ cztsimulation ] we describe the simulations and present their results for single detector units and for full czt calorimeters . in sect .
[ con ] the results are discussed . in the following all energy resolutions
are given as full - width - half - maximum energy resolutions .
from several hundred kev to @xmath010 mev the dominant interaction process of photons in matter is compton scattering . at lower energies the photo effect dominates and at higher energies pair production .
the sensitivity of past , present and upcoming hard x - ray- and @xmath8telescopes shows a gap in the `` compton '' energy range from a few 100 kev up to several tens of mev .
this energy range is highly interesting from the astrophysical point of view .
some of the scientific objectives are : the study of nucleosynthesis in core - collapse , type ia supernovae and classical novae , the search for intermediate blazars with peak emission of about 1 mev and @xmath8ray lines from nuclear interactions of cosmic rays with the interstellar medium @xcite , and the study of the 511 kev emission from sources like the galactic center @xcite .
+ -ray telescope operating in the energy range from a few hundred kev to several ten mev .
measuring the position and energy deposition in d1 and in d2 , the incident direction of the primary @xmath8ray can be constrained to lie on a circle .
additional measurements of the direction of the recoil electron would make it possible to constrain the incident direction to an arc .
, width=240 ] figure [ designeconcept ] shows the basic detector principle used in both the tigre ( tracking and imaging @xmath8ray instrument ) @xcite and the mega ( medium energy @xmath8ray astronomy ) designs @xcite .
the design uses two detectors : first , a tracker ( d1 ) where compton scattering and pair conversion takes place and where the trajectories , energies , and momenta of secondary electrons are measured ; second , a calorimeter ( d2 ) to determine the location and energy of scattered photons ( see figure [ designeconcept ] ) . the compton equation can be used to calculate the energy and incident direction of the primary photon from the combined information from the d1 and d2 detectors . depending on the knowledge of the momentum and energy of the scattered electron , the direction of the incident photon
is given either as an arc or a circle .
usually a low - z material is used for the d1 detector to maximize the probability of a single compton scattering . for d2
a high - z material is the preferred choice to maximize the probability for photon absorption in photo effect events . in the specific case of mega ,
the tracker consists of 32 layers of double - sided silicon strip detectors ( 6 @xmath5 6 cm@xmath6 each , 500 @xmath10 m thick , with a pitch of 470 @xmath10 m ) .
the calorimeter consists of csi slabs , 5 @xmath5 5 cm@xmath6 cross section , 4 - 8 cm deep .
each slab is read out by a si pin diode . over the energy range from 511 kev to 1.274 mev
, the csi calorimeter achieves an energy resolution of about @xmath08% fwhm and a spatial resolution of 2 cm @xcite .
+ + ge calorimeters could in principle be used to improve the performance of compton telescopes like tigre and mega , however , the costs of constructing such a calorimeter and the cryogenic cooling system , and lifting it into an orbit are very high .
czt with excellent spatial resolution and good energy resolution might be a better choice for d2 to improve on the performance of compton telescopes .
another application for czt calorimeters lies in the detection of mev photons in nuclear physics experiments that use `` fast beams '' ( @xmath11 0.2 ) .
presently , ge detectors are used to measure the energy and location of @xmath8rays emitted by excited nuclei .
however , ge detectors are expensive and require cryogenic cooling that makes them bulky , and complicates the use with auxiliary detectors or spectrometers . in the case of `` fast beams '' , the effective energy resolution of the experiment ( the accuracy of the energy reconstruction in the rest frame of the emitting particle ) is mainly limited by the spatial resolution with which the @xmath8ray can be localized , owing to the doppler effect . with angle @xmath12 between emitter and photon momenta for beta = 0.25 ( dotted line ) , 0.50 ( dashed line ) and 0.75 ( solid line ) .
if the spatial resolution of the detector allows one to measure the arrival direction of the photon with an accuracy of @xmath13 , the associated uncertainty in the energy of the photon in the momentum of frame is @xmath14 .
, width=288 ] figure [ doppler ] shows the differential energy change @xmath15 as a function of the angle between emitter and photon momenta for @xmath16 = 0.25 , 0.50 and 0.75 , dotted , dashed and solid line , respectively . for an angle of 50@xmath17
the differential energy change is @xmath0 0.5%/degree , @xmath0 1%/degree and @xmath0 1.4%/degree for @xmath16 = 0.25 , 0.50 and 0.75 , respectively .
thus , in fast beam experiments the effective energy resolution of the experiment is limited by the spatial resolution of the detectors .
+ the ge detector greta ( gamma - ray energy tracking array ) is the next generation @xmath8ray detector for nuclear structure research .
greta will use a two - dimensionally segmented ge detector along with pulse - shape analysis to determine the three - dimensional position and the energy of the interaction .
the prototype gretina of greta @xcite consists of a cluster of three tapered hp - ge n - type crystals with a hexagonal shape .
the crystal length is 9 cm , the diameter at the back is 8 cm .
the outer electrode is divided into 36 parts , 6 longitudinal and 6 transverse segments .
gretina achieves laboratory frame energy resolutions of 1.15 kev at 60 kev and 2.35 kev at 1.3332 mev .
the position resolution is 2 mm rms in three dimensions://@xmath18@xmath0@xmath19 . for a czt detector ,
the theoretical limit of the position resolution is the distance electrons produced in photo effects or compton events travel before transferring their energy to the valence band electrons . at 1 mev
80% of the electrons stay within 0.18 mm from the point where they are generated . for real pixellated detectors
the position resolution is limited further by pixel size , timing and possibly cathode signal resolution .
al @xcite reported position resolution of 1.27 @xmath5 1.27 @xmath5 0.2 mm for a 1.5 @xmath5 1.5 @xmath5 1.0 @xmath3 czt detector with 11 @xmath5 11 pixels . despite the clearly better energy resolution of a ge calorimeter low cost small volume czt calorimeters
can therefore achieve comparable overall effective energy resolution in fast beam experiments .
in the recent years the performance of pixellated czt detectors has been dramatically improved .
energy resolution at 662 kev of 0.8% for single pixel readout and 2.3% for combined signals of three anode pixels were reported for a czt detector of dimension 1.5 @xmath5 1.5 @xmath5 1.0 @xmath3 with 11 @xmath5 11 pixels @xcite . for the detection of higher energies
it is necessary to combine the signals of several pixels because the dominant interaction process , the compton effect , tends to spread the interaction over adjoining pixels . even at lower energies integrating over all pixel signals has a large impact on detector efficiency , because the incidents of compton events is not negligible . to gain a better understanding of the behavior of pixellated czt calorimeters ,
detailed three dimensional simulations of a large volume calorimeter have been performed .
our main objective is to derive theoretical limits for currently used czt detectors and electronics .
therefore we have chosen to use a czt calorimeter constructed out of 1.48 @xmath5 1.48 cm@xmath6 single pixellated czt detector units with 9 @xmath59 pixels and a pitch of 0.164 cm .
reported pixel pitches vary between 0.11 cm and 0.25 cm .
the pixels are held at ground and the cathode bias is -1500 v per 0.5 cm .
the thickness of the detector units was varied : 0.5 cm , 1.0 cm and 1.5 cm .
all these sizes are commercially available .
the overall size of the calorimeter considered in the following was chosen to be 10.33 @xmath5 10.33 @xmath5 45 @xmath3 .
this choice ensures that almost all interactions take place inside the detector volume .
the results are not limited to this particular design but apply to to all calorimeters as long as they are large enough to fully stop incoming @xmath8rays .
the output of our simulations are electron drift times and signals on all pixels . by `` signal ''
we understand the integrated charge over 1 @xmath10s . if not stated otherwise , we assume electronic noise of 4.5 kev and a pixel trigger threshold of 13.5 kev .
these values are typical for state - of - the - art readout systems @xcite .
we did not simulate passive material ( printed circuit boards , asics ) between the czt detector units .
the simulations can be separated in three parts . first the three dimensional electric field and the weighting potential are calculated with a finit - difference method .
the electric field is calculated at the nodes of a 3-dimensional grid with the grid size @xmath20s .
if the coordinates of the nodes are given by ( i@xmath21s , j@xmath22s , k @xmath21s ) , the laplace equation for each node with unknown potential can be approximated by the central point finite difference scheme : @xmath23 after specifying the `` boundary conditions '' , by setting the potential of the contact nodes to the applied bias potential , one gets a system of linear equations of the form @xmath24 where the index l runs over the remaining n nodes where the potential has yet to be determined .
we use the steepest descent method , to solve this system of linear equations numerically .
this iterative method minimizes the difference @xmath25 until it becomes smaller than a certain preset value . to determine the weighting potential of pixel i ,
the potential of the contact node of i is set to 1 , for the rest to 0 .
+ second , for the interaction of incident @xmath8rays and secondary particles with the detector material the low - energy extension of the geant 4.0 code @xcite is used . for each interaction
the electrons and holes are tracked in 1 ns steps through the detector .
for each step , the induced charge at the anodes is calculated .
the rise time of each pixel and the charge accumulated in 1 @xmath10s are the resulting signal values .
the electron mobility is set to @xmath26 cm@xmath6/v / s with a lifetime of @xmath27s and the hole mobility to @xmath28cm@xmath6/v / s with a lifetime of @xmath29s @xcite .
the simulations neglect any effect of material non uniformity . for
the understanding of a complex system it is often useful to study first a small part instead of the full system .
thus we begin with the discussion of a single czt detector unit .
figure [ wp ] left panel shows the potential distribution of the central pixel for a 1 cm thick detector .
the weighting potential distribution for three different lateral positions as a function of z ( equal depth of interaction ( doi ) ) is given in figure [ wp ] right panel .
one can see , that most of the signals are induced near the anode contacts ( so called small pixel effect ) @xcite , and the anode signal depends only weakly on the depth of interaction ( doi ) , if the interaction takes place less than 0.6 cm away from the cathode .
the residual doi dependence is caused by small changes of the weighting potential far away from the pixels and by electrons getting trapped as they drift to the anode .
near the anode the weighting potential rises steeply .
this causes a strong dependence of the induced charge in this region .
interactions closer to the anode induce less charge than interactions closer to the cathode .
it can be also seen , that the dependence in x - y direction increases near the anodes too .
the cathode behaves differently since it does not exhibit the small pixel effect .
czt is a single charge carrier , therefore it is necessary to correct for the doi .
for single - pixel events , one can derive the depth of interaction by using the cathode - to - anode signal ratio .
this approach does not work for multi - pixel events . in this case , the signal rise time ( proportional to the electron drift time ) can be used @xcite .
this is the approach we used in our simulations .
figure [ zdep ] , shows the anode signal as the percentage of the energy of the incident gamma as a function of doi .
the behavior can easily be understood based on the charge transport properties of czt and the weighting potential ( figure [ wp ] ) .
the gradient of the weighting potential is very steep close to the anodes and flattens with increasing z. for interactions close to the anodes , electrons drift only small distances before they impinge on the anodes .
holes contribute up to 70% of the total signal by moving through the steep gradient of the weighting potential . for interactions closer to the cathode side ,
electrons induce charge as they move towards the pixels .
holes induce relatively little charge , as they are trapped within @xmath30s ( 0.12 cm ) of being generated , and move towards the cathode through a region of rather constant weighting potential .
the relative signal contribution of electrons and holes depends on the charge integration window .
if the integration time is shorter the holes travel less distance and thus their contribution is smaller .
the weighting potential is not confined to the volume of the primary affected pixel .
( see figure [ wp ] ) therefore neighboring pixels measure signals , when an interaction takes place .
the amount of induced charge depends on the point of interaction .
the closer it occurs to the pixel edges and to the anodes the larger is the signal in the neighboring pixel . to explore this effect in more detail , figure [ inducedsignal ] , right panel , presents the signals induced in four adjacent pixels as a function of the point of interaction which is indicated as a line on the left panel .
one can nicely see how the signal is spread over adjacent pixels when charge is deposited close to the pixel boundaries .
current experiments with pixellated czt detector report energy resolution of 0.8% for single pixel @xcite and 2.3% for the combined signals of three pixels at 662 kev .
the detector size was 1.5 @xmath5 1.5 @xmath5 1 @xmath3 with 11 @xmath5 11 pixels and a steering grid .
the electronic noise was in the range of 4.5 kev .
for comparison we have simulated the energy resolution of a single detector unit , 1 cm thick , both for single - pixel events and for combining all anode signals .
doi correction is applied .
the spectra obtained are given for 662 kev in figure [ 1cmdetectorspectra ] .
ray is described in the text .
all @xmath8rays were assumed to hit the detector s central pixel . , width=576 ] the energy resolution of single - pixel events is 0.83% while for all events including multiple - pixel events , it is 1.25% . for single - pixel events the peak to compton ratio is about @xmath020 and for the full detector @xmath025 .
these results are in good agreement with the data reported by zhang et al .
this comparison shows us , that the detector performance is determined by czt general properties and detector design and not by the incidental quality of the czt material of a specific detector . + currently commercially available detectors are 0.5 cm , 1.0 cm or 1.5 cm thick .
the results of our simulations for 662 kev incident energy are given in table [ tab : energy ] .
.energy resolution for 662 kev incident energy for single - pixel and multi - pixel events for three detector thicknesses of a single czt detector unit .
[ cols= " < , < , < " , ] for 0.5 cm and 1.5 cm thick detectors we obtained single - pixel energy resolutions of 1.12% and 0.80% , respectively and multiple pixel energy resolutions of 1.34% and 1.25% .
while 1.0 cm and 1.5 cm thick czt detector units show almost identical behavior , the energy resolution of a 0.5 cm thick detector drops to 1.12% . in the following
we will discuss how the energy resolution depends on a number of different effects .
+ we begin with an effect which results from the application of a trigger threshold . when all doi - corrected signals of all triggered pixels are summed up to calculate the total energy deposition , signals below the trigger threshold
will be discarded , diminishing the summed signal .
this threshold effect becomes more prominent the closer the energy is deposited to the anode .
therefore , the energy resolution depends on the doi and thus on the z - coordinate .
furthermore , there exists a strong xy - dependence too , caused by the steep x- and y - components of the weighting potential gradient near the anode pixels . to study these two effects
, we simulated charge deposition at specified z positions and random ( x , y)-positions under one pixel .
after applying the doi correction we added all signals which exceed the trigger threshold ( three times electronic noise ) .
the results at 662 kev are presented in figure [ aufloesungvsz ] for the 1.5 cm thick detector .
on the left side , the energy resolution of a central pixel is shown as a function of the depth of interaction . of the peak of the energy spectra.,width=576 ] on the right side , the percentage of events within the photo peak ( @xmath31 ) is shown .
the energy resolution stays at about 0.8% at distances between 0.2 to 1.5 cm from the pixel side .
closer to the anode , the resolution deteriorates to @xmath01.8% .
the fraction of events within the photo peak is @xmath0100% for large z values whereas it drops to about 95% for events close to the pixel anode .
energy spectra of events close to the anodes have a low energy tail .
depending on the exact location below one pixel , a variable amount of charge is induced on adjacent pixels .
the charge is lost , if the adjacent pixel signal does not surpass the trigger threshold . the results for 0.5 cm and 1.0 cm thick detectors are virtually the same as the corresponding region of the 1.5 cm thick detector , if one uses charge integration windows of the same duration .
from figure [ aufloesungvsz ] we see that the 0.25 cm region just below the anodes gives very poor energy resolution .
this for 0.5 cm thick detectors , 50% of the detector volume performs very poorly . for 1 cm and 1.5 cm thick detectors , the fraction of poorly performing volume decreases to 25% and 16% , respectively .
the energy resolution is also limited by the fact that an incident @xmath8ray can generate multiple energy depositions below one single pixel . as multiple charge depositions under one pixel
can not be separated , the depth of interaction correction corresponds to the interaction with the longest drift time ( closest to the cathode ) .
however this correction is applied to the full signal . at 662 kev the median of the rms values of the interaction region below one pixel with a thickness of 1.5
cm is 0.004 cm in z - direction .
we calculated the relative difference between the signals induced by a interactions at point z and at point z+0.004 with x- and y - coordinates set to 0 .
the result is shown in figure [ interactionrange ] as a function of z , which gives the error in the doi correction if one corrects for a doi of z instead of z+0.004 cm .
close to the pixel anode the charge spreading effect becomes more important because of the strong gradient of the weighting potential .
the effect contributes considerably to the energy resolution below @xmath32 cm . at higher energies the spreading effect is more pronounced and its influence on the energy resolution increases .
to summerize the results above , the deterioration of the energy resolution for thinner czt detector units is due to a ) the steep gradient of the weighting potential close to the anodes , b ) the induced signals in neighboring pixels below the trigger threshold and c ) multiple interactions below one pixel . in this section
we study the performance of a full calorimeter built out of several single detector units .
the overall calorimeter size is @xmath33
this calorimeter is made of several layers of czt detectors .
each layer contains of an array of @xmath34 detector units , each with @xmath35 pixels .
as before , simulations were made varying the detector thickness ( 0.5 , 1.0 and 1.5 cm ) .
the overall size of the calorimeter was kept constant , therefore the number of layers was changed depending on thickness ( 90 , 45 and 30 for 0.5 cm , 1.0 cm and 1.5 cm ) . as before ,
we determine energy spectra by applying a doi correction to each individual pixel signal satisfying the trigger criteria .
afterwards all signals are summed up to obtain the reconstructed energy . from the peak of the energy spectrum .
the right side shows the energy resolutions of full calorimeters as function of incident energy for all three detectors designs.,width=528 ] the energy resolution is limited by several effects .
those which are due to the properties of the single detector units have been discussed in the previous section in some detail .
therefore we will limit the discussion here to effects which are peculiar to the full calorimeter .
the energy resolution of the full calorimeter as a function of incident energy are given on the right side of figure [ effizienzfwhm ] for detector designs with base units of 0.5 cm , 1.0 cm and 1.5 cm thickness .
on the left side , the detector efficiency is given , which we define as the fraction of events being reconstructed within @xmath36 of the peak of the energy spectra .
below 1 mev a calorimeter made out of 0.5 cm thick detector units shows resolutions between 2% and 5% . above 1 mev
, the value steadily increases to @xmath07.5% at 10 mev .
the efficiency lies between 76% and 60% .
calorimeters made of 1 cm and 1.5 cm thick detector units perform markedly better .
the energy resolution of both calorimeters improve from @xmath03% at 200 kev to @xmath01% at 3 mev . at higher energies ,
the resolutions increase somewhat up to @xmath01.9% at 10 mev .
the efficiency drops from @xmath094% at @xmath0200 kev to 43% at 10 mev . to understand the influence of unit thickness , it is important to understand that secondary particles do leave the detector and therefore the energy of the incident @xmath8ray is not fully deposited in the detector region .
there exist two different classes of secondary particles .
first those that are detectable by an active shield surrounding the detector . usually a shield covers sides and bottom , but not the surface through which photons enter the detector because such a shield would interact with incoming photons .
the direction in which the incoming photons stream we will call the `` forward direction '' .
secondary particles leaving the detector in the direction of the shield can excluded from the analyses , when the shield is able to detect them or they can be included in the analyses if the shield gives sufficient energy information .
thus in principle these events do not reduce energy resolution , but , they reduce the efficiency .
the second class are secondary particles which leave against the stream of incoming photons .
we call this the `` upward direction '' . since there is no way to detect these photons these events
are lost for analyses .
the drop in the efficiency at higher incident energy is mainly due to the fact that the energy of the incident @xmath8ray is not fully deposited in the detector region . therefore the reconstructed energy is lower than the incident energy .
figure [ percoutside ] shows the percentage of events leaving the detector in the upward direction .
these are mainly caused by pair production events .
the right side shows the mean energy lost in this way . at 662 kev @xmath05% of the events loose on average 217 kev . at higher energies the importance of this effect increases . at 10 mev @xmath023% of events loose on average 700 kev .
as already mentioned , there is no remedy for this and it is the main cause of the efficiency drop in calorimeters at high energies . for 0.5 cm thick detectors the energy resolution degrades at higher energies .
the increase of multiple interactions together with the steep weighting potential gradient close to the anodes which affects about 50% of the volume of 0.5 cm thick detectors reduces the energy resolution significantly . in figure
[ spectra5mevfullcalorimeter ] , an energy spectrum is shown for a full calorimeter with single detector units of 1.5 cm thickness for 5 mev incident energy .
the shaded area marks events where secondary particles leave the detector in upward direction .
with increasing energy the number of interactions increases and the readout noise increases proportional to @xmath37 where @xmath38 is the number of triggered pixels . in figure [ numpixel ] the mean number of triggered pixels per incident gamma is shown as a function of incident energy for a 1.5 cm thick detector . from @xmath0200 kev to 10 mev the mean number of pixels increases from around 1 to about 12 . for 0.5 cm and 1.0 cm thick detectors
, these values change by less than 2.5% .
finally , figure [ noise ] shows the energy resolution of a full calorimeter ( 1.5 cm thick detector units ) for single channel noise levels of 3.5 kev , 4.5 kev and 6 kev . above 2
mev the contribution of single - channel noise is not anymore significant while at lower energies the the overall performance can be improved by reducing single - channel noise . summarizing it
can be said that the efficiency of a full calorimeter is mainly limited by secondary particle leaving the detector in the upward direction while energy resolution is limited by the steep weighting potential near the anode pixels and the signal loss caused by the trigger threshold .
in our simulations we have considered weighting potential , electron and hole trapping .
the results of a single detector unit are in good agreement with experimental data .
therefore , precipitates and crystal inhomogeneities play a major role only for underperforming detector regions . the studies in this paper show , that czt has to be seriously considered as a detector material for the design of future calorimeters which operate in the range from several kev up to several mev .
we have simulated three different calorimeter designs varying the thickness of the individual detector and the number of layers , keeping constant the size of the full calorimeter constant ( 10.33 @xmath5 10.33 @xmath5 45 @xmath3 ) .
calorimeters built out of 1.0 cm and 1.5 cm thick units exhibit almost identical performance with energy resolutions of about 1.7% at 662 kev and about 1.9% at 10 mev .
the energy resolution is mainly limited by the energy loss owing to secondary particles leaving the detector upward against the direction from which the incident radiation enters the detector and by the effects of the steep change of the weighting potential close to the anode .
going to higher energies the resolution is also negatively influenced by the increasing number of events which have several interactions below one pixel because in these cases doi corrections are not well defined .
the principal result of our study is the fact that a 0.25 cm thick region below the anodes of the detector units exhibits a substantially poorer energy resolution than the rest of the detector volume . reducing the portion of this low resolution volume increases the overall performance .
a good calorimeter requires that the single detector unit is much thicker than the pixel pitch and of course much thicker than passive material .
( mounting , cables and front - end electronics ) .
the energy resolution of a czt calorimeter is shown in figure [ effizienzfwhm ] .
the resolution is limited a ) by the steep weighting potential gradient close to the anodes , which dominates in thin detectors and impairs the correction of the depth of interaction ( doi ) , b ) by the fact , that the incident energy is not totally deposited within the detector area because secondary particles leave the detector upward against the direction from which the incident radiation enters the detector c ) because signals induced in neighboring pixels when the interaction is near the pixel edges and near the anodes with intensity below the trigger threshold are lost for signal reconstruction leading to systematically reduced signals .
we found that the energy resolution of a czt calorimeter will lie between that of csi - detectors and ge detectors . in terms of spatial resolution
solid state ge and czt detectors are of course much better than scintillation detectors . where compactness and weight are issues as in a next generation space - born @xmath1-ray telescope , cadmium zinc telluride ( czt ) detectors might be the detector option of choice .
00 f. zhang , z. he , g. knoll , d. wehe , j. berry , _ 3d position sensitive cdznte spectrometer performance using third generation vas / tat readout electronics _ , ieee transactions on nuclear science , vol .
5 , october 2005 j. matteson , _ czt detectors with 3-d readout for gamma - ray spectroscopy and imaging _ , proc .
spie vol . 4784 ( 2002 )
j. r. macri , b. donmez , m. widholm , l .- a .
hamel , m. julien , t. narita , j. m. ryan and m. l. mcconnell , _ single - sided czt strip detectors _ proc .
spie , 5501 , 208 ( 2004 ) . b. doenmez , j. r. macri , m. l. mcconnell , j. m. ryan , m. widholm , t. narita , and l .- a .
hamel , proc .
spie , 5922 , 62 ( 2005 ) . s. agostinelli et al . , _
geant4a simulation toolkit _ , nim a506 , 250 - 303,2003 g. kanbach et al .
, _ the mega project _ , new astronomie reviews 48 , 275 - 280 , 2004 .
j. knoedleseder , p. jean , v. lonjou et al
. 2005 , a&a , in press t.j .
oneill et al .
, _ development of the tigre compton telescope for intermediate - energy gamma - ray astronomy _ , ieee trans .
50 2 , 251 , 1999 .
r. andritschke et al . , _ the calibration setup of the mega prototype at the high intensity c - ray source _ , new astronomie reviews 48 , 281 - 285 , 2004 .
f. schopper et al . , _ csi calorimeter with 3-d position resolution _ , nim a442 , 394 - 399 , 2000 .
k. vetter , _ performance of the greta prototype detectors _ , nim a452 , 105 - 114 , 2000 .
k. vetter , _ three - dimensional position sensitivity in two - dimensionally segmented hp - ge detectors _ , nim a452 , 223 - 238 , 2000 . w. shockley , _ currents to conductors induced by a moving point charge _ , j. appl .
( 1938 ) 635 i.r.e .
, september 1939 , 584 s. ramo , _ current induced by electron motion _ , proceedings of the i.r.e . ,
september 1939 , 584 | several astrophysics and nuclear physics applications require the detection of photons in the energy range of kev up to several mev with good position and energy resolution . for certain applications cadmium zinc telluride ( czt )
detectors might be the detector option of choice .
up to now , czt detectors have mainly been used in the energy range between a few kev and @xmath01 mev .
they operate at room temperature and achieve excellent position resolution and substantially better energy resolution than scintillation detectors . furthermore
, they can be built more compact and more economically than ge detectors and do not require cryogenic cooling . in this paper
, we describe the results of 3-d monte carlo simulations of a `` czt calorimeter '' that can be used to detect photons in the kev to several mev range .
the main objective of these studies is to evaluate the feasibility of czt calorimeters , to study their performance and detect and understand performance limiting factors .
such a calorimeter consists of many layers of closely packed pixellated czt detector units .
our simulations of single detector units reproduce experimental results , indicating that our simulations capture the main factors that limit the performance of a detector unit . for a full calorimeter
the limiting factors within a range from @xmath020 kev to @xmath010 mev are : a ) the fact , that the incident energy is not totally deposited within the detector area because secondary particles leave the detector against the direction from which the incident radiation enters , b ) signal loss when the interaction is near the pixel edges and near the anodes . in this case
signals which are induced in neighboring pixels are discarded when their intensities lie below the trigger threshold .
c ) the steep weighting potential gradient close to the anodes , which affects about 0.25 cm next to the anode and impairs there the correction of the depth of interaction ( doi ) .
this effect dominates in thin detectors ( 0.5 cm ) .
+ understanding the limiting factors we come to the conclusion that 1 cm to 1.5 cm thick detector units can be used to build a calorimeter with good performance over the energy range from @xmath020 kev to @xmath010 mev .
cdznte , czt detectors , astronomy , nuclear physics , mev @xmath1-ray astronomy 95.55.ka |
this paper deals with the following problem .
local search.__given an undirected graph _ _ @xmath4 _ _ and a function _
_ @xmath5 _ _ , find a local minimum of _ _ @xmath6__that is , a vertex _ _
@xmath7 _ _ such that _ _ @xmath8 _ _ for all neighbors _ _ @xmath9 _ of _ @xmath7__. _ _ we are interested in the number of _ queries _ that an algorithm needs to solve this problem , where a query just returns @xmath10 given @xmath7 .
we consider deterministic , randomized , and quantum algorithms .
section [ motivation ] motivates the problem theoretically and practically ; this section explains our results .
we start with some simple observations .
if @xmath11 is the complete graph of size @xmath12 , then clearly @xmath13 queries are needed to find a local minimum ( or @xmath14 with a quantum computer @xcite ) . at the other extreme ,
if @xmath11 is a line of length @xmath12 , then even a deterministic algorithm can find a local minimum in @xmath15 queries , using binary search : query the middle two vertices , @xmath7 and @xmath9 . if @xmath16 , then search the line of length @xmath17 connected to @xmath7 ; otherwise search the line connected to @xmath9 . continue recursively in this manner until a local minimum is found .
so the interesting case is when @xmath11 is a graph of ` intermediate ' connectedness : for example , the boolean hypercube @xmath18 , with two vertices adjacent if and only if they have hamming distance @xmath19 . for this graph , llewellyn , tovey , and trick @xcite showed a @xmath20 lower bound on the number of queries needed by any deterministic algorithm , using a simple adversary argument . intuitively , until the set of vertices queried so far comprises a _ vertex cut _ ( that is , splits the graph into two or more connected components ) ,
an adversary is free to return a descending sequence of @xmath21-values : @xmath22 for the first vertex @xmath23 queried by the algorithm , @xmath24 for the second vertex queried , and so on .
moreover , once the set of queried vertices does comprise a cut , the adversary can choose the largest connected component of unqueried vertices , and restrict the problem recursively to that component .
so to lower - bound the deterministic query complexity , it suffices to lower - bound the size of any cut that splits the graph into two reasonably large components . for the boolean hypercube , llewellyn et al
showed that the best one can do is essentially to query all @xmath20 vertices of hamming weight @xmath25 .
llewellyn et al.s argument fails completely in the case of randomized algorithms . by yao s minimax principle ,
what we want here is a fixed _ distribution _
@xmath26 over functions @xmath27 , such that any deterministic algorithm needs many queries to find a local minimum of @xmath21 , with high probability if @xmath21 is drawn from @xmath26 .
taking @xmath26 to be uniform will not do , since a local minimum of a uniform random function is easily found . however , aldous @xcite had the idea of defining @xmath26 via a _ random walk _
, as follows .
choose a vertex @xmath28 uniformly at random ; then perform an unbiased walk to @xmath29 . ]
@xmath30 starting from @xmath31 . for each vertex @xmath7 ,
set @xmath10 equal to the first hitting time of the walk at @xmath7that is , @xmath32 .
clearly any @xmath21 produced in this way has a unique local minimum at @xmath31 , since for all @xmath33 , if vertex @xmath34 is visited for the first time at step @xmath35 then @xmath36 . using sophisticated random walk analysis , aldous managed to show a lower bound of @xmath37 on the expected number of queries needed by any randomized algorithm to find @xmath31 .
for any @xmath38 . ]
( as we will see in section [ prelim ] , this lower bound is close to tight . ) intuitively , since a random walk on the hypercube mixes in @xmath39 steps , an algorithm that has not queried a @xmath7 with @xmath40 has almost no useful information about where the unique minimum @xmath31 is , so its next query will just be a `` stab in the dark . ''
aldous s result leaves several questions about local search unanswered .
what if the graph @xmath11 is a @xmath41-d cube , on which a random walk does _ not _ mix very rapidly ? can we still lower - bound the randomized query complexity of finding a local minimum ? more generally , what parameters of @xmath11 make the problem hard or easy ?
also , what is the quantum query complexity of local search ?
this paper presents a new approach to local search , which we believe points the way to a complete understanding of its complexity .
our approach is based on the _ quantum adversary method _ , introduced by ambainis @xcite to prove lower bounds on quantum query complexity .
surprisingly , our approach yields new and simpler lower bounds for the problem s _ classical _ randomized query complexity , in addition to quantum lower bounds .
thus , along with recent work by kerenidis and de wolf @xcite and by aharonov and regev @xcite , this paper illustrates how quantum ideas can help to resolve classical open problems .
our results are as follows . for the boolean hypercube @xmath42
, we show that any quantum algorithm needs @xmath43 queries to find a local minimum on @xmath11 , and any randomized algorithm needs @xmath2 queries ( improving the @xmath44 lower bound of aldous @xcite ) .
our proofs are elementary and do not require random walk analysis . by comparison ,
the best known upper bounds are @xmath45 for a quantum algorithm and @xmath46 for a randomized algorithm .
if @xmath11 is a @xmath47-dimensional grid of size @xmath48 , where @xmath3 is a constant , then we show that any quantum algorithm needs @xmath49 queries to find a local minimum on @xmath11 , and any randomized algorithm needs @xmath50 queries .
no nontrivial lower bounds ( randomized or quantum ) were previously known in this case . in an earlier version of this paper
, we raised as our `` most ambitious '' conjecture that the deterministic and quantum query complexities of local search are polynomially related for _ every _ family of graphs . at the time , it was not even known whether deterministic and _ randomized _ query complexities were polynomially related , not even for simple examples such as the @xmath51-dimensional square grid .
recently santha and szegedy @xcite spectacularly resolved our conjecture , by showing that the quantum query complexity is at least the @xmath52 root ( ! ) of the deterministic complexity .
given that their result generalizes ours to such an extent , we feel obligated to defend why this paper is still relevant
. first , for specific graphs such as the hypercube , our lower bounds are close to tight ; those of santha and szegedy are not .
second , we give randomized lower bounds that are quadratically better than our quantum lower bounds ; santha and szegedy give only quantum lower bounds .
in another recent development , ambainis ( personal communication ) has improved our @xmath1 quantum lower bound for local search on the hypercube to @xmath53 , using a hybrid argument .
note that ambainis lower bound matches the upper bound up to a polynomial factor .
the paper is organized as follows .
section [ motivation ] motivates lower bounds on local search , pointing out connections to simulated annealing , quantum adiabatic algorithms , and the complexity class @xmath54 of total function problems .
section [ prelim ] defines notation and reviews basic facts about local search , including upper bounds . in section [ adversary ] we give an intuitive explanation of ambainis s quantum adversary method , then state and prove a classical analogue of ambainis s main lower bound theorem .
section [ snake ] introduces _ snakes _ , a construction by which we apply the two adversary methods to local search .
we show there that to prove lower bounds for any graph @xmath11 , it suffices to upper - bound a combinatorial parameter @xmath55 of a ` snake distribution ' on @xmath11 .
section [ graphs ] applies this framework to specific examples of graphs : the boolean hypercube in section [ boolean ] , and the @xmath47-dimensional grid in section [ ddim ] .
local search is the most effective weapon ever devised against hard optimization problems . for many real applications , neither backtrack search , nor approximation algorithms , nor even grover s algorithm ( assuming we had a quantum computer ) can compare .
furthermore , along with quantum computing , local search ( broadly defined ) is one of the most interesting links between computer science and nature .
it is related to evolutionary biology via genetic algorithms , and to the physics of materials via simulated annealing .
thus it is both practically and scientifically important to understand its performance .
the conventional wisdom is that , although local search performs well in practice , its central ( indeed defining ) flaw is a tendency to get stuck at local optima .
if this were correct , one corollary would be that the reason local search performs so well is that the problem it really solves finding a local optimum is intrinsically easy
. it would thus be unnecessary to seek further explanations for its performance .
another corollary would be that , for _ unimodal _ functions ( which have no local optima besides the global optimum ) , the global optimum would be easily found
. however , the conventional wisdom is false . the results of llewellyn et al .
@xcite and aldous @xcite show that even if @xmath21 is unimodal , any classical algorithm that treats @xmath21 as a black box needs exponential time to find the global minimum of @xmath21 in general .
our results extend this conclusion to quantum algorithms . in our view
, the practical upshot of these results is that they force us to confront the question : what is it about ` real - world ' problems that makes it easy to find a local optimum ? that is , why do exponentially long chains of descending values , such as those used for lower bounds , almost never occur in practice ( even in functions with large range sizes ) ?
we do not know a good answer to this .
our results are also relevant for physics .
many physical systems , including folding proteins and networks of springs and pulleys , can be understood as performing ` local search ' through an energy landscape to reach a locally - minimal energy configuration .
a key question is , how long will the system take to reach its ground state ( that is , a globally - minimal configuration ) ? of course , if there are local optima , the system might _ never _ reach its ground state , just as a rock in a mountain crevice does not roll to the bottom by going up first . but what if the energy landscape is unimodal ? and moreover , what if the physical system is quantum ?
our results show that , for certain energy landscapes , even a quantum system would take exponential time to reach its ground state , regardless of what hamiltonian is applied to it .
so in particular , the quantum adiabatic algorithm proposed by farhi et al .
@xcite , which can be seen as a quantum analogue of simulated annealing , needs exponential time to find a local minimum in the worst case .
finally , our results have implications for so - called _ total function problems _ in complexity theory .
megiddo and papadimitriou @xcite defined a complexity class @xmath54 , consisting ( informally ) of those @xmath56 search problems for which a solution always exists .
for example , we might be given a function @xmath57 as a boolean circuit , and asked to find any distinct @xmath58 pair such that @xmath59 .
this particular problem belongs to a subclass of @xmath54 called @xmath60 ( polynomial pigeonhole principle ) .
notice that no promise is involved : the combinatorial nature of the problem itself forces a solution to exist , even if we have no idea how to find it . in a recent talk ,
papadimitriou @xcite asked broadly whether such ` nonconstructive existence problems ' might be good candidates for efficient quantum algorithms . in the case of @xmath60 problems
, the collision lower bound of aaronson @xcite ( improved by shi @xcite and others ) implies a negative answer in the black - box setting . for other subclasses of @xmath54 , such as @xmath61 ( polynomial odd - degree node ) , a quantum black - box lower
bound follows easily from the optimality of grover s search algorithm .
however , there is one important subclass of @xmath54 for which no quantum lower bound was previously known .
this is @xmath62 ( polynomial local search ) , defined by johnson , papadimitriou , and yannakakis @xcite as a class of optimization problems whose cost function @xmath21 and neighborhood function @xmath63 ( that is , the set of neighbors of a given point ) are both computable in polynomial time . given such a problem ,
the task is to output any local minimum of the cost function : that is , a @xmath7 such that @xmath64 for all @xmath65 .
the lower bound of llewellyn et al .
@xcite yields an oracle @xmath66 relative to which @xmath67 , by a standard diagonalization argument along the lines of baker , gill , and solovay @xcite . likewise , the lower bound of aldous @xcite yields an oracle relative to which @xmath68 , where @xmath69 is simply the function version of @xmath70 .
our results yield the first oracle relative to which @xmath71 . in light of this oracle separation , we raise an admittedly vague question : is there a nontrivial `` combinatorial '' subclass of @xmath54 that we can show _ is _ contained in @xmath72 ?
in the local search problem , we are given an undirected graph @xmath73 with @xmath74 , and oracle access to a function @xmath5 .
the goal is to find any _ local minimum _ of @xmath21 , defined as a vertex @xmath75 such that @xmath16 for all neighbors @xmath9 of @xmath7
. clearly such a local minimum exists .
we want to find one using as few queries as possible , where a query returns @xmath10 given @xmath7 .
queries can be adaptive ; that is , can depend on the outcomes of previous queries .
we assume @xmath11 is known in advance , so that only @xmath21 needs to be queried .
since we care only about query complexity , not computation time , there is no difficulty in dealing with an infinite range for @xmath21though for our lower bounds , it will turn out that a range of size @xmath76 suffices .
our model of query algorithms is the standard one ; see @xcite for a survey .
given a graph @xmath11 , the deterministic query complexity of local search on @xmath11 , which we denote @xmath77 , is @xmath78 where the minimum ranges over all deterministic algorithms @xmath79 , the maximum ranges over all @xmath21 , and @xmath80 is the number of queries made to @xmath21 by @xmath79 before it halts and outputs a local minimum of @xmath21 ( or @xmath81 if @xmath79 fails to do so ) .
the randomized query complexity @xmath82 is defined similarly , except that now the algorithm has access to an infinite random string @xmath83 , and must only output a local minimum with probability at least @xmath84 over @xmath83 . for simplicity , we assume that the number of queries @xmath85 is the same for all @xmath83 ; clearly this assumption changes the complexity by at most a constant factor . in the quantum model
, an algorithm s state has the form @xmath86 , where @xmath7 is the label of a vertex in @xmath11 , and @xmath87 and @xmath88 are strings representing the answer register and workspace respectively .
the @xmath89 s are complex amplitudes satisfying @xmath90 . starting from an arbitrary ( fixed ) initial state , the algorithm proceeds by an alternating sequence of _ queries _ and _ algorithm steps_. a query maps each @xmath91 to @xmath92 , where @xmath93 denotes bitwise exclusive - or .
an algorithm step multiplies the vector of @xmath89 s by an arbitrary unitary matrix that does not depend on @xmath21 . letting @xmath94 denote the set of local minima of @xmath21 , the algorithm succeeds if at the end @xmath95 .
then the bounded - error quantum query complexity , or @xmath96 , is defined as the minimum number of queries used by a quantum algorithm that succeeds on every @xmath21 .
it is immediate that @xmath97 .
also , letting @xmath98 be the maximum degree of @xmath11 , we have the following trivial lower bound .
[ degree]@xmath99 and @xmath100 .
let @xmath7 be a vertex of @xmath11 with degree @xmath98 . choose a neighbor @xmath9 of @xmath7 uniformly at random , and
let @xmath101 .
let @xmath102 , and @xmath103 for all neighbors @xmath104 of @xmath7 other than @xmath9 .
let @xmath105 be the neighbor set of @xmath7 ( including @xmath7 itself ) ; then for all @xmath106 , let @xmath107 where @xmath108 is the minimum distance from @xmath109 to a vertex in @xmath105 .
clearly @xmath21 has a unique local minimum at @xmath9 .
however , finding @xmath110 requires exhaustive search among the @xmath98 neighbors of @xmath7 , which requires @xmath111 quantum queries @xcite .
a corollary of proposition [ degree ] is that classically , zero - error randomized query complexity is equivalent to bounded - error up to a constant factor . for given a candidate local minimum @xmath7
, one can check using @xmath112 queries that @xmath7 is indeed a local minimum . since @xmath113 queries are needed anyway
, this verification step does not affect the overall complexity .
as pointed out by aldous @xcite , a classical randomized algorithm can find a local minimum of @xmath21 with high probability in @xmath114 queries .
the algorithm just queries @xmath115 vertices uniformly at random , and lets @xmath31 be a queried vertex for which @xmath116 is minimal .
it then follows @xmath31 to a local minimum by steepest descent .
that is , for @xmath117 , it queries all neighbors of @xmath34 , halts if @xmath34 is a local minimum , and otherwise sets @xmath118 to be the neighbor @xmath9 of @xmath34 for which @xmath119 is minimal ( breaking ties by lexicographic ordering ) .
a similar idea yields an improved quantum upper bound .
[ upper]for any @xmath11 , @xmath120 .
the algorithm first chooses @xmath121 vertices of @xmath11 uniformly at random , then uses grover search to find a chosen vertex @xmath122 for which @xmath10 is minimal . by a result of drr and hyer @xcite ,
this can be done with high probability in @xmath123 queries . next , for @xmath117 , the algorithm performs grover search over all neighbors of @xmath124 , looking for a neighbor @xmath9 such that @xmath125 .
if it finds such a @xmath9 , then it sets @xmath126 and continues to the next iteration .
otherwise , it repeats the grover search @xmath127 times before finally giving up and returning @xmath124 as a claimed local minimum . the expected number of vertices @xmath104 such that @xmath128 is at most @xmath129 .
since @xmath130 for all @xmath35 , clearly the number of such @xmath104 provides an upper bound on @xmath35 .
furthermore , assuming there exists a @xmath9 such that @xmath131 , the expected number of repetitions of grover s algorithm until such a @xmath9 is found is @xmath132 . since each repetition takes @xmath133 queries , by linearity of expectation the total expected number of queries used by the algorithm is therefore@xmath134 or @xmath135 .
to see that the algorithm finds a local minimum with high probability , observe that for each @xmath35 , the probability of not finding a @xmath9 such that @xmath136 , given that one exists , is at most @xmath137 for a suitable constant @xmath138 .
so by the union bound , the probability that the algorithm returns a ` false positive ' is at most @xmath139 .
we know of essentially two methods for proving lower bounds on quantum query complexity : the polynomial method of beals et al .
@xcite , and the quantum adversary method of ambainis @xcite . for a few problems , such as the collision problem @xcite , the polynomial method succeeded where the adversary method failed .
however , for problems that lack permutation symmetry ( such as local search ) , the adversary method has proven more effective .
how could a quantum lower bound method possibly be applied classically ?
when proving randomized lower bounds , the tendency is to attack `` bare - handed '' : fix a distribution over inputs , and let @xmath140 be the locations queried so far by the algorithm .
show that for small @xmath35 , the posterior distribution over inputs , _ conditioned _ on @xmath140 , is still ` hard ' with high probability so that the algorithm knows almost nothing even about which location @xmath141 to query next .
this is essentially the approach taken by aldous @xcite to prove a @xmath44 lower bound on @xmath142 . in the quantum case
, however , it is unclear how to specify what an algorithm ` knows ' after a given number of queries .
so we are almost _ forced _ to step back , and identify general combinatorial properties of input sets that make them hard to distinguish .
once we have such properties , we can then try to exhibit them in functions of interest .
we believe this gloved attack can be useful for classical lower bounds as well as quantum ones . in our _ relational adversary method _ , we assume there exists a @xmath85-query randomized algorithm for function @xmath143 .
we consider a set @xmath144 of @xmath145-inputs of @xmath143 , a set @xmath146 of @xmath19-inputs , and an arbitrary real - valued _ relation function _ @xmath147 for @xmath148 and @xmath149
. intuitively , @xmath150 should be large if @xmath66 and @xmath151 differ in only a few locations .
we then fix a probability distribution @xmath26 over inputs ; by yao s minimax principle , there exists a @xmath85-query deterministic algorithm @xmath152 that succeeds with high probability on inputs drawn from @xmath26 .
let @xmath153 be the set of @xmath145-inputs and @xmath154 the set of @xmath19-inputs on which @xmath155 succeeds . using the relation function @xmath83
, we define a _ separation measure _ @xmath105 between @xmath153 and @xmath154 , and show that ( 1 ) initially @xmath156 , ( 2 ) by the end of the computation @xmath105 must be large , and ( 3 ) @xmath105 increases by only a small amount as the result of each query .
it follows that @xmath85 must be large .
undoubtedly any randomized lower bound proved using our relational method could also be proved bare - handed , without any quantum intuition
. however , our method makes it easier to focus on what is unique about a problem , and ignore what is common among many problems .
our starting point is the `` most general '' adversary theorem in ambainis s original paper ( theorem 6 in @xcite ) , which he introduced to prove a quantum lower bound for the problem of inverting a permutation .
here the input is a permutation @xmath157 , and the task is to output @xmath145 if @xmath158 and @xmath19 otherwise . to lower - bound this problem s query complexity , what we would like to say is this : _ given any _ @xmath145__-input _ _ @xmath159 _ _ and any location _ _ @xmath109 _ _ , if we choose a random _ _ @xmath19__-input _ _ @xmath160 _ _ that is ` related ' to _ _ @xmath159 _ _ , then the probability _ _ @xmath161 _ over _ @xmath160 _ _ that _ _ @xmath162 _ does not equal _ @xmath163 _ _ is small . in other words ,
the algorithm is unlikely to distinguish _ _ @xmath164 _ _ from a random neighbor _
_ @xmath160 _ _ of _ _ @xmath159 _ _ by querying _ _ @xmath109__. _ _
unfortunately , the above claim is false .
letting @xmath165 , we have that @xmath166 for _ every _ @xmath19-input @xmath160 , and thus @xmath167 .
ambainis resolves this difficulty by letting us take the maximum , over all @xmath145-inputs @xmath159 and @xmath19-inputs @xmath160 that are related and differ at @xmath109 , of the _ geometric mean _ @xmath168 . even if @xmath169
, the geometric mean is still small provided that @xmath170 is small .
more formally : [ ambainis][ambthm]let @xmath171 and @xmath172 be sets of inputs to function @xmath143 .
let @xmath147 be a real - valued function , and for @xmath148 , @xmath149 , and location @xmath109 , let@xmath173 where the denominators are all nonzero .
then the number of quantum queries needed to evaluate @xmath143 with at least @xmath174 probability is @xmath175 , where@xmath176 to illustrate we show the following .
[ ambainis]the quantum query complexity of inverting a permutation is @xmath14 .
let @xmath177 be the set of all permutations @xmath159 with @xmath178 @xmath179 , and @xmath180 be the set of permutations @xmath160 with @xmath181 . given @xmath182 and @xmath183 , let @xmath184 if @xmath159 and @xmath160 differ only at locations @xmath185 and @xmath186 , and @xmath187 otherwise .
then given @xmath188 with @xmath189 , if @xmath190 then @xmath191 , and if @xmath192 then @xmath193 .
so @xmath194 .
the only difference between theorem [ ambthm ] and our relational adversary theorem is that in the latter , we take the _ minimum _ of @xmath195 and @xmath196 instead of the geometric mean . taking the reciprocal then gives up to a quadratically better lower bound : for example , we obtain that the randomized query complexity of inverting a permutation is @xmath197 . however , the proofs of the two theorems are quite different .
[ classadv]let @xmath198 be as in theorem [ ambthm ] .
then the number of randomized queries needed to evaluate @xmath143 with at least @xmath174 probability is @xmath199 , where@xmath200 let @xmath79 be a randomized algorithm that , given an input @xmath66 , returns @xmath201 with at least @xmath174 probability .
let @xmath85 be the number of queries made by @xmath79 . for all @xmath202 , @xmath149 , define@xmath203 now let @xmath204 be the distribution over @xmath202 in which each @xmath66 is chosen with probability @xmath205 ; and let @xmath206 be the distribution over @xmath207 in which each @xmath151 is chosen with probability @xmath208 .
let @xmath209 be an equal mixture of @xmath210 and @xmath211 .
by yao s minimax principle , there exists a deterministic algorithm @xmath212 that makes @xmath85 queries , and succeeds with at least @xmath174 probability given an input drawn from @xmath213 .
therefore @xmath152 succeeds with at least @xmath214 probability given an input drawn from @xmath215 alone , or from @xmath216 alone .
in other words , letting @xmath153 be the set of @xmath202 and @xmath154 the set of @xmath149 on which @xmath152 succeeds , we have@xmath217 define a predicate @xmath218 , which is true if @xmath152 has distinguished @xmath148 from @xmath219 by the @xmath220 query and false otherwise .
( to distinguish @xmath66 from @xmath151 means to query an index @xmath109 for which @xmath221 , given either @xmath66 or @xmath151 as input . ) also , for all @xmath222 , define a score function@xmath223 this function measures how much `` progress '' has been made so far in separating @xmath66 from @xmath180-inputs , where the @xmath180-inputs are weighted by @xmath224 .
similarly , for all @xmath207 define@xmath225 it is clear that for all @xmath35,@xmath226 so we can denote the above sum by @xmath227 and think of it as a global progress measure . we will show the following about @xmath228 : 1 .
@xmath229 initially .
@xmath230 by the end .
3 . @xmath231 for all @xmath35 , where @xmath232 is the amount by which @xmath227 increases as the result of a single query .
it follows from ( i)-(iii ) that@xmath233 which establishes the theorem . part ( i ) is obvious . for part ( ii ) , observe that for every pair @xmath234 with @xmath235 and @xmath236 , the algorithm @xmath152 must query an @xmath109 such that @xmath237 .
thus@xmath238it remains only to show part ( iii ) .
suppose @xmath239 for some @xmath35 ; we will obtain a contradiction . let@xmath240 and let @xmath241 be the set of @xmath202 for which @xmath242 .
since@xmath243 it follows by markov s inequality that@xmath244 similarly , letting @xmath245 be the set of @xmath149 for which @xmath246 , we have@xmath247 in other words , at least @xmath84 of the increase in @xmath248 comes from @xmath234 pairs such that @xmath249 , and at least @xmath84 comes from @xmath234 pairs such that @xmath250 .
hence , by a ` pigeonhole ' argument , there exists an @xmath249 and @xmath251 with @xmath252 that are distinguished by the @xmath253 query .
in other words , there exists an @xmath109 with @xmath254 , such that the @xmath220 index queried by @xmath212 is @xmath109 whether the input is @xmath66 or @xmath151 . then since @xmath249 , we have @xmath255 , and hence@xmath256 which equals @xmath195 .
similarly @xmath257 since @xmath250 .
this contradicts the definition@xmath258 and we are done .
for our lower bounds , it will be convenient to generalize random walks to arbitrary distributions over paths , which we call _
snakes_. [ snake]given a vertex @xmath259 in @xmath11 and a positive integer @xmath260 ,
a _ snake distribution _ @xmath261 ( parameterized by @xmath259 and @xmath260 ) is a probability distribution over paths @xmath262 in @xmath11 , such that each @xmath263 is either equal or adjacent to @xmath141 , and @xmath264 .
let @xmath265 be the support of @xmath261 .
then an element of @xmath265 is called a _ snake _ ; the part near @xmath266 is the _ tail _ and the part near @xmath264 is the _
head_. given a snake @xmath267 and integer @xmath35 , we use @xmath268 $ ] as shorthand for @xmath269 .
[ elgood]we say a snake @xmath270 is @xmath271__-good _ _ if the following holds .
choose @xmath272 uniformly at random from @xmath273 , and let @xmath274 be a snake drawn from @xmath261 conditioned on @xmath275 for all @xmath276
. then 1 .
letting @xmath277 be the set of vertices @xmath7 in @xmath278 such that @xmath279 , we have@xmath280 \geq9/10.\ ] ] 2 . for all vertices
@xmath7 , @xmath281 \right ] \leq\varepsilon$ ] .
the procedure above wherein we choose a @xmath272 uniformly at random , then draw a @xmath282 from @xmath261 consistent with @xmath267 on all steps later than @xmath272will be important in what follows .
we call it _ the snake _ @xmath283 _ flicking its tail_. intuitively , a snake is good if it is spread out fairly evenly in @xmath11so that when it flicks its tail , ( 1 ) with high probability the old and new tails do not intersect , and ( 2 ) any particular vertex is hit by the new tail with probability at most @xmath271 .
we now explain our ` snake method ' for proving lower bounds for local search . given a snake @xmath267 , we define an input @xmath284 with a unique local minimum at @xmath266 , and @xmath21-values that decrease along @xmath267 from head to tail . then , given inputs @xmath284 and @xmath285 with @xmath286 , we let the relation function @xmath287 be proportional to the probability that snake @xmath282 is obtained by @xmath267 flicking its tail .
( if @xmath288 we let @xmath289 . )
let @xmath284 and @xmath290 be inputs with @xmath291 , and let @xmath7 be a vertex such that @xmath292 .
then if all snakes were good , there would be two mutually exclusive cases : ( 1 ) @xmath7 belongs to the tail of @xmath267 , or ( 2 ) @xmath7 belongs to the tail of @xmath282 . in case
( 1 ) , @xmath7 is hit with small probability when @xmath282 flicks its tail , so @xmath293 is small . in case
( 2 ) , @xmath7 is hit with small probability when @xmath267 flicks its tail , so @xmath294 is small . in either case , then , the _ geometric mean _
@xmath295 and _ minimum _
@xmath296 are small .
so even though @xmath294 or @xmath297 could be large individually , theorems [ ambthm ] and [ classadv ] yield a good lower bound , as in the case of inverting a permutation ( see figure 1 ) .
[ ptb ] snakefig.eps one difficulty is that not all snakes are good ; at best , a large fraction of them are .
we could try deleting all inputs @xmath284 such that @xmath267 is not good , but that might ruin some remaining inputs , which would then have fewer neighbors .
so we would have to delete _ those _ inputs as well , and so on ad infinitum .
what we need is basically a way to replace `` all inputs '' by `` most inputs '' in theorems [ ambthm ] and [ classadv ] .
fortunately , a simple graph - theoretic lemma can accomplish this .
the lemma ( see diestel @xcite for example ) says that any graph with average degree at least @xmath298 contains an induced subgraph with _ minimum _
degree at least @xmath299 .
here we prove a weighted analogue of the lemma .
[ subgraph]let @xmath300 be positive reals summing to @xmath19 .
also let @xmath301 for @xmath302 be nonnegative reals satisfying @xmath303 and @xmath304 . then there exists a nonempty subset @xmath305 such that for all @xmath306 , @xmath307 if @xmath308 then the lemma trivially holds , so assume @xmath309 .
we construct @xmath310 via an iterative procedure .
let @xmath311 .
then for all @xmath35 , if there exists an @xmath312 for which@xmath313 then set @xmath314 .
otherwise halt and return @xmath315 . to see that the @xmath310 so constructed is nonempty , observe that when we remove @xmath316 , the sum @xmath317 decreases by @xmath318 , while @xmath319 decreases by at most@xmath320 so since @xmath321 was positive to begin with , it must still be positive at the end of the procedure ; hence @xmath310 must be nonempty .
we can now prove the main result of the section .
[ kappathm]suppose a snake drawn from @xmath261 is @xmath271__-__good with probability at least @xmath174 .
then@xmath322 given a snake @xmath323 , we construct an input function @xmath324 as follows . for each @xmath325 , let @xmath326 ; and for each @xmath327 , let @xmath328 where @xmath329 is the distance from @xmath7 to @xmath259 in @xmath11 .
clearly @xmath324 so defined has a unique local minimum at @xmath330 . to obtain a decision problem , we stipulate that querying @xmath330 reveals an answer bit ( @xmath145 or @xmath19 ) in addition to @xmath331 ; the algorithm s goal is then to return the answer bit .
obviously a lower bound for the decision problem implies a corresponding lower bound for the search problem .
let us first prove the theorem in the case that all snakes in @xmath332 are @xmath271__-__good .
let @xmath333 be the probability of drawing snake @xmath267 from @xmath334 . also , given snakes @xmath335 and @xmath336 , let @xmath337 be the probability that @xmath338 , if @xmath339 is drawn from @xmath340 conditioned on agreeing with @xmath267 on all steps later than @xmath272 .
then define@xmath341 our first claim is that @xmath9 is symmetric ; that is , @xmath342 .
it suffices to show that @xmath343 for all @xmath272 .
we can assume @xmath267 agrees with @xmath282 on all steps later than @xmath272 , since otherwise @xmath344 .
given an @xmath345 , let @xmath66 denote the event that @xmath339 agrees with @xmath267 ( or equivalently @xmath282 ) on all steps later than @xmath272 , and let @xmath346 ( resp .
@xmath347 ) denote the event that @xmath339 agrees with @xmath267 ( resp .
@xmath282 ) on steps @xmath19 to @xmath272 . then@xmath348 \pr\left [ b_{x}% & = p\left ( y\right ) q_{j}% \left ( y , x\right ) .\end{aligned}\]]now let @xmath349 denote the event that @xmath350 , where @xmath277 is as in definition [ elgood ] . also , let @xmath324 be the input obtained from @xmath267 that has answer bit @xmath145 , and @xmath351 be the input that has answer bit @xmath19 . to apply theorems [ ambthm ] and [ classadv ] , take @xmath352 and @xmath353 . then take @xmath354 if @xmath349 holds , and @xmath355 otherwise
. given @xmath356 and @xmath357 with @xmath358 , and letting @xmath7 be a vertex such that @xmath359 , we must then have either @xmath327 or @xmath360 .
suppose the former case ; then@xmath361since @xmath282 is @xmath271-good .
thus@xmath362 similarly , if @xmath360 then @xmath363 by symmetry .
hence@xmath364 the latter since @xmath365 and @xmath366 for all @xmath367 and @xmath7 .
we now turn to the general case , in which a snake drawn from @xmath368 is @xmath271__-__good with probability at least @xmath174 .
let @xmath369 denote the event that @xmath267 is @xmath271-good .
take @xmath370 and @xmath371 , and take @xmath372 as before .
then since@xmath373 by the union bound we have@xmath374
so by lemma [ subgraph ] , there exist subsets @xmath375 and @xmath376 such that for all @xmath377 and @xmath378,@xmath379so for all @xmath380 with @xmath358 , and all @xmath7 such that @xmath381 , either @xmath382 or @xmath383 .
hence @xmath384 and @xmath385 .
in this section we apply the ` snake method ' developed in section [ snake ] to specific examples of graphs : the boolean hypercube in section [ boolean ] , and the @xmath47-dimensional cubic grid ( for @xmath3 ) in section [ ddim ] .
abusing notation , we let @xmath0 denote the @xmath386-dimensional boolean hypercube that is , the graph whose vertices are @xmath386-bit strings , with two vertices adjacent if and only if they have hamming distance @xmath19 . given a vertex @xmath387 , we let @xmath388 , \ldots , v\left [ n-1\right ] $ ] denote the @xmath386 bits of @xmath7 , and let @xmath389 denote the neighbor obtained by flipping bit @xmath390 $ ] . in this section
we lower - bound @xmath391 and @xmath392 . fix a ` snake head ' @xmath393 and take @xmath394 .
we define the snake distribution @xmath261 via what we call a _ coordinate loop _
, as follows . starting from @xmath395 , for each @xmath35 take @xmath396 with @xmath397 probability , and @xmath398 with @xmath397 probability .
the following is a basic fact about this distribution .
[ mixtime]the coordinate loop mixes completely in @xmath386 steps , in the sense that if @xmath399 , then @xmath400 is a uniform random vertex conditioned on @xmath263 .
we could also use the random walk distribution , following aldous @xcite .
however , not only is the coordinate loop distribution easier to work with ( since it produces fewer self - intersections ) , it also yields a better lower bound ( since it mixes completely in @xmath386 steps , as opposed to approximately in @xmath401 steps ) .
we first upper - bound the probability , over @xmath267 , @xmath272 , and @xmath402 $ ] , that @xmath403 ( where @xmath277 is as in definition [ elgood ] ) .
[ intersect]suppose @xmath267 is drawn from @xmath261 , @xmath272 is drawn uniformly from @xmath404 , and @xmath402 $ ] is drawn from @xmath405 .
then @xmath406 } \left [ x\cap y = s_{x , y}\right ] \geq0.9999 $ ] .
call a _ disagreement _ a vertex @xmath7 such that@xmath407 clearly if there are no disagreements then @xmath408 .
if @xmath7 is a disagreement , then by the definition of @xmath368 we can not have both @xmath409 and @xmath410 .
so by proposition [ mixtime ] , either @xmath411 is uniformly random conditioned on @xmath267 , or @xmath412 is uniformly random conditioned on @xmath402 $ ] .
hence @xmath413 } % \left [ x_{t}=y_{t^{\ast}}\right ] = 1/2^{n}% $ ] .
so by the union bound,@xmath414 } % \left [ x\cap y\neq s_{x , y}\right ] \leq\frac{l^{2}}{2^{n}}% = 0.0001.\ ] ] we now argue that , unless @xmath267 spends a ` pathological ' amount of time in one part of the hypercube , the probability of any vertex @xmath7 being hit when @xmath267 flicks its tail is small . to prove this , we define a notion of _ sparseness _ , and then show that ( 1 ) almost all snakes drawn from @xmath261 are sparse ( lemma [ hammingball ] ) , and ( 2 ) sparse snakes are unlikely to hit any given vertex @xmath7 ( lemma [ sparsegood ] ) .
[ sparse]given vertices @xmath415 and @xmath416 , let @xmath417 be the number of steps needed to reach @xmath7 from @xmath109 by first setting @xmath418 : = v\left [ i\right ] $ ] , then setting @xmath419 : = v\left [ i-1\right ] $ ] , and so on .
( after we set @xmath420 $ ] we wrap around to @xmath421 $ ] . )
then @xmath267 is _ sparse _ if there exists a constant @xmath138 such that for all @xmath387 and all @xmath298,@xmath422 [ hammingball]if @xmath267 is drawn from @xmath261 , then @xmath267 is sparse with probability @xmath423 . for each @xmath416 , the number of @xmath424 such that @xmath425 is at most @xmath426 .
for such a @xmath35 , let @xmath427 be the event that @xmath428 ; then @xmath429 holds if and only if@xmath430 = v\left [ i\right ] , \ldots , x_{t}% \left [ i - k+1\right ] = v\left [ i - k+1\right]\ ] ] ( where we wrap around to @xmath431 $ ] after reaching @xmath432 $ ] ) .
this occurs with probability @xmath433 over @xmath267 .
furthermore , by proposition [ mixtime ] , the @xmath434 events for different @xmath35 s are independent .
so let@xmath435 then for fixed @xmath436 , the expected number of @xmath35 s for which @xmath437 holds is at most @xmath438 . thus by a chernoff bound ,
if @xmath439 then@xmath440 < \left ( \frac{e^{cn-1}% } { \left ( cn\right ) ^{cn}}\right ) ^{\mu_{k}}<\frac{1}{2^{2n}}%\ ] ] for sufficiently large @xmath138 . similarly , if @xmath441 then@xmath442 < \left ( \frac{e^{cn/\mu_{k}-1}}% { \left ( cn/\mu _ { k}\right ) ^{cn/\mu_{k}}}\right ) ^{\mu_{k}}<\frac{1}% { 2^{2n}}%\ ] ] for sufficiently large @xmath138 . by the union bound , then,@xmath443 for every @xmath436 triple _ simultaneously _ with probability at least @xmath444
. summing over all @xmath445 s produces the additional factor of @xmath386 .
[ sparsegood]if @xmath267 is sparse , then for every @xmath446,@xmath447 \right ] = o\left ( \frac{n^{2}}% { l}\right ) .\ ] ] by assumption , for every @xmath448,@xmath449 & \leq\frac{\left| \left\ { t:\delta\left ( x_{t}% , v ,
t\operatorname{mod}% n\right ) = k\right\ } \right| } { l}% \\ & \leq\frac{cn}{l}\left ( n+\frac{l}{2^{n - k}}\right ) .\end{aligned}\]]consider the probability that @xmath450 $ ] in the event that @xmath451 .
clearly@xmath452 = \frac{1}{2^{k}% } .\ ] ] also , proposition [ mixtime ] implies that for every @xmath453 , the probability that @xmath454 is @xmath455 .
so by the union bound,@xmath456 \leq\frac{l}% { 2^{n}}.\ ] ] then @xmath457 \right ] $ ] equals@xmath458{c}% \pr_{j}% \left [ \delta\left ( x_{j},v , j\operatorname{mod}% n\right ) = k\right ] \cdot\\ \pr_{y}% \left [ v\in y\left [ j\right ] ~|~\delta\left ( x_{j}% % , v , j\operatorname{mod}n\right ) = k\right ] \end{array}% \right ) \\ & \leq\sum_{k=0}^{n}\frac{cn}{l}\left ( n+\frac{l}{2^{n - k}% } \right ) \left ( \frac{1}{2^{k}}+\frac{l}{2^{n}}% \right ) = o\left ( \frac{cn^{2}}{l}\right)\end{aligned}\]]as can be verified by breaking the sum into cases and doing some manipulations .
the main result follows easily : [ boolean]@xmath459 take @xmath460 .
then by theorem [ kappathm ] , it suffices to show that a snake @xmath267 drawn from @xmath334 is @xmath461-good with probability at least @xmath174 .
first , since@xmath414 } % \left [ x\cap y = s_{x , y}\right ] \geq0.9999\ ] ] by lemma [ intersect ] , markov s inequality shows that@xmath462 } \left [ x\cap y = s_{x , y}% \right ] \geq\frac{9}{10}\right ] \geq\frac{19}{20}% .\ ] ] second , by lemma [ hammingball ] , @xmath267 is sparse with probability @xmath463 , and by lemma [ sparsegood ] , if @xmath267 is sparse then@xmath464
\right ] = o\left ( \frac{n^{2}}% { l}% \right ) = o\left ( \varepsilon\right)\ ] ] for every @xmath7 .
so both requirements of definition [ elgood ] hold simultaneously with probability at least @xmath174 . in the boolean hypercube case
, we defined @xmath261 by a ` coordinate loop ' instead of the usual random walk mainly for convenience .
when we move to the @xmath47-dimensional grid , though , the drawbacks of random walks become more serious : first , the mixing time is too long , and second , there are too many self - intersections , particularly if @xmath465 .
our snake distribution will instead use straight lines of randomly chosen lengths attached at the endpoints , as in figure 2 .
[ ptb ] linefig.eps let @xmath466 be a @xmath47-dimensional grid graph with @xmath3 .
that is , @xmath466 has @xmath12 vertices of the form @xmath467 , \ldots , v\left [ d-1\right ] \right ) $ ] , where each @xmath390 $ ] is in @xmath468 ( we assume for simplicity that @xmath12 is a @xmath469 power ) .
vertices @xmath7 and @xmath9 are adjacent if and only if @xmath470 -w\left [ i\right ] \right\vert = 1 $ ] for some @xmath471 , and @xmath472 = w\left [ j\right ] $ ] for all @xmath473 ( so @xmath466 does not wrap around at the boundaries ) .
we take @xmath474 , and define the snake distribution @xmath368 as follows .
starting from @xmath395 , for each @xmath85 we take @xmath475 identical to @xmath476 , but with the @xmath477 coordinate @xmath478 $ ] replaced by a uniform random value in @xmath468 .
we then take the vertices @xmath479 to lie along the shortest path from @xmath476 to @xmath480 , ` stalling ' at @xmath475 once that vertex has been reached .
we call@xmath481 a _ line _ of vertices , whose _ direction _
is @xmath482 . as in the boolean hypercube case
, we have : [ mixtime2]@xmath261 mixes completely in @xmath483 steps , in the sense that if @xmath484 , then @xmath485 is a uniform random vertex conditioned on @xmath476 .
lemma [ intersect ] in section [ boolean ] goes through essentially without change .
[ sparse2]letting @xmath417 be as before , we say @xmath267 is _ sparse _ if there exists a constant @xmath138 ( possibly dependent on @xmath47 ) such that for all vertices @xmath7 and all @xmath298,@xmath486 [ hammingball2]if @xmath267 is drawn from @xmath261 , then @xmath267 is sparse with probability @xmath423 . [ sparsegood2]if @xmath267 is sparse , then for every @xmath487,@xmath447 \right ] = o\left ( \frac{n^{1/d}\log n}{l}\right ) , \ ] ] where the big-@xmath488 hides a constant dependent on @xmath47 .
the proofs of lemmas [ hammingball2 ] and [ sparsegood2 ] are omitted from this abstract , since they involve no new ideas beyond those of lemmas [ hammingball ] and [ sparsegood ] .
taking @xmath489 we get , by the same proof as for theorem [ boolean ] : [ grid]neglecting a constant dependent on @xmath47 , for all @xmath3@xmath490
i thank andris ambainis for suggesting an improvement to proposition [ upper ] ; david aldous , christos papadimitriou , yuval peres , and umesh vazirani for discussions during the early stages of this work ; and ronald de wolf and the anonymous reviewers for helpful comments .
e. farhi , j. goldstone , s. gutmann , j. lapan , a. lundgren , and d. preda .
a quantum adiabatic evolution algorithm applied to random instances of an np - complete problem , _ science _ 292:472476 , 2001 .
quant - ph/0104129 . | the problem of finding a local minimum of a black - box function is central for understanding local search as well as quantum adiabatic algorithms . for functions on the boolean hypercube @xmath0
, we show a lower bound of @xmath1 on the number of queries needed by a quantum computer to solve this problem .
more surprisingly , our approach , based on ambainis s quantum adversary method , also yields a lower bound of @xmath2 on the problem s _ classical _ randomized query complexity .
this improves and simplifies a 1983 result of aldous . finally , in both the randomized and quantum cases
, we give the first nontrivial lower bounds for finding local minima on grids of constant dimension @xmath3 . |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.